1

FINANCE & SUSTAINABLE DEVELOPMENT

LECTURE NOTES

2007-2009

Jean-Michel Lasry Delphine Lautier Damien Fessler

2

Contents Introduction Delphine Lautier Part 1. Climate Change Policies Chapter 1. The precautionary principle and the evaluation of environmental policies Olivier Guéant Chapter 2. Open Questions on the Economics of Climate Change Policy Roger Guesnerie Chapter 3. Long Term Policy-Making Ivar Ekeland Chapter 4. Intergenerational Discounting Intergenerational discounting: a new intuitive approach U. Rashid Sumaila, and Carl Walters Remembering the future, a commentary on “Intergenerational discounting: a new intuitive

approach” Michael H Prager and Karl W. Shertzer Making future generations count, comment on “remembering the future” U. Rashid Sumaila, and Carl Walters Part 2. Economic Concepts and Methodology Chapter 1. The Precautionary Principle as a Social Norm Olivier Godard Chapter 2. Coase against Coase: An Endogenous Theory of Externalities with Dynamic

Internalisation

Jan Horst Keppler Chapter 3. Mean Field Games and Oil production Olivier Guéant, Jean-Michel Lasry, Pierre-Louis Lions Part 3. Modelization and Applications

3

Chapter 1. Uncertainties and Risks in Water Resources Management Alain Bensoussan and Nadir Farhi Chapter 2. Optimal Energy Investment and R&D Strategies to Stabilise Atmospheric

Greenhouse Gas Concentrations V. Bosetti, C. Carraro, E. Massetti, A. Sgobbi, M. Tavoni Chapter 3. The Social Costs of Greenhouse Gas Emissions: An Expected Value Approach Samuel Fankhauser Chapter 4. Assessing Sustainability, a Comprehensive Wealth Accounting Measure –

application to Mozambique Pierre-Noël Giraud and Timothée Ollivier Part 4. Carbon Markets Chapter 1. Carbon Market and Climate Negotiations Christian de Perthuis Chapter 2. Price caps and price floors in climate policy Cédric Philibert Part 5. Socially Responsible Investment Chapter 1. Socially Responsible Investing: Myths and Realities Gunther Capelle-Blancart and Stéphanie Monjon Chapter 2. SRI as sustainable and responsible insurance: responsible investing is worth a

premium Stéphane Voisin and Hubert Jeanneau Chapter 3. Linking the concepts of corporate social responsability and sustainable

development: a story of hybridization Michel Capron and Françoise Quairel Biographies

4

Introduction

This book is a compilation of the lecture notes from the seminars held at the Finance and

Sustainable Development Chair from 2007 to 2009. These seminars, initiated by Jan Keppler,

professor at the University Paris-Dauphine, were intended to provide an extensive overview of

various questions related to Chair’s main theme. While looking for diversity in the subjects,

methods and approaches presented, we did not, however, pretend to be exhaustive. Neither does

this book. Nevertheless, after two years, it turns out that the contributions cover a wide field that

can be organized around five main ideas, each representing a separate part of this book.

During the editing process, we tried to stay as close as possible to the original

contributions. As a result, the book contains several chapters that take different points of view on

the same subject. The positions adopted may, to a certain extent, be mutually contradictory, a

heterogeneity that is not surprising in a new and controversial research field. It should be stressed

that these contributions reflect their authors’ views, and not necessarily those of the Chair.

The first part of the book is concerned with climate change policies. The question of the

relevance of traditional economic theories lies at the heart of its four chapters. The second part

deals with economic concepts and methodologies associated with sustainable development. It

presents the precautionary principle, the concept of externalities and the methodology of mean

field games. The third part is devoted to models and empirical applications in various fields, such

as water resources, greenhouse gas emissions and an example of a sustainable development path.

The fourth part deals with carbon markets, together with their theoretical justification and

historical development. The last part focuses on socially responsible investment and gives

insights on its definition, its links with sustainable development, and the way it could be used for

investment strategies.

Part 1. Climate change policies

5

Unsurprisingly, climate change is deeply imbricated with time. Thinking about climate

alters economists’ temporal horizon. In such a field, the time perspective can easily extend to

fifty or a hundred years, or even longer. The main question is the relevance of traditional

economic theory, for example, cost-benefit analysis or the model of economic growth. As the

first attempt at quantifying climate change policies, the Stern review is here a natural and thus

recurrent point of reference.

This first part of the book provides an opportunity for demonstrating that, in the long run,

ethical considerations cannot be ignored by economic analysis. The choice of a discount rate is a

matter of ethics, because such a rate determines the allocation of inter-temporal goods and

services between generations, and because future generations do not participate in decisions that

will affect them. However, this does not amount to saying that an economics framework is of no

use. On the contrary, it helps clarify the issues in a situation where uncertainties are high.

Olivier Guéant proposes a methodology for evaluating environmental policies in the light

of the precautionary principle. According to its definition in the French law (1995, n°95-101), the

precautionary principle must lead to measures with an acceptable economic cost. While putting

forward a methodology suited to the evaluation of environmental policies, Guéant also underlines

the need to revise, in such a context, the economic classical modus operandi. The latter can be

disentangled into three stages: selection of a utility function, evaluation of the probability of the

event (such as environmental damage), and evaluation of the magnitude of the event in question.

While risk analysis classically emphasizes the problem of event probabilities, in the case

of the environment the main issue is that of the utility function. If the agent is a benevolent

planner aiming to introduce environmental reforms, the use of a utility function raises a least

three questions. First, what would be the validity of the utility function for assessing a radical

change in the environment or a massive economic shock following the sudden scarcity of a not

easily substitutable raw material? Second, what do we know of the agent’s utility in a context

other than the current one? Three, does the agent himself know what his preferences would be in

a radically different world?

Olivier Guéant suggests that the probability of various utility functions should be

discussed. He assumes that the benevolent planner will probabilize his own utility function due to

his ignorance of preferences when evaluating a situation that is radically different from the

6

current one, and will consider the worst case among the various possible utility functions. In this

way, Guéant provides a micro-economic basis for the precautionary principle.

Roger Guesnerie organizes his contribution around two questions. First, is the discount

rate still relevant in the case of climate change? Second, what kind of policies might be adopted

to limit greenhouse gas emissions?

When applied in a traditional way, the discount rate is largely incompatible with inter-

generational equity. For the present value of future profits is very low when the time horizon is

remote and/or the discount rate is high. Thus incentives for investment in future generations’

needs are very limited: “Cost benefit analysis leads living generations to an ethically

unacceptable selfishness” (Guesnerie). Hence there is a need for a discount rate appropriate for

the very long term and compatible with sustainable development. The author argues that four

aspects must be taken into consideration in order to reconcile ethical considerations with

economic reason: the specificity of environmental goods in the long term, the imperfect

substitutability between consumption goods and environmental goods, a low rate of time

preference, and uncertainty. He then shows that neglecting such aspects, as in the Stern review,

may result in a significant underestimation of the benefits of climate policies.

Roger Guesnerie then comes back to the intellectual debate of prices versus quantity, and

indicates what the limits of the traditional conclusions are in a context of climate policy. A

quantity policy aims at controlling the total level of emissions – by way of quotas, for example –,

whereas a price policy serves to control the cost of pollution for the agents. Theoretically, from

the point of view of incentives, both should be considered as equivalent and hence substitutes,

provided that the information on costs and benefits about pollution reduction is perfect.

Guesnerie, however, explains that in the case of climate, even in the presence of perfect

information, substitution is impossible unless it is possible to control the global price of fossil

fuel.

Ivar Ekeland’s contribution focuses on economic growth in the long run. He first defines

the long term as ranging from 50 to 200 years. He then gives evidence of two difficulties

affecting long-term policy-making: high uncertainty and non-commitment. The first is due to the

fact that generally in the long term, either predictable events have a very high dispersion or there

7

are some events that are unpredictable. The second difficulty arises from the fact that whatever

policies are adopted today, their success will depend on the willingness of future generations to

implement them – something that is impossible to ensure. Ekeland then asks whether economic

theory is relevant in the long term. In part the answer is no, because there are ethical

considerations; and in part the answer is yes, because the economic approach gives arguments

and guidelines to politicians and business leaders. It nevertheless remains difficult to apply

economic theory straightforwardly. As Guesnerie points out, discounting the future is not a

simple exercise.

Having defined the problem, Ekeland then focuses on economic growth. He first presents

the standard model: a single good in the economy, identical firms in perfect competition, and

identical individuals who can be treated as a single, infinitely long-lived individual – the

representative consumer. If there is a benevolent and omniscient planner in the economy, he will

have to find a way to maximize the inter-temporal welfare of this representative consumer, and to

know whether there is a solution to such a problem and whether any such solution is unique. In

the absence of such a planner, there needs to be an equilibrium interest rate.

This is the reason why Ekeland then focuses on the determinants of the interest rate. He

first modifies classical theory so as to take account of the environment as a separate good, and

treats the consumption good as a numéraire. He next introduces uncertainty into the growth rate

of the economy. He also introduces uncertainty into the model, in that that there may be

unpredictable events in the long run. Lastly, he brings in the particular concerns associated with

intergenerational equity. In such a context, he then looks for the existence and uniqueness of

equilibrium.

The fourth chapter is in three sections. The first, by Rashid Sumaila and Carl Walters, is

an article on intergenerational discounting. The second is a commentary on it by Michael Prager

and Karl Shertzer, while the third is Sumaila and Walters’s reply.

Sumaila and Walters present a new approach to intergenerational discounting, aiming at

explicitly incorporating the perspectives of both the current and future generations. They try to

capture the human proclivity for altruism, which occurs at all levels of society. Standard

discounting usually fails to adequately model such behavior.

8

The discounting equation formulated by the two authors is such that the future benefits of

present investments are separated into several parts. Such benefits must accrue to the current

generation – the population of stakeholders – but they also must be distributed to future

generations. The discounting rate applied to the present population is the standard discount rate.

So too is the rate applied to each group of new stakeholders once it enters the current generation.

Before that time, the future benefits accruing to this future population is discounted at a specific

discount rate, termed the “future generation” discount rate, which measures the willingness of the

current generation to postpone benefits that it could keep for itself, in favor of wealth that will go

to future stakeholders. In this way the authors are able to take account of the interests of all

generations.

Naturally, with such an approach, there is a need not only to establish the two discount

rates – the standard and the future generation rates – but also the duration of a generation, so as to

calculate the value of future benefits and to give policy-makers design management solutions for

the natural environment. Sumaila and Walters make several suggestions in this respect.

In their commentary on this paper, Prager and Shertzer first point out that the

intergenerational discounting approach provides an attractive way to link economics to policy

and ethical considerations. They then suggest a new notation for denoting the future generation

discount rate. Finally, they ask whether or not a cap should be introduced into the discounting

equation, in order to insure that the discount factor derived from the intergenerational discounting

equation does not exceed one. In their answer, Sumaila and Walter explain that a discount factor

greater than one is both practical and necessary. Indeed it makes it possible to take into account

the fact that in real life there are sometimes negative discount rates.

Part 2. Economic concepts and methodology

The second part of the book focuses on economic concepts and methodology. The first

chapter, by Olivier Godard, focuses on the precautionary principle, which has become a new

standard for risk management in the fields of environmental protection, public health and food

safety. Scientific uncertainty underlies the precautionary principle, constituting a challenge not

only for scientific expertise, but also for the traditional foundations of the legitimacy of public

action. The second chapter is devoted to externalities, that is to say, impacts on our well-being

9

that the market system is unable to allocate in an optimal way. In 1937, Ronald Coase explained

that the market stops where transaction costs are too high and that, beyond this point, we enter

into the realm of externalities. Jan Keppler develops Coase’s insight. Unlike Coase, however, he

argues that the level of transaction costs cannot be taken as given. There is thus a place for

selective and innovative government intervention. In the third chapter, by means of an application

to oil production, Olivier Guéant, Jean-Michel Lasry and Pierre-Louis Lions explain the new

“functionalities” of mean field games, i.e. their capacity to introduce into economic models the

effects of social interactions between agents, which cannot be incorporated into classical

economic modeling.

Olivier Godard sets out, firstly, to situate the precautionary principle between the two

main pure models for tackling risk management, namely the early eradication of the risk at source

and the patient management of the risk when it matures and becomes well defined. The

precautionary principle concerns the attitude to adopt in the face of “potential collective risks”.

This expression refers both to potential dangers the existence of which is scientifically uncertain

and to dangers for which, despite their known existence, it is not yet possible rigorously to

establish the likely extent of the damage they may cause. The novelty of the principle is to bring

these collective risks into the domain of public intervention.

The comparison with other risk management models enables the author to introduce the

European doctrine of the precautionary principle as developed in the field of environmental

protection. In doing so, sensitive questions are addressed. How is the general demand for

proportionality to be understood in the context of precautionary policies? To whom is the

principle addressed? Who has the legal duty to implement it? In answer to these questions, the

notions of principle, rule and criterion are clarified, insofar as a number of criticisms directed at

the precautionary principle are based on a misunderstanding as to what it actually is and are

mistaken as to what it aims to do.

Olivier Godard concludes that the precautionary principle does not lead to a break in the

link between public action and science or reason. Nevertheless, the impossibility of adopting an

objectively and scientifically established probabilistic formulation leads to public action that

takes account of other reference points: in addition to the process of scientific expertise, there are

the processes of public debate and of dialogue with interested parties. The precautionary principle

10

calls for judgment as to potential risks on a case-by-case basis and therefore gives the legal

system a critical role to play. In addition, control of potential risks by society as a whole implies a

shift in public reasoning itself, toward an approach that places less emphasis on the quest for

scientific certainty and is more open to wide-scale organized debate within society.

Jan Keppler’s contribution is a plea in favor of a dynamic approach to the internalization

of external effects. He develops Coase’s insight that the level of transaction costs in the market

determines the amount of externalities, which provides arguments against government

intervention. Contrary to Coase, however, Keppler argues that the level of transaction costs

cannot be considered as given, thus making a case for selective and innovative government

interventions to reduce such transaction costs.

Externalities are approached as intrinsically new and dynamic impacts, whose transaction

costs diminish over time, a process that can be accelerated by appropriate government action. The

author suggests that in contrast, internalization through public intervention through Pigouvian

taxation is epistemologically untenable: if externalities had the desirable informational qualities

to let governments determine optimal tax levels, these same externalities would already have

been fully internalized by the market.

The final part of the article puts forward two internalization strategies based on the

dynamic re-interpretation of the Coasean approach. The first aims at developing feedback

mechanisms between generators of externalities and those affected by them through media other

than the market. The second seeks to reduce transaction costs in order to extend the domain

where markets can operate effectively by suggesting codification strategies for the informational

complexities characterizing externalities. While some of the concrete measures proposed are not

entirely new, they have so far arisen haphazardly from historical contingencies. What this chapter

provides is their systematic exploration on the basis of a dynamic re-interpretation of the original

Coasean insight.

In their contribution, Olivier Guéant, Jean-Michel Lasry and Pierre-Louis Lions first

recall the principal characteristics of mean field games. These are a set of tools which proved to

be very useful in the context of games with infinitely many players and are suited to economic

problems characterized by the presence of a large number of agents. In such a situation, mean

11

field games tools allow the modeling of effects which were sometimes ignored because of lacks

in the analytic and numerical tools needed to handle them.

In any quantitative modeling, there is a need for an equation expressing the optimization

problem of each agent. Usually this involves one equation for each agent (or each class of agent).

Most of the time, a Hamilton-Jacobi-Bellman (HJB) equation is chosen. An equation is also

needed to express the group’s behavior. When agents are atomized, the group is represented, in

the modeling, by a distribution on the state space. The dynamics of the distribution is governed

by a Kolmogorov equation. The Hamilton-Jacobi-Bellman equation is a backward equation,

whereas the Kolmogorov is a forward one. This equation system is not new. The novelty

introduced by mean field games is that equilibrium is defined such that the two equations are

doubly coupled: individual behaviors are given for the Kolmogorov equation and, at the same

time, the distribution of agents in the state space enters into the HJB equation. This means that

agents can incorporate into their preferences the density of states of other agents at the anticipated

equilibrium.

The three authors then apply mean field games to a problem in oil production. Here, the

supply of the exhaustible resources is provided by a continuum of producers, in a pure

competitive framework. The forward/backward essence of mean field games is at work when

considering the production of a resource whose availability is limited. The backward dimension

comes from the optimization of production timing and the forward dimension is directly linked to

the evolution of individual and global oil reserves.

Part 3. Models and applications

The third part of the book brings together four contributions. Each offers a model and its

application to a specific field: water resources, greenhouse gas emissions, and a sustainable

development path. These four chapters share a common concern: that of confronting theory with

economic reality. Such a task is not easy, as sustainable development is subject to huge

uncertainties.

Water issues grow steadily as the net supply of fresh water diminishes: a society lacking

water cannot survive in the long run. Alain Bensoussan and Nadir Farhi offer a paper centered on

12

uncertainties and risk in water resources management. They first focus on uncertainties and risks

associated with water resources management, of which there are many and which are often

complicated to isolate, partly because water is in practice linked to all geographical areas. The

authors identify four of these uncertainties. Hydrologic uncertainties are due to natural events

such as stream flows and rainfall. Hydraulic uncertainties concern the design and performance of

hydraulic structures. Structural uncertainties are linked to physical failures or operational

problems such as water saturation or loss of soil stability. Lastly, economic uncertainties are

associated with construction costs, damage costs, projected revenue, operation and maintenance

costs, inflation, and so on.

The authors place special emphasis on desalination, considered as a promising response to

water scarcity in many locations. They analyze the economic and environmental issues related to

desalination, as well as the uncertainties and risks associated with desalination projects. In order

to quantify the risks associated with uncertainties in water resources systems, Bensoussan and

Fahri put forward a model that relies on real options theory. Indeed the latter can be used to

evaluate investment opportunities in desalination projects. In their model, the value of the project

depends on three variables: the demand for water in the region covered by the desalination plant,

the supply of water provided by other facilities, and the cost of desalinating water. The authors

assume that this cost depends mainly on the price of energy. In order to determine the optimal

time to invest in the project, they solve a stopping time problem, thereby obtaining a decision rule

which indicates, at a given time and on the basis of the information available, whether or not it is

advantageous to invest.

The stabilization of atmospheric greenhouse gas concentrations at levels expected to

prevent dangerous climate change has become another important, global, long-term objective.

The aim of the contribution by Valentina Bosetti, Carlo Carraro, Emanuele Masseti, Alessandra

Sgobbi and Massimo Tavoni is to investigate the economic implications of stabilizing greenhouse

gas concentrations over the next century.

In order to analyze the complex geographical and intertemporal interactions of the main

socio-economic, technological and climatic variables influencing the concentration of carbon in

the atmosphere, and with the purpose of obtaining a quantitative assessment of public policies,

the authors use a hybrid climate-energy-economy model. This is designed to identify the optimal

13

investment profiles in the energy sector and in research and development for achieving a pre-

determined carbon concentration target. The chosen framework is game theory: the model

accounts for interdependencies and spillover across twelve regions of the world, and the optimal

strategies are the result of a dynamic game through which inefficiency costs induced by global

strategic interactions can be assessed. The conclusions reached with this model are naturally

different from what can be obtained with frameworks characterized by the presence of a central

planner or the existence of a single global economy. Lastly, the authors take into account the

effects of free-riding and the implications of technological evolutions in the energy sector. They

also examine the case of a backstop energy technology.

Two stabilization targets are considered: in the first, CO2 atmospheric concentrations

should be limited to 550 parts per million (ppm) by the end of the century. The second target uses

a figure of 450 ppm. The authors conclude that such targets are attainable, but require radical

changes in the energy sector and massive investment in research and development: energy

efficiency and the decarbonization of production have to improve. These changes naturally

involve a cost. According to the estimates presented here, the total discounted costs over the next

century, as a percentage of global gross domestic product, would be 0.3% if the 550 ppm

stabilization target is to be attained, and to 2.1% for the 450 ppm level.

Samuel Fankhauser’s contribution also focuses on greenhouse gas emissions. It provides

an order-of-magnitude assessment of the potential impact of global warming. Quantifying

greenhouse gas damage is not possible without allowing for the huge uncertainty prevailing in

this field. Most studies take uncertainty into account by working with different climate scenarios.

In his article, Fankhauser uses a stochastic model and incorporates uncertainty directly by

describing uncertain parameters as random. This allows a better representation of current

scientific understanding and also enables an entire damage probability distribution to be

calculated, therefore providing important additional information on the likelihood of the

estimates and the possibility of extremely adverse events.

The author estimates the monetary cost of CO2 emissions to have been in the order of

USD 20 per ton of carbon for emissions between 1991 and 2000. This rises over time to about 28

dollars per ton in 2030. Similar figures for CH4 and N20 are also provided. Like all global

warming estimations, these results are very uncertain. Consequently, their standard deviations are

14

high. Moreover, the distribution is positively skewed, even for those scenarios excluding the

possibility of a climate catastrophe. This means that even when abstracting from current extreme

events, a deeply disastrous outcome is likely to occur. Another important conclusion is that the

results crucially depend on the choice of the discount rate, and ethical considerations will

therefore have to be taken into account in the debate.

The main application of the estimates is project appraisal. For small projects, the expected

benefits directly depend on the quantity of gas emissions avoided per year, the monetary cost of

the gas emissions, and the lifetime of the investment. For large-scale projects which might have

an influence on the future emissions trajectory, matters are more complicated. Nevertheless, the

author considers that his estimates may suffice as a rough assessment, as long as the project only

slightly affects the trajectory.

The last chapter in this part deals with a sustainable development path. In their

contribution, Pierre-Noël Giraud and Timothée Ollivier estimate the variation in the productive

base of a particular economy – that of Mozambique – between 2000 and 2005. Their aim is to

assess the sustainability of that country’s development path. They define an economy’s

productive base as the set of different capital stocks: produced, human, social and natural capital.

They consider that a development path is sustainable as long as the society’s productive base

does not shrink.

The theoretical background of their estimates is the framework developed by Arrow et al.

(2007) and the World Bank for assessing the wealth of nations. Indeed such assessments are

fundamental for sustainability issues. The authors provide a detailed empirical application of this

framework in the case of Mozambique, with several methodological improvements. They first

add various elements regarding the health dimension of human capital. They also refine the

methodology developed by the World Bank in order to estimate the natural capital.

The study relies on extended datasets collected by the authors. It also uses a number of

studies published by international organizations, national ministries, non-governmental

organizations and Eduardo Mondlane University. It shows that Mozambican growth is driven

mainly by human and physical capital accumulation, while the pressure on natural capital

remains relatively low. The growth of total factor productivity, which can be understood as the

result of an accumulation of technological capital and performing institutions, significantly

15

enhances the outcome of the different capital assets. Even if population growth has a strong

downward effect on wealth per capita, the latter keeps growing during the period. The authors

conclude that, in spite of the approximations of their calculations, Mozambique is following a

sustainable path at the present time, in contrast to many other sub-Saharan countries.

Part 4. Carbon markets

Two different kinds of solution are usually envisaged for combating climate change: taxes

and markets. Of these solutions, Europe and – at least for certain greenhouse gases – the United

States choose to rely on markets. The fourth part of this book is devoted to such markets, and

more specifically to carbon markets. Christian de Perthuis first recalls the history of the creation

of such markets, and explains how they work. Cédric Philibert assesses the advantages of a cap-

and-trade mechanism as the one used for carbon markets. He also proposes strengthening it with

the introduction of a floor for transaction prices.

Christian de Perthuis first emphasizes that, from a theoretical point of view, in a situation

of perfect competition, taxes and permits are equivalent. Moreover, in the presence of imperfect

information, taxes should be recommended. However, the solutions which were adopted to

reduce greenhouse gas emissions were largely independent of debates among economists: a

market system, composed of two segments – the European system of quota trading and the

international projects market – was rapidly imposed.

De Perthuis stresses that the creation of these economic instruments is largely due to the

Kyoto protocol, and that carbon markets should not be seen as an alternative to public action in

the face of climate change. If they provide a way of setting a carbon price that enables strong

incentives to be given to economic actors by decreasing the cost of emissions reductions, this

price depends largely on the quotas set by governments. Were European governments suddenly

to abandon their commitments, the price of carbon would collapse and the market would

disappear. If governments act in concert, the carbon market will gain in depth and effectiveness.

If they do not, this market will become fragmented and therefore less effective both economically

and ecologically.

16

The author claims that it is on the basis of these economic instruments that the

international community will be in a position to build a more ambitious climate agreement in

order to manage the many risks that climate change exposes us to. He thinks that for the next

stages of international negotiations on climate change, three parameters should play a key role:

first, the advance represented by the current carbon trading system; second, the benefits that

major emerging countries such as China, India and Brazil obtain by being able to use the

international carbon market to strengthen their emissions reduction efforts; third, the likely

introduction, during the next few years, of a federal cap on greenhouse emissions in the United

States. Without a carbon market, the chances of a post-Kyoto international agreement would be

poor.

In his contribution, Cédric Philibert assesses the long-term economic and environmental

effects of introducing price caps and floors in a hypothetical climate change mitigation

architecture.

Future unabated greenhouse gas emissions trends depend on uncertain future economic

growth, energy intensity, and the carbon intensity of the energy mix, which in turn depends on

fuel prices. Price caps have been suggested as a way to contain these uncertain costs. Under a

price cap, emissions beyond the targets would take place, but emitters would need to buy

additional allowances at a set price. While cap-and-trade would make the objective cost-effective,

if the cost were nevertheless to exceed some pre-set threshold, emissions above the targets would

be heavily taxed. Price floors could also be designed that would kick in if costs were much lower

than expected. The floors would help maintain the emission outcomes of mitigation policies and

have long-lasting effects on abatement costs through technology development.

In order to examine these assumptions, the author develops a climate policy costs model.

This highly aggregated model of the global economy makes no distinctions between countries or

sectors. It includes abatement cost curves derived from International Energy Agency expertise

and publications. To take full account of uncertainty, thousands of Monte Carlo simulations are

carried out so as to study various combinations of targets, price caps and price floors.

This quantitative analysis confirms that introducing price caps could significantly reduce

economic uncertainty: the expected costs could be decreased by about 50% and the uncertainty

on economic costs could be a whole order of magnitude lower. Reducing economic uncertainties

17

may spur on the adoption of more ambitious policies. Meanwhile, price floors would reduce the

level of emissions beyond the objective if the abatement costs turned out to be lower than

forecast. If caps and floors are commensurate with the ambition of the policy pursued and

combined with slightly tightened emission objectives, climate results could on average be similar

to those achieved with “straight” objectives (with no cost-containment mechanism).

Part 5. Socially responsible investing

The last part of this book contains three contributions on socially responsible investing

(SRI). Defining SRI, as we shall see, is a real challenge. In a first approach, Gunther Capelle

Blancard and Stéphanie Monjon characterize it as “a practice which selects investments not only

on the basis of financial criteria such as profitability and risk, but also by integrating an array of

ethical, environmental or social concerns”. The two authors try to distinguish between myth and

reality in socially responsible investing. In the second chapter, Stéphane Voisin and Hubert

Jeanneau focus on assessing the performance of socially responsible investing. They argue that,

because SRI attempts to include environmental, social and governance factors in the investment

process, the main considerations to be assessed in regard to performance should be environmental

and social. In the third chapter, Michel Capron and Françoise Quairel analyze the relationships

between sustainable development and corporate social responsibility (CSR). The European

consensus on the link between sustainable development and CSR is relatively new and is far from

universally accepted.

Recognizing the fact that socially responsible investing enjoys today a wide consensus, in

regard to the reconciling of finance and sustainable development, Gunther Capelle-Blancard and

Stéphanie Monjon nevertheless focus on the problems and challenges associated with it.

The first part of the chapter serves as a general introduction. Following a brief historical

overview, the authors refer to the different forms of SRI, present the major players and describe

this changing market. They then go on to discuss the issues raised by SRI. The most important, if

not the most complex, is conceptual: how is corporate social responsibility to be defined? The

second issue is the scope of SRI. The third, namely the connection between social,

18

environmental, economic and financial performance, is whether or not SRI could be used as a

substitute for regulation.

The authors show that in fact the SRI market share is low: a little more than 10% on a

broad definition, and considerably less for “Core SRI”. There is substantial SRI growth in

Europe, but its market share is stagnant in the US. Moreover, it seems that, for now, SRI does not

have a significant impact on firms’ cost of capital. In addition, on the basis of on a careful review

of the academic literature, it is clear that the financial performances of SRI funds are neither

better nor worse than those of traditional mutual funds.

The authors thus emphasize that although the intention of doing well by doing good is

undoubtedly admirable, it can lead to wishful thinking and blindness. They warn against two

pitfalls. First, attempting to overly inflate the SRI market may lead to a weakening of the concept

and to the development of greenwashing. Second, claiming that SRI outperforms means opening

the door to lobbies that will unquestionably use the argument to defend self-regulation. SRI must

not be used as a substitute for regulation. Finally, the authors ask whether or not the idea of over

performance is really in line with the practices dictated by SRI.

Stéphane Voisin and Hubert Jeanneau reframe the question of SRI performance by

arguing that environmental and social benefits have an intrinsic value.

The authors recall that the main objective of sustainable and responsible investing is not

to outperform. They suggest rather that it provides a “positive” extra-financial dividend in

response to investors’ extra-financial expectations and motivations. The most tangible and

material characteristic of corporate social responsibility is the protection against extra-financial

risks (litigation, regulation, etc). Picking a sustainable and responsible stock means hedging those

risks. If the market prices the risks correctly, it indicates precisely the cost of this insurance

premium. It also provides financial stability and security which, in today’s context of strong

political pressure for more sustainable financial markets, is certainly not negligible.

In their study, Voisin and Jeanneau have developed a methodology in order to identify the

nature of the risks that can come to bear on a financial materiality and to determine how they can

be relevant to a company. In theory, this protocol approach answers to all types of environmental,

social and public health issues. It does not, however, apply to most governance issues. Such risks

may prove to be damaging for some investors – for a fund, for example – as they tend not to be

19

taken into account by traditional financial analysis and common practices in the market. By

anticipating these risks, an investor can reap the benefits of a low cost insurance if he is able to

identify the companies that are most committed to achieving social and environmental objectives.

To summarize, Voisin and Jeanneau insist on the fact that SRI should stick to its core promise of

extra-financial dividends and discard any claim to outperformance.

In their article, Michel Capron and Françoise Quairel analyze the relationship between

corporate social responsibility and sustainable development. Their study aims at highlighting the

power of terminology, power all the greater in that the terminology is used in published writings.

The authors first explain that, in Europe, three main conceptions about corporate social

responsibility coexist. The first is inherited form the paternalism of the 19th century; the second is

defined as a utilitarian conception; the last is termed “sustainability”. Capron and Quairel then

present the process through which corporate social responsibility and sustainable development

were linked. In their view, referring to sustainable development in order to define corporate

social responsibility is part of an institutionalization process. According to Scott (1995),

institutions impart stability and meaning to social behavior within an organization.

Institutionalization derives from changes through which actions become “usual” in the

organization.

Through such changes, companies became sustainable development actors, and corporate

social responsibility and sustainable development became intimately linked. The authors point

out that the imputation of responsibility toward society has shifted from the political to the

corporate sphere. Although such institutionalization helps reinforce corporate legitimacy, it

nevertheless has major limitations, in that companies are not concerned with the production of

public goods. Rather they belong to the private domain and are subject to the pressure of

competition. Within a competitive context, sustainable development is presented as beneficial for

a particular company’s competitiveness, whereas what is at stake is the responsibility of all

companies. This contradiction raises the issue of monitoring and regulating a global market.

20

Part One -Climate Change Policies

Unsurprisingly, climate change is deeply imbricate with time. Thinking about climate

modifies economists’ temporal horizon. In such a field, the time perspective can easily extend to fifty or a hundred years, or even longer. The main question is the relevance of traditional economic theory, such as cost-benefit analysis or the model of economic growth. As the first attempt to quantify climate change policies, the Stern review is here a natural and thus recurrent reference.

Olivier Guéant proposes a methodology for evaluating environmental policies in the light of the precautionary principle. While putting forward a methodology suited to the evaluation of environmental policies, he also stresses the need to revise, in such a context, the economic classical modus operandi. While risk analysis typically emphasizes the problem of event probabilities, in the case of the environment, the primary issue is that of the utility function.

Roger Guesnerie organizes his contribution around two questions. The first is that of the discount rate which, when applied in a traditional way, is largely incompatible with inter-generational equity. The author proposes taking into account the specificity of environmental goods, the imperfect substitutability between consumption and environmental goods, a low rate of time preference, and uncertainty. The second question is related to the kind of policies that might be adopted in order to limit greenhouse gas emissions. Guesnerie then comes back to the price versus quantity debate and indicates what the limits of the traditional conclusions are in a context of climate policy.

Ivar Ekeland’s contribution focuses on economic growth in the long run. He first presents the standard model. He then modifies classical theory to take into account the presence of the environment as a separate good and treats the consumption good as a numéraire. He next introduces uncertainty into the growth rate of the economy and into the model. Lastly, he incorporates the special concerns associated with intergenerational equity.

Sumaila and Walters offer an approach on intergenerational discounting that aims at explicitly incorporating the perspectives of both the current and future generations. They try to capture the human proclivity for altruism to occur at all levels of society. Standard discounting usually fails to adequately model such behavior. The discounting equation developed by the authors is such that the future benefits of present investment are separated into several parts. As well as accruing to the present generation, such benefits must also be distributed to future generations.

This first part of the book is an opportunity for emphasizing that, in the long run, ethical considerations cannot be ignored by economic analysis. The choice of a discount rate is a question of ethics because the discount rate determines the allocation of inter-temporal goods and services between generations, and because future generations do not participate in decisions that will affect them. This, however, is not to say that an economic framework is of no use. On the contrary, it helps clarify the issues in a situation where uncertainties are high.

21

The precautionary principle and the evaluation of environmental policies1

Olivier Guéant

Abstract: This general readership article proposes an approach to evaluate what an economically and socially acceptable cost can be in the context of policies that aim at improving the environment. We stress the difficulty of considering a pre-determined utility function and propose an unusual methodology which is based on considering a distribution on the space of utility functions. One of our outcomes is a microeconomic foundation of the precautionary principle: the simple fact of hesitating between several utility functions invites to use the most environment-friendly one. Keywords: Precautionary principle - Long term discount rate – Ecological discount rate - Sustainable development

1 This article for a general readership is in part inspired by Guéant et al (2010).

22

The French law dated 2 February 1995 (Law No. 95-101) on the strengthening of

environmental protection is inspired by the precautionary principle, according to which "the

absence of certainty, given current scientific and technical knowledge, must not delay the

adoption of proportionate and effective measures aimed at preventing the risk of serious and

irreversible damage to the environment at an economically acceptable cost”2. Since then this

precautionary principle has been further strengthened, in that it now has a constitutional force

through the introduction of the charte environnementale into the preamble of the French

constitution in March 2005.

The debate on the precautionary principle usually takes place on scientific terrain, in that its

aim is to define appropriate responses for dealing with risks of uncertain magnitude and is

most often based on a range of more or less detailed proofs and scientific models. In the case

of environmental protection and global warming – our present concern –, the political debate

echoes the scientific one, and is likely to lead to coercive political decisions to protect the

environment and mankind.

If the scientific community and the political class are at the head of the table in this debate,

economists seem strangely absent, given that the precautionary principle, according to the

above-quoted definition, must lead to measures with an acceptable economic cost. This

seeming paradox is even more surprising in that the economic stakes are crucial and because

the main point is to make the last part of the definition meaningful. The economic issue is

arguably the heart of the precautionary principle in terms of its political application. Thus in

this general readership article, we propose putting forward a methodology for evaluating

environmental policy. We shall also revise the classical pricing approach, i.e. the financial or

actuarial approach, which is unsuited to this situation.

1. The need to revise the classical approach

The evaluation of inter-temporal policies in view of the current uncertainties is a question that

may seem classic to economists, one of whose missions is to value goods in relation to other

goods. However, in regard to the environment, the problems are far more complex, for several

reasons. Firstly, – a relatively minor objection – the goods and resources at stake are most

2 The original text, in French, is: “l'absence de certitudes, compte tenu des connaissances scientifiques et techniques du moment, ne doit pas retarder l'adoption de mesures effectives et proportionnées visant à prévenir un risque de dommages graves et irréversibles à l'environnement à un coût économiquement acceptable”.

23

often public goods or even non-commercial goods. Secondly, economists’ classical modus

operandi is not necessarily best suited to environmental issues. Indeed, it is based on the

concept of utility function, which we can assume is chosen by a benevolent planner for global

environmental reforms.

In practice, we know that a utility function should be as well calibrated as possible, carefully

summing up preferences and focusing on the parameters under scrutiny (elasticity of

substitutions between goods, elasticity of inter-temporal substitution, rate of time-preference,

etc.) – and we are well aware of the inherent difficulty of this type of calibration. Yet, even if

a utility function corresponding to agents’ preferences and thus to the preferences of a

benevolent planner could be successfully extracted from complex data, what would be the

validity of such a utility function for assessing a radical change in the environment or a

massive economic shock following the sudden scarcity of a difficult-to-substitute raw

material?

While the first objection seems debatable because we can always try to construct an ad hoc

utility function taking into account immaterial goods, even if that means calibrating from

imperfect data, the second addresses the very basis of economic rationality: what do we know

of an agent's utility in a context other than the current one?, and furthermore, – and this is the

greatest pitfall – does the agent himself know what his preferences in a radically modified

world would be?

If we apply these ideas to the environment, it is possible to create a model that accounts for an

elasticity of substitution between consumer goods and previously defined environmental

goods3. By complicating the model slightly, we obtain a utility function that would be a good,

though local, approximation of agents’ today utility function. In the absence of extreme

events, nobody is able to approximate, even imperfectly, an agent's utility function or the

value of the elasticity of substitution considered above following such an extreme event,

precisely because of the extreme nature of the event.

3 The article in this book by Roger Guesnerie and more precisely the first part of “Ecological Intuition and cost-

benefit analysis” is concerned with a two-goods model. This model is extensively used in Guesnerie (2004) and

Guéant et al. (2010) and is the basis of our discussion.

24

This remark prompts us to reconsider the standard economic analysis of costs/benefits in a

random environment. Risk analysis does indeed consist in assigning preferences, i.e. a utility

function, to agents and to evaluate the impact in terms of utility of a random serious event

whose probability of occurrence is evaluated elsewhere. In finance, for example in portfolio

management, we usually consider the impact on a saver’s utility of a bull or bear market

movement on portfolio assets, and we reason using a given utility function in order to deduce

the content of optimal portfolios. In insurance, in order to set a premium, we evaluate the

agent’s ability to pay according to his presumed utility function and on the frequency and

scale of the event occurrence covered by the insurance policy.

To sum up, the classical modus operandi is divided into three stages, as follows:

1. Selection of a utility function.

2. Evaluation of the probability of an event, such as damage

3. Evaluation of the magnitude of the event in question

Thus if we transfer the economist’s classical tools for sustainable development issues as

required, particularly in order to reach a figure for the precautionary principle, we must

choose a suitable utility function, and if we take a utility function adapted to the current

context, we will inevitably come to outdated conclusions. The reason is simple: the utility

function is unsuited to the post-damage context. Thus we can see that while risk analysis

classically places the emphasis on a discussion around event probabilities and ignores the

question of the choice of the utility function, environmentally speaking the issue is first and

foremost that of discussing the utility function. More specifically, the probability of various

utility functions should be discussed, because it is indeed possible to reconsider the rationality

concept and assume that the agent himself, or the benevolent planner representing agents, will

probabilize his own utility function due to his ignorance of preferences when evaluating a

situation that is radically different from the current one.

This slightly unorthodox method is even more justifiable if we accept the conclusions of

various studies, including those carried out by the most sceptical in regard to environmental

change. Indeed, the question is not to know whether these environmental upheavals will

occur, but rather to know how intense they will be – in terms of utility – in order to minimize

the impact on what economists call social welfare, which is the main issue here in terms of its

25

assessment… the benevolent planner’s utility function is in this case none other than a

representation of this social welfare.

2. The influence of a random variable on preferences

Determining an economically acceptable cost – which most often means acceptable to us and

future generations – boils down to considering a distribution of probabilities in the space of

utility functions and to supposing that the agents' utility function after an event is one of the

utility functions considered. As for the environment, which is our application example, the

event does not even need to be extreme for this methodology to be essential. The disutility of

pollution was unquestionably discovered at the beginning of the industrial era in large cities,

and at this time the agents’ utility function did not take environmental goods into account –

this is of course an amusing anachronism: the concept of utility function as such came much

later, even though utilitarianism as moral philosophy arose in the 19th century – since it was

public and present in constant quantities. Similarly, if global warming causes even a slight

rise in sea levels, many agents will discover the importance of environmental variables in

their utility function. This discovery could be given concrete expression by the absence of

precise answers, prior to the event, to the question: “How much would you be ready to pay to

reduce the average sea level by 2cm if the it rose by 20cm”; whereas after such an increase in

the sea level, such a question, in view of the consequences, would inevitably generate

passionate responses.

To answer the question of utility function choice or rather, as we have seen, the influence of

the choice of a particular probability measure on the ‘utility function space’, we will consider

a simple example developed in Guéant et al. (2010), consisting in a parsimonious model

suited to the evaluation of environmental policies. Consider two goods – a consumer good and

an environmental good – and suppose that the agents do not know, prior to a shock occurring

to the environmental good – a shock which for simplicity's sake we suppose to be known at a

future date, to insist specifically on the aspects of a random variable on the utility – what the

elasticity of substitution is between the two goods included in their utility function. This

uncertainty regarding the elasticity of the substitution parameter, which we assume to sum up

the utility function – in reality, we use a CES utility function with constant elasticity of

substitution between the two goods in a given period, and we add, given the inter-temporal

nature of the issue, an inter-temporal elasticity of substitution –, will disappear once the shock

26

occurs and we will know at the same time as the agent himself, who is assumed to be unaware

of his preferences in regard to this elasticity of substitution, whether it be high or low.

To simplify, we will suppose that an agent, viewed here as a benevolent planner, does not

know if he is 'environmentalist' or 'economist', i.e. whether he has a small elasticity of

substitution, and will therefore be highly interested in environmental goods if availability

decreases, or a high elasticity of substitution between the two types of goods, which

corresponds to making environmental and consumer goods, if not interchangeable, at least

relatively comparable from a utility point of view.

A question that naturally arises is the one we asked at the beginning, i.e. what is the price that

the benevolent planner is ready to pay to avoid a catastrophe, in a given period, that would

irreversibly damage the environment? It should be said that this price exactly represents the

benevolent planner’s willingness to pay and thus, if the planner fully represents the agents, the

socially acceptable or “economically acceptable” price (in the terminology used in the French

law).

The following diagram has as its x-coordinate the probability p of having a utility with low

elasticity, i.e. being 'environmentalist', as defined above, and as its y-coordinate the price that

the benevolent planner will be willing to pay to avoid a low magnitude catastrophe in 100

years, but one which would irreversibly damage the environment. As always in economics

(since economists always reason in terms of relative price), this price is normalized and we

bring it back to the one that the planner would be willing to pay to avoid a catastrophe of the

same magnitude but which would now only affect the environment for one year4.

4 In his chapter in this book Roger Guesnerie has used the variable m for a similar calculation.

27

Fig. 1: Evolution with p of the price of a forward perpetual bond in 100 years

Diagram note: In this diagram, we suppose that the discount rate, i.e. the rate of time-

preference, is 1% and the interest rate is 3%, that the inter-temporal elasticity of substitution

is 1.5 and that the representative agent hesitates between an elasticity of substitution between

the two goods of 0.8 (‘environmentalist’ case) or an elasticity of substitution of 1.2

('economist' case).

This diagram contains a lot of information and should be analyzed as follows. For an

individual, knowing for sure that he has an 'economic' orientation (p=0), the price of an

environmental perpetual bond starting in 100 years is approximately 12. This means that the

planner is ready to pay a price 12 times higher than the price he would be ready to pay to

avoid a catastrophe now, which will only have an impact for one year. For an agent knowing

for sure that he will be an ‘environmentalist’ after the catastrophic event, this multiplier goes

from 12 to 28. However, what is striking in this diagram is not the relation between these two

figures, no matter how high, but the change in the perpetual bond price with probability p of

having a utility function that gives more weight to the environment. This shift is in fact very

fast when p is close to 0, indicating that an individual who is definitely 'economist' has very

different behavior than an individual with doubts, no matter how slight. Conversely, the

simple fact of not knowing his orientation brings together the benevolent planner and the

purely ‘environmentalist’ individual, as defined by our rather imperfect terminology. This is

the dimension that will now be of interest to us.

28

As mentioned previously, in the context of global warming, it is less important to know what

will happen than to know how to quantify the disutility caused by the consequences of global

warming. The fact of not knowing, due to lack of experience, the importance of the

environmental variable means that the variable p in the diagram above cannot be too close to

0 or 1. As we have seen, the price is a highly concave function of p. Hence the simple fact of

not knowing who one really is, in terms of preferences, implies an almost identical behavior

to that of the environmentalist individual, including paying a high price for the perpetual

bond. Therefore, as an initial approximation caused by doubt, the problem of utility function

choice does not boil down to the determination of the agents’ utility function today, but to the

most environmentally demanding of utility functions that they may have after a catastrophe.

Once the utility function is chosen according to this rule, the classical actuary or risk analyst’s

methodology can be used mutatis mutandis. But as we have seen in a context where

rationality is limited by the inability to judge one’s own preferences, as is the case in most

problems in which the precautionary principle is applied, the first stage of utility function

choice is crucial because it responds to a new logic.

3. Conclusion In conclusion, let us return to the wording of the French law on the precautionary principle.

For economic calculation, and thus for the determination of an acceptable price, what comes

first is not “the absence of certainty, given current scientific and technical knowledge” as to

the potential damage (even if this is large) but doubt regarding the degree of disutility of said

damage. This doubt makes the risk analyst consider the worst case among all the possible

utility functions, before considering the question of probabilities in regard to potentially

harmful events.

In fact, we have not only answered the initial question as to the appropriate methodology for

determining the benevolent planner’s willingness to pay for environmental improvements, but

we have also, in a way, economically micro-funded a precautionary principle. The choice that

must be made regarding preferences can indeed be interpreted as a precautionary principle

rationalized by economic calculation and not by ecological intuition. Therefore this utility-

inspired redefinition of the precautionary principle gives further meaning to the 'economically

acceptable price' referred to in the law dated 2 February 1995.

29

References

[1] Dasgupta P. (2001), Human Well-Being and the Natural Environment, Chapter 6, Oxford University Press. [2] Gerlagh R. and B.C.C. van der Zwaan (2002) Long-term substitutability between environmental and man-made goods, Journal of Environmental and Economic Management 44, p329-345. [3] Gollier C., B. Jullien and N. Treich (2000) "Scientific Progress and irreversibility : an economic interpretation of the Precautionary Principle" Journal of Public Economics, 75; 229-253. [4] Gollier C. (2008), Ecological discounting. Toulouse School of Economics, Working Paper. [5] Guéant O., Lasry J.-M., Zerbib D.-O. (2007), Autour des taux d’intérêt écologique. Cahier de la Chaire Finance et développement durable, 3. [6] Guéant O., Guesnerie R, and Lasry J.M. (2010), Ecological intuition versus Economic reason, mimeo. [7] Guesnerie R. (2004), Calcul économique et développement durable, Revue Economique. 55-63, 363-382. [8] Heal G. (2006), "Intertemporal Welfare Economics and the Environment", in the Handbook of Environmental Economics, Elsevier. [9] Hoel M., Sterner T. (2007), "Discounting and relative prices", Climatic Change. [10] Hourcade J.-C., Lecoq F. (2002), Incertitude, irréversibilités et actualisation dans les calculs économiques sur l’effet de serre, in Kyoto et l’économie de l’effet de serre, La Documentation Française, Complément D. [11] Lomborg B. (2001), The Skeptical Environmentalist: Measuring the Real State of the World, Cambridge University Press. [12] Milleron J.C., Guesnerie R., Crémieux M. (1978), Calcul économique et décisions publiques, La Documentation Française. [13] Stern N. (2006), Stern Review on the Economics of Climate Change. [14] Weitzman M. (2007) The Stern Review and the economics of climate change, Journal of Economic literature, 45, 3, 703-724. [15] Weitzman, Martin L. (2008) On Modelling and Interpreting the Economics of Catastrophic Climate Change, American Economic Review 91 (1) p. 260-271.

30

Open Questions on the Economics of Climate

Change Policy

Roger Guesnerie

Abstract : The “open questions on the economics of climate change policy” that are considered here concern first the long run dimension of the climate change issues and the “prices versus quantities” debate. On the fist subject, the paper attempts to reassess the controversies on the long run discount rate, with the aim of reconciling “economic reason” and “economic intuition”. On the second point, the analysis puts the emphasis on the interaction between policies and the markets for fossil fuels Keywords : Cost-benefit analysis, long run discount rate, environmental resources, taxes, cap

and trade, fossil fuels

31

This paper consists of two essays which go through two of the major theoretical

questions on which contemporary debates about climate policy rely. The first essay has

itself three parts dealing with the issues related to the discount rate. Since the publication

of the Stern Review, controversy has resumed with a renewed intensity. I will present the

main points of the debate, insisting on my own approach of the matter without intending to

be exhaustive. The second essay focuses on a traditional debate in environmental

economics: the comparison and the good use of price control policies and alternative

policies leading to control quantities. I argue that the climate problem calls for a complete

reassessment of the standard arguments, in order to bring the interaction between the

climate policies and the fossil fuel market to the centre of the analysis.

1. Ecological Intuition and cost-benefit analysis

The discount rate is a key ingredient of cost-benefit analysis. Environmentalists

have often criticized cost-benefit analysis, arguing that discount rates or standard value

taken for discount rates lead to dismiss the environmental concerns. I confront here what

may be called “ecological intuition” with “economic reason”, as it is conveyed by cost-

benefit analysis. I first present the main elements of the debate around the discount rate

(1.1). In (1.2) I sketch an “ecological” cost benefit analysis that takes into account one key

dimension of ecological intuition. The consequences of the approach are further developed

in (1.3)

1.1. The argument about the discount rate

A discount rate brings a future profit at some weaker “present value”. Due to the

compounding of interests, a supposedly stable over time discount rate, does “crush” the

future. Let’s pick up some rough estimations: with a 10% discount rate, spending 1€ today

should generate a profit of 120€ in 50 years to be just beneficial and 14 000€ in 100 years.

With a 7% discount rate, 1€ spent today amounts to a little less than 30€ in 50 years and to

860€ in 100 years.

With a 5% discount rate, 1€ saved today should bring in 130€ in 100 years and 17 000€ in

200 years.

With a 2% discount rate, the initial investment has to be multiplied by only 2.7 in 50 years,

7.3 in 100 years but 52 in 200 years.

32

This exponential crushing of future may seem to lead to negate the coming

generations’ needs. Indeed, choosing high discount rates has often sparked a debate, for

instance about the choice of long-term investments (canals, forests). Today, the debate

reappears under a new perspective when wondering whether high discount rates and

sustainable development are compatible. According to Mrs Brundtland's famous

expression (1987), sustainable development ``fulfils today's needs without compromising

the possibility of future generations to fulfil their'' (Brundtland, 1987).

Most of the first pieces of work on climate policies and especially, the cost benefit

analysis run by Nordhaus according to the DICE model have influenced the first debates

(see for example (Nordhaus, 1993). According to his calculations with a five per cent

discount rate, the total discounted value of the climate change damages from now to

infinity would only be a rather small percentage of the world GDP in 2000. The “skeptical

environmentalist”’s argument of Lomborg, arises in a confident and not much critical

reading of the pieces of work I have just mentioned (Lomborg, 2000).

Is there any opposition between the ecological intuition and the reason or logics of

economics? As the figures just quoted may seem to prove, one might object that economic

calculations of cost benefit analysis leads the living generations to be selfish, which is

ethically inacceptable. To this argument, economists oppose two types of counter-objections.

The first one explains that cost-benefit analysis only indicate economically reachable transfer

rates between generations. In other words, the discount rate reflects the marginal productivity

of capital along the growth path taken by the economy and the transfer equivalence then

stressed are achievable: if the marginal productivity of the capital is such as it is that we can

transfer 1+r tomorrow from 1 today, then we can indeed transfer Tre )(1+ at date T. This point

of view inspired one of the traditional formulations of the Plan Français doctrine after World

War II and constituted the intellectual foundations of the operative calculation of the Plan

discount rate (Bernard, 1972). The second counter-objection, which is widely called for today,

considers not the productive possibilities but the consumption ones. To sum it up, cost-benefit

analysis simply underlines that it is useless to make sacrifices today for our descendants who

will be richer than us. The two explanations are complementary and to tell the truth,

simultaneously obtained in the theoretical model of ``first best''

The second argument directly refers to the normative point of view to which the

objection (the issue of selfishness) implicitly calls upon. Let's carry on the discussion in a

more formal way. How to give value to a consumption unit for the existing generation as

33

opposed to a consumption unit meant for the future generation5? Let's assume that the

generations, indices t, have the same preferences for an aggregated consumption of private

use goods. These preferences are represented by the following utility function )( txU . To

fully define the problem, one has to assume a benevolent planner who would arbitrate

between the generations by referring to an inter-temporal social utility function.

))((=1=

tt

t

xUeU δ−∞

∑

δ is the pure rate of time preference. As Koopmans noticed (Koopmans, 1960), it has to be

strictly positive from the moment that natural axioms of inter-temporal trade-off are used.

We will come back to that later on. Let's have a look at a trajectory that would be socially

optimal for the criteria considered but without questioning the productive mechanics

which allows this trajectory. By definition, along this trajectory, inter-generational

transfers are no longer desirable (or have only second-order effects on the social well-

being). Let's assume that along this trajectory,

tt xgx )(1=1 ++

with g as the consumption growth rate between t and t+1. Between t and t+1, the interest

rate or discount rate that implicitly determines the transfers between the t and t+1

generations and which reflects the normative choices made is:

ρδ gt

tr +

+~)

1(

providing g is small and where )(=U

Ux

′

′′−ρ is the elasticity of the marginal utility (it is

also the coefficient of relative risk aversion, the resistance to inter-temporal substitution).

With an iso-elastic utility function for which

')(1 =)(= ,]][1

1[=)( σρ

σσ

U

UxxxU tt ′

′′−

−−

Most of the discussion about inter-generational equity can be conducted starting from this

formula. It indicates the discount rate that one has to implement on a profit, passing it on

to the next generation who is richer than the present generation as its consumption has

grown following g rate. This discount rate depends on g, δ and ρ coefficient as well.

Let's comment on each of these three parameters of the formula.

5 The economic viewpoint taken here would have to be compared with the philosophical viewpoint (see Jonas (1990), Dupuy (2001)).

34

• There's to the least some uncertainty surrounding g which measures the wealth

growth of the future generation compared with the present generation.

• The rate of pure time preference for the present δ diminishes a generation's

well-being all the more so as the generation is remote and simply because it is remote. A

strictly positive delta does not lead to an identical treatment of each generation. The logic

of utilitarian equity seems to orientate the choice of an additive social well-being function.

But, forgetting about the population issue, granting the same importance to each

generation would meet the same objections as mentioned earlier. To be the most loyal to

the utilitarian vision, δ may then be identified as the probability of the planet survival.

This option echoes to a faint uncertainty but unquestionable which conditions the

emergence of the generation t and solves the logical problems of inter-temporal

aggregation on an infinite scale.

• The elasticity of the marginal utility is at the heart of the matter. The debate

about the choice of a reasonable value refers either to normative arguments or positive

arguments. The normative arguments revolve around the comparison to which extent it is

beneficial to add 1% of the wealth according to which generation this 1% will be

allocated. The positive arguments refer to the choices observed in an uncertain future.

Those choices suggest a coefficient of relative risk aversion between 2 and 6 which, along

the lines of the Bayesian hypothesis of maximisation of expected utility, identifies with the

elasticity of the marginal utility. A different version of this argument refers to the real

interest rate seen on the market. Yet, one can wonder about the limitations of this

“positive” point of view, because the logic of individual inter-temporal choices and the

one of inter-generational altruism cannot be grounded on the same considerations (this

point should be explored further on but it goes beyond the objectives of this essay).

After those general considerations, let's proceed to the numerical propositions. The

Stern Review (2006), advocates for an elasticity close to 1,3 g close to 1 and delta = 0.1.

The discount rate used is approximately 1.4%. Each of these parameters has to be

commented upon.

• A growth rate of around 1% for the century to come is low but it may be

supposed to reflect the fact that uncertainty about growth in the above reasoning should

35

lead to withhold a certain equivalent of growth, the remaining growth rate becoming lower

when the horizon becomes larger (Weitzman, 2001)6.

• The interpretation of the pure rate of time preference as suggested above, i.e

equal to the probability of the planet survival, would lead to chose a very low delta, much

inferior to the value chosen in the Stern Review. But the survival rate idea repetitively

mentioned in the Review or its post-scripts, is far from being universally accepted: many

tend to see in a rate of pure time preference rate of 1 or 2% a pragmatic but circumstantial

way to raise doubts about the fair evaluation of the coming generations' well-being (or

similarly, in the case of personal choices, a way to reflect the uncertain quality of the agent

on the determiners of his future well-being). The argument is neither fully convincing, nor

completely inadmissible. The choice of a 0.1% rate in the Stern Review reflects a certain

compromise between the two views just presented.

The normative arguments are often conflicting. A unitary elasticity implies that an

additional 1% in our generation's income and an additional 1% revenue for a twice richer

generation than ours are equivalent from the point of view of inter-generational equity.

The objection to this inter-temporal elasticity value is that it is too low and that it implies

taking too much into consideration the coming generations' well-being, or similarly,

involving too little redistribution towards for instance the existing generation !. One can

object here that standard cost-benefit analysis often implicitly grant the same importance

to the well-being of all agents (Hoel & Karp, 2001)... In addition, the preference

highlighted by the intra-generational redistribution, as one can assess it the choices

observation is even lower. Paradoxically enough, the choice of a more substantial

redistribution would lead today to more redistribution from rich countries to poor

countries but to a greater selfishness from the present generation toward the coming ones,

thus minimizing the beneficial effects of the climate policy as it is presented in the Stern

Review. However, this tendency would be thwarted by the fact that poor countries will be

more affected by climate change and that, being still relatively poor tomorrow, they will

need an even greater effort from rich countries today as the redistributive choice is

reinforced.

6 In the case where the growth rate is a random variable, of given mean g and of variance v, the preceding formula, relating the interest rate and the growth rate is modified. If u’’’>0 then the certainty rate is smaller. For example for an iso-elastic function with s’=1, (Log) r= d +s’ (E(g)-var(g)).

36

1.2. A Cost benefit analysis qualitatively compatible with the ecological

intuition

I come back on the cost benefit analysis issue, providing first a general outlook on

the general conditions of compatibility between the « economic” and the « ecological »

viewpoints (1.2.1) and then presenting a formal model that incorporates some of these

conditions (1.2.2).

1.2.1. General outlook

The article Calcul Economique et Développement Durable (Guesnerie, 2004) deals

with the cost-benefit analysis of long-term projects which aim at preserving the ecological

integrity of the planet. The suggested approach drops the aggregated good hypothesis

associated with the models starting from Ramsey (1928), in order to underline the

ecological dimension of choices. Therefore, there are two kinds of goods in the model, the

private aggregated good and the environmental good whose quantity is measured by an

index of environmental quality. Just like the private good is the aggregate of all the private

individual goods, the index of environmental quality is an aggregate of environmental

variables (the quality of the climate which relies itself on a series of indicators, the quality

of landscapes, biodiversity, water resources...). Moreover, the private good is provided on

markets while the environmental good is considered as a (non-market) public good. The

emergence of a second good into the model reintroduces the consideration of relative

prices in the analysis. In a model with n goods, the discount rate has a conventional

definition, being generally the real interest rate for a complex “numéraire” reflecting the

consumption structure of the economy. In such a context, the inter-temporal choices

necessarily refer to the whole inter-temporal price system. One can infer it from the series

of discount rates for the “numéraire” whose definition has just been mentioned, coupled

with the prices of goods relative to the numéraire at each period (Malinvaud, 1953)7.

Notice that the Plan Français traditional doctrine underlined the importance of the relative

prices prediction in the economic calculation (Milleron et al., 1978).

Guesnerie's analysis introduces four parameters, arguing that the first three ones

are necessary and often sufficient to reconcile what we can call the ecological intuition

7 The Hotelling (1931) model concerning the extraction of non renewable resources provides a well known example of a world where the relative price effect is spectacular (it cancels the discount rate effect).

37

and the economic reason. These first three parameters rely on the following

considerations:

• The specificity of environmental goods in the long-term. They are different both

from private goods (they cannot be multiplied) and from non-renewable resources (they

are not destroyed by a careful use). Available in limited quantity, in the long-term, their

relative scarcity is likely to constantly increase.

• The imperfect substitutability between standard consumption goods and

environmental goods (Neumayer, 1999). The chosen modelling option leads to consider a

series of situations indexed by the sigma variable between the two polar cases. The logic

of these polar cases can be easily understood. At one extreme of the spectrum, if private

goods and environmental goods are completely substitutable (according to a given ratio of

equivalence), there is no reason to treat differently the environmental good and the private

good in cost-benefit analysis. In the other polar case, the private good, available in

quantity x, and the environmental good, “quantity y”, are pure complements, and and

utility is Minx,y. Thus, beyond a certain level of consumption, the possibility to

increase the well-being of a generation richer than ours has to resort to environmental

good, and beyond a certain level of private wealth exclusively to environmental good.

(This is probably a way to express the point of view of the ``deep ecology'' defenders).

• A low pure rate of time preference. Intuitively, it is clear that if we grant to the

future generations' well being a lower priority which will exponentially decrease, just

because they will live in future, than to ours, (for instance, 7 times less “importance” to the

generation who will live in a century), then the weight of long-term well-being will

disappear and long term analysis will become incompatible with the inter-generational

solidarity widely constitutive of the ecological intuition.

• The last consideration which is not necessary but may turn out to be sufficient,

is related to uncertainty. In a certain way, the uncertainty about the future leads to

undervalue the long-term discount rate. This point has been presented and developed in

the article by Weitzman already mentioned. I briefly sum it here. Consider Rte− the

valorization by generation 0 of 1 given to the generation T (R being the discount rate).

Let's assume that this rate is uncertain and that it can have two values (R or r, R>r) with a

2

1 probability each. Therefore, the valorization by generation 0 of 1 given by generation T

is

38

rtRt ee −− +2

1

2

1

Or ]2

1

2

1[ )( TrRrt ee +−− + ,

Or also . )( TTre ′−

We can see that the certain equivalent to the discount rate r'(T) tends, when T goes to

infinity, to the lowest rate, which is r. As I have already said, the argument is general. It is

similar to the argument about the criterion of cautious choices with a non-expected utility,

while escaping the objection of the minimum arbitrariness. In this case and more

generally, the whole distribution depends on r'(T)8.

1.2.2. The model, cost-benefit analysis from the “reform” viewpoint.

Guesnerie's model (Guesnerie, 2004), taken up9 again in Hoel-Sterner (2006),

Guesnerie et al. (2010) describes a world with two types of goods, an aggregated

consumption good characterized by the volume of its consumption and an aggregated

environmental good whose volume is based on environmental quality. The utility function

has two levels.

First, 1

11

)(=),( −

−−

+ σ

σ

σ

σ

σ

σ

tttt yxyxv describes the equivalences between private

consumption and environmental consumption. This formula implies that if y

x decreases

by 1%, the marginal disposition to pay for the environmental good increases by σ

1 %. The

two precedent polar cases, perfect substitutability and strict complementarity match the

limit values of σ : the infinite and zero.

Then, the function )),(1

1(=),( 1 σ

σ′−

′−tttt yxvyxV gives an iso-elastic cardinal

representation of the generation t's well-being. In a one good model, σ ' is alternatively

interpreted as the elasticity of the marginal utility, the relative risk-aversion or the

resistance to inter-temporal substitution.

8 For example Weitzman (2001) shows that if uncertainty has a gamma distribution r’(T)=A/(1+BT).

9 For discussions on the ecological dimension of choices in models with several goods, see Heal (2008) and its references.

39

It is clear that the model is extremely simple, especially as the intensity of the

ecological need is measured with the σ parameter (in which one should see a kind of one-

dimensional summary statistics), when the social value of the well-being increase obtained

by a generation in comparison to what is granted to other generations relies on the other

parameter of the model: σ ' (as said before, it is likely to fit several different

interpretations).

Finally, the social well-being is measured according to the utilitarian logic with

σδ

σ′−−

∞

∑′−1

0=

),(1

1= tt

t

t

yxeU

where the t index is associated with the t generation and δ is the pure rate of time

preference which tends to 0 when the “ethical point of view tends to prevail”.

In this model, we can run a cost/benefit analysis in a growth situation with g as the

private consumption rate and a stable environmental quality. This is the ``reformist'' point

of view. It is necessary to distinguish the discount rate for the private good which governs

the inter-temporal variation of a given good price and the ecological discount rate which

indicates the variation of the relative price of the environmental good.

The ecological discount rate is used as followed for what we can call the

``canonical'' ecological calculation: the 0 generation evaluate an investment (at 0), leading

to the improvement of the environmental quality for the t generation. The social value of

the improvement is measured simply (``canonical procedure'') as is the 0 generation's

willingness to pay, discounted with the ecological discount rate. This benefit is thus

compared to the investment cost.

Guesnerie's work (2004) focuses on this ecological discount rate and its value on

the long-run, (for a fixed growth rate of consumption g). He makes a distinction between

the cases σ <1 and σ >1, the first one being associated with the idea of an ``ecological

deadlock''. Proposition number 1 (with a delta almost equal to 0) highlights the limits of

ρ (T) the ``canonical'' ecological discount rate with a T horizon: without any ecological

deadlock at all, the limit is near ]1

[σ

σ −′g , whereas under the ecological deadlock

hypothesis, this limit is near 0.

The ecological rates just emphasized are different from standard rates. For

instance, if the probability of an ecological deadlock on the long-term is null, the long-

term discount rate for private goods would mainly be equal to (gσ ') (or its Min is to take

40

uncertainty into account). On the contrary, in case of the ecological deadlock on the long-

term, the long-term discount rate for private good is σ

g.

To underline the most striking result of the analysis, we can put it this way. The

two following propositions are incompatible:

• In the long term, the private consumption optimum grows asymptotically (at a

positive rate g), there is an ecological deadlock and the ecological situation deteriorates.

• The canonical ``ethical'' ecological discount rate (that is to say the ecological

rate obtained from a vanishing pure rate of time preference) is positive.

The incompatibility is maintained if the ecological deadlock, instead of being

certain, is only possible.

The analysis of reform, however, has to face two objections. This first one is that

the analysis is elusive on the problem of transition as the concept of very long term

discount rate is not much operational. The second one is that the growth rate comes from

an endogenous choice which has apparently no reason to stay the same whether the value

of the elasticity of substitution between private and environmental consumption is above

or below 1. For instance, it is not clear whether a positive growth rate of consumption is

compatible with an elasticity of substitution below 1.

1.3. Optimal growth and the underlying cost-benefit analysis.

Guesnerie et al. (2010) embrace an optimisation point of view, in a context which has

the advantage to be extremely simple. They adopt an endogenous growth modelling of the

simple AK variety(where production is linear in capital), where, de facto, the standard interest

rate, r, is given exogenously. The world they describe is then exogenously and completely

defined by four parameters, δ, σ, σ′, r.

They demonstrate that optimisation leads to a growth asymptotically constant

whether there is an ecological deadlock or not. The asymptotic results from the

optimization (from an utilitarian social welfare viewpoint) echo to the results obtained

from the reform approach.

The characteristics of the optimal, growth path can be summed up as follows:

There is an asymptotic growth rate, given by the formulas:

- σ <1, )(=* δσ −rg

- σ >1, σ

δ

′

−rg =*

41

• Correspondingly, the asymptotic ecological discount rate B* is :

- σ <1, δ=*B ,

- σ >1, δσσσσ ′

+′

−1

)1

(1=* rB

With the logic of optimization, the analysis of transition ceases to be arbitrary.

Therefore, it can be shown that:

• Transition and growth

- σ <1 and σσ ′ >1, then g*(t) increases, and B*(T) decreases

- σ >1 and σσ ′ >1, then g*(t) decreases, and B*(T) increases.

One can wonder about the form of precaution (see (Godard, 1997), (Gollier et al.,

2000) for discussions on the concept) that the analysis of our simplified universe would

justify. Let's briefly consider three guidelines.

• Let's introduce an uncertainty on the value of σ with a non null probability of

an ecological deadlock. This uncertainty is lifted at the certain period T'. The optimization

underlines that the asymptotic ecological discount rate, seen from any period previous to

T', is the pure rate of time preference. This can be regarded as a particular form of the

``precautionary principle'' according to which the willingness to pay to avoid an

environmental damage to a remote generation (the willingness to pay for the present

generation is x) is equal the discounted value of x with a rate close to the pure rate of time

preference, even if the probability of an ecological deadlock is low.

• There is another point which refers to the precautionary principle, without

resorting to uncertainty. How much should we be willing to pay today to avoid an

``irreversible damage done to the environment'' (the expression comes from the

precautionary principle of the Barnier law) in the short term? One can write the answer

mx, where x is the (one year) willingness to pay of the present generation to enjoy the

piece of environment under threat. The following tables give the approximate inferior

boundaries for the multiplier m, using the information from the previous analysis

(mingling the asymptotic results with those dealing with transition). The scenarios are

deterministic with σ between 1.2 and 0.8 and σ ' being 1+e (e small), between 1.5 and 3

and delta equating 0.1% and r equating 2%.

Table 1. Willingness to pay for avoiding an irreversible damage to a piece of

environment whose short term “value” is x

1+e 1.5 3

42

1.2 >200x 100x >70x

0.8 ? 400x ?

The analysis is not always conclusive considering the information collected in the

study but it keeps the flavour of the ecological intuition. Let's notice that the results still

support the “ecological intuition” even when the pure rate of time preference is equal to

2% (the previous figures would then be divided by a number between 2 and 5).

• Among the other questions that can be considered in relation to the above results, let's focus on the

following: what is the willingness to pay to modify the environmental quality, equivalent to x% of the

actual GDP, a modification which would occur between year 50 and year 350? The exercise can be

conducted from figure 3 in the quoted text and suggests a multiplier coefficient of 20 in the case of the

central box in the above table. This means that if it was undergone by our generation, a possible non-

market damage of 1% of the actual GDP (an extremely optimistic evaluation of the environmental costs

of the coming climate change), would justify an investment of 10 points of the GDP in order to avoid it.

All this does not take into consideration the direct consumption costs which are not superfluous. This

suggests that the neglect in the Stern Review of the relative price effects, associated

with the public good aspect of the scarce good environment, leads to a possibly very

significant underestimation of the utility of a strong climate policy.

Of course, the studies mentioned here leave a certain number of questions open to

debate. Without pretending to be exhaustive, I will pick some of them:

• How to describe preferences in a more complex way without making it

confusing?

• How to introduce a less rough growth model which still remains an

endogenous growth and focuses on the appropriate characteristics of the phenomenon for

our analysis?

• How to treat in a more adequate way the uncertainty related to the phenomena

that we try to understand?

2. Price policy and quantity policy

In the language of economic theory of environment, the Kyoto Protocol is a

quantity policy. Indeed, it relies on the determination of quotas for the concerned

countries, being here countries who volunteered and form the B annex, with quotas

43

referring to a standard year. These quotas are meant to be exchanged on an international

market which is supposed to have an international price for carbon emerge.

Such a policy is said to be a quantity policy since the success of its implementation

allows to control the total level of the participants' emissions. A pricing policy has an

effect not on quantities but on prices. As it increases the price of a polluting good, the

imposition of a tax allows to control not the quantity of pollution for instance, but its cost

for the polluting agent. Paradoxically enough, a subsidy for cleanup has an effect similar

to the tax effect, as soon as the the marginal costs of the polluting unit are concerned.

More generally, pricing policy and quantity policy constitute the terms of an alternative

which may have equivalent incentive effects, providing the information on costs and

benefits about pollution decrease is perfect. Such a point is highlighted by economic

theory, even if of course different policies, with the same incentives effects, may have

different effects on the distribution of income. (The effects on the financial balance of a

company are not the same whether the pollution it created is imposed or subsidized,

whether the quotas it needs on the emission market are free or not). Note however that the

equivalence, from the point of view of incentive, holds when the environmental policy is

determined on condition of perfect information. We will come back to it.

An alternative to the Kyoto agreement could have been the creation of a carbon tax

harmonized on an international level. To tell the truth, the European Union had raised such

a proposition, France being the initiator during the first negotiations on climate issues. It

had been thus rejected by the United-States. Ironically enough, after having defused the

European Union's strong reluctances about Kyoto trading market, the United-States

rejected the ratification invoking then, among other arguments, the superiority of

harmonized tax solution over the Kyoto solution10. Of course, there are lots of problem

related to the implementation of such a tax whose analysis is beyond the framework of this

study. Let's come back to intellectual debate of prices vs quantity to indicate what the

limits to the traditional conclusions are in a context of climate policy.

2.1. Why Price policies and Quantity policies are not substitute?

Again, the argument echoes back to the old and influential article by Weitzman

(1974). He comes to the conclusion that the relative merits of a pricing policy and a

quantity policy, given the information on costs and benefits is not perfect, depend on the

10 This view had however been defended earlier, see for example Cooper (1998).

44

respective slope of the marginal costs and benefits curves (as a function of the amount of

effort). To sum it up, if the curve representing the marginal cost of cleanup is sloping and

if the expected marginal benefit is ``flat'' that is to say not much variable in the considered

zone, then the pricing policy is superior. On the contrary, if the marginal cost is flat and

the marginal benefit sloping, the quantity policy is superior.

The following graph illustrates the argument in the case of a flat marginal benefit

and where the two possible curves represent the uncertain marginal cost (the marginal cost

is uncertain but rapidly increases in the considered zone). As it is suggested by the graph,

considering uncertainty, the optimal quantity policy sets the quantity at q* (the optimum in

expected value) when the optimal pricing policy sets the price at p* (again, the optimum in

expected value). But, due to uncertainty, both the pricing policy and the quantity policy

will lead to erroneous choices afterwards, with the mistake depending on which hazard is

going to happen. The graph below shows a case in which the social loss associated to a

pricing policy is insignificant and inferior to the one entailed by a quantity policy.

Figure 1. “Flat” marginal benefit and uncertain marginal cost

The argument leads to the opposite conclusion in the case where the marginal

benefit curve is very sloping while the marginal costs are ``flat''. It is very intuitive: if the

marginal benefit curve is almost flat, the pricing policy which sets the price according to

Price

Quantity Q*

p* Marginal Benefit

Marginal Cost

45

the expected marginal benefit, induces small mistakes on the cleanup quantities. In other

words, a relatively constant cost of damage leads to promoting a pricing policy which

relies on the relatively assured value of this damage, when a significant variation of the

marginal benefit in the considered zone (which is a threshold-like effect) lead to promote a

quantity policy which forbids to cross the threshold. When we draw apart from the

borderline cases, according to the intuition derived from the graph, the differences in

slopes but also in the variance that affects the costs and the benefits play a part in the

comparison.

2.2. Application to climate policy

Applied to the climate policy, this analysis suggests that the quantity policy is

dominated by the pricing policy. The reason why is the following: the significant referent

variable here is the stock of greenhouse gas or carbon gas if we stick to the main ones

stored up in the atmosphere. Added throughout a certain number of years (the number of

years results from an international agreement), the emissions represent only a small

percentage of the accumulated stock and thus, the marginal benefit of the emission

limitations cannot be too much variable11.

The argument has been picked up again for a sequential model by Pizer (2001)

Newell & Pizer (2000), Hoel &Karp (2001) using credible numerical values about

uncertainty and the slope of marginal costs. Even if we can object a certain number of

things to the reasoning12, the conclusion is plausible if correctly interpreted. In case of

quantity policy, the uncertainty about the costs of emissions cuts involves a rather high

variance in the endured costs in the case of quantity policy.

We can either rely on the analysis to amend the Kyoto pattern by introducing a

“safety valve” that is, a minimum price for quotas which would frame the market (see

Philibert (2000) for a detailed discussion on this type of arrangement) or advocate for a

radical questioning of the Kyoto architecture to promote the logic of the pricing policy

which would make the harmonization of international carbon taxation come true.

11 Clearly, climatologists warn on the danger of passing some tresholds of atmospheric concentration (550ppmv, and beyond), for which the probability of an increase of global temperature of more than 2 degrees is significant, but this treshold is beyond the concentrations that are plausible within the next decade.

12 For example, the fact that the simplifying assumption of additive separability of costs, biases the comparison towards the price policy (see Guesnerie (2008).

46

Apart from the delicate problem of defining the exact outline of a harmonized

carbon tax, the main objection to this conclusion is that, contrary to what appears to be, a

carbon tax does not make a “price policy”, in the above sense, come true. In the strict

sense of the word, a “price policy” as it was defined in the previous model allows to

control the global carbon price put on the market. The price is constituted by two

elements: the fossil fuel price and the carbon tax. Of course, it is the adequacy of the sum

between these two elements to reflect the social value of the extraction cost, plus the rent

plus the external effect, which is determines the quality of the price control. Yet, as they

underline the complexity of the interactions between carbon tax and the price of fossil

fuel, the simplest models at our disposal cast a serious doubt on the possibility to

implement what would be a pricing policy as described in the previous model.

Let's explain this last point. Starting from what we know in a static pattern, the

price of a rare resource available in limited quantities does not depend on the amount of

the tax that it suffers. The tax only transfers the rare resource owner's rent to its users. In

this simple pattern, the quantity policy described earlier on is simply impossible. The price

of the fossil fuel simply cannot be controlled by taxation. In a sequential pattern in which a

rare resource available in limited quantities would be tapped and taxed, while the

remuneration of the owners would follow the Hotelling law, the situation would be more

complex. The following graph illustrates the dwindling of the rare resource, with the

unrealistic but simplifying hypothesis of an invariable demand in time, when a substitute

fuel is available at a high exogenous cost.

Graph 2. Dwindling of a rare resource with invariable demand curve and

fixed price substitute

47

According to Hotelling law, the rent of the resource increases exponentially, given

that the interest rate has a constant value, and the rent equals the price of a renewable

energy when the resource is depleted. From this diagram, we can intuitively understand

the effect of the resource taxation, following an hypothesis of a perfect foresight and a

competitive behaviour from all the actors (see Guesnerie (2008), Chakrovorty et al.

(2003)).

Here is a brief presentation some dynamic assessments from static teachings:

• Some inter-temporal fiscal arrangements are likely to leave the path taken by

the price when there's no tax unchanged and thus, are likely to transfer the rent from the

producer to the owner.

• Some arrangements allow indeed to postpone the depletion date, even if the

depletion itself is avoidable only after a ``surprise'', followed by a drastic fiscal action.

• In a nutshell, fiscal measures, let's say the carbon tax, may have a dynamic

effect, that is to say postponing the depletion of a fossil fuel but the temporal modulation

of the tax and thus, the credibility of this modulation, is a key element to the success of the

policy. Moreover, the optimistic view about the effects of such a defined price policy is

widely linked to the plausibility of the inter-temporal coordination of expectations that the

time

price

quantity

substitute price

demand

48

rational expectations hypothesis ensures in our models, when “eductive” instability in the

sense of Guesnerie (2005) makes its doubtful.

To conclude on the introductory discussion, we can underline that a commitment,

for instance an international commitment to carbon taxes, even in a certain future, is in no

way a substitute to the commitment on quantities. To be so, it would demand to control the

global price of fossil fuel which is a hard task to do. For the conception of a climate

policy, the debate of price against quantity should be started all over again, taking into

consideration the more general context of a control over the price of fossil fuel.

References

[1] Bernard, A. (1972). Calcul économique et Planification, la Documentation

Française.

[2] Brundtland, G. (1987). Notre avenir à tous, rapport des Nations Unies.

[3] Bradford, D.F. (2001). Improving on Kyoto: A No Cap but Trade Approach to

Greenhouse Gas control Princeton University, published in Guesnerie-Tulkens 2008.

[4] Chakrovorty, U., Magné, B., and Moreaux, M. (2003). Energy resource

substitution and carbon concentration targets with non stationary needs, Leerna 31,

Université de Toulouse.

[5] Cooper, R. (1998). Toward a real global warming treaty, Foreign Affairs, vol.

77 no 2, March-April C.

[6] Dasgupta, P. (2001). Human Well-Being and the Natural Environnment,

Chapter 6, Oxford University Press.

[7] Dupuy, J.P. (2001). Le catastrophisme éclairé, Le Seuil.

[8] Godard, O. (1997) (dir.), Le principe de précaution dans la conduite des

affaires humaines. Paris, Éd. de la Maison des sciences de lÕhomme et INRA-Éditions,

1997.

[9] Gollier, C., Jullien, B., and Treich, N. (2000). Scientific progress and

irreversibility: an economic interpretation of the ÔPrecautionary PrincipleÕ, Journal of

Public Economics, 75: 229-253.

[10] Guesnerie, R. (2003). Les enjeux économiques de l'effet de serre in Kyoto et

lÔéconomie de l'effet de serre, sous la direction de R. Guesnerie, La Documentation

Française, Paris.

49

[11] Guesnerie, R. ( 2004). Calcul Economique et Développement Durable, Revue

Economique, p.363-382.

[12] Guesnerie, R. (2005). Assessing Rational Expectations: 2- ''Eductive''

stability in economics, MIT Press, 453p.

[13] Guesnerie, R. (2008). The design post Kyoto climate schemes: an

introductory analytical assesment, in The design of Climate Policies, Guesnerie, R., and

Tulkens, H. (Eds.), MIT Press.

[14] Guesnerie, R., and Tulkens, H. (Eds.) (2008). The design of Climate Policies,

MIT Press.

[15] Guesnerie, R., Guéant, O., and Lasry, J.M. (2010). Ecological intuition

versus economic “reason”. Mimeo.

[16] Ha-Duong, M., Grubb, M., and Hourcade, J.C. (1997). Influence of socio-

economic inertia and uncertainty on optimal CO2-emissions abatment', Nature, Vol. 390.

[17] Harrod, R. (1948). Towards a dynamic economics, Mac Millan, London.

[18] Heal, G. (1998). Valuing the future : economic theory and sustainability.

Columbia University Press.

[19] Heal, G. (2005). Intertemporal welfare economics and the environment in the

Handbook for Environnemental Economics, volume 3, Chapter 21, 1105-1145.

[20] Heal, G. (2008). What do we know about the economics of climate change?

Mimeo Columbia University.

[21] Hoel, M., and Karp, L.S. (2001). Taxes and Quotas for a Stock Pollutant with

Multiplicative Uncertainty. Journal of Public Economics, Vol. 82, pp. 91-114.

[22] Hoel, M., and Sterner, T. (2006). Discounting and relative prices. Mimeo.

[23] Hotelling, H. (1931). The economics of exhaustible resources. Journal of

Political Economy, 137-175.

[24] Krutilla, J., and Cichetti, C. (1972). Evaluating benefits of environmental

resources with special applications to the Hells Canyon. Natural Resources Journal, 12,1-

29.

[25] Jonas, H. (1990). Le principe responsabilité. Une éthique pour la civilisation

technologique. Paris, Éd. du Cerf.

[26] Koopmans, T. (1960). Stationary utility and impatience. Econometrica, 28,2,

287-290.

[27] Lomborg, B. (2000). The skeptical environmentalist. Cambridge University

Press.

50

[28] Malinvaud, E. (1953). Capital accumulation and efficient allocation of

resources. Econometrica.

[29] Milleron, J.C., Guesnerie, R., and Crémieux, M. (1978). Calcul économique

et décisions publiques. La Documentation Française.

[30] Neumayer, E. (1999). Global warming : discounting is not the issue but

substitutability is. Energy Policy, 33-43.

[31] Newell, R.G., and Pizer, W.A. (2000). Regulating Stock Externalities Under

Uncertainty. Discussion Paper 99-10, Resources for the Future, Washington DC,

February.

[32] Nordhaus, W.D. (2002). After Kyoto: Alternative Mechanisms to Control

Global Warming. Paper prepared for the meetings of the American Economic Association

and the Association of IEA/SLT(2002)28.

[33] Nordhaus, W.D. (2006). The Stern Review on the Economics of Climate

Change. Yale University, 17 Novembre.

[34] Nordhaus, W.D. (1993). Rolling the DICE :an optimal transition path for

controlling greenhouse gases. Resource and Energy Economics, 15, 27-50.

[35] Nordhaus, W.D., and Boyer, J. (2000). Roll the DICE Again : The Economics

of Global Warning. MIT Press.

[40] Ramsey, F. (1928). A mathematical theory of saving. Economic Journal, 38,

543-559.

[37] Philibert, C. (2000). How could emissions trading benefit developing

countries. Energy Policy , volume 28, no 13.

[38] Philibert, C., and Pershing, J. (2001). Des objectifs climatiques pour tous les

pays : les options. Revue de lÔEnergie 524.

[39] Pizer, W.A., (2001). Combining Price and Quantity Control to Mitigate

Global Climate Change. Journal of Public Economics, 85,(3), 409-434.

[40] Ramsey, F. (1928). A mathematical theory of saving. Economic Journal, 38:

543-559.

[41] Stern, N. (Dir.) (2006). The Stern Review Report: the Economics of Climate

Change. London, HM Treasury, 30 Octobre, 603 p.

[42] Sterner, T., and Persson, U.M. (2007). An even Sterner Review - Introducing

Relative Prices into the Discounting Debate. Göteborg University, February.

[43] Weitzman, M. L. (1974). Prices vs. Quantities. Review of Economic Studies,

vol.41, October.

51

[44] Weitzman, M. (2001). Gamma discouting. American Economic Review, 91,

1, p. 260- 271.

52

Discounting the future: the case of climate change

Ivar Ekeland

Abstract: According to the Stern Report on climate change, the course of the next

fifty years is set: present policies will impact only in the very long term, fity to two hundred years from now. There is no market for interest rates, so far into the future, and economists must find other ways to set interest rates in a coherent way. This is paper reviews some of the methods which have been used. We start with the classical Ramsey model of economic growth, which remains a central reference in the current debate, and we study the determinants of the interest rate in that framework. We then adapt the model (and the results) to take into account various concerns, namely (a) the existence of the environment as a distinct, non-produceable good, (b) uncertainty on the parameters or on the model (c) intergenerational equity.

Keywords: climate change, long-term policy making, interest rates, growth models

53

1. Long-term policy-making. 1.1 What is the long term ? These lectures deal with the economic aspects of long-term policy-making. As the

historical notes will show, this problem has been around for many years, in fact since the beginnings of economic theory, and there is a vast literature on this subject. To state it as simply as possible, a decision-maker, either an individual or a collective entity (the government), is to make a decision today (time 0=t ), the consequences of which will kick in only at a (much) later time Tt = , and he/she has to weigh the immediate benefit of that decision against the future costs. Alternatively, the cost occurs today and the benefits at time T , and the question is then how much cost the decision-maker is willing to bear today in order to reap the benefits at time T .

What has changed, however, since these early days, is the time horizon. Up to very recently, what the economist, the engineer or the politician would consider long-term would be in the range 10 to 30 years (note for instance that the US Treasury does not issue bonds with a maturity longer than 30 years). Anything beyond that was considered beyond the horizon - just like accelerating galaxies slip beyond the boundary of the observable universe. This has changed in recent years, where the consequences of our actions beyond that horizon have become part of the agenda. Here are two examples:

• the lifetime of a nuclear plant is 40 to 60 years, after which it will have to be

decommissioned and the site reclaimed, at a considerable cost, which has to be factored in the investment decision

• the Stern Review on Climate Change states that the course for the next 50 years is

set: the inertia of the physical and biological system governing the Earth climate is such that the consequences of any policy we enact today will not be felt before 50 years have elapsed. The question is what happens after that, and the Stern Review depicts alternate scenarios spanning the 50 to 200 years period.

In these lecture notes, we will define the long term as the 50 to 200 years range. 1.2 What are the difficulties of long-term policy-making ? There are two main features which set long-term decisions apart from short and

middle-term ones. The first one is high uncertainty. This comes in two different guises: • the predictable outcomes, that is those for which probabilities can be set, have a

very high dispersion. For instance, the Stern Review states that, with a probability of 95%, under the business as usual scenario, the loss of GDP to the world econony 200 years from now will be in the range 2% to 35%

• but there also non-predictable events, which Stern calls "bifurcations", such as the

cessation or the thermo-haline convector which runs the Gulf Stream. or the melting of the West Antarctic ice sheet. We are in no position to assign probabilities to such event but we know (a) that they can occur,and (b) that their consequences would be catastrophic

54

Let me mention "en passant" that the fact of global warming as such no longer is part

of the uncertainty: it is now certain that it is occuring. At the time of this writing, there is a 50% chance that, for the first time in recorded history, the North Pole will be on open water.

The second difficulty is non-commitment. Whatever policy we enact today will presumably have to be adhered to until the desired consequences are achieved, 50 to 200 years from now, when we (or whoever has decided on these policies) no longer is there to carry them out. This means that we have to rely on future generations (and future governments) to carry them out when we are gone. There is no way we can commit unborn generations and whoever rules the planet one hundred from now to anything: they will do as they please. Whatever policy we design now for the long term has to answer the question: one hundred years from now, when the powers that be are supposed to implement these policies, is there a reasonable chance that they will do it ?

1.3 Is economic theory relevant ? There are, of course, the economic approach to these problems is not the only possible

or relevant one. Clearly, there are ethical considerations. As Keynes famously said, in the long term we are all dead. Do we care what happens after that ? Some people don't: this is the "après moi le déluge" philosophy, which has quite a number of proponents in academic and government circles. Most people do because of ethical considerations towards future generations and the planet itself. Adam Smith himself wrote that "the Earth and the fullness thereof belongs to all generations". There are also political agendas, with immediate gains or losses for decision-makers which preempt any long-term concern.

Even in the presence of ethical or political considerations, traditional cost-benefit analysis has shown itself to be a useful tool, if only to clarify the issues. To attach figures to policies does make a lot of difference. Part of the impact of the Stern Review is due to the fact that it came up with the conclusion that, under a policy of business as usual, climate change would cost 10% of GDP per year, while the cost of prevention stood at less than 1% . This is the kind of argument to which politicians and business leaders pay attention, and the scientific community should make every effort to speak to them in their own language.

Even so, applying cost-benefit analysis to long-term policies is by no means straightforward. In practice, this means discounting future benefits at a certain rate, say r . A natural choice for r would be the market rate of interest, especially for the longest maturity, which today stands at 4.6% (rate of 30 -years US Treasury bond at the time of writing). In the following table, we give the present value of 1,000,000 $ at 100,50, and 200 years, for interest rates of 10% , 4.6% and 1.4% , which is the value that the Stern Review took:

62,000249,000499,0001,4%

12411,140105,5404.6%

0.00<738,51910%

ys200ys100ys50

Clearly, an interest rate of 10% just wipes out the long term. The current market rate

of 4.6% does somewhat better, but the rate of 1,4% really make future events loom large. It is evident that the conclusions of the Stern report heavily depend on this choice of the interest rate, and that they would have been entirely different if, for instance, Stern had chose to discount at market rates. So the question now is: what justification, if any, is there for choosing such a low rate ? This is the question which we will address now.

55

1.4 Some historical notes From the beginning of economics as a separate science, it was apparent that individual

choices between present and future rewards were driven by some kind of time preference: individuals prefer to enjoy goods sooner rather than later. That theme was developed by John Rae (1834), Böhm-Bawerk (1884), Irving Fisher (1930), as a psychological trait of human nature. In 1960, Tjalling Koopmans (1960) showed that impatience can be derived from benign assumptions on preferences. In other words, preferences in economic theory are usually ascribed to immediate consumption ,c resulting in utility functions ( )cu . If one now

considers consumption schedules, ( )tc for 0≥t (possibly discrete), and tries to write down a reasonable set of axioms that preferences should satisfy, one is inevitably led to time preference as a logical consequence.

Since time preference is well established, the next question is how to translate it into a mathematical model. It is a natural idea to discount future utilities at a constant rate 0>ρ : the larger ρ , the more impatient the consumer. If 0=ρ , the consumer is indifferent between immediate and deferred consumption; in other words, he exhibits no impatience at all. The idea of setting up the question of economic growth as an optimisation problem is due to Frank Ramsey (1928). Interestingly, he chose 0=ρ as his preferred option: "we do not discount later enjoyments in comparison with earlier ones, a practice that is ethically indefensible and arises merely from the weakness of the imagination". He did, however, treat the case 0>ρ as well, and that became the standard of the industry, following Samuelson (1937).

1.5 Structure of the paper We will proceed by first giving a detailed exposition of the standard model of

economic growth. This is the topic of the next section; this model has very stringent assumptions (a single good in the economy, and constant discount rate), but as a result explicit formulas can be derived, including a formula for the long-term interest rate:

gr θρ += which serves as the basis for much of the present discussions among economists. This formula is derived and explained in Section 3, and then discussed; suitable modifications are proposed to take into account the known deficiencies of the standard model and some of the current discussions about environmental issues and climate change. Many of these modifications lead us to non-constant interest rates.

Finally, we turn to concerns of intergenerational equity. In the standard model, there is a single infinitely-live individual, who stands as a proxy for people alive today and all future generations, which is clearly very simplistic. Following an idea of Sumaila and Walters (2005), we separate the utility of people alive today and that of future generations, and we aggregate them in a single intertemporal criterion. This leads to a model similar to the standard one, but with non-constant interest rate.

At the end of this investigation, we have built an overwhelming case for considering non-constant interest rates, and this is the topic of the last section. It turns out that handling time-varying interest rates requires a change of paradigm and a new mathematical theory. Indeed, non-constant interest rates lead to time inconsistency, so that the notion of optimality changes with time. The problem then becomes a game between successive generations, which was solved by Ekeland and Lazrak (2005, 2008) and we describe the main results of these papers.

56

2. The standard model of economic growth 2.1 Firms, consumers and growth This model originates with the seminal paper of Ramsey in 1928. It is described in the

opening chapters of most graduate textbooks in macroeconomics: see for instance (Blanchard and Fisher, 1994) and (Romer, 1996) or (Barro and Sala-i-Martin, 1995) for a more detailed treatment. There is a single good in the economy, which can be either consumed (in which case it is denoted by C ) or used to produce more of the same good (in which case it is called capital and denoted by K ) . This good is produced by a large (fixed) number of identical firms in perfect competition, so that they function as a single firm which is a price-taker. The global production function is:

),(= ALKFY where K is the total capital invested in the economy, L is the labour force, and A is the productivity of labour (which will eventually depend on time, ( )tA , to reflect technological progress). It will be assumed that there are constant returns to scale, so that setting

ALYy /= (production per unit of effective labour) and ALKk /= (capital per unit of productive labour), we have:

( )kfy =

where ( ) ( ),1:= kFkf is the reduced production function. It will be assumed to be concave, increasing, and satisfying the Inada conditions:

( ) ( ) 0=00,>0, fkfk ≥

( ) ( ) ∞→→+∞ kkff '' when0,=0

The population consists of identical individuals. Total consumption is C ; up to a constant factor, it is also the consumption of each individual. Using the same scaling for production and for consumption, we find that the relevant variable is ALCc /= , the consumption per unit of effective labour. The consumption per individual is AcLC =/ , and this is the variable which enters the individual's utility function u . In the sequel, we will take the following specification:

( ) 10,>for1

=1

≠−

−

θθθ

θxxu (1)

( ) 1=forln= θxxu (2) so that the utility of each individual alive at time t is:

( ) ( )( ) ( ) ( ) θθ

θ−

−

−

11

1= tc

tAtctAu

We will consider (and compare) different consumption scenarios ( )tC . To do that, we

will treat the population as a single 13, infinite-lived individual (the so-called representative consumer), who consumes ( ) ( ) ( ) ( )tLtCtctA /= at time t . Note that this is not the total consumption of society at time t , but the average consumption of its members; the difference is quite significant, since we will assume exponential growth of the population.

13

A slight variant of the model, which leads to the same results, consists of assuming that the population consists

of N identical dynasties, each of which will be treated as a single, infinite-lived individual.

57

This single, infinite-lived individual, has a pure rate of time preference ρ . This means

that, given the choice between consuming ( )0c at time 0=t and ( )tc at time 0>t , he/she will be indifferent if and only if:

( )( ) ( )( )tcuecu tρ−=0

Let us note right now (and we will expand on this later) that ρ is NOT an interest

rate. The economy is driven exogeneously by technological progress and population

growth, both of which happen at constant rates g and n . We set: ( ) ( ) gteAtA 0=

( ) ( ) nteLtL 0= so that the utility of the representative consumer at time t is:

( ) ( )tgec

A θθθ

θ−−

−

−11

1

1

0

2.2 The planner's problem 2.2.1 Statement We now imagine a benevolent and omniscient planner, who wants to maximize the

intertemporal welfare of the representative consumer. He/she will consider the following problem:

( ) dtRamseyeec ttg

c

ρθθ

θ−−−

∞

∫−11

01

1max (3)

( ) ( ) ( ) 0=0,= kkkgnckfdt

dk+−−

( ) ( ) 00, ≥≥ tctk

The second equation represent the balance equation between savings and consumption. It states that at every moment t , (scaled) production ( )kf is fully allocated between immediate (scaled) consumption c and (scaled) capital investment dtdk/ , the correction term ( )kgn + being there to take into account growth of population and

technological progress. Of course 0k is the initial capital.

Solving this problem leads to the following result, which will be proved in the next subsections. Assuming that ( )θρ −1> g (the pure rate of time preference is high enough) we find that:

There is a single ∞k , called the equilibrium value of capital, which solves the

following equation, known as the golden rule: ( ) ngkf ' ++∞ θρ= (4)

58

The problem has a single solution ( )tk , which has the property:

( ) ∞→→ ∞ tktk when

The corresponding consumption ( )tc along the optimal path also converges:

( ) ( ) ( ) ∞∞∞ +−→ kgnkfctc =

In equilibrium, when ( ) ∞kk =0 and ( ) ∞cc =0 , we have ( ) ( ) ( )tgneCtC +0= ,

( ) ( ) gneKtK +0= and ( ) ( ) ( ) gteAtLtC 0=/ , so that total consumption and total production are growing at the rate gn + , while consumption per head is growing at the rate g .

2.2.2 Existence and uniqueness The Ramsey problem appears as a control problem, where the control is ( )tc and the

state is ( )tk . Substituting ( ) ( )dt

dkkgnkfc −+−= in the integrand, we reduce it to a problem

in the calculus of variations, where the only unknown is k :

( ) ( ) dtedt

dkkgnkf t

k

β

θ

θ−

−∞

∈

−+−

−∫1

0 1

1max

A

(5)

( ) ( ) 0,=0 0 ≥tkkk (6)

( ) ( ) 0≥−+−dt

dkkgnkf (7)

We have set ( ) 0>1:= θρβ −− g , and A is the set of admissible functions, will be

discussed later. For the sake of convenience, we also set:

( ) ( ) ( ) dtedt

dkkgnkfkI tβ

θ

θ−

−∞

−+−

−∫1

0 1

1=

Note that, since u and f are strictly concave, strictly increasing, and positive, I is a

strictly concave function of k with values in ∞+∪R . As a consequence, if the maximum is

attained on A, and A is convex, it is attained at a single point (which will usually depend on

the set A).

Does the optimal solution exist ? Of course, the answer will depend on the definition of admissibility, that is, on the choice of A. An appropriate choice for A would incorporate

both regularity conditions (how smooth is ( )tk ) and growth conditions (how does ( )tk behave

when )∞→t . One would then consider a minimising sequence ( ),tki show that a

subsequence converges to some ( )tk , and prove that ( )tk is the minimizer. This is called the direct method in the calculus of variations. Unfortunately, I do not know of this method being applied successfully in the case at hand. The main difficulty is that the integrand ( )cu has slow growth when ∞→c (it grows like a power α , with 1<a ); as a consequence, there is nothing to prevent ( )tk from being discontinuous at some point 0t , as long as the jump is

59

downwards (in other words, the consumer can consume instantaneously a non-zero quantity of good.

In the sequel, we shall prove existence in the class A = C2 by another route (the royal

road of Caratheodory) 2.2.3 A necessary condition: the Euler-Lagrange equation

Suppose there is a solution ( )tk for A = C2. The classical Euler-Lagrange equation

then holds for ( )tk (see any texbook on the calculus of variations), yielding a second-order equation:

( ) ( ) ( ) ( ) ( ) ( )( ) t't' egnkfdt

dkkgnkfue

dt

dkkgnkfu

dt

d ββ −− +−

−+−

−+−− =

(with ( ) ( ) θθ −−− 111= ccu ), which is valid as long as we do not hit the boundaries, that is, for

( ) 0>tk and ( ) 0>tc . We simplify this expression by setting ( ) ( )dt

dkkgnkfc −+−:= , so that

it becomes:

( )( )

( ) 0=ngkfcu

cu

dt

dc ''

''

−−−+ θρ (8)

We can reduce this second-order equation to a system of first-order equations:

( ) ( ) ckgnkfdt

dk−+−= (9)

( )( )ckfngdt

dc '−++− θρθ

1= (10)

All the solutions of this system can be represented on a two-dimensional phase

diagram. The vertical line ( )kfng '=++θρ and the curve ( ) ( ) ckgnkf =+− divide the

positive orthant 2+R in four regions, which have a common boundary point ( )∞∞ ck , , where:

( ) ngkf ' ++∞ θρ=

( ) ( ) ∞∞∞ +− kgnkfc = We can find out the sign of dtdk/ and dtdc/ in each region, and draw the phase diagram. It follows that ( )∞∞ ck , is the unique fixed point, and that it is unstable. There is a unique

solution ( ) ( )( )tctk , , which converges to that fixed point: ( ) ( ) ∞→→→ ∞∞ tctcktk whenand

all the others go one of the boundaries, 0=c or 0=k . Of course, we strongly suspect that ( ) ( )( )tctk , is the optimal solution to the problem

when 2C=A . We can try to compare it directly with the other solutions of the Euler-Lagrange equation (8), but this is quite ponderous (one has to investigate separately all the four regions in the phase space) and will not generalize to more general situations (for instance, with several goods). In the literature, one points points out that the Euler equation should be supplemented by two boundary conditions, one of which is known, namely

( ) 00 = ktk , while the other should describe the behaviour of ( )tk when ∞→t . This is the

celebrated "transversality condition at infinity", several versions of which have been given,

60

either in discrete (see (Ekeland and Scheinkman, 1988)) or in continuous (see (Aubin and Clarke, 1979), (Michel, 1982)) time. Unfortunately, none of them applies to the present situation, where the candidate solution is the only one which does not go to the boundary.

So we will prove that ( ) ( )( )tctk , is the optimal solution by a very different method, using the Hamilton-Jacobi equation, which we now introduce.

2.2.4 Another necessary condition: the Hamilton-Jacobi-Bellman equation If there is an optimal solution for every initial point 0k , depending smoothly on 0k ,

the function: ( ) ( ) ( ) 0

20 =0,|max:= kkCkkIkV ∈

satisfies the following relation at every point where it is differentiable:

( ) ( ) ( ) ( ) ( )( ) ( ) 0=1

and0 1/1 kVkgnkfkVkVkV ''' ρθ

θ θ−+−+

−≥

− (HJB)

and the optimal consumption is given by a feedback strategy:

( ) ( )( ) θ1/= −tkVtc ' (11)

An informal proof is as follows. Define:

( )( )

( )( ) ( ) 000 =0|max:= kkdtetcukV t

c

ρ−∞

⋅∫

(with ( ) ( ) θθ −−− 111= ccu ). Using optimality, we have:

( )( )

( ) ( )( ) ( ) ( ) ( )( ) 000,

0 =|max= kgnxkfkkdtetcuxukV t

cx+−−++ −

∞

⋅∫ εεε ρ

ε

( )

( ) ( )( ) ( ) ( ) ( )( ) 0000,=|max= kgnxkfkkdttcuexu

cx+−−++ ∫

∞−

⋅

εεε ρε

( ) ( ) ( )( )( ) 000max= kgnxkfkVexux

+−−++ − εε ρε

( ) ( ) ( ) ( ) ( ) ( )( )( ) 00001max= kgnxkfkVkVxu '

x+−−+−+ ρεε

( ) ( ) ( ) ( ) ( ) ( ) ( )( ) 000000 max= kgnkfkVxVkxVxukV ''

x+−+−−+ ρε

We end with: ( ) ( ) ( ) ( ) ( ) ( )( ) 00000max=0 kgnkfkVxVkxVxu ''

x+−+−− ρ (12)

The right-hand side splits in two terms, yielding the desired equation: ( )( ) ( ) ( ) ( ) ( )( )00000

~=0 kgnkfkVkVkVu '' +−+− ρ

where: ( ) ( ) xyxuyu

x−max:=~

−−

−− θθ

θxyxyx

x=|

1

1max= 1

θ

θθ

θ

θ

θ

−−

−

−

−

−

11

1=1

1

1= yx

and the maximum is achieved for ( ) θ−xxuy ' == , yielding θ1/= −yx , and hence formula ((58)).

61

2.2.5 A sufficient condition: the royal road of Caratheodory Theorem 1 Suppose (HJB) has a 2C solution ( )kV such that, for any 0>0k , the

solution ( )tk of the Cauchy problem:

( ) ( ) ( )( ) ( ) 0=0,= kkkVikgnkfdt

dk '−+− (13)

converges to ∞k and ( )( ) 0→− tkVe tβ when ∞→t . Then, for any starting point 0>0k , the

path given by ((60)) is optimal among all 2C paths ( )tk such that

( )( )( ) 0suplim ≥−

∞→

TkVe t

T

β

In particular, it is optimal among all interior paths (that is, paths along which ( )tk is bounded away from 0 and ∞ ).

Proof. Consider any path ( ) ( )tktc , starting from 0k . Because of equation ((59)), we

have:

( )( ) ( ) ( )( ) ( )( ) ( )( ) ( ) ( ) ( )( )( )[ ] 0>forevery00

TdttkVtkgntkftkVtkVtctcue ''tT≤−+−+−−

∫ ββ

The left-hand side can be rewritten as follows:

( )( ) ( )( ) ( )( ) 000

≤

−+ −−

∫∫ dttkVtkVdt

dkedttcue 'tTtT

βββ

( )( ) ( )( ) 0|00≤+ −−

∫TttT

tkVedttcue ββ

( )( ) ( )( )( ) ( )( )0suplim0

kVTkVedttcue t

T

tT≤+ −

∞→

−

∫ββ

Letting ∞→T , we find:

( )( ) ( )00kVdttcue t ≤−

∞

∫β

. On the other hand, setting ( ) ( )( )( )tkVitc '= , we get the path ( )tk and the equality is achieved because of equation (HJB). Hence the result.

Note that the 2C regularity of ( )kV everywhere (including at k ) has played a crucial role in the proof

Theorem 2 There is a 2C solution of the HJB equation such that all solutions of ((60)

) converge to the point ∞k defined by:

( ) ngkf ' ++∞ θρ=

This will be proved much later in the lectures. As a consequence, we have: Corollary 3 The Ramsey problem has a unique solution ( )tk in the class 2C , with

( ) ∞→ ktk when ∞→t .

62

2.2.6 Local analysis Linearizing equations ((56)) ((57)) near the stationary point, with ∞− kky = and

∞− ccx = , we get:

( ) ( )( ) xyxygnkfdt

dy ' −−+−∞ ρ==

( ) yc

kfdt

dx ''

θ∞

∞=

where we have taken into account the fact that 0=dt

dc at the stationary point. The

corresponding matrix is:

( )

−

∞∞ 0

1

kfc ''

θ

ρ

which has two real roots, one positive and one negative. Along the "optimal" trajectory, we have:

( )( )( )

−

−

+−

−

−

∞

∞

∞∞

∞

∞

cc

kktkfc

cc

kk ''

0

04

2

1exp126 2

θρρ

2.3 The equilibrium problem What if there is no planner, or if the planner has no means of implementing his/her

policy ? In that case, we will be looking for an equilibrium interest rate, that is, an interest rate ( )tr for which markets clear. By definition, if the spot rate is ( )tr , the price today of one unit

of consumption available at time t is ( )( )tR−exp , where:

( ) ( )dssrtRt

∫0=

Note that, if the interest rate is constant, ( ) rtr = , we find the usual formula, ( ) ( )rttR −exp= , but, as we shall see, there is no particular reason that this should be the case in equilibrium.

Assume the yield curve is ( )tr , which is common knowledge and let us write the market-clearing conditions. There is only one because the representative consumer and the representative firm enter forward contracts at time 0=t . On the supply side, the interest rate (spot rate on the money market) should be equal to the marginal return on investment (between t and dtt + ):

( ) ( )( ) gntkftr ' −−=

Since the firm makes no profit, its revenue is shared between labor and capital, so that the wage must be:

( ) ( ) ( ) gnkkfkftw ' ++−=

Recall that the representative citizen consumes ( ) ( ) ( ) ( )tceAtctA gt0= at time t . He/she maximizes intertemporal utility:

63

( ) dtetc t

c

βθ

θ−−∞

−∫1

0 1

1max

(with ( )gθρβ −− 1= , as above) subject to the budget constraint (intertemporal borrowing and lending is allowed):

( ) ( ) ( ) ( ) ( ) ( ) ( ) dtetwekdtetce tngtRtngtR +−∞

+−∞

∫∫ +≤00

0

Introducing a Lagrange mutiplier λ , this problem becomes (with ( ) ( ) θθ −−− 111= ccu ):

( )( ) ( ) ( ) ( )[ ] ( )[ ]dtetctweetcu tgntRt

c

+−−∞

−+∫ λβ

0max (14)

where ( )tR and ( )tw are known. Note for future reference that the exponent ( )sR is non-constant. The optimal solution to the consumer's problem, as seen from time 0=t , is:

( )( ) ( ) ( )tgntRt' eetcu ++−− λβ =

Let us transform this equation a little bit: ( )( ) ( ) ( )tgntRttcu' ++−− λβ ln=ln

( )kfgndt

dc

c'−+−− =β

θ (15)

This is just the Euler-Lagrange equation, which is satisfied by the solution to the

planner's problem. Hence: Proposition 4 The solution to the planner's problem is also a solution to the

equilibrium problem. They are both efficient (Pareto optimal). There is another way to retrieve the Euler-Lagrange equation. Consider the marginal

return on consumption. In equilibrium, it should be equal to the interest rate (and hence to the marginal return on investment). More precisely, consider the representative consumer at time

0=t , and his/her intertemporal consumption path ( ) ( )tcetc gt=~ ; let us ask ourselves how

much consumption ( )0~c∆ he/she would be willing to forgo at time 0 in other to increase

his/her consumption by ( )tc~∆ at time t . The balance equation can be written (with

( ) ( ) θθ −−− 111= ccu and ( )gθρβ −− 1= )

( )( ) ( ) ( )( ) ( ) 0=0~0~~~ ccutctcue ''t ∆−∆−ρ

( )( )

( )( )( )( )

( )dssretcu

cu

c

tc tt'

'

∫∆

∆0

exp=~0~

=0~

~β

with

( ) ( )( ) ( )tdt

cd

ctcu

dt

dtr '

~

~=~ln=θ

ββ +−

Writing ( ) ( )( ) ( )gntkftr ' +−= , and plugging in ( ) ( )tcetc gt=~ , so that:

dt

dc

cg

dt

cd

c

1=

~

~1

+

we find the Euler-Lagrange equation. It expresses that, in equilibrium, the interest rate in the economy is equal to the marginal return on investement, and also to the marginal return on consumption.

64

2.4 Bibliography As mentioned in the introduction, the standard model can be found in the textbooks,

for instance (Blanchard and Fisher, 1994), (Romer, 1996) or (Barro and Sala-i-Martin, 1995), although not in a way that would fully satisfy a mathematician. The problem of finding the right transversality condition at infinity is still open: there are many versions around (see(Aubin and Clarke, 1979), (Ekeland and Scheinkman, 1986), (Michel, 1982), (Kamihigashi, 2001)), but none which applies to the standard model (where the optimal solution is isolated, all the other ones being either unbounded or hitting the boundary). For the Hamilton-Jacobi-Bellman equation, and an introduction to the royal road of Caratheodory, see (Clarke, 1983) or (Bressan and Piccoli, 2007).

3 Determinants of the interest rate. We shall use the standard model as a benchmark, and introduce successive

modifications. 3.1 The classical theory In the preceding section, we have proved that in the framework of the standard model,

where: • the rate of growth of average consumption is constant and equal to g ,

• the utility function of the representative consumer is ( ) ( ) θθ −−− 111= ccu with

0>θ (CRRA: constant relative risk aversion) the equilibrium interest rate in the economy is given by: ( ) ( )( ) gntkftr ' −−= (16)

( )( ) dt

dc

tcg

11= +−− θρ (17)

Here ( ) ( )( )tctk , is the optimal (from the point of view of the planner) or equilibrium

(from the point of view of the representative consumer) scenario for the economy, starting from ( ) 0=0 kk . The equality in (??) and (??) expresses that in equilibrium, the marginal return

on consumption equals the marginal return on investment. Note an important consequence of these formulas: the spot rate ( )tr at time t is not

constant (even though all the other parameters in the economy, including the psychological discount rate, is constant). More precisely, consider the yield curve at time 0>t . This is the map ( )TrT t→ , defined for tT > , with:

( ) ( )dssrT

TrT

tt ∫ln1

:=

so that ( )Trtexp is the price at time t of one unit of numéraire delivered at time tT > . We

find that: • the yield curve is not flat

65

• it changes with time • as ∞→t it converges to a flat yield curve, ( ) ∞→ rTrt

• for every t , the long-term rate at time t is ( ) ∞∞→ rTrtT =lim

Here ∞r is the spot rate at the stationary point ( )∞∞ ck , . At this point, formulas (??)

and (??) become: ( ) gnkfr ' −−∞∞ = (18) gθρ += (19)

Note the remarkable fact that it no longer depends on t . In other words, in the

standard model, the long-term interest rate is constant along the optimal path and equal to the spot rate at the stationary point.

Formulas (??) and ((79)) generalize to a much broader class of models than the standard one, provided there is one consumption good and one representative consumer. Consider an infinite-lived individual, with utility function u , pure rate of time preference

0>ρ , and who is facing a schedule of consumption ( )tc , leading to an overall utility of:

( )( )dttcue tρ−∞

∫0

Let us ask ourselves how much immediate consumption ( )0c∆ he/she would be

willing to forgo in order to increase its consumption by ( )tc∆ at some later time 0>t . Assuming these are small quantities, we can work on the margin, and we get the relation:

( )( )

( )( )( )( )

t

'

'

etcu

cu

c

tc ρ0=

0∆

∆

so that the marginal return on consumption is:

( )( )

( )dssrc

tc t

∫∆

∆0

exp=0

In equilibrium, if we use the consumption good as numéraire, this should be equal to

the spot interest rate:

( ) ( )( ) ( )( )( )( )

( )tdt

dc

tcu

tcutcu

dt

dtr

'

''' −− ρρ =ln=

This is most conveniently rewritten as follows:

( ) ( ) ( )( )( )( ) ( )

( )

−+ t

dt

dc

tctcu

tcutctr

'

'' 1= ρ

( )( ) ( )tGtcηρ += (20) where:

• ( ) ( ) ( )cuccuc ''' /:= −η is a positive parameter (because u is concave), usually

called the relative risk aversion; in this context, it would be more relevant to call it the relative

66

satiation. It usually depends on the level of consumption c . In the special case of power utilities, ( ) ( )θθ −− 1/= 1ccu , it is constant and equal to θ

• ( ) ( ) ( )tcdtdctG //:= is the rate of growth of the economy Formula ((80)) is the benchmark for determining the interest rate, and is generally

accepted in the economic literature. For instance, the Stern report takes 0.1%=ρ , 1=θ and 1.3%=g , yielding 1.4%=r . Most of its critics claim that it is too low, and take 2%=ρ , 2=θ and 2%=g as more reasonable numbers, yielding 6%=r . We will discuss these

claims, and bring more economic arguments to bear, in the sequel. Meanwhile, let us make some observations:

• as soon as there is growth in the economy ( ( ) 0>tG ), we have ρ>r . For

instance, as Ramsey found out, we can have positive interest rate 0>r even if the pure rate of time preference is zero, 0=ρ .

• the interest rate rises with the growth rate g . For instance, setting 2%=ρ and

2=θ , we get 6%=r if 2%=g and 10%=r if 4%=g . Why is that so ? Well, ask yourself the following question. Historically, growth has been around 2% for the past two hundred years. Now, imagine how your own ancestors were living 200 years ago - probably in conditions which you would consider of extreme need and poverty. Would you want such miserable people to have set something aside for you ? Probably not - quite the opposite, if you were able to do something for them, you would do it. Well, it growth continues at the same rate, this is the way that our descendants will look upon us; they will be richer than we can imagine. So why should we make sacrifices for such people ? Hence the high interest rate that we are in fact charging them.

• on the other tack, the interest rate falls with the growth rate. For instance, setting

2%=ρ and 2=θ again, we get 2%= −r if 2%= −g , that is, if the economy contracts at the rate of 2% a year. So negative interest rates are not unthinkable - they might actually be needed in periods of negative growth. Think for instance of an economy where the only good is the environment, which cannot be produced, and actually has to decrease as the population growth - in such an economy, the interest rate would have to be negative. This leads us to the idea that one would actually have to use different rates for environmental goods and for consumption (manufactured) goods. The proper setting for exploring this idea is a two-goods model, and this is what we will be doing next.

3.2 Modifications 1: The environment as a separate good This section develops, in a continuous-time framework, the ideas of Guesnerie (2004),

and Guesnerie et al. (2008). We complement the standard model by adding a environment good E , along with the consumption good C . The two goods have different characteristics:

• E is a public good, and cannot be produced: it should be understood as the global

quality of the environment.

67

• C is a private good, and can be produced as in the standard model: it should be understood as an aggregate of all consumption goods.

The consumption good will be used as numéraire. For the time being, we assume that

the environment good is available in a fixed quantity E (so that the quantity will not be decreased as the economy grows).

Along the lines of Guesnerie, we choose the utility function of the representative consumer to be:

( ) ( ) θ

θ−

−

1,1

1=, ECvECu

with

( ) ( ) ααα 1/=, ECECv +

The parameter 1≤α denotes the extent to which the environment good E and the

consumption good C are substitutes. If a simultaneous and marginal changes CCC ∆−→ and EEE ∆+→ is to leave the total utility invariant, then we must have:

0=EE

vC

C

v∆

∂

∂+∆

∂

∂−

so that:

α−

∂

∂

∂

∂

∆

∆1

=/=C

E

E

v

C

v

C

E

The right-hand side can be rewritten as follows:

C

C

C

E

E

E ∆

∆−α

=

In other words, to achieve an increase of 1% in the environmental good, the

representative individual is willing to give up %α

C

E of the consumer good.

• if 1<0 ≤α , the willingness to pay for the environmental good decreases as CE/

decreases, that is, as it becomes relatively scarcer. This is the case when the environmental good and the consumption good are substitutes.

• if 0<α , the willingness to pay for the environmental good increases as it

becomes relatively scarcer. This is the case when the two goods are complements. As in the standard model, we assume that there is a production function

( )ALKFY ,= , which is positively homogeneous of degree one, and where the labour force

( ) ntLtL exp= 0 and the technological progress ( ) gtAtA exp= 0 are exogeneously given.

Introducing the reduced consumption ( ) ( ) ( ) ( ) 11= −− tLtAtCtc , as in the preceding section, so

that the average consumption at time t is ( ) ( )tctA , we find that the utility of the representative consumer at time t is given by:

68

( )( ) ( )

( ) αθ

ααα

θ

θ

/1

01 11

1=

−

−

+

−tgetc

E

AEtcu

The representative consumer's optimisation problem then becomes:

( )

( )

dteetcE

AE ttg ρ

αθ

ααα

θ

θ−

−

−∞

+

−∫

/1

01

01

1

1max

( ) ( ) ckgnkfdt

dk−+−=

3.2.1 The case 0<α In that case, we find that, for large t , the utility function can be approximated as

follows:

( )

( )

( ) tgtg etcE

Aetc

E

A ααα

αθ

ααα

α

θ

−+

+

−

0

/1

0 111 ;

The constants play no role in intertemporal optimisation, and we are left with the

criterion:

( ) ( ) dtetc tga αρ −∞

∫0max

This is precisely the standard problem again. We find that ( ) ∞→ ctc and ( ) ∞→ ktk ,

where: ( ) ( ) ( ) ( ) ∞∞∞∞ +−+−+ kgnkfcngkf ' =,1= αρ

In the absence of environmental concerns, the stationary level of capital would be ck∞ ,

given by: ( ) ( ) ngngkf c' +−+++∞ αρρ 1<=

so that ∞∞ kk c > . In other words, the presence of a non-substituable environmental good lowers

economic growth. More precisely, the economy grows as ( ) ( )tngktngk c ++ ∞∞ exp<exp , and, of course, the amount of consumption is reduced accordingly. This is known in the literature as "ecological stunting".

The interest rate at the stationary point then is: ( ) ggr +−+∞ ραρ >1=

Note that the parameter θ does not appear in thes formulas - the risk aversion of the

representative consumer does not come into play (at least not at the stationary state) ! The only relevant parameter is the substitution rate between the public and the private good. As technological progress and population growth drive up the production of consumer goods, the environment becomes comparatively more valuable, and long--term interest rates are determined only by ρ , the pure rate of time preference, g , the technological growth rate, and α - the larger α , the less an increase in consumption can compensate for a decrease in environment quality, and the higher the interest rate.

69

We can now ask ourselves whether an investment that will result in a one-time increase in consumption ( )0c∆ today and result in a permanent decrease E∆ in the quality of the environment is worth undertaking. The answer will be yes if and only if: that it is equal to:

( )( ) ( ) ( )( )( )

EdteetcAEE

ccu ttg' ∆

+

−∂

∂∆ −

−∞

∫ρ

αθαααα

θ

/1

00 1

1>00

( )

( )( )( ) ( )( )

( )

dteetcAEEEcEc

c ttg ρααθ

αααααααα

α−

−−∞−

−+∆∆

+∫

/1

00

11/1 >0

0

0

On the stationary path, ( ) ∞ctc = , we get:

( ) ( )( )

dteecAEc

EEc

E

c ttg ρααθ

ααα

α

αααα −

−−

∞∞

∞

∞

−−

∞ ++∆

∆∫

/1

00

11/1

>

3.2.2 The case 1<0 ≤α As ∞→t , we find that:

( )( ) ( ) ( ) ( )gtecAtcu θθθ

θ−−−

−111

01

1;

and we are back into the standard model.This time the environmental good simply disappears from the global picture.

3.3 Modifications 2: Uncertainty on the growth rates 3.3.1 A classical argument Let us start from Ramsey's formula: gr θρ += (21)

Assume now that we believe in the model, but are uncertain about the growth rate :g

( )2,126 σgg N

so that average consumption ( ) ( ) gtectc 0= is lognormal. Assume moreover that we are utility maximizers, and handle uncertainties à la von

Neumann-Morgenstern. We ask, as always, how much consumption ( )0c∆ we are willing to

forgo today to increase by ( )tc∆ our consumption at time t ( )( ) ( ) ( )( ) ( )[ ] t'' eTcTcuEccu ρ−∆∆ =00

with ( )θ

θ

−

−

1=

1ccu so that ( ) .= θ−ccu' This gives:

( )( ) ( )( )

( )( )[ ]tcuEcu

e

tc

c ''

t

0=

0 ρ−

∆

∆

The computation gives:

70

( )

( )[ ] rttgt

eecEc

e −−−

−

−

=00

θθ

θ

ρ

Using the well-known properties of the lognormal distribution, we find that:

[ ]

+−− 222

2

1exp= ttgeE tg σθθθ

and hence:

tgr 22

2

1= σθθρ −+ (22)

Note that we are back with non-constant interest rates: in fact, the interest rate goes to

∞− when t goes to ∞ ! It may be for this reason that the economic literature takes a different route, and simply averages the Ramsey formula ((23)), defining r as the mathematical expectation of the right-hand side. We then get the formula:

22

2

1= σθθρ −+ gr (23)

which, although simpler (we are back with constant interest rates) and popular (see Erreur ! Source du renvoi introuvable. or makes little sense to me. If we take the usual values, 2=== gθρ , we get 226%= σ−r , and since σ (the volatility of the growth rate) is of the order of a few percentage points, this correction will not be enough to reach the Stern value of 1.4% . On the other hand, formula ((22)) will drive the interest rate down to very low values, and eventually to negative ones: in the very long term, the uncertainty becomes so large that it forbids any risk-taking.

One thing is for sure: uncertainty lowers the interest rate. This corresponds to the standard fact that individuals are risk-averse. Note that this runs counter to an argument that politicians and companies have been making for many years, namely that we should do nothing about climate change, because it is not certain and it may turn out to be all right after all. From what we know, people are risk-averse for themselves, at least when the stakes (magnitude of potential losses) is large, meaning that the downside is more important to them than the upside, and it is difficult to understand why society should behave differently.

3.3.2 Pooling opinions of experts In 2001, Weitzman made the following, very general, observation (Weitzman, 2001).

Suppose you consult two experts, whom you equally trust, about which interest rate to choose, and that they come up with two different opinions, namely 1r and 2r , with 21 < rr . What value should you take ? As you trust them equally, it seems reasonable to pick the mean value, namely ( )/2= 21 rrr + . As Weitzman points out, this is wrong: what these experts are really

saying is that one dollar today is worth respectively tr

e 1− and

tre 2−

at time 0>t . So if a mean is to be taken, it should be the mean of those values, leading to a interest rate r~ given by:

( )

+

−− trtree

ttr 21

2

1

2

1ln

1=~

Note that this interest rate is not constant. It is approximately equal to ( )/2= 21 rrr +

for the short term, but for the long term is equal to the lowest rate 1r . This is the Weitzman lesson: for the long term, the lowest rate should prevail

71

Weitzman put his idea into practice. He pooled 1,800=I economists and asked them for an assessment of interest rates to be applied for investment projects. Economist i

answered with a constant rate ir . leading to a discount rate ( ) tir

i etR−

= . He found that the ir

were distributed according to a Gamma distribution with parameters ( )βα , :

( )( )

rerrf βαα

α

β −−

Γ1126

Averaging the discount rates, he then derives the aggregate interest rate:

( ) ( ) ( )( )

2/220/1

1==:=

σµ

α

µσβ

β

ttdrrArftR

+

+∫∞

In terms of the mean µ and the variance 2σ of the Gamma distribution ( )βα , . The

corresponding interest rate then is:

( )µσ

µ

βα

α

/1==

1=

2tdt

dR

Rtr

++−

Note that very long-term interest rates are 0 . The question is, how far out is the very

long term ? Within the time horizon of the Stern review, from 50 to 200 years, Weitzman finds an interest rate of 1.75% , very much in line with the value 1.4% chosen by Stern himself.

3.4 Modifications 3: Uncertainty on the model. Up to now, the modelling does not capture one of the main features of very-long term

decisions, namely the possibility of major catastrophes with unknown probabilities. The fact that these probabilities are unknown is an added ingredient to risk, which is not captured by simply assigning a priori probabilities, as in classical economic theory. Indeed, a classicial experiment by Ellsberg (1961) indicates that people have a specific aversion to ambiguity, that is to facing unknown probabilities. This is not captured by the von Neumann-Morgenstern approach to decision under uncertaintly, and the paradigm has to be changed. There is at present an active and promising literature on decision making under Ellsberg ambiguity (see (Klibanoff et al., 2005), (Gierlinger and Gollier, 2008) for instance).

Weitzman (2008) has pointed out another problem: whatever probability distribution our model works with, this will not be the one we work with. Indeed, we do not observe the distribution, all we can do is to infer it from a finite (and, in the case of climate change, pitifully small) amount of data. This means that, even if our model specifies Gaussian or Poisson distributions, which is usually the case, and which are nice because they have "thin tails" (large deviations have small probabilities) the ones we will end up working with may well have "fat tails", meaning that all long-term calculations break down.

To take a specific example, go back to the formula:

22

2

1= σθθρ −+ gr

which is based on the modelling assumption that g~N(ḡ, σ2). Weitzman'point is that, even if

we agree with that specification, we know neither g nor σ . We will have to estimate them, and for this we need not only the data but an a priori distribution.

72

A standard way (Jeffreys prior) to choose such a distribution is to suppose that σln is Gaussian. If there are N experimental values available, we are led to a classical problem in statistics (find the variance of a Gaussian variable given N experimental values), the answer to which is a Student distribution with N degrees of freedom. It is well known that this distribution has fat tails. More precisely, if we have observed we find that:

( )( )

( )( ) ( ) ( )[ ] +∞=,...,|0

11 N

''

tctctcuEcu

with the specification ( ) ( )θθ −− 1/= 1ccu . In other words, given a finite number of observations, society should be willing to give up an unlimited amount of consumption today to gain any certain amount of consumption in the future. This corresponds to an interest rate of −∞=r

3.5 Modifications 4: Equity and redistribution 3.5.1 The problems Consider again the standard model: there is an infinite-lived representative consumer,

who strives to maximize

( )( ) dtetCu tρ−∞

∫0max (Ramsey Growth Model)

Problem 1: there is no such thing as a representative consumer People are different - in their tastes (utility function u ), in their expectations

(probability p ). More importantly, some are rich, but most are poor. The first question is dealt with by aggregation theory (see the lectures by Jouini in this summer school). The second question, to my knowledge, has not attracted academic attention - except from Ramsey himself ! He devotes the last section of his seminal paper (1928) to this problem and concludes : "In such a case, therefore, equilibrium would be attained by a division of society into two classes, the thrifty enjoying bliss and the improvident at the subsistence level". It would of course be politically quite incorrect to mention the poor nowadays, and this is why academics gladly adhere to the fiction of the representative consumer.

Problem 2: no one lives for ever This means that the coefficient ρ (pure rate of time preference) will apply to different

persons in the short to middle term (where the present generation is alive) and in the middle to long term (when we are all dead, and our descendants rule or are ruled). This means that this parameter is put to two different uses:

• for weighing consequences to me of my own actions • for weighing consequences to others of my own actions In a seminal paper, Sumaila and Walters (2005) separate the (psychological)

impatience from the (ethical) concern for future generations. Their model combines three parameters:

73

• the population is renewed at the rate γ • each generation has a pure rate of time preference ρ • each generation discounts at the rate ρδ < the utility of future generations For an event which is to happen at time t , we find that the discount factor to apply is:

( ) ( )dseeetR ststt −−−−

∫+ ρδρ γ0

= (24)

( ) tt ee δρ λλ −− +−1= (25) with:

δρ

γλ

−= (26)

Note that this corresponds to a non-constant rate of time preference: ( ) ( )( )tt eetr δρ λλ −− +−− 1ln=

• ( ) δ→tr when ∞→t (long-term rate) • ( ) γδ +=0r (short-term rate) We shall assume ρδγ <+ , so that 1<<0 λ .

4 Non-constant discount rates 4.1 Time inconsistency Let us summarize the arguments for non-constant interest rates: • there is strong experimental evidence that the psychological discount rate is not

exponential, but hyperbolic: it is more like ( ) tt αµ −+1 than ( )tρ−exp

• even within the framework of the standard model, where the representative

consumer has a constant psychological discount rate ρ , he/she ends up facing non-constant interest rates (see problem (14)), unless his/her utility happens to be CRRA. So, in the case of general utility functions, there will be time inconsistency even if the psychological discount rate is constant

• aggregating the beliefs of several individuals will lead to non-constant interest

rates, as Weitzman pointed out, even if each of them has a constant psychological discount rate

74

• if the planner takes the interests of future generations into account, he/she will have to discount future welfare at a non-constant rate, even if he/she and the representative consumer both have constant rates of time preference, and even if this rate is the same !

My view is that this the case is overwhelming. Even if it were not, there is a final

argument to be made: any policy recommendation drawn from the analysis must be robust to small changes in the model: one should ask, for instance, what becomes of the Ramsey model if the psychological discount rate is not exactly exponential. One would hope that there still is an optimal strategy, which converges to a stationary state, given by a suitably modified golden rule.

Unfortunately, non-constant interest rates create a special situation, known as time inconsistency. This is best explained in the framework of intergenerational equity, as explained at the end of the preceding section. The present generation faces the problem:

( ) ( )( ) ,max0

dttcutR∫∞

( )( ) ( ) ( ) ( ) 0=0and= kktckgntkfdt

dk−+−

However, the present generation will not be around to implement the policy it has

designed today, and must rely upon others (namely future generations) to do so. But the future generations may not agree with decisions taken on their behalf many years before they were around, and decide to carry out different ones. This is the non-commitment problem, which can be avoided only if the policy which seems optimal to us today still seems optimal to them when they are in charge. Unfortunately, with a non-constant rate of time preference, this will not happen.

Take two scenarios ( )⋅1c and ( )⋅2c , both of which kick in at time T . In other words,

( )⋅1c and ( )⋅2c are defined for Ts ≥ . Say that we compare them, at some time Tt <1 , and we

find ( )⋅1c is superior to ( )⋅2c :

( )( ) ( )( )dttcuttRdttcuttRTT 2111 )()( −≥− ∫∫∞∞

(27)

Let some time elapse, and do the comparison again at some later instant Tt <2 . Is it

still true that we will find ( )⋅1c superior to ( )⋅2c ? This would mean that:

( )( ) ( )( )dttcuttRdttcuttRTT 2212 )()( −≥− ∫∫∞∞

(28)

In the case when the discount rate is constant, so that ( ) exp= rttR , the first inequality implies the second because of the special properties of the exponential function. We have:

( )( ) ( ) ( )( )dttcuedttcuttRttr

TT 12

12 =)(−∞∞

∫∫ −

( ) ( ) ( )( )dttcuee

ttr

T

ttr

1121=

−∞−

∫

( ) ( )( )dttcuttRe

T

ttr

1121 )(= −∫

∞−

so that (28) is derived from (27) by multiplying both sides by a constant. In the case of non-constant discount rates, (27) no longer implies (28) ! In fact, a

policy which is optimal for the decision-maker at time 1t , no longer is optimal for the

decision-maker at a later time 2t (even though the utility function ( )cu is unchanged). There is no control that will be simultaneously optimal for all those who will have to implement it.

75

The fact that this problem does not occur with constant interest rates is a miracle, which was already pointed out by Samuelson in 1937, with his usual foresight (Samuelson, 1937): "our equations hold only for an individual who is deciding at the beginning of the period how he will allocate his expenditures over the period. Actually, however, as the individual moves along in time there is a sort of perspective phenomenon in that his view of the future in relation to his instantaneous time position remains time invariant, rather than his evaluation of any particular year (e.g. 1940). This relativity effect is expressed in the behaviour of men who make irrevocable trusts, in the taking out of life insurance as a compulsory savings measure, etc. The particular results we have reached are not subject to criticism on this score , having been carefully selected so as to take care of this provision. Contemplation of our particular equations will reveal that the results are unchanged even if the individual discounts from the existing point of time rather than from the beginning of the period." The last sentence, of course, means that Samuelson uses constant interest rates.

So, if interest rates are not constant, and there is no commitment technology, we will have to discard optimal control theory and build a new theory to replace it. This is a heavy mathematical work, which was carried out by Ekeland and Lazrak (2005, 2008). In the sequel, we describe their results.

4.2 Equilibrium strategies As above, we shall consider a general discount function [ ] RR →∞0,: . Throughout, it

will be assumed to be continuously differentiable, with:

( ) ( ) ( ) ∞≥ ∫∞

<,0,1=00

dttRtRR

and we consider the intertemporal decision problem (as it is seen at time 0=t )

( ) ( )( )

( )( ) ( ) ( ) ( ) ( ) 0

0

=0and=

,max

kktctkgntkfdt

dk

dttcutR

−+−

∫∞

(29)

Without loss of generality we are assuming that 0== ng (just change the definition of the

production function ( )kf ) Because of time-inconsistency, problem (29) can no longer be seen as an optimization

problem.There is no way for the decision-maker at time 0 to achieve what is, from her point of view, the first-best solution of the problem, and she must turn to a second-best policy: the best she can do is to guess what her successors are planning to do, and to plan her own consumption ( )0c accordingly. In other words, we will be looking for a subgame-perfect equilibrium of a certain game.

A second idea now comes into play: we will assume perfect competition between decision-makers: none of them is sufficiently powerful to influence the global outcome. In the spirit of Aumann (1964) we will consider that the set of decision-makers is the interval [ ]T0, .

At time t , there is a decision-maker who decides what current consumption ( )tc shall be. As

is readily seen from the equation of motion ( ) ckfdtdk −=/ , changing the value of c at just one point in time will not affect the trajectory. However, the decision-maker at time t is allowed to form a coalition with her immediate successors, that is with all [ ]ε+∈ tts , , and we will derive the definition of an equilibrium strategy by letting 0→ε . In fact, we are assuming that the decision-maker t can commit her immediate successors (but not, as we said before, her more distant ones), but that the commitment span is vanishingly small.

76

We restrict our analysis to Markov strategies, in the sense that the policy depends only on a payoff relevant variable, the current capital stock and not on past history, current time or some extraneous factors. Such a strategy is given by ( )kc σ= , where RR →:σ is a continuously differentiable function. If we apply the strategy σ , the dynamics of capital accumulation from 0=t are given by:

( )( ) ( )( ) ( ) 0=0,= kkskskfds

dkσ−

We shall say σ converges to k , a steady state of σ , if ( ) ksk → when ∞→s , when

the initial value 0k is sufficiently close to k . A strategy σ is convergent if there is some k

such that σ converges to k . In that case, the integral is obviously convergent, and its successive derivatives can be computed by differentiating under the integral. Note that if σ converges to k , then we must have ( ) ( )kkf σ= .

Suppose a convergent Markov strategy ( )kc σ= , where RR →:σ is a continuously differentiable function, has been announced and is public knowledge. The decision maker begins at time 0=t with capital stock k . If all future decision-makers apply the strategy σ , the resulting capital stock 0k future path obeys

( )( ) ( )( ) 0,= 000 ≥− ttktkf

dt

dkσ (30)

( ) .=00 kk (31)

We suppose the decision-maker at time 0 can commit all the decision-makers in [ ],0, ε where 0>ε . She expects all later ones to apply the strategy σ , and she asks herself if it is in her

own interest to apply the same strategy, that is, to consume ( )kσ . If she commits to another

bundle, c say, the immediate utility flow during [ ]ε0, is ( )εcu . At time ε , the resulting

capital will be ( )( )εckfk −+ , and from then on, the strategy σ will be applied which results

in a capital stock ck satisfying

( )( ) ( )( ) εσ ≥− ttktkfdt

dkcc

c ,= (32)

( ) ( )( ) .= εε ckfkkc −+ (33)

The capital stock ck can be written as ( ) ( ) ( )εtktktkc 10= + where:

( )( ) ( )( )( ) ( ) εσ ≥− ttktktkfdt

dk '' ,= 1001 (34)

( ) ( ) ckk −σε =1 (35)

and 'f and 'σ stand for the derivatives of f and σ . Summing up, we find that the total gain for the decision-maker at time 0 from consuming bundle c during the interval of length ε when she can commit, is

( ) ( ) ( ) ( )( )( ) ,10 dttktkushcu εσεε

++ ∫∞

and in the limit, when 0→ε , and the commitment span of the decision-maker vanishes, expanding this expression to the first order leaves us with two terms

( ) ( )( )( )dttkuth 00σ∫

∞

( ) ( ) ( ) ( )( )( ) ( )( ) ( ) .)( 1000

+−+ ∫

∞

dttktktkuthkucu '' σσσε (36)

77

where 1k solves the linear equation

( ) ( )( ) ( ) 0,)()(= 1001 ≥− ttktktkf

dt

dk '' σ (37)

( ) ( ) .=01 ckk −σ (38)

Note that the first term of (36) does not depend on the decision taken at time 0 , but the second one does. This is the one that the decision-maker at time 0 will try to maximize. In other words, given that a strategy σ has been announced and that the current state is k , the decision-maker at time 0 faces the optimization problem:

( )ckPc

,,max 1 σ (39)

where

( ) ( ) ( ) ( ) ( )( )( ) ( )( ) ( ) .)(=,, 10001 dttktktkuthkucuckP '' σσσσ ∫∞

+− (40)

In the above expression, ( )tk0 solves the Cauchy problem (30),(31) and ( )tk1 solves the

linear equation (37),(38). Definition 5 A convergent Markov strategy RR →:σ is an equilibrium strategy for

the intertemporal decision problem (29) if, for every Rk ∈ , the maximum in problem (39) is attained for ( )kc σ= :

( ) ( )ckPkc

,,maxarg= 1 σσ (41)

The intuition behind this definition is simple. Each decision-maker can commit only

for a small time ,ε so he can only hope to exert a very small influence on the final outcome. In fact, if the decision-maker at time 0 plays c when he/she is called to bat, while all the others are applying the strategy σ , the end payoff for him/her will be of the form

( ) ( )ckPkP ,,, 10 σεσ +

where the first term of the right hand side does not depend on c . In the absence of commitment, the decision-maker at time 0 will choose whichever c maximizes the second term ( )ckP ,,1 σε . Saying that σ is an equilibrium strategy means that the decision maker at

time 0 will choose ( )kc σ= . Given the stationarity of the problem, if the strategy ( )kc σ= is chosen at time 0 , it will be chosen at any future time t and as a result, the strategy σ can be implemented in the absence of commitment. Conversely, if a strategy σ for the intertemporal decision model (29) is not an equilibrium strategy, then it cannot be implemented unless the decision-maker at time 0 has some way to commit his successors.

4.3 The quasi-exponential case From now on, we shall use the following specifications: ( ) ( ) ( ) ( ) ( ) ccutttR ln=andexp1exp= ρλδλ −−+− (42)

with 1<<0 λ . Using a logarithmic utility simplifies the computations, but the results extend to general CARA utilities ( ) ( )θθ −− 1/= 1ccu , with 0>θ , and presumably to more general utilities as well. The argument in Erreur ! Source du renvoi introuvable. relies heavily the fact that the discount function is quasi-exponential. It has been extended to the case:

( ) ( ) tettR ρλ −+1=

78

by [40], and it would presumably extend to discount functions of the form

( ) ( ) ( )ttPtR ii

n

iρ−∑ exp=

1=, where the iP are polynomials, although this has not been done.

Under the specifications (42), Ekeland and Lazrak have obtained an extension of the behaviour observed in the classical Ramsey model, with constant discount rate: they have found equilibrium strategies which converge to some asymptotic growth rate ∞k , independent

of the initial capital 0k . The precise value of ∞k must lie in an interval, which converges to

the golden rule (4) when the discount function becomes exponential. Denote by K is the flow associated with the differential equation (30) defined by

( ) ( )( ) ( )( )ktktf

t

kt,;,;=

,;σσσ

σKK

K−

∂

∂

( ) .=;0, kkσK

Definition 6 Take some 0>∞k . We shall say that σ is a local equilibrium strategy

converging to ∞k if there is some open interval Ω around ∞k such that σ is defined and 2C

on Ω , the flow ( )kt,;σK sends Ω into itself, and: • ( ) ( )ckPk c ,,maxarg= 1 σσ for all Ω∈k • ( ) ∞→ kkt,;σK when ∞→t , for all Ω∈k Here ( )ckP ,,1 σ is given by formula (40). Our main result then is: Theorem 7 Assume f is 3C for 0>k . Define k ≤ k by:

( ) ( ) ( )

ρ

λ

δ

λρλλδ

−+

−+11

=,1= kfkf '' (43)

Then, for every [ ]kkk ,∈∞ , there exists a local equilibrium strategy converging to ∞k . In the Ramsey case, when ρδ = , or 1=λ , we find the classical relation ( ) ρ=∞kf ' .

In the general case, the golden rule (4) is replaced by the inequality kkk ≤≤ ∞ we find a continuum of possible equilibrium strategies, and corresponding asymptotic growth rates, and their range is fully characterized. So the proof in Ekeland and Lazrak (2008) is in two parts: first showing that every possible ∞k is in that range, and then showing that every point in that

range is a possible ∞k . Note that there is still an indeterminacy, smaller of course as one nears the exponential

case, but present. This indeterminacy arises from the fact that there are no boundary condition as ∞→t , nothing to replace the transversality condition at infinity of the exponential case. A further game-theoretical argument, however, will enable us to do away with that indeterminacy, and give a definite recommendation to the policy-maker.

79

Definition 8 Let σ and 'σ be two equilibrium strategies converging to ∞k and 'k∞ .

We shall say that σ is eventually dominated by 'σ if, for any starting point k , there is some 0>t such that:

• for all ts > , the decision-maker at time s prefers 'σ to σ • if one applies strategy 'σ after time t , it remains true that at all subsequent times,

the decision-makers prefer 'σ to σ In Ekeland and Lazrak (2008), it is proved that, whenever a strategy converges to

some ∞k , it is eventually dominated by any strategy that converges to some ∞∞ kk ' > . So the only equilibrium strategy which is not dominated is the rightmost none, namely the one which converges to k . Let us express this result:

Proposition 9 All convergent equilibrium strategies are dominated, except the one(s)

which converge(s) to k . So the rational choice is now clear: it is the equilibrium strategy which converges to

the point k where

( )( )

−+

ρ

λ

δ

λ 1

1=kf ' (44)

Indeed, for any other choice, one of the future decision-makers will switch to another

strategy, in the knowledge that his/her successors will follow suit. So there is no point in applying now a strategy which one knows will not be implemented later on, even if one has to wait for the distant future.

Applying the Sumaila-Walters specifications (24), (25) and (26) gives:

( )

+−

γδρ

/1

11=kf '

which is to be compared to the golden rule (4), where we we have taken logarithmic utility, so that 1=θ , and 0=n (no technological progress), so that it becomes ( ) ρ=∞kf ' . So concerns for intergenerational equity lower the interest rate, by a factor which depends only on γδ/ .

4.4 Bibliographical notes The first paper to investigate the Ramsey model of economic growth with non-

constant discount rates is due to Barro (1999), who investigated the case of logarithmic utility. There was an earlier literature, in the discrete-time framework, originating with the seminal papers of Strotz (1956) and Phelps and Pollack (1968). Going from the discrete to continuous time proved to be mathematically challenging.Ekeland and Lazrak (2005, 2008) then introduced the idea of perfect competition between decision-makers, which enabled them to characterize seem to fully characterize equilibrium strategies in this case and to derive an analogue of the HJB equation. Some of this work was anticipated by Karp, in (Karp, 2007); note also (Karp, 2005), which applies these results to climate change.

80

5 Conclusion This introduction to long-term interest rates, incomplete as it is, would be even more

so if I failed to direct the reader to (Lind, 1982) and (Portney and Weyant, 1999), which are standard references in the field. As a personal conclusion, I would like to remind the reader once more that determining the proper interest rates to use for projects with very long-term consequences, such as those which impact the environment, is one of the most important ways that the economic profession can contribute to solving the major challenges which our planet faces today. Such interest rates should incorporate the distributional and ethical concerns of our contemporaries. In other words, economics, after decades of riding the tiger of economic expansion, should once more become a normative science and take the lead.

References [1] Aghion, Philippe and Howitt, Peter (1998) "Endogeneous growth theory", MIT

Press [2] Aubin, Jean-Pierre and Clarke, Frank (1979) "Shadow prices and duality for a

class of optimal control problems", SIAM J. Control Optim. 17 , pp. 567--586. [3] Aumann, Robert "Markets with a continuum of traders", Econometrica, 32 (1964),

p. 39-50 [4] Barro, Robert and Sala-i-Martin, Xavier (1995) "Economic growth", Mc-Graw and

Hill [5] Barro Robert J. " Ramsey Meets Laibson in the Neoclassical Growth Model",

Quarterly Journal of Economics 114 (November 1999): 1125-52. [6] Blanchard, Olivier and Fisher, Stanley (1994) "Lectures on macroeconomics",

MIT Press [7] Bressan, Alberto and Piccoli, Benedetto (2007) "Introduction to the Mathematical

Theory of Control" Applied. Math Series vol. 2, American Institute of Mathematical Sciences [8] Clarke, Frank, (1983), "Optimization and Nonsmooth Analysis", Wiley; reprinted

(1990), SIAM, series "Classics in Applied Mathematics" [9] Boehm-Bawerk, Eugen von (1884), "History and Critique of Interest Theories" [10] Ekeland, Ivar and Lazrak, Ali (2005) "Being serious about non-commitment",

http://arxiv.org/abs/math/0604264 [11] Ekeland, Ivar and Lazrak, Ali (2008) "Equilibrium policies when preferences are

time inconsistent", http://arxiv.org/abs/0808.3790

81

[12] Ekeland, Ivar and Scheinkman, Jose (1986) "Transversality conditions for some infinite horizon discrete time opimization problems", Mathematics of Operations Research 11, p. 216-229

[13] Ellsberg, Daniel (1961), "Risk, Ambiguity, and the Savage Axioms", Quarterly

Journal of Economics 75 (4), p. 643--66 [14] Fisher, Irving (1930) "The Theory of Interest" [15] Gierlinger, Johannes and Gollier, Christian, "Socially Efficient Discounting under

Ambiguity Aversion", IDEI Working Paper, n. 561, November 2008. [16] Guesnerie, Roger (2004) "Calcul economique et developpement durable", Revue

Economique, 55 (3), p. 363-382 [17] Guesnerie, Roger, Guéant, Olivier, and Lasry, Jean-Michel (2008), "Economic

Calculus and Sustainable Development", Cahiers de la Chaire de Développement Durable, Université Paris-Dauphine http://www.ifd.dauphine.fr/fileadmin/mediatheque/recherche_et_valo/FDD/CAHIER16_GUESNERIE_ET_AL.pdf

[18] Heal, Geoffrey (2007) "Climate change economics: a meta-review and some

suggestions" [19] Kamihigashi, Takashi (2001) "Necessity of Transversality Conditions for Infinite

Horizon Problems". Econometrica 69 (4), 995-1012. [20] Karp, Larry " Non-constant discounting in continuous time", Journal of

Economic Theory 132 (2007), p. 557-568 [21] Karp, Larry "Global warming and hyperbolic discounting", Journal of Public

Economics 89 (2005), 261-282 [22] Klibanoff, P., Marinacci, M., and Mukerji, S. (2005) "A smooth model of decision

making under ambiguity" Econometrica 73, p. 1849-1892 [23] Koopmans, Tjalling (1960), "Stationary utility and impatience", Econometrica 28

(2), p. 287-290 [24] Lind, Robert C., editor (1982), "Discounting for Time and Risk in Energy Policy",

Resources for the Future, Washington D.C, distributed by John Hopkins Press [25] Michel, Philippe (1982) " On the Transversality Condition in Infinite Horizon

Optimal Problems." Econometrica 50, p. 975-985. [26] Nordhaus, William D. (1991) "To slow or not to slow: the economics of the

greenhouse effect" Economic Journal 101 p. 920-937 [27] Nordhaus, William D. (1992) "Lethal model 2: the limits to growth revisited"

Brookings papers on economic activity (2) P. 1-43

82

[28] Phelps, Edmund S., and Pollak, Robert A. " On Second-Best National Saving and

Game-Equilibrium Growth.' Review of Economic Studies 35 (April 1968): 185-99. [29] Portney, Paul and Weyant, John (1999), "Discounting and Intergenerational

Equity", Resources for the Future, Washington D.C. [30] Rae, John (1834) "Statement of some new Principles on the subject of Political

Economy" [31] Ramsey, Frank (1928) "A mathematical theory of savings", The Economic

Journal, 38 (152), p 543-559 [32] Samuelson, Paul (1937) ""A note on measurement of utility", Review of

Economic Studies 4, p. 155-61 [33] Sumaila, Ussif and Walters, Carl (2005) "Integenerational discounting: a new

intuitive approach" Ecological Economics 52 p. 135-142 [34] Ramsey, Frank (1928) "A mathematical theory of saving", Economic Journal 38,

p. 543-559 [35] Romer, David (1996), "Advanced macroeconomics", McGraw and Hill [36] Strotz, Robert H. " Myopia and Inconsistency in Dynamic Utility Maximization"

Review of Economic Studies 23, no. 3 (1956): 165-80. [37] Weitzman, Martin (2001) "Gamma discounting", American Economic Review 91

(1) p. 260-271 [38] Weitzman, Martin (2008) "On modeling and interpreting the economics of

catastrophic climate change", American Economic Review 91 (1) p. 260-271 [39] Weitzman, Martin (2007) "A review of the Stern review on the economics of

climate change", Journal of Economic Literature vol. 45, no3, pp. 703-724 [40] Qinglong Zhou, personal communication, Nankai University, June 23, 2009

83

Intergenerational Discounting - A Debate This chapter is in three sections. The first, by Rashid Sumaila and Carl Walters, is an article

on intergenerational discounting. The second is a commentary on it by Michael Prager and

Karl Shertzer, while the third is Sumaila and Walters’s reply.

Intergenerational discounting: a new intuitive approach U. Rashid Sumaila, and Carl Walters Remembering the future, a commentary on “Intergenerational discounting: a new intuitive approach” Michael H Prager and Karl W. Shertzer Making future generations count, comment on “remembering the future” U. Rashid Sumaila, and Carl Walters

84

Intergenerational discounting: a new intuitive approach1,2

U. Rashid Sumaila, Carl Walters

Abstract :This paper proposes a new intergenerational discounting approach for

computing net benefits from the use of environmental resources. The approach explicitly incorporates the perspectives of both the current and future generations, as argued for by Pigou [Pigou, A.C., 1920. The Economics ofWelfare 1952 (4th edition), London: Macmillan] and Ramsey [Ramsey, F.P., 1928. A mathematical theory of saving, Econ. J., 38 543–559], and required by most national and international laws related to the use of these resources. An equation for use in the calculation of net discounted benefits is developed, which provides a ‘middle’ position whereby both the ‘reality' of ‘personal’ discounting and that of ‘social’ discounting are included in a social welfare function.

Keywords: Current and future generations; Discount factor; Environmental resources

1 Copyright © 2004 Elsevier B.V. All rights reserved. This paper was first published in Ecological Economics in 2005. We are very grateful to Professors Rashid Sumaila and Carl Walters for their permission to reprint these pages.

2 [Authors’ Acknowledgments] We are grateful to the Sea Around Us Project (and its members) at the Fisheries Centre, a Partnership of the Pew Charitable Trusts Oceana and participants at various conferences for their valuable comments and suggestions. Our thanks also go to James Amegashie, Claire Armstrong, Frank Asche, John Bishir, Trond Bjorndal, Tony Charles, Colin Clark, Derek Clark, Jon Conrad, Sjur D. Flåm, Lawrence Goulder, Robert McKelvey, Rosamond Naylor, Erling Risa, Keith Sainsbury, Daniel Walters and three anonymous referees for their helpful comments and suggestions on the idea reported herein.

85

This contribution is another attempt at grappling with the vexing problem of

discounting flows of net benefits from natural and environmental resources. It is an attempt at answering the question: how should we discount flows of benefits in order to more adequately take into account the interests of future generations with respect to their needs from natural and environmental resources.

In comparing the present values of policy alternatives, it is standard to discount net benefits that will accrue in the future compared to net benefits that can be achieved today (Koopmans, 1960; Heal, 1997). Since cost-benefit analysis (CBA) discounts streams of net benefits from a given project or policy alternative into a single number, namely, the net present value (NPV), discount rate assumptions used in these time stream comparisons can have a big impact on the apparent best policy or project (Nijkamp and Rouwendal, 1988; Burton, 1993; Fearnside, 2002). In particular, high discount rates favor myopic policies or projects that continue to exaggerate unsustainable resource use such as global overfishing (see for example, Pauly et al., 2002; Koopmans, 1974; Clark, 1973).

Discounting as described above has attracted considerable attention from economists since Böhm-Bawerk (1889) and Fisher (1930) invented intertemporal preferences. There are many arguments for and against standard discounting in the literature (see for example, Baumol, 1952; Sen, 1961; Marglin, 1963; Arrow, 1979 and Becker, 1980; Nijkamp and Rouwendal, 1988; Burton, 1993; Goulder and Stavins, 2002; Fearnside, 2002). A simple defense of discounting is that people have a positive time preference, which, it is argued, needs to be respected by the social planner (Bauer, 1957; Eckstein, 1957). Goulder and Stavins (2002) provide a concise and typical defense of standard discounting. They argue that standard discounting is meant to ensure that the present value of net benefit calculations provide a meaningful indication of whether the efficiency criterion is satisfied or not. The authors suggest that adjustments of the discount rate to accommodate other legitimate policy questions such as intergenerational equity are problematic, as they blur the distinction between the efficiency criteria, and other legitimate policy-evaluation criteria. They therefore argue that in evaluating policies, it seems better to use the market interest rate while judging intergenerational fairness by direct examination. In other words, this should be done outside the cost-benefit framework.

Many authors disagree with the arguments of Goulder and Stavins (see Schelling, 1995; Rabl, 1996; Lind, 1995). For instance, Padilla (2002) state that intergenerational problems arise in standard discounting partly because of intergenerational externalities. This externality arises because future generations do not participate in decisions that will affect them. They cannot defend their interests in current decision making even though present decisions can have irreversible impacts on their welfare. In a situation where one party is absent, the ‘Coasian’ and ‘Pigouvian’ solution cannot help Padilla (2002). Schelling (1995) makes the point that a utility discount rate measures emphatic distance, and since future generations cannot be emphatically distinguished, discounting is inappropriate for intergenerational issues. Lind (1995) makes the interesting point that standard discounting makes the implicit assumption that designated capital transfers between generations are possible. An assumption Lind finds to be somewhat incorrect. Daly and Cobb (1989) declare that the idea of discounting losses of ‘natural capital’ is to be rejected in principle. They claim that irreversible losses such as these cannot be adequately adjusted for by lowering the discount rate. Rabl (1996), and also Schelling (1995), argue that discounting within a given generation is appropriate but not so between generations.

According to Chichilnisky (1996), sustainability means that the preferences of the current generation do not dominate the preferences of future generations in determining the intergenerational distributions of resources. She used this axiom to develop an intertemporal

86

welfare, which is expressed as the weighted sum of standard net present welfare and the limiting properties of the system under consideration. The need for the interest of future generations to be included in our social welfare function (intergenerational equity), as required in many management jurisdictions via sustainability mandates, is more powerfully expressed in the case of natural and environmental resources, in particular, climatic change (see IPCC, 1996, 2001; Weitzman, 2001; Nordhaus, 1997; UNEP, 1987). This is because it is believed by many that damages to these life support resources can be irreversible (Daly and Cobb, 1989).

The above criticisms have led to the development of a number of non-standard discounting approaches. Cline (1992) argues that the pure time preference component of the discount rate should be set to zero, thereby allowing for only the opportunity cost of capital component. The implication of this argument is to drive the discount rate below the market rate. Using empirical evidence from cognitive psychology, Heal (1997, 1998) concluded that standard discounting is inappropriate. He proposed a substitute, which depends on the length of time under consideration in a logarithmic fashion. Weitzman’s gamma discounting (Weitzman, 2001) is an approach that exhibits similar behavior to Heal’s logarithmic discounting. In developing his approach, Weitzman (2001) based his argument on the fact that there are huge uncertainties about the magnitude of future discount rates.

Other discounting approaches are the Chichilnisky criterion (Chichilnisky, 1996), Rabl discounting (Rabl, 1996), where the discount rate is set equal to zero at a certain point in the future, and the Fearnside unified index (Fearnside, 2002). Other interesting approaches are those of Collard (1981), Bellinguer (1991) and Nijkamp and Rouwendal (1988). Overlapping generations models have also been used to address the problem between intertemporal efficiency and intergenerational fairness as criteria for social choice (see for instance, Burton, 1993; Howarth and Norgaard, 1995; Howarth, 1996).

1. Rationale for the new approach In contrast to Goulder and Stavins (2002), this paper makes the case that not only do

economists need to provide decision makers analysis that reveal the standing of a planned environmental project or policy with regards to the economic efficiency criteria, they also need to develop evaluation approaches that explicitly include legitimate policy questions such as intergenerational equity. Alternative methods, like the one presented in this paper, will allow comparison and trade-off analysis between results from different approaches. They can be used to answer the question, how much in ‘current generation discounted dollars’ do we need to give up in order to ensure that future generations have the benefit of inheriting ‘healthy’ natural and environmental resources. In other words, this and similar approaches can be used to determine the ‘price’ in standard discounted dollars of ensuring the sustainable use of these resources.

According to Fearnside (2002), the decision as to the relative weight to be given to short- versus long-term effects (in other words, current versus future generation interests) is a policy rather than a scientific question. For most nations in the world, this policy question has already been answered via sustainability mandates. For example, the Magnuson-Stevens Fisheries Conservation and Management Act of the USA (Anon, 1996) specifically demand that the interests of future generations be taken into account in the management of the nation’s fishery resources. Our approach provides an intuitive way to incorporate this requirement into the discounting approach.

A point made succinctly by Tol (1999) is that the choice of discount rate (and discounting approach) is both empirical and ethical. It is empirical because people do make trade-offs between present and future in their daily decisions. It is at the same time ethical

87

because the discount rate determines the allocation of intertemporal goods and services between generations. Tol states that neither the empirical nor the ethical should overrule the other in the choice of discount rate or discounting approach. Tol’s point relates to the discussion about ‘personal tastes’ and ‘social tastes’, which makes it possible to argue that people may really prefer the use of lower discount rates to evaluate societal goals and objectives, even while possessing a personal high time preference rate (see IPCC, 2001). People’s political choices reflect their ‘social tastes’, but their personal economic choices reflect their ‘personal tastes’. One reason for this is simply that the ‘frame of reference’ is different in personal and social considerations (Marglin, 1963). Thus, one may conclude that people’s social tastes are for lower rates than the market discount rate. This statement is reinforced by the fact that, at least, some members of the current generation actually care about benefits to generations yet unborn (see Popp, 2001). Our approach explicitly takes into account the ethical (social tastes) component of discounting while not neglecting the empirical (personal tastes).

Impatience of the individual is fundamental to financial decisions at all levels. It is an ingrained human attribute that allows us to instinctively account for uncertainty, lost opportunities, and other considerations relevant in resource acquisition. Standard discounting merely emulates our time preference on a larger scale; providing an analytical means to make value-based judgments.

Standard discounting though fails to adequately capture human proclivity. Viewed as any other investment, the education of children generally yields a negative net present value at most practical rate of discount, making alternative investments more attrac- tive. Yet parents and society seemingly disregard conventional financial wisdom, educating their children with little promise of return save the confidence that they have equipped them with the tools needed for survival. Indeed, “altruism” occurs at all levels of society without the concession that future benefits to our offspring carry significant value for us in the present. The discounting method we propose can more accurately model this critical behavior. Significantly, it can provide us with a responsible means to value the flow of benefits from long-term environmental policies when investors are separated in time from recipients. More practically, application of this new procedure may now offer incentives to rebuild depleted ecosystems such as the Grand Banks of Canada.

A final motivation for the current approach may be derived from Rawl’s (1972) veil of ignorance. It appears to us that if decisions by and for society are taken under the assumption that neither the current nor future generations know their position in terms of who comes first, they will all agree that our approach, which explicitly takes into account the interest of future generations, should be applied rather than the standard discounting approach. The approach aims to provide a basis for stewardship for future generations’ welfare as argued for by Brown (1992) and Coward et al. (2000).

2. Deriving the intergenerational discounting equation The intergenerational discounting equation is a generalization of the original idea

presented in Sumaila (2004). It is developed as follows. For each simulated future year, we treat the benefits as accruing to the current generation (at standard discount rates) plus to each of the annual 1/(generation time) increments of new stakeholders who will have entered the stakeholder population by that future year. Each incremental group of new stakeholders is assumed to discount future benefits at the standard or normal rate after entering the stakeholder population. In this manner, we are able to include the interest of all generations as argued for earlier in this paper. We consider these assumptions axiomatic because we agree with Marglin (1963) that a democratic government should reflect only the preferences of the

88

individuals who are presently members of the body politic, since these preferences have been shown to include the interests of future generations (see for instance, Popp, 2001).

The discounting equation resulting from these assumptions contains two factors: the standard, normal annual discount rate that is assumed to apply to all stakeholders after entering the stakeholder “future population”, and a “future generation” discount rate that represents our willingness to forego benefits that we can obtain for ourselves now in favor of benefits that would accrue to future stakeholders.

The net present value of a flow of net benefits, NPV, is defined as

tt

T

t

WVNPV ∑0=

= (1)

where tV is the net benefit in period t , and tW is the weight used to discount tV to the

net present value. Let d denote the discount factor given the prevailing standard discount rate, r :

r

d+1

1= (2a)

Then, the conventional weight or discount factor, tcW , , in a given period or year, t , is

given by t

tc dW =, (3a)

Let the annual future generation discount rate be fgr , then the future generation

discount factor, fgd , is given by

)(1

1=

fgfg r

d+

(2b)

To develop the modified Eq. (3a) for the proposed intergenerational discount factor,

we expand our earlier assumptions as follows: (i) the present generation (stakeholders) discount flows of values at the standard rate, (ii) a new generation of size G1/ enters the population every year, where G is the generation time. This cohort of people discount values at the standard rate every year after entry, (iii) the current generation (as decision makers) discounts the interests (values) of these new entrants at a future generation rate, per generation (or rate/generation time per year). It should be noted that the within-generation discounting factor already includes some expectation of possible mortality for the existing generations, so there is no need to explicitly account for this.

Putting these assumptions together mathematically, we get the expression below for the intergenerational weight or discount factor in year t , tiW , :

∆−

∆−+

−

1

1=

1

,

ttfgt

ti G

dddW (3b)

89

where )/(= dd fg∆ , the ratio between the intergenerational and the standard discount

factor. An examination of Eq. (3b) shows that the derived formula is valid for fgd greater

than or less that d . It reduces to the formula below in the case in which dd fg = .

G

tdddW

tfgt

ti

1

, =−

+ (3c)

The full derivation of Eqs. (3b) and (3c) is given in Appendix A. Eqs. (3b) and (3c)

reduce to the standard discount factor when 0=fgd . By taking the derivative of Eqs. (3b) and

(3c) with respect to G, we see that the intergenerational discount factor, tiW , , decreases with

increasing generation time. Note that Eq. (3b) avoids the comparability problem3 for all 0>fgr .

3. Comparing intergenerational and standard discounting To compare and contrast the standard or the so-called Samuelson discounting and the

proposed intergenerational discounting, Fig. 1 is developed. It plots the present value of a constant annual flow of $1 over a period of 100 years using these discounting approaches. A standard discount rate (r ) of 5% is assumed for the calculations. Under the standard approach, the flow of $1 is discounted using a rate of 5% in the usual manner. With respect to intergenerational discounting, we present three plots -- when rrfg < ( fgr =1%); == rrfg 5%,

and when rrfg > ( fgr =20%). The generation time, G, is given a value of 20 years.

Fig. 1. Present value of a constant annual flow of $1 over a period of 100 years for the

different approaches to discounting.

3This problem refers to the incompleteness of the social preference orderings. Specifically, it may be impossible

to determine which of two sequences of consumption is ‘socially better’ because with an infinite time horizon, it is quite possible that there are feasible consumption paths such that the sum of utilities is infinity. If two different consumption paths each yield utility sums that are infinite, then they become incomparable since two infinities cannot be arithmetically ordered.

90

From Fig. 1, we see that intergenerational dis- counting results in less discounting of future flows compared to the standard approach in all the cases presented, with the degree of discounting depending on the level of future generation discount rate.

4. Concluding remarks In addition to the standard discount rate, one will need to decide the future generation

discount rate and the generation time in order to calculate the intergenerational discount factor developed herein. We deem a generation time of about 20 -- 30 years to be appropriate. One could argue that most fishers and fisheries managers will not have more than this number of years to make decisions that would impact the sustainability of fisheries. An easy way to fix a value for the future generation discount rate is to make it equal to the standard discount rate. One could also carry out a survey of fishing communities, regions of a country or the whole of a country to find out what people are willing to accept as the future generation discount rate in relation to the standard rate. Finally, one could use the internal rate of return for educating people to the PhD level in a given country.

Our new discounting formula achieves many of the results of other alternative approaches to discounting, it does so in an intuitive manner by partially resetting the discounting clock (Sumaila, 2004), by taking into account the fact that a new cohort of people enter the population each year. The fact that our approach introduces two more parameters in the formula for calculating the discount factor may be seen as a disadvantage. But this, also, is the strength of the approach relative to other alternative approaches to discounting: It affords the approach invaluable flexibility, because the future generation discount rate can be chosen to reflect the particular situation at hand. For instance, its value can be fixed to capture the degree of irreversibility of the change to be brought about as a result of anthropogenic interventions. Another advantage of our approach compared to some of the alternative approaches available in the literature is that it successfully deals with the notorious comparability problem.

A potential disadvantage of our approach, like all the other alternative, is that its effect is to lower the discount rate as seen from the time perspective of the current generation, or what we call their discounting clock (Sumaila, 2004). This, it is argued, could serve as a double-edged sword with respect to conservation, because resource intensive projects that would otherwise not be profitable from the perspective of the private investor could turn out to be profitable with a lower discount rate (the so- called ‘conservationist dilemma’). But, the discount factor we propose in this paper is supposed to be applied at the level of society. Private level decisions will still be made using the higher private discount rate, within the overarching policy framework made at the level of society.

We have proposed a new discounting approach that would take into account the interests of all generations with regard to the use of environmental and natural resources. The approach recognizes the need for discounting of flows of benefits by each generation because we agree that each generation would prefer to have their benefits now rather than tomorrow. The paper also recognizes the fact that capital has an opportunity cost and therefore discounting is necessary. At the same time, the approach proposed builds in the need not to foreclose options to future generations when it comes to their future needs from the natural environment. In effect, we have produced a more balanced approach to discounting, which while recognizing the need for allowing for substitutability between natural and human made capital, does not allow for 100% substitutability. In this way, this approach can help policy makers design management solutions for the natural environment that would stop the kind of overexploitation of environmental and natural resources described in Koopmans (1974) and Clark (1973).

91

5. Appendix A. Derivation of intergenerational discounting equation The derivation of Eqs. (3b) and (3c) can be most easily understood by examining a

table of the present and future stakeholder value components that we propose should be included in tiW , :

That is, for each present stakeholder, we propose that new stakeholders numbering

G1/ should be entered into the value weighting at each future year t , with initial discount weighting (or present concern to us) of t

fgd . We assume that each of these incremental

stakeholder groups will discount value at the normal rate d after recruitment to the stakeholder population. tiW , is the sum of the elements of the t th row of this table, i.e.

GdGddGdddW tfgfg

tfg

ttti ///= 221

, +++ −− . Letting dd fg /=∆ as above, this sum can be written

simply as

G

ddW

ttt

ti

∆++∆+∆+∆+

−12

,

1=

(2)

Writing the ∆ series component of this sum as )/(1)1/(1 ∆−∆−∆− t (i.e. as the infinite

sum less the t +1 and following terms, treating ∆ as though it were always less than 1.0), we obtain Eq. (3b). Eq. (3c) is obtained by applying L’Hospital’s rule to Eq. (3b), to obtain the limit of Eq. (3b) as ∆ approaches 1.0. We have checked these algebraic results against the full table above using spreadsheets, and have confirmed that Eq. (3b) applies even when

1.0>∆ so that the infinite series simplification does not apply. References

[1] Anon, 1996. “Magnuson-Stevens Fishery Conservation and Management Act”, US Public Law, 94–265. J. Feder version (12/19/96). [2] Arrow, K.J., 1979. The trade-off between growth and equity. In: Greenfield, H.I., Levenson, A.M., Hamovitch, W., Rotwein, E. (Eds.), Theory for Economic Efficiency: Essays in Honor of Abba P. Lerner. M.I.T. Press, Cambridge, MA. [3] Bauer, P.T., 1957. Economic Analysis and Policy in Underdeveloped Countries. Duke

92

University Press, Durham, NC. [4] Baumol, W.J., 1952. Welfare Economics and the Theory of the State. Harvard University Press, Cambridge, MA. [5] Becker, R.A., 1980. On the long-run steady state in a simple dynamic model of equilibrium with heterogeneous households. Quarterly Journal of Economics 95, 375– 382. [6] Bellinguer, W.K., 1991. Multigenerational value: modifying the modified discount method. Project Appraisal 6, 101– 108. [7] Böhm-Bawerk, E.v., 1889. The positive theory of capital. 1923 Reprint of 1891 Smart Translation. Stechert, New York. [8] Brown, P.G., 1992. Climate and planetary trust. Energy Policy 20, 208–222. [9] Burton, P.S., 1993. Intertemporal preferences and intergenerational equity considerations in optimal resource harvesting. Journal of Environmental Economics and Management 24, 119– 132. [10] Chichilnisky, G., 1996. An axiomatic approach to sustainable development. Social Choice and Welfare 13, 219–248. [11] Clark, C.W., 1973. The economics of overexploitation. Science 181, 630– 634. [12] Cline, W.R., 1992. The Economics of Global Warming. Institute for International Economics, Washington, DC. [13] Collard, D., 1981. Altruism and Economy. The Pitman Press, Bath. [14] Coward, H., Ommer, R., Pitcher, T.J. (Eds.), Fish Ethics: Justice in the Canadian Fisheries. Institute of Social and Economic Research (ISER). Memorial University, St. John’s, Newfoundland, Canada. [15] Daly, H.E., Cobb, J.B., 1989. For the Common Good: Redirecting the Economy Toward Community, The Environment, and A Sustainable Future. Beacon Press, Boston. [16] Eckstein, O., 1957. Investment criteria for economic development and the theory of intertemporal welfare economics. Quarterly Journal of Economics 71, 56–85. [17] Fearnside, P.M., 2002. Time preference in global warming calculations: a proposal for a unified index. Ecological Economics 41, 21– 31. [18] Fisher, I., 1930. The Theory of Interest: As Determined by Impatience to Spend Income and Opportunity to Invest It, 1954 Reprint. Kelley and Millman, New York. [19] Goulder, L.H., Stavins, R.N, 2002. Discounting: an eye on the future. Nature 419, 673– 674. [20] Heal, G.M., 1997. Discounting and climate change. Climate Change 37, 335– 343. [21] Heal, G.M., 1998. Valuing the Future: Economic Theory and Sustainability. Columbia University Press, New York. [22] Howarth, R.B., 1996. Climatic change and overlapping generations. Contemporary Economic Policy 14, 100– 111. [23] Howarth, R.B., Norgaard, R.B., 1995. Intergenerational changes choices under global environmental change. In: Bromley, D.W. (Ed.), Handbook of Environmental Economics. Blackwell Science, Oxford. [24] IPCC, 1996. Contribution of Working Group 4 Report, http://www.ipcc.ch. [25] IPCC, 2001. Climate change 2001: The scientific basis, Contribution of Working Group 1 to the Third Assessment Report, Summary for policy makers: http://www.ipcc.ch. [26] Koopmans, T.C., 1960. Stationary ordinal utility and impatience. Econometrica 28, 287–309. [27] Koopmans, T.C., 1974. Proof of a case where discounting advances doomsday? Review of Economic Studies 41, 117– 120. [28] Lind, R., 1995. Intergenerational equity, discounting, and the role of cost–benefit analysis in evaluating climate policy. Energy Policy 23, 379– 389. [29] Marglin, S.A., 1963. The social rate of discount and the optimal rate of investment.

93

Quarterly Journal of Economics 77, 95– 111. [30] Nijkamp, P., Rouwendal, J., 1988. Intergenerational discount rates and long term plan evaluation. Public Finance 43, 195– 211. [31] Nordhaus, W.D., 1997. Discounting in economics and climate change: an editorial comment. Climatic Change 37, 328– 351. [32] Padilla, E., 2002. Intergenerational equity and sustainability. Ecological Economics 41, 69– 83. [33] Pauly, D., Christensen, V., Guenette, S., Pitcher, T.J., Sumaila, U.R., Walters, C.J., Watson, R., Zeller, D., 2002. Towards sustainability in world fisheries. Nature 418, 689–695. [34] Pigou, A.C., 1920. The Economics of Welfare 1952, 4th edition. Macmillan, London. [35] Popp, D., 2001. Altruism and the demand for environmental quality. Land Economics 77 (3), 339–350. [36] Rabl, A., 1996. Discounting of long-term costs: what would future generations prefer us to do? Ecological Economics 17, 137– 145. [37] Ramsey, F.P., 1928. A mathematical theory of saving. Economic Journal 38, 543– 559. [38] Rawls, J., 1972. Theory of Justice. Oxford University Press. [39] Schelling, T.C., 1995. Intergenerational discounting. Energy Policy 23, 395– 401. [40] Sen, A.K., 1961. On optimising the rate of saving. Economic Journal 71, 479–496. [41] Sumaila, U.R., 2004. Intergenerational cost benefit analysis and marine ecosystem restoration. Fish and Fisheries, 5, 329-343. [42] Tol, R.S.J., 1999. Time discounting and optimal emission reduction: an application of fund. Climate Change 41, 351–362. [43] UNEP, 1987. Our Common Future. Oxford University Press, New York, p. 400. [44] Weitzman, M.L., 2001. Gamma discounting. American Economic Review 91, 260–271.

94

Remembering the future: A commentary on “Intergenerational discounting: A new intuitive

approach”1

Michael H. Prager, Kyle W. Shertzer

Sumaila and Walters (2005) described a new approach to valuing benefit streams in

resource economics. In contrast to conventional methods, their “intergenerational discounting” recognizes two categories of benefits: those accruing to present stakeholders and those accruing to future generations. The new method has clear application whenever alternative management strategies for natural resources are compared in the metric of net present value (NPV).

The intergenerational approach provides a sensible way to integrate economics into policy considerations. To us, it seems far more reasonable than the conventional discounting approach, in which future benefits become of negligible value after a few generations. That aspect of conventional discounting is troublesome to many resource biologists, policy analysts, and economists.

To carry the argument further, if one believes that a major goal of economics is to quantify human preferences (and the corresponding goal of resource economics is to quantify societal preferences), it can be demonstrated by syllogism that the use of conventional discounting is logically inconsistent:

Major premise–Society has expressed a strong preference for sustainability in resource use. Minor premise–Procedures that use conventional discounting typically assign optimality to nonsustainable policies. Conclusion–Therefore, conventional discounting is inappropriate to quantify decisions on resource use.

Any doubts about the major premise of this syllogism can be put to rest by reflecting

that the vast preponderance of national laws and regulations governing resource use, as well as conventions defining international use, have sus- tainability as a major goal. This is quite conspicuous in fisheries, where the concept of maximum sustainable yield (MSY) has recovered, phoenix-like, from several deaths, a phenomenon remarkable enough to have been written about separately by Prager (1994), Mace (2001), and Punt (2001).

Besides expressing enthusiasm for the concept of intergenerational discounting, we wish to point out a few technical issues in Sumaila and Walters (2005). A careful reading of that paper reveals that the same notation ( fgr ) is used for two related quantities: the future-

generational discount rate per generation and the future-generational discount rate per year.

1 Copyright © 2006 Elsevier B.V. All rights reserved. This paper was first published in Ecological Economics in 2006. We are very grateful to Professors Michael H. Prager and Kyle W. Shertzer for their permission to reprint these pages.

95

Clearly, those quantities can be the same only if the generation time G is exactly 1 year. That seems unlikely for even the most rapidly growing human populations.

To remedy this situation, we propose the following notation: fgR to denote the future-

generational discount rate per generation (i. e., the discount rate per G years) and fgr for the

corresponding rate per year. The two are related by the generation time G, as discussed below. Most often, Sumaila and Walters (2005) discuss annual rates, and we have verified

that their derivation of the intergenerational approach is correct when fgr is considered an

annual rate. However, we could not duplicate their Fig. 1 by taking the stated fgr =1%, 5%,

20% as annual rates. We were able to duplicate it by taking generational rates fgR =1%,

5%, 20% and converting to annual rates with the relationship fgr = fgr /G before using their

Eq. (3). Thus the intergenerational rates actually used in Fig. 1 of Sumaila and Walters (2005) are smaller–by a factor of G = 20–than one might think from the text.

The relationship fgr = fgr /G is exact when instantaneous rates are used. However, it is

an approximation when simple rates are used, because of the effects of annual compounding. From the definition by Sumaila and Walters (2005) of the discount factor as )1/(1= rd + , we surmise that annual com- pounding is assumed, as the discount factor under instanta- neous rates would be computed as d = re? . We also note the use of annual increments and simple

fractions throughout the paper. Under annual compounding, the exact expression for deriving an annual rate from the generational rate is most clearly expressed in terms of discount factors. We use notation ad for the annual discount factor and Gd for the generational (G-

year) discount factor. Then, )1/(1= fgG Rd + (45)

GGa dd = (46)

and aafg ddr )/(1= − (47)

In Fig. 1 of Sumaila and Walters (2005), the error resulting from the approximation is

relatively small, amounting to a bias of about -2.2% in NPV in the curve with fgR =20%. Still,

it seems better to use the accurate conversion. Once that technical issue has been addressed, the question of appropriate values for

fgR must be considered. Statements like “an easy way to fix a value for the future generation

discount rate is to make it equal to the standard discount rate” (in Section 5 of Sumaila and Walters, 2005) are ambiguous until the time basis of each rate is specified. That ambiguity is evident in their Fig. 1 with such claims as fgrr = = 5%. The numbers are not truly equal

when their units are different, and they are different, as we have shown. Sumaila and Walters (2005) suggest that appropriate values for fgR can be obtained

directly from surveys or indirectly from examining other social issues, such as education. More specifically, they propose using the internal rate of return for education to the PhD level, an approach adopted by Ainsworth and Sumaila (2005) for the intergenerational valuation of Atlantic cod (Gadus morhua). In that application, however, as in Fig. 1 of Sumaila and Walters (2005), it is unclear whether the applied rates correspond to annual or

96

generational rates. The answer to that question seems important to the argument that appropriate intergenerational rates can be chosen by examining the value put on education of future generations.

A possible criticism of intergenerational discounting is that future benefits are counted again for each generation that participates in them. This criticism, it seems to us, misses the point. Multiple generations do receive benefits, so each generation's share should be counted and added to the others. When fgr is small, however, the present formulation can lead to

discounted benefit in the future that exceeds the original value. To avoid this, we propose that in using intergenerational discounting, the procedure should be modified to cap at unity the weights used to discount net benefits (Eq. (3) of Sumaila and Walters, 2005).

In summary, the intergenerational discounting technique is a welcome solution to a persistent problem in applied resource management. We believe it to have considerable application, e.g., in choosing among recovery trajectories for depleted fish stocks. In applying the technique, one must be careful to distinguish between the future-generation discount rate per generation and the corresponding discount rate per year, a distinction not at all clear in Sumaila and Walters (2005). When using instantaneous rates, the conversion is simply

fgr = fgr /G; when using annual compounding, the conversion is described above. Choice of a

future-generation discount rate will still be subjective; it may best be illuminated by examining other social choices, rather than on theoretical grounds. History and biology reveal that humans will sacrifice present gains for the well-being of their offspring–and for future generations in general. On those grounds we can conclude clearly that fgR is much closer to

zero than it is to infinity or even to unity. The authors are grateful for the support of NOAA's National Marine Fisheries Service,

in particular the Southeast Fisheries Science Center and the NOAA Beaufort Laboratory. We thank D. Ahrenholz, J. Smith, P. Marraro, R. Parker, D. Squires, and J. Waters for reviewing the work. Opinions expressed do not necessarily reflect the opinions of NOAA or any other government agency, nor of the reviewers. The authors have implemented the intergenerational discounting paradigm in the S computer language. The resulting function is compatible with the open-source statistics language R and is believed compatible with the commercial version of the S language, S- PLUS2. The computer code is available by email request to either author.

References

[1] Ainsworth, C.H., Sumaila, U.R., 2005. Intergenerational valuation of fisheries resources can justify long-term conservation: a case study in Atlantic cod (Gadus morhua). Canadian Journal of Fisheries and Aquatic Sciences 62, 1104–1110. [2] Mace, P.M., 2001. A new role for MSY in single-species and ecosystem approaches to fisheries stock assessment and management. Fish and Fisheries 2, 2–32. [3] Prager, M.H., 1994. A suite of extensions to a nonequilibrium surplus-production model. Fishery Bulletin 92, 374–389. [4] Punt, A.E., 2001. The gospel of maximum sustainable yield in fisheries management:

2Mention of commercial or noncommercial products does not imply endorsement by NOAA, U.S. Department

of Commerce, or any other government agency. No such endorsement is made or implied.

97

birth, crucifixion and reincarnation. In: Reynolds, J.D., Mace, G.M., Redford, K.H., Robinson, J.G. (Eds.), Conservation of Exploited Species. Cambridge University Press, pp. 67–86. [5] Sumaila, U.R., Walters, C.J., 2005. Intergenerational discounting: a new intuitive approach. Ecological Economics 52, 135–142.

98

Making future generations count: Comment on “Remembering the future”1,2

U. Rashid Sumaila, Carl Walters

Prager and Shertzer made three points on our intergenerational discounting approach

presented in Sumaila and Walters (2005). First, they found the approach a welcome solution to a persistent problem in applied resource management, and that it provides a sensible way to integrate economics with policy considerations. Second, they provided a useful clarification of our use of notation to denote the future generation discount rate in the approach. Finally, they suggested that a cap should be incorporated into the discounting equation we introduced to make sure that the discount factor derived from our intergenerational discounting equation does not exceed 1. We comment on each of these in this short response to their comments.

1. Response to Prager and Shertzer We agree with Prager and Shertzer that if the major goal of economics is to quantify

human preferences, it must be the case that conventional discounting is illogical because society has expressed a strong preference for sustainability in resource use and management. Their syllogism is beautiful: Simply put, for conventional discounting to be logical, society will have to accept that emptying the ocean, cutting down all our forest, etc., is acceptable. This is because with conventional discounting there are practical levels of discount rates at which it would be economically non-optimal not to do so (Clark, 1973; Chichilnisky, 1996; Nordhaus, 1997; Weitzman, 2001).

The point made by Prager and Shertzer regarding our unclear use of notation for the intergenerational discount rate is well taken. Their clarification of the need to more clearly define and separate the intergenerational discount rate in terms of (i) fgR – the future-

generation discount rate per generation and (ii) fgr , denoting the future-generation discount

rate per year is very helpful, and will be beneficial to the reader and all those who want to apply the method.

Finally, the suggestion by Prager and Shertzer that there is a need to cap the discount factor to ensure that it does not exceed 1 is problematic because there are real-life situations in which a negative real discount rate, and hence a discount factor greater than 1, is practical and necessary. Negative real discount rates have actually been recorded in the past. The U. S. came close to facing a negative discount rate after 9/11. Also, Japan most probably faced a negative discount rate during its prolonged recession of recent times. One can therefore see

1 Copyright © 2006 Elsevier B.V. All rights reserved. This paper was first published in Ecological Economics in 2006. We are very grateful to Professors Rashid Sumaila and Carl Walters for their permission to reprint these pages.

2 [Authors’ acknowledgments] We are grateful to the Sea Around Us Project, a Partnership of the Pew Charitable Trusts, for support. Sumaila also acknowledges the support of Oceana and the European Community's Program for International Scientific Co-operation (INCO) through Contract 003739 for INCOFISH project.

99

situations in resource management where a negative discount rate and therefore a discount factor greater than one are reasonable. For instance, society may discover that a certain vital component of the natural environment is in such a bad state that it decides to `pull all the stops' to remedy the situation by any means necessary – for the benefit of both current (in the future) and future generations. Such a move can, in most cases, be justified only by an implicit assumption of a negative discount rate.

2. Conclusion We thank Prager and Shertzer for a useful commentary on our contribution. They

have, by their commentary, made our approach more accessible to colleagues. The authors have helped to make a method that promises to provide policy makers a useful tool that can help them meet society's strong preference to ensure the long term sustainability of our natural and environmental resources. In other words, they have contributed in helping make future generations count when it comes to resource management and policy making.

References

[1] Chichilnisky, G., 1996. An axiomatic approach to sustainable development. Social Choice and Welfare 13, 219–248.

[2] Clark, C.W., 1973. The economics of overexploitation. Science 181, 630–634.

[3] Nordhaus, W.D., 1997. Discounting in economics and climate change: an

editorial comment. Climatic Change 37, 315–328.

[4] Prager, M.H., Shertzer, K.W., 2006-this volume. Remembering the future: a commentary on “Intergenerational discounting: a new intuitive approach”. Ecological Economics. doi:10.1016/j. ecolecon.2006.01.003.

[5] Sumaila, U.R., Walters, C., 2005. Intergenerational discounting: a new intuitive

approach. Ecological Economics 52, 135–142. [6] Weitzman, M., 2001. Gamma discounting. American Economic Review 91,

260–271.

100

Part Two. Economic Concepts and Methodology

The second part of the book focuses on economic concepts and methodology. It presents the precautionary principle, discusses the concept of externalities and discusses the methodology of mean field games.

The first chapter, by Olivier Godard, is concerned with the precautionary principle, which has become a new standard for risk management in the fields of environmental protection, public health and food safety. The precautionary principle concerns the attitude to adopt in the face of “potential collective risks”. It refers both to potential dangers the existence of which is scientifically uncertain, and to dangers for which, despite their known existence, it is not yet possible rigorously to establish the likely extent of the damage they may cause. Thus scientific uncertainty underlies the precautionary principle, constituting a challenge not only for scientific expertise, but also for the traditional foundations of the legitimacy of public action.

The second chapter is devoted to externalities, that is to say, impacts on our well-being that the market system is unable to allocate in an optimal way. In 1937, Coase explained that the market stops where transaction costs are too high and that beyond that point, we enter the realm of externalities. Jan Keppler develops Coase’s insight. Contrary to Coase, however, he argues that the level of transaction costs cannot be taken as given. There is thus a place for selective and innovative government interventions.

In the third chapter, through an application to oil production, Olivier Guéant, Jean-Michel Lasry and Pierre-Louis Lions explain the new “functionalities” of mean field games, i.e. their capacity to introduce into the models the effects of social interactions between agents, which cannot be integrated into the framework of classical modeling in economics. This methodology is especially suited to economic problems characterized by the presence of a large number of agents. In its application to an oil production problem, the essential forward/backward character of mean field games operates in the production of a resource whose availability is limited. The backward dimension comes from the optimization of the production timing; the forward one is linked to the evolution of oil reserves.

101

The Precautionary Principle as a social norm

Olivier Godard

Abstract: Against the diversity of beliefs on the content of the precautionary principle (PP), a doctrine has been shaped by years in Europe and France. This papers aims at introducing this doctrine that insists on the requirement of proportionality of precautionary measures and refuses any general obligation of reversal of proof. It shows how the PP borrows from two extreme behavioral models: early eradication of potential threats at source and the fitness/maturity one that focuses on a patient management of sufficiently well-developed risks. It also explains what difference it makes saying the PP is a principle, i.e. neither a rule nor a decision-making criterion.

Key Words: Precautionary principle, Europe, scientific uncertainty, risk management, principles, decision theory

102

Over the last twenty years or so the precautionary principle (henceforth referred to as PP) has become a new standard for risk management in the fields of environmental protection, public health and food safety. The principle concerns the attitude to adopt in the face of ‘potential collective risks’. This expression refers both to potential dangers the existence of which is scientifically uncertain, being neither proven nor refuted, and to dangers for which, despite their known existence, it is not yet possible rigorously to establish, by logical analysis or by statistical observation of repeated events, the likely extent of the damage they may cause. Should such risks be left to scientific analysis and be ignored in terms of collective action, while awaiting a better understanding that at a certain point will allow us to have a rational approach based on calculation of the risks? This attitude prevailed before the 1980s, within a culture of prevention strictly based on scientific risk analysis and insurance mechanisms. The novelty of the PP is to bring these ill-known collective risks affecting the environment and human health into the domain of public intervention, especially in cases of threats of irreversible damage.

This process has occurred unevenly, accompanied by confusion and debate (Godard, 2006). In Europe at least, the PP is a doctrine that has been developed and refined, particularly since the pivotal year 2000, and ratified by national and EU institutions. Nevertheless, it is still subject to contradictory interpretations, not only between the institutional doctrine and what is claimed by some militant organizations, but also between the accepted doctrine and the concrete management of certain issues, as for example genetically modified organisms for agricultural use. These issues have been intensified in France with the recognition of the PP in the Environmental Charter, which was incorporated into the country’s constitution on 1 March 20051.

Scientific uncertainty forms the background to the PP, not uncertainty of contingent events of day-to-day life. It is a challenge not only for scientific expertise, but also for the traditional foundations of the legitimacy of public action, at least for the rational-legal form that has been at the core of the development of the welfare state in Europe. Of course individual agents may forge their own subjective estimation of probabilities, from a mixture of information and prior beliefs. Such an operation may trigger personal decisions or simply lead to the formation of opinion. This individual construction of belief is deprived of direct relevance as a basis for public action in regard to collective risk. Indeed the public arena of policy-making in democracies entails a basic requirement, that of justifying goals and actions to citizens. In the debate, policy-makers have to share information, diagnoses and arguments with the members of society. A symmetrical obligation to ascertain positions and proposals is addressed to the citizens, who in turn make various demands. On the whole, both the authorities and citizens have to justify changes and actions they propose. Here a scientific approach to the issues can play a key role, aiming at objective statements that everybody should consider as the best available 1 Article 5 of the Charter states: “When the causing of harm, although uncertain in terms of scientific knowledge, could seriously and irreversibly affect the environment, the public authorities, by applying the precautionary principle and in their areas of remit, should make sure that risk assessment procedures are implemented and provisional and proportionate measures are taken in order to ward off the damage”. (Unofficial translation from French).

103

representation of reality. In this way, a common world is defined on the basis of which public deliberation may take a rational form. When scientific uncertainty makes it impossible for citizens to share an objectively determined common world, difficulties arise for public deliberation and for the justification of policy proposals, unless there is an agreement throughout society on how to address situations and risks characterized by scientific uncertainty. This specific feature of public decision-making is ignored by risk decision models which privilege either the authorities’ beliefs and preferences or some aggregate of consumers’ individual preferences. Herein lies one of the main sources of the discrepancy and tension between the PP as a social norm and rational individual decision-making models in regard to risk2.

Box 1: What is a risk?

The use of the word ‘risk’ sometimes arouses terminological disputes. If risk is defined as “a function of the probability of an adverse health effect and the severity of that effect, consequential to a hazard”, as in article 3 of EU regulation 78/2002 on food law and food safety (EC, 2002), the expression ‘potential risks’ is not consistent, since it designates situations for which it is not possible to state an objectively and scientifically established probability. If risk is defined by the uncertain nature of the consequences of an action, where the unknown element may be expressed around a central value, the term ‘potential risks’ is also inappropriate, since it relates exclusively to the possibility of harm attributable to exposure to a danger. In fact usage and public debate have given a different meaning to risk than has decision-making theory, namely the potential danger or the threat of disaster. In this paper I shall follow current usage, since it is this which structures discussion of the application of the PP.

The impossibility of adopting an objectively and scientifically established probabilistic formulation leads to public action that takes account of other reference points: in addition to the procedures of scientific expertise there are the procedures of public debate and of dialogue with interested parties. The PP calls on judgment on potential risks on a case-by-case basis and therefore gives a critical role to case law. However, the PP does not aim to break with the general demands of reason, even if these must resign themselves to achieving what is reasonable by means other than scientific calculation.

The first objective followed in this chapter is to situate the PP between the two main pure models proposed for tackling risk management, namely the early eradication of the risk at source and the patient management of the risk when it matures and becomes well defined. This comparison will enable the EU doctrine of the PP to be introduced, as developed in the field of environmental protection. In so doing, sensitive questions will be addressed. How is the general

2 For a presentation of the economic approaches expressed within a Bayesian framework, see Kast and Lapied (2006) and Gollier (2004).

104

demand for proportionality to be understood in the context of precautionary policies? To whom is the principle addressed? Who has legal duty to implement it? Beforehand the notions of principle, rule and criterion need to be clarified, insofar as a number of criticisms directed at the PP are based on a misunderstanding of what it really is and are mistaken as what it aims to do.

Section 1 introduces key distinctions between principles, rules and criteria for decision making and stresses the often overlooked reality that the PP is a principle, not a criterion by its own. Section 2 gives an apercu of the intellectual history through which a European and French doctrine has been shaped. Section 3 posits the PP as an intermediary model between two ideal-types of risk management: the attempt of early eradicating hazards at source and the fitness/maturity model. Section 4 introduces the European and French doctrine with a focus on the application field, the proportionality requirement, expertise and evidence. The final sub-section is a plea against the thesis that all persons in Europe are legally committed to directly implement the PP whatever initiatives public authorities may take or not. Section 5 concludes.

1. Principles, rules and criteria

In law, the notion of a rule can have either a wide interpretation, referring to any normative element belonging to the legal system, or a restrictive interpretation for which the rule is distinguished from other categories of legal norms. Here we are interested in the restrictive sense: the rule precisely specifies a prohibited or prescribed behavior. Hence it has a concrete focus. Thus the highway code contains numerous rules consisting of both prohibitions (e.g. driving through a red light) and obligations (e.g. fastening one’s seat belt). A good rule is free of any ambiguity in terms of what is expected. Consequently the rule system is committed to logical consistency – i.e. the principle of non-contradiction – so as to avoid putting people in a position of having to satisfy two mandatory but contradictory requirements. Once these conditions are met, the rule system does not fully define behavior: in a liberal democratic society considerable room is left for people’s choices and initiatives within the area delimited by what is prescribed and what is forbidden.

A principle is an abstract statement that seeks to provide guidance for the resolution of a concrete problem. Its useful effect cannot be obtained without taking account of the effect of other principles and of the particular circumstances of the problem. The principle therefore must be both interpreted and given concrete expression in specific applications. Because of this dependence, the principle does not dictate a specific behavior, nor does it forbid it, just as its role is not to specify all possible actions. Whereas it is not possible to have contradictory rules within the same legal system, different principles can come into conflict in terms of the orientations they provide in regard to action. For example, in EU law, the principle of free circulation of goods across borders of member states and the PP may be mutually opposed. It is then up to economic agents and citizens, as well as the relevant legal authorities, to find the best concrete ways, appropriate to a given situation, of harmonizing the requirements arising from the various principles applicable. In general, regulation by principles runs into a casuistical approach to

105

situations and problems. The judgments thereby arising should be considered as singular and attached to the context in which they have been made, without being able to be mechanically transposed into other situations. Nevertheless, similar cases can serve as comparative reference points for understanding new situations.

A decision-making criterion is different from a rule or a principle. It presupposes a choice situation in which a decision-maker is faced by a range of options for behaving or acting in different ways. The idea of choice implies that the rule system does not fully determine the decision-maker’s behavior. The purpose of the criterion is to evaluate and hierarchize all possible actions from the viewpoint of a given decision-maker. The criterion attaches itself to a value (for example, rationality, efficiency, justice, etc.) and gives it an operational way of being calculated (for example, maximizing the present value of net benefits, for economic efficiency). Ideally the criterion enables a complete ordering of all possible actions to be specified, which does not rule out two actions, judged as equivalent, receiving the same classification.

Once the preceding distinctions have been agreed, the most commonly made criticisms of principles – that they are vague, are open to different interpretations, and do not have the clarity of a rule or the clearness of ordering of a criterion – will appear as devoid of relevance, reflecting only the misunderstanding of those who formulate them. Let us therefore emphasize that the PP is a principle, not a decision-making criterion. It is therefore pointless to criticize it for not being a criterion or to try and make it a standard criterion.

The role and significance of principles remain to be considered. Why not be satisfied with simply having rules and criteria? The most obvious answer is that the purpose of principles is to provide a basis for rules and criteria by giving them a more general foundation which articulates basic social values and legitimate forms of engaging in action. Situated beyond particular circumstances, principles enable precise rules and the admissibility of criteria to be decided. They establish a meaning that clarifies existing or proposed institutional procedures. They are a reference point in a coordination process that must accommodate numerous transformations and wide empirical diversity. Most importantly, they are a means of inspiring change in an existing normative system. This is one of Nicolas de Sadeleer’s (2003) conclusions in regard to the principles of environmental law: principles primarily exist to create a dynamic; by appealing to change, they are inscribed within a dimension that is more strategic than tactical, more reforming than stabilizing. In doing so, principles do not remain immutable: they themselves are affected by the change they set in motion.

2. Landmarks in a thirty-year history

Explicit formulation of a PP goes to the 1970s in Germany, where one of the main political principles inspiring public action in relation to environmental protection was the Vorsorgeprinzip (Boehmer-Christiansen, 1994). In the name of concern for the future, this principle recognizes the need and the legitimacy of not waiting for the stage of scientific certainty on causal relationships involved before undertaking preventive action aimed at

106

environmental threats. This principle also asserts the need to engage in a medium and long-term, continuous and adaptive approach to action for safeguarding the environment, taking advantage of technological progress to introduce more stringent environmental requirements.

These two ideas were subsequently separated. The second became associated with the different currents converging under the umbrella of sustainable development, while the first was reinforced under the names of the “precautionary approach” and the “precautionary principle” through various pronouncements in international law. Particular mention should be made of 1992, with on the one hand all the major statements issuing from the Earth Summit in Rio de Janeiro in June of that year and, on the other hand, the adoption in Europe of the Maastricht Treaty, which created the European Union. It was with this latter document, which mentions it as one of the pillars of environmental protection policies, though without defining it, that the PP became a key legal concept within EU law.

Several dates mark out the consolidation of this principle in French and EU law. The French law 95-101 for the strengthening of environmental protection, the so-called Barnier law, defined the principles that should inspire actions for protecting and enhancing the environment “within the framework of laws which define its scope”. Among these principles is the PP, defined as the principle according to which “the absence of certainty, taking account of current scientific and technical knowledge, should not postpone the adoption of effective and proportionate measures aimed at averting the risk of serious and irreversible damage to the environment, at an economically acceptable cost”.

In 1998, the French Council of State – the higher level court for matters involving the state –, in its public report, made known its thinking on health law. In particular the authors questioned whether there were good grounds to consider the PP, which was strangely presented as creating the obligation to provide proof of the absence of risk3, as a new basis for medical liability. Unsurprisingly, they gave a negative answer to this question, secured by such an unacceptable understanding of the principle (Conseil d’Etat, 1998).

In May of the same year, the European Court of Justice gave its judgment on the disagreement between the British government and the European Commission in regard to the embargo measures taken in March 1996 against beef products coming from the UK, in the context of the fight against the spread of the BSE epidemic. This judgment in substance confirmed the PP, reckoning that the authorities were justified in taking health measures without waiting to have scientific certainty about causal links and the extent of damage, and thereby gave the PP an autonomous legal value in an area other than that of the environment, namely food and health safety (ECJ, 1998; Gonzalez Vaqué et al., 1999).

Following a request from the European Council, in February 2000 the Commission published a communication on the PP, which expounded its doctrine drawn up after an open

3 On page 255 the report states: ”This new concept is defined by the obligation on public or private decision-makers either to take action or to refrain from doing so according to the possible risk. In this sense, it is not sufficient to limit their conduct to taking known risks into account. They should, in addition, provide proof, considering the current state of science, of the absence of risk.”

107

consultation process (European Commission, 2000). On this basis, a “Resolution on the precautionary principle” was adopted by European heads of state and governments at the Nice summit in December 2000 (European Council, 2000). This resolution confirmed the proportionate conception already embodied in the Barnier law in France.

In 2002, the European regulation 178/2002 on food safety was adopted and the European Food Safety Authority was set up. With direct application to the national territory of all member states, this regulation explicitly made the PP one of the pillars of food safety and drew from it obligations for public institutions as well as for companies in the agri-food industry (European Council, 2002).

In June 2003, the French government, at the invitation of the French president, introduced a proposal for a constitutional law aimed at adding an Environmental Charter to the French Constitution. Adopted successively in the same terms by the National Assembly and the Senate in June 2004, this charter was solemnly incorporated into the Constitution in March 2005 following a congressional vote (Godard, 2006). It refers in Article 5 to the obligations of the public authorities in implementing the PP, thereby confirming that they are in the front line for applying the principle. The implementation of risk assessment procedures becomes a strict obligation. Similarly, emphasis is placed on the requirement for the proportionality of measures and taking precautionary action on a provisional basis.

3. Two opposed risk-management models

There are two diametrically opposed pure models for managing risk (Godard, 2007). The first aims at early eradication, by targeting the risk at source, in order to eliminate it as soon as possible. The second model bides its time, preferring to have reliable knowledge of risk before taking selective and appropriate action based on fully mature knowledge.

The problem raised by both approaches is that of defining effective, well thought-out prevention. Is this possible if done early? Is it possible if done when the risk matures? Are we not faced by an impossible choice between early action that is largely ineffective and very costly, because risky or even blind, and late action that is targeted and potentially much more effective, but liable to fail because confronted by now irreversible phenomena that cannot be controlled for want of timely action? Is it advisable to wait for scientific certainties before acting or should we tackle the risk as soon as its eventuality is recognized?

The two pure models that we have at our disposal are opposed in every way. The first, defended by the German philosopher Hans Jonas, is based on a moral imperative of eliminating any potentially apocalyptic risk, whatever the probability of its happening. The second corresponds to the standard model of rational action in the area of prevention, but encounters a negative echo in several unhappy historical experiences characterized by late action and inertia that the courts have subsequently come to view as culpable. The asbestos issue is a perfect example of this.

108

Let us now further expound these two models and try to find the best possible path for managing risk.

3.1. The early action model aimed at eradicating risk

3.1.1. Hans Jonas’s Responsibility Principle

The grave threat that the power of modern technology poses for the future of humanity, either through the destruction of the world that harbors human life (an environmental crisis or nuclear catastrophe) or through calling into question that which morally constitutes our human essence, is the point of departure for Hans Jonas’s thinking (Jonas, 1984). To this he adds an axiom: a new technology’s course of development is really only controllable at the outset; subsequently the technology acquires its own autonomy as it spreads through society and becomes virtually unmanageable. Early monitoring and action are therefore needed to safeguard the future of humanity from anticipated catastrophes that would destroy the human project. Hence Jonas’s proposal to envisage the development in our secularized societies of “a heuristic of fear”, as a complement to scientific forecasting of the impact of technology, forecasting moreover that he insists should be further developed. Indeed, whatever the research effort, the reality of the ultimate consequences of technical development seem to him always to exceed what science will be able to anticipate. The contribution of science to risk management must thus be complemented by an approach to the future that mobilizes imagination and sensitivity in the service of identifying the threat and taking it seriously, despite the fact the “the future has no lobby”. In this respect moral fear brings vigilance and triggers action; it stems from a thought-out moral position.

On the basis of such premises, Jonas’s analysis makes apocalyptic risk absolutely unacceptable; it is a risk that must be eliminated, whatever the probabilities involved and whatever the expected benefits of the technologies that cause it. From the standpoint of conceiving and gauging what action to take, the mere possibility of an apocalyptic risk should be considered as a certainty, Jonas tells us.

Jonas limits this maxim to apocalyptic risks liable to affect humanity as such. It should not be applied to other risks, which can continue to be taken according to the usual standards, which are compatible with a certain element of risk. The practical relevance of the maxim therefore entirely depends on the capacity of risk managers to ascertain, early on, which risks are potentially apocalyptic and which are not. In the event of doubt, a risk should be treated as potentially apocalyptic and this possibility should be treated as a certainty. Yet – and here there arises an inner contradiction within Jonas’s thinking (Godard et al., 2002; Godard, 2003) – the incompleteness of knowledge, the structural incapacity of scientific knowledge to embrace the full reality and spread of the technology’s effects, on which Jonas bases the argument for his “heuristics of fear”, means that the possibility of an apocalyptic outcome can in fact never be ruled out, for Jonas excludes any reference to the idea of probability. With every non-disqualified possibility needing to be viewed as a certainty, all the risks associated with a technology must

109

finally be treated as apocalyptic risks in a logic of early eradication. All new technologies must be stopped at their outset. Against his intent, the application of Jonas’s maxim would lead a paralysis of technological development. Even if we can accept many of his premises, we have to say that the action model he puts forward leads to a logical and practical impasse.

If referring to apocalyptic risks seems to limit the scope of the discussion, we should point out that replacing the word “apocalyptic” by “serious and unacceptable” leads to the same problem and the same impasse. In the absence of definitive scientific certainty, any idea of distinguishing two approaches to risk – those deemed unacceptable being a matter for elimination or radical refusal to take the risk, while others deemed acceptable would be approached in a way accepting the very existence of the risk – will be affected by the same aporia. Furthermore, Jonas’s radical conception gives rise a false sense of security. The paralysis of innovation resulting from applying it necessarily leads to making humanity wholly dependent on existing technologies, whereas many of them underlie Jonas’s moral disquiet and are the source of the present environmental crisis.

The early elimination model thus cannot claim to embody the reflexive management of risk that is needed, whether it results from a conscious moral attitude or is the involuntary outcome of an unreflected choice of method such as that which gives rise to a specific artifact systematically overemphasizing the seriousness of any risk, up to the point of triggering an overall forbidding of any new technology.

3.1.2. The earliness-seriousness artifact

Let us return to the problem of the management of the prevention time schedule by placing it within the time frame of progressively maturing scientific knowledge, and let us reveal the catastrophist bias of a certain type of risk assessment, that is apparently supported by ethical common sense yet is nevertheless profoundly inadequate. This is the case when an artifact occurs (Godard, 2003). It results from reversed interaction between the earliness of taking the risk into account and its supposed seriousness: the initial justification of the PP is that because of seriousness of possible damage resulting from suspected risks, early preventative action should be taken; the reversed and perverse relation would be that because of the earliness of consideration, any risk is suspected of having potentially apocalyptic consequences and this possibility catches becomes the focus policy-makers’ attention. Why is this so? When free rein is given to this artifact, whatever the empirically estimated risk, the greater the possibility that a risk is taken into account early on in the scientific time frame, the greater the number of possibilities for producing damage that are not yet excluded by science, the more the latter will include within them catastrophic possibilities that will seize people’s attention, the greater the risk associated with the technology under consideration will appear dangerous, and the more this risk will call for severe and restrictive measures aiming at its pure and simple elimination. One then ends up with the same aporetic result as with Jonas’s maxim.

The skewed assessment procedure giving rise to these perverse effects rests on two axioms which, taken individually, may seem reasonable enough. The first requires that public

110

appraisal of collective risk considers only the possible damage from new technologies, without concerning itself about their possible benefits. The underlying “common sense” is that the principle of business freedom assigns to the entrepreneur, and to him alone, the role of appraising the advantages of business projects or technical innovations, with public responsibility being limited to making sure that these do not create the risk of unacceptable damage for the community4. The second axiom basically takes up Jonas’s position, in requiring, in the name of the future, that a non-excluded possibility be treated as a certainty. It only needs these two axioms to be combined to end up with the earliness-seriousness artifact to deleterious effect.

To be convinced that this artifact does actually occur, one only has to look at what is currently said about the risks associated with biotechnology and nanotechnology, in particular the apocalyptic visions accompanying many of the scenarios contemplated. According to Jonas’s maxim, the possible strongest measures should be taken to neutralize such risks at their source: all research in nanotechnology and its various applications should be halted, due to the seriousness of hypothesized dangers associated with it. The same thing could be said of any other technology whose effects are still uncertain. It is difficult to subscribe to such a radical conclusion.

There is in fact an intellectually straightforward solution: this involves reversing the content of the two axioms causing the artifact. First, risk assessment should also take into account the various social benefits of technical developments, without confining itself to potential damage. Next, it is essential to hierarchize risk hypotheses according to their degree of plausibility and scientific consistency: a hypothesis lacking any confirmation cannot lead to the same level of response as a hypothesis strongly supported by observations, experiments and theoretical models. In this way, we would abandon the early elimination model in favor of the PP.

3.2. The fitness/ maturity model

The second risk management model is the fitness/ maturity model. Here the risk is met by prevention only when the level of acquired knowledge is sufficient to escape scientific uncertainty and to distinguish between established risks and unconfirmed suspicions, as well as between dangerous applications and harmless ones. This model means waiting for the moment when the framing for action on risk has become sufficiently clear to find a rational approach and allow calculations. The “good reasons” supporting this model of deferred action partake of a general concern for the efficacy of action. They also result from particular circumstances. Thus when the available knowledge does not allow a sufficiently precise early identification of the risk involved and its cause, early action is obliged to strike at random or to tackle the multiple 4 This principle does not apply in the area of health technologies and products, where on the contrary risk-benefit analysis has been adopted by medical circles as the assessment framework. But in that case, benefits as well as risks of damage are related to health impacts and do not directly take into account considerations of pharmaceutical companies’ commercial benefits.

111

potential risk factors, to the point of questioning the credibility of the action in the eyes of those who have to provide it with their support. Similarly when in practice the benefits gained and the damage incurred cannot be distinguished at an early stage, trying to avoid the latter would lead to completely abandoning the former. It may also be that the available knowledge is too incomplete to be able to establish any balance between the benefits and the damage.

The viability of this model nevertheless presupposes draconian conditions. Throughout the development of a technology, from the initial research through to the commercial distribution of various applications, public risk managers have to retain continuous control of its development. This presupposes two conditions: the real possibility of at any time stopping the technical development of the source of the risk and the ability to contain and then neutralize the residual effects attributable to the development phase before it was interrupted. This model is ineffective in the face of a major irreversibility, or when the damage engendered does not become apparent for a long time. In such cases, action deferred to the maturing of the risk becomes powerless, making administrators’ newly acquired clarity on the exact profile of the risks ineffective.

Experience has shown that this model too can be affected by a fatal paradox, perfectly illustrated by the asbestos affair. If the effects are postponed for a long time, when the health threat does become apparent, it is generally attributable to a past economic, technological, regulatory and therefore health situation. The damage perceived does not then provide a reasonable basis to call into question current technologies and systems, which are not those which created these risks – unless one gives into retrospective illusion or to blurring the distinction between old but no longer operative causes and current conditions. Since the model is ineffective for identifying the risk at its source, policy-makers tend to abandon the prevention paradigm in favor of repairing that part of damage that can be repaired, offering reparation to victims or adapting to a new and unavoidable state of the world. Nor does this model allow the future potential risks of existing and new technologies to be prevented, since such risks, that will manifest themselves too far in the future, cannot yet be empirically demonstrated and scientifically characterized. This model is therefore ultimately ineffective both for neutralizing the risk when it becomes apparent, since such risk stems from a superseded past, and for avoiding future risk, insofar as this is not yet sufficiently registered in reality and cannot simply be assimilated in any reasonable way to the past risk. In the case of asbestos, the logic of continuously improving working conditions in industry has considerably changed workers’ exposure compared to the practices in force several decades previously. Further, throughout the economic use of this substance, there remained numerous scientific uncertainties, for example on the comparative effects of different types of asbestos, a situation that has too easily led managers to the idea that not enough was known about it to decide on effective preventive action, that would be finely tuned to the characterization of risk. For this type of deferred effect, the logic of continuous improvement can act as an invisible cognitive obstacle to effective risk prevention.

3.3. The precautionary principle as an intermediate model

112

Between these two diametrically opposed models, each of which comes up against serious difficulties, there exists a third way for the prevention of potential technology-associated risks. Containing certain features of each basic model, this is the PP way.

The PP retains the idea of taking early account of risk hypotheses in order to get round the obstacle of the irreversibility of the fait accompli or deferred effects, but will drop the aim of eradication of risk, which would turn out to be an arbitrary source of blocking technological innovation.

The well thought-out prevention of risks that the PP promotes is compatible with the cautious engaging in activities suspected of creating health or environmental risks, even when such risks are little understood. However, this early account, which does not rule out engaging in potential risk activity, comes with a threefold caveat.

On the one hand, such engagement should be undertaken with due care and be accompanied by vigilant follow-up of the activities and techniques concerned in order to identify as quickly as possible the first signs of harmful phenomena or, better, ascertain in what ways the risks could materialize. On the other hand, the PP does not lead to a complacent acceptance of initial scientific uncertainty. The early consideration of risk hypotheses should be supplemented by specific programs of scientific research organized to reduce and possibly eliminate this uncertainty. In this, the PP is different from the sort of instrumentalization of scientific uncertainty in the service of the interests of actors or various causes that can be observed each time actors do not genuinely try and solve problems but use them to improve social positions of influence or power. Finally, the PP is about placing the early prevention of technological risks within an adaptive perspective of periodical revision of the measures taken in order to adjust them to new information arising from feedback from experience, monitoring devices and research programs. On the last point, the greatest attention should be paid to the leading edge of research and not only to accepted findings and established knowledge, in order to enable risk management to work on new, still unconfirmed research hypotheses and to sustain a future-oriented approach prospecting the effects of technological development. Only such projecting into the future would enable risk managers not to find themselves trapped by deferred effects and allow them to set up prevention measures without waiting for the onset of damage.

Maintaining a distance both from catastrophism and from casualness or a wait-and-see approach, the intermediary risk management model that the PP embodies looks capable of providing a proportionate response to the challenges that the risks created by new technologies give rise to. Rejecting the sharp opposition between all-out early eradication and late, totally rationalized and selective action when the risk matures, it presupposes the adoption of a regime of risk management which accompanies technological expansion in its various stages in a continuous and adaptive way, without refraining, if need be, from firmly halting certain paths of development in view of the threats they present.

4. European and french doctrine

113

The development of international, EU and French doctrine on the PP has benefited from work provided by researchers and experts on risk assessment. At a European level, there are two main reference texts: the Communication on the Precautionary Principle, presented by the European Commission in February 2000 and the Resolution on the Precautionary Principle which was adopted by European heads of state at the Nice Summit in December 2000. Admittedly, these texts have no direct legal force, but their political ratification at the highest level makes them benchmark texts. The Resolution is summarized in Box 2. In France, a noteworthy report by Philippe Kourilsky (Institut Pasteur) and Geneviève Viney (Université de Paris 1) (Kourilsky & Viney, 2000) was commissioned by the French prime minister at the time (1998).

4.1. Application field and schedule

A number of points in the European doctrine should be emphasized. The area of application of the principle is made explicit: the environment, of course, but also animal and plant sanitation and human health and food. The PP applies both to the Commission’s policies and actions and to those of the member states. It is the public authorities who are primarily and directly concerned. Indeed, they are responsible for organizing scientific assessment and defining the appropriate research framework. They are entrusted with the responsibility of politically estimating the level of protection to be provided, and therefore of defining acceptable risk, but within the limits laid down by the principles of proportionality and of consistency with the measures adopted for other similar situations. It is incumbent upon these authorities to follow up the implementation of these measures and to monitor any further research required5.

The variants found in national and international texts of political and legal import all have one key notion in common: the authorities should not wait until they have scientific certainty as to the existence of a danger and the scale of damage it may cause before coming up with a risk hypothesis and adopting measures to find out more about the danger and prevent it or limit its impact. In other words, the specific content of the PP is to insist on responding, in the current scientific context, to potential or hypothetical risks that could be serious or irreversible. Above all, the PP governs the time schedule of action in relation to risks; it requires that risk be taken into account early on, and firmly rules out the pretext of scientific uncertainty as a justification for delay (EEA, 2001). However, determining the possibility of damage must be based on relevant and serious information. Ignorance or the simple presupposition that “effects could be produced” cannot be sufficient.

5 The Agreement on sanitary and phytosanitary measures (SPS), concluded within the framework of the Marrakesh Agreement which created the World Trade Organization in 1994, creates a legal obligation, for a state, to support research programs in order to overcome scientific uncertainty in a reasonable delay, as a counterpart of the right of this state to take measures to close its borders to a suspect product in the light of as yet not fully established indications (Noiville, 2000; Belvèze, 2003).

114

Box 2: The resolution on the precautionary principle adopted by the Nice European Summit (2000)6

1. The PP is applicable to the environment and human health, as well as to animal and plant sanitation;

it is set within a sustainable development perspective.

2. The PP applies to the policies and actions of the EU and its member states; it concerns action by the

public authorities, both at an EU level and a member state level.

3. There are grounds for resorting to the PP if the possibility of harmful effects on health or the

environment is identified and if the preliminary scientific assessment on the basis of the available

facts does not enable the risk level to be ascertained with certainty.

4. To proceed to an assessment of the risks, the public authority should have an appropriate research

framework, drawing in particular on scientific committees and relevant scientific work; the public

authority is responsible for organizing risk assessment, which should be conducted independently and

transparently on a multidisciplinary basis; open debate and adversarial procedures should be

enhanced.

5. The assessment of the risk should take note explicitly of possible minority views.

6. Risk management measures should be taken by the public authorities responsible on the basis of a

political judgment as to the level of protection sought.

7. When choosing measures to be taken for the management of the risk, the whole spectrum of measures

able to achieve the desired protection level should be considered.

8. Civil society should be implicated and particular attention should be paid to consulting all interested

parties at the earliest possible stage.

9. The measures taken should respect the principle of proportionality by taking into account short and

long-term risks and by aiming at the high level of protection sought.

10. When there are several ways of obtaining the same level of health or environmental protection, the

least restrictive measures for trade should be sought.

11. Measures should be consistent with the measures already taken in similar situations or using similar

approaches, taking account of the most recent scientific developments and the evolving level of

protection sought.

12. Measures taken on the basis of the PP should be reexamined in the light of the development of

scientific knowledge. To this end, follow-up of the effects of decisions should be implemented and

further research carried out to reduce the level of uncertainty.

4.2. Proportionality and evidence

6 Extract from Godard et al. (2002, pp. 122-123).

115

Proportionality is one of the central ideas of the conception developed. In EU law, it includes three key requirements as to the measures admissible (Gonzales Vaqué et al., 1999) :

- the measure should be appropriate to its objective (notion of efficacy);

- the measure should be necessary in the sense that that there are no other measures which are equally effective and less restrictive for international trade;

- there should be a reasonable connection between the objective sought and the constraints imposed on the free circulation of goods.

Naturally, within a national context, the notion of proportionality is supplemented by dimensions other than solely that of trading goods. It should be extended to the examination of all costs and benefits of the technique under consideration and the seriousness of the damage incurred. It should also consider the scientific plausibility of the risk hypotheses, which it is advisable to assess, in the absence of reliable probabilities, on the basis of the scale and richness of the scientific work mobilized in favor of the hypotheses under consideration (Godard, 2003).

Whatever certain analysts may have said, the idea of reversing the burden of proof is not part of the PP doctrine adopted in Europe by the authorities and positive law7. In its communication on the PP, the European Commission (2000) refused, for example, to adopt a general rule as to the imputation of the burden of scientific assessment of products and processes liable to present risks; it considered that either solution could be used depending on the situation. For example, for food additives, new foods, or drugs, the existing procedures of administrative authorization prior to being put on the market should accommodate the PP; since the weight of scientific studies supplying the authorization dossier is already the responsibility of the innovation promoters, the PP does not alter the pre-existing rule. For already authorized products and processes that are on the market and widely used, it is the responsibility rather of third parties and the public authorities to provide the scientific evidence needed to identify the disputed potential risks and to establish their plausibility before the authorities can take restrictive measures. The Commission thus reserves for itself the right to settle this question on a case-by-case basis.

In fact, in the context of the PP, the idea of reversing the burden of proof plays on an ambiguity that conceals a real shift. Strictly speaking, it is logically impossible to provide prior proof of the harmlessness of a product, substance or technique, when the scientific knowledge involved is neither complete nor fixed. Numerous authors have emphasized this impossibility, which reflects existing scientific uncertainties and the intrinsic incompleteness of knowledge. Hence the real point of the PP does not consist in reversing the burden of proof but to take its distance in regard to the concept of scientific proof, whether this be burdened – harmlessness must be proven – or disburdened – the existence of damage must be proven.

7 The Wingspread conference, held in the United States in January 1998, agreed on a declaration making the PP a principle reversing the burden of proof. This declaration has since marked public debate in that country and strongly contributed to its rejection by the public authorities and industrial and medical circles. See Raffensperger and Tickner (1998).

116

Many authors who speak of the reversal of the burden of proof are primarily thinking not of proof of the absence of risk, but of setting up prior authorization procedures, calling for the preparation of a dossier by the applicant. Even though this dossier would have to consider unknown risks, it naturally would not be able to do so by way of proving the absence of risks.

Moreover, the conditions needed to ensure the credibility and transparency of scientific expertise can be strongly opposed to the idea of assigning to the producers alone the task of preparing the scientific dossier on the project or technique they want to promote (Godard, 2005). Even if they are competent to do this, they are not best placed to ensure the credibility and transparency of the information they transmit. Recent institutional reforms affecting expertise in the area of health and food risks have thus assigned to “independent” agencies and committees the task of implementing said expertise. Risk expertise should be carried out according to precise multidisciplinary rules, independence, transparency and the principles of adversarial examination and open debate; the committees should make public, if need be, minority opinions alongside majority opinions.

All this argues in favor establishing a clear distinction between assigning operational responsibility to scientific expertise and the attribution of the corresponding financial burden: risk expertise should be given within an institutional framework at a good distance from risk-creating agents, while the financing of such work can be made the responsibility of companies benefiting from the activities giving rise to the suspected risks.

4.3. Liability rules

One of the liveliest issues in regard to the PP, and one of the most confused, is the question of who is responsible for implementing it. Two positions have been defended in the past few years. For the first, only the public authorities at various levels have the right and/or the obligation to apply the PP when the requisite conditions are present. There is no longer any controversy about whether the public authorities may have recourse to the PP. But what of other agents? For the second position, held by some legal specialists (Boy, 1999), both from a normative standpoint and from the standpoint of EU and French positive law, the PP would be of direct and universal application, that is, it should be applied, without any other mediation, by any public or private individual, particularly doctors, elected local officials and heads of companies. If this is confirmed by positive law, it would have significant consequences for the exercise of people’s civil liability: there can only be responsibilities if sanctions are applied in the event of their not being met. For example, one could, according to this position, bring a lawsuit against an industrialist for not halting a process or withdrawing a product that certain people or groups consider to be of potential danger, even though the industrialist is in regulatory compliance, no damage has (yet) been caused, and scientific doubts remain as to the reality of the risks.

Up until now (2008), this latter position has not been confirmed by positive law. Nor does it conform to the doctrine ratified by the highest European political authorities (cf. the Nice Summit Resolution). In France, the constitutional charter on the environment clearly states that it

117

is the public authorities who, in applying the PP, monitor risk assessment and the adoption of proportionate measures. The very term “monitor” clearly indicates that the public authorities are not the only agents involved in the implementation of the PP, but that they have the constitutional responsibility to establish a “precautionary state” that assigns appropriate roles to all those who contribute to the creation of potential risks. Ultimately, the implementation of the PP will concern all creators of risk, but within a framework organized by the public authorities and not in its absence.

On normative grounds, four main reasons may be put forward to challenge the relevance of the direct and universal application of the PP position: legal uncertainty; collective inefficiency; the removal of politics from risk management; and the unsuitability of those supposedly in position of responsibility for the type of action to be carried out.

4.3.1. Legal uncertainty

Changing the PP into a direct source of obligation for everyone without such obligations being precisely specified in advance by the public authorities, and leaving it to the courts to say subsequently what they were, would make it a source of legal uncertainty, which would be particularly harmful to businesses. This danger stems from the fact, already emphasized, that the PP is a matter of interpretation and not of strict attributes of a rule to exact specifications. It is for this reason that the PP could be accused of becoming a break on innovation and causing the spread of an overcautious attitude on the part of investors and entrepreneurs toward anything new.

4.3.2. Ineffectiveness from the standpoint of safety, costs and incentives

When it comes to fending off collective risks, implementation by directly applying the PP would require everyone who is not a public authority to interpret the demands of the PP, even though economic agents and private individuals have a limited cognitive ability as regards the collective risks affecting the environment or the health of people who may be very remote. With collective risks, the asymmetries of information tend to reverse themselves for at least one part of the relevant information: the producers of the risk may be much less aware of it than the public authorities. In the absence of any precise public coordination framework, very different opinions would arise – different people do not share the same attitudes toward risk and the collective interest –, even though collective security depends on the consistency of the measures taken by everyone acting on behalf of the same objective. The result would be ineffective risk prevention and wastage of effort.

4.3.3. The political moment short-circuited

118

The third reason is that direct application would result in short-circuiting the strictly political moment in estimating the various aspects in play, which should lead to the choice of a package of measures. It is each person’s anticipation of the possible legal effects of his actions more than the pursuit of the common good that would tend to inspire decision-makers8. Yet the strictly political moment in formulating and organizing the PP is vital for the management of collective health and environmental risks.

Seeking and providing security of citizens in the face of risk is a cornerstone of the state. The legitimacy of the state in the eyes of its citizens depends on its capacity to respond to people’s expectations, hence a direct link with the idea of political sovereignty as the condition for the exercise of responsibility by the state in front of its citizens. Direct application subsequently sanctioned by the court would, on the contrary, give excessive power to judges in the area of risk management, without their being responsible to anyone.

Next, one of the issues of the PP is the development of new forms of interaction between the public, experts and decision-making authorities, so as progressively to allay suspicion about the conditions of exercising expertise and to allow wide-ranging discussion on the taking of risks to which the public can give its consent. In the historical context of France and Europe in this first decade of the new century, the aim of these procedures is twofold: to introduce a reasonable system of managing serious potential risks, the existence of which has not however been established; and to create the conditions for restoring confidence in the capacity of experts and the authorities to set up collective risk management that answers to the expectations of citizens.

4.3.4. A gap between actors and actions

Within the perspective of a proportionate and procedural PP, most of the actions to be undertaken by their very nature involve a public responsibility: taking measures on the country’s borders, setting up scientific research programs on collective issues, organizing health and environmental monitoring, organizing citizens’ debate and consultation, regulating, encouraging, controlling, and so on. Is it not then a category mistake to suppose that all agents (business, private individuals, etc.) could directly implement this type of action in the name of their new responsibility in regard to risks?

Conclusion

8 See the modeling proposed by Chevé and Congar (2003) of the problem of a politician who wants to avoid his responsibility being called into question at some future point.

119

The PP does not lead to a break in the link between public action and science or reason. The proper responsibility and the dignity of politicians do not lie there. On the contrary, politicians should show themselves to be more attentive to what goes on in science so as to adjust action to the production of knowledge, without abandoning their capacity for judgment. Nevertheless, control of potential risks by the entire society implies a change of public reason itself, in favor of one that is less certain of its scientific and rational methods, and more open to wide-scale organized debate within society. This requires opening up the expertise process to citizens in ways that must avoid the confusion of roles and competences, and the involvement of citizens in discussion for enhancing a reasonable management of the risks which society can agree to. There is a delicate balance to be found between risk management and public opinion management. Recent French experience, both in regard to the mad cow affair (Godard, 2001) and the GMO issue (Godard, 2008), tends to show that this equilibrium has not yet been found, so much does the PP arouse the temptation to take public opinion management for risk management and to make the category of potential risks the wild card of a management of problems that is more petty political than policy-focused.

References

[1] Belvèze H. (2003), “Le principe de précaution et ses implications juridiques dans le domaine de la sécurité sanitaire des aliments”, Revue scientifique et technique de l’Office international des Épizooties, 22(2), pp. 387-396.

[2] Boehmer-Christiansen S. (1994), “Chapter 2. The Precautionary Principle in Germany – Enabling Government”, in T. O’Riordan and J. Cameron (eds), Interpreting the Precautionary Principle. London, Earthscan, pp. 31-60.

[3] Boy, L. (1999), “La nature juridique du principe de précaution”, Natures, Sciences, Sociétés, 7(3), pp. 5-11.

[4] Chevé M and R. Congar (2003), “La gestion des risques environnementaux en présence d’incertitudes et de controverses scientifiques – Une interprétation du principe de précaution”, Revue économique, 54(6), novembre, pp. 1335-1352

[5] Conseil d’État (1998), “Réflexions sur le droit de la santé”, Rapport public 1998. Paris, La Documentation française, (Coll. ‘Études et documents n°49’), pp. 227-501.

[6] European Commission (2000), Communication on the precautionary principle, Brussels, 2 February, COM(2000)1.

[7] European Council (2000), “Resolution of the Council on the Precautionary Principle”, In : Conclusions of the Presidency, Nice European Council, 7-9 December, Nice. Annex III.

[8] European Council (2002), “Regulation (EC) No 178/2002 of the European Parliament and of the Council of 28 January2002 laying down the general principles and requirements of food law, establishing the European Food Safety Authority and laying down procedures

120

in matters of food safety”, Official Journal of the European Communities, L 31 (1.2.2002), pp. 1-24.

[9] European Court of Justice (1998), “Judgment of the Court of 5 May 1998. - United Kingdom of Great Britain and Northern Ireland v Commission of the European Communities. - Agriculture - Animal health - Emergency measures against bovine spongiform encephalopathy - 'Mad cow disease'. - Case C-180/96”, European Court Reports 1998, p. I-02265.

[10] European Environment Agency (2001), Late lessons from early warnings: the Precautionary Principle – 1896-2000. Copenhagen, (Environmental issue report No 22).

[11] Godard, O. (2001), “Embargo or not embargo?”, La Recherche, Special issue on ‘food risk’, (339), February, pp. 50-55.

[12] Godard, O. (2003), “Le principe de précaution comme norme de l’action publique, ou la proportionnalité en question”, Revue économique, 54(6), pp. 1245-1276.

[13] Godard, O. (2005), “Le principe de précaution et la proportionnalité face à l’incertitude scientifique”, in Conseil d’État, Rapport public 2005 – Responsabilité et socialisation du risque. Paris, La Documentation française, pp. 377-392.

[14] Godard, O. (2006), “Chapter 4. The precautionary principle and catastrophism on tenterhooks: lessons from a constitutional reform in France”, in E. Fisher, J. Jones and R. Von Schomberg (eds.), Implementing the Precautionary Principle: Perspectives and Prospects. Cheltenham (UK) and Northampton (MA, US), Edward Elgar, pp. 63-87.

[15] Godard, O. (2007), “Peut-on séparer de façon précoce le bon grain de l’ivraie ?”, in C. Kermisch and G. Hottois (eds.), Techniques et philosophie des risques. Paris, Vrin, (Coll. ‘Pour demain’), pp. 139-157.

[16] Godard, O. (2008), “Le principe de précaution et la controverse OGM”, Économie publique, IDEP, (21), 2007-2, pp 13-75.

[17] Godard, O., C. Henry, P. Lagadec and E. Michel-Kerjan (2002), Traité des nouveaux risques. Précaution, crise, assurance. Paris, Gallimard, (Coll. ‘Folio-Actuel’ n° 100).

[18] Gollier, C. (2004), The Economics of Risk and Time. Cambridge (Mass.), MIT Press.

[19] Gonzales Vaqué L., L. Ehring, and C. Jacquet (1999), “Le principe de précaution dans la législation communautaire et nationale relative à la protection de la santé”, Revue du Marché Unique Européen, 1, pp. 79-128.

[20] Jonas, H. (1984), The Imperative of Responsibility: In Search of an Ethics for the Technological Age. University of Chicago Press.

[21] Kast R. and A. Lapied (2006), Economics and Finance of Risk and of the Future. Hoboken (N.J.), John Wiley and sons.

[22] Kourilsky, P. and G. Viney (2000), Le principe de précaution. Rapport au Premier ministre. Paris, Odile Jacob.

121

[23] Noiville C. (2000), “Principe de précaution et Organisation mondiale du commerce. Le cas du commerce alimentaire”, Journal du Droit International, 127(2), pp. 263-297.

[24] Raffensperger C. and J. Tickner (eds) (1999), Protecting Public Health and the Environment. Implementing the Precautionary Principle. Washington DC, Island Press.

[25] Sadeleer, N. de. (2003), Environmental Principles: From Political Slogans to Legal Rules. Oxford University Press.

122

Going with Coase beyond Coase: the dynamic approach to the internalization of

external effects1

Jan Horst KEPPLER

Abstract: The article develops R. H. Coase’s insight that the level of transaction costs in the market determines the amount of externalities, thus providing arguments against government intervention. Contrary to Coase, however, we argue that the level of transaction costs cannot be considered as given, and that there is therefore a case for selective and innovative government intervention to reduce such transaction costs. Externalities are approached as intrinsically new and dynamic impacts, whose transaction costs diminish over time, a process that can be accelerated by appropriate government action. In contrast, internalization by means of public intervention through Pigouvian taxation is shown to be epistemologically untenable: if externalities had sufficient information content to allow governments to determine optimal tax levels, these same externalities would already have been fully internalized by the market. The final part of the article proposes two internalization strategies based on a dynamic re-interpretation of the Coasean approach. The first aims at developing feedback mechanisms between generators of externalities and those affected by them through media other than the market. The second seeks to reduce transaction costs in order to extend the domain in which markets can operate effectively by proposing codification strategies for the informational complexities characterizing externalities.

Keywords: External effects, incomplete information, environmental economics, transaction costs, codification, dynamic internalization

1 I would like to thank the participants in the seminar on “Theory and Measurement of Externalities” at the University Paris–Dauphine on 31 October 2007 organized by the Finance et développement durable: approches quantitatives Chair. Valuable comments were received by Françoise Forges and Damien Fessler.

123

1. Introduction – Taking Coase Further

Externalities are impacts on our well-being that the market system is unable to allocate in an optimal manner. It was the historical contribution of Ronald Coase to delineate this limitation of the market system with the help of the notion of transaction costs (Coase (1937)). The market stops where transaction costs are too high. Beyond that point we enter the realm of externalities.

This fundamental insight has usually been interpreted as implying a “hands-off” approach

for practical policymaking with respect to externalities. The argument is usually referred to as the “Coase theorem”, which centers on the proposition that in a world without transaction costs private and social costs are identical and production is maximized. It is well known that Coase’s original work never stated any such theorem and that even his later comments on the issue consist of loose criticisms concerning the earlier formulation by George Stigler, who must be credited with the original formulation of the “Coase theorem” (Stigler (1966), p. 113). Despite the complicated paternity and subsequent permutations of the Coase theorem (see Bertrand (2006) on this issue), the essential message of both Stigler and Coase on the issue of externalities is clear. First, in the absence of transaction costs, all externalities are internalized and thus disappear as externalities. Second, if transaction costs exist, their size determines the amount of existing externalities. Despite providing different answers to the question “Should economic science include considerations concerning non-codifiable transaction costs?”, which have different implications for the epistemological status of economic theory, the two statements are equivalent with respect to policymaking.

Both Stigler and Coase see the market system as the natural limit to any solution to the

problem. Externalities will disappear to the extent that the reach of the market can be extended. While Coase emphasizes the fact that transaction costs form a sort of natural barrier to this extension of the market system that will require us to learn to live with an irreducible residual of externalities, neither considers government capable of improving on the situation, in explicit opposition to the Pigouvian approach (see below). This article argues that the original Coasean approach, whether expressed by Stigler or Coase, does of course formulate valid insights but has so far been inadequately interpreted in terms of its policy implications. This is due to two interrelated reasons. First, despite his unparalleled flair and intuitive understanding of the impact of transaction costs on economic exchange, Coase unfortunately never engaged in a more comprehensive characterization of transaction costs. Far from being an incompressible residual, they constitute a dynamic, highly malleable and intrinsically public phenomenon, which has important implications for the role that public policies can play in reducing them. Second, both Stigler and Coase adopt a static framework for the analysis of the intrinsically dynamic phenomena constituted by externalities.

Treating transaction costs as phenomena linked to the transient status of the information

possessed and required by market participants permits a new approach to externalities. Transaction costs not only imply the existence of externalities, but the two notions are consubstantial. Both refer to the extra-economic reality of human interactions in constant flux that is not captured in the prices for well-defined goods. Externalities do not exist as natural

124

phenomena which can unequivocally be identified as epistemological objects for formal analysis. Rather they are permanently internalized and created by the market system itself in its incessant effort to separate goods for which prices can be found from those for which they cannot. Partitioning out a clearly defined and codified economic “good” from the complexity of the human world implies creating externalities by simply abstracting from the numerous contiguous links which connect this “good” to the totality of the natural and social environment from which it emanates.2

This separation of the economic from the non-economic does not constitute, in itself, a

problem for economic theory. On the contrary, it is a necessary condition for establishing the epistemological objects with which economic science deals and is thus a vital condition for its own legitimacy as an independent scientific discipline. The conundrum for economic theory arises only with the fact that the remainder left over after the pricing process has done its work – the unconsidered links between the economic and the extra-economic world – have welfare relevance. Externalities and the market system implicate each other in a complementary relationship whose precise limits are constantly being renegotiated. This has profound implications for the public policies addressing externalities.

Transaction costs and externalities (for reasons of convention, we will continue to refer to

the two notions side by side) are two sides of the same coin of an as yet un-codified welfare impact. Codifying a significant new welfare impact means reducing transaction costs, making it marketable and progressing with internalization. Codification in this context involves the establishment and measurement of universally accepted cause-impact relations and the formation of stable preferences. This is why externalities and transaction costs need to be approached within a dynamic perspective. Markets not only work best with codified goods, they also are also permanently engaged in a process of formatting goods – both in their objective physical manifestation and in their subjective perception.

This article will thus use a thoroughly Coasean approach to externalities, in the sense that it considers externalities to exist to the extent that transaction costs prevail. However, taking this approach further, it will constitute an argument for coherent and effective public intervention to internalize externalities. Naturally, the task of policy in this case is no longer to substitute itself for market outcomes but to enable the market system to reduce transaction costs and to proceed towards successful internalization, where this is most needed. It means taking both Stigler and Coase by the letter and profoundly respecting the logic of their respective arguments so as to arrive at policy implications that for methodological as well as ideological reasons were beyond their grasp. 2. The Pigouvian Trap The purpose of this chapter is not primarily historical, but is rather to familiarize the reader with the line of reasoning we make reference to in our argument and which can be traced to a first, emblematic contribution to the theory of external effects. The first economist to explicitly state

2 It is obvious that in this approach a zero-level of un-internalised externalities is not only undesirable but also intrinsically impossible. It would in fact mean putting a brake on the evolution of the market system.

125

that markets do not fully capture all welfare relevant aspects of a marketable good was Arthur Cecil Pigou in Wealth and Welfare (1912), an approach later elaborated on in the better known Economics of Welfare (1920). His now canonical distinction between the “private” costs of a train-ride and the “social” costs caused by the sparks flying from the train’s chimney and setting crops on fire firmly grounded the notion of externalities in the public realm. The distinction is not obvious at first sight and absolutely essential at second sight. After all, the farmers whom Pigou deemed affected by the external effects of the passing trains paid for these effects in a very tangible, private manner through the loss of their crops by fires due to flying sparks.

However, once we delve into the nature of the transaction costs that prevent

straightforward internalization through the tort liability system, their public nature becomes immediately clear. First, the legal situation may not be straightforward. This was, of course, one of Coase’s key messages in “The Problem of Social Cost”, thus separating the specific cost of establishing binding property rights (and concomitant liabilities) from other more generic transaction costs. Second, the causal relationships and the socially relevant valuations may not be established. Which part of the crop loss is actually due to flying sparks? Can this be proven? What is the residual value of the remaining crop? And so on.

In the absence of established protocols for settling the issue, the individual farmer may

well decide not to ask for compensation, given the low odds that the small-town lawyer he may be able to mobilize will prevail against the specialized legal experts of the railway company. It is obvious that without prior public measures the transaction costs in this example are very high. It also obvious that the transaction costs are a “social” problem and can only be addressed by collective action: establishing rights and liabilities, measuring damage, conventions for establishing torts. Further research on this issue, however, showed that this is precisely what happened, thus invalidating Pigou’s example. Coase himself, in “The Problem of Social Cost”, quotes from Halsbury’s Laws of England:

“If railway undertakers use steam engines on their railways [...] they are liable, irrespective of any negligence on their part for fires caused by sparks from engines (Coase (1960 (1988)), p. 136).”3

While this shows that Pigou had chosen an inappropriate example – and for a reason, as

will become clear in the following discussion – it also indicates, with respect to the unsatisfactory nature of the discussion about externalities, that Coase’s research was undertaken from a polemical perspective of showing that no policy-relevant externalities exist.4 This is, of course, wrong. Policy-relevant externalities arise and disappear at all times.

3 To be precise, that liability only applies to railways without “express statutory authority”. But this only affects the distribution of property rights and does not in the least affect the basic point: the externality Pigou used as a paradigm case for government intervention by means of a corrective tax was already at the time well known, researched and fully internalized by the legal system.

4 In fact, it was subsequently shown by Coase and others that many classic examples of externalities (such as lighthouses or the pollination of fruit trees) were, in fact, negotiated on markets like any other private good. To some extent even the comprehensive judicial tort system in Anglo-Saxon jurisdictions can be interpreted as a series of organized bilateral markets for otherwise uninternalized externalities.

126

However, let us first deal with the trap in which Pigou ensnared himself. The mistaken choice of example could only arise because of Pigou’s insistence on two crucial characteristics of externalities. First, external effects and their internalization can take place in a system of static optimization. Second, in order to do so, governments possess all the knowledge about liabilities, causal relationships and damage costs that private participants lack, from which the imposition of a shadow price in the form of a “Pigouvian tax” logically follows. Pigou thus correctly identified externalities as social issue, but then fell into the trap of subsequently treating externalities as fully codified goods, for which it only so happened that governments rather than markets would need to take responsibility.5

Coase was, of course, correct to launch an attack on this paradoxical position in 1960 by

pointing out that private parties were fully capable of optimally treating externalities if they possessed the same information and low transaction costs Pigou assumed governments to have. He was also well placed to do so given that in “The Nature of the Firm” (1937) he had already defined the limits of the market system in terms of transaction costs. In perfect symmetry, “The Problem of Social Cost” makes exactly the same argument from the other side. The market can be extended up to the point that transaction costs are sufficiently low.

The point was taken up by Kenneth Arrow in “The Organization of Economic Activity:

Issues Pertinent to the Choice of Market versus Non-Market Allocation” (1970) with customary elegance. In identifying the non-existence of markets as the underlying reason for the existence of externalities, he wrote:

“The problem of externalities is [...] a special case of a more general phenomenon: the failure of markets to exist […]. The discussions in the preceding sections [on the non-existence of markets] suggest two sources of transaction costs: (1) exclusion costs and (2) costs of communication and information, including both the supplying and the learning of the terms on which transactions can be carried out (Arrow (1970), 76f).”

Externalities are thus goods for which no markets exist due to the paucity of available

information. Indeed Arrow had a deep understanding of informational complexity and the up-front work necessary to transform it into codified nuggets of information – an essential pre-requisite for the creation of marketable goods. The following paragraph highlights the extent to which he considered such codification to be essential for the reduction of transaction costs and market exchange:

“To cooperate and to take advantage of the division of labor, there must be an exchange of information in one way or another. Let us draw upon […] information theory […] [which considers that] communication is a costly operation […]. Roughly speaking, you want to phrase the messages that occur most frequently in as short a way as possible […]. Concepts used repeatedly are given short technical names, so that a couple of words convey a whole thought […]. This is an illustration of the process that is sometimes called ‘encoding’ (Arrow (1979), 161).”

5 Elsewhere, we have commented more widely on the paradox of the Pigouvian position. See Keppler (1998), “Fixed Costs, Information and Externalities”.

127

However, Arrow also had a reputation as a theoretical economist to defend and never provided more than these tantalizing fragments on the issue of externalities. Subsequently, the Pigouvian approach to externalities prevailed despite the inherent contradiction of treating externalities as already fully codified goods. The tone was set by The Theory of Environmental Policy by Baumol and Oates in 1988, which was built on an analytical formulation of the Pigouvian approach. Works by Pearce, Freeman, Sandmo, Kneese, Costanza and many others followed and further explored different aspects of the Pigouvian approach to static optimization, characterized by the paradoxical double assumption of full information and the need for government intervention.

Equalizing marginal social cost and the marginal private cost of abatement (or marginal

private benefit) to determine the optimal amount of externalities became de rigueur the static framework of environmental economics. The cross formed by continuous marginal cost and benefit curves became the iconic representation of that approach (see below). Consequently taxes became the policy instrument of choice for the internalization of externalities, in particular environmental externalities. Environmental taxes have, of course, a number of highly beneficial effects. They have low institutional transaction costs, provide clear incentives for market participants and do not pre-empt technological choices, while at the same time “inducing” technological change in the right direction. Furthermore, they can be used to offset distortive taxes on labor, capital or transactions.6 However, their ability satisfactorily to solve the theoretical problem of the optimal internalization of externalities is limited.

Figure 1. Optimal internalization in a Pigouvian framework

Thus a massive literature was built on the underlying assumption that externalities were

well-defined goods that just happened to have been overlooked for the market system. In other words, social costs were considered well-defined, measurable, monetizable, continuous and

6 It is impossible here to provide even an outline of the richness of the literature elaborating on the Pigouvian approach, which includes work on the relative efficiency of price- and quantity-based mechanisms (for instance Weitzman, 1974), the distributional impact of regulatory instruments, taxes or tax equivalent systems such as emissions trading (Cruciani and Keppler, 2009) and tax interaction effects (Goulder, 1999).

128

differentiable. The fact that the market system would have been eager and perfectly able to internalize real-world externalities if they had displayed such desirable informational qualities was literally never even mentioned. On epistemological grounds, the contradiction was blatant: externalities could be dealt with applying all the rigor and sophistication of standard economic theory as long as they were treated akin to other economic goods, in other words, as long as they were no longer treated as externalities.

It is easy to understand why such a paradoxical situation could persist for so long. In the

minds of environmental economists, the cost of epistemological incoherence was more than offset by the benefits of increased methodological attractiveness. The analytic formulation of the Pigouvian approach allowed the environmental economic profession to enter the methodological mainstream. Differential calculus, game theory, even general equilibrium theory (in the work of Karl-Goran Maler) all found “environmental” applications. Environmental economics became respectable, and enjoyed all the power and prestige in terms of academic journals and positions that comes with respectability.7

3. Addressing the information issue... and leaving it unresolved Most environmental economists would probably bristle at the criticism that modern

environmental economics disregards the information problem. After all, the Achilles’ heel of the Pigouvian approach was rather obvious. The Pigouvian approach to externalities hinges on overcoming the information barrier concerning the value of social costs, and much theoretical and practical effort went into its measurement. Thus many attempts were made to break down the complex notion of “environmental social costs” into its component parts in the hope that they would thus become more tractable.

7 Let us be clear that the criticisms addressed to environmental economics go far beyond the standard criticism of economic theory, namely that it is too far removed from reality. While an assumption such as perfect information in general economics constitutes a necessary abstraction that to some extent generates the added value of the model even if that removes it from immediately observable reality, the same assumption in environmental economics defeats its very raison d’être, i.e. the claim of being able to include a class of welfare-relevant arguments characterized by less-than-complete information within economic discourse, arguments that are traditionally excluded from standard economic theorizing.

129

Figure 2. The components of environmental value

However, such breakdowns only underline the essential problem. If direct use values

such as hunting or tourism could be neatly separated from existence and bequest values, then the former could be dealt with by the market mechanism. The latter would, more than ever, require public intervention in the face of “market failure”, given their inability to yield well-defined social cost schedules. One could, of course, devise sensible internalization strategies by extending the market mechanism wherever possible, in an ongoing attempt to reduce the extent of the intractable core of externalities.

However, admitting the existence of a core of external effects impervious to the standard

Pigouvian approach would endanger the methodological respectability of environmental economics as a whole. Economic theory after all relies on the systematic and complete codification of its arguments. Environmental economists have thus devised a number of strategies aiming at the full codification of external effects. In other words, such strategies seek to evaluate social costs in monetary terms, oblivious to the fact that if these estimates had any credibility (not least with the public authorities presumed to be financing them) private providers would step forward to provide or protect them. Without any claim to completeness, one may mention:

1. The direct evaluation of damage through replacement or remediation costs (for

instance, damage to houses and other structures due to air pollution). 2. The measurement of ecosystem services provided by natural habitats (pollination

services, prevention of erosion, water purification, the provision of genetic resources etc.). Again, the usual measure is to take replacement costs as a lower bound of true value.

3. The travel cost method for measuring the value of tourist sites, based on

estimating the demand functions of visitors traveling different distances (see Figure 3 below).

130

Figure 3. An illustration of the travel cost method (The number of visitors from different population basins and their respective travel costs

can provide a lower bound for their utility from visiting the site in question)

4. Hedonic pricing, which relies on deriving the value of a public good, say a park, from the impact it has on related private goods, such as houses that are at different walking distances from the park. It relies on estimating econometrically the equation:

House price = α + β1*size + β2*year + β3*closeness to park. Once the price elasticity β3 is estimated and the available housing has been assessed, the

absolute price impact of the park can be derived and taken as its “value”. It is obvious that these four methods are useful for measuring the value of public goods

only to the extent that they concern the impact on private goods. In other words, they only measure the first two columns in the graph depicting the components of environmental value. Consider the example given for the hedonic pricing method: a private entrepreneur might provide the amenity of a private park to increase the value of a new housing development precisely up to the point that is determined by the hedonic pricing method in connection with a public park. The public park, however, has a much higher value, in that it is visited by school children, the elderly, tourists, amblers etc. who will never even consider buying a property in the area.

In other words, none of the four methods listed above is capable of capturing true public

goods. This is why environmental economists have developed a further method, which crystallizes the futility of the paradoxical and ultimate self-defeating attempt to quantify the class of welfare impacts that by their very definition are uncodifiable. This measure is usually referred to as contingent valuation, and due to its practical importance and epistemological exemplarity, we shall comment on it more extensively in the next section.

4. Contingent valuation On the face of it, contingent valuation neatly cuts the Gordian knot created by the

ambiguities and inadequacies of the proceeding approaches with one swift stroke of the methodological sword: if one wants to know the true value of a public good, its complete value including existence, bequest and option values in addition to direct and indirect use values, one

131

just needs to ask those benefiting from it. Using a simple questionnaire, contingent valuation superficially avoids the methodological pitfall of the four preceding approaches –applying sound private reasoning to public goods – only to stumble headlong into it on a slightly deeper level. Asking people to answer the question “How much would you be willing to pay for the public good in question?” provides a contingent valuation that captures, in principle, all the utility-relevant aspect of a public good.8 As far as the provision or protection of a public good is concerned, no aspect is excluded. However, as far as the consumer or recipient of the public good is concerned, the confusion between a private and a public good is even more entrenched.

Asking someone to compare the value of the “environment” he experiences every day in a

vague and multi-dimensional way to the marketable he goods he buys or sells, assumes that the processes of preference formation for the two categories of goods are comparable. But the same informational problem that was identified in regard to the notion of “transaction costs” on the supply side now arises on the demand side. Economic (market) value is, ever since Adam Smith, an inter-subjective notion.9 Asking someone for a conclusive answer in regard to an inter-subjective process of value formation established through a process of haggling, buying and selling that has not even begun, is a sign more of considerable epistemological than methodological naivety. What is in question is not so much the method itself, but its underlying hypotheses concerning the existence of a well-defined informational structure of preferences for public goods and their impact on utility. Contingent evaluation is only possible on the basis of a blatant confusion of use value and exchange value. The existence of the former in a generalized, non-codified and vague form that is specific to each individual, in no way implies the existence of the latter.

In the light of this fundamental weakness of contingent valuation, it hardly seems to

matter that this particular attempt at establishing monetary values for public goods also raises some tricky theoretical questions as to the framing of the question itself and to the distinction between:

a) willingness-to-pay and willingness-to-accept, b) compensating and equivalent variation, and c) Hicksian and Marshallian demand.

At first sight, it would seem that the questions “How much would you pay to progress from level A to a higher level B of environmental quality?” and “How high would the compensation need to be for you to accept a decrease from the higher quality level B to level A?” should yield comparable answers. The two answers are, however, quite different. The reason for this is that the first question is put ceteris paribus to an individual possessing a lower level of income than the second individual. The first individual’s answer would have been to the same as the second individual’s only if the question had been “How much would we need to compensate you for the fact that the promised improvement from A to B never took place?” The second

8 Our criticism abstracts here from the question of strategic behavior (for instance by indicating higher values than the “true” values) that can, in principle, be solved by more intelligent questionnaires, control questions, framing, etc.

9 See, for instance, Keppler (2008).

132

individual’s answer would have been equal if the question had been “How much would you be willing to pay to avoid the decrease from B to A?”

The reason is that in the first two examples, the ex ante utility levels of the change in

environmental quality are the relevant parameters for assessing the equivalence of environmental quality change and its monetary equivalent (compensating variation), while in the second two examples, the ex post utility levels of change are the relevant parameters (equivalent variation). The first pair of questions thus assumes Hicksian demand functions with constant utility levels, while the second pair assumes Marshallian demand functions that incorporate the changes into the future utility levels.

Figure 4. The relationships between different welfare measures I

Source: Keppler (1991).

The two graphs further elucidate the relations between theses different measures of the value of the public good of environmental quality. It is left to the reader to decide which measure provides the “true” value of the environment.

Figure 5. The relationships between different welfare measures II Source: Keppler (1991). Finally, however, such specific theoretical questions do not pose the decisive difficulty of

finding monetary expressions of the value of public goods and of the externalities that threaten to destroy or to diminish them. The key point remains that there is no escaping the Coasean verdict that what is hidden behind the veil of transaction costs cannot easily be drawn out into the light of overt monetization. Or in Coase’s own words:

133

“My point was simply that such tax proposals are the stuff that dreams are made of. In my youth it was said that what was too silly to be said may be sung. In modern economics it may be put into mathematics (Coase (1988), 185).”

5. The seductive self-sufficiency of the static Coasean approach So if externalities cannot be convincingly measured and monetized in the spirit of the

Pigouvian approach, and if therefore their smooth integration into a market system remains elusive, is the right policy approach then to leave good enough alone and refrain from any policy action to internalize externalities? In other words, do policy-relevant external effects perhaps really not exist? This has, of course, been the point of view of a libertarian tradition following George Stigler’s formulation of the “Coase theorem”, which states that in the absence of transaction costs private bargaining over the externality in question will lead to an efficient outcome regardless of the initial allocation of property rights. George Stigler had formulated the Coase Theorem as follows:

“[...] when there are no transaction costs the assignments of legal rights have

no effect upon the allocation of resources among economic enterprises (Stigler (1988), 77).” and

“[...] the magnitude of the transaction costs puts a ceiling on how large external economies [...] can be (ibid., 78).”

In essence, this formulation only states that frictionless markets are efficient and applies

this assertion to the class of utility-relevant goods hitherto referred to as external effects. The attentive reader will appreciate that Stigler’s formulation repeats precisely the same epistemological paradox that we identified earlier in Pigou’s approach. As long as there are no transaction costs, externalities can be treated just like other goods, with the result in Coase’s words that “with zero transaction costs, private and social costs will be equal... [and] the value of production would be maximized Coase (1988), 159).” These statements, while formally correct, are devoid of content, since they deal with a class of goods whose very existence depends on transaction costs.

Ronald Coase insisted in other instances on the existence of positive transaction costs and

his desire to examine their impact:

“In sections III and IV [of the “The Problem of Social Cost”], I examined what would happen in a world in which transaction costs were assumed to be zero. My aim in doing so was not to describe what life would be like in such a world but [...] to make clear the fundamental role which transaction costs do, and should, play in the fashioning of the institutions which make up the economic system (ibid., p. 13).

And more forcefully still:

134

“The world of zero transaction costs has often been described as a Coasean world. Nothing could be further from the truth. It is the world of modern economic theory, on which I was hoping to persuade economists to leave (ibid, 174).”

Coase also provided a definition of transaction costs which is now standard but was less so at the time:

“In order to carry out a market transaction, it is necessary to discover

who it is that one wishes to deal with, inform people that one wishes to deal and on what terms, to conduct negotiations leading up to a bargain, to draw up the contract, to undertake the inspection needed to make sure that the terms of the contract are being observed, and so on. These operations are often extremely costly, sufficiently costly at any rate to prevent many transactions that would be carried out in a world in which the pricing system worked without cost (ibid., 114).”

What Coase, however, failed to identify were the reasons that distinguish transaction costs

from other costs, i.e. he does not answer the question of why there is no market for transactions as there is for other factors of production.10 The reason is, of course, the absence of codification. Transaction costs concern the messy, ill-defined, all-pervasive slush of human life from which well-defined economic goods need to be extracted before they can be brought to market.11

If transaction costs exist, so too will externalities. To the extent that externalities have

impacts that outweigh the transaction costs, private bargaining will internalize them up to the cost of the transactions. Implicit in the argument is that given the imperfect informational structure of externalities, governments do not have any intrinsic advantage over the private sector in addressing them. In other words, there is – in the static perspective that both Coase and Stigler had in mind – nothing to be done about them. Ronald Coase and George Stigler, each in his own way, are both perfectly consistent with the axiom of Professor Pangloss, the mentor of young Candide in Voltaire’s eponymous novel, that “this is the best of all possible worlds”. If at any time external effects existed whose internalization would cost less than the transaction costs that created them in the first place, surely people would internalize them through private bargaining. In an efficient world, they already have.

The argument is general, simple and seductive. Its policy implications are clear and easily

implementable: hands off. The argument is also correct as long as the externality issue is treated

10 There are, of course, markets for transactions, as any broker will tell you. But what Coase refers to are those transaction costs to which the division of labor cannot be applied.

11 Nevertheless, Coase’s insistence on the existence and importance of transaction costs is considerable progress from the empty tautology of the “Coase Theorem”. In fairness to George Stigler, however, one should state that his formulation of the “theorem”, which put Coase’s work on the map, was primarily concerned with policy implications rather than with more subtle methodological or epistemological points. Stigler was always the great propagandist of the Chicago School and its laissez faire beliefs. For that purpose, an empirically empty but immediately striking formulation served better than one that was more complete and nuanced. Furthermore, the policy conclusions that Coase draws from his own work do not differ from Stigler’s.

135

with the help of a static general equilibrium view, in which ceteris paribus holds, i.e. the level of transaction costs does not change during the period of analysis. So why do most people react with intuitive disbelief to the proposition that external effects such as environmental pollution are best left alone?12 The reason is that the application of static equilibrium thinking – which is a defensible methodology in all other areas of microeconomic theory – is profoundly unsuited for discussing externalities.

Externalities are characterized by their diffuse, informationally unstructured, open and un-codified nature. This absence of informational linearity constitutes precisely the transaction costs between two parties who would otherwise both benefit from a movement towards a Pareto superior configuration through a better allocation of externalities. It does not matter whether this happens by forcing the originator to pay for the damage caused or by allowing the agent impacted to offer compensatory payments to the perpetrator to reduce the damage he causes.

However, even if one acknowledges that the uncodified nature of externalities creates

transaction costs that prevent internalization, its diffuse, informationally unstructured and uncodified nature does not allow a stable equilibrium to occur either. Human nature, being what it is, seeks to describe, analyze, categorize, define and codify welfare-relevant arguments. Semantic entities coalesce over time to ever more determined notions through which interpersonal communication can take place with a reasonable amount of certainty. Once communication at low cost is possible, haggling, bargaining and optimization will not be far away.

In other words, transaction costs are not fixed. Intrinsically, the slow but persistent work

of codification will reduce them and integrate ever greater swathes of external effects into the market system. Every single externality problem of recent decades has followed this path. Identified by fringe groups, amplified by a progressive minority, pored over by scientists and experts, taken to heart by interest groups, broken down into policy positions and legislative proposals by the political process and finally integrated by market participants – such is the lifecycle of an externality. Of course, some issues fall by the wayside (global cooling), while others take far too long to be addressed (asbestos), but the tendency is inexorably towards less uncertainty and greater codification.13

Externalities and transaction costs are intrinsically dynamic phenomena. Technological

advances, institutional progress and preference change all have a bearing on the level of transaction costs in the market. External effects need to be thought of as essentially new

12 Provoking such intuitive disbelief and maintaining their arguments was, of course, on the face of it, an exquisite pleasure for Chicago economists such as Stigler and Coase. In fact, a significant part of the Chicago aura came from the implicit or explicit assertion that even relatively simple economic thinking, as long as it was undertaken by qualified professionals steeped in such thinking, could arrive at new and far-reaching policy conclusions profoundly at odds with people’s conditioned reflexes.

13 The fact that this process is inexorable should not be seen as justifying the “hands off” approach of the Chicago School. The key point is that this process can be sustained and accelerated through public action in the interest of welfare improvement and economic progress. In fact, it would be possible to develop a theory of public institutions based on reducing transaction costs, codifying complex issues, allocating property rights and responsibilities, assisting in preference formation and thus internalising externalities.

136

phenomena. Once this simple observation is accepted, it allows for a vast array of innovative effective policy responses that avoid both the conceptual absurdity of the Pigouvian approach and the callous indifference of the Chicago approach.

6. A Neo-Coasean Framework for Dynamically Internalizing Externalities In the following, we pursue a strictly Coasean approach to the extent that we base our

argument on the proposition that the amount of externalities corresponds to the level of transaction costs in the market. However, contrary to Ronald Coase, we argue in addition that transaction costs are not exogenously given but are contingent on the actions of participants. In other words, transaction costs are endogenous to the process of internalizing externalities. Once this fundamental point is accepted, internalizing externalities simply means reducing transaction costs by whatever means available. The key difference between the classic Coasean and the neo-Coasean framework proposed here is the claim that technological, organizational and informational improvements permanently expand the set of internalizable externalities.14

In order to fully understand this approach it is helpful to go back to the definition of an externality as a “good with a welfare impact not taken into account by the agent producing it.” This simple definition specifies the fundamental difference between goods that are externalities and those that are not – the existence of a feedback mechanism between those affected by it, positively or negatively, and those producing it. Normal marketable goods, of course, have a simple and complete feedback mechanism through the market price. You provide a valuable good to me, I pay. You take something from me, you pay. In competitive markets, this reciprocity guarantees efficiency and optimality.

Such feedback mechanisms precisely do not exist in the case of externalities. You pollute

and take my fresh air, and nothing happens. I appropriate the insights of your research, and again nothing happens. The graph below demonstrates this missing link in the simplest possible manner for a power plant that negotiates its inputs (say coal) and part of its outputs (electricity) on markets with feedback mechanisms that exist, and shows the absence of such a link for the other part of its outputs, namely pollution and greenhouse gas emissions.

14 The counter-argument that the essence of economic theory is to work with static equilibria is a non sequitur since it would be equivalent to saying that economic theory should not concern itself with externalities. There are economists who take this position, and while they might not win prizes for openness of mind or for policy relevance, they are at least methodologically consistent. Externalities, which are dynamic phenomena in permanent flux, are in themselves indicators that the economic system is not in equilibrium.

137

Figure 6. The missing feedback mechanism So far, so good. The fact that the existence of externalities is equivalent to the absence of

markets had been known at least since Arrow (1970). However, the debate was left there. The attitude was: if the market cannot price it, leave it un-priced (Chicago) or let the government determine the price or, alternatively, the quantity (Pigou). But missing from the debate were the two following policy-relevant insights:

1. Feedback for the internalization of externalities can be provided through

mechanisms other than market. 2. The reduction of transaction costs can extend the reach and pertinence of the

market mechanism for the internalization of externalities.

Both insights are based on the premise that externalities are currently outside of market-based decision-making due to transaction costs of an informational nature. They also both provide helpful perspectives for establishing relevant feedback mechanisms in order to proceed towards internalization. However, while they are closely linked and complementary they are not identical.

The first insight accepts the complex informational nature of externalities and thus that

their internalization through the market mechanism is currently not an option. This does not mean that internalization as such has to remain beyond the scope of policymaking. Political and legal processes, institution-building, and deliberative processes in civil society are examples of areas where the establishment of feedback mechanisms between the parties concerned takes place. Clearly, there exists an enormous spectrum of mechanisms for internalization. Even the Chicago School with Ronald Coase and George Stigler was well aware of this fact, albeit in a rather restricted fashion: establishing and allocating property rights, perhaps the most fundamental of economic institutions, was to be the first and necessary step in any process of internalizing externalities.

The array of possibilities for the internalization of externalities through institutional and

deliberative processes in the widest sense is, of course, infinitely larger. To some extent, the creation of institutions themselves is intrinsically linked to their function of internalizing externalities. This is not the place to proceed either towards a systematic exploration of these possibilities or towards an externality-based theory of institutions. In practice, doing so requires the “detailed investigation of the actual results of handling the problem in different ways” (Coase (1960, p. 18-9). We would, however, like to convey the flavor of innovative measures for the internalization of externalities through a number of examples for standard cases of external effects. All of them are designed to create implicit feedback mechanisms between those affected by externalities and those having leverage over their provision in situations dominated by informational complexity and high transaction costs.

1. Oblige the managers of critical industrial installations (chemical plants, refineries,

nuclear plants, etc.) to live no more than five kilometers away from the plant. 2. Reserve one board seat in major companies for representatives of accredited

consumer or environmental organizations. Of course, this representative is held to the same confidentiality requirements on commercial issues as everybody else.

138

3. Formulate regular and extensive environmental and social reporting requirements (this is, of course, a process currently being implemented , albeit in a haphazard and little theorized manner).

4. Facilitate class action suits for certain kinds of environmental or social issues. Class action is precisely a way of using the legal system to internalize when transaction costs are too high for the individual claimant. Precautions need to be taken not to allow such class action suits to become indiscriminate transaction costs for normal business.

5. Proceed towards a drastic review of the patent system and the protection of intellectual property rights, with increased thresholds and shorter periods of protection, thus making it more appropriate for fast-moving digital economies with intrinsic winner-takes-all structures.

The second insight is the basis for another strategy of internalization: the reduction of

transaction costs to let markets participate in internalization. It is centered on the fact that externalities are due to transaction costs in the market, which hamper the establishment of appropriate feedback mechanisms. The aim this time is to lower the transaction costs so that markets can do their work. The challenge is to “commoditize” externalities. The shift in the approach to climate change from fringe issue twenty years ago to today’s functioning carbon markets is an excellent example of such a process. This example also reminds us that there are substantial costs in creating the informational and institutional infrastructures needed for such a commoditization process (codification).

Such approaches recognize that governments and other public institutions have a role in

codifying externality issues in order to allow their treatment through a decentralized market process. This involves the establishment of generally accepted physical or chemical cause-impact relationships, the quantitative measurement of impacts, the allocation of costs, responsibilities and benefits and, perhaps most importantly, the advancement of processes which permit the formation of stable preferences beyond a vague “unease” or “appreciation” of specific external effects. One should never forget the basic lessons of the Coasean approach: a given externality is not the result of either a cleverly engineered social injustice or a permanent blind spot of the market system but the consequence of transaction costs that are due to the newness and the informational complexity of the externality in question and the diffuse nature of the welfare impacts connected with it. Governments or, more generally, public processes can, and indeed often do, usefully develop measures to address these issues .15 Even Coase himself clearly saw clearly this interplay between public intervention and the establishment of markets:

“It is not without significance that these [financial] exchanges, often

used by examples of a perfect market and perfect competition, are markets in which transactions are highly regulated... It suggests, I think correctly, that for anything approaching perfect competition to exist, an intricate system of rules and regulations would normally be needed (Coase (1988), 9).”

15 Many of the examples advanced below will have a familiar ring to them. However, the point is not that such measures do not yet exist – they are part and parcel of the inexorable codification process that all externalities undergo. The point is that they exist in a conceptual vacuum, because they have never been theorized in the context of a theory of externalities. This impedes the systematic development, linking, streamlining and improvement of such measures.

139

Below we present a number of further illustrative examples of measures to reduce

transactions costs. 1. Undertake basic scientific research into externalities to understand basic cause-

impact relationships. The Intergovernmental Panel on Climate Change (IPCC) is a vast example of this sort of undertaking.

2. Undertake applied research in medicine, agriculture, engineering, etc. in order to acquire a solid understanding of the magnitude of the impact. Major externalities in recent years (asbestos, mad cow disease etc.) were effectively addressed once the impacts had been unequivocally established.

3. Organize political and social processes that allow responsibilities and a distribution of the costs of internalization to be established. In economic jargon this is referred to as the “allocation of property rights”, although the term implies a degree of codification rarely achieved with external effects even after such processes.

4. Advance social and individual processes of preference formation through information dissemination, public hearings, media involvement, etc. In this context, the Pigou-inspired monetization of social cost can also be partially rehabilitated, albeit as part of a much larger process of preference discovery rather than as a substitute for it.

5. Formulating systematic reporting requirements for key environmental attributes of major goods such as houses (air quality, noise level, proximity to parks, etc.), so that the market can price them.

6. Reduce transaction costs through measurement systems, transparency and disclosure requirements, standard-setting and labeling. (Examples can be found in the area of socially and environmentally responsible investment.)

7. Create markets where the codification processes have advanced sufficiently far, while being aware of the remaining transaction costs. (Markets for CO2 or SO2 emissions work fine thanks to the easy measurability of the underlying commodity. Markets for energy efficiency improvements, so-called “white certificates”, may be a different matter altogether.)

8. Distinguish, in the case of stubbornly and intrinsically complex externalities, which are likely to remain so, e.g. the loss of biodiversity, which are marketable (use values such as eco-tourism) and which are non-marketable (non-use values such as existence values). These attributes need to be addressed separately with different groups of instruments.

9. Recognize the multi-dimensionality of externalities. Establish partial use-rights that may be amenable to codification and market allocation, rather than all-encompassing property rights.

We said earlier that transaction costs and consequently the level of uninternalized

externalities are a function of the scientific, technical and informational infrastructure. The present is therefore a propitious moment. Never before have these infrastructures been as advanced as they are now. Science permits the identification of new causal chains. Technology allows ever more precise measurements. The global information society permits much faster information transmission, debate and preference formation on the one hand and new and innovative forms of organization to establish the feedback that is missing in the case of externalities on the other. A buzzword such as “stakeholder involvement” provides a glimpse of the potential of the feedback mechanisms which might serve to anticipate and manage external effects but which are neither coherently theorized nor systematically applied. This article is a

140

contribution towards a more systematic exploration of the opportunities offered when applying Coasean insights to the frontier between market and non-market allocations of goods in a dynamic perspective.

7. Conclusion Our article elaborates Ronald Coase’s thesis that the amount of externalities is determined

by the level of transaction costs in the market. Contrary to Coase’s own development of the implications of this insight, however, we do not consider this level as given. Because externalities are intrinsically new and dynamic effects, transaction costs diminish over time, a process that can be significantly accelerated by appropriate government action. Yet Coase’s great merit is to have associated externalities with transaction costs, i.e. with the intrinsic limits of the market mechanism to proceed towards an efficient allocation of goods.

In settling for a static equilibrium approach, however, ironically Coase rejoins Arthur

Cecil Pigou, whose tax-based approach he had originally set out to prove utopian at best and usually harmful at worst. Pigou’s approach is indeed epistemologically untenable, even though a large literature of environmental economics has developed around it. The Pigouvian approach finds itself caught in a basic contradiction: if externalities had the desirable informational qualities that allowed governments to determine appropriate tax levels for internalization, these same externalities would already have been fully internalized by decentralized negotiations in open markets.

The final part of our article is concerned with presenting an internalization strategy to

transform the primarily defensive Coasean approach into a constructive, forward-looking approach to external effects. This strategy consists of two complementary sub-strategies to address high transaction costs. The first aims at developing feedback mechanisms between generators of externalities and those affected by them through mechanisms other than the market, which include political and legal processes and specific regulatory requirements. The second seeks to reduce transaction costs to extend the perimeter inside which markets can effectively operate by codifying the informational complexities that characterize externalities. While some of the concrete measures proposed are not entirely new, they have so far arisen haphazardly from historical contingencies. What our article provides is their systematic exploration on the basis of a coherent conceptual approach built on a dynamic re-interpretation of the original Coasean insight. References [1] Arrow, Kenneth J. (1979), “The Division of Labor in the Economy, the Polity and the

Society” in Driscoll Jr., Gerald P. [1979], Adam Smith and Modern Political Economy, Ames, IO, Iowa State University Press, p. 153-164.

[2] Arrow, Kenneth J. (1977), “The Organization of Economic Activity: Issues Pertinent to the Choice of Market versus Nonmarket Allocation” in Robert Haveman and Julius Margolis (eds.), Public Expenditure and Policy Analysis. Boston: Houghton Mifflin, p. 67-81.

141

[3] Baumol, William and William Oates (1988), The Theory of Environmental Policy, New York: Cambridge University Press.

[4] Bertrand, Élodie (2006), “La thèse d’efficience du ‘théorème de Coase’: Quelle critique de la microéconomie?”, Revue économique 57(5), p. 983-1008.

[5] Coase, Ronald H., “The Nature of the Firm”, Economica, n.s. 4: 386-405, reprinted in Ronald H.

Coase (1988), The Firm, the Market and the Law, Chicago: University of Chicago Press.

[6] Coase, Ronald H. (1960). “The Problem of Social Cost”, Journal of Law and Economics 1(1): 1-21, reprinted in Ronald H. Coase (1988), The Firm, the Market and the Law. Chicago: Chicago University Press, p. 95-156.

[7] Coase, Ronald H. (1988), “Notes on the Problem of Social Cost”, Coase (1988), The Firm, the Market and the Law. Chicago: Chicago University Press, p. 157-185.

[8] Diamond, Peter A. and Jerry A. Hausman (1994), “Contingent Valuation: Is Some Number Better than No Number?” Journal of Economic Perspectives 8: 45-64.

[9] Freeman, M. 1979. The Benefits of Environmental Improvements: Theory and Practice. Baltimore: Johns Hopkins University Press.

[10] Keppler, Jan Horst (1991), “Wieviel Geld für wieviel Umwelt? Entschädigungskonzepte und ihre normativen Grundlagen”, Zeitschrift für Umweltpolitik (Journal of Environmental Law and Policy) 91 (4), 1991, p. 397-410.

[11] Keppler, Jan Horst (1998), “Externalities, Fixed Costs and Information”, Kyklos 52 (4): 547-563.

[12] Keppler, Jan Horst (2008), L’économie des passions selon Adam Smith: les noms du père d’Adam, Paris: Kimé.

[13] Ostrom, Elinor (1995), “Designing Complexity to Govern Complexity” in Susan Hanna and Mohan Munasinghe (eds.), Property Rights and the Environment: Social and Ecological Issues. Washington, DC: The Beijer International Institute of Ecological Economics and The World Bank, p. 33-46.

[14] Pigou, Arthur C. (1932 (1920)), The Economics of Welfare, 4th ed. London: Macmillan.

[15] Pommerehne, Werner W. (1978). Institutional Approaches to Public Expenditure, Journal of Public Economics. 9: 255-280.

[16] Stigler, George J. (1966), The Theory of Price, 3rd edition, New York: Macmillan.

[17] Stigler, George J. (1988), Memoirs of an Unregulated Economist, New York: Basic Books.

[18] Voltaire (1994), Micromégas, Zadig, Candide, Paris: Flammarion.

142

Mean field games and oil production1

Olivier Guéant, Jean-Michel Lasry, Pierre-Louis Lions

Abstract : In this paper we study the evolution of oil production in the long run. A first

optimization model is presented, that can be solved using Euler-Lagrange tools. Because these classical tools are not the best suited to the model, we adopt a mean field games approach based on two partial differential equations. An extended model is then presented to analyze the influence of new competitors which might enter the market with energy from renewable sources. The usefulness of a subsidy to potential entrants is discussed.

Keywords : Oil production, mean field games, public economics

1Acknowledgments must be made to the Chair ``Finance and Sustainable Development'' and a special thank must go

to the CFE (Conseil Français de l'Energie) for its financial support.

143

1. Introduction The economic study of oil production dates back to a seminal paper by Hotelling

published in the 1930s (Hotelling, 1931). Although many models have been constructed since then, our contribution is still rooted in Hotelling’s orginal paper and makes extensive use of the concepts introduced at that time. In fact, while the concepts are derived from the past, the methodology we present is relatively new and is an example of what is called mean field games2.

Our aim in this paper is twofold. First, we shall present a new approach for dealing with long-term rational expectations dynamic equilibria. The mean field games approach is in fact the best suited3 for addressing the problem and allows for various generalizations, as it can be seen in Guéant et al. (2009) and Giraud et al. (2009). The basic idea of this approach is to consider that the number of producers is large enough to justify a model using a continuum. This continuum hypothesis implies that a single producer cannot move the price in a given direction through his own production, and corresponds to the perfect competitive framework of economists. Moreover, this continuum hypothesis turns out to simplify the problem, since it reduces to a forward/backward system of partial differential equations (PDE). The forward/backward character of mean field games is at work here when it comes to considering the production of a resource whose availability is limited. The backward dimension comes from the optimization of the production timing and the forward dimension is directly linked to the evolution of individual and global oil reserves.

Apart from this largely methodological first goal, we want to use the model to analyze the interaction between oil producers and a competitive fringe of renewable energy producers which may enter the market. On this specific topic, we shall, in our framework, show some interesting and unusual results about the impact on oil production – and hence on pollution – of a subsidy to stimulate the entry of competitors to oil producers.

In what follows, we shall first consider a problem for which both a classical approach and a mean field games approach can be used. We derive results on the evolution of oil production and exhibit a Hotelling rent. We put forward two ways of solving the problem, the first based on Euler-Lagrange methods (Section 2) and the second on mean field games partial differential equations (Section 3). Introducing two partial differential equations that are at the core of mean field games theory opens up a number of possible improvements (see Guéant et al. (2009) or Giraud et al. (2009)). Our model will be used and generalized to study the influence of the potential entry of new competitors, particularly those developing alternative and renewable energy sources (Section 4). This framework allows us, for instance, to use powerful analytical tools to consider the negative effect in terms of carbon emissions of a subsidy to alternative energy producers, as in Ekeland (2007).

2 The model

2An overview of mean field games is provided at the end of this paper.

3In two related papers (Guéant et al., 2009) and (Giraud et al., 2009) we extend the model to deal with randomness

and non-financial criteria. These related contributions make extensive use of mean field games partial differential equations and show that this framework, rather than the usual Euler-Lagrange system, is the best suited to modeling oil production in the long run.

144

2.1 Basis of the model 2.1.1 Introduction We consider a large number of oil producers, which can be viewed either as wells or from

a more macro standpoint as oil companies. The only assumption we make is that there is a sufficiently large number of them and that one can apply simple hypotheses such as that of the continuum (mean field games modeling) and perfect competition (price-taker behavior of agents).

It's important to notice that the number of producers and hence the size of the continuum is fixed in our model. This may seem a very restrictive economic setting because it goes against the classical economic hypothesis of free entry. Though, it is natural in the field we consider since one cannot create an oil field as an entrepreneur would create a new firm. The number of wells is indeed fixed since we do not consider any exploration process.

Each of our oil producers initially has a reserve that is termed 0R and these reserves are

distributed among producers according to an (initial) distribution )(0,⋅m . These reserves will of course contribute to production q such that, for any given agent, we have dttqtdR )(=)( − .

Production choices will be made in order to optimize a profit criterion (the same for all agents) of the following form:

0)(0,)(..)))(()()((0,))(( ≥≥− −

∫ tRtqtsdsetqCtqtpMax rtT

Tttq

where: • C is the production cost function which we will then write as quadratic:

2=)(

2qqqC βα + .

• the prices )(tp are determined according to supply/demand equilibrium on the market at each moment, demand being given by a function ))(,( tptD 4 at instant t and supply naturally given by the total oil production of the producers.

2.1.2 Characterization of the equilibrium The problem we consider can be completely solved using classical Euler-Lagrange tools.

Let's consider the problem faced by a producer with reserve 0R . This problem can be summed up

by a Lagrangian (if we omit for now the condition on the positiveness of the production):

4The demand function for oil will take several forms in this text. A first form is a CES (Constant Elasticity of

Substitution) demand function : σρ −pEeptD t=),( . This form expresses that oil demand only depends on oil

price and demand grows by %σ when the price decreases by 1% . This form can be generalized to take in account the influence of different energies and substitution between them (see below). Further on, to set a variable maximum

price to oil, we may use the following form for the demand function: δσρ −−pEeptD t=),( ; a high δ standing

for a low maximum price.

145

0 00 0

= ( ( ) ( ) ( ( ))) ( ) ( )T T

rsp s q s C q s e ds R R q s dsλ− − + − ∫ ∫L

where )( 0Rλ is the shadow price of the reserve constraint 00=)( Rdssq

T

∫ . This shadow price will

play an important role in the following characterization of the solution: Proposition 2.1 (Equilibrium) The equilibrium is characterized by the following

equations where )(tpt , ),(),( 00 RtqRt and )( 00 RR λ are unknown functions and where

the levels of initial oil reserves are denoted by 0R .

0000 )(),(=))(,( dRRmRsqspsD ∫

[ ]+

−− rseRspRsq )()(1

=),( 00 λαβ

000=),( RdsRsq∫

∞

Remark: The first equation corresponds to the supply/demand equilibrium and defines )(tp . The

second equation characterizes the optimal scheme of production where )( 0Rλ is an unknown.

The third equation is the intertemporal production constraint, hence the constraint that ``defines'' the lagrange parameter )( 0Rλ .

Proof: Let's consider the problem of an oil producer with an oil reserve equal to 0R .

The optimal production levels can be found using the Lagrangian:

0 00 0

= ( ( ) ( ) ( ( ))) ( ) ( )T T

rsp s q s C q s e ds R R q s dsλ− − + − ∫ ∫L

The first order conditions are: rseRsqCspTs )(=))(()(, 0λ′−≤∀

rseRsqCsp )())((=)( 0λ+′⇒

and )()(=)))(()()(( 0 TqReTqCTqTp rT λ−−

Hence, using our specification for the costs, we get: rseRsqspTs )(=)()(, 0λβα −−≤∀

[ ] 0=)()()()( 0

rTeRTqTpTq λβα −−−

146

We deduce from this that 0=)(Tq or ∞=T . From the first condition we get that, before the end of the production, )(sq is given by

[ ]rseRspsq )()(1

=)( 0λαβ

−−

Hence, we can simply define the production function for all s by:

[ ]+

−− rseRspsq )()(1

=)( 0λαβ

In this equation )( 0Rλ depends on the initial oil stock (or reserve). This lagrangian multiplier is

given by the intertemporal constraint that equalizes the whole stream of production and the initial oil reserve:

( ) 00000=)()(

1=),( RdseRspdsRsq rsT

+

∞

−−∫∫ λαβ

Now, we need to find the prices that were left unknown. This simply is given by the demand/supply equality.

0000 )(),(=))(,( dRRmRsqspsD ∫

If we compile all these results we get the 3 equations that characterize the equilibrium. 2.2 Hotelling rent and Hubbert peak In the proof of the preceding proposition, we exhibited a relation between prices and

marginal cost. For a non-exhaustible resource, because of the pure competitive framework, marginal cost equals price at equilibrium. Here, there is a rent, called Hotelling rent, that makes the price larger than the marginal cost of production on the optimal trajectory. The formula has been given above and is:

rentHotelling

rteRtqCtp )())((=)( 0λ+′

Some remarks must be made on this rent. First the rent is increasing with time at a constant growth rate r . Second, the rent depends on the Lagrange multiplier. Because )( 0Rλ is the

shadow price of the constraint, the more important the constraint is, the larger )( 0Rλ is. Hence,

)(⋅λ is a decreasing function. This means that the rent is larger for a small producer than for a large producer.

Finally, we presented the rent in a competitive framework but it's important to notice that, in the general case, the rent is a function of the competition intensity. For instance, if one considers a monopoly, the price is usually the marginal cost times a markup. Here, both the marginal cost and the above (competitive) rent have to be multiplied by the usual markup to obtain the price. Hence, the rent of exhaustibility is in fact multiplied by the markup.

147

Now, if we look at the formula that expresses the evolution of the production, namely

[ ]+

−− rseRspRsq )()(1

=),( 00 λαβ

, we see that there are two forces at works. First, if we consider

that the price will end up growing, the production will grow with time. This is the first effect at work but the right part of the equation (the rent) pushes the production towards 0 and indeed it must be equal to 0 asymptotically. These two forces can lead to several evolution schemes for the production. Oil production can indeed decrease with time or increase first and then decrease towards 0 . The latter case is similar to what is usually termed Hubbert peak, though we are dealing with wells whose reserve is fixed at start.

2.3 Computation of an equilibrium. Eductive methods Now that we have presented the model, we must consider the problem of finding a

solution numerically in the general case. To solve this problem let's go back to the characterization of the equilibrium given in the above Proposition. We see that ),(),( 00 RtqRt

only depends on )( 00 RR λ and )(tpt . Hence we can totally separate the variables t and

0R . More precisely, if we consider an eductive algorithm (eductive algorithms will be used later

to solve coupled partial differential equations) we can consider two ``guesses'' )(⋅λ and )(⋅p to compute ),( ⋅⋅q and then update )(⋅λ and )(⋅p using respectively the constraints

000=),( RdsRsq∫

∞

and 0000 )(),(=))(,( dRRmRsqspsD ∫ .

More precisely, if for a given t , the price )(tp does not verify the supply/demand

equilibrium equation, let's say 0000 )(),(>))(,( dRRmRtqtptD ∫ , then the price has to move (it

must go up in the case considered to limit the excess of demand). Hence, we can consider adding a time θ and define a function ),(),( θθ tpt by an initial function (called initial guess)

,0)(tpt and

0000 )(),,()),(,(=),( dRRmRtqtptDtp θθθθ ∫−∂

Similarly, we can update an initial guess for λ and we get the following equation

0000 ),,(=),( RdtRtqR −∂ ∫∞

θθλθ

where we recall that

[ ]+

−− rteRtpRtq ),(),(1

=),,( 00 θλαθβ

θ

In fact the idea is more general. We can indeed consider a system of integral-differential equations for any two increasing functions pΨ and λΨ :

148

( )

[ ]

−−

−Ψ∂

−Ψ∂

+

∞

∫

∫

rt

p

eRtpRtq

RdtRtqR

dRRmRtqtptDtp

),(),(1

=),,(

),,(=),(

)(),,()),(,(=),(

00

0000

0000

θλαθβ

θ

θθλ

θθθ

λθ

θ

Considering such functions pΨ and λΨ or replacing the variable on which we elucidate (for

example, on may want to consider )),((ln θθ tp∂ instead of ),( θθ tp∂ ) is important to better

control the size of the moves in λ and p . Once a system is chosen and seems to converge in practice, the idea numerically is to

obtain the solution for )( 00 RR λ and )(tpt and hence the productions of all oil producers

using the following limits: )(=),(lim tptp θ

θ +∞→

)(=),(lim 00 RR λθλ

θ +∞→

As an example we can illustrate the evolution of total oil production in this model where we consider a CES demand function, namely σρ −pEeptD t=),( .

We took the following values for the parameters: the interest rate considered by oil producers is 5%=r , the average growth rate of the world economy is 2%=ρ , the initial marginal cost of producing an oil barrel is 10=α , 100=β to model the importance of capacity constraints, 1.2=σ because oil is not a highly elastic good and 40=E to obtain meaningful values in the model. The problem is considered over 150 years and the initial distribution of reserves has the following form:

149

Figure 1: 0m

If we consider the global production of oil producers, its evolution is plotted on Figure 2

and the associated evolution of oil prices is given by Figure 3 where we only plot the first 50 years to avoid ending up with very large values after too many decades and hence a graph that is unreadable.

150

Figure 2: Evolution of the total oil production

151

Figure 3: Evolution of prices over 50 years

3 The MFG framework Although we did not emphasize on this point, the preceding model was a mean field game

as any general equilibrium economic model. In the simple case developed above, the mean field games tools did not need to be used and classical tools were sufficient. However, when it comes to noise or externality in the model, the mean field games partial differential equations will be necessary (see Guéant et al. (2009), Giraud et al. (2009)).

3.1 The mean field games partial differential equations Let's rewrite a little bit differently the profit optimization criterion of a given agent:

dttqtdRtRtqtsdsetqCtqtpMax rt

ttq )(=)(0,)(0,)(..)))(()()((0))(( −≥≥− −∞

∫

We associate a Bellman function ),( Rtu to this criterion:

dsesqCsqspMaxRtu tsr

tqtssq)(

0,))(( )))(()()((=),( −−∞

≥≥−∫

0)(,,=)(,)(=)(.. ≥≥∀− sRtsRtRdssqsdRts

The Bellman function verifies the following Hamilton-Jacobi equation:

152

( ) 0=),()()(max),(),()(0

RtuqqCqtpRtruRtuJacobiHamilton Rq

t ∂−−+−∂−≥

The Hamiltonian of this problem is ( )),()()(max 0 RtuqqCqtp Rq ∂−−≥ . If we keep considering

the quadratic cost function that has been used so far, then the optimal control is given by:

+

∂−−

β

α ),()(=),(* Rtutp

Rtq R

where ),(* Rtq represents the instantaneous production at time t of a producer with an oil reserve R at time t . It's important to notice that R designates the reserve at time t and not the initial reserve as in the resolution presented above.

The Hamilton-Jacobi equation can be rewritten using the optimal production:

( )[ ] 0=),()(2

1),(),( 2

+∂−−+−∂ RtutpRtruRtu Rt α

β

Now, let's denote ),( Rtm the distribution of oil reserves at time t . This distribution is initially

given by a function )(0 ⋅m and then transported by the optimal production decisions of the agents

),(* Rtq . The transport equation is: 0=)),(),((),()( * RtmRtqRtmKolmogorov Rt −∂+∂

with )(0,⋅m given. These two equations are the two partial differential equations of the mean field games

theory when there is no randomness. The Hamilton Jacobi equation is a backward equation whereas the transport equation is a forward one. The link between the two is the double coupling. As usual, these equations are indeed coupled but here the double coupling is not obvious at first sight. First, m depends on u through the optimal production function *q and second, the MFG coupling is characterized by prices. In the Hamilton-Jacobi equation, the Hamiltonian depends on the price function )(tpt and we have to recall that prices are fixed by a global (as opposed to individual) equilibrium between supply and demand. Since supply depends on the production of all producers, it depends on the distribution of reserves amongst production: m .

Mathematically, )(tp is fixed so that supply and demand are equal and hence:

−⋅ ∫−

supply

dRRtRmdt

dtDtp ),(),(=)( 1

Hence, the coupling is indeed double: m depends on u and u depends on m . The type of interdependence involved here is not that common since the coupling happens into the Hamiltonian and not as a second member of the Hamilton-Jacobi equation.

If we want to sum up and rewrite the equations to focus on the interdependence, we may write the following Proposition:

Proposition 3.1 (Mean field games PDEs) The two partial differential equations associated to our oil problem can be written:

),(),( RtruRtut −∂

153

0=),(),(),(2

12

1

∂−−

−⋅+

+

−

∫ RtudRRtRmdt

dtD Rα

β

and

0=),(),(),(),(

),(

1

∂−−

−⋅

−∂+∂

+

−

∫Rtm

RtudRRtRmdt

dtD

RtmR

Rtβ

α

We can solve these PDEs (see Giraud et al. (2009) for the methodology) for the same

parameters as before apart from the demand function (we take δσρ −−pEeptD t=),( with 40=E as before but5 0.1=δ ) and we have the following evolution for the total production and

for the price6:

Figure 4: Evolution of the total oil production

5This value for δ sets a maximum price around $150 today, this maximum price increasing at rate σρ/ .

6Price is not represented after the end of production and equals the (non constant) maximum price corresponding to

0.1=δ

154

Figure 5: Evolution of the price )(tp

We see that, because of the the maximum price, producers are incited to produce faster

than before.

4 The two-energy case In this section, we consider that there is, in addition to oil producers, potential entrants on

the market. These entrants produce an imperfect substitute to oil but they use a renewable source of energy. Typically, we have in mind the competition between fossil energy and electricity to fuel cars. The reserves of oil are finite whereas electricity can always be produced (or at least there is no such reserve problem for the electricity producers). However, these two products are not exact substitute because autonomy is not the same in both case for instance. For this reason, the two types of energy producers (oil producers and alternative energy producers) may produce simultaneously.

One of our goal here is to understand when alternative energy producers will enter the market and to discuss the usefulness of a subsidy to induce earlier entry.

4.1 Demand function To model demand on the energy market we are going to use a classical economic setting

for demand functions: if we denote oil by 1 and the alternative energy by 2 then we can postulate the following demand functions:

η

σρ

−

−

enen

t

p

ppEepptD 1

211 =),,(

η

σρ

−

−

enen

t

p

ppEepptD 2

212 =),,(

155

where ση > and where enp stands for a reference price for energy:

( ) ηηη −−− + 1

112

11= pppen

These demand functions are justified by CES utility functions (see the box below) and allow us to factor in three effects:

• A wealth effect, through tEeρ . As before we have an exogenous growth of the economy at constant rate ρ .

• A substitution effect between energy and other goods which is modeled by the term σ−

enp . σ represents here the elasticity of demand for the energy aggregate.

• A substitution effect between oil and the alternative energy. This effect is modeled by η−

enp

p1 and η−

enp

p2 where we see that demand for one or the other energy depends on the

relative price of one energy compared to the reference price for energy. Here η stands for the elasticity of substitution between the two types of energy.

Box 1. Microeconomic foundations of the demand functions: To justify these demand functions let's consider the general case where we have a

continuum of sectors denoted by i and for each sector a continuum of products denoted by j . The CES utility function associated to this framework is:

didjXXU jiji

ji

σ

σ

η

η

η

η

1

11

, =))((

−

−−

∫∫

Hence if we use the aggregate 11

=−−

∫

η

η

η

η

djXX jii we see that the program of an agent with

wealth E is characterized by the first order conditions:

jiij

i

i pXdX

dXji λσ =,, 1/−

∀

jii

ji pXX λσηη

=1/1/1/ −−⇒

Hence, we have

djXpXdjX ji

jii

ji ∫∫

−−

λση

η

η

=1/1/1

djXpX ji

jii ∫

−⇒ λσ =1/1

Also we have

ηηηση

η

η

λ−−−−

−11))(11/(1/

1

= jii

ji pXX

156

djpXX jiii

ηηησηη

η

λ−−−−

−

∫⇒11))(11/(1/

1

=

η

ηησ

η

λ

−

−−−

∫⇒

1

111

=

iP

jii djpX

where iP is the aggregated price in sector i . Hence:

ii PX λσ =1

−

The real significance of iP is then given by the combination of our two relations above:

iij

ij

i XPdjXp =∫ . Now, σσσ λλ −−−

⇒1

1

== iiiii PXPPX .

Hence

σ

σσλ

−

−−

∫1

1=

P

i diPE where P is a global price level.

Combining all our results and going back to jiX , we see that:

ησ −−

i

jiij

i P

p

P

P

P

EX =

and this is exactly what we considered. Now that we have the demand part of the problem, let's turn to the supply side. 4.2 Program of the entrants We suppose that there is a club of potential entrants that collude to decide when they enter

the market. However, we suppose that once they entered the market, they cannot collude on price and that this continuum (of fixed size, here 1) is characterized by a perfect competition hypothesis (although there is no entry). This idea of collusion on entry and not on prices can be justified by large fixed costs. A large number of entrants decide to merge as a club to pay for the network infrastructure that is necessary to enter the market. However, once they have paid the fixed cost to enter, they have no reason to collude and a price competition leads to a perfect competition case.

Mathematically, each potential entrant (they are identical to simplify) faces a program of the following form:

( ) dtetqCtqtpeFe rt

T

rTT

qT

−∞

−

⋅

−+− ∫ ))(()()(max 2222)(2,

π

where the notations are similar to those used for oil producers, though we suppose that the

parameters of the cost function 22 2

=)( qqqCβ

α + are not the same as for oil to model a cheaper

or a more expensive technology and also to model, in some sense7, the idea of different capacity constraints in the two sectors.

7This is one of the roles played by the quadratic term

157

F here is a sunk cost to enter the market and this cost evolves at rate r<π to model inflation. T is the time to enter the market. Because of collusion, T chosen individually will be the same as T chosen by the club and that's the reason why each agent optimizes on T .

We can characterize the time at which entrants arrive on the market: Proposition 4.1 (Time of entry) Entrants turn up at time T when the post-entry price

after time T of their alternative energy verifies:

TFerTp ππβα )(2=)( 222 −+

Proof: First of all, once the entrants are on the market, the price )(2 tp is simply given by the

marginal cost of production. Hence, production is given by 2

222

)(=)(

β

α−tptq . Consequently, the

first order condition for T is:

222

2

)( ))((2

1=)( α

βπ π −− −−− TpeerF rTTr

TFerTp ππβα )(2=)( 222 −+⇒

There can a priori be several such T . In the simulations we are going to present later on,

uniqueness will be true empirically. 4.3 Analysis of the model We can consider any of the preceding oil production models and plug it into the 2-energy

framework. All the resulting models are of the same kind when it comes to analyzing the impact of potential entrants.

We have seen before that entrants decide to enter the market when the post-entry equilibrium price8 of their alternative energy is sufficiently high. Oil producers know this rule and adapt their strategies to this new context. Since entrants steal a part of the energy demand to oil producers, oil producers may want the entrants to arrive on the market as late as possible. To do that, they have to produce less than it would be the case if they were alone: using this strategy they keep enough oil in the ground so that, after an hypothetical entry, prices are low. This effect is especially relevant when a monopoly produces oil instead of a competitive continuum because oil producers keep reserves high to threaten the potential entrant of massive production after entry; a massive production that induces low prices and hence deter entry. On the other hand, and this is particularly relevant when agents are atomized and hence cannot play complex strategies that directly influence prices, each producer may want to get rid of his oil reserve, not to suffer from the competition of the entrant. In practice, there is a trade-off between these two effects and we can only analyze what happens on numerical examples.

4.4 Numerical results and effect of a subsidy

8An uncorrect reasoning would be to consider the prices of oil just before entry.

158

Let's consider that there is no inflation and that the two sources of energy are

characterized by (10,100)=),( 11 βα for oil and (50,50)=),( 22 βα for the alternative energy. If the fixed cost to enter is equal to 50 then the evolution of oil production is the following:

Figure 6: Evolution of the total oil production

We see that the second effect dominates in the above trade-off. Because of their rational

expectations, oil producers know when their competitors are going to enter and they individually find it optimal to get rid of their oil reserves more rapidly than before. Hence, the very presence of potential entrants implies an increase in production and hence an increase in pollution for the first 40 years in our example.

This could raise the idea of a subsidy to the alternative energy producers in order to diminish this side effect that may seem absurd since, in some sense non-polluting alternative energies... pollute.

Let's for instance consider the case of 50=F and then consider a subsidy of 25 to reduce F to 25 . We obtain the following evolution for oil production:

159

Figure 7: Impact of a subsidy from 50=F to 25=F

Here, the difference of pollution between the two cases 50=F and 25=F is very low during the first 30 years, though the subsidy induces higher pollution, but the pollution last a shorter period of time and a subsidy is arguably welfare-enhancing for any reasonable welfare function that model the damage or disutility of pollution.

Even better, when the alternative energy is cheap to produce (that is 10=2α instead of 50 ), a subsidy may remove the period of high production as on the following example where F goes from 100 to 50 thanks to the subsidy:

160

Figure 8: Impact of a subsidy from 100=F to 50=F when the alternative energy is cheap to

produce However, this welfare-enhancing effect of a subsidy may not exist for example if the

fixed cost is really high. Imagine for example that the subsidy may not be higher than 100 because of a budget constraint. Then, if 250=F , in the case where 50=2α , a subsidy may be very harmful in terms of pollution for the next 60 years or so because the induced increase in pollution is really high as below:

161

Figure 9: Impact of a subsidy from 250=F to 200=F or 150=F

4.5 Conclusion on the two-energy case 4.5.1 Disutility of pollution This two-energy case allowed us to discuss briefly the impact of a subsidy to alternative

energy producers. So far, we just reasoned on graphs and the reader may not be convinced by our arguments on the benefits or harmfulness of the subsidy. At the end of the day, one may argue that, since all the oil reserves will be exploited the overall cumulated pollution will be the same in all circumstances. However, if we introduce a disutility function for pollution with a reasonable discount rate, it's often better to postpone production of oil and/or to smooth the gas emissions.

Let's consider, as an illustration, a natural family of disutility functions for pollution. Since pollution in the atmosphere at time t is the result of past production, we can write pollution at time t as:

dsstqf s

t

t ))((exp)(= −−Π ∫ ∞−λ

where f stands for the pollution production function and where λ models the fact that pollution does have a half-life ( λ is arguably very small). Hence, a disutility function for pollution may be

162

dttU t )(exp=0

ξ−Π∫∞

or equivalently9, in the sense of disutility functions, dttqf t )(exp)(0

ξ−∫∞

. In

these equations, ξ is the discount rate to be applied to environmental goods (see ?, ? for more details on ecological discount rates).

Hence, if we consider a linear function of production for pollution and various discount rates ξ , we can compare the disutility figures associated to the different graphs presented above.

First we can write the reference case where there are only oil producers:

Discount

rate ξ Disutility (oil only)

0 29.49 0.5% 23.95 1% 19.86 2% 14.39 5% 7.29

Now, if we consider the case presented in Figure 7, the following figures10 show that the

subsidy indeed improves welfare in the sense that it reduces disutility for several admissible values of the discount rate (the values only have an ordinal sense and the difference between two figures has no meaning apart from its sign.).

Disutility Discount

rate ξ

50=50,=2 Fα

25=50,=2 Fα 0 29.47 29.47

0.5% 24.60 24.39 1% 20.82 20.50 2% 15.49 15.12 5% 8.02 7.83

On the contrary, and as announced in the comments regarding Figure 911 a subsidy of 50

or 100 is not welfare-enhancing when the alternative energy is relatively expensive and characterized by high fixed cost of entry:

9We permute the integral signs and consider the production to come ( 0≥t ).

10The figures for 0=ξ are supposed to be always the same. The differences that occur are relatively small and

linked to numerical errors in the calculations of the optimal production. 11

The results for Figure 8 are the following:

163

Disutility Discount

rate ξ

250=50,=2 Fα

200=50,=2 Fα

150=50,=2 Fα 0 29.48 29.47 29.48

0.5% 24.66 24.71 24.74 1% 20.87 20.95 21.00 2% 15.45 15.55 15.63 5% 7.83 7.90 7.97

4.5.2 Origin of the subsidy Our model is arguably simple because we do not explicit where the subsidy is coming

from and we do not consider for instance a subsidy financed by taxing oil production as it would be the case in a ``carbon tax''-like mechanism. However, as we mentioned it earlier, our model is consistent with either an exogenous subsidy, or a subsidy financed through a tax on oil producers' profits. Our conclusion should be hence understood as follows: a tax on oil producers profits to subsidize alternative energies may be harmful in terms of pollution when the alternative energy is costly to produce and need important initial infrastructure investment (the subsidy being understood as a way to partly finance the said infrastructure - network infrastructure for instance). Even though our conclusion may be discussed in a different setups if producers have a market power or if the subsidy comes from a tax on oil prices, the non-classical pollution effect we exhibit invites to reconsider the welfare effects of a subsidy to alternative energy producers, especially when the alternative energy is costly to produce compared to oil.

Appendix : A brief overview of mean field games theory Mean field games (MFG) have been introduced by Jean-Michel Lasry and Pierre-Louis

Lions (see Lasry and Lions (2006a,b; 2007a,b), Lions (2008)) as a set of tools to model games

Disutility

Discount rate ξ

100=10,=2 Fα

50=10,=2 Fα

0 29.46 29.46 0.5% 23.46 23.12 1% 19.09 18.53 2% 13.38 12.61 5% 6.54 5.65

164

with infinitely many players. These games are most often going to represent economic problems with a large number of agents and the mean field games tools will help solve the problem.

Mean field games theory does not introduce a new paradigm but rather what could be called a new ``version'' or, to quote the software industry, it is a new ``release'' of the existing paradigm in economics (utility maximization, rational expectations, ... ) and this new release is fully compatible with the preceding ones. As the software specialists put it: all the ``functionality'' of the previous ``release'' will still be there, and can be re-used in the context of the ``update''. Moreover, in this ``MFG version'', modeling can not only be re-applied as such, but it can be modified, extended and reworked to take advantage of the new ``functionalities'' provided by mean field games, i.e. by the capacity to introduce into the models the effects of social interaction between agents, which cannot be integrated into the framework of classical modeling in economics.

This compatibility with the classical paradigm of economics is seen in the principles that mean field games re-adopt, in relation to defining (static or dynamic) equilibrium: rational expectations, invariance by permutation of similar agents, atomization of agents in a continuum to mention the classical principles found in a large part of economic theory (though not in all economic theory: in industrial economics, agents are not always permutable since names matter).

Another way of putting it is that mean field games constitute a mathematical toolbox that enables us to model effects which were sometimes ignored for lack of having the analytic and numerical tools needed to handle them. From the standpoint of such a toolbox, the new contribution of mean field games is given concrete expression in the mean field games system of HJB/Kolmogorov-type forward/backward equations that characterize equilibria. Let us now briefly explain what this is.

In any (quantitative) modeling, there must always be an equation to express the optimization problem of each agent. Usually this involves one equation for each agent. If agents are grouped together by similar agent classes, there is one equation per agent class. This equation is generally a Bellman equation, since a large proportion of optimization problems fall within the framework of dynamic programming and of the various sophisticated forms of deterministic and stochastic control that are constructed and inscribed within the field of dynamic programming. Hence, Hamilton-Jacobi or the Hamilton-Jacobi-Bellman equations will be used to compute optimal behaviors.

An equation is also needed to express the group's behavior. When agents are atomized, the group is represented in the modeling by a distribution on the state space; or, if there are several agent classes, by a distribution for each class. The dynamics of the distribution is governed by a transport equation that can be called a Kolmogorov equation. In this transport equation, the optimal behaviors of agents occur as data, since it is the infinite collection (the continuum) of individual behaviors that is aggregated and which, thus aggregated, constitutes collective behavior.

Thus, the modeling of a the behavior of a group of agents naturally leads to an HJB/K (Hamilton-Jacobi-Bellman and Kolmogorov) equations system. This HJB/K system is not new: it forms part of previous ``versions''. Admittedly, for technical reasons, it has not been much used. This is because in the case where the state space is continuous, a system of non-linear partial differential equations is not for the moment an attractive tool for most modelers (apart from specialist mathematicians). Most of the time, modelers prefer to circumvent this formalism by the various mathematical techniques that are available in each specific case. However, some economists make excellent use of it for a few specific models.

165

But we have not yet answered the central questions: what is new in mean field games formalism? Essentially one thing: the density of agents on the state space can enter in the HJB equation.

Its novelty lies here. MFG equilibrium is defined by an HJB/K couple of PDEs in which the HJB and Kolmogorov equations are doubly coupled: individual behaviors are given for the Kolmogorov equation (this is not new) and, at the same time, the distribution of agents in the state space enters in the HJB equation (which is new).

This means that agents can incorporate into their preferences the density of states of other agents at the anticipated equilibrium. Therefore each agent can construct his strategy by taking account of the anticipated distribution of strategies and of the contingent decisions of other agents. Thus formulated, MFG equilibria can be seen as similar to another paradigm, that of Nash equilibria and more specifically of Isaacs-Bensoussan equilibria in n -player stochastic differential games. In fact, the MFG equilibria can be defined by moving to the limit on the number n of players in the class of differential games that are invariant by permutation of similar agents. It is this invariance by permutation (accompanied by a little continuity in the preference functions) which makes evident the distribution of agents in the space state as the object representing other agents for each atomized agent in the move to the limit.

Of course in games with a small number of players, permutation invariance is not a useful hypothesis: there are never two identical players around a poker table. Conversely, permutation invariance is very natural in most studies of large groups, especially in economics. On the other hand, the atomization of agents simplifies the strategic interactions: coughing on the part of one player can have major consequences in poker, but has no significance in a group of several thousand agents. This is why MFG equilibria equations are much easier to solve than Isaacs-Bensoussan equations. In other words, there is some hope of solving mean field games systems, and we can already do so for large classes (and we possess the feasible numerical techniques), whereas the equations of n -player differential games are in general not tractable since they describe situations in which the combinatory of strategic interactions exceeds calculation capacities.

These general considerations have been illustrated in this paper through application to some economic issues about the production of an exhaustible resource.

Indeed, this application of mean field games returns to the initial concerns that were present in the creation of mean field games that is finding new mathematical tools to handle modeling questions in the area of sustainable development, and therefore in particular in regard to exhaustible resources: one of the missions given to the Calyon-EDF Chair by its founders.

References

[1] Robert Aumann. Markets with a continuum of traders. Econometrica, 32(1/2), 1964.

[2] Ivar Ekeland. Le pétrole sera-t-il bradé ? Pour la Science, 2007.

[3] Olivier Guéant, Roger Guesnerie, and Jean-Michel Lasry. Ecological intuition versus economic reason. mimeo, 2009.

[4] Pierre-Noël Giraud, Olivier Guéant, Jean-Michel Lasry, and Pierre-Louis Lions. A mean field game approach to oil production. mimeo, 2009.

[5] Olivier Guéant, Jean-Michel Lasry, and Pierre-Louis Lions. Mean field games and applications. in Paris-Princeton Lectures in Quantitative Finance, 2009.

166

[6] Olivier Guéant. Mean Field Games and applications to economics. PhD thesis, Université Paris-Dauphine, 2009.

[7] Olivier Guéant. A reference case for mean field games. Journal de Mathématiques Pures et Appliquées, 2009.

[8] Roger Guesnerie. Calcul économique et développement durable. Revue économique, 2004.

[9] H. Hotelling. The economics of exhaustible resources. The Journal of Political Economy, 39(2), Apr. 1931.

[10] Pierre-Louis Lions. Théorie des jeux à champs moyen et applications. Cours au Collège de France, http://www.college-de-france.fr/default/EN/all/equ_der/cours_et_seminaires.htm, 2007-2008.

[11] Jean-Michel Lasry and Pierre-Louis Lions. Jeux à champ moyen i. le cas stationnaire. C. R. Acad. Sci. Paris, 343(9), 2006.

[12] Jean-Michel Lasry and Pierre-Louis Lions. Jeux _a champ moyen ii. horizon _ni et contrôle optimal. C. R. Acad. Sci. Paris, 343(10), 2006.

[13] Jean-Michel Lasry and Pierre-Louis Lions. Mean field games. Japanese Journal of Mathematics, 2(1), Mar. 2007.

[14] Jean-Michel Lasry and Pierre-Louis Lions. Mean field games. Cahiers de la Chaire Finance et Développement Durable, (2), 2007

167

PART THREE. MODELS

AND APPLICATIONS

The third part of the book contains four contributions. Each of them puts forward a model

and its application to a specific field: water resources, gas emissions, and a sustainable

development path. All the chapters in this part share a common concern: that of confronting

theory with economic reality. Such a task is not easy, as sustainable development is subject to

huge uncertainties.

To quantify the risks associated with uncertainties in water resources systems, Bensoussan

and Fahri present a model that relies on real options’ theory. In this model, the value of a

desalination project depends on three variables: the demand for water in the region covered by

the desalination plant, the supply of water provided by other facilities, and the cost of

desalinating water. The authors solve a stopping time problem in order to determine the optimal

time for investing in the project and obtain a decision rule which determines whether or not it is

advantageous to invest.

The aim of the article by Valentina Bosetti, Carlo Carraro, Emanuele Masseti and

Massimo Tavoni is to investigate the economic implications of stabilizing greenhouse gas

concentrations over the next century. In order to analyze the complex geographical and

intertemporal interactions of the main socio-economic, technological and climatic variables

influencing the concentration of carbon in the atmosphere, and with the aim of obtaining a

quantitative assessment of public policies, the authors use a hybrid climate-energy-economy

model. This model is designed to identify the optimal investment profiles in the energy sector and

in research and development for achieving a pre-determined carbon concentration target. The

framework selected is game theory.

168

Samuel Fankhauser’s contribution also focuses on greenhouse gas emissions and provides

an order-of-magnitude assessment of the potential impact of global warming. He uses a stochastic

model and incorporates uncertainty directly by describing uncertain parameters as random. This

approach allows a better approximation to current scientific understanding and also enables an

entire damage probability distribution to be calculated, thereby providing important additional

information on the likelihood of the estimates and the possibility of extremely adverse events.

In their contribution, Pierre-Noël Giraud and Timothée Ollivier estimate the variation in

the productive base of a particular economy – that of Mozambique – between 2000 and 2005.

They define an economy’s productive base as the set of different capital stocks – produced,

human, social and natural capital – and argue that a development path is sustainable as long the

society’s productive base does not shrink. On the basis of an extended dataset, they conclude that

Mozambique, in contrast to many other sub-Saharan countries, is currently following a

sustainable path.

169

Uncertainties and Risks in Water Resources Management

Alain Bensoussan & Nadir Farhi

Abstract: The purpose of this paper is to consider the issue of uncertainties in water

resources systems, and to introduce approaches for quantifying associated risks. We try to present various sources of uncertainties existing in real situations and develop related risk management methodology. A special emphasis is given to desalination, considered as a promising response to scarcity of water in many locations. We present the advantages and the disadvantages of the desalination in terms of risks and uncertainties, and we introduce the evaluation of investment opportunities in such projects, by solving an optimal stopping time problem which gives the optimal time to invest in a desalination project.

Keywords: Water Resources, Risk Management, Water stress, Uncertainties, Water and

Finance

170

1. Introduction One of the most important humanity's challenges in the future, is surely ensuring its water

needs, including domestic, industrial, agricultural, and recreational uses. Equitable and sustainable management of water resources is a major global challenge (United Nations Environment Programme). Halfway to 2015, the year when the globally agreed Millennium Development Goals are supposed to be reached, the crisis in water and sanitation remains among the great human development and environmental challenges.

Knowing that 88.7 percent of water on the Earth is salt water and two thirds of fresh water is frozen in glaciers and polar ice caps, it remains only 0.9 percent available for human use. Fresh water is a renewable resource, but the supply of clean fresh water is decreasing. Water demand exceeds supply in many parts of the world. As world population continues to rise, many more areas are expected to experience this imbalance.

In this paper we are interested by uncertainties and risks appearing in water resources systems, and the appropriate approaches susceptible to be applied in this context to mitigate them.

The paper is presented in four sections including this introduction. In the second section we identify uncertainties and associated risks related to water resources management (risks in investments, risks caused by drought situations, risks in natural disasters and terrorism,...). The third section deals with desalination. The purpose of this section is to identify the economic and environmental considerations related to desalination, as well as the uncertainties and the risks in desalination projects.

In the last section we evaluate opportunities of investments in desalination projects, by describing what can be expected from applying real options theory. We propose a model of investment in a desalination project built on three main processes which are the demand in water on the region concerned by the desalination plant, the supply of water assured by other facilities, rather than desalination, present in the region, and the cost of desalinating water supposed depending mainly on the price of energy. We solve a stopping time problem to determine the optimal time to invest in the desalination project. The solution is a decision rule telling us, at each time, wether to invest or not, basing on information taken from the values of the considered processes at this time.

1.1. Sources and Uses of Freshwater Freshwater is the water containing low concentrations of dissolved salts and other total

dissolved solids, excluding seawater and brackish water. It can also be the output of desalinated seawater. Numerically, it is defined as water with less than 0.5 parts per thousand dissolved salts.

There are various sources of fresh water. We distinguish surface water, groundwater, desalinated seawater and frozen water. The ultimate source of freshwater is the precipitation of atmosphere in the form of rain and snow. There exists Consumptive uses and non-consumptive or renewable uses of fresh water. Water is used in a consumptive way when, after the use, it is not available for another use. It is lost to sub-surface seepage, evaporated, or incorporated into a product. Water is used in a non-consumptive way if it can be treated and returned as surface water. We distinguish irrigation uses (which represent 69 percent of world-wide water use), industrial uses (15 percent), household uses (15 percent), and recreational uses.

171

1.2. Economic and Financial Considerations Water supply and sanitation require a huge amount of capital investment in infrastructure

such as pipe networks, pumping stations and water treatment works. OECD nations need to invest US$ 200 billion per year to replace aging water infrastructure. To meet the 2015 goals, developing countries should spend US$ 30 billion per year, beyond the investment for the maintenance of the existing infrastructure. Operational costs (personnel, energy, chemicals, maintenance and other expenses) are significant. The sources of funding are public funds and user fees.

Water crisis affects all types of business which cannot survive in a society that thirsts. Business is part of the solution but growing water issues and complexity will drive up the costs.

Financial Aspects Water related risks have not strongly appeared on the banks and financial institutions

radar screens, but this state of affair is likely to change. There are several reasons for such a change.

First, large investments are obviously necessary in order to meet future demands and declining supplies. Even though the public sector plays the main role in financing, eventually financing comes from taxpayer money. It affects this generation, but future generations even more. New financial tools have to be created, taking into account the long range planning of water resources infrastructures.

The second reason is that industry will be an increasingly important player in Water Resources Management. More private concessions for water systems will appear, and industry will have to invest in projects which are risky, because cash flow is obtained for long periods of time, and is affected by many uncertainties. Industry will have to invest in R&D to be more efficient in water treatment and in waste-water treatment. However, new opportunities will also arise, a good example is the fast progress of desalination plants.

The third reason is that banks, financial institutions, and industry understand that if water becomes a serious problem, then economic activity will be stalled. Industries will have to show that they save water, and banks have an important responsibility through their loans to make sure that their customers have an ethical behavior as far water matters are concerned.

Water is a part of the general trend to protect environment, but its relative importance will grow fast. We can already see an active program of research in the finance sector to consider the evolution of commodity markets. One cannot exclude a similar evolution for water. Concerning investments for new water resources infrastructure, one can see similarities with other sectors where an important funding is needed in early phases, with a long term return.

Some interesting approaches have been introduced to deal with uncertainties in this context. They concern both decisions on investment and organization of the development phase. The water sector can certainly benefit from these innovative approaches. What is needed is flexibility. The more uncertainty , the more flexibility is needed. Indeed, if decisions are taken in early phases, when little knowledge is available, then incorrect decisions may be made, resulting in large financial losses.

Several barriers and disadvantages exist in financing water infrastructures. This financing is in general very capital-intensive with assets unusable for other purposes and not removable.

172

Moreover, we have to support large negative cash flow in the initial years, and low rates of return later, and resist to tariff increases.

2. Uncertainties and Risks in Water Management The consequences of the climate change such as droughts are a source of a serious stress

on populations and on the economy. Moreover, huge and numerous uncertainties prevail, which entails difficulties in long term planning.

The increasing complexity of a water system can have a cascading effect on the consequences of uncertainties. This complexity is well felt by specialists. An evident example is that the 2007 State Water Plan of Texas (U.S.A.) contains 4,500 water management strategies. Texas Water Development Board (TWDB) has selected, as a research priority, risk and uncertainties in water resources management.

Different categories of uncertainties can exist in a water resources system. Hydrologic uncertainties are natural uncertainties such as stream flows and rainfall. They carry model uncertainties and parameter uncertainties when they are described by models. Hydraulic uncertainties concern the design and the analysis of performance of hydraulic structures, such as model uncertainties for the hydraulic structure and the flow processes. Structural uncertainties are physical failures or operational uncertainties such as water saturation, loss of soil stability, erosion or hydraulic soil failures, wave action, hydraulic overloading, structural collapse, or material failure. For example, structural failure of a levee system are either in the levee or in the adjacent soil. Economic uncertainties are generated by construction costs, damage costs, projected revenue, operation and maintenance costs, inflation, inconvenience losses, ...etc.

Risk Assessment in Water Management Many risk analysis approaches in water management exist. Among these approaches, the

risk-based design of water resources projects integrates procedures of economics, uncertainty analysis, and risk analysis. It incorporates also tradeoffs among risk, economics, and other performance measures in hydraulic structure design. Optimal risk-based design refers to risk-based design embedded into an optimization framework.

At the system-level, the diversity of risks necessitates defining a taxonomy. The physical factors taken into account are the structural components (dams for example), the operating components (turbines, flood gates, ...), and the flow components (water). The logical factors (software and human intervention) are the agility of the system which is its capacity to efficiently adapt to changing conditions, its self synchronization which is its ability to organize and synchronize complex activities from the bottom up. The environmental factors (external factors that can affect the system) are the climate (weather, natural hazards, ...) and the human interaction (incorrect responses, malevolent behavior,...). Risks are present at different levels of water systems including risks in failure, risks in natural disaster and terrorism, operational risks for water infrastructures, risks caused by drought situations, risks in investment in water projects,...etc.

The issue of droughts

173

There are many definitions of the drought. Meteorological drought is a period of lower than normal precipitation. Soil moisture/vegetative drought affects plants, wildlife, and crops. Hydrological drought refers to lower stream flows, lake levels, and water levels in aquifers. Droughts are one of the major causes of water shortages. They are not easily predictable, but they occur more and more frequently.

Some research, such as tree ring counting, is already done to develop better drought frequency and magnitude estimates. One way to mitigate the consequences of droughts is to develop drought contingency plans. They have been requested since 1997. Drought contingency plans anticipate short-term water shortages and water supply emergencies, evaluate risks and impacts of drought emergency conditions, identify measures and management practices to mitigate risks and impacts of drought-related, shortages and emergencies.

Developing drought contingency plans is certainly an appropriate approach because one can have an accurate anticipation of the impacts and therefore response measures can be planned. Nevertheless, the uncertainties in droughts are far from being addressed completely by the present strategies. Independently of the uncertainty on the beginning of a drought, there is a large uncertainty on its duration. The intensity is highly variable and depends on the geography. So a statistical approach is necessary. If the impacts are known in the short term, there remain disagreements about the ecological effects.

Mitigation and contingency plans require coordination of many stakeholders, which adds a political complexity dimension to the technical complexity. Contingency plans are also addressing primarily the question on how to react to drought after its onset, with emergency programs of allocation, restrictions, and interim supply arrangements. More research is needed on the proper preparation to prevent these situations.

The key question here is to assess the probability that the water supply system will fail, which means will run out of water. In order to address this issue, the concept of safe yield has been introduced by experts. In the 2007 State Water Plan of Texas, firm yield is defined as the maximum water volume a reservoir can provide each year under a repeat of the drought of record. Safe yield is defined as the firm yield in addition to an amount of water supply for an additional period of time. However, more research is needed to define and use these concepts in a complex system of streams and reservoirs. In addition, demand must be considered as a factor in the safe yield.

Risks in investments A key risk in implementing investment strategies is the financial risk. The size of the

investments, the long term return and the impact of the debt on future generations result in serious financial risks for municipalities, states and the nation. It concerns also industry, which needs to finance innovation, take in charge concessions and commit itself to a given quality of service. Financial engineering is needed to protect investors from the downside, and flexibility is needed to mitigate cost overruns and excessive delays. This situation is encountered in many industrial sectors.

Techniques such as real options have been introduced with success in the car industry, the aeronautics industry and the pharmaceutical industry. It is important to study the feasibility of these techniques for investments in the water resources sector.

Associated risks with financing investment in water can be summed up in: • Sub-sovereign risk: municipalities are not financially strong, • Political risk: politicians oppose tariff increases,

174

• Contractual risk and information asymmetry: Concessions are for 25-30 years, operating environment can change, bidders do not have full information on the underground network, on installations,...etc.

• Resource failure: Climate change, droughts, growing cities in developing countries, falling groundwater levels, loss of revenue,...etc.

3. Desalination Water desalination is the process of extracting freshwater from seawater or brackish

water, by removing salt and other minerals. Most of the technologies used in desalination rely on either membranes (reverse osmosis, electrodialysis, ...) or distillation (multistage flash distillation, multiple effect distillation, ...). However, other approaches exist. For instance, the co-generation process uses excess heat from power production to produce freshwater.

The selection of a desalination process depends on the site specific conditions, the type and the quantity of salt present in water, economic considerations, the quality and the quantity of water needed, and the local engineering experience and skills.

3.1. Economic considerations Desalination becomes much more viable: Today's reverse-osmosis desalination can

produce 27 times the amount of water that could be produced in 1980 for the same cost. The increase of water demand and the decrease of water supply given by other facilities

impose using seawater. Many scenarios are possible. The trivial scenario is that governments or municipalities subsidize the price of desalinated seawater. Governments can also encourage research and progress in desalination technologies. Other possibilities such as raising water rates to pay for desalination can be envisaged. For example, in Adelaide (Australia), a desalination plant will be funded by raising water rates to achieve full cost recovery (People’s Daily Online, 2001).

Desalination implementation costs include construction costs, and operating and maintenance costs (Younos, 2005). For example, an estimation of capital costs for a 25 million gallons per day seawater desalination plant is (TWDB, 2006):

• Desalination plant: US$88,000,000 • Concentrate disposal system: US$31,400,000 • Finished water transmission system: US$12,200,000 • Special studies, engineering, National Environmental Policy Act permitting, and

construction support services: US$18,400,000 • Total project cost: US$150,000,000 As examples, Israel is now desalinating water at a cost of US$0.53 per cubic meter

(Sitbon, 2005). Singapore is desalinating water for US$0.49 per cubic meter (Black&Veatch-Designed, 2006).

Because the seawater supply is practically infinite, desalination is surely the most reliable water supply facility. It is probably the most efficient one to face drought situations. In general, obtaining freshwater by desalinating seawater costs more than using other facilities. However, as we show it below, on an example taken from Cooley et al. (2006) and the Committee on

175

Advancing Desalination Technology (2008), the cost of desalinated water, computed by taking into account its reliability, can be lower than the costs given by other facilities, computed in the same way.

Example The following example is an adaptation to desalination of an example taken from Cooley

et al. (2006) and the Committee on Advancing Desalination Technology (2008). The purpose is to show how to take into account the reliability of supply water options in the evaluation of their costs.

Let us suppose that we have to respond to a drought situation, by creating a new water supply facility to satisfy a growing demand. We have to select one facility among two facilities which are: 1- desalinated seawater supply, and 2- surface water supply.

We denote by tX the water supply per year t from a local river, and we suppose that

),( 2σXNX t : where 410=X acre-feet, and 310=σ . That is, the average of the water supply is 410 acre-feet per year with a standard deviation of 10% . We assume that the quantity X

satisfies the actual demand denoted by Y , ie XY = . The objective is to ensure the availability of water in 39 out of 40 years, on average. This corresponds to a threshold of σ2= −− YX acre-feet of water to satisfy. Indeed:

1/40.=2.5%=8,000)<(=)2<( tt XPYXP σ−

The availability of only −X acre-feet per year is considered as a drought situation. It is avoided in 97.5% of the cases.

Suppose that the demand for the next decade is expected to increase with 2000 acre-feet per year to reach the quantity of 12,000=Z acre-feet per year (which can be due to a growth in population for example). Thus, if nothing is done, the reliability of the water supply system will decrease to 50% . Indeed:

1/2.=50%=10,000)<(=)2<( tt XPZXP σ−

Water managers have to maintain the reliability level at 97.5%. To satisfy this constraint, an additional quantity of water equal to 2000 acre-feet per year with a confidence of σ2 is required. Two choices are given:

• Desalination: assumed with a standard deviation equal to 5% and a cost of US$ 0.64 per cubic meter (Committee on Advancing Desalination Technology, 2008), which is approximately US$ 800 per acre-foot. If we denote by W the average of the required quantity, then W satisfies: 2000=)0.05(2 WW ××− , which gives 2,222=W acre-feet per year. So the real cost of water per acre-foot is US$ 2,222/2000800× =US$ 889.

• Surface water supply: with a standard deviation of equal to 20%, and a cost of US$ 600 per acre-feet. Similarly, we obtain 3,333=W acre-feet per year, and the real cost of water per acre-feet is US$ 1000.

This example shows that although the gross cost of desalinated seawater (US$800 per acre-foot) is higher than the gross cost of surface water (US$600 per acre-foot), this ratio is reversed by the fact that the standard deviation of the desalinated seawater (5%) is smaller than the surface water one (20%).

3.2. Environmental considerations

176

Two major environmental impacts of desalination are the impact of the concentrate

(brine), and the impact of the open ocean water intakes. Solutions reducing these impacts begin to be developed.

For the desalination plant and ocean outlet structures to be built in Sydney from late 2007, the water authority states that the ocean outlets will be placed in locations at the seabed that will maximize the dispersal of the concentrated seawater, such that it will be indistinguishable from normal seawater between 50 meters and 75 meters from the outlet points (Wikipedia, 2007).

In Perth, Australia, in 2007, the Kwinana Desalination Plant was opened. The water is sucked in from the ocean at only 0.1 meter per second, which is slow enough to let fish escape. The plant provides nearly 140,000 cubic meter of clean water per day (The Tampa Tribune, 2007).

3.3. Uncertainties and risks related to desalination Uncertainties and associated risks in desalination are numerous. They can come from the

demand and the supply in water, the price of energy, the political decisions, the progress in technology, ...etc.

Uncertainties in demand of water depend on the population growth, the industry and its use of water, irrigation,... etc. The demand in water is generally increasing as consequence of the demographic growth, which leads to increases of water demand in industry and agriculture. However, slowdowns or accelerations of the increase are uncertain, and this fact involves risks in investment in desalination projects.

The main sources of water such as water surface and groundwater are often affected by severe droughts, climate change, pollution,... etc. All of these situations being non deterministic, uncertainties are important in water supply. Any investment project in desalination must take into account these uncertainties, since the main objective of the desalination is to increase the supply in water.

Actually desalinated water costs more than freshwater given by other facilities. The cost of desalination water depends highly on the price of energy, which depends on the demand and the supply of energy, which are complex and non deterministic processes.

Progress in technology represents a risk on investment in desalination projects. Indeed, make a choice of one desalination technology imply a risk that the technology chosen will be overtaken by other technologies. According to a June 5, 2008 article in Globe and Mail, a Jordanian born chemical engineering Ph.D. student at the University of Ottawa named Mohammed Rasool Qtaisha has invented a new desalination technology that is alleged to be between 600% and 700% more efficient than current technology (Globe and Mail, 2008). On June 23, 2008 it was reported that Siemens Water Technologies had developed a new technology that desalinates one cubic meter of water while using only 1.5 kwh of energy, which, according to the report, is one half the energy that other processes use (The Straits Times, 2008).

4. Investment in Desalination Real options theory (Dixit and Pindyck, 1994), (Copeland and Antikarov, 2001), (Smit

and Trigeorgis, 2004) tends to answer to many questions in investments under uncertainty, such as: what is the optimal time to invest, what size should the project be initially, should the project

177

be expandable, when expanding, how much, ...etc. Options on projects are also evaluated using mathematical finance tools. For example, evaluate the option to defer, to expand, to default, to contract or to abandon an investment,...etc.

Important questions in investing in desalination projects are the initial size of the plant, the option of expanding, the optimal time to invest and eventually to expand,...etc. Indeed, due to the economy of scale, the unit production cost for large capacity plants can be lower, when the costs associated with water intake, pretreatment, and concentrate disposal can be substantially reduced if the plant is an expansion of an existing water treatment plant, compared to constructing a new plant.

Specific questions can also be asked, such as the option to invest in energy in parallel with investing in desalination, the option to switch between selling energy and using it to desalinate water, the option that governments invest in energy in regions concerned by desalination, in order to attract investors in desalination projects,...etc.

In the following, we propose a model of investing in a desalination project which introduces some of the issues cited above. The model allows to give answers to two main questions: What are the production and the stock capacities of the plant ? (or how big is the investment ?) and what is the optimal time to invest ?

4.1. Modeling of Investment in Desalination We suppose that a desalination plant project is to be achieved to cover the needs in water

on a certain region. The purpose in this section is to propose a real options approach which allows to obtain a decision rule on the optimal time to invest in the project.

The region to be covered by the plant can be delimited, for example, by comparing the cost of delivering desalinated water with the cost of delivering non desalinated water given by existing water supply facilities. The comparison should take into account the costs of producing, pumping and transporting desalinated and non desalinated water.

How big is the investment ? The desalination plant must have a production capacity and a stock capacity. The

production capacity is proportional to the size of the region of interest, and depends on the process of the demand of the region in water and on the process of the supply of non desalinated water given by other facilities. The stock capacity should guarantee that the remainder (non consumed) water can be stocked.

The users who will be interested by the desalination project are those who may replace some of their non desalinated water supplies by desalinated water supply. We suppose that the user takes into account, in this replacement, the price of water, which is given by (or is proportional to) the costs of producing, pumping, and transporting water to the user.

Delimiting the region concerned by the investment helps to estimate its size and the corresponding cost. Let us denote the demand of the delimited region in water at time t by )(tM . The water supply, given by other facilities rather than desalination, at time t , is denoted by )(tS .

Being affected by uncertainties, these quantities are represented by stochastic processes, which can be estimated using statistical methods applied on the demand and the supply in water in the considered region. The quantity of water concerned by the investment at time t , denoted by )(tN , is )()(=)( tStMtN − .

178

Knowing the distributions of the processes )(tM and )(tS , we first determine the capacity of the plant in producing water, which is also related to the investment cost. We suppose that the plant has a capacity K which is fixed (independent with time), and which represents the quantity of water produced during a unit of time (in a year). It can be estimated, for example, by:

0

1= ( ) ,

T

K N t dtTE ∫

where T denotes the life of the plant, which can be taken equal to ∞+ . We suppose also that when the produced quantity of water is not completely consumed, the remainder can be stocked. The capacity of the stock is fixed and is denoted by G . The stock at time t denoted by )(tg satisfies:

=0 =0

( ) = [ ( )] = ( ) .t t

s sg t K N s ds Kt N s ds− −∫ ∫

The stock capacity must be determined in such a way that we minimize the risk that the stock g overtakes the stock capacity G . To guarantee this, the stock process must satisfy:

,0,<>)( TtrGtgP ≤≤∀ (48) where r is the risk associated. Then the minimal stock capacity G corresponding to a given risk r is :

),(1max= 1

0rFG t

Tt−−

≤≤

where )(⋅tF is the cumulative distribution function of )(tg .

The costs of the project The fixed cost of investment, denoted by I , can be evaluated, depending on the capacity

K of the plant and on its stock capacity G . Denoting by ϖ the fixed cost per unit of volume to produce desalinated water, and by ς the cost per unit of volume to stock desalinated water, the cost of investment I can be taken:

.= GKI ςϖ + The variable costs by unit of volume, of desalinating water at time t , are summarized in one stochastic process denoted by )(tC , which depends mainly on the price of energy at time t .

Assumptions on the distributions We have three processes: the demand in water )(tM , the supply of water assured by other

facilities rather than desalination, )(tS , and the variable costs )(tC per unit of volume, of desalinating water. In addition, we assume that the process of water supply )(tS is given by two other processes: the seasonal weather and precipitations changing, denoted by )(tZ , and the rarely occasional weather changing such as droughts, denoted by D(t). We make the following assumptions on these processes:

• The total demand in water )(tM at time t depends mainly on the demographic growth which causes also growth in industrial and agricultural activities. It is supposed to be given by a geometric Brownian motion:

)),()((=)( tdWdttMtdM MM ⋅+σα

179

where Mα is a constant, Mσ is a constant vector and )(tW is a 2-dimensional Wiener process

modeling the uncertainty, whose components 1W and 2W are independent. • The supply of water given by other facilities rather than desalination is supposed to be

given by: ).()(=)( tDtZtS −

• The process )(tZ is the water supply given by other facilities, rather than

desalination, assured in regular no-droughts periods, which models the seasonal changing of weather. It is supposed to be deterministic depending on time:

,2cos=)( zthtZ +π where (0)= Zz is a constant giving the average value of )(tZ over time, and h is a constant satisfying 0],,[)( ≥∀+−∈ thzhztZ . This is a simple deterministic modeling which supposes that the precipitation process is periodic, of period 1 (when the unit of time is one year), and its variations over time are around the average value z .

• The process )(tD is the shortage of water caused by rarely drought situations. It is modeled by a Poisson process:

( , )

1 with probability ( ) =

0 with probability 1 ,

dtdD t

dtλ φ

λ

λ

−

where λ gives the mean arrival rate of a drought event during an interval dt , and φ gives the size of the shortage water causes by a drought occurrence.

• The process )(tC modeling the variable cost per unit of volume, of desalinating water at time t , depending on the price of energy, is also supposed to be given by a geometric Brownian motion:

)),()((=)( tdWdttCtdC CC ⋅+σα

where Cα is a constant, Cσ is a constant vector.

In the following we summarize the information given by the processes )(tM and )(tD in only one process of demand, denoted by )(tP , by assuming that demand in water )(tM increases during the drought event occurrences. Since )(tM is given by a geometric Brownian motion, and

)(tD is given by a Poisson process, we assume that the sum )()( tDtM + can be summarized in a mixed Brownian motion and Poisson process )(tP :

)].()()[(=)( tdDtdWdttPtdP PP +⋅+σα (49) We suppose also that dW and dD are independent, that is 1( ) = 0dW dDE and

2( ) = 0dW dDE . The dynamics (46) says that the demand in water )(tP fluctuates as a

geometric Brownian motion, but during each interval of time dt , a drought situation can occur with the probability dtλ , and in the case, the demand in water increases with a proportion equal to φ , to reach )(1 φ+ times its original value.

The present value of the project The profit flow (or revenue stream) of the desalination project at time t , denoted by )(tR

is given by:

180

),()()(=)()]()([=)( tZtKCtPtKCtStMtR Γ−−Γ−−Γ where Γ denotes the price of unit of volume of desalinated water, and K is, as denoted above, the production capacity of the desalination plant.

The value of a project, is the expected present value of its discounted profit. Thus, the value V of the desalination project is given by:

0 0 0

= ( ) ( ) ( ) .t t tV P t e dt K C t e dt Z t e dtE Eρ ρ ρ∞ ∞ ∞

− − −Γ − − Γ∫ ∫ ∫

where ρ is a discount rate. Since ( )( ) = (0)

tPP t P eEα λφ+ and ( ) = (0)

tCC t C eEα , we obtain the

value V depending on P and C :

,4

=),(22 ρπρ

ρ

δδ

zhC

KPCPV

CP

Γ−

+

Γ−−

Γ (50)

where )(= λφαρδ +− PP and CC αρδ −= .

4. 2. The optimal time of investing and the value of the option to invest The payoff from investing at time t is ItCtPV −))(),(( , where ))(),(( tCtPV is the value

of the project at time t , given in (47); and I is the total fixed cost of investment. In the following we introduce the process )(tQ which gives an information about the total costs of investment in the desalination project, including the variable costs given by )(tC , and the fixed cost I . The process )(tQ is defined by:

.4

)(=)(22

+

Γ+

+

Γ+ I

zh

KtCtQ C

ρπρ

ρδ

For convenience in notation, we denote also by QQ σα , and Qδ the quantities CC σα , and Cδ

respectively. Thus we write: )].()[(=)( tdWdttQtdQ QQ ⋅+σα

With these notations, we have:

.=),(=),( QK

PICPVQPUQP

def

δδ−

Γ−

Maximizing the expected present value of the payoff from investing gives the value of the investment opportunity, denoted by ),( QPF , which is the value of the option to invest:

max( , ) = ( ( ), ( )) ,F P Q U P Q e Eρτ

ττ τ − (51)

The solution of the optimal stopping time problem (51) is a feedback giving the decision to invest or not, depending on the values of the processes P and Q at a given time. The Hamilton-Jaccobi-Bellman equation is:

( , ) = max ( , ) , ( , ) .F P Q dt U P Q dt dF P QE ρ (52)

In the continuation region, corresponding to values of P and Q for which it is not optimal to invest, equation (52) is written as follows:

= .F dt dFEρ (53)

Using Ito's lemma, dF is given by:

181

.)2(2

1= 22 dQFdPdQFdPFdQFdPFdF QQPQPPQP ++++

We replace dP and dQ by their expressions, taking into account that 1 2= = 0dW dWE E , we

obtain: = ( , ) ([1 ] , )P P Q QdF PF dt F P Q F P Q dt QF dtE α λ φ α+ − + +

dtFQPQFFP QQQPQQPPPP ||||2||||2

1 2222 σσσσ +⋅++

Hence

)||||2||(||2

1 2222QQQPQQPPPP FQPQFFP σσσσ +⋅+

),()()()( QPFQFPF QQPP λρδρδρ +−−+−+

.0=),]([1 QPF φλ ++ (54) Equation (54) is a partial differential equation, for which three boundary conditions are associated. If we denote by ),( ** QP the points of 2R from which it becomes worth to invest,

then the boundary conditions consist of a value-matching condition: ),(=),( **** QPUQPF , and

two smooth-pasting conditions: ),(=),( **** QPUQPF PP , and ),(=),( **** QPUQPF QQ .

Because of the nature of the dependence of the value of the project on the processes of demand in water P and of the total cost of desalinating water Q , the optimal decision to invest should depend on these processes only through their ration QPx /= . So the value of the option to invest on the desalination project should be homogeneous of degree 1 in ),( QP . We can check that the partial differential equation (54) is reduced to a simple differential equation by putting

QPx /= and )(=)/(=),( xQfQPQfQPF . We obtain the following equation:

)()()()||||2||(||2

1 222 xfxxfx PQQQPP ′−+′′+⋅+ δδσσσσ

.0=)]([1)()( xfxfQ φλλδ +++− (55)

Denoting by *x the critical ratio **/QP , from which we stop waiting and start investing, the boundary conditions are:

,/)/(=)( **QP Kxxf δδ −Γ (56)

,/=)( *Pxf δΓ′ (57)

./=)()( ***QKxfxxf δ−′− (58)

We notice that these conditions are redundant but compatible. The solution of equation (55) is of the form βAxxf =)( , where β is a solution of the equation:

0,=)(1)()(1)(2

1 2 βφλλρβδδββ +++−−+−Σ PQ (59)

and where we denoted by 2Σ the quantity 22 ||||2|||| QQPP σσσσ +⋅+ . The equation (59) is solved

numerically to determine a solution β which is compatible with the boundary conditions. Once

β is known, the constant A and the threshold *x are determined by solving the boundary conditions ((56)–(58)). They are given by:

182

,1

/==/ ***

−

Γ

β

β

δδ PQ

KxQP (60)

.)/(= ** β

δδx

KxA

QP

−

Γ

Practically, the decision is taken as follows: as if the quantity QP/ is less than the threshold *x ,

the optimal decision is to wait; and we decide to invest only once QP/ reaches the value *x . Interpretation First, let us notice that if 1>β then the critical value to invest ),( ** QPV exceeds the

fixed cost of investment I . Indeed:

.>),(0>1> **** IQPVQK

PQP

⇒−Γ

⇒δδ

β

This means that, when the value of the project is less than ),( ** QPV , the optimal decision is to wait, even if this value exceeds the fixed cost of investment I . This is not the case of the net present value (NPV) method, when the threshold is simply given by I .

On a 2-dimensional space (plan), if we make the values of the total costs to invest Q on the x-axis, and those of the demand in water, which takes into account drought situations, and which are transformed to gains (when it is assured by the desalination plant), on the y-axis, then the boundary between waiting and investing is given by the lane QxP *= . This lane divides the plan of points ),( QP into two parts, a down-side part representing the region where the optimal decision is to wait, and an up-side one where the optimal decision is to invest. In practice, at a given time t , the values of )(tP and )(tQ are observed and the point ))(),(( tQtP is located in the plan. The decision to invest or not is then taken depending on to which region the point belongs.

Now, let us analyze the value *x . The parameters PQK δδ ,,,Γ appear explicitly on *x ,

and by recalling that )(= λφαρδ +− PP , QQ αρδ −= , and that β is given by (59), the threshold *x depends also on the parameters λαα ,, QP and φ . A main dependence of the threshold

*** /= QPx is on the ration Γ/K . Indeed, comparing QP/ with Γ/K is also comparing PΓ to KQ , what means that the decision to invest in the desalination project depends mainly, up to the other parameters, on the comparison of the total revenue of the plant PΓ with the total costs of desalinating water KQ .

Conclusion Water issues weigh increasingly on government budgets, on the world economy

(including agriculture, industry, tourism, ...), on the environment, and also on the direction of research in laboratories.

183

Because water is practically related to all areas, the uncertainties are numerous and often complicated to isolate. Nevertheless, in order to be able to face all eventualities with sufficient decision margins, we must identify these uncertainties and assess the associated risks. This approach should be a principle to hold during all the phases of management of a water system, starting from the design phase.

To be efficient in dealing with the hardest water issues such as droughts, long-term decisions have to be planned, and big investments in water project have to be done. The reliability of the desalinated seawater supply facility puts desalination in a strong position to meet the toughest water shortages. Despite recent progresses in desalination technologies, the costs of desalinated seawater remain relatively high.

A good approach to analyze investments under uncertainties is the approach of real options theory, which has been introduced with success in the car industry, the aeronautics industry and the pharmaceutical industry. This approach tends to answer to questions such as the best size of an ivestment, the optimal time to invest, the evaluation of the options on investments (defer, default, expand,...etc). We presented in this paper a simple model of investment in desalination projects, which introduces the application of real options theory. Our future interest is to study in more details the feasibility of applying this approach in water resources management.

References

[1] Black & Veatch-Designed Desalination Plant Wins Global Water Distinction, Press release. via edie.net, 2006-05-04. Retrieved on 2007-08-20. [2] T. E. Copeland & V. Antikarov, Real Options: A Practitioner's Guide, New York, 2001. [3] Heather Cooley, Peter H. Gleick and Gary Wolff, Desalination with a grain of salt, A California Perspective, Pacific Institue, 2006. [4] Committee on Advancing Desalination Technology, National Research Council, Desalination: A National Perspective, 2008, [5] http://nap.edu/catalog/12184.html. [6] A. K. Dixit & R. S. Pindyck, Investment Under Uncertainty, Princeton University Press, 1994. [7] Globe and Mail Ottawa student may hold secret to Water For All, June 5, 2008. [8] [41] Lawrence Livermore National Laboratory, Nanotube membranes offer possibility of cheaper desalination, Public Affairs (2006-05-18). [9] [42] R. McDonald & D. Siegel, The Value of Waiting to Invest, Quartery Journal of Economics, vol. 101, pp. 707--728, 1986.

184

[10] People's Daily Online, Australia Aids China In Water Management Project, 2001-08-03, via english.people.com.cn, Retrieved on 2007-08-19. [11] Schleher D. Generalized Gram-Charlier Series with application to the sum of lognormal variates, IEEE Trans. Inform. Theory, pp. 275--280, 1977. [12] Schwartz SC, Yeh YS, On the distribution function and moments of power sums with lognormal interferers, Bell Systems Technical Journal, 61(7):1441--1462, 1982. [13] Sitbon, Shirli, French-run water plant launched in Israel, European Jewish Press, via ejpress.org, 2005-12-28. Retrieved on 2007-08-20. [14] T. J. Smit & L. Trigeorgis, Strategic Investment: Real Options and Games, Princeton University Press, 2004. [15] The Straits Times, Team wins $4m grant for breakthrough technology in seawater desalination, June 23, 2008. [16] Applause, At Last, For Desalination Plant, The Tampa Tribune, December 22, 2007. [17] Texas Water Development Board (TWDB), Water for Texas 2007, november 2006. [18] United Nations Environment Program (UNEP), http://www.unep.org, [19] Wikipedia Encyclopedia, Desalination, September 2007, http://en.wikipedia.org/wiki/Desalination. [20] Tamim Younos, The economics of desalination, Journal of Contemporary Water Research & Education, 132, pp:39-45, 2005.

185

Optimal energy investment and R&D strategies to stabilize atmospheric greenhouse gas

concentrations1,2

Valentina Bosetti, Carlo Carraro, Emanuele Massetti, Alessandra Sgobbi, Massimo Tavoni

Abstract Stabilizing the atmospheric concentrations of greenhouse gases (GHGs) at levels

expected to prevent dangerous climate changes has become an important, long-term global objective. It is therefore crucial to identify a cost-effective way to achieve this objective. In this paper, we use WITCH, a hybrid climate–energy–economy model, to obtain a quantitative assessment of equilibrium strategies that stabilize CO 2 concentrations at 550 or 450 ppm. Since technological change is endogenous and multifaceted in WITCH, and the energy sector is modeled in detail, we can provide a description of the ideal combination of technical progress and alternative energy investment paths in achieving the sought stabilization targets. Given that the model accounts for interdependencies and spillovers across 12 regions of the world, equilibrium strategies are the outcome of a dynamic game through which inefficiency costs induced by global strategic interactions can be assessed. Our results emphasize the drastic change in the energy mix that will be necessary to control climate change, the huge investments in existing and new technologies implied, and the crucial role of breakthrough technological innovation.

Keywords: Climate policy,Energy R&D, Investments, Stabilization costs

1 @ 2009 Elsevier B.V. This paper was first published in Resource and Energy Economics. We are very grateful to

the authors and to the publisher for their permission to reprint this paper.

2 This paper is part of the research work being carried out by the Climate Change Modeling and Policy Research

Program of the Fondazione Eni Enrico Mattei and by the Climate Impacts and Policy Division of the EuroMediterranean Center on Climate Change. Financial support from the Italian Ministry of the Environment and from the Transust.Scan project is gratefully acknowledged. Comments from participants at seminars in Brussels, Louvain-la-Neuve, Mannheim, Rome and Copenhagen helped to improve the paper. The usual disclaimer applies.

186

Climate change may dramatically damage future generations. According to the latest

International Panel on Climate Change report (IPCC, 2007), anthropogenic emissions of greenhouse gases (GHGs) are among the main causes of climate change, even though uncertainty remains as to their exact relevance in the overall climatic process: thus it is necessary to identify when, where and how these emissions ought to be controlled in order to avoid dangerous climate changes.

The many uncertainties that still permeate the debate about the relationship between GHG concentrations and temperature change or the existence of temperature thresholds beyond which irreversible changes could occur, make it difficult to use the standard cost-benefit framework for jointly identifying the optimal stabilization target and related investment mix. Scientific uncertainties aside, the long-term stabilization target is clearly a political decision, and policymakers worldwide are indeed discussing how to tackle the climate change problem. At the 2008 G8 Summit in Japan, the leading industrialized nations agreed on the objective of at least halving global CO 2 emissions by 2050. Such an agreement follows earlier resolutions of other countries, such as the European Union (EU), Canada and Japan3. There is therefore increasing interest in, and a need for, research efforts providing information on the best strategy that different regions of the world should adopt in order to minimize the cost of achieving their own emission reduction target. In particular, it is crucial to identify the longterm investment mix in the energy sector in different world regions, taking into account the role of investments in energy R&D and the future evolution of different technologies.

For analytical purposes, this paper considers two long-term stabilization targets, both expressed in terms of atmospheric carbon concentrations. The first target is a 550 ppm (CO 2

only) concentration target. The second one stabilizes emissions at 450 ppm (CO 2 only). These two reference targets roughly coincide with IPCC Post-Third Assessment Report (TAR) stabilization scenarios C and B respectively. Although the IPCC considers even more stringent emissions pathways, our current analysis focuses on the two that we consider more politically realistic. The first target is often advocated for in the United States (see e.g. Newell and Hall, 2007), whereas the second one is close to the EU objective of keeping future temperature changes within 2 C. We then compute the welfare maximizing path of energy R&D expenditures, investments in energy technologies and direct consumption of fossil fuels that is consistent with the proposed stabilization targets.

The equilibrium R&D and investment strategies in a given region of the world depend upon many factors, such as the discount rate; the investment decisions taken in other regions or countries; and the effectiveness of R&D in increasing energy efficiency, or in providing new, low carbon, energy technologies. Equilibrium R&D and investment strategies also depend on the expected climate damages, on the pattern of economic growth in various regions of the world, and on other economic and demographic variables. In this paper, all these interdependent factors are taken into account.

To this purpose, we use WITCH (World Induced Technical Change Hybrid; see Bosetti et al., 2006a, 2007a), a climate-energy-economy model in which a representation of the energy

3The European Union, for example, has identified both its long-term target (to keep the increase of global

atmospheric temperature below 2 8C with respect to the pre-industrial level) and a short-term target consistent with the former (i.e. a reduction of 2020 emissions by 20% with respect to 1990, which may become a 30% reduction if a global agreement on climate change is achieved).

187

sector is fully integrated into a top-down optimization model of the world economy. Thus, the model yields the equilibrium intertemporal allocation of investments in energy technologies and R&D that belong to the best economic and technological responses to different policy measures. The game theory set-up accounts for interdependencies and spillovers across 12 regions of the world. Therefore, equilibrium strategies are the outcome of a dynamic game through which inefficiencies induced by global strategic interactions can be assessed. In WITCH, technological progress in the energy sector is endogenous, thus enabling us to account for the effects of different stabilization scenarios on induced technical change, via both innovation and diffusion processes. Feedback from economic variables to climatic ones, and vice versa, is also accounted for in the dynamic system.

These features enable WITCH to address many questions that naturally arise when analyzing carbon mitigation policies. Among those that this paper aims to answer are the following: what are the implications of the proposed stabilization targets for investment strategies and consumption of traditional energy sources vis-a-vis low carbon options? What is the role of public energy R&D expenditures for generating improvements in both energy efficiency and carbon intensity? And how sensitive are the economic costs of climate policies to different technological scenarios, and in particular, to hypotheses on major technological breakthroughs?

The structure of the paper is as follows. Section 2 describes the framework of our analysis and explores the implications of stabilization targets for the energy sector. Section 3 informs readers about investment needs for known technologies, while Section 4 focuses on innovation strategies. Section 5 provides estimates of the economic costs of climate policy with a focus on technological choices, and Section 6 concludes the paper. Appendix A provides background information on the WITCH model.

1. The challenge of stabilizing atmospheric GHG concentrations As previously indicated, we investigate best response strategies, particularly in the energy

sector, to achieve two stabilization targets. According to the first one, atmospheric concentrations must be stabilized at 550 ppm (CO 2 only) by the end of the century. This is roughly equivalent to a 650 ppm target if all GHG are included. The second target is more stringent and requires that CO 2 concentrations be stabilized at 450 ppm (550 ppm all gases included) at the end of the century. Fig. 1 shows Business as Usual (BaU) emissions together with emission time profiles for the two stabilization targets. These are optimal time profiles because they were obtained by computing the fully cooperative equilibrium of the game given the GHG concentration constraints, i.e. by solving a global joint welfare maximization problem where all externalities are internalized. Note that feedbacks from climate damage to the production of economic goods are taken into account when computing the optimal emission profiles4.

4We adopt the same damage function as in Nordhaus and Boyer (2000). Future damages are discounted at a

declining discount rate (starting from 3% and declining to 2%).

188

Fig. 1. World fossil fuel emissions in the three scenarios (2002–2102). Current annual fossil fuel CO 2 emissions are roughly 7 Giga Tonnes of Carbon per year

(GtC/yr). According to the model projections, without any stabilization policy (the BaU or ‘‘baseline’’ scenario), CO 2 emissions are expected to reach about 21 GtC by the end of the century, a value in line with the IPCC B2 scenario of the Special Report on Emissions Scenarios (SRES). In the case of the 550 ppm stabilization target, annual emissions slowly increase until 2060 (when they reach 10 GtC/yr) and then decrease to 8 GtC by the end of the century. If the target is 450 ppm, CO 2 emissions start decreasing immediately and reach 3 GtC by the end of the century. That is, the optimal emission profile does not allow for overshooting emissions which would trade off current and future abatement. The emission reductions required to meet the more stringent stabilization target are particularly challenging, given the expected growth rate of world population and Gross Domestic Product (GDP): per capita emissions in the second part of this century would have to decline from about 2 to 0.3 tC/cap per year5.

To achieve the two stabilization targets and the related optimal emission profile, it is assumed that all regions of the world agree on implementing a cap and trade policy. This is an obvious simplification which is useful in this paper to focus on differences in the technological make-up of the economy under the two stabilization scenarios, and on the difference in R&D portfolios. In two companion papers (Bosetti et al., 2008a,b), we analyze the implications of partial agreements, delayed action in developing countries, and uncertain stabilization targets. In this paper, the global cap and trade policy is implemented by assuming an equal per capita allocation of initial allowances.

Given the adopted climate policy, countries use the permit market to trade emissions (banking is also allowed) and determine their investments and R&D strategies, as well as their demand for permits, by maximizing their own welfare function (see Appendix A) given the

5Note that 0.3 tCyr 1− cap 1− is the amount of carbon emitted on a one way flight from the EU to the US East Coast.

189

strategy adopted in the other regions of the world. The intertemporal Nash equilibrium of the dynamic game defines the equilibrium investment strategies in each world region.

To assess the implications of the equilibrium of the game under the two concentration constraints, let us compare the impact of imposing the two stabilization targets on the dynamics of the main economic variables. Table 1 shows the changes in the variables belonging to the well-known Kaya’s identity (emissions, per capita GDP, energy intensity, carbon intensity of energy and population) for two periods: 1972–2002 (historical values) and 2002–2032 (WITCH scenarios).

Table 1. Ratio of future over past values of Kaya’ s variables in the three scenarios (BAU,

450 and 550 ppm). In the BaU, future changes of all economic variables, one of the main outputs of the

model, are consistent with historical values observed in the past 30 years6. Baseline emissions, which are also an output of the model, almost double in 30 years time, due to the exogenous population growth and to the endogenous growth in income per capita. From basic assumptions on energy technologies and initial investment costs, we derive an endogenous path of energy use in which a looser economy–energy interdependence emerges, but not an energy–carbon decoupling. The endogenous dynamics of energy and carbon efficiency are comfortingly similar to observed trends over the past three decades. Given that the characteristics of the baseline have important implications in terms of efforts required to stabilize the climate (and therefore in terms of stabilization costs), the ability to reproduce history – at least over short time horizons – is an important feature of the WITCH model.

In the 550 ppm scenario, lesser growth in emissions stems mainly from energy efficiency improvements as testified by the decrease of energy intensity (DEN/GDP column), although some decarbonization of energy is also needed. A more fundamental change is required in the 450 ppm scenario. Keeping carbon concentrations below this target can be achieved only if both energy intensity and carbon content of energy are significantly decreased.

6Although for carbon capture and sequestration only pilot projects are in place at the present moment, the technology

has been operating on a smaller scale for enhanced oil recovery for a long time now.

190

Fig. 2. Reductions of energy and carbon intensity in the next 30, 50 and 100 years, and

over the past 30 years (changes w.r.t. 2002). Fig. 2 provides some additional interesting information on the modifications required in

the energy sector, as it plots the evolution of energy intensity and carbon intensity of energy in 2030, 2050 and 2100. The BaU scenario is characterized by an improvement of energy intensity, even though slightly less pronounced than the historical one. It also shows a slight carbonization of energy over the century: although small, this effect reflects the increasing share of coal in the energy mix in the absence of climate policy (this is also consistent with the Energy Information Agency’s medium term projections; see EIA, 2007). This increase is mostly driven by the growing energy consumption of developing countries. Coming to the stabilization scenarios, they both show energy efficiency measures to be the most relevant in the short-term, but both call for the development of low carbon options in the long-term, especially for the more stringent 450 stabilization target.

The dynamic paths of energy intensity and carbon intensity of energy implied by the two stabilization scenarios require drastic changes in the energy sector. The next section will analyze the equilibrium investment paths in different energy technologies over the next century. This will allow us to identify the welfare maximizing investment strategies that different regions of the world ought to implement to achieve the two stabilization targets.

2. Equilibrium mitigation strategies with known energy technologies The energy sector is characterized by long-lived capital. Therefore, the investment

strategies pursued in the next two/three decades will be crucial in determining the emissions pathways that will eventually emerge in the second half of the century. The previous section highlighted the urgent need for a new strategy in the energy sector, targeted to de-carbonize energy production. This can be done through the extensive deployment of currently known abatement technologies (Pacala and Socolow, 2004) and/or through the development of new energy technologies. Let us analyze the equilibrium investment mix and the related shares of existing and innovative technologies in the stabilization investment portfolio.

Emission reductions can be achieved by increasing energy efficiency and by reducing carbon intensity. As shown in Fig. 2, energy efficiency improvements beyond the baseline scenario are an important component of a GHG control strategy. Many economic sectors are

191

indeed characterized by the potential for large savings at relatively low costs. Yet, especially for ambitious emission reductions, energy efficiency improvements are not enough and energy de-carbonization is essential. Supply cost curves of abatement vary widely across sectors; for example they are believed to be especially steep in the transport sector. Power generation is comparatively more promising: it is a heavy weight sector in terms of emissions and one of the few for which alternative production technologies are available.

Not surprisingly, our scenarios show a significant contribution of electricity in mitigation, as illustrated in Fig. 3. To optimally achieve a 450 ppm concentration target, almost all electricity (around 90%) will have to be generated at low, almost zero, carbon rates by 2050 (left panel). The milder 550 target allows a more gradual transition away from fossil fuel based electricity, but nonetheless shows a noticeable departure from the no climate policy BaU scenario. The role of electricity is strengthened by its growing share with respect to primary energy supply. The substitution towards electricity is especially important for the more stringent 450 scenario (Fig. 3, right panel), since it makes it possible to meet the strong emissions cuts needed in the traditional non-electric sector. Such a radical change is achieved through three already operational technologies7: nuclear energy, renewable sources (wind and solar) and carbon capture and sequestration (CCS) (see Fig. 4 that shows the power generation shares for the 550 (left) and 450 (right) scenarios).

Fig. 3. The role of electricity in mitigation.

7Although for carbon capture and sequestration only pilot projects are in place at the present moment, the technology

has been operating on a smaller scale for enhanced oil recovery for a long time now.

192

Fig. 4. Power generation shares for the 550 (left) and 450 (right) scenarios. Nuclear power becomes extremely competitive given the range of carbon prices implicit

in the adoption of climate policy, especially for the 450 case, where it eventually guarantees about 50total electricity generation. This remarkable expansion requires a 10-fold increase in present generation capacity. Twenty or more 1 GigaWatt (GW) nuclear plants would need to be built each year in the next half-century, bringing the nuclear industry back to the construction rates of the 1980s. Clearly, this gigantic capacity deployment for such a contentious technology would raise significant social and environmental concerns, to the point that the feasibility of a nuclear-based scenario would ultimately rest on the capacity to radically innovate the technology itself, as well as on the institutions controlling its global use. Alternatively, a scenario where nuclear power is constrained by political and environmental concerns would imply an enhanced deployment of wind and solar power plants and a minor increase in costs.

Renewable energies, especially wind power, have developed at an impressive rate in recent years (up to 10 GW per year), but the limited annual operating hours and costs bind their potential electricity contribution, at least in the short run. Only later in time would capacity additions reach 30 GW per year – especially via solar power – and be able to significantly contribute to the decarbonization of the power sector.

Carbon capture and sequestration makes it possible to burn coal in power plants while massively reducing carbon emissions. The decoupling of coal use and carbon emissions is particularly important for regions with a large endowment of coal reserves and because coal-fired power plants are very attractive for energy security reasons. However, the necessary investments are very large. To achieve the 550 ppm target, between 30 and 40 1 GWcoal-with-CCS power plants would need to be built each year from 2015 onwards, a value in line with the historical capacity building of traditional coal plants (roughly 50% of electricity generated in the world). A number of large-scale pilot plants should thus be put into place in the next 10 years to ensure the feasibility of such a massive deployment.

193

Fig. 5. Carbon capture and sequestration Fig. 5 further elaborates on the role of CCS. The optimal amount of injected carbon is

shown to be significant: about 2 GtC/yr (about 1/4 of today’s emissions) are stored underground by mid-century. Over the whole century, about 150 GtC are injected in underground deposits (a figure in line also with the IPCC Fourth Assessment Report Working Group III). However, in the 450 scenario, the use of this technology decreases after 2050. The reason is that a more stringent target calls for a relatively greater deployment of very low carbon technologies; renewable energies and nuclear power are thus progressively preferred to CCS, because they have lower emission factors8. Advances in the capacity to capture CO 2 at the plant (assumed at 90%) would increase CCS competitiveness; though this could be counterbalanced by potential leakage from reservoirs (our simulations show that leakage rates of 0.5% per year would jeopardize the deployment of this technology).

Summing up, an equilibrium investment strategy in the energy sector that can achieve the two stabilization targets at reasonable economic costs exists (the cost would be about 2.1% of global GDP in the 450 ppm case, using a 5% discount rate, see Section 5). This energy investment strategy is based on the massive deployment of existing technologies (nuclear, solar and coal + CCS). It requires huge investments and urgent decisions. In the next section, we will explore how the potential availability of new energy technologies, developed through adequate R&D expenditures, can modify the investment scenario in the energy sector.

3. Innovation strategies for energy efficiency and technology

breakthroughs

8A coal + CCS power plant emits roughly 1/3 of a natural gas one. Constraining the potential deployment of nuclear

and renewables would offset this effect, since the power sector would have fewer options. A similar effect would result from the deployment of very low carbon options in the non-electric sector, since it would alleviate the mitigation effort required from the power sector, as shown in Section 4.

194

The previous section has outlined the need for a profound transformation of the energy sector, particularly if an ambitious climate target is to be achieved. Massive deployment of technologies that are controversial, such as nuclear power, or whose reliability and affordability is still to be proved, such as CCS, indicate that currently known technologies alone might not suffice, especially in the midto long-term, and that the simultaneous achievement of global economic and environmental wellbeing is likely to ultimately rest on our ability to produce innovation. This is especially important for sectors that, at present, have a restricted portfolio of abatement options, such as transport. It is also important in case some of the mitigation alternatives described in the previous section do not deliver their expected abatement potential. In order to address these issues, in this section we use a richer model specification in which it is possible to invest in R&D to develop technologies that still do not exist. In WITCH, R&D investments in these breakthrough technologies, both in the electric and the non-electric sector, are primarily meant to decrease the carbon intensity of energy by providing new sources with energy at zero or low carbon emissions. We refer to these technologies as ‘‘backstops’’. They substitute nuclear power for power generation and oil in the non-electric sector. For a complete description, see Appendix A. Technology advancement needed for achieving higher energy efficiency are still possible, but not the unique choice when it comes to invest in technological development, as it was in the scenario evaluated in the previous section. We can therefore compute the equilibrium R&D investments that countries need to implement to achieve the required improvements in energy efficiency and timely market penetration for new carbon free energy technologies.

Fig. 6. Public energy R&D investments across scenarios to 2050.

195

Fig. 6 shows global public energy R&D expenditures. In the left-hand panel, we plot historical investment in R&D as share of Gross World Product (GWP); in the right-hand panel we plot optimal R&D investment in the three scenarios being examined. Historic data shows the wellknown decline in public expenditure for energy related R&D after the 1980 peak caused by the oil crises. Very low oil prices in the 1990s led to cuts in public expenditure, which have yet to regain momentum despite the oil price surge of the past few years. A very different picture of future R&D investments emerges from the two scenarios considered here. While the baseline scenario foresees low and stable investments in R&D, both climate policy scenarios require a significant innovation effort. For the 450 ppm case, energy expenditures ramp up to roughly 0.07% of GDP, the same share that prevailed in the 1980s. The public sector would thus be required to invest roughly 40–50 billion USD per year, globally, in the years to come; given the long time lags that separate research from commercialization, the innovation effort must be carried out immediately to allow for innovative technologies to become competitive in the medium term9. It should be pointed out that such investment inflow, although sizeable, is two to three orders of magnitude smaller than the investments needed to de-carbonize the energy sector using already existing technologies. The strategy based on R&D investments can thus be thought of as a hedging policy.

The less stringent 550 ppm scenario shows a more gradual innovation pathway, with expenditure rising over time to eventually reach figures similar to those in the 450 ppm scenario, only with a 20- year delay.

A key policy question is where such public R&D investments should be directed to. Table 2 shows the optimal allocation of R&D investment between energy efficiency and de-carbonization programs, in both the electric and non-electric sectors, for the 450 scenario.

It shows that the non-electric sector, particularly to substitute the transport-led non-electric oil demand, should receive most of the innovation funding initially, though over time energy efficiency innovation expenditure increases its relevance and eventually takes the lead (in 2050). The power sector is allocated a smaller but constant share. This shift in the timing is due to the very nature of investment in breakthrough technologies: a flow of investments in specific R&D is needed to continue improving energy efficiency, which exhibits decreasing marginal returns. On the other hand, investing in backstop R&D builds a stockwhich decreases the costs of the technologywith very high returns at the beginning.Once the technology becomes available and economically competitive, then investing in backstop R&D becomes less important as a channel to decrease the price of the backstop technology. In other words, R&D in energy efficiency does not have a permanent effect, while R&D in backstop does. Note also that R&D investment in backstops substitute part of the energy efficiency R&D when the 450 ppm stabilization target is to be achieved without the aid of the backstop technologies, though investments in the backstop technologies remain higher than in the BaU (see Fig. 7).

9The numbers shown include the avoided climate damages induced by the policies. However, the NPV calculations

put most of the weight on early periods for which almost no temperature decrease is achieved, so that gross economic losses are only 10–20% above the ones indicated here.

196

Table 2. Destination of R&D expenditure in a 450 scenario.

Table 3. Total costs of stabilization 2005–100: net present value, percent of GWP losses

at 5% (3% declining) discount rate.

Fig. 7. Energy R&D investments/GDP for BaU and 450 scenarios with and without the

possibility of breakthrough innovation.

197

Fig. 8. Carbon price (left) and GWP loss (right) for a 450 scenario with and without the

possibility of breakthrough innovation. The possibility of technology breakthroughs in the electricity sector also has an effect on

the optimal investments in already known technologies. For example, investments in CCS are crucially affected by the presence of backstop technologies. In the 450 scenario, CCS investment no longer displays the peak effect observed in Fig. 5. The reason for this is the presence of a carbon free backstop in the non-electric sector: it relieves the electricity sector from an excessive mitigation burden, which jeopardized CCS in the long run due to the non-perfect capture rate of carbon.

4. Economic impacts of different technological scenarios The previous sections have illustrated the need for drastic changes in the way we consume

and produce energy. They highlighted the need to mobilize substantial investment resources towards carbon free technologies. This is likely to have important implications for the economic system. In this section, we summarize the economic impact of both 550 and 450 ppm stabilization scenarios, with a particular focus on the role played by energy technologies.

Table 3 shows net present value losses of GWP for both climate policy scenarios and different technology settings10. The reference case shows how, in the 550 ppm scenario, costs are almost negligible, whereas they are significant in the 450 ppm case. The cost difference between the two mitigation policies is a direct consequence of the different magnitudes of energy sector modifications required. It also stems from the non-linearity of endogenous marginal abatement curves in the model. The 450 ppm policy requires drastic cuts in emissions, especially in the second half of the century, when emissions are stabilized at around 3 GtC/yr. With growing economies and population, this entails a significant increase in energy costs, particularly as mitigation gets more and more stringent. The effect of temporal discounting is partially compensated by the growing dimension of economic activity.

The economic effect of limiting the power sector technologies described in Section 3 is shown in the second row. Indeed, if we assume a world in which the expansion of wind and solar technologies is bound by limits to large-scale deployment, the options to expand nuclear energy

10

The numbers shown include the avoided climate damages induced by the policies. However, the NPV calculations put most of the weight on early periods for which almost no temperature decrease is achieved, so that gross economic losses are only 10–20% above the ones indicated here.

198

are limited (possibly because of political or environmental reasons) and Integrated Gasification Combined Cycle (IGCC) with CCS technologies do not become competitive11, then achieving a stabilization target is much more costly, with an increase in the order of 1.5–3 times. On the other hand, allowing for R&D investments in new low carbon technologies, that would enable breakthrough innovation, is shown to be able to substantially reduce the economic policy costs12. These differences are particularly important for the stringent 450 ppm target, which requires a fundamental restructuring of the energy sector13.

However different these scenarios may be, it should be noted that, in the short term, a strong carbon price signal would be needed to bring about what could be called a technology revolution. As shown in Fig. 8, left panel14, the carbon signal of a reference 450 scenario is very similar to that of the most optimistic case of breakthrough inventions.

Higher GWP losses will be experienced initially in the breakthrough technologies case (right panel) in order to make R&D resources available, but this would pay off in the future allowing for the substantial cost reductions shown in Table 3.

5. Conclusions This paper has investigated optimal investment strategies in the energy sector for two

climate policy scenarios. Our results show that the stabilization of CO 2 concentrations at 550

and 450 ppm (650 and 550 CO 2 equivalent) is feasible at reasonable economic costs, but that it requires radical changes in the energy sector and large investments in R&D. Both energy efficiency and the de-carbonization of energy should be pursued. Currently known technologies in the power sector such as nuclear, renewable energy sources and CCS will be essential, but very large investments – greater than the energy sector has ever experienced – will be needed. At the same time, R&D investments for the development of new technologies, especially in the transport sector, will be required. Public R&D expenditures should increase considerably, over the peak levels of the 1980s for at least three decades. Given the long time lags inherent to the innovation process, such investments should be made starting today. Our results thus support the call for R&D policies that complement climate stabilization policies and reduce the costs of limiting dangerous climate change (on this issue, see also Bosetti et al., 2009b). They also indicate that a strong price signal will nonetheless be needed if the climate change challenge is to be met, regardless of whether we expect low carbon breakthrough technologies to be available in the future, because of the inertia in the accumulation of GHG in the atmosphere and low decay rates. Let us conclude by stressing that our results differ from previous analyses of GHG stabilization policies, where a single global economy is usually assumed (Nordhaus’ RICE model

11

The specific constraints used are nuclear energy cannot expand above current generation levels, CCS is not allowed; W S can provide at most 35% of total electricity. 12

Costs become slightly negative in the 550 ppm scenario with breakthrough technological innovation and low discount rate (3% declining). This result depends on the game-theoretic structure of the model that enables us to account for inefficiencies and free riding behavior in technological innovation. In the case of a mild climate policy and low discount rate, the future benefits in terms of lower inefficiencies produced by a mild climate signal are larger than the cost of controlling emissions. 13

See Bosetti et al. (2009a) for a comparison of these results with those on the policy cost increases induced by delayed developing countries’ participation in the global effort to control GHG emissions. 14

The carbon prices displayed assume full country participation to an international carbon market. In case of fragmented or partial agreements, they would rise very significantly (see Bosetti et al., 2008b).

199

and its applications by Nordhaus himself and by Eyckmans, Yang, Finus, Tulkens and others are notable exceptions). The game-theoretic framework used to compute the equilibrium outcomes allows us to capture the noncooperative dynamic strategic interactions among the 12 regions in which the world is divided in WITCH. Regions competeon natural resources use (fuels and the global climatepublic good) and interact strategically on knowledge development and diffusion, as explained in Appendix A. The implications of this richer modeling environment are complex and cannot be appropriately addressed in this article. However, it is worth mentioning that while in our set-up climate policy performs, at least partly, as a coordinationmechanismwhich brings the extraction path of natural resources closer to the global social optimum(the social optimumindeed for the climate global public good), thus driving stabilization costs down with respect to a traditional single economy global model, the non-cooperative representation of knowledge creation and diffusion processes works in the opposite direction, and yields a sub-optimal allocationof resources to technological advancements. This latter effect –aswell as thewell-knownfreeriding incentives on the climate global public good – tends to increase stabilization costs. This explains why our estimates of stabilization costs are higher than those reported by IPCC (2007), despite the presence of endogenous technical change and forward looking decision-makers. Further work on the implications of adopting a non-cooperative game-theoretic framework is still necessary.

Appendix A. Description of WITCH Full details on the WITCH model can be found in Bosetti et al. (2006b). The description

below focuses on the overall model structure, and on the specification of endogenous technical change (ETC) processes.

A.1. Overall model structure WITCH is a dynamic optimal growth general equilibrium model with a detailed

(‘‘bottom-up’’) representation of the energy sector, thus belonging to a new class of hybrid (both ‘‘top-down’’ and ‘‘bottom-up’’) models. It is a global model, divided into 12 macro-regions. A reduced form climate module (MAGICC) provides the climate feedback on the economic system. The model covers CO 2 emissions but does not incorporate other GHGs, whose concentration is

typically added exogenously to CO 2 concentration in order to obtain overall GHG

concentration—a 450 ppm CO 2 concentration scenario is roughly assumed to correspond to a 550 ppm overall GHG concentration scenario in the simulations below. In addition to the full integration of a detailed representation of the energy sector into a macro-model of the world economy, distinguishing features of the model are:

• Endogenous technical change. Advancements in carbon mitigation technologies are

described by both diffusion and innovation processes. Learning by Doing and Learning by Researching (R&D) processes are explicitly modeled and enable to identify the ‘‘optimal’’15

15

Insofar as the solution concept adopted in the model is the Nash equilibrium (see below), ‘‘optimality’’ should not be interpreted as a first-best outcome but simply as a second-best outcome resulting from strategic optimization by each individual world region.

200

public investment strategies in technologies and R&D in response to given climate policies. Some international technology spillovers are also modeled.

• Game-theoretic set-up. The model can produce two different solutions, a cooperative

one that is globally optimal (global central planner) and a decentralized, non-cooperative one that is strategically optimal for each given region (Nash equilibrium). As a result, externalities due to global public goods (CO 2 , international knowledge spillovers, exhaustible resources, etc.) and the related free-riding incentives can both be accounted for, and the optimal policy response (world CO 2 emission reduction policy, world R&D policy) be explored. A typical output of the model is an equilibrium carbon price path and the associated portfolio of investments in energy technologies and R&D under a given environmental target16.

A.2. Endogenous technical change in the WITCH model In the basic version of WITCH, technical change is endogenous and is driven both by

learning-bydoing (LbD) and by public energy R&D investments17. These two drivers of technological improvements display their effects through two different channels: LbD is specific to the power generation industry, while energy R&D affects overall energy efficiency in the economy. The effect of technology diffusion is incorporated based on experience curves that reproduce the observed negative empirical relationship between the investment cost of a given technology and cumulative installed capacity. Specifically, the cumulative installed world capacity is used as a proxy for the accrual of knowledge that affects the investment cost of a given technology:

PRlog

n

tnKAtSC 2),(=1)(−

∑⋅+ (1)

where SC is the investment cost of technology j , PR is the so-called progress ratio that

defines the speed of learning, A is a scale factor and K is the cumulative installed capacity for region n at time t . With every doubling of cumulative capacity the ratio of the new investment cost to its original value is constant and equal to PR1/ . With several electricity production technologies, the model is flexible enough to change the power production mix and modify investment strategies towards the most appropriate technology for each given policy measure, thus creating the conditions to foster the LbD effects associated with emission-reducing but initially expensive electricity production techniques.Experience is assumed to fully spill over across countries, thus implying an innovation market failure associated with the non-appropriability of learning processes.

R&D investments in energy increase energy efficiency and thereby foster endogenous technical change. Following Popp (2004), technological advances are captured by a stock of

16

A stochastic programming version of the model also exists to analyze optimal decisions under uncertainty and learning. However, it was not used within the context of this paper. 17

Due to data availability constraints, only public R&D is modeled in the current version of WITCH. However, private R&D would be expected to respond in a qualitatively similar way to climate change mitigation policies.

201

knowledge combined with energy in a constant elasticity of substitution (CES) function, thus stimulating energy efficiency improvements:

ρρρ αα 1/]),()(),()([=),( tnENntnHEntnES ENH + (2)

where EN(n,t) denotes the energy input, HE(n,t) is the stock of knowledge and ES(n,t) is

the amount of energy services produced by combining energy and knowledge. The stock of knowledge HE(n,t) derives from energy R&D investments in each region through an innovation possibility frontier characterized by diminishing returns to research, a formulation proposed by Jones (1995) and empirically supported by Popp (2002) for energy-efficient innovations in the United States:

))(1,(),(),(=1),( && DR

cbDR tnHEtnHEtnaItnHE δ−++ (3)

where DR &δ is the depreciation rate of knowledge, and b and c are both between 0 and

1 so that there are diminishing returns to R&D both at any given time and across time periods. Reflecting the high social returns from energy R&D, it is assumed that the return on energy R&D investment is 4 times higher than that on physical capital. At the same time, the opportunity cost of crowding out other forms of R&D is obtained by subtracting four dollars of private investment from the physical capital stock for each dollar of R&D crowded out by energy R&D, DR&ψ , so

that the net capital stock for final good production becomes: )),(4),(())(1,(=1),( && tnItnItnKtnK DRDRcccc ψδ −+−+ (4)

where Cδ is the depreciation rate of the physical capital stock. New energy R&D is

assumed to crowd out 50% of other R&D, as in Popp (2004). The WITCH model has been extended to carry out the analysis presented in this paper to include additional channels for technological improvements, namely learning through research or ‘‘learningby- searching’’ (LbS) in existing low carbon technologies (wind and solar electricity, electricity from integrated gasifier combined cycle (IGCC) plants with carbon capture and storage (CCS)), and the possibility of developing breakthrough, zero-carbon technologies for both the electricity and nonelectricity sectors.

A.3. Breakthrough technologies In the enhanced version of the model used for this paper, backstop technologies in both

the electricity and non-electricity sectors are developed and diffused in a two-stage process, through investments in R&D first and installed capacity in a second stage. A backstop technology can be better thought of as a compact representation of a portfolio of advanced technologies. These would ease the mitigation burden away from currently commercial options, but they would become commercially available only provided sufficient R&D investments are undertaken, and not before a few decades. This simplified representation maintains simplicity in the model by limiting the array of future energy technologies and thus the dimensionality of techno-economic parameters for which reliable estimates and meaningful modeling characterization exist. Concretely, the backstop technologies are modeled using historical and current expenditures and

202

installed capacity for technologies which are already researched but are not yet viable (e.g. fuel cells, advanced biofuels, advanced nuclear technologies, etc.), without specifying the type of technology that will enter into the market. In line with the most recent literature, the emergence of these backstop technologies is modeled through so-called ‘‘two-factor learning curves’’, in which the cost of a given backstop technology declines both with investment in dedicated R&D and with technology diffusion (see e.g. Kouvaritakis et al., 2000). This formulation is meant to overcome the limitations of single factor experience curves, in which the cost of a technology declines only through ‘‘pure’’ LbD effects from technology diffusion, without the need for R&D investment (Nemet, 2006). Nonetheless, modeling long term and uncertain phenomena such as technological evolution is inherently difficult, which calls for caution in interpreting the exact quantitative results and for sensitivity analysis (see below)18. Bearing this caveat in mind, the investment cost in a technology tec is assumed to be driven both by LbS (main driving force before adoption) and LbD (main driving force after adoption), with ttecP , , the unit cost of

technology tec at time t, being a function of the dedicated R&D stock ttecDR ,& and deployment

ttecCC , :

d

tec

Ttec

e

tec

Ttec

tec

Ttec

CC

CC

DR

DR

P

P−−

−

×

,0

,

,0

2,

,0

,

&

&= (5)

where the R&D stock accumulates with the perpetual inventory method and CC is the

cumulative installed capacity (or consumption) of the technology. A two-period (10 years) lag is assumed between R&D capital accumulation and its effect on the price of the backstop technologies, capturing in a crude way existing time lags between research and commercialization. The two exponents are the LbD index (_d) and the learning by researching index (_e). They define the speed of learning and are derived from the learning ratios. The learning ratio lr is the rate at which the generating cost declines each time the cumulative capacity doubles, while lrS is the rate at which the cost declines each time the knowledge stock doubles. The relation between d, e, lr and lrS can be expressed as follows:

ed lrSandlr −− −− 2=1 2=1 (6)

The initial prices of the backstop technologies are set at roughly 10 times the 2002 price

of commercial equivalents. The cumulative deployment of the technology is initiated at 1000 TWh, an arbitrarily low value (Kypreos, 2007). The backstop technologies are assumed to be renewable in the sense that the fuel cost component is negligible. For power generation, it is assumed to operate at load factors (defined as the ratio of actual to maximum potential output of a power plant) comparable with those of baseload power generation. This formulation has received significant attention from the empirical and modeling literature in the recent past (see, for instance, Criqui et al., 2000; Barreto and Kypreos, 2004; Klassen et al., 2005; Kypreos, 2007; Jamasab, 2007; Söderholm and Klassen, 2007). However, estimates of parameters controlling the learning processes vary significantly across available studies. Here, averages of existing values 18

This is especially true when looking at the projected carbon prices and economic costs at long horizons—typically beyond 2030, while the short-run implications of long-run technological developments are comparatively more robust across a range of alternative technological scenarios (see below).

203

are used, as reported in Table 1. The value chosen for the LbD parameter is lower than those typically estimated in single factor experience curves, since here technological progress results in part from dedicated R&D investment. This more conservative approach reduces the role of ‘‘autonomous’’ learning, which has been seen as overly optimistic and leading to excessively low costs of transition towards low carbon economies19. Backstop technologies substitute linearly for nuclear power in the electricity sector, and for oil in the non-electricity sector. Once backstop technologies become competitive thanks to dedicated R&D investment and pilot deployments, their uptake is assumed to be gradual rather than immediate and complete. These penetration limits are a reflection of inertia in the system, as presumably the large deployment of backstops would require investment in infrastructures and wide re-organization of economic activity. The upper limit on penetration is set equivalent to 5% of the total consumption in the previous period by technologies other than the backstop, plus the electricity produced by the backstop in the electricity sector, and 7

A.4. Spillovers in knowledge and experience In addition to the international LbD spillovers mentioned above, WITCH also features

international spillovers in knowledge for energy efficiency improvements. The amount of spillovers entering each world region is assumed to depend both on a pool of freely available world knowledge and on the ability of each country to benefit from it. In turn, this absorption capacity depends on the domestic knowledge stock, which is built up through domestic R&D according to a standard perpetual capital accumulation rule. The region then combines knowledge acquired from abroad with the domestic knowledge stock to produce new technologies at home. For details, see Bosetti et al. (2007b).

References

[1] Bosetti, V., Carraro, C., Galeotti, M., Massetti, E., Tavoni, M., 2006a. WITCH: aWorld Induced Technical Change Hybrid model. The Energy Journal 13–38 Special Issue on Hybrid modeling of energy–environment policies: reconciling bottom-up and topdown.

[2] Bosetti, V., Massetti, E., Tavoni, M., 2007a. The WITCH model: structure, baseline and solutions. FEEM Working Paper 10-2007, Milan.

[3] Bosetti, V., Carraro, C., Sgobbi, A., Tavoni, M., 2008a. Delayed action and uncertain targets. How much will climate policy cost? FEEM Nota di Lavoro 69.08, CEPR Working Paper 6973 and CESifo Discussion Paper 2403.

19

Problems involved in estimating learning effects include: (i) selection bias, i.e. technologies that experience smaller cost reductions drop out of the market and therefore of the estimation sample; (ii) risks of reverse causation, i.e. cost reductions may induce greater deployment, so that attempts to force the reverse may lead to disappointing learning rates a posteriori; (iii) the difficulty to discriminate between ‘‘pure’’ learning effects and the impact of accompanying R&D as captured through two-factor learning curves; (iv) the fact that past cost declines may not provide a reliable indication of future cost reductions, as factors driving both may differ; (v) the use of price – as opposed to cost – data, so that observed price reductions may reflect not only learning effects but also other factors such as strategic firm behavior under imperfect competition.

204

[4] Bosetti, V., Carraro, C., Tavoni, M., 2008b. Delayed participation of developing countries to climate agreements: should action in the EU and US be postponed? FEEM Nota di Lavoro 70.08, CEPR Working Paper 6967 and CESifo Discussion Paper 2445.

[5] Bosetti, V., Carraro, C., Tavoni, M., 2009a. Climate policy after 2012. Timing, technology, participation. CESifo Economic Studies.

[6] Bosetti, V., Carraro, C., Duval, R., Sgobbi, A., Tavoni, M., 2009b. The role of R&D and technology diffusion in climate change mitigation: new perspectives using the WITCH model. OECD Working Paper, January 2009.

[7] Energy Information Administration, 2007. Annual Energy Outlook 2007. Available at: www.eia.doe.gov.

[8] IPCC, 2007. IPCC Fourth Assessment Report—Working Group III, IPCC, Geneva.

[9] Newell, R., Hall, D., 2007. U.S. Climate Mitigation in the Context of Global Stabilization. Resources for the Future, Washington, DC.

[10] Nordhaus, W.D., Boyer, J., 2000. Warming the World. MIT Press, Cambridge.

[11] Pacala, S., Socolow, R., 2004. Stabilization wedges: solving the climate problem for the next 50 years with current technologies. Science 305 (5686), 968–972.

Appendix references

[12] Barreto, L., Kypreos, S., 2004. Endogenizing R&D and market experience in the ‘‘bottom-up’’ energy-systems ERIS model. Technovation 2, 615–629.

[13] Bosetti, V., Carraro, C., Galeotti, M.,Massetti, E., Tavoni, M., 2006b. WITCH: aWorld Induced Technical Change Hybrid model. The Energy Journal 13–38 Special Issue on Hybrid modeling of energy–environment policies: reconciling bottom-up and topdown.

[14] Bosetti, V., Massetti, E., Tavoni, M., 2007b. The WITCH model: structure, baseline, solutions. FEEM Working Paper Series 10/2007, FEEM, Milan.

[15] Criqui, P., Klassen, G., Schrattenholzer, L., 2000. The efficiency of energy R&D expenditures. Proceedings of the Workshop on Economic Modeling of Environmental Policy and Endogenous Technical Change. Amsterdam, November 16–17, 2000.

[16] Jamasab, T., 2007. Technical change theory and learning curves: patterns of progress in electric generation technologies. The Energy Journal 28 (3), 51–71.

[17] Jones, C., 1995. R&D based models of economic growth. Journal of Political Economy 103, 759–784.

[18] Klassen, G., Miketa, A., Larsen, K., Sundqvist, T., 2005. The impact of R&D on innovation for wind energy in Denmark, Germany and the United Kingdom. Ecological Economics 54 (2–3), 227–240.

[19] Kouvaritakis, N., Soria, A., Isoard, S., 2000. Endogenous learning in World Post-Kyoto Scenarios: application of the POLES model under adaptive expectations. International Journal of Global Energy Issues 14 (1–4), 228–248.

205

[20] Kypreos, S., 2007. A MERGE model with endogenous technical change and the cost of carbon stabilization. Energy Policy 35, 5327–5336.

[21] Nemet, G.F., 2006. Beyond the learning curve: factors influencing cost reductions in photovoltaics. Energy Policy 34 (17), 3218–3232.

[22] Popp, D., 2002. Induced innovation and energy prices. American Economic Review 92 (1), 160–180.

[23] Popp, D., 2004. ENTICE: endogenous technological change in the DICE model of global warming. Journal of Environmental Economics and Management 48, 742–768.

[24] Söderholm, P., Klassen, G., 2007. Wind power in Europe: a simultaneous innovation–diffusion model. Environmental and Resource Economics 36 (2), 163–190.

206

The Social Costs of Greenhouse Gas Emissions: An Expected Value Approach

Samuel Fankhauser

Abstract: This paper provides an order-of-magnitude assessment of the marginal social

costs of greenhouse gas emissions. The calculations are based on a stochastic greenhouse damage model in which key parameters are random. This allows a closer representation of current scientific understanding and also enables calculation of a damage probability distribution. Thus, we account explicitly for the uncertain nature of the global warming phenomenon. We estimate social costs of CO2 emissions in the order of 20 $/tC for emissions between 1991 and 2000, a value which rises over time to about 28 $/tC in 2021-2030. Similar figures for CH4 and N2O are also provided. As a consequence of the prevailing uncertainty, the standard deviation of the estimates is rather high. The distribution is positively skewed, which implies that the currently predominant method of using best guess values will lead to an underestimation of the expected costs of emissions.

Keywords: social cost of carbon, greenhouse gas emission costs, mathematical models, socio-economic factors

207

Introduction There is now a wide and growing body of literature on the potential impacts of global

warming. Most notably this includes the work by the U.S. Environmental Protection Agency (Smith and Tirpak, 1989) and by the Intergovernmental Panel on Climate Change (IPCC), whose Working Group Two is entirely devoted to the impacts of climate change (IPCC, 1990c).

In addition there are numerous studies on particular aspects of the problem, including for example Parry et al. (1991) and Parry (1990) on agriculture, Titus et al. (1991) on sea level rise, Peters and Lovejoy (1992) on biological diversity, Waggoner (1990) on water, and World Health Organisation (1990) on health effects, to name only a few. In most parts this work is on a descriptive level, though, or limited to a quantification in physical terms. Few attempts exist to a monetary quantification of global warming damage (Nordhaus, 1991a, b; Cline, 1992a; Titus, 1992; Fankhauser, 1993, 1992).

On a policy level, a monetary assessment of greenhouse damage is crucial. A comparison between the costs of greenhouse prevention and the benefits from avoided warming is only feasible if damage can be expressed in monetary terms. Similarly, a monetary estimate is required to assess individual abatement projects such as those financed by the Global Environment Facility (GEF). Considerable effort has recently been put into analysing the social costs of the fuel cycle, with the aim of deriving externality adders which are to be put onto the price of fossil fuels to internalise the social costs of fuel consumption (see e.g. Hohmeyer, 1988; PACE, 1990; Pearce et al, 1992; Lockwood, 1992). The studies typically concentrate on classic air pollutants like NO x and SO x . To complete the picture an additional adder would be required reflecting the

social costs of global warming. The aim of the present paper is to fill this gap and provide an order-ofmagnitude

assessment of the social costs, or the shadow value, of greenhouse gas emissions. Assessing greenhouse damage is not possible without accounting, in one way or another, for the huge uncertainty prevailing in the global warming debate. Although scientists have achieved a remarkable consensus with respect to many aspects, our ignorance of global warming impacts is still vast, particularly with respect to regional and long-term impacts. Most studies allow for uncertainty by working with different climate scenarios. In the present paper we chose a different approach and incorporated uncertainty directly by describing uncertain parameters as random. Using a stochastic model of this type has several advantages. First, it allows a better representation of current scientific understanding. Scientific predictions usually take the form of a best guess value supplemented by a range of possible outcomes. Concentrating on the best guess value therefore neglects a large part of the information provided, while, on the other hand, a stochastic model can make full use of it. Secondly, and probably more importantly, a stochastic model allows the calculation of an entire damage probability distribution, thereby providing important additional information on the likelihood of the estimates and the possibility of extremely adverse events.

208

Notes: Marginal damage costs depend on the level of emissions. In the usual case of an upwards sloping

MD curve they are rising with the level of emissions, as shown in Figure 1. Optimisation models calculate marginal costs at the point where marginal damage equals marginal abatement costs-the dashed situation labelled ``opt''. The present model calculates marginal costs at the emissions level actually observed. Since emissions are uncertain, and so is the exact shape of the marginal damage curve, calculations will result in a range of possible cost figures, each of which is assigned a certain probability of occurrence --- the probability distribution shown to the left of the Figure. Care should nevertheless be exercised when interpreting the figures presented below.

Although, as we believe, based on the best available scientific information, they cannot provide anything better than a rough order-of-magnitude assessment. A distinction should also be drawn between the actual marginal costs of greenhouse gas emissions and the shadow value along the optimal emissions path. This paper concerns the former, as explained above in Figure 1. Our results give therefore little indication about the socially optimal carbon tax on an international level, the calculation of which would require an optimal control model (see Nordhaus, 1992, 1993a,b; Peck and Teisberg, 1992, 1993a,b)1. Arguably, a figure on the actual costs may be more relevant for individual abatement projects, however. As will become clear later, the shadow value of greenhouse gas emissions depends on the amount of emissions discharged in the future. Optimal shadow values would therefore only be relevant for actual projects if the world was indeed to follow the optimal emissions trajectory calculated in the model. There is no guarantee that this will be the case. The current approach, which treats future emissions as uncertain, seems

1A conceptually different approach would be to impose an exogenously given carbon concentration or emissions

constraint. The shadow value of carbon is then determined according to Hotelling's rule, see Anderson and Williams (1993).

209

therefore more realistic. Arguably, the resulting range will also encompass the optimal path, see Figure 1.

Also note that, for the same reason, the figures are only relevant for small scale abatement projects, which do not significantly affect the trajectory of future emissions. The appraisal of large projects such as an international carbon agreement, which affect future emission levels, is somewhat more complex, and would require an adjustment of the future emission trajectory. Given that with the exception of the top four emitters no country accounts for more than 4% of total greenhouse gas emissions, large projects in the above sense will however be the exception, and are arguably confined to internationally concerted efforts.

The structure of the paper is as follows. In the next section we review existing estimates of both the costs of CO 2 concentration doubling and of estimates of the shadow value of carbon. We then introduce the stochastic model utilised in this paper, and present the resulting estimates. The last section outlines policy implications and concludes.

1. Existing Damage Estimates 1.1. The Damage from a Concentration Doubling Scientific research on greenhouse impacts has so far almost entirely concentrated on the

benchmark case of warming under an atmospheric CO 2 , concentration of twice the preindustrial

level (2xCO 2 ). As a consequence studies on the economic costs of global warming have tended

to concentrate on the same benchmark. By far the best studied aspects of 2xCO 2 damage are the impacts on agriculture (e.g. Kane et al., 1992; Parry, 1990; Parry et al., 1988) and the costs of sea level rise (e.g. IPCC, 1990b; Titus et al., 1991; Rijsberman, 1991). There are nonetheless some studies which try to provide a more comprehensive picture of global warming damage by including all damage aspects. The pioneering paper in this area is Nordhaus (1991a,b). Still mainly concentrating on the costs of agriculture and sea level rise, he estimated an overall damage of global warming in the order of a quarter percent of GNP. To allow for the many non-market impacts neglected in the study this value is raised to 1%, with a range of error of 0.25-0.2%. The figures are based on U.S.-data, but Nordhaus claims that they may hold worldwide. Improvements on Nordhaus' back-of-the-envelope estimate have been provided by Cline (1992a) and Titus (1992), two papers again focusing on the United States, and by Fankhauser (1993, 1992), who distinguishes between several geopolitical regions. Despite considerable differences in individual damage categories, the three studies roughly agree on the overall result, all predicting a 2xC0 2 damage in the order of 1% to 2% of world GNP.

On a regional level, the impacts of climate change for the four U.S. states Missouri, Iowa, Nebraska, and Kansas have been analysed in the so-called MINK study, probably the most comprehensive regional damage estimation project undertaken so far2. Concentrating on agriculture, forestry, water and energy, the study concludes that global warming (approximated by the climate of the 1930s) would probably not reduce regional income by more than 1%. The impacts on agriculture and water would be most strongly felt.

2See Rosenberg and Crosson (1991) for a summary, and also the Special Issue of Climatic Change 24(1/2) (1993)

for a more detailed description.

210

Despite the attention equilibrium studies like the 2xC0 2 , case enjoy in the literature, they are not directly relevant for practical purposes. For the appraisal of abatement projects it is more important to know the marginaI costs per tonne of emission, rather than total damage. We turn to this aspect next.

1.2. The Damage per Tonne of Emission Most studies estimating the social costs of greenhouse gas emissions do so in an optimal

control framework, and primarily aim at calculating the socially optimal greenhouse emissions trajectory over time. In such a set up the social costs of emissions are equivalent to the pollution tax required to keep emissions on the optimal path. The pioneering paper on the shadow price of CO 2 emissions is again Nordhaus (1991a,b). Using a simplified approach which does not constitute a fully fledged optimal control model, he calculates social costs of 7.3 $ per tonne of carbon ($/tC) emitted. Imposing different assumptions on the rate of discount and the 2xC0 2 damage leads to a range of 0.3 $/tC to 65.9 $/tC. Implying that abatement should only be undertaken as long as costs do not exceed 7.3 $/tC abated, the estimates formed the backbone of Nordhaus' claim that global warming may not, after all, be such a big problem, and may justify only a modest policy response.

This view has been fiercely criticised by many authors (see for example Ayres and Walter, 1991; Daily et al., 1991; Grubb, 1993). The main objection concerned Nordhaus' 2xC0 2 estimate which has repeatedly been attacked as being too low. Only few of the criticisms appear to be based on sound analysis, though, and more important than the problems with 2xC0 2 damage are probably the shortcomings of the model itself (see Cline, 1992a). Particularly questionable is the assumption of a resource steady state, which inter alia implies a constant level of CO 2 emissions over time. Obviously this is unrealistic. The IPCC for example predicts an

increase in annual CO 2 emissions from about 7 gigatonnes (GtC) in 1990 to about 9-14 GtC by 2025 (IPCC, 1992). The simple (linear) structure of the climate and damage sectors also implies that costs will remain constant at 7.3 $/tC throughout. Climate processes are clearly non-linear, and the costs of CO 2 emissions will thus depend on future concentration and warming levels, i.e. they will vary over time. Subsequent estimates suggest that they may in fact rise over time. That is, a tonne of CO 2 added to an already large stock of atmospheric CO 2 is likely to cause a higher damage than a tonne emitted under a low concentration level.

These objections are also relevant to the study by Ayres and Walter (1991), whose calculations are based on the Nordhaus model. The paper has additional shortcomings. For example, by considering both the costs of sea level rise protection and the costs of climate refugees from coastal regions, the authors appear to double count at least some of the sea level rise impacts. On the whole, their cost estimate of 30-35 $/tC must therefore be regarded as suspect.

The shortcomings of the earlier model were recognised and corrected in Nordhaus' subsequent approach, the DICE (Dynamic Integrated Climate Economy) model (Nordhaus, 1992; 1993a, b). DICE is an optimal growth model in the Ramsey tradition, extended to include a climate module and a damage sector which feeds climate changes back to the economy. The shadow values of carbon following from DICE are in the same order as Nordhaus' previous results, starting at 5.3 $/tC in 1995 and gradually rising to 6.8 $/tC in 2005 and 10 $/tC in 2025 (see Table 1). Note that figures for future periods are current value estimates, i.e. they denote the

211

social costs valued at the time of emission. The DICE model was also used by Cline (1992b), who concludes that Nordhaus' choice of parameter values may have lead to an underestimation of the true costs. Unfortunately, Cline's paper only reports alternative emission trajectories, but not the corresponding shadow values.

Figures slightly higher than those by Nordhaus were suggested by Peck and Teisberg

(1992, 1993a,b), who came up with a shadow value of carbon of about 10$/tC in 1990, rising to about 22 $/tC by 2030 (see Table 1). The CETA (Carbon Emission Trajectory Assessment) model, on which their calculations are based, possesses a similar climate and damage sector as DICE, but is more detailed on the economy side by incorporating a carefully modelled energy sector3. Differences between the estimates appear to be mainly due to different assumptions about the size of 2xC0 2 damage. Common to both papers is the assumption of a 3% utility discount rate, a figure which may be rather high, according to many authors (Cline, 1992a; Hoel and Isaksen, 1993).

2. A Stochastic Greenhouse Damage Model 2.1. General Description Greenhouse gases are so-called stock pollutants. That is, global warming damage is not

caused by the flow of emissions as such, but by their accumulation in the atmosphere. Consequently a tonne of emissions has its impact not only in the period of emission, but over several time periods --- as long as the gas, or fractions of it remain in the atmosphere. The

3This latter feature is of course of less relevance in the present context.

212

damage costs of a tonne of emission of gas i, S i , are thus a present value figure --- the discounted

sum of future incremental damage, (0))/( iEtD ∂∂ ,

0

( )=

(0)t

t

i

D tS e dt

E

τδ−∂

∂∫ (61)

where δ is the discount rate. The equation can be further disaggregated, as follows. A marginal increase in base period emissions will first of all lead to an increase in the

atmospheric concentration of gas i . The impact on concentration at time 0>s , (0))/( ii EsQ ∂∂ ,

mainly depends on the removal rate of gas i and on possible interactions with other gases. The effect of a higher atmospheric concentration in gas i on global mean temperature (which stands as an index for a whole set of climatic changes) is stsQtT i >),()/( ∂∂ . It depends on a variety of

factors, including the radiative forcing capability of gas i , the atmospheric concentration of other greenhouse gases, climatic feedback effects and the inertia of the climate system. If, finally, an increase in global mean temperature causes a marginal damage of )()/( tTtD ∂∂ , the incremental damage at time t from a marginal increase in base period emissions of gas i is,

dsE

sQ

sQ

tT

tT

tD

E

tD

i

i

i

t

i

∂

∂⋅

∂

∂

∂

∂

∂

∂∫ (0)

)(

)(

)(

)(

)(=

(0)

)(0

(62)

see Schmalensee (1993). The marginal costs of an incremental tonne of emissions are then

0 0

( ) ( ) ( )=

( ) ( ) (0)

tti

i

i i

D t T t Q sS ds e dt

T t Q s E

τδ−

∂ ∂ ∂⋅

∂ ∂ ∂∫ ∫ (63)

The estimation of the social costs of greenhouse gas emissions thus requires several

pieces of information. First we require a module which represents the world climate system and transforms an increase in emissions into incremental atmospheric concentration and then warming. The climate system being a highly complex process, models aiming at its simulation, such as Global Circulation Models (GCMs), tend to be quite big. Smaller scale representations are often able to provide a sufficiently close approximation, though, and our model is based on such an approximation.

Second, we need to know more about the shape of the damage function. Research has almost uniquely concentrated on 2xCO 2 , and information about the damage before and after this benchmark is therefore extremely difficult to obtain. Assumptions are often rather ad hoc. Long-run damage in particular also heavily depends on the assumed discount rate, the choice of which is a highly controversial issue.

Finally, because the climate process is non-linear, the shadow value of a tonne of emissions today depends on the level of emissions in future periods, and we need predictions about these. Contrary to the optimal control problems introduced in the previous section, future emissions have to be exogenously determined in the present model. The social costs of emissions in a certain period are then achieved by increasing emissions by one tonne in this period and comparing the stream of annual damages before and after the increase.

Future emissions, discount rate, shape of damage function and many of the climatic parameters are highly controversial variables about which little is known with certainty. Uncertainty is explicitly taken into account in our model by modelling all key parameters as random. In the basic case we assumed triangular distributions for all variables. The triangular specification appeared to be an obvious choice, given that most scientific predictions take the

213

form of a lower bound, upper bound and best guess estimate, values which can directly be used as distribution parameters without requiring further alterations. Where no upper and lower bounds are available we usually assumed a rather modest range of ± 10%. Alternative distributions will later be introduced for some parameters to incorporate low probability/high impact events. In the remainder of this section the model is introduced in full detail.

2.2. Future Emissions The model distinguishes between 10 sources of emission, including CO 2 from fossil fuels

and deforestation, CH 4 from several sources, N 2 O, and three types of CFCs. Emissions follow an exogenous business-as-usual (BAU) path, where the growth rates of each emission source were calibrated to mimic the IPCC (1992) scenarios4. The model extends over τ = 230 years, and growth rates may vary over time. For CO 2 emissions from fossil fuel combustion the

procedure is slightly more complex in that the rate of growth of emissions at time tgt, , consists

of several factors, as follows ttttt pyfcg +++= (64)

where tc is the percentage change in carbon intensity (carbon emitted per energy unit), tf the

increase in energy efficiency (energy used per unit of GNP), ty the change in per capita income

and tp the rate of population growth. The relationship follows from taking time derivatives of

the following tautological relation (see Proops et al., 1993),

=carbon energy GNP

carbon emissions populationenergy GNP population

⋅ ⋅ ⋅ (65)

Note that, while all growth rates are random, they may not be independent of each other.

For example, a rapid decrease in carbon intensity will not only manifest itself in a significantly negative value for tc , but may also lead to a lower growth rate of CH 4 emissions from coal

mining and a higher growth rate of CH 4 from gas leakage. The reason for this is that a lower carbon intensity is likely to be achieved through switching from carbon intensive coal to cleaner gas, which will affect the methane emissions from these sources. In a similar way CH 4 from

biomass burning is correlated to CO 2 from deforestation, and growth rates in later periods depend on those observed previously. The model correlates the realisations of such variables, i.e. if CO 2 emissions from deforestation are assumed to be high in a particular model run, the

realisation for the CH 4 -biomass parameter is likely to be high as well. As mentioned above the emission rates determined in this way describe a BAU path. That

is, they do not incorporate measures which might be taken to reduce the emission of greenhouse gases, although they cover measures to combat air pollution and ozone layer depletion, which will affect emissions indirectly. There is no reason to a priori exclude greenhouse gas abatement, and to take this option into account BAU emissions were weighted against an abatement path. Emissions follow BAU with probability π , and with probability (1-π ) they are reduced through an abatement strategy. Emissions of gas i at time i

tEt, , are therefore

4An Appendix with a detailed description of parameters and parameter values is available upon request.

214

iit

it STABBAUE ⋅−+⋅ )(1= ππ (66)

where iSTAB represents emissions under the abatement policy. For CO 2 from fossil fuels this means stabilisation at 1990 levels, in the case of emissions from natural sources a complete halt of deforestation. No weighting was taken in the case of N 2 0 and for CFCs. For the latter stringent abatement regimes are already incorporated in BAU.

Accelerated abatement will not be costless and weighted future income will therefore be below BAU income. The costs of deforestation were estimated at 4 $/tC (Cline, 1992a). For the costs of carbon abatement we used a back-of-the-envelope method suggested by (Cline, 1993). He estimated an elasticity of income with respect to energy inputs of about 0.06-0.08, based on the share of energy in total production value. He further assumed an elasticity of energy availability with respect to carbon of about 0.5. Carbon abatement of x percent would thus reduce GNP by (0.07).(0.5) ⋅x percent.

2.3. Atmospheric Concentration The accumulation of emissions in the atmosphere is modelled through a set of equations

of the following form

iti

it

i

it EQ

LQ β+⋅

− −1

11= (67)

That is, the atmospheric concentration of a gas, i

tQ , is assumed to depreciate at a constant

rate 1/ iL , where iL is the lifetime of gas i 5. iβ is a conversion parameter which transforms

emissions (measured in tonnes) into concentration, measured in parts per million (ppm). One gigatonne of emissions corresponds to 5.66 ppm, divided by the atomic weight of the gas, thus in the case of CO 2 for example 1 GtC = 5.66/12 = 0.47 ppm.

For two gases the concentration equation differs from the generic form given above. In the case of methane the equation was extended to contain two decay factors, one representing atmospheric depreciation and the other the accumulation of CH 4 in the soil (Wigley and Raper,

1992). In the case of CO 2 the notion of a constant depreciation rate would he misleading,

because CO 2 is a very stable compound which does not easily decay. Rather, CO 2 is transferred from the atmosphere into other reservoirs (e.g. oceans), from where it may return into the atmosphere. To represent this carbon cycle, Maier-Reimer and Hasselmann (1987) have suggested to divide emissions into five categories, each with a constant, but different atmospheric lifetime. That is, each category is modelled as in equation (7), and atmospheric concentration is a weighted sum of all five categories (see also Lashof and Ahuja, 1990). The Maier-Reimer and Hasselmann formulation, together with their suggested parameter values, was adopted here6.

5The atmospheric lifetime of a gas is defined as the time it takes to reduce a given increase in concentration to

1/e=36.8% of its initial value (IPCC, 1990a). 6The formulation was assumed to hold for all emissions since preindustrial time. Concentration in the base period

1990 is therefore already the weighted sum of the 5 categories, as introduced above. Using historic emission rates we

derived a CO 2 concentration level of 349 ppm in 1990, close to IPCC's (1990a) estimate of 353 ppm.

215

Uncertainty was introduced by treating the weights as random variables, while lifetimes are assumed to be fixed.

2.4. Radiative Forcing As a next step, increases in concentration have to be transformed into an increase in

radiative forcing, that is into an increased flux of energy reemitted from the atmosphere back to the earth. The relevant representations for this were taken from IPCC (1990a). Because there appears to be relatively little controversy about this step, the parameter values in the following equations were assumed to be non-random. In the case of CO 2 the relationship between concentration and radiative forcing, F, is logarithmic

2

22 =

CO

p

CO

tCO

tQ

QlnF µ (68)

where 2CO

pQ denotes preindustrial concentration. For CH 4 and N 2 O the relationship is

iip

it

iit OQQF −− )(= φ (69)

where iO is a relatively complicated overlap term between CH 4 and N 2 O, described in ?. For CFCs finally, the relationship is linear

)(= ip

it

iit QQvF − (70)

The forcing effect of CFCs is diminished through an ozone feedback. Because CFCs react

with stratospheric ozone, which is itself a greenhouse gas, they have an indirect cooling effect. Wigley and Raper (1992) postulate that the ozone depletion potential of a CFC is proportional to the number of chlorine atoms it contains and propose the following approximation

jt

j

CFCt QnaFB ⋅⋅∑= (71)

where a is a factor of proportion (assumed to be random), and jn the number of chlorine atoms in CFC j .

Total radiative forcing in a period is simply the sum of all individual contributions t

it

it FBFF −∑= (72)

It should be noted that this representation neglects the potential cooling effect from

sulphur emissions. Through an increased backscattering of incoming solar radiation sulphur aerosols may cause a noticeable reduction in observed warming (see IPCC, 1992).

2.5. Temperature Rise The characterisation of temperature rise was taken from Nordhaus (1992), whose

representation itself draws on Schneider and Thompson (1981). The climate system is represented by a multi-stratum system consisting of three layers, the atmosphere, upper oceans and deep oceans. Increased radiative forcing warms up the atmosphere, which in turn heats the upper oceans, which then warm up the deep oceans. Both ocean layers impose a certain amount

216

of thermal inertia on the system, which therefore only adjusts gradually to an increase in forcing. The system is represented by two partial adjustment equations

−−−+ −−−− )(

1= 1111

lt

ut

lu

ttuu

tu

t TTR

TFR

TTθ

λ (73)

−+ −−− )(

1= 111

lt

ut

l

ll

tl

t TTR

RTT

θ (74)

u

tT and ltT are the temperature of the upper and lower layer of oceans, respectively

(relative to the preindustrial level). uR and lR denote the thermal capacity of the layers, and θ the transfer rate between upper and deep oceans. λ is a climate feedback parameter, related to the climate sensitivity of the system. That is, it indicates by how much temperature changes for a given increase in radiative forcing.

2.6. Annual Damage As mentioned above most research on global warming damage has focused on 2xCO 2 ,

and the representation of annual damage was consequently centred around this point. Studies estimating the monetary damage of 2xC0 2 were surveyed above. They all perform more or less the same hypothetical exercise, and estimate the costs of global warming damage which would occur if a society with the economic structure of today was, at some point in the future, exposed to a temperature increase of 2.5 to 3 C, the temperature rise usually associated with 2xC0 2 . To generalise from this particular set up, we need to expand the representation in three directions. First, we have to model how damage changes under temperatures different from that associated with 2xC0 2 . Second, we have to capture the impacts of time and the rate of temperature rise, i.e.

we want to know how damage alters if 2xC0 2 is reached earlier, or later, than initially assumed. Third, we need to know how damage develops as the economy changes and population grows. The three components are incorporated in a damage function of the form

)*()(1= ttu

ttt

TkD −+⋅

Λφ

γ

(75)

where the two parameters Λ and *t represent two key assumptions of the 2xC0 2 studies surveyed earlier, viz. the amount of warming associated with concentration doubling and the time at which it is assumed to occur. The values are set at Λ =2.5 C and *t =2050. Note that when

Λ=utT and *= tt annual damage in period t becomes ttt kkD .= therefore represents the 2xC0 2

estimates introduced above, adjusted for economic and population growth. The two parameters γ

and φ then determine the damage outside the 2xC0 2 benchmark. Parameter γ defines the relationship between temperature and damage. If temperature

rises by 1%, damage rises by γ %. Little is known about the value of γ , although it can reasonably be assumed that 1γ , i.e. that damage is convex in temperature. The work by Cline (1992a) and a poll of experts by Nordhaus (1993c) both suggest a value in the order of 1.3.

217

Simulation studies have typically used values between 1 and 3 (Nordhaus, 1991a,b; 1992; 1993a,b; Peck and Teisberg, 1992; 1993a,b).

A temperature rise of 3 C by 2100 is not the same as 3 C obtained in 2025. Damage may be considerably reduced if the system is given enough time to adapt. This is particularly the case for ecosystems, for which the rate of temperature change may be more crucial than the absolute change itself. Under a sufficiently slow change natural systems will be able to acclimatise or migrate to more favourable areas. A rapid change on the other hand may lead to the extinction of some of the more vulnerable species. In the case of economic damage, a slow rise in temperature may allow a gradual adjustment within the normal process of depreciation and reinvestment, which will be considerably cheaper than the immediate and premature replacement of equipment. While such considerations are not directly modelled, they are on an ad hoc basis embodied in the parameter 4, a factor which augments impacts if they occur earlier than initially assumed (i.e. before *t ), and diminishes them if they are delayed. φ is assumed to be random with a best guess value of 0.006, a figure which was derived from Nordhaus' poll of experts (Nordhaus, 1993c).

This leaves the third question of how damage changes as a consequence of economic development and population growth. In Nordhaus (1992, 1993a,b) damage is assumed to grow directly proportional to income, while in the CETA model it is scaled with a labour input index, i.e. it grows in proportion to the labour force (expressed in efficiency units, see Peck and Teisberg, 1992, 1993a,b). In the present model we distinguish between market based damage, apparent in the national accounts, and non-market based damage, such as people's valuation of a species or a warmer climate. The former is assumed to grow in proportion to GNP,

)ˆ(1=1

ttYt

Yt py

k

k++

−

(76)

where tp denotes the rate of growth of population, as introduced earlier. ty is the rate of growth

of per capita income, taking into account the possible costs of carbon abatement. For non-market impacts the formula is slightly more complex. If people's willingness to

pay for non-market goods was constant over time, non-market damage would simply grow at the same rate as population. However, this will not be the case. As people become richer over time, their willingness to pay will also change. Defining Yε as the income elasticity of people's willingness to pay to avoid non-market damage, the growth factor of non-market damage should therefore be written as

)ˆ(1=1

ttYPt

Pt py

k

k+⋅+

−

ε (77)

Estimates of Yε for environmental goods in general are surveyed by Pearce (1980).

Without finding any conclusive results, the survey seems to suggest an elasticity value in the order of 1=Yε . Total greenhouse damage is simply the sum of the two elements, P

tYtt kkk += .

The initial values 0k , Yk0 and Pk0 were derived from the 2xC0 2 studies surveyed earlier. Based

on the damage classification and estimates of Fankhauser (1993, 1992) we assumed that 62% of the damage is non-market related and 38% income related.

2.7. Discounting

218

Greenhouse damage spans over an extremely long time horizon of several centuries. Discounting future impacts is therefore a crucial issue. In our model we follow the standard consumption equivalent technique, as described in Lind (1982). In the context of global warming this method has been used by e.g. Cline (1992a).

In an ideal world without distortions, damage flows could be discounted either at the social rate of time preference, or at the marginal product of capital. The two rates would be equal. In reality they will generally be different, however, and the appropriate rate of discount for greenhouse damage will therefore depend on whether climate change affects investment or consumption goods. Consumption based damages should be discounted at the social rate of time preference, and investment related damage at the rate of return on capital. Alternatively, all flows could first be transformed into consumption units (by multiplying them with the shadow price of capital) and then discounted uniformly at the social rate of time preference. This is the method followed here. In modelling this process, two parameters therefore have to be defined, the social rate of time preference, δ , and the shadow value of capital, ν .

The social rate of time preference, or consumption discount rate, consists of two elements. The first one, called the pure rate of time preference, deals with the impatience of consumers and reflects their preference of immediate over postponed consumption. The second element has to do with decreasing marginal utility of income. A decreasing marginal utility of income implies that a dollar of additional income will create less additional utility, the higher the initial income level. Consequently, if income is rising over time, future impacts should be valued less.

Under the usual assumption of a CRRA (constant relative rate of risk aversion) utility function the social rate of time preference, δ , can be written as (see Lind, 1982)

tt y= ωρδ + (78)

ρ is the pure rate of time preference, reflecting impatience, and ω the income elasticity

of marginal utility or rate of risk aversion. Note that, because the rate of growth of per capita income may vary over time, the rate of discount changes over time as well. Growth may also be affected by future abatement policies or by greenhouse impacts. The discount rate is therefore a function of the corrected growth rate ty , rather than the BAU rate ty .

Although we do not know much about the rate of risk aversion, this aspect of discounting is relatively uncontroversial. ω is typically set at 1, which corresponds to a logarithmic utility function, and we follow this example here by setting 1=ω for the central case. The height of the pure rate of time preference, on the other hand, is highly controversial. One school of thought would argue that a positive rate is not acceptable on ethical grounds, as it implicitly attaches less importance to future generations (Cline, 1992a). Other authors point out that only a positive rate is consistent with observed savings and interest rate data, and Nordhaus (1992, 1993a,b) and Peck and Tiesberg (1992, 1993a,b), for example, both use a rate of 3%=ρ . For a further discussion, see e.g. Broome (1992), Cline (1992a) or Markandya and Pearce (1991). In the present model, ρ is a random variable with upper and lower bounds of 0 and 3% , respectively, and a best guess of 0.5%.

The second element which we need to define is the shadow value of capital, ν . The shadow price of capital basically represents the present value of the future consumption stream associated with a $1 investment, discounted at the social rate of time preference (Lind, 1982). In the present model ν is calculated in the way suggested by Cline (1992),

219

t

Nt

Nt r

r

δ

δν

−

−

+−⋅

+−

)(11

)(11= (79)

where r is the rate of return on capital and N the average lifetime of an investment project. See Cline (1992a) for details. Wether or not an expenditure has to be transformed primarily depends on its character, as outlined above. Climate impacts affecting investment have to be transformed, while reductions in consumption need not, as they are already expressed in the correct units. Following Cline (1992a) we base our estimate on the currently observed average savings rate and assume that 20% of market based damage is investment related, i.e. has to be multiplied by a factor ν . Eighty percent of income-related damage and all non-market damage is assumed to be consumption based. The 2xCO 2 damage variable tk can then be rewritten as

Ytt

Yt

Ptt kkkk ν0.2)0.8(= ++ (80)

3. The Social Costs of Greenhouse Gas Emissions 3.1. The Social Costs of CO 2 We have used the model of the previous section for Monte Carlo simulations of the social

costs of CO Ð2 emissions over four decades, from 1991 to 2030. The results are shown in Table 2. As expected, damage per tonne of emission is rising over time, from about 20 $/tC between 1991 and 2000 to about 28 $/tC in the decade 2021-2030. The rise is mainly due to income and population growth, i.e. the fact that tk is rising over time. The impact of higher future

concentration levels on the other hand is ambiguous. In some constellations with a low parameter γ the logarithmic relationship between forcing and concentration may dominate over the concavity of the damage relationship, and a higher concentration may actually lead to a decrease in marginal damage. If it was not for economic and population growth, the shadow value would fall over time in these cases. The figures for future periods are again current value estimates and denote the social costs valued at the time of emission.

Of course, expected value figures alone do not tell a complete story. The optimal policy response is likely to differ depending on the confidence in the results, the distribution of possible outcomes and the probability of high impact events. What is lacking is thus some information about the probability distribution of greenhouse damage. A probability distribution can be obtained directly from our stochastic model, and the relevant statistics are also shown in Table 2. Distributions for emissions in the two decades 1991-2000 and 2021-3030 are depicted in Figure 2. The Figure shows rather wide distributions. Standard errors are around 14 to 19, reflecting the generally low level of confidence into these figures (see Table 2). Not surprisingly the standard error is increasing over time as the estimates for more distant periods are more widely spread than that for the decade 1991-2000. The shape is clearly asymmetric and skewed to the right, with coefficients of skewedness in the order of 2.57. Loosely, this means that the probability of an extremely disastrous outcome is higher than that of an extremely modest result.

7For comparison, the skewedness of a symmetric distribution is zero.

220

221

Our estimates are somewhat higher than those of existing studies like Nordhaus (1992,

1993a,b) and Peck and Teisberg (1992, 1993a,b). Partly this is due to different assumptions on the value of some key parameters. The pure rate of time preference, for example, is set at 3% in DICE and CETA, a value which constitutes the upper bound for this parameter in our study (see above). On the other hand we used more moderate assumptions about the slope of the damage function. Conceptually more important is a second source of discrepancy, which arises from the fact that our figures represent expected values, while the other estimates are best guesses. As shown in Figure 2 global warming damage is not distributed symmetrically but skewed to the right. Under these circumstances the mean will be greater than the mode, and expected value figures are therefore bound to be higher than a best guess estimate which ignores this asymmetry. The higher value of our figures are thus also a consequence of the incorporation of high-impact events. In our model the difference between the expected value and a non-random best guess is about 25%. Encompassing extreme events thus appears to be crucial, and expected value estimates should be favoured over best guess assessments.

222

3.2. Other Greenhouse Gases CO 2 is not the only greenhouse gas, albeit by far the most important one, accounting for

more than half of the total effect. Other gases contributing to the greenhouse effect include CH 4 ,

N 2 O and CFCs. The usual way of dealing with greenhouse gases other than CO 2 is to transform

them into CO 2 -equivalents by using Global Warming Potentials (GWPs). According to this

index suggested by the IPCC one tonne of CH 4 and N 2 O is equivalent to 11 tonnes of CO, and 270 tonnes of CO,, respectively (IPCC, 1992).

However, it has repeatedly been pointed out in the literature that GWPs are not an adequate index if we are concerned with the damage caused by each gas (Reilly, 1992; Schmalensee, 1993; Hoel and Isaksen, 1993). GWPs are a measure of the relative radiative forcing capability of gases. Relative forcing, however, corresponds to the relative damage potential only if the latter were a linear function of the former. As we showed earlier this is clearly not the case. The incremental damage caused by a marginal increase in radiative forcing will therefore depend on factors like previous levels of forcing and the degree of warming already encountered. The relative damage contribution of a gas will thus vary over time and in general be different from its relative forcing capacity.

To exemplify this point, we have computed the social costs of other greenhouse gases directly by calculating the present value marginal damage caused by a marginal increase in other greenhouse gas emissions. Table 2 shows the results for methane and nitrous oxide. Again the figures are rising over time, in the case of methane from 108 $/tCH, in 1991-2000 to 176 $/tCH, in 2021-2030. The damage from nitrous oxide rises from 2,895 $/tN in 1991-2000 to 4,489 $/tN. In terms of CO 2 equivalents or “Greenhouse Damage Potentials” this corresponds to values of

20-23 for CH 4 and 333-377 for N 2 O, depending on the time of emission8. They are roughly in line with the estimates of Hoel and Isaksen (1993), but, for the reasons given above, are considerably different from GWPs. The fact that they are higher than GWPs mainly reflects the impact of discounting, which lessens the long run impacts of long-lived gases like CO 2 . Note that the figures only refer to the direct impacts of a gas and neglect indirect effects such as the impact of methane on water vapour and tropospheric ozone. They are restricted to the social costs from global warming, and do not include costs from other environmental problems.

3.3. Sensitivity Analysis I: Discounting One of the key parameters in the above estimates is the discount rate, not only because of

the ongoing controversy about its appropriate level, but also because of the high sensitivity of the results with respect to it. Our preferred way of discounting has been introduced in the previous section. It involves the transformation of all damage into consumption equivalents by multiplying with the shadow value of capital, when appropriate. The relevant discount rate is then the consumption rate, δ .

8As with Global Warming Potentials the figures compare the impact of one tonne of each gas, i.e. they compare a

tonne of CH 4 or N 2 O with of a tonne of CO 2 . The figures in Table 2 on the other hand use the more common

measures of tCH 4 for methane, tN for N 2 O, and tC for CO 2 .

223

Recall the distinction between the pure rate of time preference (parameter ρ ), the social rate of time preference, δ , and the rate of return on capital, r . As outlined earlier, the main controversy is about the value of ρ , and in particular whether ρ should be greater than or equal to zero (see e.g. Cline, 1992a; Broome, 1992; Markandya and Pearce, 1991). Our sensitivity analysis thus concentrates on this parameter.

As indicated earlier, our results were obtained assuming a probability distribution for the pure rate of time preference which is triangular with upper and lower bounds of 0% and 3% , respectively, and a best guess value of 0.5%. To illustrate the impacts of different discount rates we have re-run the model for p held fixed at different values9. The results are summarised in Table 3.

If ρ is set at 3% the expected shadow value of 1990s CO 2 emissions falls from 20 $/tC to a mere 5.5 $/tC. By 2030 the figure has risen only slightly to 8.3 $/tC. Interestingly, this is almost exactly the Nordhaus (1992, 1993a,b) result. Assuming a pure rate of time preference of zero on the other hand yields an expected value of 48.8 $/tC in 1990, and 62.9 $/tC in 2030, about eight times more than under the high rate. Using a higher rate also leads to a considerable reduction in the standard deviation, and thus the confidence interval. This is because more uncertain future impacts are weighted less under a higher discount rate. The lower skewedness, on the other hand, is mainly due to the fact that parameter ρ , which had a skewed distribution in the random case, is now fixed.

The high sensitivity of the results with respect to discounting should come as no surprise. It is a direct consequence of the long-term character of global warming and the fact that damages will only occur several decades into the future. The results clearly underline the importance of the discounting question and the crucial role ethical issues ought to play in the future debate on global warming.

9The rate of risk aversion, ω was also kept fixed in this exercise, at 1=ω .

224

3.4. Sensitivity Analysis II: Greenhouse Angst Although the parameter values underlying the above results broadly reflect the current

understanding of global warming, there is still an element of subjectivity inherent in them. In particular, by assuming a triangular distribution for random parameters they neglect the possibility of a climate catastrophe. It has often been noted that, given the complexity of the climatic system and the unprecedented stress imposed on it, surprises cannot be fully excluded, particularly in the long run (beyond 2xCO 2 ). Catastrophic scenarios implied in the literature include the melting of the antarctic ice-sheet, a redirection of the gulf stream and the release of methane from previously frozen materials through the melting of permafrost soils. Such instances of greenhouse angst are not necessarily irrational, and the probability of a catastrophic outcome is clearly greater than zero.

The easiest way to incorporate such fears is by using probability distributions with a domain greater than zero, i.e. to assume that parameter values are bounded below but unbounded upwards. Even extremely high parameter values then still occur with a positive probability. A distribution with this property is the lognormal, and as a sensitivity test we have run the model assuming a lognormal distribution for three key variables: Climate sensitivity, 2xC0 2 damage and the slope of the damage function, thus allowing for catastrophic outcomes with respect to

225

climate, with respect to impacts and with respect to the existence of thresholds10. The distributions were calibrated such that the lower bound remains unchanged and the most likely value equals the scientific best guess, as before, while the probability of extremely high outcomes was gradually increased. The results of this exercise are summarised in Figure 3. The Figure shows the mean and 90% confidence interval of the social costs of 1990s CO 2 emissions under the different scenarios considered. With respect to the mean, the difference between the lowest scenario A, which roughly corresponds to the triangular case used before, and the most extreme scenario considered is about 60% . If, for example, we allow a 1% chance (in each case) that 2xC0 2 rises temperature by more than 7 C, that a 2.5 C rise causes damage of more than 4.25% of GNP, and that the damage function rises more steeply than to the power 3.5, the social costs of CO 2 emissions will rise to about 33 $/tC. As expected the 95th percentile rises stronger than the mean, by about 80%, thus further increasing the skewedness of the distribution. On a whole, the analysis therefore clearly underlines the importance of low probability/high impact events.

4. Policy Implications and Conclusions The paper estimates the monetary costs of greenhouse gas emissions. As a rough

benchmark figure we suggest a value of 20 $/tC for emissions between 1991 and 2000. In subsequent decades the value rises to 23 $/tC, 25 $/tC and finally 28 $/tC for emissions in the third decade of the next century. Like all greenhouse damage estimates these results are highly uncertain and the confidence intervals attached to them are correspondingly wide. The stochastic character of our model allowed the explicit calculation of a damage probability distribution. It was shown that the distribution is skewed to the right, even for the runs neglecting the possibility of a climate catastrophe. That is, even when abstracting from actual extremes, an extremely disastrous outcome is still more likely to occur than a correspondingly modest result. Incorporating the possibility of a future climate catastrophe considerably increases both the mean and the skewedness of the distribution. In the most extreme case considered expected damage rose to about 33 $/tC. It was also confirmed that the results crucially depend on the choice of the discount rate, and ethical considerations will therefore have to stage prominently in the future debate.

10

An extremely steep damage function can be seen as an approximation of a threshold at T= Λ C.

226

The main application for the estimates is probably project appraisal. For small projects the

interpretation of the figures is straightforward. For a reforestation project sequestering 1 mtC per year over 30 years, for example, we can expect benefits of 20 m$/yr in the first decade, 23 m$/yr in the second and 25 m$/yr in the third. Total (undiscounted) benefits are therefore

227

200+230+250=680 m$. Investment decisions can then be made in the usual way by comparing the relative net benefits of rival projects. The analysis is more complicated with respect to large scale projects big enough to affect the future emissions trajectory. Because the shadow value of carbon depends on future emissions the social costs of CO 2 emissions will change with the implementation of the project and would have to be recalculated for the new emissions trajectory. In this way Cline (1992a) has found favourable benefit cost ratios for a suggested freeze of carbon emissions at 4 GtC/year. For projects affecting the trajectory only slightly the above estimates may suffice as a rough assessment, though. The procedure is then the same as above.

The appraisal of individual abatement projects has to be distinguished from the task of designing an optimal policy response to global warming. Our model does not deal with this latter question, and the figures provided give therefore only little, if any, indication of the socially optimal carbon tax. Calculating a socially optimal emissions trajectory would require the use of an optimal control model like CETA or DICE, and both models provide a first assessment as to what the optimal emissions trajectory might be (see Peck and Teisberg, 1992, 1993a,b; Nordhaus, 1992, 1993a,b; Cline, 1992b). However, not least because they are based on non-random parameters neither model offers a fully satisfactory approach to the uncertainty issue, particularly with respect to low risk/high impact outcomes. This is underlined by the fact that optimal trajectories differ considerably between scenarios. What is needed is a model which directly incorporates uncertainty, rather than working with scenarios. To our knowledge no such model exists at present. Further research efforts should thus be directed into two directions. Firstly into projects aiming at reducing existing greenhouse uncertainties, and secondly into projects evaluating the optimal policy in the light of them.

References

[1] Anderson, D., and Williams, R. (1993). Cost Effectiveness and the Investments of the Global Environmental Facility. A Background Paper for the Scientific and Technical Advisory Panel of the GEF, Washington DC: The World Bank.

[2] Ayres, R., and Walter, J. (1991). “The Greenhouse Effect : Damages, Costs and Abatement”, Environmental and Resource Economics 1 : 237-270.

[3] Broome, J. (1992). Counting the Costs of Global Warming, Cambridge : White Horse Press.

[4] Cline, W.R. (1992a). The Economics of Global Warming,Washington, DC : Institute for International Economics.

[5] Cline, W.R. (1992b). Optimal Carbon Emissions over Time : Experiments with the Nordhaus DICE Model, Mimeo, Institute for International Economics, Washington, DC.

[6] Cline, W.R. (1993). The Costs and Benefits of Greenhouse Abatement : A Guide to Policy Analysis, Paper presented at the OECD/IEA Conference on “The Economics of Global Warming”, June 1993, Paris.

[7] Daily, G.C., Ehrlich, P.R., Mooney, H.A., and Ehrlich, A.H. (1991). “Greenhouse Economics : Learn before you Leap”, Ecological Economics 4 : 1-10.

[8] Fankhauser, S. (1992). Global Warming Damage Costs : Some Monetary Estimates, Centre for Social and Economic Research on the Global Environment, University College London and Universily of East Anglia, CSERGE GEC Working Paper 92-29.

228

[9] Fankhauser, S. (1993). “The Economic Costs of Global Warming : Some Monetary Estimates”, in Y. Kaya, N. Nakicenovic, W.D. Nordhaus and F. Toth (eds.), Costs, Impacts and Benefits of CO2 Mitigation, IIASA Collaborative Paper Series, CP93-2, Laxenburg, Austria.

[10] Grubb, M. (1993). “The Costs of Climate Change : Critical Elements”, in : Y. Kaya, N. Nakicenovic, W.D. Nordhaus and F. Toth (eds.), Costs, Impacts and Benefits of CO2 Mitigation, IIASA Collaborative Paper Series, CP93-2, Laxenburg, Austria.

[11] Hoel, M., and Isaksen, I. (1993). Eflcient Abatement of Different Greenhouse Gases, Memorandum No. 5, Department of Economics, University of Oslo.

[12] Hohmeyer, 0. (1988). Social Costs of Energy Consumption : External Effects of Electricity Generation in the Federal Republic of Germany, Berlin : Springer.

[13] Intergovernmental Panel on Climate Change (1990a). Climate Change : The IPCC Scientific Assessment, Cambridge : Cambridge University Press.

[14] Intergovernmental Panel on Climate Change (1990b). Strategies for Adaptation to Sea Level Rise. Report of the Coastal Zone Management Subgroup, WMO and UNEP, Geneva.

[15] Intergovernmental Panel on Climate Change (1990c). The IPCC Impacts Assessment, Canberra : Australian Government Publishing Service.

[16] Intergovernmental Panel on Climate Change (1992). Climate Change 1992. Supplementary Report to the IPCC Scientific Assessment, Cambridge : Cambridge University Press.

[17] Kane, S., Reilly, J., and Tohey, J. (1992). “An Empirical Study of the Economic Effects of Climate Change on World Agriculture”, Climatic Change 21 : 17-35.

[18] Lashof, D.A., and Ahuja D.R. (1990). “Relative Contributions of Greenhouse Gas Emissions to Global Warming”, Nature 344 (5) (April) : 529-53 1.

[19] Lind, R.C., ed. (1982). Discounting for Time and Risk in Energy Policy, Washington, DC : Resources for the Future.

[20] Lockwood, B. (1992). The Social Costs of Electricity Generation, Centre for Social and Economic Research on the Global Environment, University of East Anglia and University College London, CSERGE-GEC Working Paper 92-09.

[21] Maier-Reimer, E., and Hasselmann, K. (1987). “Transport and Storage of CO2 in the Ocean. An Inorganic Ocean-Circulation Carbon Cycle model”, Climate Dynamics, 11(2) : 63-90.

[22] Markandya, A., and Pearce, D.W. (1991). “Development, the Environment and the Social Rate of Discount”, The World Bank Observer 6(2) : 137-152.

[23] Nordhaus, W.D. (1991a). “To Slow or not to Slow : The Economics of the Greenhouse Effect”, Economic Journal 101(407) : 920-937.

[24] Nordhaus, W.D. (1991b). “A Sketch of the Economics of the Greenhouse Effect”, American Economic Review, Papers and Proceedings 81(2) : 146- 150.

[25] Nordhaus, W.D. (1992). The DICE Model : Background and Structure of a Dynamic Integrated Climate Economy Model of the Economics of Global Warming, Cowles Foundation Discussion Paper, No. 1009, New Haven, Connecticut.

229

[26] Nordhaus, W.D. (1993a). “Optimal Greenhouse Gas Reductions and Tax Policy in the ‘DICE’ Model”, American Economic Review, Papers and Proceedings 83(2) : 313-317.

[27] Nordhaus, W.D. (1993b). “Rolling the ‘DICE’ : An Optimal Transition Path for Controlling Greenhouse Gases”, Resources and Energy Economics 15 : 27-50.

[28] Nordhaus, W.D. (1993c). Survey on Uncertainties Associated with Potential Climate Change, mimeo, Yale University, New Haven, Conn.

[29] PACE (1990). Environmental Costs of Electricity, Pace University Centre for Environmental and Legal Studies, New York : Oceana Publishers. 21

[30] Parry, M. (1990). Climate Change and World Agriculture, London : Earthscan.

[31] Parry, M.L., Carter, T.R., and Konijn, N.T. (1988). The Impacts of Climate Variations on Agriculture, 2 Volumes, Dordrecht : Kluwer.

[32] Pearce, D.W. (1980). “The Social Incidence of Environmental Costs and Benefits”, in T. O’Riodan and R.K. Turner (eds.), Progress in Resource Management and Environmental Planning, Vol. 2, Chichester : Wiley.

[33] Pearce, D.W., Bann, C., and Georgiou, S. (1992). The Social Costs of Fuel Cycles, London : HMSO.

[34] Pearce, D.W., Swanson, T.S., McGuire, A., and Richardson, J. (1991). Economics, Environment and Health. A Report for the European Regional Office on the World Health Organisation, CSERGE, University College London and University of East Anglia.

[35] Peck, S.C., and Teisherg, T.J. (1992). “CETA : A Model for Carbon Emissions Trajectory Assessment”, The Energy Journal 13(1) : 55-77.

[36] Peck, S.C., and Teisberg, T.J. (1993a). “CO2 Emissions Control : Comparing Policy Instruments”, Energy Policy 21(3) : 222-230.

[37] Peck, S.C., and Teisberg, T.J. (1993b). “Global Warming Uncertainties and the Value of Information : An Analysis Using CETA”, Resource and Energy Economics 15 : 71-97.

[38] Peters, R.I., and Lovejoy, T.E. (1992). Global Warming and Biological Diversity, New Haven : Yale University Press.

[39] Proops, J.L.R., Faber, M., and Wagenhals, G. (1993). Reducing CO2 Emissions. A Comparative Input-Output Study for Germany and the UK, Berlin : Springer.

[40] Reilly, J.M. (1992). “Climate-Change Damage and the Trace-Gas-Index Issue”, in J.M. Reilly and M. Anderson (ed.), Economic Issues in Global Climate Change, Boulder : Westview.

[41] Rijsberman, F. (1991). “Potential Costs of Adapting to Sea Level Rise in OECD Countries”, in OECD, Responding to Climate Change : Selected Economic Issues, Paris.

[42] Rosenberg, N.J., and Crosson, P.R. (1991). “The MINK Project : A New Methodology for Identifying Regional Influences of, and Responses to, Increasing Atmospheric CO2 and Climate Change”, Environmental Conservation 18(4) : 313-322.

[43] Schneider, S.H., and Thompson, S.L. (1981). “Atmospheric CO2 and Climate : Importance of the Transient Response”, Journal of Geophysical Research, 86 (20) (April) : 3135-3147.

230

[44] Schmalensee, R. (1993). “Comparing Greenhouse Gases for Policy Purposes”, The Energy Journal 14(1) : 245-255.

[45] Smith, J.B., and Tirpak, D.A. ed. (1989). The Potential Effects of Global Climate Change on the United States, Washington, DC : US EPA. 22

[46] Titus, J. (1992). “The Cost of Climate Change to the United States”, in S.K. Majumdar, L.S. Kalkstein, B. Yarnal, E.W. Miller, and L.M. Rosenfeld (eds.), Global Climate Change : Implications Challenges and Mitigation Measures, Pennsylvania : Pennsylvania Academy of Science.

[47] Titus, J.G., et al. (1991). “Greenhouse Effect and Sea Level Rise : The Cost of Holding Back the Sea”, Coastal Zone Management 19 : 172-204.

[48] Waggoner, P.E. (1990). Climate Change and US Water Resources, New York : John Wiley & Sons.

[49] World Health Organisation (1990). Potential Health Effects of Climatic Change, WHO, Geneva.

[50] Wigley, T.M.L., and Raper, S.C.B. (1992). “Implications of Climate and Sea Level Rise of Revised IPCC Emissions Scenarios”, Nature 357(28 May) : 293-300.

231

Assessing Sustainability, a Comprehensive

Wealth Accounting Measure – application to Mozambique

Timothée Ollivier, Pierre-Noël Giraud

Abstract: We estimate the wealth of Mozambique in 2000 and 2005 in order to assess the sustainability of its development path. Our methodology builds on Arrow et al. (2007). We show that Mozambican growth is driven mainly by human and physical capital accumulation, while the pressure on natural capital remains low. Moreover, changes in knowledge and institutions significantly enhance the outcome of the different capital assets, while population growth has a strong downward effect on wealth per capita. We conclude that Mozambique, unlike many other sub-Saharan countries, has followed a sustainable growth path in recent times. Keywords: natural capital, sustainable development, Mozambique, comprehensive wealth accounting

232

There is a growing literature on how to measure development, and how to assess the

sustainability of this development. The limitations of gross domestic product (GDP) as a welfare

indicator have been pointed out long years ago by welfare economists. There is now a consensus

in the political sphere on the need to develop other indicators to measure the evolution of present

welfare and the sustainability of current development paths. The most recent example is the ‘Sen-

Stilgitz’ Commission set up by the French President Nicolas Sarkozy in 2009. We focus in this

paper on the sustainability issue. If one tries to assess whether the development path of a country

is sustainable or not, one has to adopt a definition of sustainability. The definition we consider is

derived from Dasgupta and Mäler (2000): “each generation should bequeath to its successor at

least as large a productive base as it inherited from its predecessor”. They define the productive

base (or wealth) as the set of the different capital stocks of the economy: not only produced

capital, but also human capital (education level, knowledge, health, etc.), social capital

(institutions, level of trust, etc.) and natural capital (mineral resources, soil resources, forests, fish

resources, etc.).1 A development path is then said to be sustainable as long as the society’s

productive base (per capita) does not shrink.

David Pearce, among others, laid the theoretical foundations of wealth accounting exercises

(Pearce and Atkinson, 1993). Other contributions were Asheim (1994), Hamilton and Atkinson

(1996), Hamilton and Clemens (1999) and various others subsequently. Kirk Hamilton and his

team at the World Bank also implemented an impressive study on the issue, both theoretical and

empirical. Their contributions are summed up in the book Where is the Wealth of Nations?,

published in 2006. They provide natural capital, total wealth and adjusted net saving estimates for

210 countries. More recently, Arrow et al. (2004) and Arrow et al. (2007) have made significant

improvements, broadening the scope in terms of the assets considered.

This paper is in line with those contributions. It is a detailed empirical application of the

above-mentioned framework, with several methodological improvements, in the case of

Mozambique between 2000 and 2005. In contrast to most of the green accounting literature, our

focus is not only on physical and natural capital. . Human capital is also a central asset in our

analysis. Building on Arrow et al. (2007), we improve the treatment of the health dimension of

1 Institutions are not always considered as capital. Dasgupta (2009), for example, considers them as a social infrastructure guiding the allocation of resources.

233

human capital. We also investigate how institutions, knowledge capital and other intangible

assets affect the sustainability of Mozambique’s growth path.

Before entering into detail, it may be useful to provide a brief overview of the situation in

Mozambique. Since peace came in 1992, Mozambique has been one of the world’s fastest

growing economies, with an average growth rate of 8% over the past decade. It is often presented

as an African “success story”. Nevertheless, poverty remains widespread and the country is still

heavily dependent on donor aid. GDP per capita was $397 in 2007, among the lowest in the

world. Economic growth is driven mainly by foreign-financed “mega-projects” and large aid

inflows. It is also very dependent on its natural capital, for at least two reasons. On the one hand,

most current mega-projects involve exhaustible natural resources (mainly gas, coal and heavy

sands). On the other hand, the population strongly relies on its renewable natural resources, since

75% of the population works in the agricultural sector, which contributed 26% of GDP in 2005.

Social indicators have been on an upward trend in Mozambique with, for example, a significant

increase in the number of children in lower primary grade. As regards health, most indicators –

infant mortality rate, maternal mortality rate, malnutrition – have improved significantly.

However, mortality rates are still high, and AIDS is a critical problem, having a significant

adverse impact on life expectancy.

In order to assess the sustainability of Mozambique’s development path, we collected

extended datasets and numerous studies from international organizations (World Bank, Food and

Agricultural Organization, French Agency for Development, etc.), national ministries

(agriculture, fisheries, environment and forestry, national institute of statistics, health, finance,

etc.), non-governmental organizations (World Wide Fund for Nature, Justice Ambiental) and

Eduardo Mondlane University. We have also had in-depth discussions with experts on the

reliability of the data collected. This work thus relies on a comprehensive compilation of almost

all existing studies and databases on Mozambican natural, human and physical capital.

The paper is organized as follows. Section 2 recalls the theory of wealth accounting and the

importance it has for sustainability issues. In Section 3 we present the details of the methodology

used to estimate the different assets, and how we introduce technological progress and population

growth into the accounting. The results are presented and discussed in Section 4.

234

1. Theoretical framework

In this section we present the overall theoretical framework and the assumptions used in this

study. We start with a precise definition of the sustainability criterion used (2.1). Then we present

how intertemporal social welfare variations are assessed, and the link with capital assets

variations (2.2).

1.1. Which sustainability criterion?

Let us define intertemporal social welfare Vt at t as:

∫∞

−−=t

tst dsescuV )())(( δ

where u is a utility function, t is time, c is a vector including marketed goods consumption flow,

but also non-marketed goods or services consumption such as ecosystem services, and δ is the

discount rate. Economic growth will be considered sustainable at time t as long as dVt/dt≥0. In

other words, to be sustainable, the productive base transmitted to the next generation should be

able to generate at least the same intertemporal welfare. To assess the sustainability of a

country’s development, one has thus to estimate dVt/dt. This is what we do for Mozambique

between 2000 and 2005.

1.2. How is intertemporal social welfare variation calculated?

To assess Vt and its variation over time, we need to know the state of the productive base of the

economy at time t, and to assume a resource allocation mechanism to forecast the evolution of

these stocks. We describe the economy’s productive base by distinguishing three different capital

assets: produced capital K (buildings, machines, roads, etc.), human capital H (education, health,

etc.) and natural capital N (exhaustible and renewable natural resources, ecological services). The

consumption path, and thus intertemporal welfare, is determined by the evolution of the

economy’s productive base. At any given time, the output generated by this productive base is

allocated between consumption and investment in the different capital stocks. The rules

governing the allocation of the different resources are what we defined previously as the resource

allocation mechanism. It can be governed by optimizing behaviours or by exogenous rules that

235

make it non-autonomous. If we assume that the resource allocation mechanism is non-

autonomous, it means that Vt is an explicit function of time. Thus we have:

]),(),(),([ ttHtNtKVVt =

The allocation rules can be non-autonomous for several reasons. Dasgupta (2009) gives five

examples: an exogenous technological or institutional change, global public goods, capital gains,

population change, and uncertainty. If we differentiate (1) with respect to time, we obtain:

dt

dH

H

V

dt

dN

N

V

dt

dK

K

V

t

V

dt

dVt

∂

∂+

∂

∂+

∂

∂+

∂

∂=

iK

V

∂

∂ is the marginal increase in intertemporal welfare from one unit increase of the capital stock,

and can thus be interpreted as the shadow price of the capital stock. Intertemporal welfare Vt

variation results from the evolution of the three capital stocks defined previously – we will call

them ‘comprehensive wealth’, a term coined by (Arrow et al., 2007) – and exogenous factors

described by t

V

∂

∂. For the latter, in this study, we consider technological change, population

growth and climate as public goods. We are not able to consider capital gains because of data

limitations.

The first step is to assess the variation of the different capital stocks: human, physical and

natural. This is a way to measure the net investment in these different assets2. The second step is

to include the impact of exogenous factors. Here, we rely on Arrow et al. (2004) and Arrow et al.

(2007) who offer a framework for introducing technological and institutional change through the

growth of total factor productivity (TFP). We add the impact of climate change on the

Mozambican economy. Finally, we make an adjustment to account for demographic change in

order to obtain an estimate of the change of the productive base relative to population. At this

stage, unlike most green accounting literature, we did not make any assumptions regarding the

optimality of the economy.

2. Methodology

2 Some authors refer to it as genuine (adjusted net) saving, genuine investment or inclusive investment.

(1)

236

In this section we present the methodology used to evaluate the physical (3.1), human (3.2), and

natural capital (3.3). The last section deals with the calculation of the exogenous factors (3.4).

2.1. Physical capital (and urban land)

We use the perpetual inventory method, which derives capital stocks from the accumulation of

investment series. The aggregate capital stock value in period t is given by:

∑=

− −25

0

)1(i

iiitI α

where I is the value of past investment at constant prices (gross capital formation from (World

Bank, 2005)) and αi the depreciation rate. We derive depreciation rates over time from Jones

(2006). As in World Bank (2006), we assume that urban land value represents 24% of produced

capital (Kunte, 1998). Moreover, we are interested in produced capital owned by Mozambicans,

and not the stock owned by foreign investors outside the country. In the same way, Mozambican

residents own some physical capital outside the country. As in Arrow et al. (2007), our notion of

sustainability focuses on the changes in the productive base owned by a given country’s

residents. We use Lane (2006), which constructs net holdings of international assets. We can then

calculate the physical capital adjusted for international holdings in 2000 and 2005.

2.2. Human capital

The OECD (Organization for Economic Cooperation and Development) (1998) defines human

capital as “the knowledge, skills, competences, and other attributes embodied in individuals that

are relevant to economic activity”. In this study, we focus mainly on the educational and health

dimension of human capital. During the colonial era, education of the indigenous population in

Mozambique was neglected, with resulting low literacy rates. Even after the abolition of the

indigenato in 1961, education was limited to primary schooling, so as not to produce educated

opponents of the colonial power. After independence, education became a priority for FRELIMO

(Frente de Libertaçao de Moçambique) (Jones 2006). Basic education indicators have only

recently begun improving. For example, the number of children in lower primary grades rose

from 1.7 million in 1997 to 2.8 million in 2003. The number of schools has been increasing, and

237

the net enrolment rates for lower primary grades reached 69% in 2003 compared to 44% in 1997.

Nevertheless the quality of education remains low. Completion of primary schooling is still low,

and the number and qualifications of teachers have not increased proportionally.

The method used to assess human capital is similar to Arrow et al. (2007), which is itself

built on the seminal work of Mincer (1974). The idea is that a human being, like other kinds of

capital asset, generates a stream of income over time. A person’s human capital stock depends on

the average educational attainment and the return to education. We assume here a perfect labour

market, so the marginal productivity of human capital equals wages. The value of human capital

is estimated through the formula pH*H in which:

- the stock of human capital H equals P.eδA, where P is the working population, δ the rate of

return on education and A the average educational attainment of the working population,

- the shadow price pH of one unit of human capital equals ∫+ mt

t

rt dtew. , where: w is the annual

rental value of one unit of human capital (equals total wages divided by the total stock of

human capital), r the discount rate, and m the average remaining working years until

retirement or death.

The annual rental value of human capital is assumed to be constant between 2000 and 2005. The

evolution of human capital value will then result from the evolution of the stock (thus the

evolution of the educational attainment of the population) and the evolution of the shadow price

of human capital (not the rental value, because this is assumed to be constant, but through the

average remaining working years, which is closely linked to life expectancy in Mozambique).

2.3. Natural capital

Natural capital includes exhaustible resources, renewable resources (forests, land resources) and

environmental services produced by ecosystems (water filtration, waste assimilation, etc.).

Market prices for natural assets are often missing. Thus the different natural resources are valued

as the present value of resource rents during the asset’s lifetime:

∑= +

−T

tii

iii

r

qCqp

)1(

)(

238

where pi is the price at time i, qi is production, C the production costs and r the discount rate. For

each natural resource, we apply the following assumptions: a constant rental rate over time3, a 25-

year accounting period (2005-2030), the value of the resource at the end of the discounting period

is zero, and a 4% discount rate. The methodology we use for natural capital calculation is thus

very similar to that of the World Bank (2006). We consider the following resources: cropland,

pastureland, forests (timber, non-timber forest resources NTFR), protected areas, fish and mineral

resources. The detailed calculations, data (on prices, production costs, production quantities) and

sources are given in Appendix A.

2.4. Exogenous factors

Having shown how to assess the value of natural, physical and human capital stocks, we now

describe successively three exogenous factors: technological progress, which enhances overall

productivity (3.5.1), population growth in order to obtain per capita figures (3.5.2), and damage

from carbon emissions (3.5.3).

2.4.1. Technological and institutional progress

Technological change has to be understood in a broad sense. It concerns every change which

enhances the productivity of the different production factors. It can involve new technologies as

well as better performing institutions. It may be understood as the result of the accumulation of

production factors usually considered as residual in growth accounting studies. We assume

technological change to be exogenous and costless, through the growth of total factor

productivity (TFP). The costless hypothesis seems reasonable for Mozambique, since we can

assume that most technological progress in Mozambique results from technology transfers from

foreign direct investments. Arrow et al.. (2004) demonstrate under a set of assumptions (such as a

Hicks neutral technology and an elasticity of output with respect to all forms of capital equal to

one) that the growth rate of intertemporal social welfare Vt equals the growth rate of

comprehensive wealth plus the TFP growth rate. We use TFP calculations from a recent growth

accounting exercise (Jones, 2006), in which TFP growth rate is measured through a Cobb-

Douglas production function that includes produced capital, labour force and human capital

3 Rental rate = economic rent / output *100

239

(measured through a human capital index based on the mean years of schooling). In this case,

growth is explained by the accumulation of physical capital, the labour force and ‘educational

capital’. TFP captures the accumulation of other types of capital (the residual), mainly: natural,

social (through institutions), and knowledge (technological progress). TFP is thus broad and

heterogeneous. It should be borne in mind that this specification of the production function does

not include natural capital, so that the TFP growth rate produced by Jones (2006) may provide a

biased estimate of the growth of intangible capital. As a consequence, in Appendix B we provide

a correction of the TFP growth rate.

2.4.2. Population growth rate

We assume that population growth is exogenous. In our study, we are interested in the growth of

real wealth per capita V/P (P is the population and V the total wealth). Under several hypotheses

(the growth rate is assumed to be constant, per capita consumption is independent of population

size, and transformation possibilities among goods and services exhibit constant returns to scale),

we can write:

P

P

V

VP

P

V

P

V

dt

PVd

−=−=

2

)/(

To obtain the per capita wealth growth rate, the wealth growth rate has to be adjusted downward

by subtracting population growth rate. These assumptions are somewhat unrealistic, although

widely used. It is not, however, within the scope of this paper to refine the introduction of

population growth in wealth accounting exercises, a far from easy task (see Arrow et al. (2003;

Asheim et al. (2007; Ferreira et al. (2008)).

2.4.3. Carbon damage

We use the methodology developed in Arrow et al. (2007). Their idea is to index the climate

change cost of one particular country on global emissions. Nordhaus and Boyer (2000) estimate

that global warming will cost 1.5% of World GDP, and 3.5% of the GDP for African countries

(we use the most conservative IPCC4 scenario, corresponding to a doubling of CO2 emissions).

We use this approximation for Mozambique. Thus we can conclude that the climate change cost

4 Intergovernmental Panel on Climate Change.

240

for Mozambique will represent 0.027% of the world cost. If we assume CO2 emissions in the

world from 2000 to 2005 to be equal to 6.6 billion tons (WDI, 2005) and a marginal damage cost

of $50 per ton of carbon dioxide (Tol, 2005), this then gives a global damage of $545 billion for

the period 2000-2005 The climate change cost for Mozambique is thus $41 million.

3. Results In this section, we present the results. We start with our estimation for physical (4.1), human

(4.2), and natural capital (4.3). We compile the results, adding technological change and

population growth (4.4). We discuss the sensitivity of the results (4.5).

3.1. Physical capital

Table 1 shows the results for the evolution of physical capital between 2000 and 2005.

Table1: Physical capital in 2000 and 2005

All physical capital

($ million) Mozambican owned physical capital

($ million) 2000 15,245 7,425 2005 24,584 17,082

[Source: author’s calculation]

We see a large increase in physical capital in Mozambique between 2000 and 2005. Another

significant characteristic is the relatively high share of foreign capital, linked to the large number

of megaprojects (especially in the mining sector) in the country.

3.2. Human capital

The first step is to asses the total stock of human capital. Because of data limitations, we were

obliged to focus on the population aged over 15 in constructing the working population. From

Jones (2006) we obtain a distribution of the average educational level for the working population

into four categories: skilled workers subdivided into those with primary, secondary or higher

241

education as against unskilled workers5. From the data on consumption (not wages) and

educational level in the Mozambican population (given in (Jones, 2006)), we derive a 12.5% rate

of return on education. This figure is consistent with other regional studies such as

Psacharopoulos (2004). Results are shown in Table 2.

Table 1: Human capital stock in 2000 and 2005

Average

educational level Average per capita

human capital

Active population (over 15)

Human capital stock

2000 2.2 1.20 8,790,000 10,596,429 2005 2.6 1.24 9.288,000 11,556,343

[Source: Jones (2006) and author’s calculation]

The second stage is to assess the value of one unit of human capital. One major problem is

that people in Mozambique are mostly self-employed, so that it is difficult to obtain an average

annual wage or the total wage bill for the country as a whole. We therefore take the labour share

from the growth accounting framework of Jones (2006). One of the baseline cases involves a

simple Cobb-Douglas production function with constant returns to scale. The labour share is

assumed to be 60% of total output. From the Cobb-Douglas properties and assuming that wages

reflect the marginal product of labour, we can conclude that the total wage bill amounts to 60%

of GDP. This is rather crude, but more consistent than any of the surveys on incomes that we

found. We derive average remaining working years for the age 30-35 population segment –

which corresponds to the average age of the working population both in 2000 and 2005 – from

WHO life tables and population pyramids (US Census Bureau database). Data used and results

are shown in Table 3.

Table 2: Computation steps of the shadow price of human capital

5 We assume that primary education amounts to 7 years of schooling and secondary or higher education to a minimum of 11 years of schooling.

2000 2005 Rental value of one unit of human capital ($2005) 367 345

Remaining working years 34.1 31.9

242

[Source: author’s calculation]

Results from Table 2 and Table 3 are then used to assess human capital value and compiled in

Table 7. This reveals a significant increase in human capital between 2000 and 2005. Such

evolution is driven by two opposing factors. On the one hand, there was a significant increase in

the overall educational level. Indeed, investment in the education sector was substantial during

the 1990s in Mozambique. On the other hand, the shadow price of human capital decreased,

mainly a result of a fall in life expectancy, probably due to AIDS (the prevalence rate is around

16% for adults) and tuberculosis.

3.3. Natural capital

Below we detail the results of our own estimations for 2005 (4.3.1). We then discuss the

evolution of the different capital stocks (4.3.2).

3.3.1. Value of natural capital for 2005

Table 4 presents the composition of natural capital for 2005 (for comparison, we also include

World Bank’s figures6).

Table 4: Breakdown of natural capital

Net present value Present study

per capita ($2005)

World Bank per capita ($2000)

Present study ($2005 million)

Mineral resources 940 --- 17,860 Timber 347 340 6,593 Forest land NTFR 133 392 2,527

6 Care should be taken in comparing the figures since ours are for 2005 and those of the World Bank for 2000.

Shadow price of one unit of human capital ($2005)

6,837 6,651

243

Protected area 30 9 570 Cropland 694 261 13,186 Agricultural

land Pastureland 109 57 2,070 Marine

resources Fisheries 19 --- 361

Total 2,272 1,059 43,168

[Source: author’s calculation]

Land resources are the most important part of Madagascar's natural wealth, with cropland

constituting around a third of its total natural capital value. Mineral resources (especially through

gas, coal and heavy sands) also represent an important share, accounting for more than 40% of

the natural capital. The importance of forest is understandable, given its share in GDP, but the

relatively low figure for NTFR is more surprising. Fisheries also represent only a small

proportion of natural wealth, mainly because rents are low.

3.3.2. Value of changes in natural capital between 2000 and 2005

Having calculated the value of Mozambique’s natural capital for 2005, it is possible to derive the

value of the different natural capital stocks in 2000 by tracing back the evolution of these stocks.

We focus on subsoil assets (mainly natural gas), cropland and forest resources.

Subsoil asset depletion - We rely on World Bank estimates of resource extraction for a range of

fossil fuels and minerals. Depreciation of these resources is computed as the product of price

minus the average cost of extraction multiplied by the volume extracted: (P-AC)R where P is the

resource price, AC is average cost and R is the volume extracted. For exhaustible resources

(mainly natural gas at the moment for Mozambique), we use World Bank datasheets (compiled

for the calculation of genuine savings and available on the World Bank website). Post-2005, the

exploitation of coal and heavy sands is likely to increase the depletion of exhaustible natural

capital.

Cropland degradation - To estimate the cost of soil degradation on cultivated areas, we use the

net nutrient replacement cost method. Folmer (1998) provides figures on nutrients (nitrogen,

244

phosphorus, potassium) depletion on a national scale. These are converted into fertilizers bags

(see Table 5)7.

Table 5: Nutrient balance for cultivated fields and associated per hectare cost

N (kg/ha)

P (kg/ha)

K (kg/ha)

P2O5

(kg/ha) K2O

(kg/ha) NPK

(kg/ha) Fertilizers (kg/ha)

Cost ($/ ha)

33 6.4 25 15 30 77 172 72

[Sources: (Folmer, 1998) and author’s calculation]

Some important limitations drove us to consider only the relatively small permanent crop area of

around 235,000 hectares8. Assuming a $0.42 per kg fertilizer price, we obtain a low annual

cropland capital depletion of around $17 million per year.

Forest stock depletion – We distinguish two different stocks: the roundwood stock (of

commercial value) and the woody biomass stock (mainly for fuelwood). We assess the evolution

of each stock, balancing annual wood harvest against annual regeneration. On a national scale,

there is no depletion of these two stocks. For roundwood stock, quantities harvested each year

(135,000 m3), even if we assume high rates of illegal logging, are below annual regeneration

(500,000 m3) as reported in the last National Forest Inventory (Marzoli, 2008). For the woody

biomass stock, the annual potential biomass productivity on a national scale (46,921,000 tons)

remains much higher than current fuelwood consumption (14,003,000 tons) (Wisdom report). As

a consequence, we do not consider there to be any depletion of these two stocks9.

The aggregate results for natural capital depletion (mineral and cropland resources) are shown in

Table 6.

7 The coefficients used to convert the nutrient content of the soil into forms in which they exist as fertilizers (N,P205, K20) are: kgP*2.29=kg P2O5, kg K*1.2=K2O and a bag contains 15% N, 15% P and 15% K. 8 First, nutrient depletion assessed is on a yearly basis, although there may not be any depletion of nutrient stocks on a longer time scale (because of fallows and rotations), but only for cultivated fields. Second, chemical fertilizers may not be the cheapest substitute (organic fertilizer or new lands would be certainly more appropriate). 9 For the roundwood stock, there may not be a significant depletion of the overall stock (including all commercial species), but many local observers point to a rapid degradation of the quality of the forest. There would appear to be a depletion of the most valuable roundwood species.

245

Table 6: Natural capital depletion between 2000 and 2005

Depletion value ($2005 million)

Subsoil assets -520 Wood capital 0

Soil degradation -85 Natural capital -605

[Source: author’s calculation]

3.4. Compilation of the results

First, we present the evolution of natural, physical and human capital assets (previously named

‘comprehensive wealth’)(4.4.1). Damages from global carbon emissions are also included in this

section. Second, we account for population growth and technological change to obtain the change

of per capita total wealth including the effect of technological progress (4.4.2).

3.4.1. Evolution of comprehensive wealth Table 7: Change in comprehensive wealth

[Source: author’s calculation]

2000 ($2005 million)

2005 ($2005 million)

Variation ($2005 million)

Human capital 72,448 76,857 +4,409

Natural capital 43,773 43,168 - 605 Physical capital 7,425 17,082 +9,658

Carbon damage --- --- -41

Comprehensive wealth 123,646 137,107 13,421

246

We can see from Table 7 that there has been a large increase of both human and physical capital

stocks. Human capital has increased because of the increase in the average educational level,

physical capital certainly because of large-scale projects. Depletion of gas stocks accounts for

most of natural capital depletion, which appears to be relatively low10. In sum, the stock of

natural, human and physical capital has been significantly increasing.

3.4.2. Accounting for population growth and technological change

As we did previously, we correct the TFP growth rate estimate from Jones (2006). Following the

formula obtained in Appendix B, we obtain a much larger growth rate of 4.8%. This can be

explained by the fact that there is considerable physical capital accumulation in Mozambique.

Indeed, as we introduce natural capital into the production function, it reduces the relative

contribution of physical capital in the output (as we assume constant return to scale). Thus TFP

growth has to be higher, reflecting higher positive externalities of the fast growing physical

capital. The results are shown in Table 8.

Table 8: Growth rate of comprehensive wealth per capita adjusted for residual growth

[Source: author’s calculation]

10 The exploitation of gas started only in 2004, so that its contribution to natural capital depletion is likely increase significantly in the future. Coal and heavy sands extraction has not yet begun.

Growth rate (%)

Comprehensive wealth growth rate +2.2

TFP growth rate +4.8

Population growth rate -2.4

Per capita comprehensive wealth growth rate

accounting for TFP

+4.6

247

The comprehensive wealth growth rate hardly covers the population growth rate. The increase in

total wealth per capita therefore relies mainly on the TFP growth rate. In our calculations, it

reflects both an increased productivity of physical and human capital, consistent with the large

amount of foreign direct investment in the country and an increase of the social capital.

3. 5. Sensitivity of the results

Our results rely on a set of crucial assumptions and the data used may sometimes be disputable.

Ideally we should carry out systematic sensitivity analysis for every critical parameter. They are

in fact so many of these that we prefer here to offer an overall qualitative assessment of the

robustness of our results. For physical capital, the methodology used is standard and the usual

caveats apply. Our calculations are particularly sensitive to depreciation rates. We use rough

figures from Jones (2006), but it is very possible that our calculations overestimate physical

capital stocks, because shocks (economic or climatic such as the 2000 floods which destroyed

much infrastructure) are poorly accounted for. For human capital, one major assumption is the

working population considered. Because of data constraints, we had to focus on workers aged

over 15. However, children or youths under 15 should also be included, since they are either

working or building their human capital (with higher expected future income). If it were possible

to include the children under 15 in our results, this would certainly increase human capital

accumulation, since primary and secondary school enrolments are currently increasing. For

natural capital, the data are particularly constraining and our accounting is not exhaustive. For

example, we could not consider groundwater, fisheries stock depletion or deforestation. These

are, however, rarely considered as major environmental issues in the country, so we can assume

that these would not be very significant. In sum, it seems to us that natural capital variation

would not change significantly with better data and different assumptions. Finally, the TFP

growth rate used as a proxy for technological and institutional progress is a core parameter in our

work. TFP estimates from growth accounting exercises (as well as our adjustment method to

account for natural capital) are highly sensitive to the underlying data and assumptions (on the

factor shares for example). This issue clearly should be further investigated.

4. Conclusions and perspectives

248

This paper builds on the work carried out by Arrow et al. (2007). We add some

methodological elements in regard to the health dimension of human capital, and refine the

methodology developed by the World Bank for estimating natural capital. The paper offers

interesting material for analyzing and characterizing Mozambique’s current development path

and assessing its sustainability. Despite the approximations made, we can conclude that

Mozambique, unlike many other sub-Saharan countries, is probably following a sustainable

path at the start of this century. Its growth is driven mainly by human and physical capital

accumulation, while the pressure on renewable natural capital remains relatively low.

Compared to most sub-Saharan African countries in which TFP growth is often negative, TFP

growth in Mozambique is relatively high, indicating that there is a significant accumulation of

technological and social capital. Further work is needed to go deeper into the composition of

this intangible capital or residual, which is a key parameter in our analysis. Finally, although

population growth is high, growth of per capita comprehensive wealth (accounting for

technological progress) also remains high.

This study is, of course, not exhaustive. We had to neglect the depletion of several natural

capital stocks, such as fisheries and pastureland, because of lack of data. Nor did we take into

account water and air pollution, which could be important issues. The study could be

improved in several key areas and further work is called for. However, we think it represents

a first step toward a tool for accurately and comprehensively assessing the dynamics of

Mozambique’s growth path.

Appendix A: Details, data and sources for natural capital calculations

Cropland - The main crops in Mozambique are maize, cassava, mapira, various kinds of beans, peanuts,

rice, cotton, cashew nuts, potatoes and tobacco. We consider crops covering more than 60,000 hectares.

We assess rental rates on the basis of various production cost studies and local market prices (Gergely,

2005; FAO producer prices). Total rent in 2005 for each crop is estimated through the formula: Total rent

(crop i) = mean yield (crop i)*local market price*rental rate*crop i area. To project total rents into the

future, we use current production trends (over the last five years) for each crop.

Table A1: data used for cropland estimates

249

[Sources: TIA 2005, FAOSTAT, SIMA, Gergely (2005), Arlindo (2007), Coughlin (2006), Benfica (2005)]

Pastureland – Beef, goat meat and milk constitute the main output from pastureland in Mozambique. As

we found no comprehensive data on production costs, we use the rental rate from the World Bank (2006)

of 45%. Future rent projections are forecast using current production volume trends.

Table A2: data used for pastureland estimates

[Sources: FAOSTAT, (World Bank, 2006)]

Timber resources - We distinguish industrial roundwood from fuelwood production. For legal logging,

we use national statistics from the forest ministry. We assume illegal logging is 40% of legal logging

(MacKenzy, 2006). The sustainability of wood production is introduced through the lifetime of the

resource. We evaluate the time to exhaustion with current production trends, annual regeneration, and total

wood stock (from DNTF (2008) and Marzoli (2008)). Rental rates are assumed to be 40% for industrial

roundwood (Bila, 2003) and 50% for fuelwood production.

Non-timber forest resources - We use two studies valuing NTFR: Suich (2006) in Bazaruto, Vilanculos

and Chirendzene districts and Lizon (2002) in the Gilé district. These consider direct values only: fruit,

crop 2005 area (Ha) Yield

(tons/Ha)

Producer price ($/ton) Production growth rate

Maize 1,749,534 1,004 153 0.0186

Cassava 1,038,851 7341 113 0.1603

Sorghum/mapira 364,370 637 146 0.0616

Beans (all types) 659,151 500 441 ---

Peanuts (all

types)

433,092 341 475 -0.0206

Pumpkin 103,413 1,831 164 0.0193

Rice 278,368 902 296 -0.0177

Cashew 54,616 1,193 238 0.0289

Potatoes 78,938 13,046 352 0.0043

Tobacco 85,234 1,388 1,671 0.0444

Sesame 65,027 661 129 0.0954

Output Price ($/ton) 2005 production

(tons)

Rental rate

(%)

Total rent production

growth trend (%)

Beef 4,052 38,100 Sheep meat 6,931 768

Milk 518 68,765

45

0

250

wild animals, honey, raffia and bark, etc. (first table below). As we have no information on the time spent

collecting these products (which is the main production cost), we use a 50% rental rate (based on figures

from other southern African countries). We do not add indirect values (such as watershed protection)

because it is already included in cropland (or other types of activity) downstream value (if we consider the

environmental service ‘protection against erosion’). To extrapolate from these household surveys to a

country-wide scale is a risky task. We combine the average NTFR value consumed per household with an

assessment of the importance of NTFR in the different provinces from the last national forest inventory

(second Table below).

Table A3: value of the services from forests considered

Lizon (2002) WWF (2006)

Unit: $/household/year

Bazarut

o

Chirindze

ne

Vilancul

os

Averag

e value

Food 58 27 20 0 30

Medicinal plants --- --- --- --- ---

Material and construction wood 11 46 173 91 65

Wood fuel 44 126 170 132 123

[Adapted from Lizon (2002) and WWF (2006)]

Table A4: percentage used to extrapolate at the national scale

% use of NTFR for rural

households North Centre South

Food 21% 38% 52%

Fodder 2% 1% 4%

Medicinal plants 29% 20% 32%

fuel 19% 18% 1%

Construction wood and ustensils 25% 21% 8%

[Adapted from Marzoli (2008)]

Protected areas - In World Bank (2006), protected areas are valued at the lower end of per-hectare

returns to pastureland and cropland - a quasi-opportunity cost. Instead, we propose a rough estimate of the

net present value of the network of protected areas. IUCN (2008) suggests some of the principal benefits

from the main protected areas are: ecotourism benefits (net revenues from the tourist industry amount to

$45 million) and the existence value of the parks through environmental NGO investments (it reflect the

willingness of people in rich countries to pay for the protection of biodiversity). WWF (2008) give an

indication of the operating costs of the parks, around $5.3 per hectare per year. This figure is based on

251

three national parks and thus does not reflect the heterogeneity of the parks (national parks, reserves and

hunting reserves). To obtain the net present value of the protected areas, we assume that: their opportunity

cost is low (mostly because of the quantity of land available), the return on capital invested is 15%, the

growth rate of the rent is 5% per year (which is conservative in view of the projections for tourism by the

Ministry for Tourism).

Fish resources - Production data are from Wilson (2008), based on statistics from the Instituto de

Investigacao Pesqueira and Instituto Nacional de Desenvolvimento da Pesca de Pequeno Escala. We

upwardly adjust catches by artisanal fisheries, since official statistics do not cover the whole coastal area.

In accordance with local experts, we add 40,000 tons to recorded catches. We also use data on the value of

fish harvested to derive prices (Wilson, 2008). From Wilson (2008) and consultation with local experts,

we use a 10% rental rate for industrial fisheries and 5% for the artisanal ones.

Mineral resources - Bucuane (2007) has carried out subsoil assets valuation for Mozambique, following

the methodology developed by the World Bank (based on (Vincent, 1996), which is a refinement of

equation (5)). We use values from Bucuane’s medium scenario.

Appendix B: Proof of the adjustment of the TPF estimate to account for

natural capital omission in the production function

We use the TFP estimate derived in Jones (2006), the most recent work we could find. To fit our

framework, we use the Cobb-Douglas production function case. The production technology is

described through the following production function:

Yt=AtKta(htLt)

b with a+b=1

where A is Hicks-neutral technological change, Kt the physical capital, Lt the working

population and h a human capital quality index. Let us define g(x) as the growth rate of x. The

growth rate of the production Yt is thus: g(Yt)=g(At)+a.g(Kt)+b.[g(ht)+g(Lt)].

Adding the flow from natural capital Nt, the production function becomes:

Yt=AtKta(1-r)(htLt)

b(1-r)Ntr

where r is the share of natural resources in production. The growth rate of production becomes:

g(Yt)=gc(At)+a(1-r)g(Kt)+b(1-r)[g(ht)+g(Lt)]+rg(Nt).

Equalizing the two expressions of g(Yt), we obtain an expression for the corrected TFP growth

rate: gc=g+a.r.g(Kt)+b.r[g(ht)+g(Lt)]-r.g(Nt)

252

For the computation, we assume that: a=0.4, g(K)=0.45, g(L)=0.01, g(h)=0.018 (from

(Jones, 2006)). For the flow from natural resources, we assume that r=0.2 and g(N)=0.033. 0.2

corresponds to the share of agriculture in GDP (thus it includes agricultural land (pastureland as

well as cropland), fisheries and forests resources)11. 3.3% is the growth rate of cropland through

extensification (according to data from Aviso Previo). We use it as a proxy for the rate of change

of the flow derived from the renewable resources. Exhaustible resources are not included here

because the exploitation (of gas) really only started in 2005 and the TFP growth rate estimate was

assessed for the 1999-2005 period. The magnitude of the different terms is given in the table

below.

Table B1: correction of the TFP estimate to account for natural capital

11 The share of natural capital in comprehensive wealth in 2005 was around 30%,.Hence 20% may be an underestimate.

Growth rate (%)

g 1.6 +a.r.g(K) 3.5

+b.r[g(h)+g(L)]

0.3

-r.g(N) 0.6 gc 4.8

253

References [1] Asheim, G.B., Buchholtz, W., Hartwick, J.M., Mitra, T., Cees, W., 2007. Constant saving rates andquasi-arithmetic population growth under exhaustible resource constraints. Journal of Environmental Economics and Management 53, 2, 213-239.

[2] Arrow, K.J., Dasgupta, P., Goulder, L.H., Daily, G., Ehrlich, P., Heal, G., Levin, S., Mäler, K.G., Schneider, S., Starrett, D., Walker, B., 2005. Are We Consuming Too Much?. Journal of Economic Perspectives, 18, 147-172.

[3] Arrow, K.J., Dasgupta, P., Goulder, L.H., Mumford, K., Oleson, K., 2007. China, the U.S., and sustainability: Perspectives Based on Comprehensive Wealth. Stanford Working Paper.

[4] Benfica, R., Zandamela, J., Zandamela, J., 2005. The economics of smallholder households in tobacco and cotton growing areas of the Zambezi valley of Mozambique. Research Report 59. Ministerio de Planificaçao e Desenvolvimento.

[5] Bucuane, A., Mulder, P., 2007. Exploring natural resources in Mozambique: will it be a blessing or a curse? Discussion papers 54. Ministerio de Planificaçao e Desenvolvimento.

[6] Coughlin, P.E., 2006. Agricultural intensification in Mozambique. Infrastructure, Policy and Institutional Framework – When Do problems Signal Problems Signal Opportunities?. Report commissioned by the African Food Crisis Study (Afrint).

[7] Dasgupta, P., Mäler, K.-G., 2000. Net National Product, Wealth, and Social Well-Being. Environment and Development Economics, 5, 69-93.

[8] Dasgupta, P., 2009. The Welfare Economic Theory of Green National Accounts. Environmental and Resource Economics, 42, 3-38.

[9] DNTF (National Directorate of Land and Forest), 2008. Consolidation phase – Wood energy component WISDOM. Mozambique final report (AIFM).

[10] Ferreira, S., Hamilton, K., Vincent, J.R., 2008. Comprehensive Wealth and Future Consumption: Accounting for Population Growth. The World Bank Economic Review, Vol. 22, 233-248.

[11] Folmer, E.C.R., Geurts, P.M.H., Francisco, J.R., 1998. Assessment of soil fertility depletion in Mozambique. Agriculture, Ecosystems and Environment, 71, 159-167.

[12] Gergely, N., 2005. Economic Analysis of comparative advantage for major agricultural cash crops in Mozambique. Background paper for World Bank.

[13] Jones, S., 2006. Growth accounting for Mozambique (1980-2004). DNEAP discussion paper.

[14] Kunte, A., Hamilton, K., Dixon, J., Clemens, M., 1998. Estimating National Wealth: Methodology and Results. Environment Department Paper 57, World Bank, Washington, DC.

[15] Lane, P.R., Milesi-Ferretti, G.M., 2006. The External Wealth of Nations Mark II: Revised and Extended Estimates of Foreign Assets and Liabilities, 1970–2004. IMF Working Paper 06/69.

[16] Lizon, J.G., 2002. Rural livelihood dependence on wildlife resources in Gilé district, Mozambique and policy implications. Paper presented at an IFAD Workshop in Nairobi.

254

[17] Mackenzie, C., 2006. Chinese Takeaway! Forest governance in Zambezia.

[18] Marzoli, A., 2008. Avaliaçao integradas floresta de Moçambique. DNTF.

[19] Mincer, J., 1974. Schooling, Experience, and Earnings. New York: Columbia University Press.

[20] Nordhaus, W.D., Boyer, J., 2000. Warming the World: Economic Models of Global Warming. MIT Press, Cambridge, MA.

[21] OECD (Organization for Economic Cooperation and Development), 2001. The Well-being of Nations: the role of human and social capital. Paris: OECD

[22] Pearce, D.W., Atkinson, G., 1993. Capital theory and the Measurement of Sustainable Development. Ecological Economics, 8, 103-108.

[23] Psacharopoulos, G, Patrinos, H.A., 2004. Returns to Investment in Education: A Further Update. Education Economics, 12, 111–134.

[24] Tol, R.S.J., 2005. The Marginal Damage Costs of Carbon Dioxide Emissions: An Assessment of Uncertainties. Energy Policy, 33, 2064-2074.

[25] UICN, 2008. Assessment of economic benefits of Mozambique’s protected areas.

[26] Ogle, A., Nhantumbo, I., 2006. Improving the competitiveness of the timber and wood sector in Mozambique. Report prepared for the Confederation of Mozambican Business Associations under the Mozambique Trade and Investment project.

[27] Vincent, J., 1996. Resource Depletion and Economic Sustainability in Malaysia. Development Discussion Paper 542, Harvard Institute for International Development, Cambridge.

[28] Wilson, J., 2008. Financial flows in Mozambique’s fisheries. Report for Norad-Marema-MdP.

[29] World Bank, 2005. World Development Indicators 2005. The World Bank, Washington, D.C.

[30] World Bank, 2006. Where is the wealth of nations? Measuring Capital for the 21st Century, The World Bank, Washington, D.C.

[31] WWF (World Wildlife Fund), 2008. Pilot financial plan for conservation areas in Mozambique (2008-2017).

255

PART FOUR.

CARBON MARKETS.

Two kinds of solutions are commonly taken into consideration for combating climate

change: taxes and markets. Of these two, Europe has chosen to rely on markets. The fourth part

of this book is devoted to these markets, and more specifically to carbon markets. Christian de

Perthuis first recalls the history of the creation of such markets, and explains how they work.

Cédric Philibert assesses the advantages of a cap-and-trade mechanism as the one adopted for

carbon markets. He also suggests strengthening it with the introduction of a floor for transaction

prices.

Christian de Perthuis emphasizes that carbon markets should not be seen as an alternative

to public action in the face of climate change. The author argues that it is on the basis of these

economic instruments that the international community will be in a position to build a more

ambitious agreement for managing the many risks that climate change exposes us to. He is of the

opinion that for the next stages of international negotiations on climate changes, three parameters

should play a key role: first, the advance represented by the current carbon trading system;

second, the benefits that major emerging countries such as China, India and Brazil experience by

being able to use the international carbon market to reinforce their emissions reduction efforts;

third, the likely introduction, within the next few years, of a federal cap on greenhouse emissions

in the United States. Without a carbon market, the chances of a post-Kyoto international

agreement would be poor.

In his contribution, Cédric Philibert assesses the long-term economic and environmental

effects of introducing price caps and floors in a hypothetical climate change mitigation

architecture. He develops a climate policy costs model. This highly aggregated model of the

256

global economy makes no distinctions between countries or sectors. It includes abatement cost

curves built on International Energy Agency expertise and publications. To take full account of

uncertainty, thousands of Monte Carlo simulations are carried out in order to study various

combinations of targets, price caps and price floors. This quantitative analysis confirms that

introducing price caps could significantly reduce economic uncertainty. In the meantime, price

floors would reduce the level of emissions beyond the objective if the abatement costs turned out

to be lower than forecast.

257

Carbon Market And Climate Negotiations

Christian de Perthuis

Abstract: Within the context of the upcoming Copenhagen Conference and the outstanding issue of shaping climate change mitigation post-2012 – the final year of the commitment period covered by the Kyoto protocol – this paper puts into context the the various economic instruments available for tackling climate change, and the effective emergence of (a) carbon market(s), a central consideration in post-2012 negotiations.

The paper gives an overview of the various types of economic instruments used to tackle environmental problems: regulation, taxes and tradable permits; tracing their origin in economic theory and giving concrete examples of their application in the context of national and international efforts in environmental protection, and more particularly, in climate change mitigation. Specific attention is given to the use of economic instruments in the implementation of the Kyoto Protocol to the United Nations Framework Convention on Climate Change. The Kyoto Protocol gave impetus to the creation of the European Emissions Trading Scheme, the first large-scale carbon trading system worldwide, and an international benchmark for the price of carbon. Experiences from the implementation and operation of the European carbon market provide valuable insight for European and non-European actors and a concrete tool which the European Union can use in its continued efforts in climate change mitigation, which extend well beyond the 2012 period envisaged by the Kyoto Protocol. A second carbon pricing mechanism arising from the Kyoto Protocol is the international projects market, of which the Clean Development Mechanism (CDM) is currently the main component. The CDM provides the only link between industrialised countries of the north and the developing countries of the south in international climate negotiations.

The climate negotiations in Copenhagen will bring together a number of important decision makers to decide the fate of the climate change effort beyond 2012. Carbon markets, while instruments for inciting efficient emissions reductions, also facilitate the emergence of compromise andwill thus play a key role in the international negotiations.

Keywords: climate change, economic instruments, post-Kyoto, post-2012, carbon markets

258

The Kyoto Protocol adopted in 1997 established commitments aimed at implementing the

objectives of the United Nations Framework Convention on Climate Change (UNFCCC).1 The Protocol defines fixed objectives for the 38 most industrialized countries (listed in Annex B of the Protocol) to collectively reduce by at least 5% their overall emissions of 6 greenhouse gases in relation to 1990 levels. Non-Annex B countries do not have set objectives. These reductions must occur over the period 2008-2012. The United States is the only developed country which has not ratified the Protocol. The commitment period covered by the Kyoto protocol expires in December 2012. The rules that will then apply will be discussed during the Copenhagen Conference scheduled for December 2009. The challenge is to extend the greenhouse gas emissions commitment in order to reduce the current trajectories of world emissions. In view of the divergence between the stated positions of the various nations, it is difficult to be optimistic. Nevertheless, the Kyoto protocol enabled economic instruments to be set up, which have been developed in the field. These instruments set a greenhouse gas emissions price. They have two component principles: the European system of quota trading and the international projects market. It is on the basis of these economic instruments that the international community will be in a position to build a more ambitious climate agreement in order to manage the many risks that climate change exposes us to.

1. Free use of the atmosphere: a tragedy of the commons In his celebrated essay The Tragedy of the Commons, Garret Hardin describes the

predation mechanisms on natural resources resulting from the fact they are free [13]. He draws on the example of the shared pastures surrounding English villages up until the end of the 18th century. Under this system, each herdsman had access to the “common'' for grazing his stock. In a situation of demographic stagnation and with few animals per hectare, this social system provided villagers with a degree of security. Everyone had free access to a shared resource.

In a growth situation, this system tended to self-destruct: because access to the common was free, no herdsman took account in his economic calculations of the cost this resource levied on the community. It was in each herdsman's economic interest to graze his livestock as long as a positive marginal revenue remained, i.e. a few blades of grass remained in the pasture. The inevitable outcome is overgrazing, which reduces the fertility of the pasturage to zero and leads to the destruction of the collective resource.

To understand the economic problem presented by climate change, one simply has to replace the words “village'' and “common'' in Hardin's example with “planet'' and “atmosphere''. The growth in the number of inhabitants of the planet and their enrichment threatens a very special collective good: the stability of the climate. The atmosphere is not infinite, any more than was the common pasturage. Its capacity to regulate temperatures is therefore altered by the accumulation of our waste greenhouse gases. Yet, like the members of the village community, so long as the free and unlimited use of this reservoir predominates, we have no economic incentive 1 The United Nations Framework Convention on Climate Change (UNFCCC) is the primary international treaty on global climate change. Signed in Rio de Janeiro in 1992, the Convention’s objective is the “stabilization of greenhouse gas concentrations in the atmosphere at a level that would prevent dangerous anthropogenic interference with the climate system.” See the UNFCCC’s website: www.unfccc.org

259

to reduce emissions. But because of the inertia of the climate system, it is not our generation that will suffer the consequences of the emissions it produces. It will be future generations.

In order to escape this “tragedy'', the villagers can first of all organize themselves to limit the use of the pasture, for example by setting of a rotation system for grazing. To make the system effective, common rules must be established and necessarily adhered to. Its implementation will limit the freedom of action of each villager. It will probably include a sanctions mechanism for anyone breaking the rules. This first type of arrangement constitutes a regulations-based approach.

At the international level, a regulations-based approach was adopted to combat the destruction of the ozone layer. The 1987 Montreal Protocol gradually banned the use of CFC gases in their main industrial applications. It was considered an effective response by the international community: according to the fourth report of the Intergovernmental Panel on Climate Change (IPCC), CFC emissions were reduced from 7.5 billion tonnes 2CO eq2 to 1.5

billion tonnes in 2004. If the Kyoto protocol that regulates emissions of carbon dioxide ( 2CO ) and five other greenhouse gases over the period 2008-2012 were to have a similar impact, the trajectory of global emissions would be considerably affected. But this will not happen. Due to the withdrawal of the United States in 2001 and the very generous concessions made to Russia and Ukraine, the implementation of the protocol will have only a minor effect on the trajectory of global 2CO emissions (if any). Its contribution is of a different nature. It is through the carbon pricing mechanism that its implementation has made itself felt.

2. Taxes or permits? A second possible mechanism for protecting the communal pasturage would be to

introduce a levy that would apply to any villager wishing to use the collective asset. The rate of the levy would need to be set in such a way that its cost enables the number of livestock to be adjusted to the amount of forage that the communal pasture can produce on a sustainable basis. Introducing this levy will have redistributive effects: its payment will exclude the poorest villagers' animals from the pasturage. On the other hand, the revenues obtained can be allocated to programs of value to the community. In this case, the levy will bring a second benefit, a “second dividend'' in economists' terminology, to the community. This aspect of the levy corresponds to the logic of introducing a tax.

The use of a tax for protecting environmental resources was advocated as early as the 1920s by the English economist Pigou. His idea was to protect environmental assets by incorporating them into the production of goods, not only the standard elements of cost of production, but also an estimation of the social cost that environmental damage represents for society. This pricing of environmental externalities by means of the tax allows a price to be given for the protection of the environment. This particular route, as a way of confronting climate change, has not been taken at an international level, despite being recommended by a number of

2CFC emissions were regulated by the Montreal Protocol to stop the destruction of the ozone layer. CFCs are also greenhouse gases. The application of the Montreal Protocol thus contributes to action against greenhouse gas emissions when CFCs are not replaced by substitutes that also contribute to the greenhouse effect. It is customary to

convert non- 2CO greenhouse gas emissions into carbon equivalent tonnes, known as 2CO eq, on the basis of their

warming power over 100 years.

260

economists. However, some European countries, namely Sweden, Norway, Denmark and more recently France and Ireland have introduced such carbon taxes into their domestic legislation.

The third possible arrangement is to create a market which will yield a price for the common good one wishes to protect. This is the route that historically was taken, not only in England, but in the majority of European countries in the early days of the industrial revolution. The traditional organization of the village with its common pastures was gradually replaced by a system of private land ownership. This transformation was produced by the “enclosures'' movement, which appeared in England from the 15th century onwards. The term is a reminder that one of the first consequences of the privatization of common land was the construction of enclosures designed to protect the enclosed land from incursion and grazing by the livestock of the village. There resulted from it an increase in agricultural productivity that made possible the transfer of manpower from agriculture to industry.

This third, market route is the one that has been taken by the international community to fight against climate change. It has, of course, not taken the form of a privatization of the atmosphere, which cannot be divided up into lots or protected against greenhouse gas emissions by means of enclosures. Rather it has taken the form of emission permits markets, a route explored in the 1960s by the economists Ronald Coase and John Dales [5], and successfully put into practice in the United States since 1995 to combat acid rain produced largely as a result of

2SO emissions from power plants. From a theoretical standpoint, it is simple to show that, in a situation of perfect

competition, using a tax or using a system of permits are strictly equivalent. But in a context of uncertainty, from the moment when information is no longer perfect, the situation changes completely. In a well-known article, Weitzmann showed that the choice between taxes and permits depends on the shape of marginal cost curves and of marginal damage curves[19]. In the case of climate change, the marginal cost of reductions increase rapidly as effort increases while the future damage from climate change is only indirectly correlated with current emissions: it is the accumulation of greenhouse gases in the atmosphere that counts, more than the annual volume of emissions. This is the reason why both Weitzmann and Nordhaus recommend using a tax rather than a system of permits to combat emissions. But as we shall see, the options adopted in reality are largely independent of debates among economists. Market systems are seen as being able to offer the political compromises essential to launching collective action. It is for this reason that they were rapidly imposed.

Box 1. Taxation vs. a permits market: basic economic analysis This analysis is carried out with the help of Graphs 1 and 2. The cost curve C1 links the

cost of emissions reduction to the total volume of emissions. Its slope is negative: the right-hand part of the curve shows emissions that can begin to be reduced at a lower cost, for example by eliminating energy waste. Once these initial reductions have been implemented, more costly operations will have to be undertaken, involving for example changes in equipment or organization. And so on, as one moves from right to left along the curve.

261

Fig 1. Emissions price and volume at equibrilium The curve D1 shows the damage generated by emissions. It has a positive slope. Since

climate warming is produced by the accumulation of emissions, it is therefore not the first emissions but the most recent that have the greatest effect on the climate. Eliminating these gives rise to a higher overall social benefit. On the graphs, the slopes of curves C1 and D1 are equal.

The desirable emissions quantity 1Q , the “optimal quantity'' in economists' terminology, is

the crossover point of the two curves at 1A . If one reduces emissions more, to the left of 1Q , one goes too far: the cost of emissions reduction is higher than the benefit that society will obtain from the elimination of marginal damage. Conversely, if one moves further than 1Q one loses the social benefit that the community would obtain from emissions reductions situated to the right of

1Q . The aim therefore is to reach point 1A . To get there, two routes are open to the public authorities.

• Introduce a tax 1P . Economic agents who have a marginal cost of reduction lower

than 1P have a financial interest in eliminating their emissions in order to avoid paying the tax.

Agents who have a marginal cost higher than 1P will continue to emit. Indeed it is in their interest to pay the carbon tax, which is lower than their cost of reduction. Agents adjust their emissions according to the price signal of the tax and bring their emissions to 1Q .

• The public authorities can also set the quantities while leaving the market to take care of the price. In this case they set the overall emission ceiling 1Q , which is imposed on all actors in the economy. This ceiling represents the total right of use of the atmosphere for storing

262

greenhouse gas discharges. It will then be apportioned in the form of permits among emitters, who can either use these permits to legally cover their emissions, keep them or sell the permits on the market. Each actor will decide to buy or sell his permits by comparing the market price to his own marginal cost. Those who have a marginal cost lower than 1P will be sellers, and those who

have a marginal cost higher than 1P will be buyers. The market price will therefore rapidly

converge toward 1P , whatever the level of the tax. If the markets are efficient and if the public authorities are perfectly informed, setting up

a system of tradable permits or the introduction of a tax are therefore strictly equivalent. In reality, the markets are not totally efficient. Setting them up generates transaction costs

that can turn out in practice to be higher than the cost of collecting a tax. Moreover, if the market is too limited or if it lacks liquidity, it will not give rise to a sufficiently stable price signal that the actors would internalize as they would a tax. Hence there are practical conditions to be met in order for the system of tradable permits to work. But it is by taking account of uncertainty that it becomes possible to decide in favor of one instrument rather than another.

In actual fact, the public authority does not know with any certainty either the distribution of emissions reduction costs or the distribution of the damage that emissions give rise to. This uncertainty can be analyzed by means of Graph 2. A second cost curve C0, situated below C1, has been added. C1 was the curve anticipated by the public authority. C0 is the real curve which turns out to be lower than the anticipated curve (this is generally so in practice). We thus have the elements, following the economist Weizman who in 1974 constructed this analytic framework, to evaluate the comparative advantages of the two systems.

Fig. 2. Costs of imperfect information

263

The optimal point which balances costs and marginal damage reduction is now 0A . Due

to the lack of information available, the action taken by the public authority has not led to the optimum.

• If the public authority sets up a quota system, real emissions remain fixed at 1Q and

the market equilibrium price falls to *1P . This price is lower than 0P , which is the optimal price.

The emissions 1Q are higher than the desired emissions 0Q . The result of this is a loss for the

community, measured by the area of the triangle T1 (blue hatching). • If the public authority introduces a tax 1P , this is higher than the desired price 0P . The

reduction effort from then on becomes higher than what is economically desirable. The emissions *1Q are lower than 0Q and society suffers a loss measured by the area of the triangle T2 (green

hatching). On our graph, the two areas are the same, since the slope of the cost curve is identical to

that of the damage curve. This is the only case, in a situation of uncertainty, where the tax and the permit system are economically equivalent. If the slope of the cost curve is higher than that of the damage curve, the area of T1 is greater than the area of T2. Conversely, if the slope of the damage curve is higher than that of the cost curve, in other words, if the damage suddenly rises above an emission threshold, it is preferable to use permits.

In the short and medium term, the marginal cost curve is likely to be more steeply sloped than that the marginal damage curve. Given existing technologies, there are few or no easy substitutes that may be adopted in order to significantly reduce emissions produced from the use of fossil fuels. In addition, the amount of damage grows slowly when emissions increase because of the inertia of the climate system. Hence, to minimize the costs of uncertainty, some economists such as William Nordhaus recommend introducing a tax rather than a permit market in order to set a price for carbon.

The preceding reasoning is valid only in the short and medium term. In the long term, if a public authority is able to set a credible emissions reduction target, the marginal cost curve flattens out and the damage curve becomes steeper. The economic appeal of the permits system increases, as Nicholas Stern reminds us. One can also add that the introduction of a market instrument is the surest way of disseminating information to all actors and public authorities on the real distribution of emissions reduction costs. There are also strong economic arguments in favor of a tradable permits market system.

3. The Kyoto protocol's flexibility mechanisms The first international attempt to price carbon dates from 1992. It was a European

initiative and took the form of a proposal to the European Commission to gradually institute a harmonized tax on industrial 2CO emissions in the European Union. It came up against head-on opposition from industry. It also engendered the hostility of the majority of member states, which were disinclined to give up part of their sovereignty regarding taxation, even in the name of environmental protection. As a result, in 1997 the Commission formally abandoned this project of a harmonized European 2CO tax.

264

Logically, the European Union defended the principle of a harmonized global tax on carbon in the multilateral negotiations that led in December 1997 to the signing of the Kyoto protocol. There it came up against the twofold opposition of developing countries hostile to any sharing of a constraint and of the US delegation which was in favor of setting a ceiling on greenhouse gas emissions and of using a system of internationally-traded emission permits to limit the costs. After much discussion, it was this type of architecture that was adopted at the signing of the Kyoto protocol in 1997. At the time it was viewed as a victory for the principal American negotiator, who was none other than vice-president Al Gore. The European Union was then fairly rapidly converted: as of June 1998, the Commission was completing an enquiry process aimed at setting up a European system of permit trading. Following the withdrawal of the United States from the Kyoto protocol in 2001, Europe paradoxically became the world's main propagator of negotiable permit markets.

The Kyoto protocol commits the group of industrialized countries and countries in transition (the so-called “Annex B'' countries) to a market economy to reduce their average emissions from 2008 to 2012 by 5% compared to the reference year 1990. It thereby restrains the free and unlimited use of the atmosphere that prevailed previously. To limit the cost of the obligation, the protocol makes provision for “flexibility mechanisms'', defined in articles 6, 12 and 17. These are the basis for carbon markets.

Article 17 authorizes, within certain limits, trading of emission rights among Annex B countries, which have been set for the 2008-2012 period by their cap obligations. By doing so, it transposes to a national scale the cap and trade system which had until then applied to 2SO emissions from power plants in the United States. A country which bears high reduction costs will be able to meet part of its obligations by buying Kyoto emission rights from a country that is better positioned. Article 17 lays the foundations of an international carbon market among countries having emissions reduction obligations.

Articles 6 and 12 complement this first mechanism by creating a projects system. The idea is to allow countries or voluntary actors capable of reducing greenhouse gas emissions to obtain credits which can be priced on the international market. These credits should financially encourage countries such as China, India, Brazil, Russia and Ukraine3 to launch emission-reducing projects without waiting to be actually restrained by an international treaty. The purchase of credits by the industrialized countries of Annex B should at the same time enable them to reduce the cost of attaining their emission reduction targets. For example, it is economically rational to begin by capturing methane from Chinese mines for a dollar for every tonne of 2CO avoided rather than look for emission reductions at 80 euro per tonne of 2CO avoided in western Europe.

The mechanisms of the new carbon economy have certain similarities to those of currency creation. By ratifying the Kyoto protocol, each country acknowledges an environmental debt constituted by emissions of the six greenhouse gases covered by the protocol. Through the flexibility mechanisms, the moral debt in relation to future generations acquires financial substance. It must be settled in carbon currency (emission permits) which must be refunded in amounts equal to the emissions.

3Russia and Ukraine, as members of Annex B, have reduction obligations under the Kyoto protocol. These obligations are purely formal and do not impose real emissions constraints to these two countries which have been allocated huge emission rights.

265

4. The launch of the European 2CO trading system The European Union chose to prepare for the 2008 launch of the first Kyoto trading

period by establishing its own emission permits market in January 2005, the European Union Emissions Trading Scheme (EU ETS). This program applies to 11,500 industrial installations, representing 42% of the European Union's greenhouse gas emissions. It applies only to 2CO (and

to a small extent ON2 from 2008 onward). The sectors covered are electric power-generating companies (a little over 60% of allocated quotas) and energy-intensive industry, including steel, cement and glass manufacturers. Each installation has been given an emissions ceiling instantiated by the allocation of a certain number of quotas --- each quota giving the right to emit a tonne of 2CO --- which it must not exceed each year. To be in compliance, an installation can either reduce its emissions to the level of its ceiling or buy another installation's quotas to reduce its own below the ceiling.

Paper

2%

Ceramics

1%

Glass

1%Cement

9%

Combustion

69%

Oil refining

8%

Iron and steel

10%

Fig. 1: European industries subject to quotas in 2007 Source : Mission Climat de la Caisse des Dépôts from the CITL

The European carbon market covers two periods. 2005-2007 was a time of start-up and

learning. The second, 2008-2012, is the period of obligations under the Kyoto protocol. From 2008, the European quota market will therefore integrate itself with the flexibility mechanisms proposed by the protocol. Within each of these two periods, industries may take up unused quotas from one year to cover their emissions for the following year. On the other hand, they are not permitted to carry over unused quotas from the first period to the second. This so-called “non-bankability'' rule between the two periods is crucial for understanding the market during its first three years4.

4 For a more complete analysis, see De Perthuis C., Convery F., Ellerman D., The European Carbon Market in Action: Lessons from the First Trading Period, Mission Climat of Caisse des Dépôts, University College of Dublin, Center for Energy and Environmental Policies Research of MIT, March 2008.

266

With 262 million tonnes of CO2 traded, 12% of the quotas allocated to industry were traded in 2005. In 2006, the volume of transactions soared, rising to 818 million tonnes of 2CO ,

nearly 40% of the quotas allocated to industry. It reached 1.4 billion tonnes of 2CO in 2007. These figures mean that the European trading system is by far the largest emission permits market in the world. The World Bank estimates that this market captured more than 80% of the value of the global trade in carbon from 2005 to 2007. As a result, the European market has become the international benchmark for the price of carbon.

Fig. 2: The world carbon market in 2007 Source : World Bank

When the market was launched, the price of a tonne of 2CO was 7 euros. The price

initially rose rapidly, under the impact of demand from electric power generating companies, and remained above 20 euros a tonne. In spring 2006, the market acquired the first full information on real emissions in 2005: during the first year of operation, quotas allocated were 4% higher than real emissions. The spot price immediately fell by more than half. This first alert was followed by other corrections. The mild wet weather and the fall in the price of gas significantly reduced the demand for quotas on the part of power-generating companies during the winter of 2006. In 2007, the quota price of the first period was on average below one euro a tonne.

267

Figure 3: The price of carbon on the European market Source: Bulletin mensuel Tendances Carbone The first period allocations were the result of bargaining between industry, the member states and the Commission. These involved a moderate constraint for industry except for the electric power sector where significant demand for quotas emerged during the first two years. In other respects, some countries were manifestly more flexible, indeed more lax, than others in the first period allocations. In view of this experience, the Commission adopted a more standards-based procedure for the second period allocations. Overall, quota allocations were reduced by 9% in the second period (-15% in France). For this reason, the second period quota price of 2CO was close to 25 euros a tonne in the first half of 2008. The entry of European industry into the world recession then triggered a sharp fall of carbon prices at the end of 2008. But the banking provisions between the second and the third periods helped the market find a new equilibrium at prices over 13 euros a tonne as from the end of April 2009. All major industry and finance players now no longer consider carbon to be free in Europe, and expect that it will continue to be costly in the future. This is a major achievement.

At more than 20 euros a tonne, most large companies subject to quotas take the emissions price into account in their day-to-day management decisions. Studies reveal that this has already triggered significant emissions reductions: in each of the years 2005 and 2006, some 75 million tonnes of CO2 are likely to have been abated by industry[9]. Although significant, this is not sufficient to radically alter their energy choices and orient investment toward less carbon-intensive technologies.

All in all, Europe has in three years managed to create a credible system that functions in a community of 27 countries. These nations have arrived at the necessary compromises to overcome their sometimes conflicting interests. Admittedly, this system is regularly criticized, particularly among the 12 new member states of the European Union, which have been obliged to

268

accept the rules of the carbon market in the name of the acquis communautaire. But its legitimacy is strong. The 2CO quota trading system benefits from the support of most environmental organizations. It is managed by a competent community administration that is open externally. Lastly, it is supported by the commitments of heads of state who, at the European Council meeting of March 2007, adopted the target of a unilateral 20% reduction in greenhouse gas emissions by 2020, compared to 1990 levels. For these reasons, no one expects a step backwards after 2012: throughout the European continent, the era of free carbon emissions has very much come to an end.

5. The international projects market The other main pillar of carbon finance is the international projects market which has

developed since 2003 within the framework of the Kyoto protocol's flexibility mechanisms. The Clean Development Mechanism (CDM) is the principal component of this. The CDM allows for the crediting of emissions reductions obtained through voluntary projects implemented in developing countries which are not subject to obligations to reduce their own emissions. Industrialized nations may use credits generated through the CDM to meet a portion of their emission commitments. The CDM provides the only link between the industrialized countries of the north and the developing countries of the south in international climate agreements.

By the end of 2009, some 4600 CDM projects were registered, of which a little over 1800 have been approved by the United Nations agency responsible for the mechanism. Taken as a whole, these projects represent an emissions reduction potential of more than 2.5 billion tonnes of

2CO equivalent between now and 2012, thanks to the CDM. In order of size, this is slightly more than 1% of world greenhouse gas emissions. Needless to say, this is not hugely significant in terms of the overall stakes, but it is unquestionably a first step.

Looking at the type of projects developed so far provides some surprises. China is established as the leading world supplier of CDM credits, with nearly half the market in 2006 and 2007. It is followed by India, South Korea and Brazil. This concentration of supply results from the disproportionate weight of some fifteen very large-scale projects enabling industrial gas emissions (HFC and ON2 ) from large factories to be reduced at low cost. This windfall effect has undoubtedly occurred to the detriment of projects that are more formative for the future of energy systems in developing countries. It has, moreover, largely left the least developed countries on the sidelines in terms of participation in the CDM.

Despite its rapid acceleration, the CDM has not had a structuring effect on the development of energy infrastructure in developing countries, which are investing massively in new installations that will continue to burn fossil fuels over the coming decades. There remains therefore much room for progress. Three ways forward are currently being studied: providing more flexibility and incentives for the development of small-scale projects; facilitating the grouping together of basic operations into genuine sectoral programs that could obtain credits; and finding a way of providing credits for avoided deforestation, an area in which some progress was made at the December 2007 climate change conference in Bali, Indonesia.

Once issued, carbon credits linked to Kyoto projects should be given value through actors who are willing to buy them to meet their compliance targets. If voluntary initiatives are excluded, two main types of actors may need to procure emissions reduction credits validated by the Kyoto system: countries obliged by the Kyoto protocol to reduce their emissions and industrial companies subject to emissions constraints. These two types of buyers are found in the

269

60 or so “carbon funds'' which have been developed around the world following the launch of the Prototype Carbon Fund by the World Bank in 1999. The great majority of investors are European, followed some way behind by Japanese investors.

The preponderance of European buyers of Kyoto credits is clearly seen in the setting of prices. The value of Kyoto credits is established according to the price of contracts for emission quotas on the European market, reduced by a premium that takes into account the specific risks of the emission reduction project. The growth of the European 2CO trading system has thus greatly contributed to the launch of the Kyoto projects market by providing a reliable carbon price for project actors.

6. Carbon markets in post-2012 negotiations Since January 2005, carbon markets have been rapidly deployed on the ground, whereas

international negotiations have not made significant progress. The December 2007 Bali conference in particular gave an impression of irreconcilable differences between countries. However, even in the event of a setback in international climate negotiations, no one expects a return to the situation prevailing before 1997, when there was free and unlimited use of the atmosphere for storing greenhouse gas emissions. Carbon markets will continue to function, but in different ways depending on whether or not an international treaty on the climate is agreed upon.

In January 2008, the European Commission put forward its “energy and climate'' package. Measures concerning the European carbon market are incorporated into a much wider policy targeting three objectives: reducing EU-27 greenhouse gas emissions by 20% compared to 1990 levels; raising the proportion of renewable energy used in Europe to 20% by 2020; and increasing energy efficiency by 20% within the same time frame. The initiative is known as the “three twenties''.

Regarding greenhouse gas emissions, the energy and climate package firstly reinforces the environmental constraint on industry. It aims at moving from a system in which free allocation is the rule and auctions the exception to the reverse situation: according to the proposal, all allocations for the electric power sector must be auctioned from January 2013 onward, with a more gradual introduction of auctions for other industries. To help attain the 20% reduction in total emissions, the European ceiling would have to decrease by slightly less than 2% per annum between 2013 and 2020.

The detailed architecture of the European quota trading system will become clear only once the outcome of international negotiations is known. In the event of a “satisfactory'' post-Kyoto international agreement, European heads of state have committed themselves to a 30% reduction in greenhouse gas emissions in Europe compared to 1990. In such a scenario, the constraints weighing on European industry would be proportionately increased, but so too would the flexibility mechanisms. In the event of an international agreement, the Commission would accept half the additional effort by industry to be covered by the purchase of credits from project mechanisms. This has a twofold function: on the one hand to serve as a carrot to compensate emerging countries like India and China which are benefiting from these project mechanisms; and on the other to limit the rise in emission costs and the increase in the price of carbon in Europe.

The question is of course what counts as a “satisfactory international agreement''. The first condition for this is that all industrialized countries participate. Yet despite very general

270

declarations, such an agreement at the December 2009 climate negotiations in Copenhagen is by no means guaranteed. Neither Canada nor Australia --- a recent signatory to the Kyoto protocol --- will meet their Kyoto targets and will finish the 2008-2012 period with emission levels per inhabitant higher than the United States. Japan is in an ambiguous position: as a champion of energy efficiency, it has already widely mobilized the least costly short-term methods to reduce emissions. The United States remains, where perhaps the greatest surprises will occur. In the United States, a voluntary carbon market has been in operation in Chicago since 2001. The Chicago Climate Exchange has a limited reach and the carbon price there is low. In January 2009, 10 northeastern American states launched a mandatory carbon market covering electric power plants – the Regional Greenhouse Gas Initiative (RGGI) – and the State of California plans to launch its trading scheme for greenhouse gases by 2012. RGGI has received less media coverage than the Californian project, but is more advanced in concrete terms. Under the presidency of Barack Obama, these regional experiments have every chance of merging into a federal system. The US House of Representatives has already adopted the draft of the "American Clean Energy and Security Act of 2009", also called the Waxman-Markey project, which has to be discussed and approved by the US Senate before entering into operation. The draft aims at establishing a broader carbon trading scheme than the one operating in Europe, covering 85% of the country's greenhouse gas emissions. If the discussion draft is passed in its current form, the US will come to the international negotiating table in Copenhagen with an internal system requiring its domestic sources to reduce their emissions of greenhouse gases by 17% by 2020 relative to 2005 levels. This reduction commitment is of similar magnitude to the one adopted by Europe (-19% relative to 2005 levels). However, the US effort is much lower in relation to 1990 levels because the emissions of the European Union declined slightly between 1990 and 2005, whereas US emissions increased by 16%. By following such a path, the US can put some credibility behind its commitments and exert a positive influence at the forthcoming Copenhagen Conference.

Conclusion: a fragmented market or one unified by a post-Kyoto agreement?

Carbon markets are sometimes presented as alternatives to public action in the face of

climate change. Such a view is misleading. Carbon markets are instruments through which a carbon price enables strong incentives to be given to economic actors by decreasing the cost of emissions reductions. But by setting emissions ceilings, it is governments which determine the amount of emissions reductions and therefore the carbon price needed to achieve them. Were European governments suddenly to renounce their commitments, the price of carbon would collapse and the market would disappear. If governments act in concert, the carbon market will gain in depth and effectiveness. If they do not, carbon markets will become fragmented and therefore less effective both economically and ecologically.

Carbon markets also play a key role in international climate negotiations, since they facilitate the emergence of compromise. The economic value given to greenhouse gas emissions allows bargaining that can bring together initially very divergent positions. It was this type of compromise that enabled countries like Russia and Ukraine to be brought into the Kyoto protocol. For the next stages of international negotiations, three parameters must be taken into consideration: the advance represented by the European carbon trading system; the benefits that major emerging countries such as China, India and Brazil realize by being able to use the

271

international carbon market to reinforce their emissions reduction efforts; and the likely introduction during the coming years of a federal cap on greenhouse gas emissions in the United States. Without a carbon market, the chances of a post-Kyoto international agreement would be poor. Thanks to the existence of these markets, the chances are possibly higher than the gulf between the positions of different governments would lead us to suppose. Next stop Copenhagen, December 2009.

References [1] Aldy, J., Stavins, R. (2007) “Architectures for Agreement, Adressin Global Climate

Change in the Post-Kyoto World”, Cambridge University Press. “ [2] Bellassen, B. and Leguet B. (2009). “Comprendre la compensation carbone”,

Pearson.“ [3] Cochran, I.T., and Leguet, B. (2007). “Carbon Investment Funds: The Influx of

Private Capital”, Mission Climat Research Report, no. 12 (November 2007). [4] Convery, F.J., and Redmond, L. (2007). “Market and Price Developments in the

European Union Emissions Trading Scheme”, Review of Environmental Economics and Policy, Volume 1, Issue 1 , Oxford University Press.

[5] Dales, J.H. (1968). “Pollution, Property and Prices”, University of Toronto Press. [6] De Perthuis, C., Convery, F., and Ellerman, A.D. (2008). “The European Carbon

Market in Action: Lessons from the first Trading Period'', Interim Report, Mission Climat CDC, UCD, MIT-CEEPR, March 2008.

[7] De Perthuis, C. (2008). “Le puzzle des marchés du carbone”, Pour la Science, no. 365 (March 2008), pp. 44-50.

[8] De Perthuis, C. (2009). “Et pour quelques degrés de plus... Nos choix économiques face au risque climatique”, Pearson. (forthcoming English Edition : “Economic Choices in a Warming World”, Cambridge University Press)

[9] Ellerman, A.D., and Buchner, B. (2008). “Over-allocation or Abatement: A Preliminary Analysis of the EU ETS based on the 2005-06 Emissions Data”, Environmental and Resource Economics.

[10] Ellerman, A.D., Convery, F., De Perthuis, C. et alii (2009). “Pricing Carbon, The European Emission Trading Scheme”, Cambridge University Press.“

[11] European Commission (2000), “Green Paper on greenhouse gas emissions trading within the European Union”, COM(2000)87, 3 March.

[12] European Commission (2004a) “Commission decision for the monitoring and reporting of greenhouse gas emissions pursuant to Directive 2003/87/EC of the European Parliament and of the Council”, 2004/156/EC dated 29 January 2004.

[13] Hardin, G. (1968). “The Tragedy of the Commons”, Science, Vol.162, 13 December 1968

[14] Kruger, J., Oates, W.E., and Pizer, W.A. (2007). “Decentralization in the EU Emissions Trading Scheme and Lessons for Global Policy”, Review of Environmental Economics and Policy, I.1 (Winter 2007), pp. 112-133.

[15] “ [16] Stern, N. (2006). “The Stern Review on the Economics of Climate Change”,

Cambridge University Press.

272

[17] Trotignon, R., and Delbosc, A. (2008). “European CO2 Market and the CITL: The Trial Period Under Scrutiny”, Mission Climat Research Report, no. 13 (April 2008). Forthcoming, accessible at http://www.caissedesdepots.fr/missionclimat.

[18] Trotignon, R (2009), “Comprendre le changement climaique”, Pearson. “ [19] Weitzman, M.L. (1974). “Prices vs quantities”, Review of Economic Studies, vol.41,

October 1974, pp. 447-491.

273

Prices Caps and Price Floors in Climate Policy12

Cédric Philibert

Abstract: This study assesses the long-term economic and environmental effects of

introducing price caps and price floors in hypothetical climate change mitigation architecture, which aims to reduce global energy-related CO2 emissions by 50% by 2050. Based on abatement costs in IPCC and IEA reports, this quantitative analysis confirms what qualitative analyses have already suggested: introducing price caps could significantly reduce economic uncertainty. This uncertainty stems primarily from unpredictable economic growth and energy prices, and ultimately unabated emission trends. In addition, the development of abatement technologies is uncertain.

With price caps, the expected costs could be reduced by about 50% and the uncertainty on economic costs could be one order of magnitude lower. Reducing economic uncertainties may spur the adoption of more ambitious policies by helping to alleviate policy makers’ concerns of economic risks. Meanwhile, price floors would reduce the level of emissions beyond the objective if the abatement costs ended up lower than forecasted.

If caps and floors are commensurate with the ambition of the policy pursued and combined with slightly tightened emission objectives, climatic results could be on average similar to those achieved with “straight” objectives (i.e. with no cost-containment mechanism).

Keywords: climate change; economic uncertainty; emission allowances; GHG emissions;

mitigation policy; price caps; price floors

1 © OECD/IEA, December 2008. This paper was first published as an International Energy Agency Information

paper, under the title Price Caps and Price Floors in Climate Policy - A Quantitative Assessment. We are very grateful to Cédric Philibert and to the IAE for their permission to reprint these pages

2 This chapter builds on a longer study written by Philibert (2008). The author would like to reiterate his gratitude to all experts named in his earlier paper on the same topic, and expresses additional gratitude to Metin Celebi and Denny Ellerman for their suggestions and to Martin Weitzman for his understanding and support.

274

Climate change mitigation policies aim at reducing the severe risks associated with uncontrolled climate change. As they would require deep change in mankind’s energy structure they will entail costs. Climate-friendly energy technologies, from energy efficiency to renewable sources to nuclear to carbon capture and storage will develop, but their future costs cannot be predicted accurately. Moreover, business-as-usual emissions are unknown, and depend on uncertain economic growth, relative price of energy fuels, and our energy mix. Therefore, mitigation costs are highly uncertain when quantitative emission targets are set years in advance.

Price caps have been suggested as a way to contain these uncertain costs3. Under a price cap, emissions beyond the quantitative targets or cap would take place, but emitters – firms or households at the domestic level, countries at the international level – would need to buy additional allowances at a set price. While cap-and-trade would make the objective cost-effective, if the cost were nevertheless to exceed some preset threshold, emissions beyond the targets would be possible but heavily taxed.

While price caps would kick in if costs end up much higher than expected, price floors could also be designed that would kick if costs where much lower than expected. Price floors would help maintain, in probabilistic terms, the emission outcomes of mitigation policies, and have long lasting effects on abatement costs through technology development.

Price caps would alleviate the economic risks of mitigating climate change; however, they would also shift the uncertainty on the side of emissions, and ultimately the climate and the environment. Arguably this may explain the reluctance of many – individuals or NGOs or even government institutions – to consider introducing price caps and price floors in either greenhouse gas domestic emissions trading schemes or the future international climate change mitigation architecture

On the other hand, the reduction of expected costs to mitigation policies may allow for setting slightly more ambitious targets (Philibert and Pershing, 2002). Such a “tightening” of targets could permit designing a climate mitigation regime that is as effective with respective to the climate, but less risky from an economic standpoint, than a regime with “straight” targets, i.e. targets that must be met whatever they cost. These possibilities, however, suggested by intuition, needed to be quantitatively assessed. This was the purpose of some modelling work undertaken at the International Energy Agency (Philibert, 2008), on which this paper further elaborates.

Thus, the present study quantitatively assesses price caps and prices floors in the future global climate mitigation architecture, using a simple model of greenhouse gas mitigation costs building on the entire IEA expertise and flagship publications such as the World Energy Outlook

3 The literature on hybrid instruments dates back to Roberts and Spence, 1976, which is rooted in Weitzman’s framework to choose between economic instruments under uncertain abatement costs (Weitzman, 1974). For a review, see Philibert, 2006. For a more recent assessment of the benefits of using price caps and floors, see Fell and Morgenstern, 2009.

275

and the Energy Technology Perspectives 2008 – the ACTC Model (for “Abatement Costs Temperature Changes”). This model, which also draws on the Fourth Assessment Report of the Intergovernmental Panel on Climate Change to calculate the warming committed in the climate system by 2050, is presented in the first part of this chapter. The execution of the model entails performing thousands of simulations, where uncertain parameters take random values. The results of these simulations are presented in the second part of the chapter. The last part is devoted to comments and conclusions.

1. Methodology

For this research we developed a model of costs of climate mitigation policies, the ACTC Model (for “Abatement Costs Temperature Changes”). The ACTC Model is a highly aggregated model of the global economy, with no distinction of countries or sectors. It projects the growth rate of the global economy and future global energy-related CO2 emissions. As such, it takes no account of land-use, land-use change and forestry emissions or sequestrations, or of other greenhouse gases.

The ACTC Model includes abatement cost curves built on IEA expertise and publications (IEA 2007; IEA 2008), in particular its Energy Technology Perspectives 2008. Uncertainty ranges are further specified according to the Fourth Assessment Report (AR4) of the Intergovernmental Panel on Climate Change (IPCC). Probability functions have been defined that give greater weight to the IEA’s forecasts.

The ACTC Model affords an opportunity to study the costs of quantitative emissions objectives at a global level, objectives which can be either “certain” (“straight”), or “loose” (if there are price caps). It includes an assessment of CO2 concentration levels and resulting temperature changes, but no attempt to monetise policy benefits. Appendix 1 provides a fuller description of the ACTC Model. The model was simply completed with fully specified emission pathways to zero emissions by 2091-2100. While this allows studying the longer term climate effects of introducing price caps and floors from 2011 to 2050, only the economic consequences of the various targets and prices till 2050 are considered in the study.

To take full account of uncertainty, thousands of Monte Carlo simulations were performed to study each combination of targets, price caps and price floors under scrutiny. That is, each uncertain parameter takes random values during the simulations, and all results are compiled to provide probabilistic results in terms of total costs, as well as long term committed temperature changes. The temperatures changes indicated in this paper are not the realised warming at the time considered (2050 or 2100) but the long term equilibrium warming irreversibly committed at the time.

2. Model Results

The simulations were performed considering three cases. The first plays the role of a benchmark: it considers how global warming would evolve, according to the model, if no policy aiming at reducing greenhouse emissions were undertaken. In the second case, we first selected intermediate targets only on the basis of abatement cost optimisation over time and best guess

276

values for all model parameters. We then ran the ACTC model to find out the expected abatement costs when uncertainty is taken into account. In the third case, we assess the efficacy of employing prices caps and floors for the four periods to 2050. We show the results of our testing various levels to find out the likely effects on both abatement costs and actual emissions. Finally, we set up caps and floors at almost exactly the same level, in order to mimic the behaviour of a pure carbon tax.

2.1. No policy case

The ACTC model, reflecting IEA forecasts, indicates that in the absence of policy, global energy-related CO2 emissions would continue to grow, reaching 42 Gt CO2 by 2030 and 60Gt CO2 by 2050 (under best guess). This would lead to CO2 concentrations ranging from 499 to 579 ppm by 2050, and from 662 to 1067 ppm by 2100. Figure 1 represents the probability distribution of the temperature change committed by 2100 in the no policy case (temperature changes in Celsius degrees are in abscissa, the taller the bar the greater the probability). It is virtually certain that the warming would exceed 2°C, and only about 20% chances it does not exceed 4°C. The risk it exceeds 5.25°C is 50% (median value). There are almost 20% risks that the warming exceeds 7°C.

Figure 1: Warming committed by 2100 in the no policy case

2.2. Straight Targets

We first ran the ACTC Model to find the intermediate target values (2011-2020, 2021-2030 and 2031-2040 periods) that minimise the net present value of overall abatement costs up to 2050, on the basis of best guess values of marginal abatement costs in each period. We also find the target values for the following decades that lead to zero emissions during the 2091-2100 decade while minimising the net present value of the overall abatement costs up to 2050, supposing that abatement cost curves beyond 2050 are identical to that of the decade 2041-2050.

Next, we ran three thousand Monte Carlo simulations using the ACTC model to take uncertainties into account. GDP growth rate per decade and carbon intensity, but also coefficients driving the MAC curve, were assigned random values. No price cap was factored in.

277

The net present value of total abatement costs till 2050 for halving 2050 emissions from current (2005) levels has an expected (weighted average) value of USD 7 885 billion, against a best guess value of 2 754 billion. The main reason for that important difference rests in the shape of marginal abatement cost in the ACTC Model, directly derived from that in IEA’s Energy Technology Perspectives 2008 (see Figure 2). When BaU emissions are lower than expected, costs savings are small, while cost increases when BaU emissions are higher than expected may be considerable.

Figure 2: the marginal abatement cost curve of halving global energy-related emissions by 2050 in ETP 2008

It is also interesting to consider total abatement costs as a percentage of the World Gross

Product. The mean value is 0.39%, and the considerable dispersion extends from minus 0.019% to 5.47%.

Let us now consider the environmental results of this policy. The CO2 concentration reaches 462 ppm by 2050. It is exactly known, as one might expect since there are straight targets all the way long. By 2100 it is almost exactly known and ranges from 479 to 484 ppm4. The equilibrium temperature change has a median value of 2.72°C, as shown on Figure 3. Chances of not exceeding 2°C are close to 16%, risks of exceeding 4°C close to 11%.

4 The probability that the concentration is less than 484 ppm is less than 10%. This happens in scenario with low economic growth in the second half of the century, as a result of the construction of the ACTC model. Emission reductions in the first decades, which are supposed to last forever, offset a greatest share of the business-as-usual emissions in the last decades, than is required to achieve the intermediate targets on the way to elimination of emissions by 2100.

278

Figure 3: Warming committed by 2100 with straight targets half of 2005 levels by 2050

The key findings of this second part of the study are the following :

- total expected abatement costs, when uncertainties are taken in account, notably on economic growth, are two to three times higher than best guess estimates when all uncertain parameters take their most likely value.

- halving global energy-related CO2 emissions by 2050 slows the global warming significantly, augmenting the chances of not exceeding 2°C and reducing the risks of exceeding 4°C.

- 2.3. Price caps and price floors

In this section, we assess the efficacy of employing prices caps and floors for the four periods to 2050. We show the results of our testing various levels to find out the likely effects on both abatement costs and actual emissions.

Price caps could be simple compliance payments to governments (for domestic sources) and/or to some international entity (for governments) at the end of commitment periods – at prices set up and known by all at their outset. Contrary to compliance “penalties”, they would waive the obligation to surrender allowances to cover emissions, on a tonne-per-tonne basis. Price floors could be reserve prices (minimum prices) in periodic auctioning, thus creating no liability for government (no subsidy needs).

We combined price caps and price floors. Price floors would augment the costs of climate change mitigation for a given target, everything else being constant (e.g. price cap level). However, for a given expected environmental outcome, a combination of targets, price caps and price floors would entail lower expected costs than having only targets and price caps – for in that case, the targets would need to be even tighter (Philibert, 2008; see also Burtraw et al, 2009).

We set price caps at USD 80, 120, 180 and 260 for the periods 2011-2020, 2021-2030, 2031-2040 and 2041-2050, respectively, and price floors at half these levels, i.e. USD 40, 60, 90 and 130. These values are, respectively, above and below the best guess marginal expected abatement costs for the respective decades.

Let us look first at the emissions during the first period (2011-2020), as shown on Figure 4. They have a mean value of 260.1 Gt CO2, only 1% above the 257.835 Gt CO2 target. In about 24% of the cases, the target would be exceeded by 1 Gt CO2 per year or more, while in about 13.4% of the cases, emissions would be 1 Gt CO2 per year below the target (Figure 4).

279

Figure 4: Actual emissions 2011-2020 with USD 80 price cap and USD 40 price floor

Emissions during the last decade have a mean value 160.5 Gt CO2, still higher than the

135.680 Gt CO2 target, which is reached or beaten in 43.7% of the cases. However, there would be a 5.7% risk that emissions would be twice as much or more than the target (Figure 5).

Figure 5: Actual 2041-2050 emissions with price caps and price floors

Price caps are very effective in reducing the economic risks. Over the entire period to

2050, the net present value of total expected abatement costs would be USD 2 292 billion. Expressed as a percentage of the World Gross Product, total abatement costs now have a mean value is 0.12%, and the dispersion now extends only to 0.19%, i.e. its spread is almost twenty times smaller than with straight targets.

Results expressed in temperature changes by 2100 are worth considering (Figure 6). They are very close to results with straight targets, with a median value of 2.75°C against 2.72°C in the previous case.

280

Figure 6: Warming committed by 2100 with price caps and price floors

With the aim of achieving the same, or better, environmental results as with straight

targets, we optimise intermediate targets towards half of the 1990 levels by 2050. We consider in turn three levels of caps and floors5.

1. In the first case, we maintain the level experienced so far, considering as a distinct possibility that governments primarily want to keep control over marginal costs, say for political reasons. Total discounted abatement costs over 2011-2050 are USD 2 553 bn. Concentration by 2100 range from 443 to 525 ppm, and temperature change by 2100 has a median value of 2.69°C (Figure 7).

While in the previous case the sharp reduction in expected abatement costs had for counterpart a slight increase in the climatic risks, in this case the climate risks are slightly lower, and the costs much lower than with straight targets.

16.8% 72.5% 10.6%

2.00 4.00

equilibrium temperature change 2100

Figure 7: Warming with tighter targets and unchanged caps and floors

5 Note that they all differ from the level considered in Philibert, 2008.

281

2. In the second case we set up the price caps and floors levels at respectively plus 50% above and 50% below marginal costs. Total discounted abatement costs over 2011-2050 are increased to USD 4 638 bn Concentration by 2100 range from 444 to 517 ppm, and temperature change by 2100 show a median value of 2.63°C (Figure 8).

0 1 2 3 4 5 6 7

Median = 2.6320

Figure 8: Warming with tighter target and symmetric caps and floors

3. Finally, we set up caps and floors at almost exactly the same level, in order to mimic the behaviour of a pure carbon tax – set on purpose on the best-guessed marginal abatement costs that would result from an attempt to halve global energy-related 2050 emissions from 1990 levels. The dispersion in emissions is now very important, notably in the last decade, as illustrated on Figure 9.

Figure 9: Emissions 2041-2050 with quasi tax

Total discounted abatement costs over 2011-2050 are USD 4 212 bn . Concentration by 2100 range from 423 to 530 ppm, and temperature change by 2100 has a median value of 2.62°C (Figure 10).

282

Figure 10: Climate by 2100 change with quasi-tax

These results confirm earlier qualitative analyses. Uncertain emission outcomes in a

decade due to price caps create smaller uncertainty on concentration levels, while greenhouse gases slowly accumulate in the atmosphere. Further, this uncertainty on concentration levels is essentially unnoticeable in the final analysis in terms of temperature changes, whether one considers median values or risks of exceeding specific values. Uncertainty on equilibrium temperature change by far dominates uncertainty about concentration levels.

If policy makers primarily want to control marginal costs and set levels of price caps and floors so as to form a price corridor or price collar above and below best-guessed marginal abatement costs, say on the basis of halving 2050 emissions from 2005 levels, it does make sense to tighten the quantity targets in order to improve the environmental outcome with respect to “straight” targets – or at least not deteriorate them. However, if they want to makes use of the economic advantages for price caps and floors to further ameliorate the environmental outcomes, they can increase the levels of price caps and floors or adopt even more ambitious quantitative emission objectives, or both.

We then looked for a combination of targets that would entail the same expected costs as straight targets and deliver presumably better environmental results. One combination that goes a long way in that direction consists in setting the 2050 target at a quarter of the 1990 levels, or 52.6 Gt CO2 in ten years. Optimal intermediate targets were computed to minimise the net present value of abatement costs to 2050, using best guess values. Price caps were set at USD 150, 240, 360 and 600, price floors were set at USD 50, 80, 120 and 200 for the periods 2011-2020, 2021-2030, 2031-2040 and 2041-2050, respectively, so as to be above and below the new best guessed marginal expected costs.

The results show that emissions reach 88 Gt CO2 on average in the period 2041-2050, or 43% of 1990 levels (30% of 2005 levels), as shown in Figure 11. This ambitious target is reached in about 40% of the cases. However, there is a 17% chance that the original, straight target representing half of 2005 levels would be exceeded.

283

Figure 11: Actual emissions 2041-2050 with target 25% 1990 levels, price caps and floors

The net present value of expected abatement costs to 2050 is USD 6 762 billion, which is

still lower than halving emissions from 2005 levels with certainty. Concentrations end up by 2050 in the range 430 to 494 ppm, with mean value 454 ppm. Resulting temperature change committed by 2100 shows a median value of 2.49°C; with chances of not exceeding 2°C at 24.1% and risks of exceeding 4°C at 7.3% (Figure 12). These results are better than those obtained in halving 2050 emissions from 1990 with straight targets, which would entail overall expected abatement costs (NPV) of USD 10 671 billion.

Figure 12: Warming committed by 2100 with target 25% of 1990 levels, price caps and floors

Finally, the ACTC Model has been modified to allow simulating an allowance reserve as

suggested by Murray, Newell and Pizer, 2008. The additional allowances in the reserve would be put on the market if the carbon price reaches some trigger level; however, if the reserve is exhausted, the price is not limited anymore. The system behaves like a “limited” price cap system. In the framework of targets set to halve global energy-related CO2 emissions by 2050 from 2005 levels, with price caps and price floors similar to those described at the beginning of this section, an allowance reserve set at 10% of total amount of allowances for a given period was tested; Figure 13 shows its effects on the distribution of emission outcomes.

284

220

230

240

250

260

270

280

290

Mean = 259.9555

Figure 13: Emissions 2011-2020 with a 10% allowance reserve

A growing reserve, from 10 to 25% of allowed emissions, was also tested. Results are

shown on Table 1 and compared with those obtained with a more ambitious policy (halving global emissions from 1990 levels) and plain price caps. They show that more ambitious policies with unlimited price caps and price floors could be preferred over schemes with price caps limited in size due to an allowance reserve.

Policy Price caps

Price floors

Discounted abatement costs to

2050 (Mean)

Delta-T committed by 2100

(Median) Straight

targets $ 7 885 bn 2.72°C

10% allowance reserve

$80 to 260 $40 to 130

$ 6 282 bn 2.71°C

Growing reserve

(10-25%)

$80 to 260 $40 to 130

$ 5 122 bn 2.72°C

Tighter targets

$80 to 260 $40 to 130

$ 2 153 bn 2.69°C

Table 3: Allowance reserve vs. (unlimited) price caps

Our study of prices caps and floors leaded us to the following conclusions:

- a proper combination of target with price cap and price floor can be designed to offer comparable probabilities of meeting a given temperature outcome at lower expected costs, and whit much narrower uncertainty on total discounted abatement costs, than or straight target. - abatement costs savings due to price caps and, if possible, price floors, allow for setting more ambitious objectives. - an even tighter target with price caps and floors to 2050 would provide environmental results slightly better than halving emissions from 1990 levels, at expected costs lower than those of halving emissions from 2005 levels with certainty, and much lower cost uncertainty.

285

Discussion and Conclusions A number of observations can be made from these results. The first is that mitigation

action, whatever its form and costs, makes an enormous difference in the resulting climate change by the end of this century.

The second is that price caps can significantly reduce cost uncertainty. As shown more in detail in Philibert 2008, price floors help maintain the environmental effectiveness of the policy. While for a given target price floors increase the costs, for a given environmental result price floors contribute to keep costs low.

Price-driven variations in emissions have little influence on policy outcomes (temperature changes), if price cap and price floor levels are roughly consistent with the ambition of the policy. As previously pointed out by earlier work (Pizer, 2002; Newell and Pizer, 2003), this is primarily explained by the stock nature of the issue of climate change: the slow building up of CO2 concentrations smoothes short term emission changes. Our results further point to the role of the uncertainty on the Earth’s climate sensitivity, which by far exceeds and somehow masks the uncertainty on emission levels.

Tighter targets with price caps and price floors entail lesser economic risks and similar or slightly better climate results. With respect to economic risks, taxes are similar to permits with price corridors, and both dominate straight targets.

In these results, the reduction of expected costs can only in part result from so-called time flexibility. Banking and borrowing, for example, could reduce expected costs if the volatility of carbon prices is du to successive economic shocks of opposite signs. If however business-as-emissions show a progressive departure over time from expectation, being constantly above or below expected trend, time flexibility will do little or nothing to reduce costs, if at the end of the day all deviations from targets must be corrected and the “integrity” of the targets must be restored6.

In running the ACTC Model with price caps and floors, the reduction in expected costs results primarily from the flexibility given to adjust the emission levels to the actual abatement costs. This is also what makes it difficult to accept by many. However, this is a logical consequence of accepting that abatement costs are considered in defining the level of abatement. If only benefits were considered, the optimal emission level would be the lowest one -- for the sake of the climate it would be better to halt GHG emissions right away. But this is usually not considered possible, for the costs would be too important.

But if abatement costs are considered in setting the objective, then it is legitimate to adjust the level of abatement to those costs. Uncertainties prevent to perform ex ante a full cost-benefit analysis. To some extent the difficulties in assessing the monetary value of the damages associated with climate change are likely to persist. Nevertheless, as abatement costs are considered in setting quantitative emissions targets, it seems legitimate to adjust them if these costs deviate from expectations. This is what price caps and floors do in a manner that remains predictable by investors and do not undermine the confidence market players may have in the stability of climate change mitigation policies.

In conclusion, short term certainty on emissions appears less important than long term policy ambition. The usual criticism is that marginal damage from climate change may be highly

6 For recent analyses of time flexibility and other cost-containment options, see Whitesell and Davis, 2008, and Fell, McKenzie and Pizer, 2008.

286

nonlinear and grow very rapidly beyond some thresholds. This would justify the choice of a policy delivering sure results, such as straight targets of firm caps on emissions. However, Pizer (2003) showed that there would be a preference for straight targets only if the thresholds – for example GHG concentration thresholds – themselves were know with full certainty. If this is not the case, the most ambitious policy remains preferable.

This appears all the more the case if, as is likely, the nonlinear damage would be triggered by the temperature change. Price caps and floors seem apt to help governments accept slightly more ambitious policies. Given the uncertainty on the Earth’s climate sensitivity, these policies may have slightly better expected results than the policies with more “certain” results due to straight targets. Despite the flexibility allowed in emission levels, the “more ambitious less certain” policies are in practice less likely to meet the unknown temperature change level that would trigger these most-feared nonlinear damages.

This quantitative assessment, however, does not provide all answers. It is limited to carbon dioxide emissions from fossil fuel combustion. It does not address the effects of limiting price volatility on investors’ behaviour. Nor does it address complex implementation issues related to linking a global, international regime, with domestic, nation-wide emissions trading schemes, or linking these schemes altogether, when price caps and price floors are involved. These issues deserve further work. Appendix: the ACTC model A.1. Abatement Costs This appendix describes the model of abatement costs and temperature changes (ACTC) that is being used in this paper to assess price caps and price floors in global climate change mitigation architectures, starting with a brief description of the model structure, then considering how the model addresses business-as-usual emissions and abatement costs, with best guess values, uncertainty ranges, and probability distributions. The model is implemented on a spreadsheet. A commercial add-in allows specifying uncertainty distributions for any variable, running Monte Carlo simulations, and facilitates the collection and presentation of the results. A.2. Model structure Our model is a highly aggregated model of the global economy. It does not distinguish countries or sectors, as if all energy-related CO2 emission sources, in all countries, were similarly capped with perfect emissions trading. The business-as-usual emission pathways and abatement cost functions are consistent with the literature, and in particular, the IEA publications – World Energy Outlook 2006 and World Energy Outlook 2007, Energy Technology Perspectives 2008– which gather and express the whole energy expertise of the IEA, as well as the literature assessed by the IPCC Fourth Assessment Report. A.3 Business-as-usual emissions assumptions

287

Baseline energy-related CO2 emissions (in billion tonnes, or Gigatonnes, or Gt) are calculated by multiplying the World Gross Product (WGP), expressed in USD billion (using purchase power parities by the energy intensity of the economy and the carbon intensity of the energy mix. The energy intensity of the economy is expressed in tonnes oil equivalent per thousand dollars (toe/1000 USD), and the carbon intensity of the energy supply is expressed in tonnes CO2 per tonne oil equivalent (tCO2/toe). The initial World Gross Product, world energy production and world energy-related CO2 emissions are set at USD 54 618 billion, 11 468 million tonnes oil equivalent (toe) and 25 billion tonnes of carbon dioxide (Gt CO2) by 2005, respectively. The World Gross Product is estimated to grow 4.2% per year to 2010, then by 3.3% to 2030, 2.9% to 2050, and 2.25% to the end of the century. Such decline expresses the expected maturation of developing economies as the gap between them and already industrialised countries reduces over time and the demographic growth levels off. These assumptions are slightly different from those in Energy Technology Perspectives 2008 (4.2% growth to 2015, then 3.3% to 2030, then 2.6% to 2050) but lead to almost exactly the same WGP by 2050: a four-fold increase over the 2005 level. The rate of autonomous energy efficiency improvements (AEII) is set constant at 1.7% per year as in Energy Technology Perspectives 2008, almost 1% expressing technical improvements, and the remainder expressing structural changes. The carbon intensity of the energy supply slowly augments till 2030, then more rapidly as coal starts being used on a massive scale to produce liquid fuels, from 2.38 t CO2/toe in 2011-2020 to 2.4 in 2021-2030, 2.55 in 2031-2040, 2.7 in 2041-2050 and 2.8 t CO2/toe from 2051 to 2100. The resulting best-guess CO2 emissions postulated by the model in this baseline scenario are 42 Gt CO2 by 2030, 62 Gt CO2 by 2050 and 90 Gt CO2 by 2100. The first number is consistent with the World Energy Outlook 2007 projection and the median projections of the most recent IPCC scenarios (IPCC 2007, p. 32). The second is consistent with the projection in the Energy Technology Perspectives 2008. The third number is close to the 75th percentile of the recent IPCC scenario (see Figure 6.2). We then introduced uncertainty, with “Beta general” probability functions centred on the best guess values of the GDP growth rates. Minima and maxima are specified at 1.5% and 5.1% to 2030, 1.2 % and 4.6% for the next two decades, 1% to 4% for the rest of the century. We also introduced “Beta general” probability functions for the carbon intensity values, reflecting a growing uncertainty over time. Minima and maxima were specified as follows:

288

Table 6.1: Carbon Intensity Values Running 3000 Monte Carlo simulations produced minima and maxima business-as-usual emissions. By 2030, the range is 29 – 58 Gt CO2, by 2050 it is 39 – 99 Gt CO2, and by 2100 it is 46 – 177 Gt CO2

Figure 6.1: Business-as-usual energy-related CO2 emissions While some may deem our estimates of the uncertainty of the global economy growth rate to be rather large, they somehow compensate for the absence of explicit uncertainty relative to the AEEI — and our relatively optimistic assumptions about it. The IPCC quotes values between 0.5

289

to 1.9%; further, Pielke et al. (2008) have criticised the IPCC for being overly optimistic about both energy efficiency and carbon intensity evolutions. However, these possible biases seem to cancel each other out in forecasting global energy-related CO2 emissions. The match with the post-SRES scenarios shown in Figure 6.1 looks good (industrial CO2 emissions are small compared to energy-related CO2 emissions, accounting for respectively 2.8% and 56.6% of the total GHG emissions in 2004, and thus do not blur the comparison). Indeed, our uncertainty range is narrower than in the IPCC report, in part because the IEA expertise gives little credibility to spontaneous decreases in energy-related CO2 emissions before the end of this century, given the relative abundance and low costs of coal.

Figure 6.2: Energy-related and industrial CO2 emission scenarios (IPCC, 2007) A.4 Abatement cost functions In constructing the abatement cost function, our main assumption was that abatement costs would rise relatively rapidly with the amount of abatement undertaken in a short period of time – while they could possibly diminish over time. Hence, the first choice was to decide on the length of these periods. They could be seen either as “commitment periods”, or more appropriately, as the time lag between the setting of an objective and the end of the period in which this objective must

290

be achieved. The Kyoto Protocol has an initial five-year commitment period – and a fifteen-year time lag between its adoption at Kyoto and the end of its first period (although one would only count seven years and ten months from its entry into force). In any case, the long time lag between Kyoto and the end of the first period may remain an exception. Meanwhile, many have suggested that longer time periods would be more effective (Buchner, 2007). Hence, we took ten-year periods in round figures, starting 1st January 2011. A.5 Marginal abatement curve for 2041-2050 Our guide in shaping the abatement cost function was the abatement cost curve proposed in the Energy Technology Perspectives 2008 (the most recent and accurate global study on abatement technologies) for emission reductions below baseline by 2050, as represented – with its uncertainty range – in Figure 6.3. This figure is a greatly simplified schematic representation. The curve, consisting of hundreds of options conveys two important messages. First, costs are relatively flat up to the ACT Map scenario objective to stabilise emissions at 2005 levels in 2050. But they rise quickly as the additional emissions reduction technologies implicit in the BLUE Map scenario are required. Second, although there is a high degree of uncertainty about the cost of the cheapest reduction measures, they are clearly negative. There is less uncertainty about the cost of technologies needed to achieve the ACT Map target. But costs become more uncertain again as the measures needed to achieve the BLUE Map scenario emission reduction objectives come into play. Nonetheless, ETP 2008 makes clear that on the right-hand side of the picture mostly representing abatement in the transport sector, “the lower end of the range of [marginal costs] has a much higher likelihood than the upper end”.

Figure 6.3: ETP 2008 Abatement Cost Curve by 2050

291

How would costs evolve over time? Endogenous technical changes due to learning and increased R&D efforts responding to carbon prices and emission reduction policies tend to reduce long-term costs. But these cost reductions from technical progress might be counteracted by resource exhaustion. Contrary to common belief, exhaustion also applies to renewable energy sources, when good sites get more scarce or remote, or when the proportion of intermittent sources in electric generation makes it hardly manageable. We needed to take into account the counter-balancing of these effects and introduce them into the ACTC model. The simplest method was to pile up our emission reductions from one period to another. In other words, in period n we took the business-as-usual trend and deducted the emissions that had been abated in period n-1. Thus we made sure to take full account of the long-lasting effects of the previous investments made to reduce emissions. Moreover, we applied the very same function to the next period. This may be questioned. Is it realistic to consider that the first tonne to be reduced in a new period – say, January the 1st, 2021 – could cost much less than the last tonne to be reduced in the former period – say, 31st December, 2020? The answer is clearly yes – as the question is not properly framed. What matters is not so much time as the amount of emission reductions in a given period of time. The “last” reduction in one period is the more expensive, not necessarily the latest. New technical improvements and new opportunities arising from the optimal rotation of capital stock provide in all periods large potential for cheap reductions. The simplest way to mimic this is to re-start our cost function at the beginning of each “period”, following somehow the scheme suggested on Figure 6.4, which describes how R&D and learning processes constantly recreate cheap abatement opportunities (after Grubb, 1997). This is exactly what we endeavoured to reproduce. We then had to derive abatement cost curves for our four ten-year periods to 2050 from this bulk of information. Indeed, instead of considering the wide difference between emissions in, say, the period 2041-2050 with a theoretical business-as-usual baseline, as if nothing will happen between now and 2041, we added emission reductions from one period to another. To do this we multiplied quantities by 10, but we cannot expect that all cheap cost options – such as profitable energy savings – all come first in the first period, leaving only costlier reductions to the next. In particular, capital stock turnover and new technology developments would likely provide new cheap options in each period. Hence we could, for example, consider that the negative to rather low cost potential for a ten-year period would amount to a fourth of the total potential, or 37.5 Gt CO2, i.e. 3.75 Gt CO2 per year.

292

Figure 6.4: Technical improvements recreate cheap opportunities Source: after Grubb, 1997. Reprinted from Energy Policy with the publisher’s permission. While this may be true in the last period, when business-as-usual emissions would be more than half that of today, should we keep such numbers for shorter term negative-to-no cost potential? Keeping that volume constant over the four periods is in fact assuming a decreasing availability (in proportion to unabated emission trends), as the low-hanging fruits get collected first. But other IEA estimates confirm these numbers. For example, the IEA in its Energy Efficiency Policy Recommendations in support of the G8 Plan of Action (IEA, 2008a) assumes a 8.2 Gt CO2 amount of negative to no cost reduction, while our modelling would put the total over 20 years at 7.5 Gt CO2 – seemingly not an overestimation. Another interesting point of comparison here might be the Alternative Policy Scenario in the World Energy Outlook 2006 (IEA, 2006). Annual energy-related CO2 emissions are reduced from baseline in this scenario by 6.3 Gt CO2 per year by 2030. Energy end-users would have spent a cumulative amount of USD 2.4 trillion in energy savings but would save USD 8.1 trillion in their energy bills – an overall negative cost of USD 5.7 trillion. Similarly, we could expect that in each cost range we will find a potential for a ten-year period that is 2.5 times the size of the potential indicated on the MAC per year suggested in ETP 2008. Hence, essentially, we first changed the scale of this graph so that the negative-to-zero (or very low) cost would correspond to a potential of 37.5 Gt CO2 in 10 years, and the USD 200 to 500 range goes for 125 Gt CO2. As a result, the negative-to-zero (or very low) cost potential seen on the ETP MAC curve by 2050 can still be fully realised, provided that 25% is achieved in each of the four decades leading up to 2050. The same applies to each reduction potential corresponding to each MAC level on the curve.

293

When we selected optimal targets for the four decadal periods to 2050, however, we found that technical progress and discounting made the optimal abatement levels uneven between periods. The required total abatement of the last decade was pushed to about 160 Gt CO2 depending on the exact parameters retained for the abatement cost function. As a consequence, the marginal cost in the last period would increase significantly and, while still in the range USD 200 to 500 it does not set a best guess value close to the lower end of the range, as it should according to ETP 2008. Hence, we modified the cost function for this period so as to make it conform more to the existing information – in practice we ended up, with a marginal cost of USD 253 (best guess) for a total emission reduction of 160 Gt CO2 in the 2041-2050 period (as is shown below). We wanted to approximate this curve with functions that would provide us with equations expressing the amount of abatement for a given marginal cost (in case of taxes or price caps) that are easily “solvable” algebraically – if possible. This is to avoid the complications of using the pure force of computers for heuristically seeking solutions while we are performing thousands of Monte Carlo simulations. Fortunately, some relatively simple functions provide an acceptable approximation. We used a composite function in two segments. When abatement was less than 37.5 Gt CO2 (over ten years) we used a biquadratic function, that is:

where x represents the amount of abatement in Gt CO2. When abatement was more than 37.5 Gt CO2 we used another biquadratic function, that is:

This function yields the following values in Table 6.2 (in round numbers) for the indicated values of x. The 112 Gt CO2 abatement value is set to represent the effort of bringing emissions back to their 2005 level in the ETP 2008 ACT scenario with a calculation derived from the adjustment made for the 2041-2050 curve (under best guess). The amount of yearly reductions that corresponds to the ACT scenario, or 35 Gt CO2, was multiplied by 10 to get a decadal figure, then divided by four (because of four ten-year periods), then multiplied by 160/125 for adjustment. ETP 2008 gives USD 50 for the most likely value of the marginal abatement costs by 2050 in the ACT scenario.

Table 6.2: Significant marginal abatement cost values for a 10-year period We introduced uncertainty on this 2041-2050 marginal abatement cost curve, calibrating a low and a high abatement cost functions for 160 Gt CO2 over the 2041-2050 year-period so as to get the minima and maxima suggested by ETP 2008 for 50 Gt CO2 per year in 2050 – USD 200 and 500. This is rendered by the following functions:

294

Using the same values for x as in Table 6.2, these low and high functions return the values shown in Table 6.3 (in round numbers) and are represented on Figure 6.5 by the red and green lines, respectively. The blue line represents the best guess marginal abatement cost curve, and is closer to the lower end of the uncertainty range than to its upper end, as suggested in ETP 2008. The negative to very low cost potential varies from negative values to the value of USD 5, taking into account the possibility that negative or even zero costs were illusory. The horizontal axis was rescaled in comparison to Figure 6.3.

Table 6.3: Minimal and maximal MAC values 2041-2050

295

Figure 6.5: Minimal and maximal MAC values 2041-2051 Uncertainty about abatement costs have been introduced in the model as “Pert” probability functions between the high and low MAC curves for the period 2041-2050. Pert functions are elaborated on Beta general functions but facilitate the specification of best guess values that are not necessary equal to the mean values. Proportionate uncertainty ranges have been introduced for previous and following periods. A.6 Marginal abatement cost curves 2011-2020, 2021-2030, 2031-2040 While we can take this curve as a proxy for the MAC in the decade 2041-2050, we cannot assume that technical improvements only “re-create” low-cost options all along – indeed, this happens mostly due to capital stock turnover. What technical improvements are expected to do is to move down the costs of the expensive options – from carbon dioxide capture and storage to solar electricity to fuel cells for transport and the like. The cost curve at the end of the first half of this century is expected to result from intensive R&D efforts and learning-by-doing processes – and the whole ETP modelling effort is based on learning rates for all these technologies, i.e. rates at which costs are being reduced when cumulative production is doubled. Hence, we could expect that if we were trying to factor in much of these technologies in today’s energy mix, we would be confronted to much higher MAC. For example, in the BLUE scenario of ETP 2008, which would halve global energy-related CO2 emissions by 2050, the cost of deploying carbon dioxide capture and storage (CCS) technologies or the costs of concentrating solar electricity are divided by four by comparison to current levels, the cost of photovoltaic (PV) modules by six, the cost of fuel cells for vehicles probably by even greater numbers. The costs of

296

associated CO2 emissions reductions are likely to be reduced by even greater factors in some cases. For example, when, or if the costs of grid-connected renewable energy technologies are sufficiently reduced to make them competitive, the cost of associated emission reductions becomes null or negative. Hence we can make reasonable assumptions that our cost curve should be moved up for earlier periods. We used the following functions for abatement beyond 37.5 Gt CO2 per decade:

The parameter choices for these MAC functions have been fine-tuned to fit the information available in the IPCC AR4, and in particular, the yearly abatement potential by 2030 at various cost levels seen in Figure 6.6. At a cost of USD 20/t CO2, the ACTC model indicates that 50 Gt CO2 may be avoided from the first decade (2011-2020), and 54 Gt CO2 from additional efforts during the second decade (2021-2030). As the effects of the abatement in the first decades are added to those of the second, the total potential during the period 2021-2030 amounts to 104 Gt CO2, or, on average 10.4 Gt CO2 per year. The IPCC AR4 (Contribution of Working Group III, p. 77) reports from bottom up studies an economic potential from 9 to 17 Gt CO2-eq and from top-down studies 9 to 18 Gt CO2-eq. Arguably, if abatement potentials are proportionate to emissions shares, as energy-related CO2 emissions represent 56.6% of all GHG emissions,47 these potentials would include abatement potential of energy-related CO2 emissions of 5.1 to 10.2 Gt CO2. It thus seems that the model errs a little on the high side of the abatement potential, i.e. the low side of marginal abatement costs. At a cost of USD 50/t CO2, the ACTC model indicates that roughly 62 and 69 Gt CO2 may be avoided in the decades 2011-2020 and 2021-2030, respectively, i.e. a total of 131 Gt CO2, or 13.1 Gt CO2 per year. The IPCC gives 13 to 26 Gt CO2-eq per year from bottom-up studies, 14 to 23 Gt CO2-eq from top-down studies. This may represent energy-related CO2 abatement potential of 7.8 to 15.6 Gt CO2. Our model is close to the middle of those estimates. Finally, at USD 100/t CO2-eq, the ACTC model indicates abatement potential of 74 and 82 Gt CO2 respectively, or a decadal total in the second period of 156 Gt CO2, or 15.6 Gt CO2 per year. The IPCC indicates abatement potentials of 16 to 31 Gt CO2-eq from bottom-up studies, 17-26 Gt CO2-eq from top-down studies, likely to represent about 8.9 to 17.2 Gt energy-related CO2 7 IPCC AR4, Contribution of Working Group III, p.28.

297

abatement potential, and again our model is well in the range, though on the upper side (of potential).

Figure 6.6: Global economic mitigation potential in 2030 from bottom-up studies (left) and topdown studies (right), as reported by the IPCC Fourth Assessment Report (2007) The various MAC curves (best guess values) for the next four ten-year periods are represented in Figure 6.7. The left-hand sides of the curves merge, as the function when abatement is less than 37.5 Gt CO2 per decade is the same for all periods. On best-guess values, these MAC curves end up close to USD 680 for the 2031-2040 period, to USD 1 284 for the 2021-2030 period, and to USD 2026 for the 2011-2020 period. These values would be the marginal cost of abating 160 Gt CO2 in ten years – which will only happen in the 2041-2050 period.

298

Figure 6.7: Marginal Abatement Cost curves for the four 10-year periods We introduced uncertainty ranges and probability density functions for each of the other MAC curves. As for the period 2041-2050, whose functions are repeated below for the sake of immediate comparisons, the uncertainty range for other periods is skewed towards the high side, though to a lesser extent. The functions for decadal abatement quantities (x) over 37.5 Gt CO2 are the following:

A.7 Correlating uncertainties

299

Some combinations of scenarios are more likely to happen than others, however. For example, more rapid growth, while it pushes emissions up, also provides for a more rapid turnover of capital stock and thus more frequent opportunities to implement technical improvements, and eases the funding of research and development. Conversely, slower economic growth would slow down the turnover of capital stock and make the financing of research and development more difficult. To account for these, we introduced a relatively strong degree of negative correlation (-0.7, on a scale that could be from 0 to -1) between the uncertainty about economic growth and the uncertainty about marginal abatement costs – making it more likely that abatement costs would be low if growth were to be fast, and that abatement costs would be high if growth were to be slow. For simplicity, this correlation was only introduced for abatement with positive costs. Introducing this correlation reduced the overall uncertainty (everything else being constant). Rapid growth increases baseline emissions and thus the amount of abatement, moving up the marginal cost curve, but this cost curve is less likely to be shaped by the highest coefficient. Conversely, slow growth reduces baseline emissions and thus keeps the amount of abatement close to the low beginning of the cost curve, but this is less likely to be shaped by the lowest coefficient. This should help avoid overestimating the possible benefits of introducing price caps. A.8 Discounting Discounting is one of the most disputed areas in the economic assessment of climate change.8 The Stern report has been both praised and criticised for its use of a very low discount rate in assessing the costs of climate change. However, a growing body of literature, dating back to Krutilla (1967), Fisher and Krutilla (1974) and Boiteux (1976), points out that environmental assets which are not reproducible nor substitutable by the means of our industry should be given a value that grows over time at a rate close to the discount rate.9 Thus, the Stern report may be “right for the wrong reasons”, and there would be no need to set an arbitrarily low discount rate to perform an acceptable assessment of climate change damage. Still, one must pick a value for some “ordinary” discount rate. France revised its official discount rate in 2005, down from 8% to 4% as a basis for the first 30 years, then slowly declining to 2% until it reaches about 3% for 100-year horizons. The main argument for making the discount rate declines the uncertainty about long-term per capita economic growth. The report that led to that decision also pointed out that “the discounting procedure must be understood as a whole that comprises the discount rate and a system of relative prices in which the price of the environment, in particular, grows significantly faster than other prices.” (Lebègue, 2005). The United Kingdom has come to relatively similar conclusions; its discount rate is 3.5% for the first 30 years, 3% for the next 40 years, 2.5% for the next 50 years, 2% for years 126 to 200, 1.5% for the following century, and then 1% forever. 8 Discounting is the process of finding the present value of an amount of cash at some future date. It thus plays an important role in the economic assessment of climate change, given the long-term perspective implied.

9 For a discussion, see Philibert, 2006b; for an assessment of the Stern report along these lines, see Sterner and Persson, 2007.

300

However, these rates are for France or United Kingdom, two highly developed countries, and they may not apply to developing countries, where capital is scarcer and growth rates higher. For example, the Great Britain H.M. Treasury’s (2003) “Green Book” makes clear that “for international development assistance projects, a discount rate derived from estimates of the social time preference rate appropriate to the recipient economy should be used”. Not only has climate mitigation many aspects of an “international development assistance project”, but the growth rate of the global economy is greater than that of industrialised countries alone. Hence, we took 5% as an overall discount rate, which is consistent with World Energy Outlook analysis for the four decades from 2010 to 2050, and very close to the discount rates embodied in Nordhaus’ Dice model (Nordhaus, 2007). Indeed, relatively small variations around this value would change some numbers, but not the insights provided by the ACTC Model. In any case, the discounting issue has a relatively smaller importance in this study than in studies aiming at establishing the absolute level of (avoided) damage or policy benefits. As we are not trying to estimate the value of climate damage but focus instead on temperature changes to compare various policies, results are much less dependent on choices relative to discounting. A.9 Total costs, price caps and floors The ACTC model calculates total abatement costs for each period by integrating the marginal abatement cost function(s) and the net present values of abatement costs to 2050 and 2100, using the discount rate of 5% mentioned above. Introducing price caps is relatively straightforward. Depending of the value taken by stochastic parameters during the simulation with respect to economic growth, carbon intensity and marginal abatement costs, the model returns for actual abatement either the value of the target – if the marginal cost has not reached the level of the price cap – or the business-as-usual emission level, minus actual reductions in earlier periods, minus the level of abatement that corresponds to the level of the price cap. Introducing price floors follows essentially the same process. It necessitates setting a minimum level that brings in the totality of the negative to low cost abatement potential. We set that level at USD 10, which is twice the postulated “low cost” level. If the marginal abatement cost is between the price floor and the price cap, the model returns the target value for actual emissions. If the marginal abatement cost is below the floor price, the model returns the business-as-usual emission level, minus the actual reductions in earlier periods, minus the level of abatement that corresponds to the level of the price floor. A.10 Concentration levels and temperature changes The ACTC model constantly adds 60% of the CO2 emissions to the CO2 content or the atmosphere. According to the IPCC AR4 (Contribution of Working Group I, Chapter 2, p.139) the apparent “airborne fraction”, defined as the ratio of the annual increase in atmospheric CO2 to the CO2 emissions from annual fossil fuel and cement manufacture combined has averaged about 60% since the 1950s. This ratio may of course evolve over time, either decreasing if global emissions were to significantly decrease, or increasing as a result of some climate feedback – so

301

that the sense of the forthcoming evolution can hardly be predicted. In any case, this simple relationship is sufficiently precise for the next 40 years and the purpose of this study. The estimated equilibrium temperature change (relative to pre-industrial times) in the ACTC Model is directly derived from the probability density function of the equilibrium climate sensitivity provided by the IPCC AR4 (Contribution of Working Group I, Chapter 10, p. 798) – Figures 6.8 and 6.9. If C is the atmospheric CO2 concentration (in ppm), s the climate sensitivity, 275 ppm the pre-industrial CO2 concentration and ∆T the temperature change, the model computes as follows:

Figure 6.8: Climate sensitivity in the ACTC Model

302

Figure 6.9: Equilibrium climate sensitivity in the IPCC AR4 The analysis focuses on the “warming committed by 2050”, that is, the warming already in the “pipeline”, the climate system at that time. Realised (“transient”) warming at that time will be lower, as thermal inertia of the Ocean slows it. However, this committed warming takes no account of emissions subsequent to 2050. The warming committed by 2100 will likely be greater, but its computation would require not only estimating the emissions (and captures) of CO2 from energy and industry, but also land use and land use change, which are beyond the scope of this study. Furthermore, the ultimate, very long-term warming may be even greater than the warming committed by 2100, as the equilibrium climate sensitivity provided by the IPCC AR4 does not include all “slow feedbacks”; for a discussion, see Hansen et al., 2008. References

[1] Burtraw, D., 2009, Addressing Price Volatility in Climate Change Legislation, Prepared for the U.S. House of Representatives Committee on Ways and Means, Resources for the Future, 26 March, www.rff.org/RFF/Documents/RFF-CTst_09-Burtraw.pdf

303

[2] Burtraw, Dallas, Karen Palmer and Danny Kahn, 2009, A Symmetric Safety Valve, Resources for the Future, Washington D.C., February, www.rff.org/RFF/Documents/RFF-DP-09-06.pdf

[3] Fell, Harrison, Ian A. McKenzie and William A. Pizer, 2008, Prices vs. Quantities vs. Bankable Quantities, Resources for the Future, Washington D.C., July, www.rff.org/RFF/Documents/RFF-DP-08-32-REV.pdf

[4] Fell, Harrison and Richard Morgenstern, 2009, Alternative Approaches to Cost-Containment in a Cap-and-Trade System, Resources for the Future, April, www.rff.org/RFF/Documents/RFF-DP-09-14.pdf

[5] IEA, 2007, World Energy Outlook – China and India Insights, IEA/OECD, Paris

[6] IEA, 2008, Energy Technology Perspectives – Scenarios & Strategies to 2050, 2008, IEA/OECD, Paris

[7] Murray, Brian, Richard Newell and William A. Pizer, 2008, Balancing Cost and Emissions Certainty – An Allowance Reserve for Cap-and-Trade, Resources for the Future, Washington D.C., July, http://www.rff.org/RFF/Documents/RFF-DP-08-24.pdf

[8] Newell, Richard G., and William A. Pizer, 2003, ‘Regulating stock Externalities under uncertainty.’ Journal of Environmental Economics and Management, Vol. 45: pp.416-432

[9] Philibert, Cédric, 2006, Certainty vs. Ambition – Economic efficiency in mitigating climate change, LTO Working Paper Series 03, IEA, Paris, October, http://www.iea.org/textbase/papers/2006/certainty_ambition.pdf

[10] Philibert, Cédric, 2008, Price Caps and Price Floors in Climate Policy – A Quantitative Assessment, IEA Information Paper, IEA, Paris, December, http://www.iea.org/textbase/papers/2008/price_caps_floors_web.pdf

[11] Philibert, Cédric and Jonathan Pershing, 2002, Beyond Kyoto – Energy Dynamics and Climate Stabilisation, IEA/OECD, Paris, http://www.iea.org/textbase/envissu/cop9/files/beyond_kyoto.pdf

[12] Pizer, William A., 2002, ‘Combining Price and Quantity Control to Mitigate Global Climate Change’, Journal of Public Economics, Vol. 85: pp. 409-434.

[13] Pizer, William A. 2003, Climate Change Catastrophes, Discussion Paper 03-31, Resources for the Future, Washington D.C., May, http://www.rff.org/RFF/Documents/RFF-DP-03-31.pdf

[14] Roberts, Marc J. and Michael Spence, 1976, ‘Uncertainty and the Choice of Pollution Control Instruments’, Journal of Public Economics, Vol. 5: pp. 193-208.

[15] Weitzman, Martin L., 1974, ‘Prices vs. Quantities’, Review of Economic Studies, Vol. 41: pp. 477-491.

304

[16] Whitesell, William and Stacey Davis, 2008, Cost-Containment in Cap-and-Trade Systems: A Review of the Options, The Center for Clean Air Policy, Washington D.C, October, http://www.ccap.org/docs/resources/542/Cost%20Containment%202008%20(2).pdf

305

Part Five. Socially Responsible Investing

The last part of this book brings together three contributions on socially responsible investing

(SRI). Gunther Capelle Blancard and Stéphanie Monjon try to disentangle myth and reality in

socially responsible investing. Stéphane Voisin and Hubert Jeanneau focus on assessing the

performance of socially responsible investing. Michel Capron and Françoise Quairel analyze

the relationships between sustainable development and corporate social responsibility.

Gunther Capelle-Blancard and Stéphanie Monjon first provide a general account of socially

responsible investing. After a brief historical overview, the authors introduce the different

forms of SRI, present the major players and describe this changing market. They then discuss

the issues raised by SRI, the most important, if not the most complex, of which is conceptual:

how is corporate social responsibility to be defined? The second issue is the scope of SRI. The

third is the connection between social, environmental, economic and financial performance.

The last issue is whether or not SRI could be used as a substitute for regulation.

Stéphane Voisin and Hubert Jeanneau argue that environmental and social benefits have an

intrinsic value. They recall that the main objective of sustainable and responsible investing is

not to outperform and suggest that it provides a “positive” extra-financial dividend in

response to investors’ extra-financial expectations. The most tangible and material

characteristic of corporate social responsibility is the protection against extra-financial risks.

In their study, the authors aim to identify the nature of the risks that can bear on a financial

materiality and to determine how they can be relevant to a company. This approach responds

to all types of environmental, social and public health issues, though it does not, however,

apply to most governance issues.

In their article, Michel Capron and Françoise Quairel analyze the relationship between

corporate social responsibility and sustainable development. Their study aims at highlighting

the power of terminology, power all the greater in that the terminology is used in published

writings. According to the authors, referring to sustainable development to define corporate

306

social responsibility is part of an institutionalization process. Institutionalization derives from

changes through which actions become “usual” in an organization. In this way, companies

have become sustainable development actors. Corporate social responsibility and sustainable

development have become intimately linked. Ultimately, the imputation of responsibility

toward society has shifted from the political to the corporate domain.

307

Socially Responsible Investing: Myths and Realities1

Gunther Capelle-Blancard, Stéphanie Monjon

Abstract: Socially Responsible Investing (SRI) enjoys a large consensus and is often presented as being able to conciliate finance and sustainable development. Actually, the SRI market share stays low (slightly more than 10%), not to say very low if we consider only “Core SRI” (very few percent). Its growth is relatively high in Europe, but its market share is stagnating in the US. Accordingly, for now, SRI does not have a significant impact on firms’ cost of capital. Additionally, a careful review of the academic literature shows clearly that the financial performances of SRI funds are neither better –nor worse– than those of traditional mutual funds.

Doing well by doing good: the intention is undoubtedly admirable. However, it could lead to wishful thinking and blindness to unintended consequences. In this article, we shall warn against two pitfalls: first, attempting to overly inflate the SRI market may lead to a weakening of the concept and to the development of greenwashing; second, claiming that SRI outperforms opens the door to lobbies that will certainly use the argument to defend self-regulation. SRI must not be used as a substitute for regulation.

Keywords: sustainable and responsible investment, ethical investing, business ethics, corporate social responsibility, financial performance, investment screens, mutual funds inflows.

1 We would like to thank Jézabel Couppey-Soubeyran, Stéphanie Giamporcaro-Sauniere and Stéphane Voisin for

their helpful comments, and Barbara Balvet for her outstanding research assistance.

308

Ethics is the subject of a great many articles in the press, as well as of conferences and essays and has become the focus of public debate. Moreover, the issue has been approached from every angle. The ethics of political leaders, scientists, corporations, sports, etc. have been closely examined, and financial markets are coming under scrutiny. In stock markets the trend is currently toward socially responsible investing.

Socially responsible investing1 (SRI) refers to a practice which selects investments not only on the basis of financial criteria such as profitability and risk, but also by taking account of an array of ethical, environmental or social concerns. Although SRI is still relatively uncommon, it has been the subject of a good deal of research. Since the financial collapse caused by the subprime turmoil, SRI has sometimes even been considered as an answer to the moral crisis of capitalism. In any case, SRI enjoys a large consensus: to conciliate finance and sustainable development.2

Doing well by doing good: the intention is undoubtedly admirable, though to such an extent that it may result in wishful thinking. There are currently a number of widespread myths about SRI.

• Is SRI a growing trend? This is true in Europe, but its market share was almost zero a decade ago. In the US, the growth of SRI funds has been slower than that of traditional funds since 1999. Moreover, although SRI began in the US, European investors are now leading the way.

• Is the SRI market share greater than 10% in industrialized countries? The answer is yes, but only if a broad definition is accepted. If only restrictive strategies or the retail market are considered, SRI market share is considerably less.

• Does SRI outperform? No, SRI funds do not perform better than the market as a whole. The good news, however, is that if SRI funds are sufficiently diversified, their financial performance is no worse either. In sum, it is possible for investors “to put their money where their mouth is”, without sacrificing financial performance.

• Does SRI have a significant impact on firms’ cost of capital? No, because SRI market share is too small.

The above-mentioned assertions are based on a careful review of the literature, but the myths are well ingrained. Some SRI enthusiasts are sometimes tempted to suggest that progress is nourished by utopias. Is this any reason to exaggerate the influence of SRI, to embellish its growth or overestimate its financial performance? The problem is that wishful thinking may cause blindness to unintended consequences (Merton, 1936). In this paper, we identify at least two pitfalls.

• Attempting to overly inflate the SRI market may lead to a weakening of the concept and to the development of greenwashing.

• Claiming that SRI outperforms opens the door to lobbies who will use the argument to defend self-regulation. SRI should not, however, be used as a substitute for regulation. Besides, is the very idea of outperformance really in line with the practices dictated by SRI?

The first part of this paper serves as a general presentation. After a brief historical overview, we specify the different forms of SRI, present the major players and describe the changing market. We next discuss the issues raised by SRI: first, the definition of corporate social responsibility, and second, SRI's scope. We then examine the connection between social,

309

environmental, economic and financial performance, and conclude by discussing the substitutability between regulation and self-regulation.

1. Presentation

This part begins with a brief account of the background to SRI, followed by a short presentation of the various SRI-related investment strategies and major players. It concludes with a description of the market and its evolution.

1.1 A brief history

The origin of SRI is conventionally traced back to the 1920s in the United States.3 The Quakers, who were strict about not engaging in any activities forbidden by their moral codes, excluded investment in “sin stocks”, such as alcohol, tobacco and gambling. However, it was really not until the 1970s that the first mutual funds meeting extra-financial criteria appeared in the United States. The first was launched in 1971: the Pax World Fund, founded by two Methodists. The initial goal was not to offer financial performance better than – or even comparable to – other funds, but to propose investments in keeping with the principles of a socially aware client base. As well as the traditional “sin stocks”, others firms were also excluded: those linked to defense/armaments or the Apartheid regime in South Africa (Rudd, 1979; Grossman and Sharpe, 1986; Theo, Welsh and Wizen, 1999).

From the mid-1980s, the scope of SRI began extending to social questions (for example, the exploitation of child labor in sweatshops) and environmental issues (following major industrial disasters such as Bhopal and Chernobyl). However, it is only since the 1990s and still in the United States that some fund managers, prompted by alternative networks independent of the major financial groups, have offered active portfolio selection, not just in terms of exclusion, but based on extra-financial considerations.

In fact the post-1990s SRI boom coincides with the debates on sustainable development, climate change, corporate social responsibility (CSR), fair trade, etc. Since 2000, SRI has also relied on the “irrational exuberance” of stock markets, the explosion of the Internet bubble, the accounting scandals (Enron, WorldCom, Parmalat, etc.) and the financial crisis.

1.2 The different forms of SRI

Many key market players have now been converted to SRI (at least partly). Funds managed according to these principles are termed “socially responsible funds”, “sustainable development funds” or “ethical funds”, and attempt to conciliate financial and extra-financial performance. Financial performance means optimizing the profitability/risk relationship, while extra-financial performance is about complying with certain requirements related to environmental protection, social action and corporate governance (the ESG criteria). Three main approaches can be distinguished, although in practice they are combined.4

The first and the simplest approach is screening. Negative screening is based on excluding certain sectors or firms, on the grounds that they are not “ethical”. Positive screening consists in encouraging industries because of their pioneering involvement in sustainable development, e.g. renewable energies and clean transportation. Funds that practice positive screening are in fact very similar to thematic funds.

310

The best-in-class approach consists in selecting firms with the best financial, environmental and social performance, without excluding any sector. This approach attempts to promote social responsibility in all the firms while favoring the most “virtuous” firms in each sector.

Another approach is shareholder activism, by which investors endeavor to make corporate policy evolve by becoming more involved in driving and shaping firms, as well as putting pressure on managers if need be (Karloff, 2006). As co-owners, shareholders may orient firms toward social responsibility. Shareholder activism ranges from simple dialogue with management through to actively exercising voting rights and putting forward resolutions related to ESG criteria at general meetings of shareholders. Shareholder activism is generally much more widespread in Anglo-Saxon countries than in Europe. It began to take shape in the 1970s during the civil rights and anti-apartheid movements.

1.3 Political initiatives

As well as creating a new supply of financial securities, SRI makes itself felt through initiatives of a political nature that aim at putting pressure on financial markets and firms in order to move them toward sustainable development. As a result, most Western countries have enacted laws to promote SRI.5 In France, for example, article 116 of the law on New Economic Regulations (May 2001) stipulates that listed French firms must provide social and environmental information along with their annual reports. In addition, institutional investors met under the aegis of the United Nations in 2005 to establish the principles of responsible investment (PRI). In May 2008, there were 362 signatories (133 asset owners, 152 investment managers and 77 professional service partners) representing US$13 trillion in assets under management (AUM). It should be noted that ten trillion dollars out of these thirteen trillion are related to European investors.

1.4 Extra-financial analysis and rating agencies

Along with the development of SRI funds during the 1990s, suppliers of extra-financial information came onto the scene.6 They evaluate ethical, social and environmental aspects of firms by analyzing public documents, as well as by sending out questionnaires and conducting discussions. Then, those assessments are sold to fund managers.

Producers of SRI equity indexes should also be mentioned. The first SRI equity index, the Domini Social Index (DSI), resulted from the efforts of the American group KLD (Kinder, Lydenberg, Domini and Co). It is composed of 400 North American firms selected on the basis of two types of criteria. First, sectors such as alcohol, tobacco, gambling or nuclear power are directly excluded. The firms are then selected according to their attention to environmental concerns or their social policies, for example. Given the success of SRI, traditional stock market information providers also began providing SRI indexes, led by Dow Jones (the Dow Jones Sustainability Index family) and the FTSE Group (FTSE4Good).

1.5 SRI's Market Share

It is far from easy to define SRI market share despite the efforts made recently by dedicated professional associations such as the Social Investment Forum in the US, the Eurosif network in Europe and ASrIA in Asia. The difficulty arises of deciding under what circumstances a fund manager is practicing SRI. Is it enough for the manager to declare himself concerned by

311

ESG criteria and to dialogue with firms or should only restrictive SRI strategies be taken into consideration? Depending on the approach, SRI market share will vary considerably.

Most articles on SRI mention a market share between 10% and 20% in industrialized countries. In the United States, for example, the Social Investment Forum reports more than $2,700 billion in AUM in accordance with SRI principles in 2007 (see Figure 1), which accounts for about 10% of the total AUM. In Europe, according to Eurosif, SRI amounted to €2,665 billion in AUM in 2007 (close to $4,000 billion dollars), which corresponds to a market share of 17.5%.

More than 90% of the SRI market is currently dominated by institutional investors such as pension funds or government funds (Norway’s Government Pension Fund, the French Fonds de réserve pour les retraites, etc.), along with foundations, religious institutions and so on. However, such investors often do not adopt very restrictive SRI strategies. In contrast, SRI mutual funds are designed for retail investors and consequently must have a relatively well-substantiated strategy in order to stand out from traditional funds.

After eliminating strategies that, despite being SRI-related, are not really restrictive for fund managers, the SRI market share falls considerably. Eurosif takes an interesting approach, in that it does not simply offer all-inclusive statistics, but also provides estimates on what it describes as Core SRI (as opposed to Broad SRI): multi-criteria exclusion, positive screening, thematic funds and best-in-class selection. Core SRI is estimated at €512 billion in 2007, which is 3.4% of the total AUM (see Figure 2). If multi-criteria exclusion strategies are ignored, AUM amounts to €154 billion. The Social Investment Forum does not give any statistics on Core SRI in the United States, but focusing on investment funds can provide some idea.8

Figure 1. Socially Responsible Investment in the US and in Europe (1995-2007)

639

1185

21592323

21642290

2711

617

1219

3924

0

500

1000

1500

2000

2500

3000

3500

4000

1995 1997 1999 2001 2003 2005 2007

US

Europe

USD billions

Source: Social Investment Forum Foundation and Eurosif European SRI Survey, 2008 (market coverage is not consistent: 8 countries were covered in 2002, 9 in 2005 and 13 in 2007; assets are converted using year-end exchange rates).

312

Figure 2. SRI Strategies applied in Europe

Source: Eurosif European SRI Survey, 2008. Note: the total of individual strategies added together may be greater than the total of Core and Broad SRI due to overlaps.

Generally speaking, the SRI retail market, via investment funds, is in the order of only a few percent. For example in the United States, the number of funds that practiced SRI was 260 in 2007, for $201.8 billion in AUM. According to the Investment Company Institute, there were 9,300 investment funds in the United States at that time, for $13,000 billion in total AUM. In other words, SRI funds in the United States accounted for only 1.5% of AUM at the end of 2007.9 This percentage is about the same in most industrialized countries.10

For a decade now, SRI has progressed more quickly in Europe than in the United States. In 1995, SRI in the United States totaled $639 billion in AUM. After that, growth averaged 36% per year up to 1999, but has been only 2.9% per year since then. Similarly, only 55 investment funds in the United States practiced SRI in 1995 out of a total of 6,200 with $12 billion in AUM, compared to $3,000 billion for traditional funds. Here again growth was very rapid at the outset: the number of SRI funds almost tripled between 1995 and 1997 and the AUM went up by 700%, compared to “only” 57% for funds regarded as traditional. However, since then (see Figure 3) the growth of SRI funds has been slower than that of traditional funds.

Figure 3. Development of Non-SRI and SRI funds in the US (1995-2007)

313

0

50

100

150

200

250

300

1995 1997 1999 2001 2003 2005 2007

Non-SRI funds

SRI funds

Assets = 100 in 1997

Source: Investment Company Institute (Non-SRI funds, including mutual funds, ETF's, and closed-end funds), and Social Investment Forum Foundation (SRI funds, including mutual funds, ETF's, closed-end funds, other pooled products and alternative investments).

In Europe, things have developed differently. Growth seems to have been both significant and steady: AUM grew from €336 billion in 2002, to €1,035 in 2005 and to €2,665 in 2007.11 Europe is now the leading world market for SRI and the situation does not seem likely to change for the time being. American fund managers are the most skeptical about the future of SRI. According to a poll, two-thirds of them think that taking social or environmental filters into account will never become current practice in fund management, whereas this applies to only one out of three European fund managers.12

Whatever the case may be, the question arises as to whether the financial crisis that began in 2007 will change the situation. Most observers consider that there is a moral crisis in financial capitalism. Is this an opportunity for SRI? Possibly, if, as Landier and Nair (2008) claim, SRI is a way of reconciling people with financial markets. At the moment this appears highly unlikely, since it would seem that the crisis has increased people's mistrust of stock markets instead. Because SRI investors are undoubtedly more sensitive to this subject than others, the relative share of SRI may well stabilize or even decrease.

2. Practicing SRI: why and how?

2.1 What is corporate social responsibility?

The champions of SRI, following in the footsteps of the supporters of corporate social responsibility (CSR), are opposed to the standard economic conception of a firm made famous by Milton Friedman (The New York Times Magazine, 13 Sept. 1970): “The social responsibility of business is to increase its profits”. The primacy granted to shareholders is today accepted by most academic economists and this idea is at the basis of research on corporate governance (Hart, 1995; Shleifer and Vishny, 1997).

314

Nevertheless, the partisans of SRI reject this conception and instead attempt to “re-embed”13 economics in society. The underlying idea is to ensure that firms consider not only the interest of shareholders, but also that of stakeholders, i.e. everyone who participates directly or indirectly in the life of the firm: employees, suppliers, consumers, the community and the authorities. The idea that it is possible to conciliate economic, social and environmental concerns harmoniously, with due respect for future generations, is in line with the sustainable development movement.

Many reasons have been put forward to explain the enthusiasm for alternative economic structures, but they can be aggregated into three categories. First, there are sociological reasons: the loss of a traditional frame of reference, for example the decrease in religious practice and absence of political projects. This first category also includes the identity crisis in modern society due to globalization. Next, are the reasons related to increasing environmental risks: never before have global threats like climate change and loss of biodiversity been so serious. Finally, there are economic reasons: the general economic context is unfavorable, with the looming threat of job insecurity, exclusion and mass unemployment, to name but a few. In addition, during the second half of the 20th century, firms saw their power grow considerably. Galbraith (1967) was one of the first to recognize these changes. The acquisition of power by big business in turn gave rise to civil society's call for more responsibility from these firms.

At first glance the stakeholder theory is appealing, but things are not so simple. Wanting to serve the interests of all is praiseworthy, but what can be done when these interests are contradictory? Which decision-making policy should be adopted? Who should make the final decision and to whose benefit? The problem is that the prerogatives, rights and objectives of stakeholders cannot be clearly defined, in contrast to those of shareholders. The corporate finance theory, based on the concept of shareholder value, is in a position to provide answers to the preceding questions. On the other hand, the stakeholder theory finds it difficult to resolve these questions, because it is very complex to determine unequivocally the mandate of leaders in this framework. This aporia leads to an impasse in the decision-making process, reduces the return on investment and increases the corporate cost of capital.14

To sum up, the two ways of thinking collide on the definition of corporate social responsibility, and neither side hesitates to call on popular wisdom. For the partisans of the stakeholder theory: “With great power comes great responsibility” (Stan Lee), while their opponents remind us that “the road to hell is paved with good intentions”.

2.2 SRI: you only get out of it what you put into it

The principal difficulty arising from the stakeholder theory is its ever-changing objectives. The problem can be simplified by breaking it down into three successive levels. 1. Which objectives should be selected? 2. How should previously defined objectives be prioritized? 3. How much compliance should be demanded for these objectives?

1. It is always very tricky to distinguish what is “ethical” from what is not, and even more so because this distinction often raises cultural considerations.15 Consequently, there are marked differences between countries. For example, excluding firms involved in animal testing is an important concern for militant Anglo-Saxon investors, but it does not cause many people to react in continental Europe. Opinions are also strongly divided on investing in nuclear power. It is prohibited by most Anglo-Saxon and Swiss funds, but French SRI fund managers often tend to support power generation by nuclear power plants, arguing that they help control

315

climate change since they do not emit greenhouse gases in the way that fossil-fuel-fired plants do.

2. Assuming that the objectives are clearly defined and consensual, the challenge of prioritizing them still remains. For example, the KLD rating agency, which produced the DSI index, used no less than thirteen positive or negative criteria. These cover a very wide gamut: corporate governance, community, diversity, employees, environment, human rights, products, alcohol, firearms, gambling, military, nuclear and tobacco. A grade is given to every criterion from an aggregation of several indicators, and the grades are then added up to calculate the final grade. In other words, every criterion is weighted in the same way. As a result, a major polluter might be included in the DSI index if the firm pampers its employees! This problem is inherent to the selection process and is reinforced when the best-in-class approach is used, since the relative performance of each firm must be taken into account with respect to the sector. This only adds to complexity, if not to confusion. Whatever the case, the process of successive aggregation causes serious theoretical problems, since it concerns eminently qualitative elements.

Finally, every fund and every index has opted to emphasize environmental protection, corporate governance, human rights, etc. to a greater or lesser degree (Statman, 2006). The tendency is increasingly in favor of thematic funds. In any case, the composition of the main ethical indexes is very similar to the content of traditional indexes.

3. More or better? In the same way as sustainable development or the precautionary principle, SRI is the subject of widespread media coverage. Should this be interpreted as global awareness or trivialization – or even impoverishment – of the practice of SRI? Although there is today a niche that is jealously guarded by the pioneers of SRI, many other players in the movement would like to see increasing use of extra-financial criteria in classical models of financial analysis.

The “purists” fear that wider dissemination may mean weaker requirements and that SRI could be co-opted by the major financial institutions and wind up serving only commercial ends. However, the other danger is one of isolation. Some say that for SRI to have a genuine impact it must become more extensive, even if the criteria become less stringent. How then can stringent requirements be reconciled with wide dissemination? This is one of the main questions that will determine the success of SRI in the medium term.

3. Environmental, social, economic and financial performance

3.1 Does it pay to be green?

Protecting the environment and promoting “citizen values” are very often considered to be incompatible with business objectives. Thus from a traditional perspective, if shareholders, leaders, unions or politicians want firms to set extra-economic objectives, they will necessarily have to choose between a number of public benefits and private costs. Since the 1990s though, this choice is less and less perceived as inevitable. Some believe that economic performance, protection of the environment and promotion of social values could effectively be conciliated.

Porter and van der Linde (1995) were among the first to put forward the hypothesis of a positive relationship between economic and extra-economic corporate performance. Environmental innovations in particular are a source of competitiveness and gains in productivity for the firm: this is the concept of eco-efficiency. The concept was initially

316

proposed in relation to environmental problems, but can easily be extended to societal considerations.

A strategy emphasizing careful management of environmental and social impacts can be profitable, due as much to increased income as to decreased cost. The advantages in terms of income are essentially related to gains in market share (increased visibility, the “first mover advantage”). The advantage in terms of costs is twofold. i) Better control of the costs incurred in the event of accidents and conflicts. Future statutory constraints can be anticipated, insurance and litigation costs can be reduced and the cost of strikes can be lowered. Meanwhile, good industrial relations can be established, thereby improving employee productivity, etc. ii) Savings in terms of equipment and energy. Recycling reduces certain costs and clean technologies are also likely to be more efficient.

3.2 Too good to be true?

Firms that implement “virtuous” strategies are seen by some as providing society with a double dividend. The idea of such win-win strategies is of course enticing. But is it realistic? This highly ideological issue has given rise to lively debate among economists and specialists in corporate strategy. Their positions vary widely, ranging from disdainful skepticism to unwavering trust. Many studies have attempted to answer the question empirically, but the methods are very diverse and so it is difficult to come to a clear conclusion.16 The first point of disagreement is semantic. The notion of social responsibility remains fuzzy and harks back variously to ethics, morals, deontology and paternalism, depending on the case. Some people refer only to legal liability, but for most the commitments made beyond legal or contractual obligations are the most important. The second source of disagreement concerns performance measurement. It is already difficult to evaluate corporate economic and financial performance (based on market value or book value?), but it is even trickier to assess social performance. There are a variety of positive indicators, such as hiring disabled workers, compliance with certain environmental requirements, etc., as well as negative screens, e.g. fraud, exclusion of certain sectors. There are also indicators that are all-inclusive (reputation index) or singular (trade with South Africa), qualitative (job security) or quantifiable (volume of pollutant emissions), provided by the firms themselves in their activity reports, by advisory councils, NGOs, public authorities or international organizations. Testing procedures are a third source of disparity: even if a correlation can be established between economic and extra-economic performance, it is very tricky to demonstrate a causal relationship.

3.3 SRI fund performance

The question for investors is not so much to know if the firm gains anything by being virtuous, but rather to know if SRI is profitable. At first glance, these two questions seem closely interconnected. It is tempting to think that if virtuous firms report good economic performance compared to others, then it follows that SRI must be profitable. In fact, this is not true. In the most favorable case and assuming the eco-efficiency hypothesis is verified, all investors – whether or not guided by strictly economic motives – will buy the stocks issued by virtuous firms. Stock prices will rise, which will translate into decreasing yields and poor performance for investors.

Additionally, when SRI fund performance is examined, it should be noted that excluding certain firms – indeed certain sectors – reduces the opportunity to diversify. Diversification is one of the fundamental concepts in the modern theory of portfolio choice. The theory came into being in the 1950s with the work of forerunners Harry Markowitz (Nobel Prize in 1990),

317

James Tobin (Nobel Prize in 1982) and William Sharpe (Nobel Prize in 1990). The concept boils down to the following proverb: “Don’t put all your eggs in one basket”. The three authors showed that the best strategy for investors consists in holding a portfolio made up of the widest possible choice of assets, thereby allowing for reduced risk without sacrificing expected yield. This theory revolutionized financial practice.17

As a result, the best attitude for fund managers is to track and mirror the market index passively. In other words, there is no need to choose what goes into the portfolio; the only requirement is to hold the largest number of assets possible. This is why most financial investment counselors today recommend investing in index funds.18

Financial theory therefore suggests it is impossible to “beat the market” and SRI funds must not be an exception to the rule. This also means – good news for ethical investors – that as long as SRI funds are sufficiently diversified, they have no reason to perform poorly.19 In other words, it is entirely possible to align financial investments and moral convictions. SRI funds do not perform better or worse than others, financially speaking.

This theoretical result can easily be verified empirically. Figure 4 shows the Aspi Eurozone and the Dow Jones Euro Stoxx indexes, with the first index constructed using 120 stocks out of the 320 that comprise the second. As financial theory predicts, there are no significant differences in performance between the two indexes and this should come as no surprise. In fact, SRI indexes are very similar to traditional ones, especially when they are of the best-in-class type. The stocks that make up the Aspi Eurozone, one of the principal European SRI indexes, account for 75% of the DJ Euro Stoxx stock market capitalization. Furthermore, no less than 45 of the 50 biggest firms in the euro zone (which represent 93% of the DJ Euro Stoxx 50 stock market capitalization), are included in the Aspi!20 The correlation coefficient between the returns of the two indexes is therefore naturally very nearly one.

318

Figure 4. The Aspi Eurozone and Dow Jones Euro Stoxx indexes

0

100

200

300

400

500

600

700

800

Aspi Eurozone

DJ Eurostoxx

Note: The indexes (dividends included) are expressed in base 100 as of January 1st 1992. On the top left is the scatter plot of the logarithmic returns of the two stock market indexes. The straight line corresponds to a linear regression with an estimated slope of 1.03 and a R² equal to 98%.

Capelle-Blancard, Laaradh and Monjon (2009) lists more than forty academic papers on SRI fund performance published between 1992 and 2008, all of which are very uniform in their methodology (see Table C in the Appendix). More recent studies have access to larger samples and use more sophisticated ways of measuring performance, but the main conclusion of the vast majority of them is that SRI fund performance is no better or no worse than that of traditional funds.

The results of these empirical studies suggest that, on the whole, the financial performance of SRI funds is comparable to that of conventional funds. However, as suggested by several recent research studies (King and Lenox, 2001; Lanoie, Ambec and Scott, 2007), the relevant question is not “Does it pay to be good?” but rather “When does it pay to be good?” It is in this spirit that Capelle-Blancard et al. (2009) examined the performance of SRI funds in France. Firstly, this study examines the financial performance of SRI funds marketed in France over the period from 2000 to 2007. A wide range of performance yardsticks (including conditional measurements) is used. Secondly, in contrast to previous studies, SRI funds according to their orientation, e.g. social, environmental, corporate governance, etc., as well as the quality of the SRI management systems, e.g. diversity and appropriation of sources, SRI selectivity, the SRI management process, transparency and pedagogical aspects are identified. Then an econometric estimate of the determinants of SRI financial performance is carried out. The findings show that average SRI fund performance is not significantly different from that of the market and this confirms the results of previous studies. Above all, the results indicate that neither the funds’ thematic orientation, nor the quality of management systems has any effect on financial performance. On the other hand, the financial performance of funds that operate by excluding certain sectors is significantly poorer than others.

y = 1,03x - 0,00

R² = 0,98

-0,10

-0,05

0,00

0,05

0,10

-0,10 -0,05 0,00 0,05 0,10

319

4. SRI: a new form of regulation?

Over and beyond the issue of whether SRI funds perform better or worse than traditional investments, SRI’s impact on firms should be addressed. Two questions are often asked. Can SRI serve as a substitute for political action? Is SRI in a position to get firms to reorient their strategy to better respond to stakeholders’ interests?

To answer the first question, we will use a simple graphic demonstration coming from Statman (2000). Figure 5 shows the supply (S) and demand (D) for capital. Beginning with the intersection of curves S₁ and D₁ a comparison can be made between: (a) the effect of SRI and (b) the effect of a tax on the cost of capital, the level of the firm’s investment and the profitability for shareholders.

For “non-virtuous” firms, SRI translates into an increase in the cost of capital associated with the leftward movement of the capital supply curve (S₁ towards S₂). Consequently, some investment projects will have to be abandoned (going from level I₁ to I₂), but at the same time expected profitability for the investors who invest in the firm increases (from R₁ towards R₂). 21

Figure 5. Compared effects of SRI and political action on the cost of capital for a polluting firm.

a) Effect of investment action

b) Effect of political action (rules, taxes)

Source: Staman (2000).

If the government enforces a tax on projects that pollute, for example, this will also translate into an increase in the cost of capital at the same time as a decrease in the demand for capital (from D₁ towards D₂) and a fall in the level of investment (from level I₁ to I₂). In this case, an increase in expected profitability is not anticipated. It is the government that benefits from the surplus (i.e. the difference between C₂ and R₂) via taxation.

In the model above, the implicit assumption is that firms do nothing to gain the approval of SRI investors. Heinkel, Kraus and Zechner (2001) examine the opposite case where firms are able to improve their production technology, but this change is costly. Firms do not, however, adopt “responsible” behavior unless it allows them to attract enough investors to lower the

320

cost of capital significantly. In this context, everything depends on the share of SRI. The authors consider that at least a quarter of the funds must be managed according to SRI principles for firms to have an incentive to modify their production processes.

Assessing the impact of SRI investors on the cost of capital is rather complex. Empirically, however, their impact can be fairly well estimated by studying the reaction of stock markets when indexes are redefined. Is belonging to an ethical index rewarded by the market? Is being excluded from such an index perceived negatively by investors?

Between 2000 and 2005, the DJSI, FTSE4Good and Aspi indexes changed their composition 25 times, with over 800 stocks being awarded the “socially responsible” label. At the same time, about 400 stocks that no longer met the criteria were excluded. Using event study methodology, we show that announcing these changes did not have any significant impact on the market price of the firms’ stock (Capelle-Blancard and Couderc, 2009). On the other hand, when the indexes are effectively redefined (some time after redefinition is announced), significant though temporary pressure can be seen on prices, a sign that ethical indexes can be used as benchmarks and are replicated by certain funds.

5. Conclusion: Maybe not such a good idea?

Doing well by doing good: let us hope so. This article focuses on the problems and challenges involved in socially responsible investment. The most important issue, if not the most complex, is conceptual: how is corporate social responsibility to be defined? Maximizing the share value is perhaps too reductive, but we do not (yet) have any other convincing alternatives. Without calling for a complete reworking of corporate finance theory, we suggest three fields of research that warrant further study.

Exclusion or Best-in-Class? There is quite a wide consensus in continental Europe on SRI: the most interesting and innovative approach is to select firms by the best-in-class model rather than to avoid certain sectors of activity. This approach is defended in particular by Landier and Nair (2008) and in fact has one major merit: it allows fund managers to take advantage of diversification. This is true to such an extent, however, that SRI can no longer even be distinguished from traditional funds! Basically, the problem is that ad hoc rules and the multiplication of criteria are detrimental to the consistency of strategy. On the other hand, exclusion funds have the merit of simplicity, even though this may mean poorer performance. The same is true for solidarity funds.

Is the social responsibility of SRI funds to outperform the market? Among the various publications on SRI, those dealing with fund performance are by far the most numerous.22 Advocates of SRI usually consider that good financial performance is likely to promote SRI. But is this really the objective? Is there not an element of delusion here? Is it not “fairer” to put the emphasis on the financial sacrifices imposed by the ethical approach? Whatever the case may be, this is not quite so simple theoretically: monetary incentives can have many perverse effects, in particular that of discouraging “pro-social” behavior (cf. Bénabou and Tirole, 2006). Indeed, altruism, reputation or self-esteem can be the powerful motives which lead people to be socially responsible investors. In this respect, it would be interesting to study the effect of SRI fund performance on fund inflows.23

What justifies market sanctions? The firms excluded by SRI fund managers are often already subjected to constraining legislation, while the sectors supported by proponents of SRI are often largely subsidized. At first glance, this is normal to the extent that SRI investors as well as legislators attempt to improve collective welfare. This raises problems of overall consistency, however: should steel industry firms that emit huge quantities of CO2 be

321

excluded from SRI funds? The answer is yes, if climate change is believed to pose a serious threat to the ecosystem. These same firms already pay for their emissions; negative externalities are now internalized due to the emission-permit trading market. So what justifies their being subjected to a double penalty? Is it the failure of the (permit) market that was tailor-made to correct the failures of another (steel production) market? It is perfectly admissible for SRI to exert additional pressure on firms, in the form of a new “market discipline”, but in this case it is necessary to ensure that SRI does not become a substitute for state intervention.

322

Endnotes

1. The expression Sustainable and Responsible Investment is also more and more used.

2. There are several other forms of investing that attempt to conciliate finance and sustainable development,

such as solidarity or community funds (designed to finance environmental or social projects), charity mutual

funds (part of the earnings revert to humanitarian or charitable associations), and microfinance funds (which

make loans to microfinance institutions). We only examine SRI here, which – quantitatively at least – is the

prevailing form in industrialized countries.

3. This explanation of origins arguably has an arbitrary side to it. Setting the date back in the 1920s might imply

that reflection on the relations between religion and finance had been non-existent before then. This is not so:

all religions have long preached in favor of ethics in finance. See, for example, the position of the Catholic

Church and Islam on the interest charged on loans.

4. It should come as no surprise to find the big three put forward by Alfred Hirschman in Exit, Voice and Loyalty.

5. Table A in the Appendix shows the principal regulations in force in industrialized countries.

6. The financial information market is naturally a concentrated one. Among the main European agencies, Vigéo,

Innovest, SAMet and Eiris should be mentioned.

8. In Canada, Broad SRI totaled 503.4 billion Canadian dollars (about €334 billion) in 2006, 11% of which for the Core. In Australia and in New Zealand, Broad SRI accounted for 72.2 billion Australian dollars (about €41 billion) in 2007, 27% of which for the Core. In Japan, Broad SRI represented 840 billion yen as of September 31, 2007 (about €5.5 billion).

9. In the United States, SRI funds made up 3.2% of total overall funds at the end of 2007, which means that

they are smaller than traditional funds on the average: SRI fund AUM amount to $776 million, i.e. about half as

much as traditional funds.

10. We do not have any overall data for Europe, but to take the case of France, there were 175 SRI funds

marketed by 48 fund management firms at the end of 2007, for total assets under management of €10 billion.

There were 40 funds in 2001 and only 7 in 1997 (source: Novethic). The first SRI fund came into being in France

in 1983 thanks to Sister Nicole Reille of the “Notre Dame” Congregation. Today all big banks offer their

customers “ethical” investments, in their great majority based on the best-in-class approach. Despite this

dramatic development, SRI represents about 1% of the total assets under management.

11. European data are not consistent from one survey to the other: the first investigation involved 8 European

countries, the second 13 countries and the third 15.

12. Jane Ambachtsheer, 2005, “SRI: What Do Investment Managers Think?”, Mercer.

13. The term refers to the theory of “disembedment” put forward by Karl Polanyi (The Great Transformation,

1944).

14. For a formal presentation, see Tirole (2001, 2006). See also Moore and Rebérioux (2007) for a critical

approach to the debate on the shareholder theory.

15. Table B in the Appendix presents the principal filters and criteria that are usually selected.

16. See Margolis and Walsh (2003), Orlitzky, Schmidt and Rynes (2003), Laroche and Allouche (2008) for an

overview.

17. On the history of modern finance see Peter L. Bernstein (Capital Ideas, 1995). John Maynard Keynes was

himself very critical of the diversification principle: “I am in favor of having as large a unit as market conditions

will allow (…) To suppose that safety-first consists in having a small gamble in a large number of different

[companies] where I have no information to reach a good judgment, as compared with a substantial stake in a

company where one’s information is adequate, strikes me as a travesty of investment policy”. J.M. Keynes,

1983, The Collected Writing of John Maynard Keynes, Vol. XII, texts collected by D. Mogridge, New York,

Cambridge University Press.

18. Moreover, the development of SRI can be seen as a reaction by investors to passive management, which is impersonal by nature.

323

19. The marginal gains associated with diversification decrease very quickly as the number of stocks included in

the portfolio increase. Depriving the portfolio of some stocks is therefore not detrimental (except possibly if

these stocks are negatively correlated with the stock market). See Bello (2005).

20. In comparison, half of the firms that make up the S&P 500 index are also listed in the Domini 400 index

(Statman, 2006).

21. In other words, the strategy of “vice funds” (which are diametrically opposed to the SRI approach and

invest specifically in activities connected to tobacco, gambling, arms manufacture, etc.) should theoretically be

profitable. The provocative nature of these funds has succeeded in attracting media attention. In this sense,

the promoters of such funds have attained their objective, but assets under management are still very small.

See Chong, Her and Philips (2006) and Hong and Kacperczyk (2007).

22. This profusion of research has a somewhat surprising outcome, given how much the responses converge.

However, the feasibility of the study, i.e. the fact that the subject lends itself to econometric tests and that

data are available, etc., evidently plays a major role in the choice of questions that are dealt with.

23. See the empirical study by Benson and Humphrey (2008) on the determinants of investment flows to SRI

funds.

324

References

[1] Barnett, M.L., & R.M. Salomon, 2006, Beyond dichotomy: The curvilinear relationship between social responsibility and financial performance, Strategic Management Journal, 27(11): 1101-1122.

[2] Bauer, R., J. Derwall & R. Otten, 2007, The Ethical Mutual Fund Performance Debate: New Evidence from Canada, Journal of Business Ethics, 70(2): 111-124.

[3] Bauer, R., K. Koedijk & R. Otten, 2005, International Evidence on Ethical Mutual Fund Performance and Investment Style, Journal of Banking and Finance, 29(7): 1751-1767.

[4] Bauer, R., R. Otten & A. Tourani Rad, 2006, Ethical Investing in Australia: Is There a Financial Penalty?, Pacific-Basin Finance Journal, 14(1): 33-48.

[5] Bello, Z., 2005, Socially responsible investing and portfolio diversification, Journal of Financial Research, 28 (1): 41-57.

[6] Bénabou, R., & J. Tirole, 2006, Incentives and Prosocial Behavior, American Economic Review, 96(5): 1652-1678.

[7] Benson, K., & J. Humphrey, 2008. Socially responsible investment funds: Investor reaction to current and past returns, Journal of Banking and Finance, 32(9): 1850-1859.

[8] Capelle-Blancard, G. & N. Couderc, 2009, The impact of socially responsible investing? Evidence from stock index redefinitions, Journal of Investing, 18(2), 76-86.

[9] Capelle-Blancard, G., K. Laaradh & S. Monjon, 2009, The impact of the screening process on the performance of ethical funds, Working Paper, Université Paris 1 Panthéon-Sorbonne.

[10] Chong J., M. Her & G.M. Philips, 2006, To sin or not to sin? Now that’s the question, Journal of Asset Management, 6(6): 406–417.

[11] DiBartolomeo, D. & L. Kurtz, 1999, Managing Risk Exposures of Socially Screened Accounts, Northfield Working Paper.

[12] Galbtraith, J.K., 1967, The New Industrial State, Princeton University Press.

[13] Gregory, A., J. Matatko, & R. Luther, 1997, Ethical Unit Trust Financial Performance: Small Company Effects and Fund Size Effects, Journal of Business Finance and Accounting, 24: 705-725.

[14] Grossman, B. & W. Sharpe, 1986, Financial Implications of South Africa Divestment, Financial Analysts Journal, July/August.

[15] Guerard, J.B. Jr., 1997. Is there a cost to being socially responsible in investing? Journal of Investing, 6(2): 11-18.

[16] Hamilton, S., H. Jo & M. Statman, 1993, Doing Well While Doing Good? The Investment Performance of Socially Responsible Mutual Funds, Financial Analysts Journal, 49 (6): 62-66.

[17] Hart, O., 1995a, Corporate Governance: Some Theory and Implications, The Economic Journal, 105: 678-689.

[18] Heinkel, R., A. Kraus & J. Zechner, 2001, The Effect of Green Investment on Corporate Behavior, The Journal of Financial and Quantitative Analysis, 36(4): 431-449.

[19] Hong, H. & M. Kacperczyk, 2007, The Price of Sin: The Effects of Social Norms on Markets, Working Paper, Princeton University.

[20] Karpoff, J., 2006, The Impact of Shareholder Activism on Target Companies: A Survey of Empirical Findings, Working Paper, University of Washington.

[21] King, A. & Lenox, M. 2001. Does it Really Pay to be Green? The Journal of Industrial Ecology, 5(1): 105-116.

[22] Kreander, N., G.H. Gray, D.M. Power & C.D. Sinclair, 2005, Evaluating the Performance of Ethical and Non-SRI Funds: A Matched Pair Analysis, Journal of Business, Finance and Accounting, 32(7): 1465-1493

[23] Kurtz, L., 1997, The Impact of Social Screening on Growth-Oriented Investment Strategies, The Journal of Performance Measurement, Spring 1997.

[24] Landier, A. & V.B. Nair, 2008, Investing for Change: Profit from Responsible Investment, Oxford University Press.

[25] Lanoie, P., S. Ambec & I. Scott, 2007, When and Why Does it Pay to be Green? Working Paper, Cirano.

[26] Laroche, P. & J. Allouche, 2008, Corporate Social Responsibility and Firm Performance: A Meta-Analysis, Working Paper.

[27] Luck, C & N. Pilotte, 1993, Domini Social Index performance, Journal of Investing, Fall.

325

[28] Luther, R., J. Matatko & D. Corner, 1992, The Investment Performance of UK Ethical Unit Trusts, Accounting, Auditing & Accountability Journal, 5(4), 57-70.

[29] Merton R.K., 1936, The Unanticipated Consequences of Purposive Social Action, American Sociological Review, 1(6), 894-904.

[30] Margolis, J. & J. Walsh, 2003, Misery loves company: Rethinking social initiatives by business, Administrative Science Quarterly, 48: 268-305.

[31] Markowitz, H., 1952, Portfolio selection. The Journal of Finance, 7(1): 77-91.

[32] Moore, M., & A. Rebérioux, 2007, « The Corporate Governance of the Firm as an Entity: Old Issue for the New Debate », in Y. Biondi, A. Canziani et T. Kirat (eds.), The Firm as an Entity: Implications for Economics, Accounting, and Law, Routledge: 348-374.

[33] Moskowitz, M., 1972, Choosing Socially Responsible Stocks, Business and Society Review, 10: 71-5.

[34] Orlitzky, M., F.L. Schmidt, & S.L. Rynes, 2003, Corporate Social and Financial Performance: A Meta-analysis, Organization Studies, 24(3): 403-441.

[35] Porter, M.E. & C. van der Linde, 1995, « Towards a new conception of the environment-competitiveness relationship », Journal of Economic Perspectives, 9(4): 97-118.

[36] Porter, M.E. & C. van der Linde, 1995, Towards a new conception of the environment-competitiveness relationship, Journal of Economic Perspectives, 9(4): 97-118.

[37] Rudd, A., 1979, Divestment of South African equities: How risky? Journal of Portfolio Management, 5(3): 5.

[38] Schröder, M., 2006, Is there a Difference? The Performance Characteristics of SRI Equity Indices, Journal of Business Finance and Accounting, 34(1-2): 331-348.

[39] Shleifer, A. & R. Vishny, 1997, A Survey of Corporate Governance, The Journal of Finance, 52: 737-783.

[40] Statman, M. 1987. How many stocks make a diversified portfolio? Journal of Financial and Quantitative Analysis, 22(3): 353-363.

[41] Statman, M., 2000, Socially Responsible Mutual Funds, Financial Analysts Journal, 56 (3): 30-39.

[42] Statman, M., 2006, Socially Responsible Indexes: Composition, Performance, and Tracking Errors, Journal of Portfolio Management, Spring.

[43] Teoh, S.H, I. Welch, & C. P. Wazzan, 1999, The Effect of Socially Activist Investment Policies on the Financial Markets: Evidence from the South African Boycott, Journal of Business, 72(1): 35-89.

[44] Tirole, J., 2001, Corporate Governance, Econometrica, 69(1): 1-35.

[45] Tirole, J., 2006, The Theory of Corporate Finance, Princeton University Press, 640 pages.

326

Appendix

Table A. SRI regulations

Country SRI related regulations

Australia In a 2001 bill it is stated that all investment firms’ product disclosure statements should include a description of “the extent to which labor standards or environmental, social or ethical considerations are taken into account”.

Since 2001, all listed companies on the Australian Stock Exchange are required to make an annual social responsibility report.

Belgium In 2001, Belgium passed the ‘Vandebroucke’ law, which requires pension funds to report the degree to which their investments take into account social, ethical and environmental aspects.

France In May 2001, the legislation “New Economic Regulations” came into force requiring listed companies to publish social and environmental information in their annual reports.

Since February 2001 managers of the Employee Savings Plans are required to consider social, environmental or ethical considerations when buying and selling shares.

Germany Since 1991, the Renewable Energy Act gives a tax advantage to closed-end funds to invest in wind energy.

Since January 2002, certified private pension schemes and occupational pension schemes ‘must inform the members in writing, whether and in what form ethical, social, or ecological aspects are taken into consideration when investing the paid-in contributions’.

Italy Since September 2004 pension funds are required to disclose non-financial factors (including social, environmental and ethical factors) influencing their investment decisions.

Netherlands In 1995, the Dutch Tax Office introduced a ‘Green Savings and Investment Plan’, which applies a tax deduction for green investments, such as wind and solar energy, and organic farming.

Sweden Since January 2002, Swedish national pension funds are obliged to incorporate environmental and ethical aspects in their investment policies.

UK In July 2000, the Amendment to 1995 Pensions Act came into force, requiring trustees of occupational pension funds in the UK to disclose in the Statement of Investment Principles “the extent (if at all) to which social, environmental and ethical considerations are taken into account in the selection, retention and realization of investments”.

The Trustee Act 2000 came into force in February 2001. Charity trustees must ensure that investments are suitable to a charity’s stated aims, including applying ethical considerations to investments.

In 2002, The Cabinet Office in the UK published the Review of Charity Law in 2002, which proposed that all charities with an annual income of over £1 m should report on the extent to which social, environmental and ethical issues are taken into account in their investment policy. The Home Office accepted these recommendations in 2003.

The Association of British Insurers (ABI) published a disclosure guideline in 2001, asking listed companies to report on material social, environmental and ethical risks relevant to their business activities.

US Section 406 of the Sarbanes-Oxley Act, which came into effect in July 2002, requires companies to disclose a written code of ethics adopted by their CEO, chief financial officer and chief accountant.

This table summarizes the regulatory initiatives regarding SRI taken by national government in western countries. Source : Renneboog, Horst, Zhang (2008).

327

Table B. SRI screens

Screens Definitions

Tobacco Avoid manufacturers of tobacco products Alcohol Avoid firms that produce, market, or otherwise promote the consumption of alcoholic beverages Gambling Avoid casinos and suppliers of gambling equipment Defense/weapons Avoid firms producing weapons for domestic or foreign militaries, or firearms for personal use Nuclear power Avoid manufacturers of nuclear reactors or related equipment and companies that operate

nuclear power plants Irresponsible foreign operations

Avoid firms with investments in government-controlled or private firms located in oppressive regimes or firms which mistreat the indigenous peoples of developing countries

Pornography/adult entertainment

Avoid publishers of pornographic magazines; production studios that produce offensive video and audio tapes; companies that are major sponsors of graphic sex and violence on television

Abortion/birth control Avoid providers of abortion; manufacturers of abortion drugs and birth control products; insurance companies that pay for elective abortions (where not mandated by law); companies that provide financial support to Planned Parenthood

Labor relations and workplace conditions

Seek firms with strong union relationships, employee empowerment, and/or employee profit sharing. Avoid firms exploiting their workforce and sweatshops

Employment diversity Seek firms pursuing an active policy related to the employment of minorities, women, gays/lesbians, and/or disabled persons who ought to be represented amongst senior management

Corporate governance Seek companies demonstrating “best practices” related to board independence and elections, auditor independence, executive compensation, expensing of options, voting rights and/or other governance issues. Avoid firms with antitrust violations, consumer fraud, and marketing scandals

Business practice Seek companies committed to sustainability through investments in R&D, quality assurance, product safety

Human rights Seek firms promoting human rights standards. Avoid firms which are complicit in human rights violations

Environment Seek firms with proactive involvement in recycling, waste reduction, and environmental cleanup. Avoid firms producing toxic products, and contributing to global warming

Animal testing Seek firms promoting the respectful treatment of animals. Avoid firms with animal testing and firms producing hunting/trapping equipment or using animals in end products

Renewable energy Seek firms producing power derived form renewable energy sources Biotechnology Seek firms that support sustainable agriculture, biodiversity, local farmers, and industrial

applications of biotechnology. Avoid firms involved in the promotion or development of genetic engineering for agricultural applications

Community involvement Seek firms with proactive investments in the local community by sponsoring charitable donations, employee volunteerism, and/or housing and educational programs

Shareholder activism The SRI funds that attempt to influence company actions through direct dialogue with management and/or voting at Annual General Meetings

Non-married Avoid insurance companies that give coverage to non-married couples Healthcare / pharmaceuticals

Avoid healthcare industries (used by funds targeting the “Christian Scientist” religious group)

Interest-based financial institutions

Avoid financial institutions that derive a significant portion of their income from interest earnings (on loans or fixed income securities). (Used by funds managed according to Islamic principles)

Pork producers Avoid companies that derive a significant portion of their income from the manufacturing or marketing of pork products. (Used by funds managed according to Islamic principles)

This table summarizes the investment screens used by SRI mutual funds. Source: From Renneboog, Horst, Zhang (2008). Data are compiled from Social Investment Forum (2003, p. 42) and the Natural Capital Institute (www.responsibleinvesting.org).

328

Table C. SRI mutual funds performance: A survey of empirical studies

Authors Sample # SRI funds (non-SRI)

Model Main Result

Details

Luther et al. (1992) UK

(1984-90) 15 (0)

CAPM ns Average αSRI is 0.03%. Small cap bias.

Hamilton et al. (1993) US

(1981-90) 32

(320) CAPM ns Average αSRI (αnon-SRI) is -0.06% (-0.14%) before 1985 and -0.28% (-0.04%) after.

Luther & Matatko (1994)

UK (1984-92)

9 (0)

CAPM ns Average αSRI and αnon-SRI are not different from 0 in average. Small cap bias.

Mallin et al. (1995) UK

(1986-93) 29

(29) CAPM ns αSRI (αnon-SRI) ranges from -0.28% to 1.21% (-0.41% to 1.56%).

Gregory et al. (1997) UK

(1986-94) 18

(18) 2-factors ns αSRI (αnon-SRI) ranges from -0.71% to 0.24% (-0.40% to 0.51%). Small cap bias.

Sauer (1997) US

(1991-94) 110 (0)

CAPM ns α of the Domini Social Equity mutual fund is -0.12% or 0.02% according to the benchmark (Vanguard Index Extended or VI 500).

Reyes & Grieb (1998) US

(1986-95) 15

(15) Cointegration ns SRI and Non-SRI funds monthly returns are not co-integrated.

Goldreyer et al. (1999) US

(1981-97) 49

(180) CAPM ns

Average αSRI (αnon-SRI) is -0.04% (0.23%). SRI funds using positive screens consistently outperform those without (α is -0.01% and -0.07%, respectively).

Havemann & Webster (1999)

UK (1988-98)

15 (15)

Stat. ns Lower risk and lower return.

Cummings (2000) Australia (1986-94)

7 (0)

3-factors ns αSRI ranges from -0.6% to 0.2%. Older trusts (established in 1986) outperformed the other trusts.

Statman (2000) US

(1990-98) 31

(62) CAPM ns

Average αSRI (αnon-SRI) is -0.42% (-0.62%). Average αSRI is -0.48% when using DSI as benchmark.

Foerster & Asmundson (2001)

Canada (1995-99)

6 (0)

CAPM ns Financial performance is not significantly different from market benchmark (TSE300).

Otten & Koedijk (2001) Netherlands (1994-00)

4 (4)

CAPM ns Financial performance is not significantly different from non-SRI funds.

Tippet (2001) Australia (1991-98)

3 (0)

CAPM – The average of the three largest Australian ethical mutual funds significantly under-performed the All Ordinaries index by 1.5% per year.

Turcotte et al. (2001) France

(1994-98) 7

(0) Stat. ns Financial performance is not significantly different from market benchmark (CAC 40).

Benjaminson & Westerdahl (2002)

Sweden (1999-02)

9 (0)

CAPM ns αSRI ranges from -0.046% to 0.058%. Older funds seem to perform better than funds launched more recently.

329

Authors Sample # SRI funds (non-SRI)

Model Main Result

Details

Plantinga & Scholtens (2002)

Belg. Fra. Nld. (1994-00)

_ (784)

Style ns Most funds have a significant exposure to the SRI index.

Young & Proffitt (2003) US

(2000-03) 32 (0)

Stat. ns Financial performance is not significantly different from market benchmark (Morningstar).

Burlacu et al. (2004) US

(1997-02) 50

(1688) CAPM ns

Average αSRI (αnon-SRI) is -0.03% (-0.04%). Neither type of fund displayed any ability to time the market.

Miglietta (2004) Europe

(1996-04) 65 (0)

FF – Underperformance. Small cap bias.

Muñoz et al. (2004) Spain

(2000-02) 12 (0)

CAPM ns αSRI ranges from -0.57% to 0.04% (FTSE4Good as benchmark) and -0.32 to 0.11%. 9 out of the 12 are negative.

Schröder (2004) US, Ger. & Swtz.

(1990-02) 46 (0)

FF ns αSRI (αnon-SRI) ranges from -2.06% to 0.87% (-0.41% to 1.56%). 38 out of the 46 are negative.

Bello (2005) US

(1994-01) 42

(84) CAPM ns

αSRI ranges from -0.87% to 0.99% (DSI as benchmark) and from -0.91% to 1.08% (S&P 500 as benchmark). Average αSRI (αnon-SRI) is -0.09% (-0.17%) with DSI as benchmark and -0.10% (-0.16%) with S&P 500 as benchmark.

Kreander et al. (2005) Europe

(1996-98) 40

(40) CAPM ns

Average αSRI (αnon-SRI) is 0.20% (0.12%). Neither type of fund displayed any ability to time the market.

Bauer et al. (2005) Ger., UK & US

(1990-01) 103

(4384) FFC ns

Average αSRI is 0.29%, 0.09% and -0.05% for Germany, UK domestic and US domestic funds. Higher expense ratio for SRI funds. Small cap bias and growth orientation.

Gregory & Whittaker (2005)

UK (1989-02)

32 (5)

FFC ns Average αSRI (αnon-SRI) is -0.1% (-0.1%) for domestic and international funds. For domestic funds, past ‘winning’ SRI funds outperform ‘losing’ SRI funds to a greater extent than their control portfolio counterparts.

Scholtens (2005) Netherlands (2001-03)

12 (0)

CAPM ns The performance differential between SRI and non-SRI funds is not statistically significant. SRI funds seem to be tilted toward value stocks.

Barnett & Salomon (2006)

US (1972-02)

61 (0)

CAPM U-shaped

As the number of social screens used by an SRI fund increases, financial returns decline at first, but then rebound as the number of screens reaches a maximum. Community relations screening increased financial performance, but environmental and labor relations screening decreased financial performance.

Bauer et al. (2006) Australia (1992-03)

25 (281)

FFC ns Average αSRI (αnon-SRI) is -2.17% (-0.61%) for domestic and -1.42% (-4.40%) for international funds.

Chong et al. (2006) US

(2002-05) VICEX

(DSEFX) CAPM-ARCH ns

The Vice Fund (VICEX) has outperformed both the Domini Social Equity Fund and the S&P500 Index (ΑVICEX = 8.64%), while the Domini Social Equity Fund has underperformed (Α = -0.84%, ns).

330

Authors Sample # SRI funds (non-SRI)

Model Main Result

Details

Geczy et al. (2006) US

(1963-01) 34

(860) > 4 factors ns

Average αSRI (αnon-SRI) is 0.21% (0.08%). The SRI constraint imposes large costs. Restricting the SRI universe to the funds that screen out “sin” stocks (e.g. alcohol, tobacco or gambling) increases the monthly cost of the SRI constraint by 10 basis points or more.

Lozano et al. (2006) Spain

(2000-03) 14 (0)

Stat. ns Financial performance is not significantly different from market benchmark.

Mill (2006) UK

(1996-04) 1

(3) FF ns

Examines the financial performance of a UK unit trust that was initially ‘‘conventional’’ and later adopted SRI principles. Mean risk-adjusted performance is unchanged by the switch to SRI.

Bauer et al. (2007) Canada

(1994-02) 8

(267) > 4 factors ns Average αSRI (αnon-SRI) is -0.21% (-0.18%).

Girard et al. (2007) US

(1984-03) 117 (0)

Style SRI funds have less diversification. SRI fund managers showed poor stock selection and market timing.

Jones et al. (2007) Australia (1986-05)

89 (9278)

FFC – SRI funds significantly underperform the market in Australia, particularly during the period 2000-2005. Average αSRI is -0.07% over the whole sample period and -0.12% in 2000-2005.

Scholtens (2007) Netherlands (2001-05)

7 (0)

FF ns α SRI ranges from -0.11% to 0.02%.

Gil-Bazo et al. (2008) US

(1997-05) > 61

(> 1100) FFC +

SRI funds may outperform their conventional peers, but only when they are operated by management companies specialized in the management of SRI funds.

Derwall & Koedijk (2008)

US (1987-03)

24 (5)

> 4 factors ns Average αSRI (αnon-SRI) is -1.08% (-1.28%) for pure bond funds and 0.11% (-1.25%) for balanced funds.

Fernandez-Izquierdo & Matallin-Saez (2008)

Spain 13

(2051) Style ns

Average αSRI (αnon-SRI) is -0.03% (-0.02%) with net returns and -0.01% (-0.04%) with gross returns.

Renneboog et al. (2008) 17 countries (1991-03)

440 (16036)

FFC – / ns

SRI funds in the US, the UK, and in many continental European and Asia-Pacific countries underperform their domestic benchmarks by -2.2% to -6.5%. In France, Japan and Sweden, the risk-adjusted returns of SRI and non-SRI funds are not statistically different. Moreover, SRI investors are unable to identify the funds that will outperform, whereas they show some ability in identifying funds that will perform poorly. Finally, corporate governance and social screens yield lower risk-adjusted returns.

331

SRI as Sustainable and Responsible Insurance

Responsible investing is worth a premium

Stéphane Voisin, Hubert Jeaneau

Abstract: This article addresses the questions surrounding the financial and extra-financial value

of Sustainable and Responsible Investment (SRI). Beyond the outperformance issue, it contends that SRI

benefits are essentially an insurance against ESG (Environmental, Social, Governance) risks. For the

investor, those risks are a source of market volatility that creates investment opportunities. Often

neglected in the debate, environmental and social dividends are important features of SRI, which also

drive the demand for SRI products.

Keywords: SRI, investment, performance, extra-financial, Environmental, Social, Governance

332

The outperformance myth has haunted the field of SRI investment from the outset. Although it has been perpetuated by recent academic debate, the myth has been killed off many times (Orlitzky and al. 2003, Allouche and Laroche 2005, Margolis and al. 2007). It is dead and should be buried once and for all in order for SRI to flourish. While "positive" performance is an open issue, outperformance is, in our view, a dead end, impossible to prove and moreover, it is easy to demonstrate the opposite. It is only by leaving aside the search for outperformance that Sustainable and Responsible Investing may finally find its own investment philosophy. One thing is clear, unlike investment biases such as "value" or "growth", its main objective is not to outperform: rather, it provides a "positive" extra-financial dividend in response to investors' extra-financial expectations and motivations.

One of the most surprising points in the debate on SRI performance is that only financial performance is actually discussed. Because SRI attempts to include environmental, social and governance factors in the investment process, the main considerations to be assessed in regard to performance should therefore be environmental and social. We therefore see the need to reframe the question of SRI performance, building on Cheuvreux Environmental Social and Governance (ESG) research experience and drawing on many years of academic research. And the answer comes quite easily: Socially Responsible Investing has a price because it is valuable, and whatever is better is worth a premium.

For instance, as a consumer, you are willing to pay a premium for organic food, fair trade goods, ethical services and so on. You should therefore be ready to pay the same premium for investing in a sustainable and responsible way. Moreover, studies show that customers expect to pay that premium for sustainable expenditure and that their willingness to do so vanishes when the premium disappears. This example is a good illustration of the conclusion of Cheuvreux’s research commitment to highlight and demonstrate the real value, definition and objective of SRI. We propose that environmental and social benefits have an intrinsic value that should be reflected in higher stock prices and, equally, there is a price to pay for this. If the responsible consumer and the responsible investor have anything in common, then a price premium is a selling point for a Sustainable and Responsible equity asset.

But there is much more to the argument of the fundamental underperformance of SRI. The most tangible and material characteristic of Corporate Social Responsibility (CSR) is often the protection against extra-financial risks (litigation, regulation, etc.). To express this at the investor level, the best definition for SRI investment might be: buying insurance against ESG risks. Picking a sustainable and responsible stock means hedging those risks. If the market prices the risks correctly, it indicates precisely the cost of this insurance premium. It also provides financial stability and security, which, in today's context of strong political pressure for more sustainable financial markets, is certainly not negligible.

It is no accident that the first winner of the Moskowitz Prize109 was very simply entitled: "Is there a cost to being socially responsible in investment?" (John B. Guerard, 1997). It was the moment to provide an answer, namely: Yes… but there are also substantial advantages.

109 Delivered by the Haas School of business, University of California at Berkeley, the Moskowitz prize is a global award recognizing outstanding quantitative research in the field of socially responsible investing

333

In this article, we reframe the question of SRI performance to address the more pragmatic and urgent issue of SRI positioning as an investment strategy. First, we posit that ESG materiality is, in essence, an insurance against extra-financial risks. We develop the idea that SRI means a lower risk profile and generally lower returns. Then, this article underlines the role of ESG research in identifying the value and optimizing the cost of that insurance. For the investor, the main challenge becomes seizing the market opportunities that SRI themes are likely to create. Finally, we contend that extra-financial reporting is a cornerstone for Sustainable and Responsible Investment.

1. The SRI financial benefit is essentially an insurance premium Academic research has now a fair grasp on the relationship between financial performance and corporate social responsibility. In this section, we review the academic evidence that quantitative analysis has failed to prove that doing good leads to doing well. To those global and quantitative approaches, Cheuvreux opposes a sector approach to assess the financial impact of carefully selected ESG issues. What are the implications for the SRI investor? We emphasize that a clear distinction should be made between a company’s performance and its attractiveness as an investment.

1.1 Academic research confirms the inadequacy of global approaches The quest for a correlation between financial and CSR performance (using accounting-based or market-based measures) has yielded mixed results. According to Margolis, Elfenbein and Walsh's (2007) thorough survey of 167 studies over 35 years: most studies conclude that the relationship is not significant (67%). 31% show a positive relationship and very few (2%) point to a negative relationship. The overall effect is positive, although not strong enough to allow any conclusions (the weighted average of r equals 0.10 in the Margolis et al. meta-analysis). This correlation becomes even smaller when the issue of causality is addressed (companies with strong financial performance might subsequently invest more in CSR). The specific subject of risk vs. performance has not been thoroughly examined, and many studies use risk-adjusted results (see figure 1).

Figure 1: Breakdown of the results of 167 studies on the performance of CSR over a 35-year period

334

Positive

relationship

31%

Negative

relationship

2%

Statistically

non

significant

67%

Source: Margolis, Elfenbein and Walsh, 2007

The probability of shifting the academic consensus with a new study is small. The results of meta-analysis are now fairly stable (see Orlitzky et al. 2003, Laroche and Allouche 2005, and Margolis and Walsh 2007), and there is less incentive to carry out a new global analysis. However, there is a sparkling interest to conduct a new study as soon as new data is available (see figure 2).

Figure 2: Number of studies on CSR and financial performance over the last 35 years: has the peak passed?

1

5 54

53

57

89

8

19

28

1514

6

16

9

0

5

10

15

20

25

30

1972-73

1974-75

1976-77

1978-79

1980-81

1982-83

1984-85

1986-87

1988-89

1990-91

1992-93

1994-95

1996-97

1998-99

2000-01

2002-03

2004-05

2006-07

Number of studies

Source: Margolis, Elfenbein and Walsh, 2007

Hope for this kind of exercise may lie in qualitative shifts in the methodology used (involving more complex models or relationships (Laroche and Allouche, 2005)) or in the gradual improvement in

335

the data and reporting on companies' extra-financial indicators. We do not really expect a miracle in the area of quantitative studies, mainly because the risks and issues at hand are far from homogeneously distributed across sectors, which is likely to constitute a long-term barrier to any kind of generalisation on this subject. Sector approaches are an appealing alternative to such global approaches and, in Cheuvreux’s experience, they yield far more tangible results.

1.2 Cheuvreux’s methodology

The materiality of an ESG issue is therefore neither systematic nor easy to assess. The answer to the so-called missing link is so simple, indeed almost tautological, that one can wonder why the question is still being asked. Namely, CSR analysis and ratings alone cannot yield financial results; they have to be combined with and enriched by the expertise of financial analysts dedicated to their sectors, in order to understand the financial consequences of CSR policies (Crifo, Ponssard, 2008). There is, of course, no free lunch: this kind of analysis requires a complex, qualitative and resource-consuming process.

Almost always, the ESG dividend has a short-term cost, and it is sometimes difficult to tell whether it will pay in the long run: risks, by definition, may not materialise. Nevertheless, ESG issues represent a risk for unprepared companies that has to be discounted and a business opportunity for the fortunate few.

Working under those hypotheses, Cheuvreux designed a materiality protocol to assess Environmental and Social risks (see Table 1). Intangible risk analysis is the right arm of extra-financial research (the left arm being intangible capital analysis, which we discuss in detail below). Cheuvreux has developed a holistic methodology to identify the nature of the risks that can bring to bear a financial materiality and to determine how they can be relevant to a company. This protocol approach in theory meets all types of environmental, social and public health issues. It does not, however, apply to most governance issues.

Table 1 : Cheuvreux’s methodology to assess extra-financial risks

CA Cheuvreux’s Materiality Protocol

1. HARD LAW Market mechanisms, environmental taxes, banning

E.g.: Bonus/penalty taxes for cars

2. SOFT LAW New standards, international agreements, European targets

3. LIABILITIES Risk exclusion, fee increases, antitrust actions, class actions

E.g.: Paying back taxes after fiscal evasion

4. PHYSICAL Natural catastrophes, climate stress, damages, social tensions

5. REPUTATION Brand attraction, boycott

336

5. SOCIOLOGICAL SHIFT Consumer shift, pricing power: the strongest drivers?

Source : Cheuvreux Research

To better illustrate Cheuvreux’s methodology, let us give two examples. The first one is the analysis of the implementation of a carbon constraint to mitigate emissions and adapt to climate change is emblematic of our approach. The objective is to assess the financial impact and economic shift on all sectors and companies exposed. As CO2 has a price fixed on a market, it is feasible to evaluate the impact of the carbon constraint on companies. This cost is far from negligible for energy-intensive sectors: Cheuvreux estimate that CO2 costs will reach EUR250bn over 2013-2020 (EUR35bn p.a.). This can significantly reduce margins, as well as being a source of short-term windfall profits, even for the most energy-intensive companies. A few sectors will bear the bulk of the constraint: electric utilities, heavy industries (steel, cement, pulp and paper), etc. Another example concerns the pharmaceutical sector. Cheuvreux raised the question of whether aggressive tax planning by most pharmaceutical majors is sustainable. There is indeed a risk of a backlash, given the positive social externalities benefiting the sector (notably in the form of public funds, direct public purchases, clinical trials and state support for research). These risks can visibly materialise, as when GlaxoSmithKline faced the largest payment of back taxes in IRS history, after trying to shift profits out of the US.

Most of the time, a company foregoes higher returns to protect itself from social and regulatory liabilities, e.g., in the pharmaceuticals industry, by paying taxes correctly, or, for heavy industries, by having a low-carbon energy mix. These companies have, in this regard, a less risky profile, and ceteris paribus, display more stable cash flows. For most sectors and for a few carefully selected issues, ESG topics have a way of materialising and impacting the margins of a company – and possibly its investors.

1.3 A premium necessarily means a price for it

Can an investor benefit from that insurance premium? Can the lower CSR risk profile be cashed in by SRI funds? A general answer might be to trust the market to put the right price on a stock, and more specifically, on the SRI insurance premium.

In addition, CSR performance and SRI fund performance are two different issues that should be treated separately, and including them jointly in a meta-analysis makes little sense. How well a company performs and its attractiveness for the investor are largely unrelated questions. Everything depends on the share price. As Warren Buffett (Lowenstein, 1995) puts it, the magic of the stock market is that sometimes you can buy a dollar bill for 40 cents. But the opposite may also be true! It does not really matter how well a company performs if you pay far too much to own the stock.

337

On the hypothesis of market efficiency, there are no grounds for believing that SRI should be a pocket of higher performance. If indeed SRI stocks are insurance against a range of downside risks, such SRI issues are likely to be priced by the market. And academics generally support this view, with the hypothesis of market eco-efficiency. This is most obvious in the European utility sector, which fully priced, as soon as 2006, the carbon constraint (see figure 3).

Figure 3: Evidence from the carbon market. Once the potential materiality of a SRI issue is identified, the market applies a premium to the best positioned stocks

EdisonEnelEndes a

Union Fenos aEDP

E.ON

RWE

Electrabel

Iberdrola EDF

Verbund

Fortum

5

6

7

8

9

10

11

12

13

14

02004006008001000

CO2-intens i ty 2006 (kgCO2/MWh)

EV

/EB

ITD

A 2

00

8

Source: Cheuvreux, Carbon Research, carbon factor before crisis

For an investor, outperformance often comes from stocks that are temporarily ignored by the market despite their quality… until the market finds out and corrects it. SRI stocks benefit from a special interest from SRI investors, but they are also under the scrutiny of traditional investors.

Conversely, it is the sectors shunned by ethical investors (alcohol, tobacco and gaming) that have historically, and logically, outperformed the market (in the US, Hong and Kacperczyk, 2006). According to this study, the valuation of so called sin stocks suggests neglect. Their price/book ratios are, on average, 16% lower than those of other companies (after adjusting for characteristics such as ROE and R&D / Sales) for the 1962-2003 time period.

In the long term, the efficiency of the market ensures that ESG issues cannot be a sustained pocket of outperformance. As evidenced by the utility sector (figure 3), the market eventually corrects such anomalies. On the other hand, there are few reasons to believe that SRI, which openly displays longer-term performance objectives, can correct structural biases in the financial markets, such as market myopia (cf. on the subject the field of behavioural finance). SRI is indeed subject to the general constraints of the financial industry in terms of performance appraisal.

The constraints applying to SRI fund managers are, from that standpoint, very similar to their peers. From this standpoint, the relatively high turnover of French stocks by traditional institutional investors (7 months) is an issue. We do not know the specific statistics concerning so-called responsible investors, but it is very unlikely that these investors – who are subject to the same

338

pressure on the performance of their funds, measured annually – manage to double this holding period, which would still be relatively short in view of their long-term objectives.

1.4 From Sustainability to Solvency: the premium link

We made a point that investors should not expect higher returns from SRI: they should expect better returns. The financial attraction of SRI is to provide sustainable and responsible insurance: SRI gives steadier dividends in the long term and means better solvency. Logically, this should result in a price premium and, for the investor, imply lower returns.

Equity investors often resist this logic on the mistaken basis that if CSR performance, in the real economy, leads to leadership in sustainability, then SRI performance should lead to market performance. This confusion, however, disappears as soon as we shift the logic to the bond market, and there the argument is difficult to refute: there is no doubt that a company with better solvency benefits from lower interest rates. Would it be entirely paradoxical to expect the more virtuous companies to pay higher interest? Should corporate issuers pay an extra fee for being more sustainable?

Credit agencies quietly incorporate some ESG factors into their methodology, endorsing the principle that positive extra-financial performance widens the "distance to default" of a company and indicates better solvency. There is a virtuous Catch-22 situation behind the reasons why ESG performance leads to solvency: either the sustainability leaders are the most likely to maintain market leadership, or only market leaders can afford to implement and achieve sustainable objectives. No matter: in both cases it means there is a also a valuable solvency premium attached to SRI investment.

2. The cost of the SRI premium can be optimised with ESG research

The optimisation of the SRI premium cost relies on the fund manager's ability to stay ahead of the market: buy cheap and eventually sell high. That is why, on the whole, as beating the market is a zero-sum game, SRI fund performance is in line with its benchmark.

2.1 Clear financial risks for a portfolio… but unseen financial performance

A strong point of the PRI (Principles for Responsible Investment)110, as well as the EAI (Enhanced Analytic Initiative)111, is the claim as to the capacity of the ESG approach to identify a new type of potential risks for companies. These risks may prove to be even more damaging for a fund’s performance, as they tend not to be taken into account by traditional financial analysis and common practices in the market. Typically, such risks would not be reflected in a company’s beta – and thus not identified as such – since they are not correlated with market risks. From that standpoint, SRI has been an adequate response, being flexible enough to answer a growing demand from the investor community. In fact, a relevant analysis of corporate social responsibility can be a sustainable competitive advantage in addressing a growing set of issues.

110 In 2006, the United Nations Environment Programme launched its Principles for Responsible Investment. The Principles provide a framework to help investors incorporate environmental, social, and governance (ESG) factors into the investment process.

111 EAI is a group of international asset owners and asset managers who work together to encourage investment research that considers the impact of extra-financial issues on long-term company performance

339

The effectiveness with which public health or environmental issues, particularly in relation to the challenge of climate change, find regulatory expression, emerges as a major financial risk. It is actually an argument against CSR to say that it is a tool at the service of capitalism so as to anticipate regulatory changes (Plihon, 2004) – which it is anyway. The propensity of social responsibility to convert into civil responsibility (Voisin, 2006), with a growing financial materiality (antitrust charges, class actions), emerges as a favourite domain for the SRI analyst. Investors clearly have a vested interest, regardless of their mandate, in instilling this approach.

2.2 A field for investment opportunities

In theory, SRI insurance is fully priced in the market. In practice, however, taking ESG factors into account can lead to outperformance. It depends both on the capacity of ESG issues to generate market anomalies and on the ability of the fund manager to spot and seize those opportunities:

- by risk anticipation

CSR is a corporate tool to anticipate and prepare regulatory shifts – and, indeed, consumption shifts. Obesity is one of the health concerns that can potentially reshuffle the cards among the players in the food industry. The uncertainty created by the strengthening of the legal and regulatory environment surrounding obesity typically implies higher risks for groups whose sales are related to unhealthy products.

- and opportunity identification

Regulatory and consumption shifts create risk as well as business opportunities. Characteristically, the carbon constraint is not only based on energy efficiency and adaptation, but is also project-based, meaning it can be defined in a positive fashion. More recently, the green momentum as well as the perspective of economic recession produced the conditions for green recovery paths. Renewable energies are a key area in this context, amounting to 8% of the total U.S. stimulus package.

2.3 Sector approach and financial analysis are necessary steps

At Cheuvreux we focus on identifying key ESG issues and indicators on a sector basis (see Table 2). This is a necessary step prior to identifying those companies that are the best positioned to meet these long-term issues. But beware of that step: it is a complex process that often resists attempts at quantification.

For example, to understand how the CO2 constraint affects the auto sector requires combining many sources of expertise and analysis: assessing the cost and feasibility of the different technical options for curbing CO2 emissions, understanding the advancement and credibility of projects in place, evaluating the automakers' distance from the CO2 objective , calculating how it impacts on their operational margins, estimating their financial flexibility for pursuing environmental efforts, and so on.

Table 2: Key ESG issues for a few sectors

340

SECTORS ISSUES

1. Addressing the CO2 challenge: overall target of 130g / km by 2012

2. Commitment to promoting eco design: End-of-life Vehicle Directive

AUTOMOTIVE

3. Labour management of cyclicality in terms of social management

1. Environment liabilities: disclosure, management and provisioning

2. Commitment to hazardous product substitution (REACH implementation)

CHEMICALS

3. Investment in clean and green chemicals

1. Company product portfolio: fair trade, organic, healthy food, etc.

2. Impact on the environment: information and improvement FOOD

3. Food poverty vs. land competition and water shortage

Source: Cheuvreux research

Although the major European banks are now active in the area of extra-financial research, it is difficult to integrate these factors into traditional financial analysis. Most of the time, they cannot be plugged into a valuation model. All in all, it is easier and more relevant to compare balance sheets and P&L than indicators of social or environmental performance.

In some cases, CSR offers some business opportunities (green or clean technologies, for example). This falls back into traditional financial analysis, in order to identify the best positioned or undervalued companies. However, it depends to a great extent on the political momentum on environmental issues. There is no reason to believe that these sectors/technologies should display abnormal, above-average returns. The question of financial performance is thus very marginal.

Through ESG risk anticipation, an investor can reap the benefits of low-cost insurance. It is not by chance that some hedge funds have a taken keen interest in these issues. However, their motivation for incorporating ESG is certainly cynical and they do not expect, or value, any dividend other than a financial one. But they do contribute to the mainstreaming of the issue and therefore to the absorption factors of ESG anomalies.

3. The ultimate equation: ESG = f(a3), anomaly, anticipation and absorption

In this section, we present our view of the dynamic of SRI investing. SRI financial gains are not systematic, but ESG issues have a strong potential to create opportunities for the investor.

3.1 ESG issues create volatility and therefore market opportunities…

In a landmark article, Richard Roll (1988) showed that the explanatory power of betas and overall CAPM was low, with the adjusted R2 of the model being around 0.35. According to Roll, the market does not fully account for variations in stock returns, leaving room for further analysis. The research of other explanatory factors led, among others, to the Fama French model. ESG analysis may partially help to fill the gap.

341

One might suggest that extra-financial issues are a significant source of stock price volatility and will remain so over a mid- to long-term timeframe. This volatility generates market opportunities for the investor, and will sustain the need for extra-financial analytic tools.

The scope of the subject and the complexities of the issues involved should be the main drivers of that resilience and future. One of the features and the challenge of extra-financial research is the scope of the issues it addresses. Not only does it require multiple areas of expertise (legal, social, environmental), but it is also constrained by limited resources (around 300 extra-financial analysts in Europe). There is undeniably a vast field of SRI issues ahead.

3.2 …until anomalies are absorbed by the market

It is dangerous to underestimate the capacity of the market to price in ESG issues. Several mechanisms will support this integration on a long-term basis: through increased visibility and normalisation. Broker research plays a role in this movement.

- The potential for regulation and normalisation is still high

Relevant ESG issues have strong potential for being expressed in regulation. Carbon is the obvious example. Even for CO2, it may be just the beginning, given the challenges of global warming… but the road to Copenhagen is long. The prospect for further regulations is strong in a number of sectors/issues: nuclear waste, waste and recycling, health issues in food, water management, energy efficiency, etc. Once regulation is passed, the ESG issue at stake is factored into the market (see Figure 3).

- Reporting initiatives and normalisation are likely to be other drivers

Normalisation of reporting is recent, dating from the early 1990s and emerging first with corporate initiatives. Corporations also rely on sector initiatives (chemicals, for example, as early as the 1990s).

Shareholder activism (the Cadbury report in 1992 in the UK, and the Enron, Parmalat and Vivendi scandals) has accelerated access to information in regard to governance. Since the beginning of the current decade, expectations have also grown with respect to environmental issues (climate change among others) and led to significant progress in environmental reporting. On the downside, social issues (human resources, for example) are currently less developed.

Since 2002, listed companies in France must account for their social and environmental impacts (following the NRE law, New Economic Regulations). Seven years have passed, and its outcome is symbolic and positive. The “laundry list” nature of this law, which is sometimes not particularly relevant for many economic sectors, has certainly tempered enthusiasm in regard to this regulation.

Overall, reporting is part of a virtuous circle. On the one hand, belonging to an SRI index and the growth of SRI assets provides a strong incentive to improve reporting. On the other hand, improved reporting and normalisation foster sustainable and responsible investment. Through initiatives and normalisation, and maybe one day through their incorporation into accounting standards, ESG reporting contributes to the assimilation of SRI by the market.

342

3.3 From risk to value

According to the PRI, ESG issues can affect the performance of an investment portfolio. PRI signatories recognise that their fiduciary role is to consider ESG issues in the investment process. Going further, Kofi Annan hinted in the PRI declaration that mainstreaming ESG factors should lead to a recognition by the market of the potential value of CSR issues.

As part of our analysis, we consider both extra-financial value and risks. For the most part, extra-financial value derives from intangible assets or capital. Following the initiative of the French Observatory on Intangible Assets, we look at a range of intangible capital, thereby complementing pure financial analysis.

In practice, the frontier between risk and value analysis is porous. The relationship between risk and value is complex, as evidenced by the climate change challenge. The carbon issue has materialised and is clearly priced in by the market. As a result, Cheuvreux now considers carbon, in absolute terms, as a form of capital, a source of future value (see table 3). The difference between risk and value is thus partly methodological: as risks materialise they translate into a premium or discount on the market.

Table 3 : Cheuvreux’s approach to extra-financial value analysis

Extra-financial value analysis

Type of capital Features

Customer capital Range of loyal clients with solvency

Human capital Key competencies, motivation, etc.

Brand capital Brand recognition and value

Intellectual capital Innovation potential is a significant source of competitive advantage

Organisation capital Management, control, quality process, etc.

Shareholder capital Loyal and supportive shareholders

Carbon capital CO2 today + future credits are an asset that can be valued

Source: Cheuvreux Research

This "extra-financial value" is significant. Other things being equal (at comparable levels of earnings, for example), similar companies often trade at a premium or discount between one another. This spread is a combination of analysts’ and the market's confidence in a company, which can rarely be explained through balance sheet analysis alone. Again, the challenge is not to justify an already priced discount or premium, but to spot hidden sources of value ahead of their potential absorption by the market, and leverage them.

4. What is a social and environmental dividend worth?

343

One major investor motivation for taking account of ESG is to encourage CSR behaviour. This implies either putting pressure on the worst companies or encouraging those that display best practices. Such a motivation has two implications: 1) it stresses the need to meet extra-financial expectations by providing investors with an extra-financial dividend disclosing the ESG performance of the companies they invested in; and 2) it recognises the value of such a dividend, the reporting of which is, at this stage, a considerable challenge to formalise. Recognising that social and environmental dividends have a value does not necessarily mean that we are willing to pay for them. But is there an alternative?

On the other hand, entering the virtuous cycle that encourages corporations to go one step further on CSR issues and that leads to a recognition by the market of the potential value of ESG issues, as suggested by Kofi Annan, necessarily implies some investment in the first place. First you sow, then you reap.

It is, in the authors’ opinion, the cornerstone of SRI business development to report on these environmental and social benefits: quantification and independent auditing are key. Unfortunately, these benefits are much more complex to measure and report on than financial performance.

4.1 The difficulty of communicating on extra-financial dividends

Corporate reporting is the primary source of information for extra-financial analysts, rating agencies and investors alike. It is the raw material for SRI analysis and processes, and as such provides a good reflection of the areas of improvement to be made.

If collecting data on companies is a tricky issue in SRI analysis, it is even more complex to evaluate the environmental and social (E&S) benefits yielded by a whole portfolio. In that respect, exclusion funds may be the exception. The investment process is easier to understand, explain and sell, since it does not involve stocks in morally questionable businesses (tobacco, gambling, alcohol or arms).

More complex and positive approaches, such as best-in-class, favouring companies most committed to achieving social and environmental objectives, are a more difficult sell. Few, if any, companies in an investment universe will be irreproachable, and the methodology will always be open to criticism or at least question. This approach remains, however, the most appropriate technique for distinguishing between companies on relevant ESG issues and triggering the necessary positive changes in their behaviour in regard to their stakeholders.

For the purpose of clarity and transparency in the process, it can be fruitful to drop a global ESG approach and focus instead on specific issues. Carbon is a leading metric that favours these initiatives, and in general playing the green theme goes down well. In comparison, social issues receive far less attention.

The Low Carbon 100 Europe Index is an illustration of such attempts. It requires that the set of companies invested in outperform by 42% their benchmark (the largest 300 European companies) in terms of carbon emissions. It reproduces the structure of its benchmark and picks the best-in-class in terms of carbon emissions from among the different sectors and subsectors.

344

Themes can also be attractive. They offer an approach very similar to traditional investment, only with a clear environmental bias. The idea is to invest in those companies whose products are seizing the opportunities offered by sustainable development (renewable energies, recycling, water management, and so on).

Social issues have received far less systematic attention from investors, with the exception of a few funds: for example, in France, Natixis AM’s "Insertion emploi" Natixis AM and AXA IM’s Social Fund.

The development of such products and generally higher standards of transparency of SRI funds underline the crucial importance of extra-financial reporting. It reveals that investors’ extra-financial motivations are a key component of the demand for SRI products.

4.2 What childcare and SRI investment have in common

In their Freakonomics, Stephen Dubner and Steven Levitt highlighted the unintended consequences of deterring moral behaviour by setting financial incentives. Daycare centres are usually concerned with parents who are bad at picking up their offspring on time: the children become anxious and a teacher has to stay on. One obvious solution is to impose a fine on those coming late. As it turned out, in those centres that tried it, fining late parents was totally counterproductive: before long, the number of late pickups doubled.

Thus economic incentives can deter altruistic or ethical behaviour. In a similar way SRI should stick to its core promise of extra-financial dividends and discard its claims of outperformance. It should count on the overall positive benefits of responsible investment and not discount the non-financial motivation that drives it.

345

Conclusion

The article’s main point is that an SRI investor should not expect higher but better returns. Primarily, Sustainable & Responsible Investment generates environmental and social dividends. If reflected at all in stocks valuation, we contend that SRI should logically rhyme with higher stock prices, to account for the intrinsic value of those extra financial benefits. In addition, SRI offers an insurance against extra-financial risks. As such, it provides an enhanced security for the investor: once more, this insurance has a price that should add to the value of a stock. Just like fair trade products, a bit like higher grade corporate bonds, high SRI ranked equities should trade with a premium. Therefore, SRI universe should fundamentally yield lower returns and this should not entail, for the 2 reasons mentioned above, its attractiveness as an investment. On another level, we argue that ESG issues are a source of market disturbances, create volatility and therefore investment opportunities. The field and the scope of ESG issues, (such as climate change challenge, heath, water management, corporate governance and so on) are broad and require many areas of expertise (legal, social, environmental, financial). This suggests the sustained need for extra-financial analytic tools and further research in this area.

346

References

[1] Allouche J., Laroche P. [2005], A meta-analytical examination of the link between Corporate Social and Financial Performance, Cahiers de Recherche du Gregor, 2005-03.

[2] Crifo P., Ponssard J.P. [2008], RSE et / ou performance financière: points de repère et pistes de recherche, Ecole Polytechnique, Cahiers no. 2008- 15.

[3] Guerard J.B. [1997], Is there a Cost to Being Socially Responsible in Investing?, Journal of Investing.

[4] Hong and Kacperczyk [2006], The price of sin, the effects of social norms on markets, working paper, Princeton University, 2006.

[5] Lowenstein R. [1995], Buffet, The Making of an American Capitalist, Random House, New York.

[6] Margolis J., Elfenbein H., Walsh J. [2007], Does it pay to be good? A meta-analysis and redirection of research on the relationship between corporate social and financial performance, working paper, Harvard Business School.

[7] Orlitzky M., Schmidt F., Rynes S. [2003], Corporate Social and Financial Performance: a meta-analysis, Organization Studies, 24(3): 403-411.

[8] Plihon D. [2004], Le nouveau capitalisme, Repères, La découverte.

[9] Roll R. [1988], R2, Journal of Finance, Vol. 43, no. 2, July 1988.

[10] UNEPFI, Mercer [2007], Demystifying Responsible Investment Performance, a review of key academic and broker research on ESG factors.

[11] Voisin S. [2006], Impulsions et perditions de l'investissement socialement responsable, Regards, Cahiers de Syndex, no. 9.

347

History, challenges and limitations of the

"corporate social responsibility" and "sustainable development" coupling1

Françoise QUAIREL, Michel CAPRON

Abstract : In Europe, the concepts of corporate social responsibility (CSR) and sustainable

development (SD) are closely linked. Yet this approach to CSR has only taken shape in the last

ten years and it is not shared by everyone. The objective of this article is to present the different

stages of the CSR and SD "coupling" process, highlighting the role of "institutional

entrepreneurs" (World Business Council for Sustainable Development and SustainAbility), who

have contributed along with large international organizations (UN, OECD and the European

Commission) towards the institutionalization of big businesses as actors in SD and to analyze its

challenges and limits. The major challenge faced by the CSR-SD coupling is the political role

attributed to big businesses in terms of international regulation in the absence of real global

governance and international company law. CSR's political approach, based on self-regulation

and the "win-win" strategy argument, has many limitations for the implementation of SD policies

at a global level.

Keywords: Corporate social responsibility, Sustainable development, Stakeholders, International regulation, Institutionalization, Sustainability

1 This article was written as part of the ANR Program “Le potentiel régulatoire de la RSE”.

348

Today in Europe when we talk about sustainable development policies in business, we

immediately associate this with the concept of social responsibility (in its broadest sense, that is,

encompassing the societal and environmental aspects). There is a consensus of opinion on the

European continent in defining corporate social responsibility (CSR) as "the contribution of

businesses to sustainable development"2. Whether it be on managerial agendas, or in the stance

taken by NGOs, academic papers, standardization measures, and papers and reports issued by

public authorities, corporate responsibility is currently indissociable from the notion of

"sustainable development" (SD).

The link between these two concepts which, up until the beginning of the new millennium,

referred to two clearly distinct semantic fields, is now generally accepted by all parties. The CSR

semantic field is adopting more and more aspects from that of SD. CSR is therefore to be

assessed in the light of the three pillars of sustainable development: economic prosperity, social

justice, and ecological sustainability. This combination makes up what could be described as a

"coupling" of the meanings of both elements3. In general, this coupling takes on the attributes of

the progressive institutionalization of the corporate role in recognizing and, indeed, taking

responsibility for sustainable development matters.

This European consensus on the integration of sustainable development into CSR is on the one

hand a rather recent phenomenon and on the other still a long way from being universally

recognized by all the parties involved (businesses and stakeholders), who are working throughout

the world to develop CSR. Moreover, CSR's different traditional approaches describe corporate

responsibility as the response to other types of justifications and criteria of a very diverse nature:

ethical, managerial interest, answers to stakeholder expectations, embedding of business in

society, and so on.

In this article we intend to analyze the stages of the CSR-SD coupling process (the first fruits of

which go back to the 1990s), to present the parties involved and to assess the implications that

can be inferred from this representation of the corporate role in SD. In particular, we will

2 The title of the July 2002 communication from the Commission of the European Union is “the business contribution to the sustainable development”

3 The association may at times be highly ambiguous when the managerial agenda shifts the sustainable development of the planet towards that of businesses.

349

demonstrate that this representation gives a political framework to the development of CSR in the

sense that it is accompanied by a transfer of responsibility from governments to business. We will

also present the limits of such a transfer and, lastly, will look at its consequences on the definition

of the concept of sustainability. We will draw on the description and analysis of the discourses

that have accompanied the different stages of this pairing between CSR and SD since the

beginning of the 1990s. These discourses have been drawn from texts from the world of business

and from international organizations and conferences at a global as well as a European and

French level.

After having presented the different traditional approaches to CSR in the first part, we will

analyze the successive stages in the CSR-SD coupling leading to the institutionalization of "the

responsible corporate actor" as a SD actor in the second part. The third part will deal with the

challenges to and limitations of this representation.

Part 1: Traditional approaches to corporate social responsibility make no reference to

sustainable development

The currently dominant representation in Europe of corporate social responsibility as its

contribution to SD was by no means universally recognized in time and space. The three main

traditional proposals of CSR that can be drawn from the literature on the subject make no

reference to SD. Historically, an "ethical" approach inherited from 19th century paternalism

(1.1), a utilitarian strategic approach (1.2) and a so-called "stakeholder" approach (1.3) can be

identified. Although the approaches are different, they do share common ground (Capron et al.,

2007).

1.1. The "ethical" approach, inherited from paternalism

The approach inherited from the paternalism of the 19th century is generally founded on moral

and religious values and appeals to the personal ethics of the leader. Bowen (1953), who was

asked by the American evangelical churches to develop a social doctrine to counter the 1891

Rerum Novarum Encyclical established by the Catholic Church, is generally considered to be the

first theorist of this movement, which later comes to be known as "business ethics" (Pasquero,

350

2005). It is founded on biblical precepts (the stewardship principle and charity principle) by

considering the business as a "moral entity" duty-bound to ensure the well-being of its workers,

their families and their neighborhood, including the well-being of the community. The measures

and social actions undertaken by businesses can only be of a voluntary nature and are therefore

situated "outside the business" (e.g. through foundations): patronage and philanthropy are the

ultimate stage of CSR (Carroll, 1979). These measures are generally intended to repair, not

prevent, the damage caused by economic activity. This idea that arose in the United States is still

the most widely held view.

1.2. The utilitarian strategic approach

This approach is based on the idea that corporate social behavior should be used to assist business

performance. It maintains that the demands for profitability will be better achieved by

maximizing the social effectiveness of individuals and the organization. The choice of social

action is made using a cost/benefit analysis that demonstrates its business relevance and

usefulness to managers. In particular, the business should look after its brand image and

reputation, and preserve trust among its stakeholders, namely the groups and individuals liable to

influence or be influenced by the execution of the organization’s objectives (Freeman, 1984). The

business should therefore pay attention to their expectations in order to preserve (or gain) the

legitimacy that allows it to obtain the resources necessary for its business activity and therefore to

guarantee its long-term existence. This approach, which appeared in the 1970s in Europe and the

United States (where it gave rise to the managerial movement known as "Social Issues

Management"), is in keeping with the classical idea of business economic and financial

performance; the justification is based on the "business case": an argument that justifies CSR

voluntary measures by all the competitive benefits that are supposed to come from them. We are

purportedly in a win-win situation: society wins, business wins.

1.3. The CSR approach as an answer to stakeholders' expectations

351

Stakeholder theory has become a benchmark in CSR4; it provides a framework for defining CSR

and its integration into strategic management. It puts the business at the centre of a group of

actors (the stakeholders) who have implicit and explicit expectations regarding the business and

who have greater or lesser power to be heard and influence leaders. In this approach, integrating

"stakeholders" into corporate governance rather than "shareholders" alone legitimates CSR. This

theory aims at combining the ethical and utilitarian approaches: it embodies several proposals

that can be linked to the ethical or utilitarian views of CSR. The typology proposed by Donaldson

and Preston (1995) distinguishes between "the normative approach" and "the instrumental

approach". In the normative approach, stakeholders' interests have an intrinsic value that carries

the interests and welfare of society as a whole; stakeholders are the bearers of the welfare

challenges of society. In "the instrumental approach", CSR is defined as the integration into

strategic management of the dialogue with these stakeholders.

As we have already seen, the "Social Issues Management" (utilitarian) movement states that the

long-term existence of a corporation depends on its ability to respond to the expectations of its

stakeholders. Identifying and managing relations with the latter is central to all the standards and

guidelines in CSR implementation (GRI, AA 1000, SD 21000, ISO 26000, etc.). This approach

provides a strong framework for CSR discourses and representations, whatever the country.

"Accountability"5, which is often used with the concept of answerability, is an account-giving

behavior; being accountable has become CSR's key principle.

However, CSR's "stakeholder" theory has a number of limitations (Pesqueux, 2006): limitations

regarding the legitimacy of the actors who are supposed to represent the key challenges faced by

society, limitations related to challenges that are not upheld by the stakeholders and, above all,

the underlying hypothesis that the sum of the welfare interests of private individuals adds up to

the common good; yet we know, from the Condorcet paradox and Arrow's impossibility theorem,

that the sum of individual preferences is insufficient in forming the basis for the general interest

of the whole. While the expectations expressed by the stakeholders are helpful in identifying the

4 See S. Mercier: "Stakeholder theory, a synthesis of the literature" (2006).

5 “Balancing Performance Ethics and Accountability”, Zadek (1998). “Accountability” is the first principle of responsibility for organizations selected by the committee draft of the ISO 26000 standard. Accountability is defined as: “responsibility of an organization for its decisions and actions, and the state of being answerable to its governing bodies, legal authorities, and, more broadly, its other stakeholders regarding these decisions and actions” (ISO, 2008)

352

environmental and social stakes, they cannot claim to represent general welfare and the common

good.

The different approaches of CSR presented in the first part are not mutually exclusive within

managerial discourse; rather they tend to join forces in strengthening the legitimacy of CSR

policies. This overview highlights the uniqueness of the views debated in Europe today, in which

the primary justification for implementing CSR policy is linked with sustainable development.

The "sustainable development" concept has become a matter of consensus in this, a reference

point for all the parties involved in social and economic life; the different representations held by

the parties involved make this consensus possible without any further clarification of meaning.

The word is omnipresent in management vocabulary: CSR policies are drawn up by "Sustainable

Development" departments in large groups, and CSR messages are found on the SD pages of

websites and in published SD reports.

Thus the CSR concept is closely linked in these discourses to the concept of SD that has become

the embodiment of the general interest to be taken into account by "responsible companies". How

did this imbrication of the two concepts occur and what are its implications?

Part 2 The key stages of the institutionalization process of "the responsible business" as

an SD player

The reference to SD in defining CSR can be analyzed as a process of institutionalization, as

defined in neo-institutional sociological theories. In defining the institution, Scott (1995) refers

either to the legal framework or to a set of social norms and accepted rules; he underlines the fact

that "institutions consist of cognitive normative and regulative structures and activities that

provide stability and meaning to social behavior" (Scott, 1995, p. 33). Institutionalization results

from the processes through which actions become "routine" within an organization or an

organizational field. Legitimacy, i.e. the “perception or assumption that the actions of an entity

are desirable, proper, or appropriate within some socially constructed system of norms, values,

beliefs, and definitions” (Suchman, 1995, p.574), becomes a crucial lever for modifying

behavior. The symbolic dimension of decisions and structures takes on greater importance than

their material dimensions (Meyer and Rowan, 1977). A large part of the work done within neo-

353

institutional theory deals with the creation and transformation of institutions. This work

underlines the importance of historical processes and offers an analysis of the influence of actors,

described as "institutional entrepreneurs", on the transformation or creation of new institutions.

Such “institutional entrepreneurs” “mobilize allies; they are highly embedded agents; they are

able to get the main actors in this field to follow their view; they develop discursive strategies,

discourse and texts to legitimize their approach; they theorize the institutional project” (Leca et

al., 2008)6.

This theoretical framework can be applied to the coupling phenomenon between CSR and SD

forming part of the institutionalization process, which can be broken down into the following

steps:

- the development of discourses on principles or emerging practices,

- the actions of institutional entrepreneurs (World Business Council for Sustainable

Development7 and Elkington8) whose legitimacy is recognized by the actors in the organizational

fields; the institutional entrepreneurs develop theories that include SD in "responsible" business

strategies,

- setting up lobbying strategies with international or European authorities contributing to the

communication of the concept,

- developing assessment and scoring instruments9, management standards (SD21000, ISO

26000, GRI) that will integrate SD into CSR's semantic field.

Three periods can be identified that correspond to the three institutionalization stages of

business responsibility in sustainable development: before 1990, when the two concepts were

separate (2.1); the 1997-2002 period, during which the two terms became associated with one

another (2.2); and from 2005 onwards, where the texts clearly reveal a shift from the recognition

of SD to its adoption (2.3).

6 See Leca et al. (2008), which presents a very complete review of the literature, covering all the articles on the institutional entrepreneur published between 1988 and 2008 in the principal academic journals.

7 WBCSD.

8 See Stage 2: 1997-2002.

9 Vigeo, an extra-financial ratings agency, presents itself as a group “measuring companies’ performance in the field of sustainable development and social responsibility”.

354

2.1. Before 1990: two separate concepts implemented by different parties

The history of the CSR-SD link is quite brief. Until the 1990s, the concepts of CSR and

sustainable development were confined to separate spheres moving in parallel: the one mainly

driven by universities in the United States (the Business Ethics movement), starting in the 1960s,

and pioneer businesses (grouped under the Business Social Responsibility association); the other

emerging within the scientific community of the IUCN (International Union for Conservation of

Nature) also during the 1960s, based on the ideas of several unorthodox economists of the time

(Ignacy Sachs and his eco-development concept in particular). Made popular in 1987 by the

Brundtland report10, the concept of sustainable development obtained worldwide recognition at

the United Nations Conference on Environment and Development in Rio de Janeiro in 1992.

Article 30.3 of Agenda 21, launched at this conference, states that “Business and industry,

including transnational corporations, should recognize environmental management as a key

determinant to sustainable development” (ONU, 1992). Agenda 21 recognises the benefits of the

initiatives taken by some pioneering businesses, but stresses the importance of the role of

governments in creating the conditions and economic tools for the re-internalization of

externalities according to the "polluter pays" principle.

2.2 Stage 2: 1997-2002: Toward a linking of the concepts of CSR and SD

The year 1997 is pivotal for the linking of the CSR and SD concepts. It was marked by the

publication of two books from the institutional entrepreneurs of the World Business Council for

Sustainable Development and SustainAbility, as well as by that year’s session of United Nations

General Assembly, an event that signaled the beginning of a series of important steps towards the

coupling of the CSR and SD concepts. We shall first present the World Business Council for

Sustainable Development and SustainAbility (2.2.1), and then describe in detail the key events

of the CSR-SD coupling (2.2.2).

10 United Nations World Commission on Environment and Development, Our Common Future / Brundtland Report, 1987.

355

2.2.1 The mobilization of institutional entrepreneurs: World Business Council for Sustainable

Development and SustainAbility

It was during this period that all the defining features of the institutionalization process were set

up: the convergence of the actions of the two parties that we can describe as "institutional

entrepreneurs" – the World Business Council for Sustainable Development and SustainAbility –,

together with the development of their theories and their influence on the international

organizations on the one hand and the tools for implementing these on the other.

The emergence of the World Business Council for Sustainable Development (WBCSD) played a

decisive role in bringing together the concepts of CSR and SD. Created in 1995, shortly after the

Rio Declaration on Environment and Development (1992), the WBCSD was made up of 120 of

the world’s largest multinational companies at the time11. Endowed with substantial resources, it

forged for itself a key position in the international debate on sustainable development,

particularly on environmental questions (Zaccaï, 2002). Its policy is based on the statement:

“maintain credibility with society which is necessary to sustain business operations”

(Schmidheiny, 1992, p.84). Its Swiss founder, Stephan Schmidheiny, took part in the Rio

Conference in 1992 on behalf of business, and his influence was already evident in article 30 of

Agenda 21, where the best practices of certain large pioneering companies, "enlightened leaders

of enterprises", are put forward to demonstrate their contribution to sustainable development.

Thus paradoxically in comparison with NGOs, the World Business Council for Sustainable

Development has without doubt done more to raise corporate awareness (at least among larger

companies) of the challenges of environmental protection, even if it has been criticized in some

quarters for having subtly acted against all practical global measures to combat climate change

(Levy, Egan, 2000).

The influence of the World Business Council for Sustainable Development on international

authorities has grown since that time. The "responsible entrepreneurship" concept developed in

May 1998 by UNEP borrows a lot from the texts published by the WBCSD, particularly "Signals

11 In 1992, Schmidheiny assembled 50 big European, Japanese and North-American multinationals at the Business Council for Sustainable Development (BCSD) to represent the business community at the Rio summit. The basic ideas underlying their view of their responsibility (largely environmental) are presented in the book: "Changing Course: a global business perspective on development and the Environment". In 1995 BCSD became the World Business Council for Sustainable Development.

356

of Change" (1997)12. This report lays down the theoretical basis of the "business case": it insists

on the potential for innovation provided by SD and on the potential for restoring the legitimacy of

multinationals, by entrusting them with the mission of preserving the environment and reducing

poverty. It also lists all the benefits, in terms of profitability, that proactive businesses will obtain

from this strategy. The best practices presented are intended to demonstrate their feasibility, and

the "win-win" logic is described as a holistic vision for business practice and SD13. But the

WBCSD also emphasizes that these endeavors must remain voluntary, and that governments

should create favorable market conditions and not intervene by means of coercive laws. Two key

tools complement these notions: the creation of eco-efficiency indicators for managing

environmental strategies and the promotion of product life-cycle studies (Schmidheiny et al.,

1997).

SustainAbility and its founder John Elkington is the second key player in the institutionalization

of SD in CSR. Elkington founded this strategic consultancy company in 1987 in the United

Kingdom. It was initially the driving force behind the "stakeholder" approach and particularly the

NGO-business dialogue for restoring the legitimacy of certain big businesses, particularly Shell

after the Brent-Spar oil platform crisis in 1995 (Aggeri et al., 2005). Elkington then published a

book14 in 1997 in which he recommends the "triple bottom line", i.e. the need to assess the results

of a business on the basis of three criteria: economic, environmental and social. This view of

sustainable development and the need to incorporate these three dimensions into business

objectives was very quickly and extensively taken up in Europe and applied to CSR management

policies and implementation tools. The coordination of economic, social and environmental

policies that formed the basis for the Brundtland report15 was symbolically rephrased by

SustainAbility in 1995 as "People, Planet, Profit", the title of Shell's first SD report.

2.2.2 Key events in the CSR-SD coupling

12 The writers of this report are S. Schmidheiny, R. Chase (BP) and L. DeSimone (3M).

13 “It is a shift from a fractured view of environment and development issues to a holistic view of business and sustainable development” (Schmidheiny et al., 1997, p. 9).

14 Cannibals with forks: the triple bottom line of 21th century business (1997).

15 Zaccaï (2002, p. 101) provides a diagram of the triangle that has become famous, taken from a 1993 World Bank document that breaks down sustainable development into three components (or three pillars).

357

Alongside the World Business Council for Sustainable Development, SustainAbility and

Elkington made major contributions to closing the gap between CSR and SD during this period,

in terms both of the political role of large businesses and of the managerial justification and

implementation of social responsibility. Four key events have marked the recognition of the

convergence of the CSR and SD concepts.

(1) Within the framework of the mandate entrusted by the 1997 United Nations General

Assembly16, the 1998 Commission on Sustainable Development established the responsibility of

businesses in recognizing SD, in a document17 that sets out a framework for what corporate social

responsibility should be. In this document, the Commission referred to a statement by Koffi

Annan, UN General Secretary at the time, – “Responsible entrepreneurship is therefore a business

strategy to achieve sustainable development” – and called on all companies to voluntarily define

strategies to embody the SD triptych – economic, social, environmental – in all their departments

and operations. The works of the Commission have left their mark on United Nations doctrine on

responsible entrepreneurship. The "Global Compact18", proposed in 2000 by Koffi Annan, is also

in keeping with this thinking. Even if the term sustainable development does not appear in the

Global Compact principles, it underpins the entire objective of identifying common values that

could become indispensable by establishing a protocol for businesses in a globalization context.

Global Compact's political dimension is fundamental in that for the first time businesses were

considered to be members of the global political community, when hitherto only nation states had

been involved (Frydman, 2007).

(2) In June 2000, in its Guidelines for Multinational Enterprises, the OECD declared that its

objective was to improve the framework for foreign investors and also “to enhance the

16 UN General assembly at its 19th special session (23-28 june 1997): Programme for the Further Implementation of Agenda 21.

17 Responsible entrepreneurship, Background paper N° 4, 6th session.

18 The Global Compact asks big businesses to contribute to the promotion and implementation of "globalisation with a human face" by voluntary commitments, particularly in the struggle against poverty. Drawing inspiration from the Declaration of Human Rights, the fundamental principles of the ILO and the Rio Declaration on environment and development, it includes nine principles: two are concerned with respect for human rights, four with the fundamental right to employment, and three with the protection of the environment. A tenth principle regarding the struggle against corruption was subsequently added.

358

contribution to sustainable development made by multinational enterprises” (p.11), as well as “to

implement best practice policies for sustainable development that seek to ensure coherence

between social, economic and environmental objectives” (p. 12).

(3) The Report of the World Summit for sustainable development in Johannesburg in August

2002 highlighted the weakening of governments and the growing power of big businesses and

NGOs in sustainable development. The Johannesburg plan of action (UN, 2002) proposed to

“encourage industry to improve social and environmental performance through voluntary

initiatives” (p. 21). Faced with the issue of financing operations, governments promoted public-

private partnerships. The idea that governments were not essential to promoting SD in view of

the relative ineffectiveness of international authorities is something that stands out clearly in the

conclusions of this summit (Godard, 2003). There appears to be a gap between well-intentioned

international policies and the resources allocated to them: international authorities are unable to

control the commitments of governments, which are more concerned with the interest of their

own industrial sectors than the welfare of the global community. This summit assigned

considerable responsibility to large multinational companies in terms of SD.

Thus at an international level, the leading authorities have progressively deployed sustainable

development in their discourse as a common value that is recognized and accepted by business.

They ask big business voluntarily to take on the responsibility for the planet’s SD needs in the

knowledge that representations of SD are more directed towards environmental aspects. This

movement is the consequence of the weakening of their authority in the context of globalization,

as well as the growing influence of thinking that that would enhance the power of market

regulation. Europe followed a parallel path during this period with a more balanced view of the

relation between the social pillar and the environmental pillar of SD.

(4) The July 2002 policy document from the European Commission entitled “Corporate Social

Responsibility: A business contribution to Sustainable Development” is a sign of the belated

adoption of the CSR-SD coupling in the policies of European institutions. The Green Paper

“Promoting a European framework for Corporate Social Responsibility” published in July 2001

by the European Commission as follow-up to the European Council meeting in Gothenburg

on15-16 June 2001, where the SD strategy for the European Union had been approved, still did

359

not convey an impression of any strong coupling of CSR and SD. But it served as a reminder that

“the debate on the role of business in achieving sustainable development is gaining importance

on the global stage” (p. 15) and was in keeping with the United Nation's 1998 definition of a

responsible business (referred to above). Structured around social and environmental

considerations, the Green Paper also called for stakeholders to be consulted on EU strategy in

matters relating to CSR.

In order to clarify this aspect of the European position of 2001-02, it is appropriate to recall some

of the key stages. The social component of sustainable development was more particularly

developed in the European Union in the mid-1990s. Spurred on by Jacques Delors, president of

the European Commission, some of the heads of leading French enterprises were encouraged to

sign a Manifesto against social exclusion. This call was extended in 1994 to all the big European

businesses. This resulted in the formation of the European Business Network for Social Cohesion

(EBNSC), which several years later became CSR Europe, a pivotal organization between the

European business community, trade unions and European NGOs in the discussions following

the publication of the Green Paper on CSR. The social component of SD was therefore

introduced through the issue of social cohesion. It is also of note that it was the Directorate-

General for Employment and Social Affairs which drafted the Green Paper and not the

Directorate for the Environment.

In addition, the European SD strategy specified in the Commission's statement of June 2001 (in

preparation for the Council of Gothenburg) was a clear reminder of the EU's role in promoting

SD at the European and global level; it declared that "political leadership is essential" (p. 4) and

proposed measures to be taken at the EU level. The SD responsibility of businesses is only one

example of possible action: “Public policy also has a key role in encouraging a greater sense of

corporate social responsibility and in establishing a framework to ensure that businesses integrate

environmental and social considerations in their activities. […] Business should be encouraged to

take a pro-active approach to sustainable development in their operations both within the EU and

elsewhere” (p. 8). The breakdown of responsibilities is clear: SD is above all part of the

responsibility of the political authorities. The July 2002 statement that referred to the Green Paper

was a decisive step at the European level in the linking of CSR and SD: its title "Corporate Social

Responsibility: A business contribution to Sustainable Development" is unambiguous in this

respect. It states that “CSR is intrinsically linked to the concept of sustainable development” (p.

360

6) and emphasizes that CSR constitutes a new form of governance within the context of

globalization.

We can therefore conclude that, following this 2001-02 period in Europe, CSR is defined as a

taking into account of SD by businesses; the connection is no longer a matter for debate. All

management tools are structured around the "triple bottom line" and the "business case"

arguments restate the economic advantage for businesses in incorporating SD into their practices.

2.3. Stage 2. 2005 onward: From “taking SD into account” to “taking responsibility for SD”

Since 2005 the shift of responsibility from the political level the business level has become

increasingly marked. At the EU level, in its statement of March 2006, the European Commission

called “on the business leaders and other key stakeholders of Europe to engage in urgent

reflection with political leaders on the medium- and long-term policies needed for sustainability

and propose ambitious business responses which go beyond existing minimum legal

requirements” (Commission UE, 2006, p. 4). It pointed out that CSR was essential for achieving

SD objectives.

Thus not only is the CSR-SD linkage not called into question, but the rejection of any additional

restrictive European regulations, the promotion of voluntary initiatives, and the principle of

"responsible competitiveness" in a "sustainable market economy" (EU Commission, 2006),

confirm that SD implementation is based primarily on business strategies. The lack of

international consensus on what is to be done following the Kyoto Protocol and on the United

Nations Human Rights Norms for Corporations project is clearly handing over responsibility for

promoting the sustainable development of the planet to the big business by means of self-

regulation.

Over this period, the increase in the number of instruments, e.g. specific codes of conduct for

businesses or sectors, international agreements signed with international union federations, multi-

stakeholder initiatives and CSR guidelines, strengthen regulatory measures by integrating into

business tools the principles of substantive norms intended for nation states (such norms of

behavior include: international agreements on environmental protection, respect for fundamental

human rights and minimum social rights).

361

Part 3. Challenges to and limitations of the CSR-SD coupling

The coupling of the CSR and SD concepts is not without consequence on the

representations and implementation of each of them. It may give rise to different interpretations

depending on their degree of mutual integration, the perception of how they interact and above all

the subsumption, or even the absorption, of one by the other. The major challenge of the SD-CSR

coupling is the political role attributed to big business in international regulation (3.1). The

limitations of leaving a large part of the implementation of sustainable development up to

businesses are significant at the level of international politics as well as conceptually (3.2). This

observation will lead us to look at the question of the appropriation of the idea of sustainability

by management policies (3.3).

3.1. The political dimension of the coupling

The institutionalization of the business role in the implementation of SD can be seen as the sign

of an ongoing transition from standard international law towards a global law where nation states

would no longer be the only parties involved, given that their political and legal power to control

is cast into doubt within the context of globalization (Frydman, 2007). It is remarkable that there

are no social or environmental international authorities with effective control of the application of

international measures regarding SD (with a role equivalent to that of the WTO for the regulation

of trade). Nor is there any legal definition of transnational business responsibility.

International organizations (UN, ILO, OECD, etc.) lack the means to control government; they

have left it up to the largest companies, which theoretically have the power and the means to take

responsibility for regulating certain issues relating to SD: human rights, working conditions or

destruction of the environment. Businesses have filled the vacuum of global law with self-

regulation and its more open mode of operation in the form of their commitments to principles

like those of the Global Compact, the publication of specific and sectorial codes of conduct, and

the signing of agreements with other stakeholders.

This political dimension of SD integration into business responsibilities coincides with the

expectations of a certain number of large companies. If it turns out that the latter are capable of

voluntarily passing self-regulatory measures, then the introduction of further restrictive

362

legislation (on a national or regional level) would appear to be unnecessary. Further restrictive

legislation would be even less necessary given that the business case is endeavoring to

demonstrate that self-regulation favors business competitiveness. In a certain way, the CSR-SD

coupling reinforces the legitimacy of large corporations when they are accused of having a short-

term outlook under the pressure of financial markets and globalization. This type of argument is

regularly put forward by the World Business Council for Sustainable Development and other

groups of proactive multinational corporations such as CSR Europe.

3.2. The limitations of the implementation of SD by businesses.

The idea of SD implementation by businesses driven by the CSR-SD coupling has several

limitations, both from the standpoint of its effective implementation and conceptually. We will

firstly present the effective limitations of this implementation (3.2.1). A second sub-section will

allow us to look at the debate on the ISO 26000 standard project which demonstrates that the

coupling, in conceptual terms, has not been established at the international level (3.2.2).

3.2.1 Effective limitations

Godard and Hommel (2005) have highlighted the ambiguity of their perception of CSR-SD

integration: they point out that certain businesses have introduced practices into their policies that

are increasingly respectful of the environment and the social conditions of their business activity,

yet at the same all the main SD indicators of the planet are worsening. Voluntary proposals have

their limitations; decisions are made according to economic objectives and only those practices

with a "win-win" result are adopted. For these authors, business is not the right place for

implementing SD; combining CSR and SD by saying that the long-term future of businesses

depends largely on their degree of involvement in SD is, therefore, a "mystification" (Godard,

2005, p.13).

Although self-regulatory practices unquestionably produce certain limited results, there are also

many counter-examples of businesses and groups of businesses which have attempted to slow

down global regulation by the creation of influential lobbies to oppose the measures that could be

set up by international and national authorities (e.g. action taken by the chemical industry lobby

to limit the scope of the REACH directive, pressure applied by large multinational corporations

363

against the “human rights standard” project at the UN sub-commission, the outcry against the

new Employment Law in China by the International Chamber of Commerce and the American

Chamber of Commerce, and so on).

The concession of an international regulatory role to private companies in matters of SD is

questioned by the theoretical framework of global public goods (Samuelson, 1954); this

highlights the failure of the market to produce or protect these goods and stresses the need for

regulation introduced by public institutions (Tubiana and Severino, 2002). In a context of fierce

competition, it is not in the interest of certain private entities to take on responsibility for the

different aspects of SD; yet they tend freely to take advantage of the situation when proactive

companies make commitments to engage in actions that are more socially or environmentally

responsible. As pointed out by Brodhag et al. (2004) and Vogel (2006), in a competitive market

SD will receive limited recognition from businesses if voluntary proposals are not accompanied

by substantive incentives provided by governments or establishing minimum standards for

businesses which do not toe the line. However, the formation of an international multilateral

system is taking too long for such powers to be capable of being exercised.

As we have already seen, the political role of big business in matters of SD and particularly

environmental issues has been carried out by international organizations with limited powers.

The World Business Council for Sustainable Development has provided significant leverage in

promoting self-regulation in this area, but it is essentially at the European level that the

institutionalization of the CSR-SD coupling has been achieved. At the international level, the

development process of the ISO 26000 standard on the social responsibility of organizations was

confronted with the clash between the European and North American concepts of CSR.

3.2.2 CSR-SD coupling: the ISO 26000 project

The projected ISO 26000 standard on companies’ social responsibility seeks to define social

responsibility, its principles and its key areas in a document that aspires to be universally

relevant. In the text currently under discussion, the CSR-DD coupling is not established.

Conceptions of social responsibility vary according to the country and party involved and the

negotiations have led to compromises. The most recent definition adopted refers to SD, but adds

in the health and well-being of society. In many countries, the notion of SD is reduced to its

environmental aspect and fails to recognize the three "pillars" on which it is founded.

364

The "ethical" and "stakeholder" approaches are also referred to in this definition19. It is

noteworthy that the term "sustainable development" appears only 39 times in the project, whereas

the term "stakeholders" appears 148 times. The ISO 26000 project nevertheless clearly states:

"Although many people use the terms social responsibility and sustainable development

interchangeably, and there is a close relationship between the two, they are different concepts.

Social responsibility has as its focus the organization, not the globe. Social responsibility is,

however, closely linked to sustainable development because an overarching goal of an

organization’s social responsibility should be to contribute to sustainable development" (p. 7).

It is interesting to see, in the development process of an international and multi-stakeholder

(businesses, unions, consumers, NGOs, governments, experts) compromise, that a number of

representations of CSR are opposed to each other and that the CSR-SD coupling is not

institutionalized, even if the SD standpoint is seen as an overarching principle that is supposed to

coordinate stakeholders' expectations and form a framework for collective action around a

consensus and shared values. The SD concept is sufficiently vague and fluid to allow this mix of

different representations. It can then be understood as a convention that ensures there is

coordination between civil society and the market economy in a context of deregulation (Wolf,

2007).

3.3 Appropriation of the notion of sustainability by managerial discourse

Finally, in order to shed light on the issues at stake with the CSR-SD coupling, it may be useful

to take another look at this shift of the concept of sustainability from the agendas of "ecologists"

to those of management. This shift could be interpreted as an appropriation by the managerial

agenda of a criticism of the capitalist system, similarly to the agenda described by Boltanski and

Chiapello in regard to Fordist organizational business models and day-to-day alienation

(Boltanski and Chiapello, 1999). The concept of sustainability was developed in the work of

19 Definition of the social responsibility of the Committee Draft / ISO 26000 (12-12-2008): “Responsibility of an organization for the impacts of its decisions and activities on society and the environment, through transparent and ethical behaviour that

- contributes to sustainable development, including health and the welfare of society, - takes into account the expectations of stakeholders, - is in compliance with applicable law and consistent with international norms of behaviour and - is integrated throughout the organization and practised in its relationships”.

365

unorthodox economists (M. Strong, I. Sachs) in the 1970s, and put into question growth and

development models that could result in the planet's resources running out (Vivien, 2005).

Although this concept is still the subject of debate, it is generally understood that sustainability

presupposes a new relationship between mankind and nature that subordinates economic

development to its physical limits, together with a changed distribution of wealth between and

within generations.

The writers subscribing to this movement consider, in the same vein as the works of Polanyi

(1944) and Granovetter (2000), that business is embedded in society and its very existence

depends on this fact. Conversely, businesses have a duty to care for the preservation of the

environment. To ensure their long-term survival, managers have an interest in avoiding any

activity liable to destroy resources in a lasting or irreversible manner. They are duty-bound to

contribute to sustainable development, the preservation of social cohesion, the production and

maintenance of global and local common goods. It is also in their "clearly understood" interest to

incorporate into their managerial approach not only economic but also social and environmental

objectives, since their businesses can only prosper in a healthy, viable and fertile environment.

Such thinking chimes with ancient biblical precepts of “the universal destination of goods” and

with Robespierre’s idea that the freedom to appropriate should not be at the expense of those

lacking such freedom (Gross, 2009). The question as to what is covered by the general interest or

the common good is, of course, a fundamental one; for those who stand for sustainability, it is a

political issue and is therefore to be decided by the responsible citizen. But this is an inherently

complex issue that we will not deal with here.

The approach to CSR via the concept of sustainability is different from the utilitarian strategic

approach in two ways. It is on the one hand a long-term concern: what is at stake is not an

immediate competitive advantage but the existence over time of businesses that are dependent on

the survival of the planet and its population. On the other hand, the risks attached to economic

activity should be anticipated and, if possible, controlled by public regulatory procedures that

specify the conditions for the practice of business.

Managerial discourse around SD often appears close to this conception, but it tends to emphasize

the economic dimension (profitability) at the expense of the other two (social and ecological).

Although the World Business Council for Sustainable Development has constructed the business

366

case20 for the convergence of the three dimensions, limitations and dilemmas come to light as

soon as any conflict between them arises. Companies devote less time to considering the

environmental and social dimensions when the expected benefits are outweighed by the costs

(McWilliams and Siegel, 2001; Stoeckl, 2004; Vogel, 2006). Thus managerial references to SD

emerge more as a means of adjusting traditional management practices in keeping with

maintaining the economic system rather than a genuine transformation. The managerial agenda

on SD, widely promoted by the World Business Council for Sustainable Development and

SustainAbility, is aimed above all at preserving the economic system while taking into

consideration the risks associated with social criticism; it works like "a symbolic band-aid"21.

CSR is justified by the declared convergence between its economic and financial objectives and

the social and environmental objectives of SD. "SD" is an undisputed principle within business;

hence it emerges as a means of justifying the actions of those responsible for SD with regard to

shareholders and, internally, to finance departments.

But the main consequence of this is that managerial discourses and practices today tend to specify

the standards of the common good, thereby reversing what the conception of sustainability aimed

to promote, namely the subordination of economic considerations to social and ecological

concerns.

Conclusion

Finally, we have been able to ascertain that responsibility in relation to the implementation of

sustainable development has progressively shifted from the political domain toward the business

sector through the institutionalization of the notion of "responsible business" as a main actor in

sustainable development. However, although the coupling of CSR and SD leads to the legitimacy

of big business being reinforced, it quickly reaches its limits, because the private nature of

20 “Companies recognize that their long-term competitiveness ultimately depends on their ability to understand these societal changes, to contribute to the global good and to carve out new competitive space for productivity, innovation and growth. Taking societal responsibilities seriously helps companies sustain their economic success.” (Site WBCSD: www.wbcsd.org)

21 Laufer (1996) cited by Acquier and Aggeri (2008).

367

business and, above all, the competitive system it operates within does not predispose companies

to produce public goods.

In regard to representations, even if sustainable development is present as a formal reference

point, it cannot be used as the basis of a universal conception of CSR. The emphasis put on the

"business case" as an argument encouraging businesses to make voluntary proposals results in the

SD issue being overturned. For when SD is situated in a competitive environment, it has proved

favorable to the competitiveness of individual firms, whereas what is at stake for CSR-SD is the

responsibility of the entire business community. So-called business responsibility might therefore

be considered as coming about through voluntary endeavors, when in fact the responsibility of

businesses is governed by legal obligations. The problem thus arises of how to regulate the world

market and therefore the international norms that might close the gap between the social and

environmental conditions for competition throughout the world.

References

[1] Acquier A., Aggeri F. (2008), “Une généalogie de la pensée managériale sur la

RSE”, Revue française de gestion, N° 180, p. 131-157.

[2] AFNOR (2007), 4ème projet de travail de l’ISO 26000, Doc. DDRS N195 (private distribution).

[3] Aggeri F. et al. (2005), Organiser le développement durable, Vuibert, Paris.

[4] Berger P.L., Luckman T.(2004), “From the social construction of reality”, The

new economic sociology, ed. Dobbin F., Princeton and Oxford, Princeton University Press.

[5] Boltanski L., Chiapello E., (1999), Le nouvel esprit du capitalisme, Gallimard,

Paris. [6] Bowen H.R. (1953), Social Responsibility of the Businessman, New York, Harper

& Row. [7] Brodhag C., Gondran N., Delchet K. (2004), “Du concept à la mise en oeuvre du

développement durable : théorie et pratique autour du guide SD 21000”, VertigO, la revue en sciences de l’environnement, Vol. 5, N° 2, 12 p.

[8] Capron M., Quairel-Lanoizeléee F., (2007), La responsabilité d’entreprise, La

Découverte, Paris.

368

[9] Carroll a.b. (1979), “A three dimensional conceptual model of corporate performance”, Academy of Management Review, vol. 4, n° 4, p. 497-505.

[10] European Union Commission (2001), Communication: A sustainable Europe for a

better world: A European strategy for Sustainable Development, Brussels, http://europa.eu

[11] European Union Commission (2001), Green Paper: Promoting a European

framework for Corporate Social Responsibility, Brussels http://europa.eu

[12] European Union Commission (2002), Communication: Corporate Social Responsibility: A business contribution to Sustainable Development, Brussels http://europa.eu

[13] European Union Commission (2006), Communication: Implementing the

Partnership for Growth and Jobs: Making Europe a Pole of Excellence on Corporate Social Responsibility, Brussels http://europa.eu

[14] Commission on Sustainable Development (1998), Responsible entrepreneurship,

Background paper N° 4, 6th session.

[15] Donaldson T., Preston L.E. (1995), “The stakeholder theory of corporation: concepts, evidence and implications” Academy of Management Review,Vol.20, N°1, pp.65-91.

[16] Donaldson T., Dunfee T.W.(1999), Ties that bind: a social contracts approach to

business ethics, Harvard Business School Press.

[17] Elkington J. (1999), Cannibals with Forks. The Triple Bottom Line of 21st Century Business, Capstone Publishing Limited, Oxford.

[18] Etchegoyen A. (2003), “Le sens du développement durable”, Les Echos, 18 March

2003.

[19] Fligstein N. (1997), “Social skill and institutional theory”, American Behavior Scientist, Vol. 40, N° 4, p. 397-406.

[20] Frydman (2007), “Stratégie de responsabilisation des entreprises à l’ère de la

mondialisation” in T. Berns et al. Responsabilités des entreprises et co-régulation, Bruylant, Brussels p.1-50.

[21] Godard O. (2003), “Les multinationales et le développement durable: un jeu

ambigu”, cahier 2005-021, Chaire développement durable, Ecole Polytechnique, Paris.

[22] Godard O., Hommel T. (2005), “Le développement durable de Rio de Janeiro (1992) à Johannesburg (2002)”, cahier 2003-017, Laboratoire d’économétrie, Ecole Polytechnique, Paris.

369

[23] Granovetter M., (2000), Le Marché autrement, Desclée de Brouwer, Paris.

[24] Gross J.P., (2009), “Le libéralisme égalitaire des Jacobins”, Manière de voir, Le

Monde diplomatique, n° 106, August-September, p. 10-13.

[25] ISO (2008), “ISO/TC Committee Draft 26000, ref. ISO/TMB WG SR, http://isotc.iso.org

[26] Leca, Bernard, Battilana, Julie and Eva Boxenbaum. "Agency and Institutions: A

Review of Institutional Entrepreneurship." Harvard Business School Working Paper, No. 08-096, May 2008.

[27] Levy D. and Egan D. (2000), “Corporate politics and climate change”, in Higgott

R.A., Underhill G.R.D. and Bieler A. (eds), Non-state Actors and Authority in the Global System, London, Routledge, p. 138-153.

[28] McWilliams A. et Siegel D. (2001). “Corporate social responsibility: a theory of

the firm [29] perspective”, Academy of Management Review, vol.26, n°1, p. 117-127.

[30] Mercier S. (2006), “La théorie des parties prenantes, une synthèse de la

littérature”, in Bonnafous-Boucher M. et Pesqueux Y., Décider avec les parties prenantes, La Découverte, p. 157-172.

[31] Meyer J.W., Rowan B. (1977), “Institutionalized Organizations: Formal structure

as myth and ceremony”, American Journal of Sociology, vol. 83, n° 2, p. 340-363.

[32] OECD (2000), Guidelines for multinational enterprises, Paris.

[33] UNO (1992), Report of united nations conference on environment and development, RIO de Janeiro 3-14 June

[34] UNO (1997), Programme for the further Implementation of Agenda 21 , 19th

special session

[35] UNO (2002), Implementation plan of the world summit on sustainable development, Johannesburg, A/CONF. 199/20 –17.

[36] Pasquero J. (2005), “La responsabilité sociale de l’entreprise comme objet des

sciences de gestion: un regard historique” in M.F. Turcotte et A. Salmon, Responsabilité sociale et environnementale de l’entreprise, Presses de l’Université du Québec.

[37] Pesqueux Y. (2006), “Pour une évaluation critique de la théorie des parties

prenantes”, in Bonnafous-Boucher M. et Pesqueux Y., Décider avec les parties prenantes, La Découverte, p. 19-40.

370

[38] Polanyi K., (1944), La grande transformation, French trans., 1983, Gallimard, Paris.

[39] UNEP (1998), Responsible entrepreneurship, Background paper N° 4, 6th session.

[40] Samuelson P.A. (1954), “The Pure Theory of Public Expenditure”, Review of

Economics and [41] Statistics, vol. 36, p. 387-389.

[42] Schmidheiny S., Chase R., DeSimone L. (1997), Signal of change, 57 p.,

www.wbcsd.ch

[43] Scott, W.R. (1995), Institutions and Organizations, Thousands Oaks, London, Sage.

[44] Stoeckl N. (2004), “The private costs and benefits of environmental self

regulation: Which firms have most to gain ?”, Business Strategy and Environment, N°13, p.135-155.

[45] Suchman M.C. (1995), “Managing legitimacy: strategic and institutional

approaches”, Academy of Management Review, vol. 20, n° 3, 1995, p. 571-610.

[46] Tubiana L., Severino J.M. (2002), “Biens publics globaux, gouvernance mondiale et aide publique au développement”, in Conseil d’Analyse économique, Gouvernance mondiale, La Documentation Française, Paris, p. 349-373.

[47] Vivien F.-D., (2005), Le développement soutenable, La Découverte, Paris.

[48] Vogel D., (2006), The Market for Virtue, Brooking Institution Press, Washington

D.C.

[49] Wolf D. (2007), “Le concept de développement durable: identification des conditions stratégiques de diffusion de ce type de convention grâce aux jeux évolutionnistes”, Gestion 2000, Nov-Dec. 2007, p .35-47.

[50] Zaccaï E. (2002), Le développement durable. Dynamique et constitution d’un

projet, PIE-Pieter Lang, Brussels.

[51] Zadek S. (1998): “Balancing Performance Ethics and Accountability”, Journal of Business Ethics, N°17, p.1421-1441

371

Biographies

Alain Bensoussan is Distinguished Research Professor and the Director of ICDRiA at the University of Texas at Dallas. He is Professor Emeritus at the University of Paris Dauphine and has an extensive research background in stochastic control, probability and stochastic processes. Professor Bensoussan served as President of National Institute for Research in Computer Science and Control (INRIA) from 1984 to 1996; President of the French Space Agency (CNES) from 1996 to 2003; and Chairman of the European Space Agency (ESA) Council from 1999 to 2002. He is a member of the French Academy of Sciences, French Academy of Technology, Academia Europae, and International Academy of Astronautics. His distinctions include IEEE Fellow, SIAM Fellow, Von Humboldt award, and the NASA public service medal. Professor Bensoussan is a decorated Officer of Legion d’Honneur and Officer Bundes Verdienst Kreuz.

Valentina Bosetti holds a Ph.D. in Computational Mathematics and Operation Research from the Università Statale of Milan and a Master’s Degree in Environmental and Resources Economics from University College, London. At FEEM since 2003, she works as a modeler for the Sustainable Development Program, leading the Climate Change topic and coordinating a research group on numerical analysis of carbon mitigation options and policies. She has also collaborated with a number of other institutes such as the Euro-Mediterranean Center on Climate Change, the NOAA and Italian Universities. Her main research interest is socio-economic modeling of climate change with particular emphasis on innovation, uncertainty and irreversibility. She is currently a visiting researcher at Princeton Environmental Institute.

Gunther Capelle-Blancard is Professor of Economics at the University Paris 1 Panthéon-Sorbonne where he is Director of the Master “Monnaie-Banque-Finance”. He is also Deputy Director of the CEPII, a French research center in international economics. He was previously scientific advisor to the Prime Minister's French Council of Economic Analysis. His primary areas of interest are about financial markets and financial intermediaries. Recently, he has explored some issues at the frontier between finance and environment and published articles in several academic journals including the Journal of Environmental Economics and Management and the Journal of Investing.

Michel Capron (Doctor in management science, entitled to supervise research) is Professor Emeritus and researcher at the Institut de Recherche en Gestion, Paris-Est University. Over the last ten years, he has carried out most of his research on corporate social responsibility and sustainable development. He has published numerous papers on these topics and two books (with Françoise Quairel-Lanoizeléee). He is the Chairman of the Scientific Committee of the International Network of Research on Organisations and Sustainable Development (INROSD). He is a member of the French delegation for ISO 26000 negotiations.

Carlo Carraro holds a Ph.D. from Princeton University and is currently President of the University of Venice. He is Director of Research of the Fondazione ENI Enrico Mattei and Research Fellow of CEPS (Centre for European Policy Studies, Brussels), CEPR (Centre for Economic Policy Research, London), and CESifo (Centre for Economic Studies). He is Director

372

of the Climate Impacts and Policy Division of the EuroMediterranean Centre on Climate Change (CMCC) and Vice Chair of IPCC WG III.

Ivar Ekeland is Director of the Pacific Institute of the Mathematical Sciences (PIMS) and the Canada Research Chair in Mathematical Economics at the University of British Columbia. He is a former President of Université Paris-Dauphine, and a former Director of the research centers CEREMADE and Institut de Finance de Dauphine. He has received prizes from the French Academy of Sciences, the French Mathematical Society, and the Belgian Academy of Sciences. He is a foreign member of the Norwegian Academy of Sciences and holds honorary doctorates from UBC and from the University of Saint-Petersburg for Economics and Finance. Ivar Ekeland is the founding editor of the “Annales de l’Institut Henri Poincaré-Analyse non linéaire” and is a member of the editorial board of many other publications. He has also written several books which are reflections on, or popularizations of, mathematics. For these contributions, Dr Ekeland was awarded the “Prix Jean Rostand” by the Association des Ecrivains Scientifiques de France and the “Prix d’Alembert” by the Societé Mathématique de France. He is also a regular contributor to the journal Nature as well as to the magazine Pour la Science.

Sam Fankhauser is a Principal Research Fellow at the Grantham Research Institute on Climate Change at the London School of Economics, and Chief Economist of Globe International, the international legislator forum. He is also a member of the UK Committee on Climate Change, an independent public body that advises the British government on its greenhouse gas targets. A former Deputy Chief Economist at the European Bank for Reconstruction and Development (EBRD), he served on the 1995, 2001 and 2007 assessments of the Intergovernmental Panel on Climate Change (IPCC). He studied economics at the University of Berne and the London School of Economics, and holds a Ph.D. from University College, London.

Nadir Farhi prepared his doctoral work in modeling and control of road traffic at the INRIA Center in Paris. He received his Ph.D. from the University of Paris 1 Panthéon - Sorbonne and currently researches water resource management as a postdoctoral research fellow with the International Center for Decision and Risk Analysis, the University of Texas at Dallas.

Damien Fessler holds a doctorate in economics, and carries out research on the history of economic ideas and economic epistemology. He is currently engaged in post-doctoral work within the Chaire “Finance et Développement Durable” at Université Paris-Dauphine, in which he occupies the position of General Secretary.

Pierre-Noël Giraud, graduated from École Polytechnique, École des Mines and Université de la Sorbonne. In 1978, he set up and headed the CERNA (Research centre in industrial economics) at ParisTech/École des Mines de Paris. He is now Professor of Economics at Mines ParisTech and at Université Paris-Dauphine. Adviser to the French government and the European Commission in the fields of environmental and industrial policies, he is member of the French Academy of Technologies. His research work first focused on the economics of commodities, in particular energy, and environmental economics. For more than ten years, he has also worked on the globalization of firms and its macroeconomic and political consequences, as well as on urbanization in the emerging countries. He has written seven books and many articles on these topics.

Olivier Godard is senior researcher with the Centre National de la Recherche Scientifique (France) (beginning in 1979), and Associate Professor, Department of Humanities and Social Sciences of Ecole Polytechnique (Paris) (beginning 2003). He holds a Ph.D. in economics from

373

University of Paris-Sorbonne (Paris 1) (1993). His recent fields of research include the theory of sustainable development, policy-making under uncertainty and controversies, the precautionary principle (environmental issues, food safety), scientific expertise under uncertainty, the use of economic instruments (taxation and tradeable permits) for environmental policies (acid rain, global warming), justification of public action in the environmental field, and intergenerational and international justice in the context of global warming policies.

Olivier Guéant graduated in both Mathematics and Economics from the Ecole Normale Superieure in Paris. He holds a Master’s in Economics from Paris School of Economics and a Ph.D. in Applied Mathematics from Université Paris-Dauphine. Olivier Guéant also graduated in Finance from ENSAE and was a special student in economics at Harvard University. He was a Lecturer in "computer science applied to economics" at Harvard University. He is now partner of MFG R&D and continues to be involved in several projects of the “Finance and Sustainable Development” Chair as well as teaching Microeconomics at Sciences-Po Paris.

Roger Guesnerie is currently holder of the “Théorie économique et organisation sociale” Chair at Collège de France, director of studies at EHESS, and president of the Paris School of Economics. A former student at École Polytechnique and École Nationale des Ponts et Chaussées, he has taught throughout the world. His work in public economics, on the theory of mechanisms and on general equilibrium, has made him one of the most renowned French economists.

Hubert Jeaneau holds a degree in Applied Mathematics from Université-Dauphine. He then undertook a Ph.D. on extra-financial risks while teaching finance, accounting and statistics at the University of Bordeaux. He has conducted various research projects with both the Europlace Institute of Finance and CA Cheuvreux. He is currently working in the asset management industry.

Jan Horst Keppler is Professor of Economics at Université Paris-Dauphine, Senior Researcher at the Centre for the Geopolitics of Energy and Primary Resources (CGEMP) and Co-Director of the Master’s “Energy, Finance, Carbon”. His primary research interest is in the functioning of energy and carbon markets. He regularly advises international organizations and industry on energy and environment issues (EEA, EDF, NYSE Euronext OECD Nuclear Energy Agency, RTE, World Bank, Caisse des Dépôts and others). Recently, Professor Keppler has worked on the costs of power generation as well as on the link between carbon and electricity prices and has co-edited The Econometrics of Energy Systems (Palgrave Macmillan, 2007). He is a member of the Scientific Council of the Finance and Sustainable Development Chair at Dauphine and contributes regularly to energy and environmental issues through academic publications, the press, radio and television.

Delphine Lautier is Professor of finance at Université Paris-Dauphine. A member of the Steering Committee of the “Finance and Sustainable Development” Chair, she is researcher at DRM-Finance (Dauphine) and at the Fime Lab (Dauphine, Crest, EDF). Since 2000, she has also been an associate research fellow at MinesParisTech. Her main areas of research are energy derivative markets and the term structure of commodity prices. She has published a number of books and articles on this topic.

Jean-Michel Lasry is Senior Scientific Advisor at CALYON and Chairman of the Steering Committee of the “Finance and Sustainable Development” Chair. He was previously a member of the Executive Committee of CALYON Markets Activities for four years as well as the Global

374

Head of Research & Capital Management. Previously, he was a Deputy CEO of CPR Bank in Paris for four years. From 1994 to 1996, Jean-Michel Lasry was the CEO of the Caisse Autonome de Refinancement. From 1990 to 1993, he was a member of the Executive Committee of CDC Banking Divisions, as well as a Board Member of CDC Gestion. He worked as a Professor at Université Paris-Dauphine and École Polytechnique for 17 years. He published more than 100 scientific papers in Mathematics and Economics Journals.

Emanuele Massetti is Senior Researcher at the Sustainable Development Unit of Fondazione Eni Enrico Mattei. He holds Ph.D. in Economics from the Catholic University of Milan, an M.Sc. in Economics from University College, London and an M.A. in Economics from Brown University. His main research interests are in Environmental Economics. He is one of the authors of WITCH, an Integrated Assessment Model to study optimal climate change policies. He is also studying the impacts of climate change and optimal adaptation strategies in the agricultural sector. In 2007/2008 he was Research Affiliate at the Yale School of Forestry and Environmental Studies.

Stéphanie Monjon is research fellow at CIRED (Centre International de Recherche sur l’Environnement et le Développement), a CNRS research center specializing in environment economics. She holds a Ph.D. from the University Paris 1 Panthéon-Sorbonne. She was previously visiting fellow at the University College London and senior economist at the French Environmental Agency (ADEME), in charge of the economic issues related to climate change. Her primary areas of interest are about industrial competitiveness and climate policy, carbon markets and more recently socially responsible investment. She is regularly involved in expert networks for the industry or European institutions on climate policy issues.

Timothée Ollivier, ex-student of Ecole Normale Supérieure in Lyon and the French Institute of Forestry, Agricultural and Environmental Engineering (Engref), is currently finishing his Ph.D. in natural resource and development economics at MinesParisTech and Paris-Dauphine University (to be defended in 2009). He works on the link between natural wealth, growth and sustainability, with a focus on Sub-Saharan countries.

Christian de Perthuis is Associate Professor at Université Paris-Dauphine where he directs the Energy-Finance-Climate Master’s program, and is scientific advisor to the Climate Mission, a research team working on the economics of carbon at the Caisse des Dépôts. With contacts built up with various foreign research centers (CEEPR at MIT, University College of Dublin, Stanford University, Öko Institute), he is recognized as a leading international specialist on the economics of climate change. He is the author of numerous articles and books, including Et pour quelques degrés de plus…Nox choix économiques face au risque climatique (Pearson, 2009) and, in collaboration with Frank Convery and Denny Ellerman, Pricing Carbon (Cambridge University Press, February 2010).

Cédric Philibert, born 1954 and a former science journalist, advised the French Environment Minister from 1988 to 1990. In 1990 he published two books on climate change and on renewable energies. From 1992 to 1998 he advised the CEO of the French Agency for the Environment and Energy Efficiency, then joined UNEP and, in 2000, the IEA, with responsibility for the “evolution of climate policy”. In 2002 he published the IEA’s Beyond Kyoto with Jonathan Pershing. In 2005 he co-authored with Richard Baron the IEA’s publication Act Locally Trade Globally. Since 2007, Cédric Philibert has also been teaching Energy and Climate at Sciences-Po, Paris. In April 2009 he moved into the Renewable Energy Division at the IEA. Qualified in political sciences, he studied economics and published numerous papers in peer-

375

reviewed and other journals. He is married and has three children.

Michael H. Prager received a BS in Humanities and Science from the Massachusetts Institute of Technology, and a Ph.D. in Oceanography from the University of Rhode Island. His research interests include fishery management, stock assessment, population ecology, and operations research. Since 1987, he has been with the U.S. National Marine Fisheries Service (part of the National Oceanic and Atmospheric Administration) as a stock-assessment biologist, leader of two stock-assessment groups, and now, a senior scientist in the Southeast Fisheries Science Center.

Françoise Quairel is an Associate Professor and a member of the research centre DRM (Dauphine Research in Management) at Paris-Dauphine University. Her teaching and Research interests are in Corporate Social Responsibility. She is member of ISO 26000 French Committee (AFNOR). She has published numerous articles and contributions to books. In particular, she is co-author with M. Capron of two books: Mythes et réalités de l’entreprise responsible (La Découverte, Paris, 2004) and Responsabilité sociale de l’entreprise (Collection Repères, La Découverte, Paris, 2007).

Alessandra Sgobbi holds a PhD in Analysis and Governance of Sustainable Development at School for Advanced Studies in Venice Foundation (Italy) and a M.Sc. in Environment and Resource Economics, University College London (UK). She collaborates with FEEM on various projects in the field of the natural resources management and climate change modelling and policy. Currently, she works at the European Commission, EuropeAid Cooperation Office, focusing on development interventions in the fields of climate change adaptation, sustainable consumption and production, energy efficiency and the “grey” environment.

Kyle W. Shertzer received a Ph.D. in Biomathematics in 2001 from North Carolina State University. His research interests are in fish stock assessment, fishery management, and theoretical ecology. He works as a stock-assessment biologist for the U.S. National Marine Fisheries Service in Beaufort, North Carolina.

Dr Ussif Rashid Sumaila is Director of the Fisheries Centre, the University of British Columbia, Vancouver, Canada. He is deeply interested in how economics, through integration with ecology and other disciplines, can be used to help ensure that environmental resources are sustainably managed for the benefit of all generations. Sumaila has authored several articles and has won various awards including the Leopold Leadership Fellowship, the Pew Fellowship for Marine Conservation, and the Peter Wall Institute Senior Early Career Scholar Award. He has given talks at the UN, the White House, the U.S. Congress, the Canadian Parliament and the WTO. His work has been cited by the Economist, the Boston Globe, the International Herald Tribune, Maine Sunday Telegram, the Financial Times, the Globe and Mail, VOA, CBC News and the Vancouver Sun.

Massimo Tavoni is research associate at the Princeton Environmental Institute and senior researcher at FEEM. His research concerns energy and climate change economics. He focuses on the evaluation of international climate mitigation policies, with a focus on technological evolution and uncertainty, and the role of tropical deforestation. He is also interested in the consumption patterns of environmentally stressful goods, especially in countries in economic transition. Massimo holds a Laurea cum Laude in Engineering from the University of Bologna, an M.Sc. in Mathematical Economics from the London School of Economics, and a Ph.D. in Political Economics from the Catholic University of Milan.

376

Stéphane Voisin is head of SRI Research at Crédit Agricole Cheuvreux, the European equity brokerage and research house, which is recognized throughout Europe for the quality and independence of its research. Stéphane Voisin holds an MBA in law and finance with 20 years experience in equity markets. Prior to joining CA Cheuvreux, Stéphane Voisin was an executive director at JP Morgan and at Paribas in London. He has authored a number of articles and reports on social, governance and environmental themes and he chairs various advisory and scientific committees for NGOs and social initiatives.

Dr Carl Walters is a Professor at the Fisheries Centre whose areas of research include the development of rapid techniques for teaching systems analysis and mathematical modeling to biologists and resource managers. A member of several of NSERC's grant committees since 1970, he has done extensive fisheries advisory work for public agencies and industrial groups. He has also conducted over two dozen three-to-ten-day workshops in the past decade, for the International Canadian Fisheries Service, US Fish and Wildlife Service and the International Institute for Applied Systems Analysis. In 1992, he gave the keynote address to the American Fisheries Society, entitled: Where have all the Coho Gone? He is the editor of The Open Fish Journal and has been on the editorial boards of the Journal of Applied Mathematics and Computation, the Northwest Environmental Journal, the Canadian Journal of Fisheries and Aquatic Sciences, and Marine and Coastal Fisheries. Dr. Walters is a Fellow of The Royal Society of Canada.

377

[QUATRIÈME DE COUVERTURE]

This book brings together the lecture notes from the seminars held at the Finance and Sustainable

Development Chair from 2007 to 2009. The first part is concerned with climate change policies, with

particular emphasis on the relevance of traditional economic theories. The second part deals with economic

concepts and methodologies associated with sustainable development. It presents the precautionary

principle, discusses the concept of externalities and introduces the methodology of mean field games. The

third part is devoted to models and empirical applications in various fields, such as water resources, gas

emissions, and a sustainable development path. The fourth part deals with carbon markets, along with their

theoretical justification and historical development. The last part focuses on socially responsible investment

and gives insights in regard to its definition, its links with sustainable development, and the way it could be

used for investment strategies.

The Finance and Sustainable Development Chair is a meeting point for debate, reflection and

research. It brings together researchers, thinkers and international actors involved in the fields of finance

and sustainable development. The Chair was created under the aegis of the Europlace Finance Institute

Foundation. Its academic partners are the University Paris-Dauphine and Ecole Polytechnique. It is

supported by Calyon, EDF, and Caisse des Dépôts et Consignations.

Jean-Michel Lasry is senior scientific adviser at Calyon and chairman of the steering committee on the International Research Network’s Finance and Sustainable Development program. Delphine Lautier is professor of finance at Université Paris-Dauphine. Damien Fessler is junior researcher and coordinator of the International Research Network’s Finance and Sustainable Development program at Université Paris-Dauphine.