project-based mechanisms for emissions reductions: balancing trade-offs with baselines

17
Energy Policy 33 (2005) 1807–1823 Project-based mechanisms for emissions reductions: balancing trade-offs with baselines Carolyn Fischer* Energy and Natural Resources Division, Resources for the Future, 1616 P Street, NW, Washington, DC 20036, USA Abstract Project-based mechanisms for emissions reductions credits, like the Clean Development Mechanism, pose important challenges for policy design because of several inherent characteristics. Participation is voluntary, so it will not occur without sufficient credits. Evaluating reductions requires assigning an emissions baseline for a counterfactual that cannot be measured. Some investments have both economic and environmental benefits and might occur anyway. Uncertainty surrounds both emissions and investment returns, and parties to the project are likely to have more information than the certifying authority. The certifying agent is limited in its ability to design a contract that would reveal investment intentions. As a result, rules for benchmarking emissions may be systematically biased to overallocate, and they also risk creating inefficient investment incentives. This paper evaluates, in a situation with asymmetric information, the efficacy of the main baseline rules currently under consideration: historical emissions, an average industry emissions standard, and expected emissions. r 2004 Elsevier Ltd. All rights reserved. JEL classification: D8; Q4 Keywords: Clean Development Mechanism; Asymmetric information; Tradable emissions permits 1. Introduction The Kyoto Protocol set legally binding emissions targets for industrialized countries (Annex I Parties), amounting to a reduction from 1990 levels of at least 5% by 2008–2012. To help meet these ends, it envisioned flexibility mechanisms that include not only emissions trading among participating countries, but also project- based emissions reductions in countries or sectors not subject to an emissions cap. The Clean Development Mechanism (CDM) allows investment projects that reduce or sequester carbon emissions in developing (non-Annex 1) countries to generate Certified Emissions Reductions (CERs) that count toward compliance goals in developed countries. 1 Similarly, ‘‘Joint implementa- tion’’ provides for Annex I Parties to implement projects that reduce emissions in other industrialized countries, in return for emission reduction units. 2 By exploiting these flexibility mechanisms, adherents expect to reduce substantially the costs of complying with the targets. The CDM also forms the key instrument to have the developing countries become part of the Kyoto Protocol. Offset programs are not new to emissions trading. In the US, they have been incorporated into effluent compliance and watershed trading programs, particu- larly for new or expanding sources. The RECLAIM program for the Los Angeles air basin has allowed the scrapping of old vehicles to generate mobile source emission reduction credits to offset stationary source emissions of criteria air pollutants. 3 However, the Kyoto Protocol elevates project-based mechanisms to a whole new scale in terms of potential emissions reductions and in its international scope. Although such projects may offer significant gains in terms of lower cost abatement opportunities—as well as development benefits in the case of CDM—they also pose important challenges for policy design if an additional goal is to maintain the stringency of the emissions cap. These issues are also ARTICLE IN PRESS *Tel.: +1-2023285012; fax: +1-2023285024. E-mail address: [email protected] (C. Fischer). 1 Article 12. 2 Article 6. 3 RECLAIM stands for ‘‘Regional Clean Air Incentives Market’’ and is administered by the South Coast Air Quality Management District. 0301-4215/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.enpol.2004.02.016

Upload: carolyn-fischer

Post on 21-Jun-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

Energy Policy 33 (2005) 1807–1823

ARTICLE IN PRESS

*Tel.: +1-202

E-mail addre1Article 12.2Article 6.

0301-4215/$ - see

doi:10.1016/j.enp

Project-based mechanisms for emissions reductions: balancingtrade-offs with baselines

Carolyn Fischer*

Energy and Natural Resources Division, Resources for the Future, 1616 P Street, NW, Washington, DC 20036, USA

Abstract

Project-based mechanisms for emissions reductions credits, like the Clean Development Mechanism, pose important challenges

for policy design because of several inherent characteristics. Participation is voluntary, so it will not occur without sufficient credits.

Evaluating reductions requires assigning an emissions baseline for a counterfactual that cannot be measured. Some investments have

both economic and environmental benefits and might occur anyway. Uncertainty surrounds both emissions and investment returns,

and parties to the project are likely to have more information than the certifying authority. The certifying agent is limited in its

ability to design a contract that would reveal investment intentions. As a result, rules for benchmarking emissions may be

systematically biased to overallocate, and they also risk creating inefficient investment incentives. This paper evaluates, in a situation

with asymmetric information, the efficacy of the main baseline rules currently under consideration: historical emissions, an average

industry emissions standard, and expected emissions.

r 2004 Elsevier Ltd. All rights reserved.

JEL classification: D8; Q4

Keywords: Clean Development Mechanism; Asymmetric information; Tradable emissions permits

1. Introduction

The Kyoto Protocol set legally binding emissionstargets for industrialized countries (Annex I Parties),amounting to a reduction from 1990 levels of at least5% by 2008–2012. To help meet these ends, it envisionedflexibility mechanisms that include not only emissionstrading among participating countries, but also project-based emissions reductions in countries or sectors notsubject to an emissions cap. The Clean DevelopmentMechanism (CDM) allows investment projects thatreduce or sequester carbon emissions in developing(non-Annex 1) countries to generate Certified EmissionsReductions (CERs) that count toward compliance goalsin developed countries.1 Similarly, ‘‘Joint implementa-tion’’ provides for Annex I Parties to implement projectsthat reduce emissions in other industrialized countries,in return for emission reduction units.2 By exploitingthese flexibility mechanisms, adherents expect to reduce

3285012; fax: +1-2023285024.

ss: [email protected] (C. Fischer).

front matter r 2004 Elsevier Ltd. All rights reserved.

ol.2004.02.016

substantially the costs of complying with the targets.The CDM also forms the key instrument to havethe developing countries become part of the KyotoProtocol.Offset programs are not new to emissions trading. In

the US, they have been incorporated into effluentcompliance and watershed trading programs, particu-larly for new or expanding sources. The RECLAIMprogram for the Los Angeles air basin has allowed thescrapping of old vehicles to generate mobile sourceemission reduction credits to offset stationary sourceemissions of criteria air pollutants.3 However, the KyotoProtocol elevates project-based mechanisms to a wholenew scale in terms of potential emissions reductions andin its international scope. Although such projects mayoffer significant gains in terms of lower cost abatementopportunities—as well as development benefits in thecase of CDM—they also pose important challenges forpolicy design if an additional goal is to maintain thestringency of the emissions cap. These issues are also

3RECLAIM stands for ‘‘Regional Clean Air Incentives Market’’

and is administered by the South Coast Air Quality Management

District.

ARTICLE IN PRESSC. Fischer / Energy Policy 33 (2005) 1807–18231808

important for the value of investments in mitigationprojects and CERs, which are already being under-taken by energy companies, governments, and otherorganizations.The challenges arise from certain characteristics

inherent to CDM and project-based mechanisms moregenerally.4 First of all, CDM participation is voluntary,and the emissions credits determine both the incentive toparticipate and the effective cap on emissions. Second,many emissions-reducing projects provide other benefitsbeyond the value of emissions credits, so some of thoseinvestments might occur anyway. Third, evaluatingproject-based reductions requires assigning a baselinefor the emissions that would occur in the absence of theproject; that counterfactual cannot be measured, andconsiderable uncertainty surrounds it. Fourth, access toinformation will likely be asymmetric: the third-partymonitor that will certify emissions reductions will knowless about the project fundamentals than the investingand recipient parties. Fifth, the certifying agent islimited in its ability to design a contract that wouldelicit truthful information from CDM participantsregarding their investment intentions. The certificationauthority can only set the amount of abatement credits,while market forces determine the value of emissionsreductions.The combination of those factors means that rules for

determining baseline emissions—the benchmark againstwhich actual emissions will be measured and reductionswill be certified—may be systematically biased. Thecosts of poor baseline determination range from under-allocation, which means forgoing some worthy projects,to overallocation, which expands the global emissionscap and may encourage some unjustified investments.This paper uses a stylized model to evaluate, in asituation of uncertainty and asymmetric information,the efficacy of the main baseline rules currently underconsideration:

(1)

4F

case.

historical emissions;

(2) an average emissions standard for the industry; and (3) expected emissions.

5 ‘‘Baseline’’ is the official terminology of the CDM for the

benchmark emissions allocation of a project. We use this term as

such throughout the paper. In comparison to the traditional use of a

The first and third methodologies rely on project-specific information, while the second uses an industry-wide standard to generate a benchmark. The analysisindicates that using individual emissions histories as abaseline generally provides good investment incentives,but it may substantially overallocate to those that wouldfind such investments profitable in the absence of aCDM policy. Policies that rely on industry-wideaverages overallocate to some firms and underallocateto others, resulting in poorer investment incentives, lesscost-effective abatement, and possibly allocation of too

rom this point forward, however, we will emphasize the CDM

many CERs unless the standard is sufficiently strict. Thelatter two problems are exacerbated when firms canparticipate in the program without making any requiredinvestments.A baseline of expected emissions is formed by

gathering project-specific information to make a reason-able estimate of future emissions. This method strikes acertain balance between the historical and the averagingmethods. Increased accuracy reduces the inefficiencies ininvestment incentives compared with averaging andreduces overallocation compared with historical emis-sions. However, this method would require that thecertifying authority has access to necessary informationat costs that do not outweigh the benefits of greateraccuracy.The next section presents background for CDM

projects and the economic and uncertainty issues thatare likely to accompany them. The subsequent sectiondevelops a stylized model to assess the impact ofbaseline allocations on incentives to engage in emis-sions-reducing investments, given information asymme-tries. The final section discusses the welfare implicationsof different policies for baseline determination.5

2. Background

The CDM, as conceived in the Kyoto Protocol, hasdual objectives. One is to help the parties in Annex I (thedeveloped countries) achieve their commitment targetsat a lower cost than relying fully on efforts conducted athome. Another objective is to provide sustainabledevelopment opportunities for the non-Annex I parties.To receive CER, projects are expected to offer realemissions reductions as well as benefits accruing to hostcountries, such as the transfer of environmentally soundtechnology and know-how (Art. 12).Both the quantity of abatement to be certified and the

benefits to the developing country partner would beevaluated on a project-by-project basis. For the latterevaluation, the type of benefits considered and recog-nized may be important. For example, a requirement fortechnology to be transferred could imply a very differentset of potential projects than merely allowing for amonetary transfer in exchange for reductions. For thetypes of projects that serve dual purposes of productiv-ity enhancement and environmental improvement, asstipulated by the protocol, the evaluation of emissionsreductions may then be quite complicated because of theproblem of disentangling influences.

‘‘baseline scenario’’—which in this case would be no CDM—these

baseline rules in effect offer a standard methodology for approximat-

ing the emissions in that no-policy scenario.

ARTICLE IN PRESSC. Fischer / Energy Policy 33 (2005) 1807–1823 1809

In principle, ensuring compliance in Annex I countries isa relatively straightforward task: because they are subjectto an overall emissions cap, one needs to monitor onlyemissions.6 Knowing what would have happened in theabsence of the cap helps in estimating the opportunitycosts to the economy, but it is not necessary forcompliance (except, perhaps, in the thorny area of carbonsinks). The CDM, however, does not operate under a cap;because it is a project-by-project mechanism, to certifyactual reductions to credit against Annex I caps, one mustknow not only actual emissions but also the emissions thatwould have occurred in the absence of the project. Theprocedures regarding additionality and baselines for aCDM are detailed in Annex decision 17/CP.7:7

48. In choosing a baseline methodology for a projectactivity, project participants shall select from amongthe following approaches the one deemed mostappropriate for the project activity, taking intoaccount any guidance by the Executive Board, andjustify the appropriateness of their choice:

(a)

6F

track

coun

(Rus7U

Existing actual or historical emissions, as applic-able; or

(b)

Emissions from a technology that represents aneconomically attractive course of action, taking intoaccount barriers to investment; or

(c)

The average emissions of similar project activitiesundertaken in the previous five years, in similarsocial, economic, environmental and technologicalcircumstances, and whose performance is amongthe top 20 per cent of their category.

The options are represented here in a stylized form ashistorical, expected, and average emissions, respectively.Historical emissions are self-evident, and probably thesimplest to calculate. Assessing the ‘‘economically attrac-tive course’’ represents an attempt to generate expectationsabout what baseline emissions would have been, takinginto account pre-existing investment incentives. Finally,average emissions use similar, non-CDM projects as abaseline; restricting the pool of comparable projects to topperformers then serves to reduce the average allocation.The intent is to encourage projects with lower GHGintensity than top performers and to ensure that projectsthat would occur on their own do not get allocations.

2.1. Baselines and welfare

True additionality requires that the allocated baselinenot exceed what would have happened had the firm not

or the most part, emissions monitoring can be achieved by

ing the entry of fossil fuels into the economy; most Annex I

tries have sufficient infrastructure for compiling accurate data

sia may be a notable exception).

NFCCC (2001).

participated in CDM. Unfortunately, that informationis highly uncertain. As a result, the certifying authorityruns the risk of guessing too high or too low in awardingCERs (equivalently, guessing too low or too high onbaseline emissions). Guessing too high a baseline andgiving away too many CERs means expanding theoverall emissions cap for Annex I countries, whichimplies some marginal increase in the potential damagesfrom climate change. Guessing too low a baseline, onthe other hand, risks forgoing the benefits of a cost-effective greenhouse gas-reduction project and valuablelocal benefits due to insufficient compensation.Consider the baseline problem from the vantage point

of the global community, ignoring for now issues of theinternational distribution of effort, benefits, and costs.In weighing the risks of under- or overallocation, onemust consider the net costs or benefits of expanding thecap compared with the cost savings from shifting thelocation of abatement effort. If the true baseline isallocated, then overall emissions do not change. How-ever, in shifting some abatement from areas where themarginal costs are high to areas where marginalabatement costs are low (even zero), we have made thetotal costs of achieving that overall cap strictly lower.Therefore, total global welfare is strictly higher. Acorollary to this result is that one can allocate somewhatmore than the true baseline without lowering welfare.How much more depends on whether the marginalbenefits from reducing emissions (reduced potentialclimate change burden) are higher or lower than themarginal costs of reducing emissions. If they areapproximately equal, the extra damages from looseningthe cap are roughly offset by the cost savings from doingless abatement. If marginal costs are higher, welfareimproves from loosening the too-tight cap. If marginaldamages are higher, welfare is lost by expanding the cap,and fewer permits in excess of the baseline can beallocated without lowering overall welfare.The full cost of inappropriate baseline determination

includes both economic and environmental costs.Receipt of the baseline allocation is conditional onparticipation, and participation is also conditional ontechnology transfers; all these decisions are voluntaryand depend in part on the attractiveness of theallocation. Thus, baseline rules do not merely redis-tribute the rents and compliance costs of greenhouse gaspolicy, they may also introduce complex incentives forpotential participants. The cost of inappropriate in-centives is not just unintended redistribution of wealthin permits but wasted resources.

2.2. Uncertain and unequal information

The critical problem is that although actual emissionscan be verified, actual reductions cannot. One cannot

ARTICLE IN PRESSC. Fischer / Energy Policy 33 (2005) 1807–18231810

predict what baseline emissions would have been anybetter than one can predict what emissions will be.Many factors contribute to uncertainty about the

baseline and about the profitability of an investmentproject. Fuel prices may fluctuate, consumer tastes maychange, exchange rate movements can affect exportingindustries, regulation or deregulation can affect elec-tricity production and prices, and other macroeconomicvariables have an impact on demand for energy-intensive products. Many of these factors are just asuncertain to the firm or project manager as to theauthority that certifies the emissions reductions. If thatwere all, the certifying authority could assign a baselinerule based on mutually shared expectations, and theinvestor would act according to those expectations.In many instances, however, the participants in a

project (hereafter ‘‘the firm’’) may have better informa-tion than the certifying authority.8 Firms may differ inthe state of their existing production capital, in theirmanagement’s ability to implement a change in technol-ogy, or in local demand conditions. The profitability ofa given technology transfer can vary across firms, andthe firms are more likely to know what their particulareconomic gains are. If that information is private andcannot be obtained without significant cost, a discre-pancy is created between the certifying authority’sexpectations about what the firm would do in theabsence of the CDM project and the firm’s ownexpectations. This discrepancy poses serious challengesfor designing baseline rules.This baseline issue is complicated by the dual nature

of many technologies. Some technologies are single-purpose and reduce emissions without significantlyaffecting the rest of the production process: the retrofitof scrubbers to remove pollutants at the ‘‘end of thepipe,’’ for example. However, these options are essen-tially non-existent for greenhouse gas emissions. Typi-cally, the reduction of greenhouse gases entails a majorchange in the use of fossil fuel inputs and their relatedproducts. Many investments that would produce sig-nificant emissions reductions through major productionprocess changes could also then produce significanteconomic gains, and vice versa. For example, animproved production technology may also be moreenergy efficient, reducing both emissions and the costs offuel inputs. Although productivity increases and emis-sions decreases are both good things, determiningenvironmental additionality can then be tricky.

8Project participants may include foreign direct investors, host

country governments, and/or firms. We focus on cases where these

actors have better information about the project than the authority

certifying the emissions reductions. Others consider cases where

information asymmetries exist between the parties to the project

themselves (e.g., the host country may have better knowledge of the

true profitability than the foreign investor) see Hagem (1996).

Investment in productive capital or new technology iscostly, and the project investors weigh the economicvalue of the project against those costs. Some projectswill pass this economic test and go forward regardless ofclimate change policy. However, in addition to theeconomic benefits, there may also be environmentalbenefits to the project. The firm will not value them inthe absence of some intervention or opportunity, such asjoining a CDM program. If a firm participates in aCDM project, the total return then equals the economicvalue plus the sales of CERs. The baseline allocationthus determines whether otherwise unprofitable projectsare implemented.On the other hand, not all reductions require

significant fixed costs or new technologies—changes invariable production factors can also reduce emissions.Such alternatives might not be profitable on their own,but would be worthwhile in return for emission credits.Although the CDM is intended to be a mechanism fortechnology transfer, it is unclear to what extent a majorinvestment in equipment or training is a prerequisite forconsideration as a CDM project. A developing country(or company) might also on its own be able to adjust itsbehavior to reduce fossil fuel use, if allowed toparticipate and receive credits. Thus, both the designof the baseline allocation rules and the participationrequirements are important for determining who par-takes in CDM projects, how many additional reductionsare achieved, and how many are certified.

2.3. CDM contracts and asymmetric information

Asymmetry in access to information is a potentiallyserious problem for calculating CERs. Because partici-pation in CDM projects is voluntary, certain baselineallocation methods run the risk of selection bias––attracting participants who would be predisposed tomaking such investments and having low emissionsanyway. For example, suppose CDM candidates have agood idea what their baseline emissions would be, butthe certifying authority does not. A firm that is planningto be a lower emitter would like to pretend to be a higherfuture emitter, in order to get a larger allocation.9

Economic theory offers ideas for designing contractsto induce project proponents to reduce such distortion,or even to fully reveal what type they are.10 Normally,this involves offering the firm a menu of differentcombinations of quantities (e.g., production, effort, orabatement) and corresponding prices. The two contract

9Wirl et al. (1998) show that in the absence of an exogenously set

baseline, countries participating in a jointly implemented project have

incentives to cheat in announcing their intentions, both to each other

and to the certifying authority.10Much of the theory of optimal contract design is derived from the

theory of optimal regulation of monopolies. See Laffont and Tirole

(1993) and Loeb and Magat (1979).

ARTICLE IN PRESSC. Fischer / Energy Policy 33 (2005) 1807–1823 1811

variables (price and quantity) are used to identify firmswith different preferences over those aspects. In thiscontext, each contract in the menu would have to specifya certain fixed transfer payment (and thereby a certainprice) for each fixed amount of abatement.The attention given to project-based mechanisms in

the literature has focused mostly on the impact ofasymmetric information and costly enforcement be-tween the project partners in a bilateral exchange.11 It isgenerally assumed that abatement can be measured expost, information about abatement costs is incomplete,and the investor has declining marginal benefits toabatement.12 Three problems make these kinds oftraditional models either impractical or inapplicable toa decentralized CDM with third-party validation.First, in a competitive market for CDM projects and

other fungible international carbon credits and allow-ances, the price of emissions reductions is likely to befixed by international markets; a contracting partywould find it difficult to deviate from prevailing permitprices. In fact, these revelation methods are unnecessaryif the investing country’s abatement cost function isrelatively flat in the area of the corresponding abatementamount. It could offer a constant price per unit ofabatement (the international market price), and anefficient amount would be achieved in the absence ofany information about the agent’s costs: this is thebeauty of decentralized markets. Bargaining with‘‘menus,’’ as described above, is useful only if the goalis to obtain part of the inframarginal rents or if marginalbenefits are declining.13

Second, the certifying authority is not designing a fullcontract for abatement. It determines the categories ofprojects that are eligible—coal plant modernizationversus renewable energy—and the procedures by whichactual emissions reductions are calculated and certified.In defining the baseline against which abatement ismeasured, the certifying authority cannot induce twotypes of eligible projects to make the different conditionsknown by choosing different packages of provisions,since a larger baseline is always preferred, regardless oftype. A believable information gathering and punish-ment mechanism would be necessary to design proce-

11For example, Hagem (1996) considers the problem of contract

design for an investing country firm that does not know the true cost to

the host country partner of conducting abatement. Liski and

Virrankoski (forthcoming) focus on transaction costs in bilateral

bargaining. Janssen (1999) considers third-party enforcement of

bilateral transactions.12Millock and Hourcade (2001) are an exception, introducing

uncertainty over abatement levels into the bilateral framework.13The existence of these competitive opportunities is why developing

countries’ CER suppliers are better off creating credits for competitive

sale than with isolated bilateral deals where foreign investor can

capture much surplus. See Narain and Van ‘t Veld (2001). Babu and

Saha (1996) look at the effect of bargaining power on the distribution

of gains from bilateral negotiations over abatement and price.

dures that induce the revelation of firm-specific factorsthat affect baselines.Third, and most problematic, these revelation mechan-

isms assume that actual abatement can be established,albeit with a cost to monitoring. Unfortunately, thetarget of the necessary monitoring and enforcementmechanism is a hypothetical behavior, not an actual one.Although actual emissions can be observed, emissions inthe absence of the project cannot, and actual abatementmay therefore be impossible to know.14

3. Allocation and the investor: a stylized model

Since project-by-project information gathering is verycostly, some general rules for determining project-levelemissions baselines have been proposed: historicalemissions, average emissions, and expected emissions.Each rule has its own biases in the departure from actualbaselines and in incentives to participate. To evaluatetheir effects, we develop in this section a model thatcharacterizes the incentives of the investing party, whichfor simplicity we call ‘‘the firm.’’ Projects may of coursebe undertaken by a variety of entities, includinggovernments, joint partnerships, multinational corpora-tions, and local businesses; the key assumption is merelythat the parties undertaking the investment and abate-ment activities are distinct from the authority certifyingthe amount of emissions reductions.15

A formal analytical model of the firm’s investmentproblem is developed first for a general case, and thenapplied to the specific baseline allocation rules. Weanalyze in particular the role of fixed costs in thedecision to participate in a CDM project. In the firstscenario, a significant investment is required to partici-pate, representing a technology-transfer mandate. In asecond scenario, smaller projects without major fixedcosts—like process improvements—are also eligible,representing no transfer requirement. We focus oninvestment-related costs and assume the firm does notbear any transactions costs related to monitoring orcertification.16

We use the analytical model to build intuition aboutthe basic incentives created by allocation and participa-tion rules. To demonstrate these effects, the model isstylized and simplifications are made with respect to therange and distribution of potential projects. Subse-quently, numerical simulations are performed to illus-trate how project investment, participation and

14This concern is raised by Bohm (1994).15See also footnote 5.16This assumption would hold if monitoring costs are either

negligible or borne by international sponsors of the certifying

authority. The main effect is to eliminate additional modeling

complexities, and the implications will be discussed after the model

is presented.

ARTICLE IN PRESSC. Fischer / Energy Policy 33 (2005) 1807–18231812

outcomes vary with respect to the effectiveness of theinvestment and its alternatives at reducing emissions.

3.1. Model with investment prerequisite

The investment in a new, cleaner production processbrings two potential payoffs to the firm: (1) cost savingsreflecting reduced use of energy inputs, which we assumeare proportional to the emissions reduced; and (2) anincrease in general productivity, which yields additionalprofits. These profits vary from one investment projectto another because of market conditions, managers’ andworkers’ skills, or technical expertise.To formalize a model of the firm’s decision making,

let these key terms be represented with the followingvariables:

k=

17The

rules, but

the next

investment cost (annualized);

p= economic benefits; m= initial emissions; r= share of emissions remaining after investment; s= savings per unit of emissions reduced; t ¼ market price of emission permits; AðmÞ= baseline allocation.

18 In other words, the share of firms with (or the probability of) mpx

isR x

0 dm ¼ x:19The only difference in these expressions with a non-uniform

Restated mathematically, if the firm makes a fixedinvestment than costs k in annualized terms, it receivesannual flows of economic benefits and energy savings ofpþ ð1� rÞsm: We assume that the firm knows the valueof all those variables with certainty, but the internalprofit variable is unknown to the certifying authority.Initial emissions rates also vary by firm, affecting thetotal returns of the investment, but they are assumed tobe uncorrelated with the firm-specific profit. Thecertifying authority can verify initial emissions. How-ever, without knowing the profit variable—and therebywhether the firm would invest anyway—it cannotdetermine actual baseline emissions.In a general formulation, let AðmÞ be the firm’s

baseline allocation under a given rule, which may be afunction of the initial emissions rate. The CERs, whichcan be sold at the market price of permits, are thedifference between the allocation and remaining emis-sions. Thus, the additional value of participation istðAðmÞ � rmÞ: Although theoretically one could haveunderallocation serious enough to prevent participationwhen an investment already made, we restrict ourconsideration to abatement rates and baseline charac-terizations such that rmoAðmÞ:17 Thus, we focus on thequestion of whether to invest or not.Let #s ¼ ð1� rÞs be the effective rate of energy savings

from the technology. In general, a firm will invest if the

simplification applies to the average and expected emissions

it is not in actuality very restrictive, as will be explained in

part.

gains—the combined profits, energy savings, and theCER sales—outweigh the costs; i.e., if pþ #smþ tðAðmÞ �rmÞ > k: Whether or not an equivalent investmentproject passes this test depends on each firm’s individualcharacteristics, the prices it faces, and the allocationregime.

3.1.1. Equilibrium with uniform distribution

To assess the impact of these factors, we need toevaluate the equilibrium investment and emissions,given a continuum of projects with varying profiles ofeconomic benefits and emissions reductions. In theabsence of knowing what the actual distribution ofpotential projects looks like, consider the followingdemonstration using independent, uniform distributionsof emissions rates and profits, with mA½0; 1�; pA½0; 1�;18

k ¼ 1 and 0oso1: This normalizes the variables to theinvestment cost, so that p is the share of those costsrecouped by profits, m is the share of maximumemissions, and s represents the energy savings in termsof profits from reducing the emissions share. Theassumption that each variable is bounded by 1 in thisexample means that it is not worthwhile to make anyinvestment for profits or energy savings alone; bothmust be present.The uniform distribution has convenient properties.

For example, the share of projects with profits exceedingthe investment threshold is easy to interpret. In theabsence of a CDM policy, given any m; firms withpXð1� #smÞ will invest; with the uniform distribution, #smis the share of those firms with emissions m meeting thisthreshold, or the odds that the particular project willpass the investment hurdle. Thus, average initialemissions are

R 1

0 ðmÞ dm ¼ 1=2; and investment is theintegral of the odds of investing over the range ofemissions rates:

R 1

0 ð#smÞ dm ¼ #s=2:When CDM participation is conditional on invest-

ment, a firm will invest if pXð1� #sm� tðAðmÞ � rmÞÞ:Thus, with the uniform distribution,19 total investmentequals

Z 1

0

ð#smþ tðAðmÞ � rmÞÞ dm ¼#s

2þ t

Z 1

0

AðmÞ dm �r2

� �:

ð1Þ

Thus, investment in all regimes is necessarily higherthan in the no-policy regime, and it is increasing in thecumulative (or average) allocation.

distribution of emissions would be the inclusion of a frequency

distribution function, i.e., f ðmÞ dm: However, if profits are not

uniformly distributed, the expression for the probability of investment

is more complicated.

ARTICLE IN PRESSC. Fischer / Energy Policy 33 (2005) 1807–1823 1813

The total CERs for the industry equal the differencebetween the baseline allocations and actual emissionsfor the investing firms:Z 1

0

ð#smþ tðAðmÞ � rmÞÞ ðAðmÞ � rmÞ dm ð2Þ

and actual emissions reduced equal the differencebetween pre- and post-investment emissions for theinvesting firms:Z 1

0

ð#smþ tðAðmÞ � rmÞÞ ð1� rÞm dm: ð3Þ

The net reductions to global emissions are thedifference between actual reductions from the CDMproject and the CERs, since those are used to offsetreductions in Annex I countries:Z 1

0

ð#smþ tðAðmÞ � rmÞÞ ðm� AðmÞÞ dm: ð4Þ

In other words, net reductions are defined as theoverall impact on global emissions compared with theinitial state with no investment. The difference resultsfrom both the direct reduction in emissions by investorsand the increase in emissions by the recipients of thecertified reductions.

3.1.2. Policy scenarios

We now use this framework to compare five policyscenarios.20

3.1.2.1. No policy. In the absence of CDM, investmentwill go ahead if the profits and energy savings justify it—i.e., if pþ #sm� k > 0: The value of emissions reductionsdoes not play a role (effectively, t ¼ 0).

3.1.2.2. Optimal and additional. Ideally, every firm forwhich the investment is justified from a social standpointwould invest—i.e., if pþ #smþ tð1� rÞm� k > 0: How-ever, to preserve additionality, one would not want toallocate permits to those that would invest under nopolicy. Thus, the firms one would want to target arethose whose emissions are costly enough in environ-mental terms that they would justify investment, eventhough profits and energy savings alone would not: pþ#smþ tð1� rÞm� k > 0 > pþ #sm� k:

3.1.2.3. Historical emissions. The historical emissionsrule allocates to each firm a baseline reflecting itsindividual emissions history, which in this framework isits level of initial emissions m: The firm will then invest ifpþ #smþ tð1� rÞm� k > 0: The full environmental value

20All of the rules at hand could, in theory, be implemented in lump-

sum form, with fixed allocations, or rate-based form, with allocations

depending on production levels. Since we are focusing on the

investment rather than production questions, we abstract from this

issue; some implications are discussed in the conclusion.

of emissions reductions enters into consideration, low-ering the investment hurdle according to the firm’sinitial emissions. Thus, all the firms that can justify theinvestment in terms of social costs and benefits under-take it. Those that would invest under no policy still do,and their profits are raised by the value of thenet allocation of CERs. That net allocation is bydefinition always positive, assuring participation.

3.1.2.4. Average emissions. An industry average base-line offers %m credits to each project joining the program.This average allocation is determined based on industry-wide parameters and is independent of the firm’scharacteristics. We will allow this average to be definedin different potential ways, such as by average initialemissions, or by the average emissions of relatively cleanfirms; the point is that the allocation is identical for allparticipants. Compared to no policy, this baselinelowers the hurdle across the board, without respect tothe individual emissions saved. If investment is aprecondition for participating in the CDM program, afirm will invest and join if pþ #sm� k þ tð %m� rmÞ > 0:Thus, some of the firms with emissions that aresufficiently costly to the environment do not haveenough incentive to invest: tm > k � p� #smþ trm > t %m:Meanwhile, some firms with relatively low emissions aregiven the extra push to invest, something they would nothave done under the historical emissions baseline:tmok � p� #smþ trmot %m:The policy restriction we impose to ensure participa-

tion is that the average emissions standard is set higherthan the post-investment emissions of the highestemitter ( %m and r such that %m > rm for all m). Otherwisestated, we assume the technology is always cleaner thanthe standard.21

3.1.2.5. Expected emissions. Under an expected emis-sions rule, the certifying authority uses the informationavailable to it about the firm’s characteristics to assesswhat emissions would be in the counterfactual, *m:Although these emissions are not actually uncertain,they are unknown to the certifying authority. Theauthority is here assumed to observe initial emissionsand determine expected emissions, knowing the oddsthat an investment would be made in the absence ofpolicy. In other words, expected emissions are definedhere as a weighted average of initial emissions andpost-investment emissions, where the weights derivefrom the probability distribution of profits: Ef *mg ¼ðPrfp > k � #smgÞrmþ ðPrfpok � #smgÞm: Consequently,

21Else some firms would invest but not participate, as under the no-

policy scenario. Then the investment decision is whether pþ sð1�rÞmþ tMax½ %m� rm; 0� > k: For lower reduction rates, some high

emitters might want to invest but not join the program. Since smaller

reductions will be incorporated with variable cost changes, we consider

this additional complication unnecessary at this time.

ARTICLE IN PRESS

Table 1

Investment decisions under different baseline rules

Allocation rule Investment decision

AðmÞ: Invest if

No policy 0 pþ #sm > k

Historical baseline m pþ #smþ tð1� rÞm > k

Industry average baselinea %m pþ #smþ tð %m� rmÞ > k

Expected emissions baseline mð1� sð1� rÞ2mÞ pþ #smþ tðð1� rÞm� sð1� rÞ2m2Þ > k

aWe maintain the assumption that %m > r; meaning that all investments reduce emissions below this standard and are assured a positive

net allocation of CERs.

C. Fischer / Energy Policy 33 (2005) 1807–18231814

the net allocation of CERs is always positive in ourexample. However, we note that if expectations areformed by other means, participation may not beassured. With our uniform distribution, these expecta-tions simplify to a function of the firm’s observableinitial emissions: Ef *mg ¼ mð1� sð1� rÞ2mÞ:22

3.1.3. Policy comparison

Table 1 summarizes the stylized allocation rules andthe decision of the firm to invest and participate in aCDM project.From these allocation rules, we can already see some

of the implications for net reductions as laid out in (4).Since having no policy produces some reductions and noallocations, net reductions are necessarily positive. Sincehistorical emissions set AðmÞ ¼ m; net reductions are bydefinition zero—and thereby less than with no policy.Since expected emissions are less than historical emis-sions (AðmÞom for all m), that rule always producespositive net reductions. Finally, since the averageindustry emissions standard may allocate more or lessthan initial emissions, it can generate positive ornegative net reductions. A question remains whetherany of the baseline rules, though they all generate moreabatement activity in the developing country, actuallygenerate more net reductions than the no-policy case.

3.1.3.1. A clean technology example. The differencesbetween the baseline rules can be illustrated analyticallywith a simple example using an investment in a clean,non-emitting technology.23 For example, the projectcould install wind, solar, or renewable energy technol-ogies for generation, displacing coal-fired plants or otherfossil fuel sources. When the investment reducesemissions to zero, the baseline allocation is equal tothe quantity of CERs granted. Table 2 presents theresults, derived from assuming that r ¼ 0; substitutingthe allocation definitions, and solving for investmentcosts, allocations, and emissions under the different

22Recall from the definition of expected emissions, Ef *mg ¼ ðPrfp >k � #smgÞrm þ ðPrfpok � #smgÞm ¼ ð1 � #smÞmþ #smrm ¼ mð1 � sð1 �rÞ2mÞ:

23While this example may still seem restrictive, it introduces the key

methods and intuition; the effects of incomplete emissions abatement

will be evaluated numerically.

scenarios. Note that net reductions reflect the impact onglobal emissions compared with the initial state, ratherthan with no policy.With no policy, s=2 is the share of firms with high-

enough combined profits and emissions to investregardless, abating emissions by s=3: With the historicalemissions baseline, an additional t=3 units of abatementoccur through more investment. However, both theyand all other investing firms choose to participate inCDM and are allocated their historical emissions. Sinceall covered reductions are offset by extra emissions byCER purchasers, global emissions rise by the amount ofabatement that would have occurred anyway, s=3:The industry average emissions baseline has several

effects. For comparison, consider the standard ofaverage historical emissions, or %m ¼ 1=2: That ruleproduces investment equal to the historical baseline,but the distribution is different. High-emitting firmshave less additional incentive to reduce emissions, whilelower-emitting firms have more incentive to invest.24

Since the investment decision is less sensitive to initialemissions, actual abatement is lower, despite equivalentlevels of investment. Meanwhile, initially high-emittingfirms that would invest anyway are not allocated asmuch, but firms with low initial emissions are over-allocated emissions. With investment as a preconditionfor joining the program, the net effect is a much lowerallocation, which then translates into more net reduc-tions than with the historical baseline—but not morethan with no policy.25 Finally, as the standard isreduced, we see that investment, and with it CERs andabatement, are also reduced—however, net reductionsmay rise. To see if the average allocation can offer trulyadditional reductions, let %m� represent the industryaverage standard that leads to the greatest net emissionsreductions. Maximizing the expression for net reduc-tions with respect to the average emissions standard andsolving, we get %m� ¼ ðt � sÞ=ð4tÞ: Substituting this valueback into that expression for net reductions, we find that

24Unless they have the option to join the program without investing,

which is explored in the next section.25Without that precondition, we will see in the next section that

emissions are higher and generally higher than with the historical

baseline.

ARTICLE IN PRESS

Table 2

Effects of baseline rules on investment in a clean technology

Investment CERs Emissions reduced Net reductions

No policy s

20 s

3

s

3Historical baseline s þ t

2

s þ t

3

s þ t

30

Industry average s

2þ t %m

s

2þ t %m

� �%m

s

t %m2

s

ðtð1� 2 %mÞ � sÞ %m2

Expected emissions s þ t

2�

st

3s þ t

3�

st

2�

s2

s2t

5

s þ t

3�

st

4sðs þ tÞ

4�

s2t

5

26For initial units of reduction, this assumption should be quite

accurate. In the absence of a value to emission reductions, firms should

be just indifferent between conservation costs and energy savings. If

marginal costs of abatement are increasing, it does not reflect the full

costs of the reductions. However, it is a useful simplifying assumption.27This assumption does not significantly change the flavor of the

results. If the variable cost options are not merely alternative but

additive opportunities, participation in a CDM project can induce

firms to find reductions in addition to those afforded by the

investment. This incentive means that some gain is made with the

firms that would have invested anyway, reducing some of the

overallocation problem.

C. Fischer / Energy Policy 33 (2005) 1807–1823 1815

this standard does indeed produce net reductions inexcess of the no-policy case of ðt � sÞ2=ð16tÞ: Further-more, we see that the reductions-maximizing standard ishigher when the emissions price is relatively high, andlower when firms realize relatively more cost savings ontheir own.The expected emissions rule still over-rewards some

firms and under-rewards others, reducing investmentcompared with the historical baseline. However, thedegree of inaccuracy is less than with the industryaverage rules, and the more appropriate investmentresults in lower actual emissions, though still not so lowas the historical baseline. But since overallocation istempered, total emissions may be lower than with nopolicy. Also interesting to note is that, unlike with thehistorical baseline, net reductions are sensitive to thepermit price through the formation of expectations. Onthe other hand, the average emissions rule makes thepermit price a factor only if the standard differs from thehistorical average; in all other cases, the extra incentivefor abatement is fully offset by extra allocations.Perhaps unsurprisingly, allowing the interested parties

to choose their own method of baseline allocation wouldprovide for the greatest amount of investment and ofoverallocation. Since expected emissions are alwayslower than historical emissions, the firm would alwaysprefer either historical or average emissions: Max½ %m;m�:Thus, below-average emitters would want the averagerule and above-average emitters would choose historicalemissions. As a result, this rule would provide all of theinvestment of the historical baseline, plus the additionalinvestment of the average emissions baseline for thoseinitially below the standard. Thus, additional reductionswould be made, but many more CERs would beallocated. Since net reductions are zero for the historicalbaseline and negative for below-average emitters, netreductions would necessarily be negative. Expressedmathematically, net reductions would be

Z %m

0

#smþ tð %m� rmÞð Þðm� %mÞ dm

þZ 1

%mð#smþ tð1� rÞmÞÞðm� mÞ dmo0: ð5Þ

3.2. Model without investment precondition

Opportunities to reduce emissions may exist that donot involve capital investments, like modifying theproduction process or switching to higher-quality fuel.Then, some firms that see little economic benefit toinvesting could be given an incentive to make worth-while emissions reductions, if participation did notdepend on investment.Suppose now that undertaking a fixed investment is

not a requirement for participation in a CDM project.Firms that choose to participate but not invest then faceincentives to use other methods to reduce their emissionsand generate CERs. The option to undertake such cost-effective reductions then also affects the decision ofwhether to invest or not.Consider an alternative method that is available to

reduce emissions by ð1� aÞ without incurring a fixedcost; instead, it involves variable costs, which forsimplicity we assume are proportional to emissionsand just equal to the corresponding energy savings.26

This method can represent the ‘‘low-hanging fruit’’—smaller-scale options that incur no significant net costs,but also reap no net benefits without a reward toemission reductions. We also assume that this method isless effective at reducing emissions than the fixed-costtechnology ðroaÞ; in fact, unlike the investmenttechnology, it may not generate sufficient reductions toguarantee a positive net allocation of CERs under eachof the baseline rules. Furthermore, we assume that thereductions from the investment would be made insteadof, rather than in addition to, those available fromvariable-cost efforts.27

ARTICLE IN PRESSC. Fischer / Energy Policy 33 (2005) 1807–18231816

The firm then faces two decisions. First, the firm willchoose whether to participate. Profits from participationare the net receipts from permit sales, since the othercosts are just outweighed by the energy savings. Thus,the firm will join the program and engage in variable-cost reductions if the allocation it gets would exceed itsremaining emissions: AðmÞ > am:Next, the firm decides whether to invest. If it would

not participate without investing, the investment deci-sion remains identical to that in the previous section. Ifit would participate anyway, the hurdle is now thatprofits from investing must be greater than those fromparticipating, not just greater than zero. In this case, thefirm would invest if pþ #sm� k þ tðAðmÞ � rmÞ >tðAðmÞ � amÞ: Rearranging, we see that the baselineallocation drops out of the investment decision: investif pþ #sm� k þ tða� rÞm > 0: Importantly, this metric isalso the socially efficient investment decision, meaningthat for any baseline rule, if the firm would participateanyway, the decision to invest is always cost effective forthe additional reductions.

3.2.1. Equilibrium without investment prerequisite

When investment is not a prerequisite for the CDM,equilibrium behavior reflects the additional participa-tion decision. Let

%m denote the cutoff level of emissions

below which firms would participate in the CDMwithout investing, given a baseline rule. With theinvestment prerequisite, that cutoff was predeterminedas

%m ¼ 0; without it, the cutoff depends on the allocation

mechanism. With the uniform distribution, total invest-ment equals the share of those who would otherwiseparticipate who find the investment project preferable,plus the share of those who would not otherwiseparticipate who find the project worthwhile:

Z%m

0

ð#smþ tða� rÞmÞ dmþZ 1

%mð#smþ tðAðmÞ � rmÞÞ dm:

ð6Þ

The total CERs for the industry then equalZ%m

0

ðAðmÞ � ð#smþ tða� rÞmÞrm

� ð1� #sm� tða� rÞmÞamÞ dm

þZ 1

%mð#smþ tðAðmÞ � rmÞÞðAðmÞ � rmÞ dm; ð7Þ

while actual emissions reduced equalZ%m

0

ðð#smþ tða� rÞmÞ ð1� rÞm

þ ð1� #sm� tða� rÞmÞ ð1� aÞmÞ dm

þZ 1

mð#smþ tðAðmÞ � rmÞÞ ð1� rÞm dm: ð8Þ

%

3.2.2. Comparing policies

For our different baseline rules, Table 3 summarizesthe participation and investment decisions, using thesame definitions as in the previous section where Ef *mg ¼mð1� sð1� rÞ2mÞ as before. Combining the participationdecision and the baseline definitions gives the participa-tion cutoff levels of emissions as shown in Table 4.Note that with the historical baseline, firms will

always participate, whether investing or not. With theindustry average baseline, everyone will participate if thevariable-cost adjustments always reduce emissionsbelow the average (in this case, if ap %m). For smallerreduction rates, then, those starting with lower emis-sions will be more likely to participate. With expectedemissions, participation occurs when ð1� sð1�rÞ2mÞmXam; full participation in this case holds if 1� a >sð1� rÞ2: That is, the variable-cost method must reduceemissions sufficiently relative to the expected reductionsfrom the investment alternative.The option to join but not invest creates both positive

and negative effects. On the positive side, cheaperreductions are made available and the investmentdecision is improved for imperfect baseline allocations.On the negative side, overallocation can increasesubstantially for firms with relatively low initial emis-sions under the industry average baseline rule.

3.2.2.1. An extreme example. For example, consider theextreme case in which the variable-cost method achievesnegligible reductions, while the investment is perfectlyclean. In other words, a ¼ 1 and r ¼ 0: Furthermore,consider the most generous definition of the industryaverage: %m ¼ 1=2: Then the only firms to participatewithout investing would be those with emissions that arealready below the industry average with that baselinerule. In the equilibrium solution presented in Table 5, wesee that the investment is lower by t=8 and actualemissions reduced are lower by t=48; compared to theindustry average baseline with the investment prerequi-site in Table 2. However, allocations are significantlyincreased, further decreasing net emissions reductions.We can see this result more generally by subtracting

the CERs from the actual reductions to get netreductions:Z

%m

0

ðm� AðmÞÞ dm

þZ 1

%mð#smþ tðAðmÞ � rmÞÞ ðm� AðmÞÞ dm ð9Þ

As with the investment precondition, net reductionsare by definition zero with the historical emissionsbaseline (since AðmÞ ¼ m) and positive with the expectedemissions allocation (since AðmÞom). SubstitutingAðmÞ ¼ %m into the previous equation, we can rewritethe net reductions under the average emissions rule as

ARTICLE IN PRESS

Table 3

Investment decisions without investment prerequisite

Participate if Then, Invest if Else, Invest if

No policy n.a. pþ #sm > k pþ #sm > k

Historical baseline ao1 pþ #smþ tða� rÞm > k pþ #smþ tð1� rÞm > k

Industry average amo %m pþ #smþ tða� rÞm > k pþ #smþ tð %m� rmÞ > k

Expected emissions amoEf *mg pþ #smþ tða� rÞm > k pþ #smþ tðEf *mg � rmÞ > k

Table 4

Cutoff emissions rates and baseline allocations

%m

Historical baseline 1

Industry average %m=aExpected emissions

Min1� a

sð1� rÞ2; 1

� �

Table 5

Incentives with average emissions baseline

a ¼ 1; r ¼ 0;

%m ¼ 1=2Investment Allocated

baseline

Emissions

reduced

Net

reductions

No Prerequisite s

2þ3t

8

1

8þ11s

48þ

t

6

s

3þ11t

48

5s

48þ3t

48�

1

8Investment

prerequisite

s þ t

2

s þ t

4

s

t

4

s

12

C. Fischer / Energy Policy 33 (2005) 1807–1823 1817

Z 1

0

ðm� %mÞ dm�Z 1

%mð1� #sm� tð %m� rmÞÞ ðm� %mÞ dm

¼1

2� %m� wð %mÞ; ð10Þ

where wð %mÞ ¼R 1

%m ð1� #sm� tð %m� rmÞÞ ðm� %mÞ dm: Since

%m ¼ %m=a; m > %m for all m >

%m; ensuring w > 0: Thus, net

reductions can only be achieved if the standard issufficiently strict: %mþ wð %mÞo1=2: (Furthermore, they aremaximized when �wð %mÞ ¼ 1:) Since the average ofhistorical emissions is 1/2, this rule necessarily raisesnet emissions.As further analytical solutions to the general problem

are too complex to be useful, we next employ numericalsimulations to explore the impact of different levels of a;r; and %m; with and without an investment requirementfor participation.

3.3. Numerical comparison

To illustrate the results, we apply the parametervalues s ¼ 0:2 and t ¼ 0:8 to the solutions from theuniform distribution example, reflecting the idea thatsocial benefits of reductions are likely to be large relativeto the private benefits. For the average emissionsbaseline, we consider two standards: %m ¼ 0:5; theaverage of initial (historical) emissions, and %m ¼ 0:2;

reflecting a stricter standard of the 20th percentile ofinitial emissions.

3.3.1. Complete reductions from investment

First, consider the case of an investment in atechnology that completely eliminates emissions. Thefollowing figures depict on the one side the shares ofinvestment and participation, and on the other side theactual, certified and excess emissions reductions thatresult under the different baseline rules. ‘‘Excessreductions’’ denote net reductions in excess of thosethat occur under no policy. Figs. 1 and 3 present the caseof an investment prerequisite, while Figs. 2 and 4assume there is an alternative method of abatement thatis 10% effective and no investment is required toparticipate (Figs. 1–4).With the investment prerequisite, we see that the

historical industry average baseline (i.e., %m ¼ 0:5)encourages as much investment as the historical emis-sions baseline, but it is less effective at generating actualreductions. However, it does not overallocate as manyCERs, ensuring a positive net reduction. The expectedemissions policy generates less investment than either ofthe other baseline rules, and actual reductions that fallin between those of the historical and average baselines;however, since fewer CERs are given, net reductions arelarger.As expected, eliminating the investment requirement

increases participation and reduces investment. Sincethe variable-cost reductions are relatively small, as fewerfirms opt to invest, actual emissions reductions aresmaller. However, all but the expected emissions base-line overallocate to a greater extent, lowering netreductions. In fact, the historical industry averagebaseline allocates so many CERs that net reductionsare negative. For the expected emissions baseline,eliminating the prerequisite allows for more cost-effective reductions and more accurate allocation,resulting in an improvement in net reductions.It is interesting to note that in only one case do total

emissions fall compared with no policy: when invest-ment is a prerequisite for participation and a sufficientlystrict average emissions standard is used to determinethe baseline. In this case, less abatement occurs, but evenfewer reductions are certified. In all other regimes,allocations outweigh (or at least offset) actual reduc-tions. As discussed earlier, the net cost of this expansion

ARTICLE IN PRESS

No Policy Historical Expected Average(initial)

Average (low)-0.150

-0.100

-0.050

0.000

0.050

0.100

0.150

0.200

0.250

0.300

0.350

Reductions CERs Excess Reductions

Investment Prerequisite

s = 0.2, t = 0.8; � = 0.9, � = 0

Fig. 3. Investment and participation rates with clean technology.

No Policy Historical Expected Average(initial)

Average (low)-0.150

-0.100

-0.050

0.000

0.050

0.100

0.150

0.200

0.250

0.300

0.350

Reductions CERs Excess Reductions

No Prerequisite and10% Effective Abatement Alternative

s = 0.2, t = 0.8; � = 0.9, � = 0

Fig. 4. Emissions reduction with clean technology and no prerequisite.

Investment

Investment Prerequisite

s = 0.2, t = 0.8; � = 0.9, � = 0

0.0000.100

0.2000.3000.4000.500

0.6000.7000.800

0.9001.000

No Policy Historical Expected Average(initial)

Average (low)

Fig. 1. Investment rates with clean technology and prerequisite.

0.0000.100

0.2000.3000.4000.500

0.6000.7000.800

0.9001.000

No Policy Historical Expected Average(initial)

Average (low)

Investment Participation

No Prerequisite and10% Effective Abatement Alternative

s = 0.2, t = 0.8; � = 0.9, � = 0

Fig. 2. Emissions reduction with clean technology and prerequisite.

C. Fischer / Energy Policy 33 (2005) 1807–18231818

in terms of welfare depends on the difference betweenthe marginal benefits and costs of abatement.On the other hand, if participants, as opposed to the

certifying authority, can choose the baseline methodol-ogy, overallocation will always occur. Above-averageemitters will select the historical baseline, while below-average emitters will choose the industry average—andengage in too much investment if that is a prerequisitefor participation. Using the same parameter values, 60%of firms invest (compared to 50% with the historicalbaseline) with the prerequisite; emissions reductions riseto 0.35, but CERs are inflated to 0.49, resulting in excessreductions of �0.21, or three times the increase in globalemissions over the no-policy regime compared to thehistorical baseline.Figs. 5 and 6 show that when the alternative method

is sufficiently effective (i.e., for a small enough a suchthat participation always occurs, except in the case of astrict average emissions rule), the investment andabatement decisions become identical across baselinerules. The only difference then lies in the allocation, for

which the expected emissions rule performs the best. Thestrict average standard does not overallocate much, butneither does it encourage as much abatement.

3.3.2. Incomplete reductions from investment

Next, consider the case in which the investment reducesemissions only partially. To illustrate the impact ofincomplete proportional reductions, Figs. 7–10 assume50% reductions in emissions (i.e., just enough such thatno investors will retain above-average emissions).As we know, the historical baseline, with its efficient

investment incentives, always accomplishes the mostactual reductions, while the industry historical averagebaseline without the investment prerequisite overallocatesthe most compared with actual reductions, the resultbeing the highest total emissions. When reductions arelarge, the historical baseline certifies the most reductions,but when reductions are half of initial emissions, theindustry average baseline allocates the most in absoluteterms. The expected emissions baseline generates morereductions and fewer allocations than either of the

ARTICLE IN PRESS

-0.100

-0.050

0.000

0.050

0.100

0.150

0.200

0.250

0.300

0.350

Reductions CERs Excess Reductions

No Policy Historical Expected Average(initial)

Average (low)

No Investment Prerequisite and50% Effective Abatement Alternative

s = 0.2, t = 0.8; � = 0.9, � = 0

Fig. 6. Emissions reduction with more effective alternative.

Investment

0.000

0.100

0.200

0.300

0.400

0.500

0.600

0.700

0.800

0.900

1.000

No Policy Historical Expected Average (initial)

Investment Prerequisite and 50% Effectiveness

s = 0.2, t = 0.8; � = 0.9, � = 0.5

Fig. 7. Investment rates with less effective technology and prerequisite.

Investment Participation

0.000

0.100

0.200

0.300

0.400

0.500

0.600

0.700

0.800

0.900

1.000

No Policy Historical Expected Average (initial)

No Prerequisite and 10% Effective Abatement Alternative

s = 0.2, t = 0.8; � = 0.9, � = 0.5

Fig. 8. Investment and participation rates with less effective technol-

ogy.

Investment Participation

0.0000.100

0.2000.3000.4000.500

0.6000.7000.800

0.9001.000

No Policy Historical Expected Average(initial)

Average (low)

No Investment Prerequisite and50% Effective Abatement Alternative

s = 0.2, t = 0.8; � = 0.9, � = 0

Fig. 5. Investment and participation rates with more effective

alternative.

C. Fischer / Energy Policy 33 (2005) 1807–1823 1819

averaging baselines. In the absence of an investmentprerequisite, the expected emissions regime ensures fullparticipation—and thereby efficient abatement incen-tives—and also allocates no more than is justified.Finally, although we have not explicitly incorporated

transaction costs for CDM participation, including anymonitoring or enforcement costs borne by the partici-pants, the analysis offers intuition about their effects aswell. Transaction costs essentially make the participa-tion decision a second investment decision.28 Some firmsthat would invest might not choose to participate if thevalue of the net CER allocation does not cover thetransaction costs. Effectively, then, the costs of investingin the absence of CDM are smaller—and the reductionsthat would occur anyway are larger—relative to whatoccurs with the program. Second, in the absence of an

28Mathmatically, this creates extra cutoff points in the distribution

for participation, much like in the section without the investment

prerequisite.

investment prerequisite, the alternative abatementmethod would now involve a fixed cost, so fewer wouldchoose that. However, for firms that would participatein any case, the investment decision remains asoriginally presented, since the transaction cost isincurred regardless. Thus, the essential differencesamong the baseline rules are characterized similarlywith transaction costs as with investment costs alone.

4. Other issues

Although we have focused on investment incentives,the design of baseline rules can affect other behavior aswell. For example, if current emissions practices affectthe future baseline, then an incentive would exist toincrease emissions before joining a CDM project. This iswhy a historical emissions approach should be based onemissions that occurred prior to implementation ofCDM, and why expected emissions need to be based on

ARTICLE IN PRESS

-0.150

-0.100

-0.050

0.000

0.050

0.100

0.150

0.200

0.250

0.300

0.350

Reductions CERs Excess Reductions

No Policy Historical Expected Average (initial)

Investment Prerequisite and 50% Effectiveness

s = 0.2, t = 0.8; � = 0.9, � = 0.5

Fig. 9. Emissions reduction with less effective technology and

prerequisite.

-0.150

-0.100

-0.050

0.000

0.050

0.100

0.150

0.200

0.250

0.300

0.350

Reductions CERs Excess Reductions

No Policy Historical Expected Average (initial)

No Prerequisite and 10% Effective Abatement Alternative

s = 0.2, t = 0.8; � = 0.9, � = 0.5

Fig. 10. Emissions reduction with less effective technology and no

prerequisite.

C. Fischer / Energy Policy 33 (2005) 1807–18231820

factors that cannot be manipulated. Second, even if thebaseline amount is predetermined, if the annual alloca-tion is conditional on continuing operations, it may notoffer proper long-run incentives.29

Another important design choice is the option toimplement a baseline method in rate-based, rather thanlump-sum, form. For example, instead of using a fixedbaseline, permits might be allocated according to afirm’s historical emissions rate multiplied by actualoutput (or perhaps use of some input), in order toincorporate trends in demand and production. Althoughthis rule might in some ways make baselines morerealistic and responsive to external conditions, it isimportant to recognize that the allocation then alsofunctions as a production (or input) subsidy. As aconsequence, the firm has less incentive to reduce overallemissions through conservation.30

However, in certain circumstances, some of whichmay be quite valid for CDM situations, an outputsubsidy could enhance the gains from a project. Forexample, the developing country’s plant owner may be apoorly regulated monopoly, underproviding its output(e.g., electricity) and overcharging customers. Makingemissions costly (or reductions valuable) can thenexacerbate this problem by driving up prices. An outputsubsidy can help bring prices and production more inline with what society would want, benefiting consumersin the developing country—or at least insulating themfrom increased prices due to the CDM project.31

Another circumstance that may justify an output-based

29See Baumol and Oates (1988, Chapter 14) for the problem of long-

run inefficiency of grandfathered permits conditional on entry into and

non-exit from the regulated industry.30See Fischer (2001).31See Gersbach and Requate (2004) on optimal rebating under

imperfect competition. Of course, in the developing country context,

better regulation, management, and competition would likely do much

more to help consumers.

allocation is if the firm joining the CDM program facescompetition from other non-participating domesticproducers, whose emissions thereby remain unregulated.If these other firms are close competitors and havehigher emissions rates, an output subsidy can help divertproduction back toward the now-regulated participatingfirm, further reducing overall emissions.32 If, however,the participating firm were a manufacturing facilitycompeting primarily with firms in Annex I countries, anoutput subsidy would generally not be warranted toprevent leakage (although it might offset other imper-fections in developing countries’ output markets).Rate-based versions of these allocation methods have

the same complications as their lump-sum counterpartsin terms of investment and participation decisions. Tothe extent that it would be more accurate, thosecomplications will be mitigated. However, to the extentthat the rule is overly generous, the subsidy impactlooms larger. Which effect dominates depends on theproject characteristics and market environments in thecandidate countries. Indeed, the strength of the parti-cular subsidy will depend on both the price of CERs andthe baseline rule. Historical emissions rates are project-specific, so firms with higher previous emissions willreceive a stronger output subsidy. An industry averageemissions rate offers a consistent output subsidy acrossfirms in that industry, but it may offer less efficientinvestment incentives and less accurate allocations. Atrade-off thus arises between accuracy of allocation andappropriateness of the output support.

5. Conclusion

Selecting a reasonable method for determining base-line emissions is critical to the success of incorporating

32See Bernard et al. (2001).

ARTICLE IN PRESS

33 It is interesting to compare this baseline rule to more familiar

tradable performance standards (TPS) systems, since both allocate

permits based on an average industry standard. The main differences

are twofold. First, the CDM program is voluntary, which creates a

selection bias for relatively clean firms to join the program. Second,

permits can be traded outside the sector. With a traditional TPS, if

clean firms selected into the program, the standard would be more

easily achieved and the price of permits would fall. With CDM, the

permit price is determined by international markets, so the average

emissions rate is an allocation rather than a binding performance

standard. See also Fischer (2003).

C. Fischer / Energy Policy 33 (2005) 1807–1823 1821

project-based emissions reductions strategies into anyemissions trading program. In the international pro-gram for reducing greenhouse gases, the CDM is a mainfocus for baseline rules, but the issue also applies tostrategies for carbon sinks and potential offset projectsin developed countries. The obstacles to accuratebaseline determination include general uncertainty,asymmetric information between the certifying author-ity and participants, costly administrative and informa-tion-gathering activities, and insufficient policy tools toensure revelation of true emissions. As a consequence, apolicy balance must be struck between the accuracy ofthe abatement measures, which are needed to preservethe environmental integrity of the projects, and theefficient performance of the system, which is neededto assure cost-effective and productive investmentstrategies.In response to these challenges, three main options for

baseline rules have been advanced: (1) historicalemissions, (2) expected emissions, and (3) a standardfor average industry emissions. They represent differenttrade-offs between the root problems of inaccuratebaseline accounting: inefficient participation and ineffi-cient investment. For investments that would be profit-able anyway, an allocation of credits inducesunnecessary participation in the CDM and non-addi-tional CERs, which expand the global emissions cap.For marginal firms, the value of the permit rentsdetermines whether the investment costs are incurred.The welfare loss from inaccurate baseline assessmentfor these projects involves not only expansion—orcontraction—of the global cap, but also the undertakingof inefficient investments from overallocation or theforgoing of worthwhile opportunities due to under-allocation.A rule using historical emissions as a baseline

primarily risks overparticipation; however, the incen-tives for investment and abatement behavior aregenerally efficient. The exception would be when truebaseline emissions would be changing (or deviating fromthe historical trend) in the absence of any investment.For a firm whose emissions would grow faster, the rulemight underallocate and discourage participation; for afirm whose emissions would decline anyway withoutinvestment, an overallocation would induce participa-tion that might not otherwise be worthwhile.The industry average emissions rule tends to perform

poorly from an incentives standpoint. It providesinsufficient incentives for relatively high emitters toparticipate, since they are allocated only a fraction oftheir actual counterfactual emissions. A very dirty firmmay have inexpensive options for reducing its emissions,but it must push its emissions below the average toreceive a net transfer, and that transfer must then offsetthe full cost of all reductions to make participationworthwhile. Meanwhile, it over-rewards not only firms

that would reduce their emissions in the future anyway,but also those that are already relatively low emitters,since they could do little in terms of reduction and stillreceive a net transfer for having below-average emis-sions. Requiring major investments as a prerequisite forparticipation would deter some below-average emittersfrom joining, but it would encourage others to under-take costly investments that are not justified, since thevalue of the allocation outweighs the environmentalgains. Tightening the standard reduces the overalloca-tion problem, but it also exacerbates the underallocationproblem; the corresponding improvement in net reduc-tions comes at the cost of less abatement activity.33

With an expected emissions rule, the third-partyverifier would attempt to gather project-specific as wellas industry-specific information and make an educatedguess about the likely baseline emissions in the absenceof policy. To the extent that expectations are inaccurate,marginal investment incentives will not be efficient.However, to the extent that expectations are moreaccurate than the other, cruder methods, the allocationof CERs will be more appropriate and result in lessexpansion of the global emissions cap. On the otherhand, this rule is likely to be more expensive toimplement because of the costs of information gather-ing. Also, this approach risks inviting strategic behaviorthat attempts to create ‘‘better’’ sets of expectations.Thus, this method would require that the certifyingauthority have access to necessary information at coststhat do not outweigh the benefits of greater accuracy.Clearly problematic would be to allow participants, as

opposed to the certifying authority, to choose thebaseline methodology. They would select the rule thataffords them the largest baseline, ensuring overalloca-tion and encouraging too much investment if that is aprerequisite for participation.Designing eligibility requirements presents another

opportunity for screening projects worthy of receivingany baseline allocation. One could require a majorinvestment rather than minor improvements. Forexample, building a wind farm that clearly wouldotherwise be uneconomic could qualify, but not updat-ing the boilers in a power plant. However, by includingonly ‘‘clearly uneconomic’’ projects, one risks excludingmany projects that deserve to qualify and cost less

ARTICLE IN PRESSC. Fischer / Energy Policy 33 (2005) 1807–18231822

(being less uneconomic). Participation requirementswould better serve the goal of promoting cost-effectiveemissions reductions by focusing on eliminating the‘‘clearly economic’’ projects.Sustainable development benefits are intended as

another requirement for eligibility in a CDM project.Such ancillary benefits can be important and representanother yardstick for participation in addition toemissions abatement. They can legitimately affect thepublic priorities for projects and can affect the welfarecosts of erring in the baseline allocations. Understand-ing how requires knowing not only the size of thesustainable development benefits, but also how theymight be correlated with the private economic andenvironmental benefits driving investment and policychoices.It has been shown that allocation and participation

rules alone cannot induce firms to truthfully reveal theirbaseline emissions. Although a third-party certificationauthority would not be able to set the transaction price,one could incorporate a quality parameter to CERs,which would function like a price differential. Byallowing the certification authority to contract overtwo different project aspects—baseline emissions andabatement quality—one may restore power to designmore efficient revelation mechanisms. This option willneed to be the subject of future research.Since participation in these emissions reduction

projects is voluntary, any baseline rule is likely to errby allocating too much overall. The cost savings mustthen justify not only the costs of administrating theprogram but also the costs of expanding the cap. Thosedepend on the difference between prospective marginaldamages from future climate change and marginal costsof abatement, and on the size of the misallocation.However, one could argue that the cost of having a bitmore global emissions may be a lot lower than the costof encumbering the nascent institutions for developingcountries’ participation with complex project approvalcriteria that reward rent seeking, distort markets, and soforth.34 In this case, the focus should be on simplicityand good investment incentives, such as with a basichistorical emissions rule. If additional, unbiased infor-mation can be obtained at low cost, it could be used tohelp eliminate clearly economic projects and improvethe accuracy of expected baselines.Indeed, the real benefits could come from the

successful experience and development that lead non-Annex B parties to join in the emissions cap, thuseliminating the need for determining baselines forindividual projects. Also deserving mention are thesustainable development benefits the countries get evenbefore graduating into the cap. Although harder toquantify and not included in this calculation, these

34See Kopp et al. (2003).

benefits offer a reason to lean toward a generousapproach. In the meantime, however, further study isneeded to evaluate which baseline rules are mostappropriate in which situations. In particular, for thedifferent types of projects, we need to better understandthe correlation between abatement investments andother productivity enhancements, as well as the natureof uncertainties and information availability. Further-more, we need to recognize the impact of marketimperfections and institutional differences in developingcountries and how they might affect participation,investment, and allocation.

Acknowledgements

Thanks to Mike Toman, Christoph Boehringer, andseminar participants for useful comments.

References

Babu, P.G., Saha, B., 1996. Efficient emission reduction through

joint implementation. Environment and Development Economics

1 (4, October), 445–464.

Baumol, W.J., Oates, W.E., 1988. The Theory of Environmental Policy

second ed. Cambridge University Press, Cambridge.

Bernard, A., Fischer, C., Vielle, M., 2001. Is there a rationale for

rebating environmental levies? RFF Discussion Paper 01-31,

Resources for the Future, Washington, DC.

Bohm, P., 1994. On the feasibility of joint implementation of carbon

emissions reductions. Climate Change: Policy Instruments and

their Implications, Proceedings of the Tsukuba Workshop of IPCC

Working Group III, Center for Global Environmental Research,

Environment Agency of Japan, Tsukuba, Japan.

Fischer, C., 2001. Rebating environmental policy revenues: output-

based allocations and tradable performance standards. RFF

Discussion Paper 01–22, Resources for the Future, Washington,

DC.

Fischer, C., 2003. Combining rate-based and cap-and-trade emissions

policies. Climate Policy 3S2, S89–S109.

Gersbach, H., Requate, T., 2004. Emission taxes and optimal

refunding schemes. Journal of Public Economics 88 (3–4), 713–725.

Hagem, C., 1996. Joint implementation under asymmetric information

and strategic behavior. Environmental and Resource Economics 8,

431–447.

Janssen, J., 1999. (Self-) enforcement of joint implementation and

clean development mechanism contracts. Fondazione Eni Enrico

Mattei Note di Lavoro: 14/99.

Kopp, R., Morgenstern, R., Pizer, W., Ghersi, F., 2003. Reducing cost

uncertainty and encouraging ratification of the Kyoto protocol. In:

Chang, C.-C., Mendelsohn, R., Shaw, D. (Eds.), Global Warming

and the Asian Pacific. Edward Elgar Ltd., Aldershot.

Laffont, J.-J., Tirole, J., 1993. A Theory of Incentives in Procurement

and Regulation. MIT Press, Cambridge, MA.

Liski, M., Virrankoski, J., forthcoming. Frictions in project-based

supply of permits. Environmental and Resource Economics.

Loeb, M., Magat, W.A., 1979. A decentralized method of utility

regulation. Journal of Law and Economics October, 399–404.

Millock, K., Hourcade, J.-C., 2001. Bargaining Institutions under the

Clean Development Mechanism. CIRED, Mimeo. Paris.

ARTICLE IN PRESSC. Fischer / Energy Policy 33 (2005) 1807–1823 1823

Narain, U., Van ‘t Veld, K., 2001. Long-term risks to developing

countries from participating in the clean development mechanism.

Working Paper, Resources for the Future, Washington, DC.

UNFCCC, 2001. Modalities and Procedures for a Clean Development

Mechanism, as Defined in Article 12 of the Kyoto Protocol Report

of the Conference of the Parties in Its Seventh Session. FCCC/CP/

2001/13/Add2.

Wirl, F., Huber, C., Walker, I.O., 1998. Joint implementation:

strategic reactions and possible remedies. Environmental and

Resource Economics 12, 203–224.