emissions pricing, spillovers, and public investment in environmentally friendly technologies

Available online at www.sciencedirect.com

er.com/locate/eneco
Energy Economics 30 (2008) 487–502
www.elsevi

Emissions pricing, spillovers, and public investment inenvironmentally friendly technologies☆

Carolyn Fischer

Resources for the Future (RFF), 1616 P Street, NW, Washington, DC 20036, United States

Received 29 March 2006; received in revised form 17 May 2007; accepted 2 June 2007Available online 16 June 2007

Abstract

In a second-best world of below-optimal pollution pricing, the public return to R&D may be greater thanunder Pigouvian pricing, due to excess benefits of increasing abatement, or it may be lower, since privateactors lack the incentives to take full advantage of the new, cleaner technologies. This paper uses a simplemodel to demonstrate the interaction between environmental policies, R&D externalities, and the socialreturn to innovation. The results indicate that strong public support for innovation in abatement techno-logies is only justified if at least a moderate emissions policy is in place and spillover effects are significant.If emissions policy can adjust to cost changes, the public role may be enhanced, since private actors willprefer to avoid ratcheting. Social gains from innovation increase only if innovation can allow emissions tobe more fully priced. Technology policy is more effective with fuller emissions pricing and is better viewedas a complement to than a substitute for mitigation policy.© 2007 Elsevier B.V. All rights reserved.

JEL classification: Q28; O38; H23

Keywords: Emissions price; Technology; Innovation; Spillovers; R&D policy; Climate change

☆ Support from the National Commission on Energy Policy is gratefully acknowledged. This article was revised whileparticipating in the project “Environmental economics: policy instruments, technology development, and internationalcooperation,” sponsored by the Centre for Advanced Study (CAS) at the Norwegian Academy of Science and Letters inOslo in 2005.

E-mail address: [email protected].

0140-9883/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.eneco.2007.06.001

mailto:[email protected]

http://dx.doi.org/10.1016/j.eneco.2007.06.001

488 C. Fischer / Energy Economics 30 (2008) 487–502

1. Introduction

Public investment in research and development (R&D) is often driven by the fact that thesocial value of innovation often surpasses what the innovators themselves can appropriate.Studies of commercial innovations suggest that, on average, only about half of the gains to R&Dreturn to the originator, although appropriation rates vary considerably over different types ofinnovations.1 With respect to innovations that may benefit the environment, this rationale ofspillover effects is combined with the argument that, since the damages of emissions are not fullyinternalized by private markets, there is an added impetus for public investment in less pollutingtechnologies.

Research in environmental economics on technological change has focused on the private andsocial incentives for innovation created by different environmental policies.2 Empirical studieshave shown that technological progress and diffusion respond to price incentives and to certainregulations.3 Innovation incentives are recognized as important criteria for policy selection,4 andmarket-based environmental policies are revealed to perform better than command-and-controlregulation.5 Several theoretical papers compare innovation and adoption incentives amongdifferent market-based instruments, like emissions taxes and auctioned or grandfathered permits.6

Others have drawn attention to how the presence of induced technological change affects optimalemissions policy.7 Typically, they assume that the tax or emissions price achieves some level ofoptimality with respect to pollution abatement, at least in some (often pre-innovation) sense, if notin the dynamic sense.8 Alternatively, the environmental policy goal may be to achieve a second-best solution in the presence of an externality in the market for R&D, where direct R&D policiesare either absent or insufficient to target the specific environmental and innovation problems.

Meanwhile, in reality, environmental externalities are often underpriced; occasionally, publicsubsidies may actually exacerbate them.9 Not surprisingly, innovation in “environmentallyfriendly” technology is then lacking. Consequently, governments use various innovation andadoption incentives to make up for the shortfall. These tools may include direct subsidies, taxcredits, technology forcing regulation, and price or market share guarantees for the use ofparticular technologies.10 Often, where political constraints prohibit policies from creatingsignificant costs to business for polluting behavior, innovation subsidies are then called upon withgreater urgency to try to correct the environmental problem.

Parry et al. (2003) show that the total gains to innovation are naturally limited to the currentcosts of abatement and the benefits from eliminating the remaining emissions. These potential

1 See Griliches (1992) and Nadiri (1993), as well as Hall (1996) and Jones and Williams (1998).2 For more discussion of the literature, see Kemp (1997), Ulph (1998), and the recent review by Jaffe et al. (2003).3 For example, Jaffe and Stavins (1995) found energy efficiency in new residential construction to be sensitive to

energy prices and technology adoption costs, but not to command-and-control regulations like state building codes.Newell et al. (1999) found that innovation in energy-using consumer durables is sensitive to energy prices and labelingrequirements.4 Stavins (2000), Bohm and Russell (1985).5 See, e.g., Downing and White (1986), Magat (1978) and Zerbe (1970) for comparisons of market-based policies to

command-and-control regulations.6 See, e.g., Milliman and Prince (1989), Biglaiser and Horowitz (1995), Jung et al. (1996), Fischer et al. (2003),

Requate and Unold (2001), Fischer and Newell (2004).7 See, for example, Goulder and Schneider (1999), Popp (2004), Gerlagh (2006).8 Some of the dynamic problems are addressed in Petrakis and Xepapadeas (1999) and Kennedy and Laplante (1999).9 Fischer and Toman (1998), OECD (2003).

10 Many of these examples have been used to promote renewable energy sources and low- or zero-emissions vehicles.

489C. Fischer / Energy Economics 30 (2008) 487–502

gains are further reduced by the fact that innovation and diffusion are lengthy–and costly–processes, and future benefits are discounted. Thus, in few situations are the gains to innovationlarge compared to the current gains from internalizing the pollution externality. However, thoserelative gains can be quite different when policies do not take advantage of all the current gainsfrom abatement.

While there is a growing awareness that pricing emissions creates private incentives forinnovation, it is not well understood how emissions pricing (or lack thereof) affects the socialpremium to innovation. In a second-best world of below-optimal pollution pricing, the role forpublic R&Dmay be greater, due to the excess benefits of increasing abatement; on the other hand,it may well be smaller, since private actors will still not have the incentives to take full advantage ofthe new technologies that are developed. This paper uses a simple model to demonstrate theinteraction between environmental policies, R&D externalities, and the social return to innovation.

The next section analyzes the social and private benefits from innovation as a function of thespillover rate and degree of emissions pricing, when those prices are fixed. It reveals that a gooddeal of public support for R&D in emissions control technologies is warranted only if theemissions price will reflect a significant share of the damages and spillovers are important. Foralternative technologies more generally, support is warranted if the market price for conventionaltechnologies reflects a significant share of their overall social costs, including emissions damages.The third section evaluates those benefits when technological progress enables politically con-strained regulators to make environmental policy more stringent. Ratcheting tends to diminishprivate incentives for innovation, and it also lowers the overall social return to innovation—unless abatement targets are expanded so much as to raise marginal costs. On balance, the role forpublic investment may be enhanced by expectations of policy adjustments. The final sectionoffers conclusions as applied to the climate policy debate, arguing that technology policy is anecessary complement to, not a substitute for, a proper mitigation policy.

2. Model

We use a transparent model of linear benefits and marginal costs of abatement to solve for thesocial premium to innovation–that is, the additional social gains above and beyond the privategains–given a rate of emissions pricing and of spillovers. Since the focus will be on the marginalgains from abatement and the returns to lowering abatement costs, we can abstract from the costsof innovation and inframarginal emissions, albeit recognizing that they affect the optimal equi-librium levels of innovation and welfare. We consider the private and social innovation incentivesunder an emissions tax, which allows total abatement to respond to cost changes; if the regulationinstead set a quantity target, the only discrepancy between the two returns to innovation would bethe spillover effects.

Assume that abatement, A, has constant marginal benefits to society of b. Let the costs ofabatement be

CðAÞ ¼ cð1� rÞ2

A2 ð1Þ

where c is the initial slope of the marginal abatement cost curve and r is the rate of potential costreduction.

Let p be the price of emissions (and the value of abatement to the firm). The firm must pay thisprice on all of its initial emissions, M, net of abatement and any freely allocated (grandfathered)


permits or tax threshold, F. This policy could be set up as a tax or emissions permit program, or asubsidy, in which case the net emissions liability (M−F) is less than abatement. The importantassumption for now is that this emissions price is fixed.11 Consequently, the net emissionsliability does not affect incentives and can be ignored.

The private market, portrayed here as a representative firm, chooses the level of abatement tomaximize profits:

p ¼ pA� cð1� rÞ2

A2 � pðM � FÞ ð2Þ

leading to

A ¼ pcð1� rÞ ð3Þ

Rewriting profits, we get

p ¼ p2

2cð1� rÞ � pðM � FÞ: ð4Þ

Let us assume that a private innovator can capture 1−σ of the private gains to reducing costs,where σ is the spillover rate.12 Thus, the private innovator's gains to increasing r are

ð1� rÞApAr

¼ ð1� rÞp22cð1� rÞ2 : ð5Þ

Social welfare is

W ¼ bA� cð1� rÞ2

A2 ¼ pð2b� pÞ2cð1� rÞ : ð6Þ

The marginal social benefits of innovation are reflected in the additional gains from reducingcosts:

AWAr

¼ pð2b� pÞ2cð1� rÞ2 : ð7Þ

Let p=ϕb, where ϕ is the share of the marginal damages reflected in the market price ofemissions. We define the social premium, R, for cost reductions as the difference between thesocial and private values for reducing abatement costs a bit more:

RuAWAr

� ð1� rÞApAr

¼ /ð2� /ð2� rÞÞG⁎ ð8Þ

where G⁎u AWAr jp¼b ¼ b2

2cð1�rÞ2. In other words, G⁎ is the welfare gain that would arise from an

additional percentage reduction in costs when emissions are optimally priced at Pigouvian levels.

11 We will relax this assumption in the next section, looking at endogenous policy adjustments.12 Fischer et al. (2003) note that in a market equilibrium in which the technology may be adopted or imperfectlyimitated, the effective appropriation rate will be endogenous. However, for our purposes we will take it as given.

Fig. 1. Gains to abatement cost-reducing innovation.


An elegant result from this model is that, while the slope of the marginal cost and benefitcurves are essential for determining the marginal gains to innovation, the share that is the socialpremium is invariant to all but the rate of emissions pricing and spillovers. Similarly, this sharewill also be invariant to other factors that might affect the level of the marginal gains to innovationin a multi-period model, as discounted flows or lagged benefits will have proportional impacts onG⁎. Thus, although we have a single-period model, we can think of A as being abatement over thehorizon relevant for the innovation, and p and b as representing the (expected) discounted pricesand marginal damages over that horizon.

Graphically, we see that the marginal social gains to innovation include the private gains(inclusive of spillovers), as well as the difference between marginal benefits and the price ofabatement, multiplied by the additional abatement that occurs with lower costs.We also see that theoverall gains are less than those that would occur if the price equaled the marginal benefit (Fig. 1).

In the following series of figures, we explore how the social premium–expressed as the shareof the optimal, marginal social value of innovation when the marginal cost of the externality isfully internalized (R /G⁎)–varies with the two parameters of concern: the rate at which the pricereflects marginal damages, and the spillover rate.

When innovators in the private market can capture all the returns to their R&D, the socialpremium displays an inverted U-shaped function with respect to the rate of internalization of theexternality (Fig. 2). Logically, if emissions are not priced at all, any innovation will not be used, so

Fig. 2. Social premium for cost reductions — no spillovers.

Fig. 3. Social premium for cost reductions — complete spillovers.


investment is not worthwhile. Meanwhile, if emissions are fully priced, the private market hassufficient incentive to invest, so no social premium exists. Correspondingly, the social premiumremains small when the emissions price is very far or very close to the Pigouvian level, and it ishighest when the emissions price is half of marginal damages.

On the other hand, if the private market has no incentives of its own to innovate, sincespillovers are complete, the social gains from abatement cost reduction increase monotonicallywith emissions pricing (up to the Pigouvian level). Furthermore, if about a third of the marginaldamages are priced, the additional benefits to society of reducing abatement costs are at least halfof the total benefits with full emissions pricing (see Fig. 3).

The other combinations of spillover shares and degrees of social cost pricing are shown as acontour graph in Fig. 4, in which lighter shades indicate a larger social premium. It shows clearlythat, for any degree of spillovers, if prices do not reflect a significant share of the social costs, littleis gained from reducing abatement costs. (For internalization rates less than 30%, for example, thesocial premium does not exceed half of optimal gains.) On the other hand, if prices reflectsubstantial portions of the full social costs (like 70%), spillover effects must be quite large for thesocial premium to be significant. We see not only that the lighter areas with the greater socialpremium begin where moderate shares of both social cost pricing and spillovers are present, butthey also intensify as those shares are increased together. Taken another way, the more significant

Fig. 4. Social premium for cost reductions (all combinations of price and spillover shares).

Table 1Change in R&D spending compared to optimal climate and technology policies

Example with 50% spillover Rate(σ=.5)

Renewable electricityand conservation a

Reducingcoal inputs b

Carbonsequestration

Social value with $25/ton C $ 0.075 $ 1.84 $ 25unit of measurement /kWh /million Btu c /ton

Market price without climate policy $ 0.070 $ 1.17 $ 0Percentage of social value priced (ϕ) 93% 64% 0%Social premium relative to optimal pricing 113% 133% 0%Price with $15/ton C $ 0.073 $ 1.57 $ 15Percentage of social value priced (ϕ) 97% 86% 60%Social premium relative to optimal pricing 105% 123% 132%

a Price figures are taken from Fischer and Newell (2004).b Coal input prices reflect those delivered to electric utilities. Production-weighted average emissions factors were

derived from “Table FE4: Average Carbon Dioxide Emission Factors for Coal by Rank and State of Origin,” Hong andE. R. Slatick (1994) and “Table 10. Coal Production by State, Coal Rank, and Group, 2000,” EIA (2000). Average Btu pershort ton of coal were obtained from “Table C6 Gross Heat Content of Coal, 1992-2001,” EIA (2001). The price per shortton is from “Table 34: Average Price of Coal Delivered to End Use Sector by Census Division and State, 2002, 2001” EIA(2002).c The Btu measure reflects gross heat content of coal.


the spillover problem is, the more the social gains to public R&D in environmentally friendlytechnologies can be enhanced by more complete emissions pricing.

The relevant price share need not always be the share of emissions damages reflected in theprice of emissions, but rather the share of social costs that are reflected in the price of the emissions-intense activity or input. These concepts can be illustrated and refined using the case of carbonemissions. Table 1 illustrates some examples, supposing that the marginal damages from green-house gas emissions were $25 per ton of carbon. Total social costs include both production costsand the costs of carbon emissions. By this metric, the current market price of electricity thenreflects about 93% of the social costs; the current price of coal to electric utilities reflects roughlytwo thirds of the social costs; and the price for capture and sequestration activities reflects none ofthe costs. In this case of no carbon price, there would be additional social benefit from innovationto lower the cost of renewable energy, improve energy efficiency, and reduce reliance on coal,above the private savings from reduced input use. Furthermore, spillovers from fuel-savingtechniques or technologies would add to the social premium. On the other hand, from the point ofview of reducing the carbon emissions from the use of coal (i.e., “end-of-pipe” technologies likecarbon capture and storage, or emissions-reducing processes that do not offer other cost-effectiveefficiency improvements), without any incentive to use them, little is to be gained from innovation.

To illustrate how large are the gains to R&D relative to a policy of full carbon pricing, Table 1assumes that spillovers are half of the social value to innovation.13 The ratio of the social premiumto that under optimal pricing then gives an indication of howmuch public support for R&D shouldincrease with the discovery of the climate damages–given a $0 or $15 carbon price–relative to howmuch it should increase if we could simultaneously price carbon at the full $25/ton.14 In other

13 This corresponds to rough estimates in the literature: Griliches (1992) and Nadiri (1993), as well as Hall (1996).14 Note that spillovers have the opposite effect on this ratio as on the absolute return to public R&D support. With lowerspillovers, baseline public support is lower, making the ratio of second-best technology policy higher. With completespillovers, the government always pays for all research, so a low emissions price undermines the return to any innovationand reduces the desired spending on R&D.


words, if we think our current suite of technology policies reflects a reasonable response to thetradeoffs posed by other societal problems like spillovers, how much more should we increaseenergy-related technology spending to respond to climate change, given the mitigation policy wechoose? For example, suppose we would want to increase technology support by 100% with fullcarbon pricing.15 Then in the absence of an explicit carbon price, efforts to reduce the cost ofdisplacing fossil energy generation reap additional gains of 13%, and efforts to improve theefficiency of or reliance on coal inputs reap an additional 33%. Since electricity and coal inputshave significant costs already, there is a good amount of incentive to use alternatives or improveefficiency, so public investment is worthwhile. With a moderate price of $15, private innovationincentives improve, as do abatement incentives, so technology policy does not need to over-compensate as much. On the other hand, sequestration only occurs if carbon reduction has a value,so a $15 price means that the gains from public R&D spending in this area are 32% greater thanwith full pricing, but with no emissions price, that R&D is useless.

Of course, assessing the appropriate time horizons of the innovation and the policies is alsoimportant here. For innovations that improve current technologies–like commercializing windpower and other renewable energies and promoting energy efficiency–the emissions policies ofthe near term are the determining factors. However, for technological advances that have longer-term implications–or might replace current technologies–what matters are expectations about thepolicies that will be in place when the invention can be commercialized. Thus, even withoutcarbon pricing now, the expectation of prices in the future can make current public R&D supportworthwhile. Furthermore, it may be that those policies will be influenced by the future state ofabatement costs.

3. Policy constraints and price adjustment

Another argument for public investment in climate-friendly technologies when carbon pricesare low is that it could make more stringent regulation politically viable. Stakeholders are likely toresist policies that are too costly, so if abatement becomes cheaper, they may accept harder targets.Similarly, policymakers may resist mitigation subsidy programs that cost too much in publicfunds; lower-cost abatement technology can in such cases enable those limited funds to purchasemore abatement.

While the dislocation costs of carbon pricing are targeted to specific agents in the economy, theburden of public R&D costs is broad. From a political standpoint, then, a technology policy maybe easier to implement than a strong mitigation policy. The previous discussion revealed that thedesirability of that technology policy depends on the scope of the mitigation policy. However, iftechnological improvement can enable more efficient mitigation policies, the current scope fortechnology policy is expanded, reflecting not only the gains to utilizing the technology underbetter carbon pricing conditions, but also the benefits from improving the mitigation incentives.

A question is then whether the improved credibility of significant mitigation incentives willspur private innovation as well. In part, the answer depends on how those policy expectations areformed—and why a credible mitigation policy is limited in its current implementation.

In this section, we evaluate the social premium to innovation when the environmentalregulation adjusts according to the policy constraint. The approach differs from most previousresearch on dynamic policy adjustment and induced innovation, in which policy makers want to

15 This example is not completely unfounded; in Fischer and Newell (2004), the effect of a $25/ton C price is to roughlydouble optimal R&D spending.


respond by equating marginal costs and benefits;16 however, the methods are similar. We continueto abstract from questions of the policy horizon and discounting; although these dynamic issuesare certainly important, the key intuition is summarized in this simple, three-stage game of policyadjustment. In the first stage, the firm or the social planner chooses an amount of innovation. Inthe second stage, the regulator adjusts the emissions price, abatement target, and allocation ofemissions permits. In the third stage, the firm conducts abatement, and profits and welfare arerealized. In this framework, let us analyze the innovation incentives in general and then underspecific policy adjustment mechanisms.

3.1. General case

Abstracting from the form of the constraint on policymakers underlying how the price is adjustable,consider that the price is now a function of the degree of innovation, so p(r). Let p(0)=p0 be equalto the price in the fixed-price scenario. Let us denote the adjustable price equilibrium with thesubscript “adj”. Substituting p(r) into Eq. (6), we have Wadj=p(r)(2b−p(r)) / (2c(1−r)) . Then themarginal social benefits from innovation are larger than with the fixed price if p′(r)N0:

AWadj

Ar� AW

Ar¼ bð1� /Þ

cð1� rÞ pVðrÞ: ð9Þ

In other words, if innovation does allow regulators to raise prices closer to their Pigouvianlevel, additional welfare gains arise from environmental R&D.

From the private perspective, an important issue is the extent to which firms recognize theirinnovations will change the market equilibrium and induce a price adjustment. Recognizing theprice adjustment may be an incentive or disincentive for innovation, depending on whether firmshave positive emissions liabilities. For example, a price fall has the added effect of loweringemissions payment costs; on the other hand, net sellers of emissions permits (or subsidy recipients)receive additional benefits from a price rise. Assuming the innovator captures a share of the totalchange in private rents, the additional return from recognizing the price adjustment is

ð1� rÞ ApadjAr

� ApAr

� �¼ �pVðrÞðM � F � AÞ: ð10Þ

If we assume that all permits are grandfathered in equilibrium, the price adjustment does notaffect the return to innovation in the private markets. However, if the polluter is a net emissionstaxpayer, the social premium to innovation is necessarily raised by the possibility of upward priceadjustments—and lowered by downward price adjustments. A final caveat regards the nature ofthe R&D market and whether innovators are really able to appropriate a constant share of the totalgains. In a decentralized market, the emissions price adjustment will also affect the extent towhich other firms benefit from adopting the innovation; Fischer et al. (2003) term this change inthe willingness to pay by adopters to license an innovation the “adoption price effect.” Thus, anupward adjustment may have an added inducement to private innovation, while a downwardadjustment reduces private innovation incentives. In both cases, an adoption price effect wouldtend to offset some of the effect of the price adjustment on the marginal social gains to R&D andtemper the impact on the social premium.

16 See, e.g. Montgomery and Smith (in press), Fischer et al. (2003), Petrakis and Xepapadeas (1999) and Kennedy andLaplante (1999).


In summary, using innovation policy to promote stronger environmental policy does notalways imply a bigger return to public investment in R&D. Only if emissions prices can risecloser to the level of the marginal benefits are the overall welfare gains from additional innovationlarger. On the other hand, the social premium may rise in more cases of policy adjustment, sinceprivate incentives for innovation may be lessened by the adjustment mechanism. The net effectdepends on the difference between Eqs. (9) and (10) (and any adoption price effect).

Of course, the adjustment mechanism depends on the policy constraint imposed on (or self-imposed by) the regulator. We consider several examples of rationales for price adjustment: a totalcost constraint, a total emissions constraint, a consumer price impact constraint, and fixed costs toimplementing the policy. We find that only the latter two kinds of rationales can provide foremissions prices that rise with innovation.

3.2. Total cost constraint

One target of policy concern may be the total cost burden of regulation, resulting in a constraintthat profits be held constant at the initial level of regulation. Here, then, it is important to accountfor the distribution of the emissions rents. Since taxing the remaining (inframarginal) emissionsonly lowers profits—and thereby the scope for abatement under a total cost constraint—let usassume that the policymaker exempts this threshold amount from taxation, setting F=M−A. Forexample, this policy could be implemented with a tradable emissions permit scheme in which allof the permits are allocated to the firm in a lump-sum fashion.17

Formally, the policymaker continues to choose p to maximize abatement, subject to the costconstraint, and adjusts the emissions cap (and allocation F) according to this price. Let us denotethis scenario with the subscript “cost.” Since no permit liability remains with grandfatheredpermits, the net costs are simply those of abatement. After innovation, profits with grandfatheredpermits are

pcost ¼ � cð1� rÞ2

A2 ¼ � p2

2cð1� rÞ : ð11Þ

To see how p adjusts to innovation, let P0 be the initial permit price. Initial profits are thendefined as π0=−p02 / (2c). As abatement costs fall, the policymaker tightens the emissions abate-ment requirement (thereby adjusting the price), maintaining constant profits, such that πcost=π0.

18

That adjusted price is

pcost ¼ p0ffiffiffiffiffiffiffiffiffiffiffiffi1� r:

pð12Þ

Note that this constraint implies that the permit price falls as costs fall (pcostVðrÞ ¼ �p0=ð2 ffiffiffiffiffiffiffiffiffiffiffi

1� rp Þ), but not to the full extent of the cost reduction, since some of the reduction is used to

allow abatement to increase. As a consequence, from Eq. (9), since the permit price falls as policyis adjusted, the marginal benefits of innovation are lower than they would be if the price remained

17 Equivalently, an emissions tax could be levied on emissions above a grandfathered amount, with refunds availablebelow. With automatic adjustment to cost changes, the two policies will be identical.18 We note that with linear marginal costs, this policy constraint is equivalent to what would arise if total abatementsubsidies (pA) had to be constant. A difference arises in the constant level of profits to the firm: With a subsidy programand F=M profits would be π=p0A0−p2 / (2c(1−r)). However, the policy adjustment response to cost reductions (whichdetermines the innovation incentives) would be the same in either scenario.


fixed. In another sense, since costs are fixed, there exist no gains from reducing abatement costs,only from increasing abatement.

If a tax is levied on inframarginal emissions (or some permits are auctioned, rather thangrandfathered) so that firms have a positive net emissions liability, the constant-profit constraintcould allow prices to rise as costs fall. However, this implies that to maintain any level of profits,the emissions tax must be lower to account for the transfer payments, meaning some abatement isforegone.

If industry realizes the policy response to cost reductions–i.e., that innovation will not changetotal costs–it will have no incentive to innovate. In this case, the social premium equals themarginal welfare gain from cost reductions (Rcost=∂Wcost /∂r). Thus, this expression is themaximum possible social premium, since any private innovation that occurs (such as due tomyopia about the adjustment process or adoption rents from other regulated firms) would reducethe social need for innovation.

To compare to the social premium with the exogenous price policy, we can rewrite Eq. (7) plusEq. (9) as

AWcost

Ar¼ /G⁎ð1� rÞ1=2 ð13Þ

where ϕ=p0 /b. In this case, the marginal gains (and social premium) to innovation are a linearfunction of the rate of emissions pricing. Furthermore, the slope falls as more innovationoccurs.

If we compare these gains to the social premium without price adjustment, we get

Rcost � R ¼ /ðð1� rÞ1=2 � 2þ /ð2� rÞÞG⁎: ð14Þ

The premium with this endogenous policy is lower than with the exogenous price policy whenϕb1 /2, but it becomes higher for greater levels of social cost pricing and lower spillover rates.19

In other words, reduced costs lead to higher levels of abatement when prices are fixed, soadditional innovation leads to larger increases in welfare with exogenous emissions pricing.However, when marginal damages are more fully priced, the private market has better incentivesto innovate when prices are fixed, while the government must provide the support for innovationwhen price adjustments are expected.

3.3. Total emissions constraint

For another example, consider the common case of an emissions cap–a fixed quantity ofabatement, rather than a fixed price–and denote this equilibrium with the subscript “cap”. FromEq. (3), which defines A, and correspondingly that A=A0=p0 /c, we get

pcap ¼ ð1� rÞp0 ð15Þwhich implies that innovation causes prices to adjust downward evenmore than with constant totalcosts (p ′cap(r)=−p0). Thus, assuming no net emissions liability, the social premium is lower thanwith a fixed emissions price. Furthermore, the absence of the opportunity to expand abatement,combined with the presence of some private incentives to innovate, means that less of a role for

19 I.e., when ϕN1− (1−r)1 / 2 / 2.


public investment exists under a fixed quantity policy than when emissions policy adjusts to keepcosts constant. Of course, this comparison assumes that the initial price of permits, which reflectsthe constraint on policymakers, is the same.

Some of these results are not so surprising. As is well recognized in the literature on innovationincentives, a fixed-price policy induces more innovation than when prices decline, the mostextreme case of which is a fixed-quantity policy, when permit rents are grandfathered.20 Thereason for this phenomenon is that abatement expands most when emissions prices remain fixed.A constraint on policy that limits costs or limits abatement then calls for less innovation. Only ifthe constraint (or specifically the adjustment to the constraint) eliminates more private incentivethan social return does the social premium to R&D rise.

Perhaps more surprising is that these common characterizations of the political constraints oncarbon pricing do not call for emissions prices to rise and generate a larger social benefit toinnovation than a currently fixed emissions price––even a below-optimal one. The idea thatinnovation can allow regulators to impose much more stringent regulation (that is, expandingabatement to the extent that emissions prices and marginal costs rise) must rest on more complexpolitical motivations or cost structures. Two such potential structures come to mind: a constrainton the marginal costs of production, or fixed costs to initiating the policy.

3.4. Marginal cost constraint

This first example lies with a greater concern about downstream impacts than the costs to aparticular sector. The assumption of a constant emissions price, as in previous section's analysis,seems to suppose that policymakers are constrained by the marginal abatement costs they mayimpose on polluters. This situation might arise from a focus on marginal production costs, out ofconcern for industrial competitiveness or the burden on consumers, since, when emissions are afunction of output, marginal abatement and tax costs are reflected in higher output prices.

Although we have abstracted from product markets, we can still consider the effect of amarginal production cost constraint. In our general framework, profits have represented thechange in producer surplus due to the policy, meaning abatement costs can represent both directcosts and opportunities foregone. In focusing on marginal production costs, however, we need tobe more explicit. Let Q be output and suppose that M=mQ. The policy-induced change inmarginal production costs to be evaluated will be −∂π /∂Q.

The form of abatement is now relevant, and two types are possible. One involves abatementcosts that are independent of output, as might be the case for some end-of-pipe treatments forwhich the cost depends on effluents removed. The other involves abatement costs that dependupon output, such as technologies that remove a certain share of emissions.

When abatement is independent of output, then π=p(A+F−mQ)−c(1− r)A2 /2 and −∂π /∂Q=pm; in other words, the product price rises according to the value of the embodied emissionsin an additional unit of output. To keep this regulatory burden on marginal production costsconstant, p must then also be constant. Note that this result also holds if we write A=aQ andmaximize with respect to output and the abatement rate a.

Suppose instead that abatement costs depend on output levels in a different manner, so thatp ¼ pðAþ F � mQÞ � cð1� rÞ A2

2Q0

Q (where Q0 is baseline output, so that c retains the samemeaning as before) . In this case, marginal abatement costs depend on the abatement rate. Taking

20 Again, an exception is if innovators recognize that they can reduce their own emissions payments, and those savingsoutweigh the lower royalties they can charge adopting firms when prices fall (Fischer et al., 2003).


the first-order conditions for abatement and output, we have A ¼ pcð1�rÞ

QQ0

� �and � Ap

AQ ¼ pm�cð1� rÞ A2

2Q0

Q2. Substituting in the expression for A, we get

� ApAQ

¼ pm� p2

2cð1� rÞQ0: ð16Þ

In this case, marginal costs decrease with r, meaning that technological change can allowemissions prices to rise while holding product prices constant. Totally differentiating the con-straint (Eq. (16)), we see that p′(r)=p2 / (2(1− r)(mc(1− r)Q0−p2)). This expression is positive, aslong as marginal production costs are increasing in the emissions price.

3.5. Other rationales

A related example is if technological advances reduce baseline emissions rates (m)–that is, inthe absence of an emissions pricing policy–rather than reducing the costs of further abatement.(This corresponds to a shift, rather than a pivot, in the marginal abatement cost curve.) At a givenprice, this shift lowers total costs without changing abatement incentives, which would allowprices to rise under a total cost constraint. It also lowers per-unit production costs. More generally,if additional innovation lowers marginal production costs (as opposed to abatement costs),emissions prices can rise without impacting consumers and downstream firms. If abatementincentives are also unaffected, then the adjustment necessarily raises the social premium toinnovation. This example raises a caveat to the preceding analysis regarding the functional formsused, particularly the effect of innovation.

Another example of a situation allowing innovation to increase the politically acceptable emissionsprice is if initiating a mitigation policy involves significant fixed costs. In this case, onemight want todelay implementation until innovation enables sufficiently large abatement benefits. Private in-centives to delay the regulation then can reduce their incentives to innovate, although the increasedlikelihood of a significant mitigation policy may encourage some to prepare. The social premium canthen be bolstered by the lack of private innovation and the desire to implement the policy.

This last example is more relevant to policy initiation than policy adjustment. In consideringpolicy ratcheting, the preceding analysis reveals it is important to understand the specific goalsand constraints of the policymakers. In most situations, it would be unusual for cost-reducinginnovation to lead to an increase in marginal costs, as implied by an emissions price rise. If thereal issue is burdens on downstream consumers or competitiveness, then certain kinds oftechnological progress–but not all–may allow for fuller emissions pricing and thereby meritgreater prominence in the public technology policy portfolio.

4. Conclusion

A good deal of emphasis in environmental policy is often placed on promoting technologicalsolutions that reduce the costs of emissions abatement. While the true costs and benefits ofreducing emissions determine the scope of the gains from lowering abatement costs, the actualreturns to investments–both public and private–depend as much on the policy environment. Thissimple discussion reveals that the relative importance of a public role for environmentally friendlyR&D is highly sensitive to both the degree to which the emissions externality is internalized forprivate markets and the extent to which spillovers prevent private actors from reaping the fullbenefits of their innovations.


Private incentives to reduce abatement costs depend on how expensive carbon emissions areand to what extent they can capture the gains to their innovations. If the price of additionalemissions does not reflect the full social cost, less abatement will be performed, creating lessincentive to reduce abatement costs. Spillover effects further reduce the private incentive toinnovate. From a social perspective, insufficient abatement means that more is to be gained fromcost reductions that expand abatement, since the marginal benefits outweigh the costs; however,since a cost reduction will have less of an impact when carbon prices are low, there is also lessscope for cost savings.

As a result, the role for publicly supported innovation is strongest when some spillover effectsare present and at least a moderate share of the social costs–including the marginal damages ofemissions–is reflected in the price. At the extremes, if emissions are unpriced, even if spilloversare complete, no public support for R&D to reduce the costs of pure abatement technologies isjustified, since the innovations will not be used. If emissions are fully priced and there are nospillovers, then no market failure remains to justify public support. However, the case for manytechnologies–particularly those related to energy use–will tend to fall somewhere in between, andthese principles can help us identify which kinds of public investments are most likely to pay off.These principals should be weighed along with other factors beyond the scope of this paper thatalso influence the desirability of public and private R&D, including the potential for costreductions, the cost of R&D, the likelihood of success, and option values for backstoptechnologies.

If the degree of regulation is constrained by the costs (private, fiscal, consumer, or political),then innovation can allow for stricter abatement targets. However, if the industry realizes thatpolicy makers will respond to technical advances by ratcheting abatement targets, it may have lessincentive to innovate, preferring instead to avoid regulation. The net effect may raise or lower thesocial premium to innovation. Of course, energy market actors may not be so coordinated, andindividual innovators may not expect their R&D to have sufficiently widespread impacts as tolower industry abatement costs and induce a significant climate policy response.

From a social perspective, the gains to innovation arise not only from reducing abatementcosts, but also from increasing abatement. Relative to a fixed emissions price, those latter gainsonly increase if abatement increases to a greater extent—that is, if the price adjusts upwards.Thus, if technological progress can help stakeholders accept an even higher emissions price,closer to the marginal damage costs, public support for mitigation-related R&D can be yetmore worthwhile. However, there are few straightforward rationales for why cost-reducinginnovation should lead to higher emissions prices, and thereby higher marginal abatementcosts.

Ultimately, it is the pricing of emissions to reflect the environmental burden that improveswelfare and determines the utilization of innovations. This study demonstrates, in basic terms,that expanding public support for environmentally friendly technologies is not necessarily theappropriate response to a lax emissions policy. It adds to a substantial literature showing thattechnology policy is a poor substitute for mitigation incentives: relying on abatement effort aftercosts are brought down requires large up-front investments and forgoes many cost-effectiveopportunities to reduce emissions (see, e.g., Schneider and Goulder, 1997; Popp, 2006; Fischerand Newell, 2004). Nonetheless, technology policy plays an important role in enabling societyto meet uncertain future challenges, and perhaps in enhancing the credibility of future policytargets. While mitigation policy must be the engine for reaching environmental goals; tech-nology policy can help that engine run faster and more efficiently, but it only helps if the engineis running.


References

Biglaiser, Gary, Horowitz, John K., 1995. Pollution regulation and incentives for pollution control research. Journal ofEconomics and Management Strategy 3, 663–684.

Bohm, Peter, Russell, Clifford F., 1985. Comparative analysis of alternative policy instruments. In: Kneese, A.V.,Sweeney, J.L. (Eds.), Handbook of Natural Resource and Energy Economics. North-Holland, Amsterdam.

Downing, Paul G., White, Lawrence J., 1986. Innovation in pollution control. Journal of Environmental Economics andManagement 13, 18–29.

Fischer, Carolyn, Newell, Richard, 2004. Environmental and Technology Policies for Climate Change. RFF DiscussionPaper, vol. 04–05. Resources for the Future, Washington, DC.

Fischer, Carolyn, Toman, Michael, 2001. Environmentally and economically damaging subsidies: concepts and illus-trations. RFF Climate Issues Brief #14. In: Toman, Michael A. (Ed.), Climate Change Economics and Policy: an RFFAnthology. Resources for the Future, Washington, DC.

Fischer, Carolyn, Parry, Ian W.H., Pizer, William A., 2003. Instrument choice for environmental protection whentechnological innovation is endogenous. Journal of Environmental Economics and Management 45 (3), 523–545.

Gerlagh, R., 2006. ITC in a global growth-climate model with CCS. The value of induced technical change for climatestabilization. The Energy Journal, Special Issue on Induced Technological Change and Climate Change 55–72.

Griliches, Zvi., 1992. The search for R&D spillovers. Scandinavian Journal of Economics 94, S29–S47 (supplement).Hall, Bronwyn H., 1996. The Private and Social Returns to Research and Development: What Have We Learned? In:

Smith, Bruce L.R., Barfield, Claude E. (Eds.), Technology, R&D, and the Economy. The Brookings Institution and theAmerican Enterprise Institute, Washington, DC.

Jaffe, Adam B., Stavins, Robert N., 1995. Dynamic incentives for environmental regulations: the effects of alternativepolicy instruments on technology diffusion. Journal of Environmental Economics and Management 29, S43–S63.

Jaffe, Adam B., Newell, Richard G., Stavins, Robert N., 2003. In: Mäler, K.-G., Vincent, J.R. (Eds.), Technological changeand the environment. Chapter 11 in Handbook of Environmental Economics, vol. 1. Elsevier Science, Amsterdam,pp. 461–516.

Jones, C.I., Williams, J.C., 1998. Measuring the social return to R&D. Quarterly Journal of Economics 13 (114),1119–1135.

Jung, Chulho, Krutilla, Kerry, Boyd, Roy, 1996. Incentives for advanced pollution abatement technology at the industrylevel: an evaluation of policy alternatives. Journal of Environmental Economics and Management 30, 95–111.

Kemp, Rene, 1997. Environmental Policy and Technical Change: A Comparison of the Technological Impact of PolicyInstruments. Edward Elgar Publishing, Cheltenham, UK.

Kennedy, Peter W., Laplante, Benoit, 1999. Environmental policy and time consistency: emission taxes and emissionstrading. In: Petrakis, E., Sartzetakis, E., Xepapadeas, A. (Eds.), Environmental Regulation and Market Power:Competition, Time Consistency and International Trade. Edward Elgar Publishing, Northanpton, MA.

Magat, Wesley A., 1978. Pollution control and technological advance: a dynamic model of the firm. Journal of Envi-ronmental Economics and Management 5, 1–25.

Milliman, Scott R., Prince, Raymond, 1989. Firm incentives to promote technological change in pollution control. Journalof Environmental Economics and Management 17, 247–265.

Montgomery, W.D., Smith, A.E., in press. Price, quantity and technology strategies for climate change policy. In Human-Induced Climate Change: An Interdisciplinary Assessment, edited by M. Schlesinger, H. Kheshgi, J. Smith, F. de laChesnaye, J.M. Reilly, T. Wilson, and C. Kolstad. Cambridge: Cambridge University Press.

Nadiri, M. Ishaq., 1993. Innovations and technological spillovers. Working Paper, vol. 4423. National Bureau ofEconomic Research, Cambridge, MA.

Newell, Richard G., Jaffe, Adam B., Stavins, Robert N., 1999. The induced innovation hypothesis and energy-savingtechnological change. Quarterly Journal of Economics 941–975 (August).

OECD, 2003. Environmentally Harmful Subsidies: Policy Issues and Challenges. OECD Publishing, Paris, France.Parry, Ian W.H., Pizer, W.A., Fischer, C., 2003. How large are the welfare gains from technological innovation induced by

environmental policies? Journal of Regulatory Economics 23 (3), 237–255.Petrakis, Emmanuel, Xepapadeas, Anastasios, 1999. Does government precommitment promote environmental inno-

vation? In: Petrakis, E., Sartzetakis, E., Xepapadeas, A. (Eds.), Environmental Regulation and Market Power:Competition, Time Consistency and International Trade. Edward Elgar Publishing, Northanpton, MA.

Popp, D., 2006. R&D subsidies and climate policy: is there a “free lunch”? Climatic Change 77 (3–4), 311–341.Popp, D., 2004. ENTICE: endogenous technological change in the DICE model of global warming. Journal of

Environmental Economics and Management 48, 742–768.


Requate, Till, Wolfram, Unold, 2001. On the Incentives by Policy Instruments to Adopt Advanced Abatement Technologyif Firms are Asymmetric. Journal of Institutional and Theoretical Economics JITE 157 (4), 536–554 (19).

Stavins, Robert N., 2000. Market-based environmental policies. In: Portney, Paul R., Stavins, Robert N. (Eds.), PublicPolicies for Environmental Protection. RFF Press, Washington DC.

Schneider, S.H., Goulder, L.H., 1997. Achieving low-cost emissions targets. Nature 389, 13–14.Ulph, David, 1998. Environmental policy and technological innovation. In: Carraro, C., Siniscalaco, D. (Eds.), Frontiers of

Environmental Economics. Edward Elgar, Cheltenham, UK.Zerbe, Richard O., 1970. Theoretical efficiency in pollution control. Western Economic Journal 8, 364–376.

emissions pricing, spillovers, and public investment in environmentally friendly technologies

Documents