Ecological Economics 54 (2005) 175 – 195

www.elsevier.com/locate/ecolecon

SURVEY

Dynamic incentives by environmental policy

instruments—a survey
Till RequateT
Department of Economics, University of Kiel, Olshausenstrabe 40, 24118 Kiel, Germany
Accepted 1 December 2004
Available online 26 April 2005

Abstract

In this paper I survey and discuss recent developments on the incentives provided by environmental policy instruments for
both adoption and development of advanced abatement technology. A main conclusion to be drawn from the literature is that
under competitive conditions market based instruments usually perform better than command and control. Moreover, taxes
may provide stronger long term incentives than tradable permits if the regulator is myopic. If the government can anticipate
new technology or is able to react on it optimally, regulatory policies by virtue of administered prices (taxes) and policies by
setting quantities (issuing tradable permits) are (almost) equivalent. The literature also shows that under competitive
conditions there is no difference between auctioning permits and grandfathering. Moreover, timing and commitment of
environmental policy is not crucial for adoption under competitive conditions. Commitment has positive incentive effects,
however, if an R&D sector has market power. In the presence of market imperfections the ranking of the different policy
instruments is ambiguous.
D 2005 Elsevier B.V. All rights reserved.

Keywords: Emission taxes; Auctioned permits; Grandfathering; Emission standards; Technology adoption; R&D

JEL classification: L5; Q2; Q28

1. Introduction known that among the wide array of pollution control

instruments, economists prefer those which provide
This paper addresses recent developments on the incentives through prices rather than through com-
incentives provided by environmental policy instru- mand and control. The main advantage of market-
ments to spur both R&D and the adoption of emission based instruments such as emission taxes, subsidies
reducing or energy saving technology. It is well on abatement, and different regimes of tradable
permits (notably free and auctioned permits) is their
(static) cost efficiency. The theoretical prediction that
T Tel.: +49 431 8804424; fax: +49 431 8801618. firms will take advantage of the efficiency gains of
E-mail address: [email protected]. those instruments has been well confirmed by
0921-8009/$ - see front matter D 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.ecolecon.2004.12.028
176 T. Requate / Ecological Economics 54 (2005) 175–195

empirical observations, in particular on markets for excellent survey on the empirical literature I refer to
tradable permits. Meanwhile those instruments are Jaffe et al. (2002).
much better understood and more widely accepted by With respect to theoretical studies it is expedient
both the public and by policy makers, as can be seen to distinguish between adoption of new, though
from a growing number of countries which employ yet existing technology, on the one hand, and research
these instruments. Notably, Norway, Sweden, Finland, and development of new technology, on the other.2
the Netherlands, France, Slovenia, and even Russia Downing and White (1986), Malueg (1989), Milliman
have implemented emission taxes for various pollu- and Prince (1989) and more recently Jung et al. (1996)
tants, in particular SO2 and NOx (although the made the first attempts to rank environmental policy
variation of tax rates is quite large, ranging from a instruments, in particular, taxes, subsidies, auctioned
few cents in some countries up to about 2000 EURO permits, free permits, and emission standards with
per ton of NOx in Sweden). Targeting CO2-emissions, respect to their incentives to adopt less polluting
notably Denmark, Germany, and the Netherlands technology. Their results, to be summarized below,
charge taxes on energy. By contrast, the United States have later been challenged by different authors who
were the first to introduce markets for pollution made the point that those authors mainly compare
permits on a large scale, notably for SO2 and NOx . aggregate cost savings of a whole industry and that
Other countries such as Australia use permit markets these cost savings are not equivalent to incentives
for local and regional pollutants such as salt effluents of a single firm to adopt new technology in
into rivers. New Zealand issues land development equilibrium.
permits whereas Iceland uses tradable permits to More recently other authors have included R&D
allocate fishing quotas. Very recently the European into the analysis of environmental policy incentives.
Union has launched a directive for CO2 permit trading Their contributions can once more be divided into
in Europe to be started in January 2005. This will then two strands of literature. One road goes along the
be the world’s largest market for pollution allowances. methodology of microeconomics, in particular
Kneese and Schulze (1975) have pointed out industrial organization, using concepts of game
early that, besides the issue of static efficiency, the theory in order to analyze strategic behavior in
extent to which policy instruments bspur new equilibrium. Most of these models are partial
technology toward the efficient conservation of the equilibrium models which are static or quasi-
environmentQ is one of the most important criteria dynamic in the sense that they allow for sequential
on which to judge the performance of environ- decisions taken by a regulator, by an innovating
mental policy instruments. It took quite a while, sector, and by the firms which adopt new technol-
however, until researchers started to inquire those ogy. Except for Parry et al. (2003) the vast majority
long run incentives and to point out the differences of authors do not explicitly capture the aspect of
between the different policies. By several reasons time. A second road follows the methodology of
there is still little that we know empirically to endogenous growth theory. Since it would burst this
which extent those instruments perform differently. survey to treat both approaches, I will concentrate
Firstly there is hardly any chance to make experi- on the partial equilibrium microeconomic contribu-
ments and empirical comparisons of instruments tions, and I will not pursue the growth approach in
under similar economic conditions. Secondly, sev- this paper.
eral countries regulate the same externality by Hence this paper is organized as follows.
several instruments.1 Hence this paper concentrates Section 2 starts with introducing some basic con-
on the theoretical inquiries on adoption and innova-
tion incentives of environmental policy. For an
2
This terminological distinction is not always sharp in the
literature. For example pioneers in this area such as Downing and
White (1986) talk about binnovationQ whereas Milliman and Prince
1
For example Germany, uses energy taxes, feed in subsidies for (1989) use the notion of btechnical changeQ in their studies although
electricity generated by wind and solar power, and from 2005 on it is mainly incentives for badoptionQ of existing technology which
also tradable permits to regulate the emissions from fossil fuels. is the subject of both contributions.
 T. Requate / Ecological Economics 54 (2005) 175–195 177

cepts, such as abatement cost functions, a listing of prices, there is no need to explicitly pay attention to
different regulatory pollution control instruments, the output market.3
the question of how to compare incentives for Pollution-reducing technological progress can now
adoption, diffusion and R&D, and finally a be defined by declining abatement costs for any level
classification of different forms of timings and of emissions. Before investment, a typical firms’
commitment strategies. In Section 3 I survey the technology is represented by its abatement cost curve
results on incentives of adoption whereas in Section C 0. Adoption of new technology leads to a lower
4 I deal with the literature on environmental R&D marginal abatement cost curve C I (see Fig. 1) with
and the interplay between technology adoption, on
 CIVðeÞb  C0VðeÞ for all ebemax
0 :
the one hand, and incentives to engage in research
and development of advanced abatement technol- In addition, buying and installing the new tech-
ogy, on the other. In Section 5 I draw some nology involves a fixed cost F N 0. If, by contrast,
conclusions and give some outlook for further pollution is proportional to output and if there is no
research. further short run abatement technology, pollution-
reducing technological progress can be modelled by
decreasing emission coefficients (i.e. reducing the
2. Some basic concepts emission–output ratio). In this case, lower emission
coefficients do not necessarily lead to declining
2.1. Definitions and model frameworks marginal abatement costs if emission levels are
already low.
For a polluting firm there are basically two
strategies to reduce pollutants: one is to reduce 2.2. Pollution control instruments
gross emissions by reducing output, another one is
to keep output constant and to reduce emissions by Unregulated market forces do not necessarily
employing an abatement technology. In general, a induce adoption of pollution reducing technology, a
mix of both will be optimal. Experts usually feature which calls for regulation. Economists dis-
distinguish between two types of abatement tech- tinguish mainly between two types of pollution
nologies, end-of-pipe technologies, on the one hand, control instruments: command and control and
and process integrated technologies, on the other. market based instruments. The most common instru-
The latter leads to a decline in gross emissions ments of command and control are technological
whereas, with the former, gross emissions remain standards (a regulatory authority might prescribe the
unchanged and are subsequently decreased, for firms to adopt the best available technology),
example by using a filter. In both cases a firm’s
abatement cost is nothing else than its forgone profit
incurred by reducing emissions. A firm’s abatement 3
To become a bit more formal we denote by C( q,e) the joint
cost function C i (e i ) represents the cost to reduce costs which a typical firm incurs to produce q units of output with
emissions from the laissez-faire emission level emax i no more than e units of emissions. For this definition it is not
to some lower level e i be max i and satisfies the relevant, whether the firm has an end-of-pipe or an integrated
following properties: C i (e i )N0, C V(e i i )N0 and
technology. The firm’s profit is then given by p( q,e)=pqC( q,e).
C i U(e i )N0 for e i bemax , and C (e )=0 for e i zemax . If In the absence of regulation the competitive firm chooses q and e
i i i i
such that p=C q ( q,e) and C e ( q,e)=0 providing a maximal profit
the product market is competitive, decisions on p max and an emission level e max. It is plausible to assume C q N0 and
output are optimal. In this case, the abatement cost C qq z0. Further, C e ( q,e)N0, C ee N0, and C eq ( q,e)N0 for ebe max.
functions account for the optimal output adjustment. We refer to C e ( q,e) as the marginal abatement costs. If now the
It is straightforward to show that such an abatement firm is constrained to emit no more than e units of the pollutant, the
˜
cost function can be derived from a firm’s joint cost rule induces output q(e) and a reduced profit p (e)=p( q(e),e). The
˜
full abatement cost can then be defined as C̃(e)=p maxp (e), which
function which may incorporate both cost of pro- is the forgone profit resulting from reducing emissions from e max to
duction and cost of abatement. Note further that if e. It is easy to show that the reduced abatement cost function C̃(e)
firms are small and thus cannot influence the output has the same properties as C( q,e) with respect to e.
178 T. Requate / Ecological Economics 54 (2005) 175–195

virtue of their static efficiency. For, under compet-

itive conditions, market based instruments lead to
equalization of marginal abatement costs across
firms, a necessary condition for achieving an
aggregate emission target at least costs. Moreover,
it is easy to see that regulation by prices (taxes or
subsidies) and regulation by quantities (tradable
permits) is equivalent if markets are competitive
and the number of firms is fixed. Under free entry,
taxes and permits are also equivalent, as Spulber
(1985) has shown, whereas subsidies will usually
induce excess entry. The dynamic properties, in
particular the innovation incentives of these instru-
ments are much more complex and, therefore, are
subject to discussion below.
Fig. 1. Single firm’s investment incentive under taxes.

2.3. How to compare incentives for adoption, diffu-

emission standards (firms are constrained by an sion and innovation
absolute upper emission level), and finally, so-called
generation performance standards, sometimes When comparing the incentives for adoption,
referred to as relative standards. The latter means diffusion, and innovation of different policy instru-
that firms’ face a cap on the ratio of emissions per ments, it is natural to begin with the incentives for
output. This sort of standard is most commonly adoption. In a first step this can at best be analyzed
applied in reality. Market based instruments, by from the perspective of a single firm. The incentive
contrast, provide incentives to reduce emissions for adoption is simply given by the firm’s additional
through prices, and firms are free to decide how total profit from switching to the new, exogenously
much they want to emit or to abate. The most given technology. Let us study this incentive in
commonly used market based instruments are emis- particular for market based instruments like an
sion taxes, subsidies on abatement of emissions, and emission tax or a system of tradable permits, and let
tradable permits. Under emission taxes and abate- p denote a tax rate or a permit price, respectively.
ment subsidies the prices for emissions are admin- Assuming interior solutions, firms’ cost minimization
istrated by a regulator. If she levies a linear tax per requires marginal abatement cost to equal the price of
unit of a pollutant or pays a subsidy per unit of emissions for both technologies:
abated emissions, then each firm has to pay (gets)
the same marginal price for each unit of pollution it  C0Vðe0 Þ ¼ p ð1Þ
emits (abates). Under permits, by contrast, a firm  C0VðeI Þ ¼ p ð2Þ
must hold one permit for each unit of pollution it
wants to emit. Usually firms can trade those permits which yields emission levels e 0( p) and e I ( p). Then a
with other firms. There are two allocation schemes typical firm decides to adopt the new technology if
for permits: free allocation according to historical and only if
emission or output levels (often referred to as ˇ ˇ
grandfathering or benchmarking, respectively) or C0 e0 pÞÞ þ pe0 ð pÞ  CI ðei ð pÞÞ  peI ð pÞ  FN0
auctioning off the permits in which case the firms ð3Þ
have to pay for each unit of the pollutant they are
going to emit. In contrast to the tax rate, the market whereas firms are indifferent about adopting the new
price of permits is determined endogenously by the technology if
market mechanism. Economists usually prefer mar-
ket based over command and control instruments by C0 ðe0 ð pÞÞ þ pe0 ð pÞ ¼ CI ðeI ð pÞÞ þ peI ð pÞ þ F ð4Þ
 T. Requate / Ecological Economics 54 (2005) 175–195 179

Inequality (3) and Eq. (4) are the driving forces for the such that the marginal (expected) benefit equals the
analysis.4 Eq. (4) determines the number of firms marginal cost of R&D. Denicolo (1999, p. 186) points
which adopt the new technology in equilibrium. The out that the innovation incentive of the different
rate of diffusion refers to the percentage of firms instruments can be correctly measured by the inno-
adopting the new technology. The incentives for vator’s respective profits if her bR&D investment
adoption and the rate of diffusion are interrelated. In cannot affect the nature of the innovation and hence
particular in the case of tradable permits the market the reduction in effluent emissions that it entails.Q
price of permits depends on the number of firms Otherwise, bit is the marginal profit that matters to
which adopt the new technology. determine the innovator’s incentive to invest.Q
The incentive for innovation refers to the benefit a It is important to note that the distinction between
firm enjoys from developing and inventing a new investment into adoption and into innovation is not
technology. Firms engaging in research and develop- always sharp. Some authors such as Phaneuf and
ment of pollution-reducing technologies are either part Requate (2002), Petrakis and Xepapadeas (2003), or
of the polluting industry itself, or they engage in R&D Gersbach and Requate (2004) assume cost functions
exclusively in order to sell or license their new of the form C(e,k) where e denotes emissions and k
technology to a different polluting sector. In the first can be interpreted as both investment into abatement
case, the innovator’s benefit is determined by the equipment or R&D effort. In the literature survey
change of her compliance costs (including short run following below, we subsume a paper under a model
abatement cost, revenues from emission subsidies or of innovation (in contrast to a model of adoption)
permit sales, or expenditures for emission taxes or when the respective model contains at least one of the
permit purchases, respectively) and the change of her following aspects. Either, there is a stochastic element,
profit on the output market. In addition, the innova- i.e. the size of innovation, its date, or the R&D
tor’s benefit might increase by accruing license fees, success is uncertain, or secondly, a patent is granted
and it may be reduced by other firms which imitate his on the innovation, or thirdly, spillovers occur, or
technology. If the innovator’s objective is mainly to finally, imitation is possible.
sell or license his new technology, his profit depends
on the polluting firms’ willingness to pay for the new 2.4. Possible timing and commitment strategies of the
technology, which in turn is determined by the regulator
adoption costs, by the change of compliance costs,
and by the profits those firms accrue from adopting When analyzing diffusion, innovation, and tech-
the new technology. Again, the possibility of imitation nological progress, it is important to distinguish the
has a diminishing effect on the innovator’s profit. Fig. possible timing and commitment strategies of the
1 illustrates the change of an adopting firm’s com- regulator. This basically boils down to the question
pliance costs under an emission tax. Depending on the about who is the first to move, the regulator or the
kind of regulation, a potential innovator will put a firms. If the regulator moves first and is able to make
certain effort into R&D. A higher effort may either a commitment to the level of her policy instrument,
lead to a higher degree of innovation (for example a we refer to ex ante regulation or ex ante commitment.
more radical shift of both the abatement and the A myopic regulator does not anticipate a new
marginal abatement cost curves), or induce either an technology and therefore commits ex ante to a level
increased probability or an earlier date of success. A of her policy instrument which is optimal with respect
profit maximizing innovator chooses an effort level to the conventional technology. If, by contrast, the
firms move first by engaging in R&D or by adopting a
new technology and the regulator moves second by
4
It is important to note that we assume that each adopting firm is adapting the level of her policy instrument to the
small and can neither influence the tax rate nor the price for permits. respective R&D outcome or to the rate of adoption of
Otherwise, we could not assume to have the same price for
emissions before and after adoption of the new technology.
new technology, we talk about ex post regulation. If
Amacher and Malik (2002), by contrast, (see below) consider the regulator has no incentive to change her behavior
regulation of a single firm. There, the rationale is a bit different. after firms have moved, her policy is called time
180 T. Requate / Ecological Economics 54 (2005) 175–195

consistent. The adjustment of her policy in case of a 3.1.1. The alleged argument
time inconsistent policy is sometimes called ratchet- The basic argument set out by Milliman and Prince
ing. Note, that ex post regulation is always time (1989) and Jung et al. (1996) is outlined in Fig. 1.
consistent. The early literature usually considers the MAC0 denotes the marginal abatement cost curve of a
regulator as the natural first mover whereas more representative firm running a conventional abatement
recent contributions emphasize the importance to also technology whereas MACI is the marginal abatement
study the regulator’s reaction on innovation and cost curve after adoption of some advanced abatement
technology adoption. technology.
Let us first look at the investment incentives
provided by an emission tax. If a particular firm
3. Results of models on adoption and diffusion switches from conventional to advanced technology,
its savings in variable abatement cost plus tax
In this section we focus on theoretical micro- payments are represented by the area A+B depicted
economic partial equilibrium models which serve to in Fig. 1. If installing a new technology causes a fixed
analyze the incentives for adoption and diffusion of set-up cost FN0, it will be profitable for the firm to
the environmental policy instruments. The models invest in the new technology if and only if FbA+B.
summarized in this section differ, on the one hand, Now consider pollution control by auctioning off
with respect to the behavior of the regulator who is permits. Let r 0 denote the original price for permits
assumed to either act myopically or to engage in ex (see Fig. 2) before the advanced technology was
ante or ex post regulation. On the other hand, they available. Let us assume for a moment that only one
differ with respect to whether or not they pay attention small firm, which is not able to manipulate the permit
to the output market, or whether or not they include price, has access to the new technology. Then the
uncertainty. We begin to discuss the models of myopic incentive for this single firm to install the new
regulation.5 Table 1 summarizes the different features technology is the same as in case where the firm is
of the models described below including the policy regulated by a tax. If, however, a considerable number
rankings as far as those are available. of firms adopt the new technology, the equilibrium
price for permits must go down. Jung et al. assume
3.1. Myopic regulation such a price, say r I , to be exogenously given (see
again Fig. 2). Then indeed, each firm that has adopted
The first contributions dealing with adoption of a the new technology has lower variable costs than in a
new abatement technology are those by Downing and situation where no other firm has adopted the new
White (1986), Milliman and Prince (1989), and technology. The cost difference is equal to the area
Malueg (1989). For different types of pollution A+B+C depicted in Fig. 2. This cost difference,
control instruments the authors compare the aggregate however, is not equivalent to a single firm’s incentive
cost savings incurred by industry when adopting a to adopt the new technology. In other words, the
new technology. By ex ante assuming an industry criterion for whether or not to adopt the new
wide adoption of the new technology, Milliman and technology is not determined by the inequality
Prince arrive at the following (descending) ranking of FbA+B+C. Rather, this incentive is given by the
policy instruments with respect to those cost savings: marginal firm’s cost saving if it adopts the technology
(1) auctioned permits, (2) emission taxes and abate- given that other firms have already decided to adopt,
ment subsidies, (3) free permits and emission stand- i.e. if the price has already fallen to r I . This cost
ards. Jung et al. (1996) come to a similar ranking saving of the marginal firm is displayed by area
employing a more formal analysis. They find, how- A 1bA+B in Fig. 3. Hence the incentive to invest under
ever, that free permits outperform (absolute) emission permits must be lower than under taxes.
standards. Note that if A 1bFbA+B holds, some but not all
the firms will adopt the new technology in equili-
5
This ranking refers to adoption of new technology. Milliman and brium. Assuming that all the firms will adopt the new
Prince also study other scenarios which yield different rankings. technology–as some authors do–is tantamount to
 T. Requate / Ecological Economics 54 (2005) 175–195 181

Table 1
Models of adoption and diffusion
Authors Policy instruments Timing of Special Marginal Policy ranking
game/behavior attention to damage
of regulator output market
Amacher and Malik Emission taxes Ex ante, ex post No Increasing 1. Ex post
(2002) 2. Ex ante
Carraro and Soubeyran Adoption subsidies and Ex ante Yes Constant Ambiguous
(1996) discriminatory emission taxes
Downing and White Emission taxes, abatement Myopic No Increasing No welfare ranking
(1986) subsidies, emission standards,
free permits
Jung et al. (1996) Emission taxes, abatement Myopic No Increasing 1. Auctioned permits
subsidies, emission standards, 2. Taxes, subsidies
auctioned permits, free permits 3. Free permits
4. Standards
Gersbach and Glazer Free permits Ex post Yes Constant, –
(1999) increasing
Kennedy (1999) Free/auctioned permits and Ex ante No Constant Both adjustment
different adjustment rules rules efficient
Kennedy and Laplante Emission taxes, permits Ex ante, ex post No Constant, Taxes and permits
(1999) increasing are efficient if there
are many firms
Laffont and Tirole Permits, permits and futures, Ex ante No Increasing 1. Options
(1996a) permits and price support 2. Permits and futures
policy/options 3. Permits
Milliman and Prince Emission taxes, abatement Myopic No Increasing 1. Auctioned permits
(1989) subsidies, auctioned 2. Taxes, subsidies
permits, free permits, 3. Free permits,
emission standards standards
Malueg (1989) Permits Myopic No No damage –
function
Montero (2002a) Auctioned permits, free Exogenous Yes Emission cap For competitive output
permits, emission standards, emission target markets, and uniform
performance standards permit allocation:
1. Free permits,
auctioned permits
2. Absolute standards
3. Relative standards
Ambiguous otherwise
Petrakis and Xepapadeas Emission taxes Ex ante, ex post Yes Increasing 1. Ex ante commitment
(1999) 2. Ex post taxation
Phaneuf and Requate Permits and banking Ex ante No Increasing Ambiguous
(2002) of permits
Requate and Unold Emission taxes, abatement Myopic, ex ante, No Increasing 1. Free permits,
(2001) subsidies auctioned permits, ex post auctioned permits,
free permits taxes
2. Standards
Requate and Unold Emission taxes, abatement Myopic, ex ante, No Increasing 1. Free permits,
(2003) subsidies, auctioned permits, ex post auctioned permits
free permits, emission 2. Taxes
standards 3. Standards
van Soest (in press) Emission taxes, emission Ex ante Yes No damage Ambiguous
standards function
van Soest and Bulte – – Yes No damage function –
(2001)
182 T. Requate / Ecological Economics 54 (2005) 175–195

number of firms which adopt the new technology

must be determined endogenously. Requate and
Unold (2003) study several scenarios: The first one
corresponds to those studied by both, Milliman and
Prince (1989) and Jung et al. (1996) who both assume
the regulator to be myopic.
We now briefly summarize the incentives provided
by the different policy instruments in case of a naive
regulator. Both the socially optimal outcome and the
firms’ performance depends on the fixed set-up costs:
If this fixed cost is sufficiently high, it will be socially
optimal that no firm invests, and indeed no firm will
invest under either policy. If the fixed cost is zero or
sufficiently low, all firms should invest and they will
Fig. 2. Single firm’s total cost reduction if all firms invest under invest even under moderate environmental policy. For
permits. intermediate values of fixed costs, however, –i.e. for
the interval (F,F) displayed in Fig. 4–the outcomes
resulting from decentralized decision making under
assuming the fixed set-up cost to be sufficiently low, the different policy instruments and the socially
i.e. FbA 1. If one assumes, however, that ex ante all optimal allocation are all different. This result is
the firms will adopt the new technology, it is of little displayed in Fig. 4 where the number of firms which
interest to ask the question which policy instrument adopt the new technology in social optimum and
will provide a higher incentive to adopt the new under the different policy instruments is plotted as a
technology. function of the fixed set-up costs: Curve A denotes the
socially optimal number of firms whereas the curves
3.1.2. Adoption incentives in equilibrium B, C, and D denote the number of firms that invest
Both Kennedy and Laplante (1999) and Requate under permits, taxes, and a standard, respectively.
and Unold (2003) challenge the approach by those The result is formally derived in Requate and
authors by pointing out that equilibrium considera- Unold (2003), but the intuition is simple: In social
tions must be taken into account when studying the optimum, the lower the fixed costs, the more firms
incentives to adopt new technology and, therefore, the should adopt the new technology. For intermediate
values of fixed costs partial adoption is optimal.
Moreover, the lower F, the lower the marginal
damage, and the lower the optimal emission level.

Fig. 4. Number of firms investing in the advanced abatement

Fig. 3. Single firm’s investment incentive under permits. technology.
 T. Requate / Ecological Economics 54 (2005) 175–195 183

The formerly optimal tax, however, is sufficiently We see that rd ê cancels out. Thus Eq. (5) is
high to induce all firms to adopt the new technology equivalent to Eq. (4) and we immediately see that the
even for those cases where partial adoption is optimal. incentive to adopt a new technology is the same for
Under permits, by contrast, the permit price falls when free (grandfathered) and for auctioned permits.
more firms adopt the new technology. Since the Matters are slightly different if an innovator has
regulator does not change the total supply of permits, market power as is assumed by Montero (2002a) and
the permit price falls below the socially optimal level Fisher et al. (2003). In that case there is an endowment
and makes non-adopting firms free ride on those firms effect, and the choice of the regime may affect the
which adopt the new technology. Under an absolute incentive to innovate.
emission standard, either all or no firms adopt the new
technology, depending on how sharp the standard is 3.1.4. Uniform emission standards
set. We see that taxes (or equivalently subsidies on Finally, consider a uniform emission standard ē,
abated emissions) provide the strongest incentives to sometimes referred to as bcommand and control
adopt new technology, whereas, depending on the policyQ. In this case a firm will be indifferent between
parameters either permits or standards provide the staying with the conventional technology and adopt-
lowest incentives. Note that for a considerable range ing the advanced technology if
of parameters both taxes and standards induce more
C0 ðēe Þ  CI ðēe Þ  F ¼ 0 ð6Þ
firms to adopt new technology than permits. Although
environmentalists may prefer taxes and standards for Since C 0V(ē)NC V(ē),
I the LHS of Eq. (6) de-
this reason, from a broader economic point of view we creases in ē. This means that the cost advantage of the
cannot say that taxes and standards are generally new technology decreases as the emission standard is
better than permits. It depends on parameters which of relaxed. This implies, however, that there exists a
those instruments leads to the lowest welfare loss standard ¼e (depending on F) such that no firm will
compared to the first best allocation. adopt the new technology if ēNe¼ , but all firms will
adopt the new technology if ēbe¼ . As a consequence a
3.1.3. Grandfathering uniform standard does not necessarily have the lowest
According to Malueg, Milliman and Prince, and adoption incentives, in contrast to what has often been
Jung et al., free permits provide lower incentives to claimed in the literature (e.g. Milliman and Prince,
adopt new technology than auctioned permits. The 1989; Jung et al., 1996).
argument they propose is that innovation leads to
depreciation of the firms’ permits. However, innovat- 3.2. Anticipating new technology, timing, and
ing depreciates the firms’ permits anyway, irrespective commitment
of whether these permits are auctioned off or
distributed for free. In contrast to what Jung et al. We now turn our attention to scenarios where the
claim, the original permit price is completely irrele- regulator anticipates the evolution of a new technol-
vant with regard to the incentive to adopt a new ogy. We focus on the issue of timing and commitment
technology in equilibrium. To see this, let ẽ be the and the difference between ex post and ex ante
firm’s (identical) initial endowment of permits. As the regulation. Kennedy and Laplante (1999) analyze a
firms are alike before innovation, there will be no similar model as Requate (1995) where symmetric
trade before the new technology is available. Since the firms behave as price takers on the output market and
advanced technology leads to lower marginal abate- choose whether or not to adopt a new abatement
ment costs for each emission level, the investors must technology that lowers their marginal abatement costs
be sellers and the non-investors must be buyers of but incurs fixed investment costs. Considering a
permits. In an equilibrium with partial adoption, firms scenario where the regulator anticipates the new
must be indifferent about adopting the new technol- technology and makes an ex ante commitment to the
ogy or not, i.e. level of her policy instrument, the authors study the
time consistency of emission tax and permit policies.
C0 ðe0 Þ þ r½e0  êe  ¼ CI ðeI Þ  r½êe  eI  þ F: ð5Þ They find that, if the environmental damage function
184 T. Requate / Ecological Economics 54 (2005) 175–195

is linear, ex ante commitment is time consistent for adopting the bdirtierQ technology and emitting the
both instruments, no matter how many polluting firms corresponding socially optimal amount of emissions.
are operating in the market. If, instead, the damage If the regulator commits ex ante to an emission tax
function is convex, time inconsistencies arise if the rate and if the damage function is strictly convex, then
number of firms is relatively small. Ex post regu- it is possible that neither the first nor the second best
lation, by contrast, referred to as bratchetingQ, is time outcome will be achieved: The Pigouvian tax, optimal
consistent by definition. However, firms then tend to with respect to one technology, may prompt the firm
underinvest in the new technology under permits (too to adopt the other technology, and the emission level
little adoption) and to overinvest under taxes (too would not be socially optimal. With ex post regu-
much adoption).6 lation, by contrast, i.e. when the firm moves first, the
For a continuum of firms no inconsistencies arise first or the second best outcome are the only possible
even when the damage function is convex. In a model equilibria. The authors further show that the firm is
with many (a continuum) of asymmetric firms, always better off under ex post regulation, whereas the
Requate and Unold (2001) come to a similar result. regulator, depending on parameters, may either prefer
They even show that both, optimal ex ante and ex ante or ex post regulation.
optimal ex post regulation lead to first-best allocations
for all of the following environmental policy instru- 3.3. Imperfect competition
ments: emissions taxes, subsidies on abatement,
auctioned permits and free permits. Uniform stand- In all the models discussed so far, the authors have
ards, of course, cannot induce the optimal rate of either explicitly or implicitly assumed that the output
adoption due to their static inefficiency. In the markets are perfectly competitive. A couple of models
symmetric version of the model Requate and Unold pay explicit attention to imperfect competition on the
(2003) confirm the optimality of ex ante and ex post output market and its consequences for adoption and
regulation for the case of permits. Under taxes, regulation. Requate (1997) reconfirms the phenom-
however, a first best allocation can only be obtained enon of multiple equilibria for ex ante commitment to
for the case of ex post regulation. For the case of ex a tax policy, in particular if there is free entry.
ante regulation, taxes may induce many equilibria, Petrakis and Xepapadeas (1999) investigate regu-
some of them inefficient in case that partial adoption lation of a single monopolistic polluting firm that can
is socially optimal.7 choose among a menu of new technologies where the
Amacher and Malik (2002) analyze the incentive emission-per-output-coefficient is lower than the one
for adoption for a single firm under an emission tax of the conventional technology, and where the cost is
only. The firm can choose between a bcleanerQ the higher the smaller this coefficient. Assuming
abatement technology that incurs high fixed but low specific functional forms, in particular linear damage,
marginal cost and a bdirtierQ technology which incurs and focusing on emission taxes only, the authors
low fixed but high marginal cost. A first best is compare ex ante commitment to ex post regulation
achieved if the firm adopts the bcleanerQ technology with respect to the optimal level of investment, the
and emits the respective socially optimal emission optimal emission tax, and welfare. Since the monop-
level. The second best outcome is defined as the firm olist can influence the tax rate under ex post
regulation, she will always invest more than in the
case of ex ante commitment. Therefore, both the
6
The fact that the slope of the damage function is crucial may be
emission tax and welfare are always lower under ex
reminiscent of Weitzman’s (1974) seminal paper, as a referee post regulation compared to the case of ex ante
suggested. In fact, there is little in common: Weitzman found that commitment.
the ratio of the slopes of the marginal abatement cost function and Montero (2002a) is mainly interested in the
of the marginal damage functions may be crucial for instrument investment incentives of tradable permits and two
choice. Here the question of whether or not the slope of the marginal
damage function is positive is crucial for whether or not there is a
kinds of standards, emission and performance stand-
problem of time consistency. ards, but rules out the possibility of taxation. Besides
7
Feess and Gleaves (2000) independently found similar results. allowing for perfect and imperfect competition on the
 T. Requate / Ecological Economics 54 (2005) 175–195 185

output market, he is one of the very few researchers Requate and Unold (2003) who implicitly assume
who also models imperfect competition on the permit perfect competition on the output market.
market. In his model ex ante symmetric firms produce In a recent comment on Montero (2002a), Bruneau
a homogenous good and emit a pollutant. The firms (in press) holds the view that performance standards
can choose among different abatement technologies generate a greater incentive to innovate than permits
that are associated with lower marginal abatement even under perfect competition. Moreover, he extends
costs than the conventional one. In contrast to the Montero’s analysis to the case of increasing marginal
other approaches, however, there is no damage costs and argues that in this case both auctioned and
function, and thus optimal pollution levels are not free permits dominate performance standards. Bru-
considered. Rather, the regulator aims at enforcing an neau, however, neither carries out a complete equili-
exogenously given aggregate emission standard. The brium analysis nor a welfare comparison, and thus
R&D incentives are compared by means of the does not rank the instruments with respect to total
respective marginal profits. The total effect of a social costs or welfare. Since performance standards
typical firm’s investment decision consists of a direct lead to inefficient abatement levels in the static case, it
cost effect and, in case of imperfect competition, of cannot be ruled out that they lead to inefficiently high
strategic effects on both the output and the permit investment.
market. The direct effect, always prompting the firms
to invest, is the same for permits and for an absolute 3.4. Including uncertainty
emission standard. Under performance (or relative)
standards it is lower. The strategic effect with respect A couple of contributions take into account differ-
to the output market causes the firms to invest more ent kinds of ex ante uncertainty in dynamic models
under standards, but less under permits. In the former with two or three periods where the uncertainty will
case the investment lowers whereas in the latter case it usually be resolved in the second period: Laffont and
raises the competitors’ output. The strategic effect on Tirole (1996a) assume uncertainty about the benefits
the permits market is in particular relevant under accrued by the firms from emitting pollutants.
auctioned permits: since all firms are permit buyers, Kennedy (1999) considers uncertainty about environ-
they benefit from a lower permit price due to higher mental damage, whereas in Phaneuf and Requate
investment. Hence, for imperfect competition on both (2002) the abatement cost is subject to uncertainty.
markets Montero finds the following (partial) ranking: van Soest and Bulte (2001) and van Soest (in press)
Both, an emission standard and a regime of auctioned take into account the uncertainty about technological
permits provide a higher incentive to invest than free progress. The approach by Bulte and van Soest is an
permits whereas a performance standard may provide important step in the direction to model ever ongoing
a higher, a lower, or the same investment incentive R&D effort, for, R&D is usually not once and for all,
than both free permits and the emission standard. and R&D failure may be just temporary.
Finally, both types of standards may provide a higher, Laffont and Tirole (1996a) analyze optimal regu-
a lower, or the same incentive to invest than auctioned lation of many (a continuum of) (potentially) polluting
permits. If there is perfect competition on both asymmetric firms in a two period model where in both
markets, only the direct effect matters. Thus an periods, firms can either emit one unit or nothing.
emission standard, free and auctioned permits provide Firms are regulated by emission permits. In the first
the same incentives whereas the incentive provided by period full abatement requires the firms to cease
the performance standard is lower. Interestingly, the production whereas in the second period they can use
results are qualitatively the same as in the case where an abatement technology which they had to purchase
there is imperfect competition only on the permit in the first one. In the first period firms know their
market. For imperfect competition on the output benefit from pollution in period 1 but not that of
market only, Montero finds that the incentives for period 2. Only the probability distribution of those
tradable and auctioned permits are equivalent whereas benefits is known. This uncertainty is resolved in the
an emission standard may provide a higher incentive second period. A novelty of this model is that the
than permits. These findings are in line with those of shadow cost of public funds is taken into account.
186 T. Requate / Ecological Economics 54 (2005) 175–195

Thus the regulator chooses the optimal Ramsey permit argues that the first rule is unlikely to be implemented
price (respectively the corresponding amount of since firms will be rewarded if the expected damage
permits) in period 1. The authors show that if the turns out to be higher than expected. I think, however,
permit market is a pure spot market, i.e. if the that expropriating permits from firms is as least as
regulator issues permits in each period and in the difficult since it violates the firms’ property rights.
first period is not able to make a commitment about Phaneuf and Requate (2002) examine the effects
the second-period permit price, then excessive invest- of banking permits on the incentives to invest in
ment incentives are created. A too high permit price advanced abatement technology if there is aggregate
prompts the firms to bypass the permit market via uncertainty about the abatement cost. To this end
adoption of the abatement technology. By committing they set up a three period model where in the pre-
to a lower second-period permit price, especially by regulation period 0, firms can invest in an abatement
introducing a futures market, the regulator can technology. In the periods 1 and 2, respectively, the
enhance welfare by discouraging unwanted invest- uncertainty about the abatement cost is resolved.
ment. However, committing to a lower second-period Banking allows firms to postpone investment until
price in the first period is not time consistent. For, a more information on the abatement costs is revealed.
marginal increase in the number of permits in the The authors find that, if the discount factor is
second period increases welfare. A price support sufficiently small, and if period 1 costs are revealed
policy or options to pollute, issued in period 1, allow to be low, there will be positive banking but no first
the regulator to solve this problem of time incon- period investment. If instead period 1 costs are
sistency. It is important to note, however, that the revealed to be high, there will be no banking but
over-investment result hinges on the assumption that positive first period investment. The analysis of the
the regulator wants to collect money form auctioning constrained socially optimal response of the firms to
off permits in order to mitigate the social costs of the resolution of uncertainty (i.e. total endowment of
public funds. In other words, the regulator wants to permits is fixed) shows that the regulator would like
slow down diffusion of advanced technology because some banking, but that private and social optimal
this erodes his tax base! responses to the resolution of uncertainty are not
Kennedy (1999) sets up a two period model where identical. This can lead to sub-optimal levels of
in the first period the (constant) marginal damage may investment in improved technology. A unique con-
be high or low. Given this uncertainty, a large number clusion concerning the savings of social cost through
of polluting firms have to decide whether or not to banking is not possible, though. For a quadratic cost
adopt an improved abatement technology which and damage function the authors show that banking
incurs fixed costs and lower marginal abatement leads to lower costs for society if the damage
costs. Kennedy assumes a very special scheme of function is relatively flat. Otherwise non-banking is
permit trading, rarely used in real existing policy preferred.
frameworks: one permit allows the emission of one van Soest and Bulte (2001) study the problem of
unit of pollution in each period of time (rather than technology adoption for the case that future techno-
just for one period). When uncertainty about the logical advances are uncertain. For this purpose they
marginal damage is resolved in the second period, the apply the option value approach developed by Dixit
regulator can adjust the number of permits. Kennedy and Pindyck (1994), which has been applied to the
considers two kinds of adjustment rules which both problem of technology adoption by Farzin et al.
implement the social optimum if announced in period (1998) and improved by Doraszelski (2001). Van
1; firstly, open market operations, i.e. buying back Soest and Bulte show that even if adoption of new
permits if the marginal damage turns out to be high technology pays according to the criterion of net
and auctioning off more permits if the damage is low, present value comparisons, it may not be profitable if
and secondly, a proportional adjustment rule, where further improvements are likely to occur. Hence the
firms lose a fraction of their permit endowment if the firm is better off by postponing the adoption decision
damage is high, or get more permits proportional to to the point of time when an even better technology
their initial endowment if damage is low. Kennedy is available. Using this calculus they offer an
 T. Requate / Ecological Economics 54 (2005) 175–195 187

explanation why firms do not invest, although it regulation scheme is characterized by a decreasing
seems favorable to do so. In a companion paper van probability of setting a stricter standard as firms reach
Soest (in press) uses this approach in order to a certain stage of technology development.
compare policy instruments with respect to the
incentive to adopt improved energy-efficient tech-
nologies. Surprisingly, he finds that firms tend to 4. Innovation, R&D, and adoption
postpone the adoption of improved technology if
environmental policy is more stringent. Moreover, So far we have considered scenarios where the
comparing energy taxes to absolute energy use new technology was given and firms had only to
standards, he finds that there is no unambiguous decide whether or not to adopt it. In this section we
ranking of the two instruments when it comes to will study innovation and R&D before firms adopt
stimulate early adoption of new technology. the new technology. The question about how to
model innovation depends on whether innovation is
3.5. Regulation and hold up a private or a public good. If innovation is mainly
firm specific and thus a private good, it can be
Gersbach and Glazer (1999) identify incentives for modelled in a stylized way by letting production and
firms to hold up innovation. In their model the cost functions depend on some investment level k.
regulator aims at prompting all (ex ante symmetric) To adapt this to our previous model we would write
firms of an oligopolistic, polluting industry to adopt C(e,k) with C k b0 and C ek b0, i.e. both abatement
a certain abatement technology at a fixed cost and and marginal abatement costs are decreasing with
to abate the corresponding optimal amount of more innovative investment. One can show easily
emissions. If a firm does not invest, it can only that the results from the last section carry over to this
abate emissions by reducing its output. A crucial model.
but most unusual assumption is that the social Matters are different if from the social perspective
benefit of the last unit of output exceeds its social innovation is a public good, and once it has been
cost of pollution. Hence, the regulator will never invented, other firms could in principle use it without
force a non-investing firm to abate emissions. incurring considerable costs.8 In these kinds of models
Anticipating this, firms have an incentive to not authors either assume that ex ante there is only one
invest into the new technology. Gersbach and Glazer innovator (e.g. Denicolo, 1999; Fisher et al., 2003, or
show that this hold-up problem can be overcome by Requate, in press), or that several — usually identical
ex post issuing free permits, assuming a competitive R&D firms — engage in a patent race, but ex post one
equilibrium on the permit market. In equilibrium all innovator, the winner of the race, prevails in the
firms invest if the number of firms in the market is market. Then this innovator can either sell or license
as least two. the technology to other firms which decide whether or
Although entering the realm of R&D, which we not to adopt it. Since in that case the innovator usually
will deal with more properly in the next section, Cadot can exercise some market power, commitment and
and Sinclair-Desgagne (1996) tackle a similar prob- timing of the regulation game are crucial concerning
lem. They look at a situation where the regulator both the incentives to engage in R&D and the
wants to implement a stricter emission standard. incentives to adopt new technology.
Without regulation firms have no incentive to adopt The various models summarized in this subsection
new technology or to engage in R&D whereas differ, firstly, with respect to which policy instruments
immediate regulation constitutes too high a burden are subject to investigation, secondly, which timing
for the firms and is thus not credible. Thus, the firms and commitment strategies are feasible for the
can again hold up the regulator by doing nothing. To regulator and which strategy she is assumed to pursue,
solve this problem, the authors construct a dynamic
incentive scheme that can be implemented as a
Markov perfect equilibrium. In equilibrium both the 8
Innovators can, of course, exclude other firms from using it, and
regulator and the firms can use mixed strategies. The they can charge license fees to rent it out.
188 T. Requate / Ecological Economics 54 (2005) 175–195

thirdly, whether R&D success is stochastic or deter- 4.1. Innovation incentives and welfare gains under
ministic, forth, whether the marginal damage is perfect competition on the final market
constant or increasing, or finally whether or not
special attention is paid to the output market. We Biglaiser and Horowitz (1995) and Parry (1995,
summarize the different contributions in the next 1998) were the first to rigorously combine the issues
sections. Table 2 gives an overview over the different of innovation and adoption. Biglaiser and Horowitz
features of the models. (1995) consider a competitive polluting industry with

Table 2
Models of innovation and diffusion
Authors Policy instruments Timing of R&D Marginal Special Policy ranking
game/behavior stochastic damage attention
of regulator to output
market
Biglaiser and Emission taxes Interim, ex ante Yes Constant No –
Horowitz (1995) plus technological
standard
Denicolo (1999) Emission taxes, Ex post, ex ante No Increasing Yes For damage not
auctioned/free permits too high:
1. Taxes
2. Permits
For high damage:
1. Permits
2. Taxes
Fisher et al. (2003) Emission taxes, Myopic, ex post No Increasing No Ambiguous
auctioned permits,
free permits
Katsoulacos and Emission taxes cum Ex ante No Increasing Yes –
Xepapadeas (1996) R&D subsidies
Innes and Bial (2002) Emission taxes and Ex ante Yes Increasing Yes 1. Taxes cum standards
emission standards 2. Taxes
Laffont and Tirole Permits, advanced Interim, ex ante Yes Increasing No 1. Options
(1996b) allowances (future 2. Advanced allowances
market for permits), 3. Spot market permits
options, incentive
contract, permits
and securities and
licensing tax
Montero (2002b) Emission taxes, Exogenous No – Yes Ambiguous
emission standard, emission target
auctioned permits,
free permits
Parry (1995) Emission taxes Ex ante Yes Constant, No –
increasing
Parry (1998) Emission taxes, free Myopic, interim Yes Constant Yes Ambiguous
permits,
performance standards
and output quota
Parry et al. (2003) Not specified Ex post No Constant, No –
increasing
Requate (in press) Uniform emission Ex ante, interim, Yes Increasing No 1. Menu of tax rates
taxes, menu of taxes ex post 2. Uniform taxes
auctioned permits 3. Permits
 T. Requate / Ecological Economics 54 (2005) 175–195 189

an exogenously given number of ex ante symmetric marginal damage. This ambiguity may be caused by
firms each of which can engage in R&D. The new the assumption of free entry which induces a further
technology is randomly drawn from a cumulative market imperfection. Requate (1997) also found in a
distribution which is identical for all firms. Thus, model of emission taxes in a Cournot model with
innovation size and R&D success are stochastic. A pollution and free entry, and even without R&D, that
patent is granted to a successful firm, and other firms the second best tax rate may exceed or fall short of
can use this technology when paying a license fee and marginal damage.
installation costs. Since adoption costs are independ- In a variant of this model Parry (1998) allows for
ent of which particular technology is adopted, the incomplete diffusion, and besides taxes, he also
regulator always wants an adopting firm to adopt the studies tradable permits and an instrument mix
best available technology. If the regulator makes an consisting of a performance standard and a production
interim commitment (i.e. after R&D but before quota. For permits and performance standards he
adoption) to both an emission tax equal to marginal distinguishes the cases of ex ante and interim commit-
damage and to a technological standard which either ment. Employing numerical simulations he finds that
specifies the firms which have to adopt the best emission taxes yield higher welfare than permits, with
available technology or which specifies the firms the difference depending crucially on the potential
which have to adopt the blowest acceptableQ technol- size of innovation. The same holds for performance
ogy, then the efficient levels of pollution, production, standards under ex ante commitment. For interim
and adoption will be induced. The level of R&D commitment, however, the difference in efficiency
effort, however, is too low compared to the socially almost vanishes. For the case of emission taxes he
optimal level. shows that imitation does not necessarily imply large
Parry (1995) sharply separates the polluting from inefficiency in the R&D market. Therefore, imitation
the R&D sector. In contrast to most of the other does not call for bresearch subsidies or tightening
contributions there is free entry on both markets. Each environmental regulation beyond the Pigouvian levelQ
upstream firm conducts one R&D project to develop a (p. 252).
new abatement technology for the polluting down- In contrast to the previous authors, Fisher et al.
stream sector but does not need it for its own (2003) abstract from patent races and consider a
production. Both the probability of R&D success model with a large number of competitive polluting
and the industry’s R&D cost rise with the number of firms, one of which, called the innovator, is able to
R&D firms. If a firm is successful, it is granted a engage in R&D in order to improve its own abatement
patent and becomes a monopolist. The ex ante technology in the first place. The other symmetric
symmetric downstream firms can adopt the new polluting firms can either pay a license fee to adopt
technology by paying a license fee. Parry shows that the new technology, or they can freely use an
a rising tax rate leads to a smaller number of polluting imperfect imitation. Despite this possibility complete
firms in the downstream market. Since those with the diffusion is socially optimal. Considering emission
highest willingness to pay for the new technology stay taxes, auctioned and free permits as policy instru-
in the market, a higher tax also induces a higher ments the regulator is assumed to be myopic by ex
license fee. Parry studies only the tax instrument with ante committing to the Pigouvian levels with respect
ex ante commitment to the tax rate before the to the conventional technology. For the case of
upstream firms engage in R&D, and he derives the constant marginal damage the authors find that if no
second best optimal tax rate for the cases of linear and imitation is possible, emission taxes induce the first-
convex damage and with and without the possibility best outcome. If imitation is possible, taxes dominate
of costless imitation. If imitation is not possible, the free permits with respect to welfare. Depending on
second best optimal tax rate turns out to be smaller how imperfectly the innovative technology can be
than marginal damage. This is a typical finding for imitated, welfare under taxes might exceed or fall
second best optimal taxation under imperfect com- short of welfare under auctioned permits. Emission
petition. If imitation is possible, however, the second taxes are superior if imitation is easy. For the case of
best optimal tax rate may be smaller or greater than increasing marginal damage the authors find that the
190 T. Requate / Ecological Economics 54 (2005) 175–195

steeper the marginal damage curve the more do the level of his policy instrument, or that the regulator
permits dominate the tax regime. It is worth to note moves after observing R&D success and/or the degree
that in equilibrium imitation never occurs. It only of adoption, a setting we referred to as ex post
serves as an outside option for the polluting firms and regulation. It has been shown by Requate and Unold
thus drives down the license fee. Furthermore, in (2001, 2003) that under pure adoption and compet-
contrast to Requate and Unold (2001, 2003), Fisher et itive conditions ex ante and ex post regulation are
al. (2003) find that free and auctioned permits do not (almost) equivalent. If, by contrast, there is only one
perform equivalently in this model since the innovator firm to be regulated, this equivalence breaks down, as
is able to exercise market power, and his price strategy Amacher and Malik (2002) have shown. If we study
depends on his own initial endowment of permits. innovation, there are typically few firms which engage
This endowment effect is similar to the one pointed in R&D and even fewer that will be successful. Hence
out in Hahn’s (1984) model who considers permit we would expect the timing to be crucial for the
markets with one firm exercising market power. The incentives to innovate. This is the focus of the papers
endowment effect found by Fisher et al. is likely to be by Denicolo (1999) and Requate (in press) who
small, however, if the share of permits owned by the compare the different timings and commitment
R&D firm is small. strategies with respect to welfare. Following Parry
Laffont and Tirole (1996b) study the innovation (1995), Denicolo (1999) and Requate (in press)
incentives of permits in a regime where, just as in their sharply distinguish between a sector that develops
companion paper (Laffont and Tirole, 1996a), the and a sector that uses the new technology. This
regulator takes into account the shadow cost of public separation is well supported by empirical evidence.
funds and thus is interested in reducing the burden on Lanjouw and Mody (1996) found that 81% of all
tax payers by taking advantage of the permit revenues. innovations in air emissions clean up was developed
A single upstream firm engages in R&D and might by machinery industry but only 5% have been used by
invent a pollution-free technology. The authors show the same sector. For water cleaning technology we
that if the regulator commits ex interim (i.e. after obtain 83% and 2%, respectively, and for energy
R&D but before the innovator’s pricing decision) to a production from renewable resources the numbers are
permit price, the innovator does not engage in R&D 85% and 8%, respectively.
because he will make no profit. Since permits and Denicolo (1999) was the first to explicitly compare
innovation are perfect substitutes, the innovator will ex ante and ex post regulation for both emission taxes
always undercut the permit price which drives down and tradable permits in a model with an upstream
the price to zero. Laffont and Tirole further find that if monopolistic R&D firm and many polluting down-
the regulator prior to R&D sells a certain amount of stream firms. The perfectly competitive firms produce
permits and commits himself not to issue additional an output with constant returns to scale. Emissions are
permits on the spot market after R&D success, then proportional to output. The new technology has a
the innovation incentive is always smaller than lower emissions-per-output ratio than the conven-
optimal, and the induced level of adoption may be tional one, and the degree of reduction depends on
sub-optimal as well. Moreover, emissions are too R&D investment. Denicolò finds that taxes and
high. However, the regulator can restore the first best permits are equivalent for ex post regulation. For ex
outcome by offering an optimal ex ante incentive ante commitment, by contrast, the instruments are not
contract to the innovator, committing to purchase the equivalent and both always lead to underinvestment in
invention at a certain price and to sell the licenses at R&D. If the regulator commits to the second-best
the optimal Ramsey price. optimal level of the instruments, it depends on the
social cost of pollution whether taxes perform better
4.2. Timing and commitment or worse than permits.
Requate (in press), similar to Parry (1995), studies
As has become clear from the above summaries, the relationship between adoption of new technology
most authors either assume that the regulator moves and R&D incentives. For this purpose he studies a
first and is able to ex ante commit to both the type and monopolistic upstream innovator which invests R&D
 T. Requate / Ecological Economics 54 (2005) 175–195 191

effort in order to find a new, exogenously given aggregate emission target by either charging a tax or
abatement technology. The R&D effort determines the by issuing the corresponding number of permits.
probability of success. If research was successful, the The main result of that analysis is that commitment
innovator produces the new technology at constant to a menu of tax rates (timing B) dominates all other
marginal cost. Then a large number of asymmetric policy regimes. Moreover, the tax regime outperforms
polluting downstream firms decide whether or not to the permit regime for both interim regulation (timing
adopt the new technology. The incentive for both, to C) and commitment to a menu of policy levels
adopt new technology and to engage in R&D, is contingent on R&D success (timing B). The reason
analyzed for emission taxes and auctioned permits. is that under a tax regime and commitment before the
The industry process can be divided into four innovator engages in pricing of his new technology,
stages: (i) R&D activity, (ii) pricing of new technol- the innovator is not able to influence the tax rate
ogy, (iii) adoption of new technology, and (iv) whereas he can influence the permit price by pricing
decision of downstream firms on short term abatement higher or lower, or equivalently, by holding down or
(see Fig. 5). This structure gives rise to four different raising output. To be more precise, by pricing higher,
timings of environmental policy: (A) ex ante commit- other things equal, demand for the technology falls.
ment before R&D, using a uniform tax (or permit) But at the same time demand for permits by those
policy; (B) ex ante commitment before R&D, using a firms which still hold the old technology rises. Higher
menu of different tax rates (or permit policies, demand for permits leads to higher permits prices and
respectively) contingent on R&D success; (C) interim thus raises the willingness to pay for the advanced
commitment after observing R&D but before adoption abatement technology. The bottom line is that the
of new technology, and finally (D) ex post regulation innovators’ effective inverse demand function for the
after observing both R&D and the rate of adoption. In new technology is more elastic under permits than
all cases Requate discusses both regulation by prices under taxes, and thus the distortion through monopoly
(taxation) and regulation by quantities (issuing trad- power on the market for new technology is more
able permits). It is easy to see that the two policies severe under the permit regime compared to the tax
must be equivalent in scenario (D). This is so because regime.
when the regulator is the last to move, he knows both Moreover, Requate (in press) finds that under
the marginal damage and the aggregate marginal interim commitment the second best tax rate exceeds
abatement costs. Hence he is able to implement each marginal damage under some mild assumptions on the

Regulator Regulator Regulator

Timing Timing C Timing D
A B
U-firm D-firms D-firms
U-firm output/ adoption abatement
R&D effort price
s
ces

B
suc

1
Regulator
D

A
R&

B
0
Timing C,D
D-firms
R&

abatement
D
fai
lur
e

Fig. 5. The structure of the different regulation games.

192 T. Requate / Ecological Economics 54 (2005) 175–195

demand for new technology. The intuition here is that effort. The regulator ex ante commits to both an
the innovator produces less units of the new technol- emission tax and an R&D subsidy (which can also
ogy than socially optimal. Hence the regulator can take negative values). The authors find that the
enhance the demand for new technology by raising optimal tax rate falls short of marginal damage while
the tax rate. Under timing A and B, by contrast, the tax the optimal subsidy is positive if the spillover effects
rate may exceed or fall short of marginal damage. are sufficiently high, and negative otherwise. The
Here the reason is that under these timings the intuition for this result is as follows: on the one hand,
regulator tries also to influence the R&D effort. Note firms tend to underinvest in R&D because the private
that timing A is similar to Parry’s set-up, who obtains return from R&D is smaller than the social return and
a tax rate lower than marginal damage, but incurs a because firms do not account for consumers’ surplus.
further market imperfection through sub-optimal On the other hand, firms strategically tend to over-
market entry. invest in R&D in order to increase market shares.
Requate further argues that a first best allocation With a similar set-up Petrakis and Xepapadeas
can be restored if the regulator has control over three (2003) study a model of innovation with market
policy instruments: an emission tax in order to reduce power on the output market. Surprisingly they find
pollution, an output subsidy in order to give incen- that cases exist where time consistent policies lead to
tives to the monopolist to increase output, and finally higher welfare than commitment strategies.
a profit tax or subsidy, in order to equalize the Innes and Bial (2002), by contrast, study innova-
monopolist’s private value of innovation to the social tion incentives for a Bertrand (i.e. price setting)
value of innovation. Nevertheless, even with three duopoly with homogenous products. By investing in
instruments, the timing of the regulator does matter: R&D the firms can find a certain incremental
Since under commitment to a tax policy the demand innovation that would lower both their marginal cost
function for new technology faced by the innovator is of production and their marginal abatement cost.
less elastic, the output subsidy required to obtain the Since R&D success is stochastic, no, one or both
first best allocation will be smaller compared to a firm(s) may be successful. If firms are ex post
permit policy. Thus, if the regulator faces social costs symmetric, the regulator is able to implement efficient
to raise public funds, he will prefer a timing which is pricing and efficient levels of both production and
less costly in terms of total subsidies to be paid to abatement by levying the corresponding Pigouvian
industry in order to obtain a first best, or at least less tax. If, however, firms are ex post asymmetric, the
distorted allocation. regulator is not able to implement the first best
allocation by levying a tax only. The emission tax that
4.3. Imperfect competition on the final goods market is optimal with respect to the new technology would
enable the winner of the R&D race to serve the entire
Whereas the contributions studied so far have market and to produce less than efficient. The
abstracted from the output market, some authors regulator can solve this problem by combining an
investigate incentives to innovate with the special emission tax lower than marginal damage with a non-
focus on imperfect competition on the output market. uniform relative standard which requires the winner to
Katsoulacos and Xepapadeas (1996) assume Cournot comply with the first-best standard and the loser with
competition whereas Innes and Bial (2002) assume a laxer standard. The authors further look at a setting
Bertrand competition with homogenous goods. Mon- where the regulator is not able to observe R&D
tero (2002b) considers both the case of price outcomes without any cost. Surprisingly however, the
competition with differentiated commodities and regulator can nevertheless induce the firms to report
Cournot competition. their technologies truthfully and thus can implement
Katsoulacos and Xepapadeas (1996) study a the first-best allocation by ex ante committing to a
Cournot (i.e. quantity setting) duopoly with emissions similar policy, i.e. by committing to levy an emission
proportional to output. Both firms are able to reduce tax, to set a non-uniform relative standard contingent
their emissions-per-output ratio by investing in R&D on the firms’ technology reports and to monitor the
and enjoy spillovers through the other firm’s R&D firms with a certain probability.
 T. Requate / Ecological Economics 54 (2005) 175–195 193

Building on his companion paper (Montero, cost savings rather than by investigating the firms’
2002a), Montero (2002b) allows for R&D spillovers incentives to invest in equilibrium. I have argued that
and assumes imperfect competition on both the the comparison of environmental policy instruments
output and the permit market throughout his analysis. leads to quite different results if the number of firms
He studies Cournot as well as price competition with which adopt new technology is determined endoge-
differentiated products. Besides emission standards nously through equilibrium considerations.
and permits he also analyses emission taxes. With all the special results it seems to be difficult
Montero finds that if the marginal cost of both firms to draw clear conclusions on which policy instru-
is constant, there is no strategic effect when levying ments dominate other policy instruments. I think,
an emission tax. Taxes may provide more, less, or however, one can draw the main conclusion that
the same incentive to invest in innovation than instruments which provide incentives through the
emission standards and auctioned permits whereas price mechanism, by and large, perform better than
free permits offer less incentive than taxes. With command and control policies. Even though, it is
price competition on the output market, by contrast, important that the regulator either anticipates the new
taxes provide a higher incentive than an emission technologies to a certain extent or that he reacts in an
standard which in turn provides a higher incentive optimal way on invention and adoption of new
than free permits. Auctioned permits again can offer technology. Under competitive conditions and perfect
more, less or the same incentive than taxes. Montero foresight, different authors established the result that
concludes, that in the Cournot case, either emission ex ante commitment and ex post optimal policies
standards, taxes, or auctioned permits can provide generate equivalent or at least similar allocations.
the highest incentive, whereas in the case of price Under imperfect market conditions, the policy con-
competition this holds either for taxes or for clusions are less clear cut.
auctioned permits. Under myopic environmental policies or long term
I would like to close this section by emphasizing a commitment to the levels of policy instruments, by
recent contribution by Parry et al. (2003) who open up contrast, emission taxes tend to provide a stronger
a somewhat anti-innovative perspective by showing incentive to invest in both R&D and adoption of new
that within a dynamic framework including the aspect technology as compared to emission allowances. The
of time explicitly, in many cases the net present value reason is that the permit price falls if new technology
of innovation is small. They argue that environmental diffuses, providing a lower incentive for firms with
improvements by even optimally employing new old technology to invest in pollution reducing
technologies may be small compared to optimally technology.
exploiting existing abatement opportunities. After all One shortcoming of the research is that one of the
they conclude, that bthese findings appear to contra- most commonly used instruments, namely the relative
dict earlier assertions by some economists that (or generation performance) standard is usually not
technological advance might be more important than studied. One reason is that many researchers look at
achieving optimal pollution control in the design of the polluting sector only and do not pay attention to
environmental policiesQ (p. 252). the output market. To model the relative standard, it is
necessary to also take into account the firms’
decisions on output. According to my knowledge,
5. Concluding remarks Montero (2002a) was the only one who studied this
kind of standard. He finds, however, that this policy
In this paper I surveyed recent developments on the instrument does not perform very well compared to
incentives provided by environmental policy instru- market based instruments such as tradable permits.
ments to adopt advanced abatement technology and to Note that most of the contributions on adoption and
engage into R&D to develop such new technology. I R&D incentives follow the tradition of the industrial
started with some critical remarks on the ranking of organization literature since they are concerned with
environmental policy instruments in the traditional dynamic games of a finite number of stages. In reality,
literature, a ranking derived by comparing aggregate however, we observe a permanent process of research
194 T. Requate / Ecological Economics 54 (2005) 175–195

firms engaging in R&D and polluting firms adopting and Market Structure. Kluwer Academic Publishers, Dordrecht,
new technology. Furthermore, a regulator will not be pp. 131 – 141.
Carraro, C., Soubeyran, A., 1996. Environmental policy and the
able to commit to the level of his policy instruments for choice of production technology. In: Carraro, C., Katsoulacos, Y.,
once and for all. Hence an important path for further Xepa-Padeas, A. (Eds.), Environmental policy and market
research is to account for such processes in a more structure. Kluwer Academic Publishers, Dordrecht, pp. 151 – 180.
complex dynamic framework in order to improve Denicolo, V., 1999. Pollution-reducing innovations under taxes or
permits. Oxford Economic Papers 51 (1), 184 – 199.
dynamic environmental policy design. As mentioned
Dixit, A., Pindyck, R., 1994. Investment Under Uncertainty.
above, the approach by van Soest and Bulte (2001) who Princeton University Press.
employ the option value approach, developed by Dixit Doraszelski, U., 2001. The net present value method versus the
and Pindyck (1994), is one of the most promising steps option value of waiting: a note on Farzin, Huisman, and Kort
into that direction. (1998). Journal of Economic Dynamics and Control 25 (8),
At the beginning of this survey we quoted Kneese 1109 – 1115.
Downing, P.B., White, L.J., 1986. Innovation in pollution control.
and Schulze who emphasized that the long term Journal of Environmental Economics and Management 13 (1),
incentives provided by environmental policy instru- 18 – 29.
ments to spur the development of new technology are Farzin, Y.H., Huisman, K., Kort, P., 1998. Optimal timing of
possibly more important than the criteria of static technology adoption. Journal of Economic Dynamics and
efficiency. Contrasting from this view, Parry et al. Control 22 (5), 779 – 799.
Feess, E., Gleaves, S., 2000. Environmental Policy and Incentives
(2003) stress that the welfare gain from innovation is to Adopt Advanced Abatement Technology—ex post Efficient
sometimes not much greater than the welfare gain of Adjustment Can Lead to ex ante Afficient Investment Decisions.
efficiently abating pollutants by means of conven- mimeo, University of Aachen.
tional technologies. I think this is an important point. Fisher, C., Parry, I., Pizer, W., 2003. Instrument choice for
Resources to engage in R&D are scarce. Hence environmental protection when technological innovation is
endogenous. Journal of Environmental Economics and Manage-
environmental technological progress may crowd out ment 45, 523 – 545.
other strands of welfare enhancing technological Gersbach, H., Glazer, A., 1999. Markets and regulatory hold-up
progress. It is often the same people (and political problems. Journal of Environmental Economics and Manage-
parties) who urge for the adoption of particular ment 37 (2), 151 – 164.
technologies such as wind and solar power, and at Gersbach, H., Requate, T., 2004. Emission taxes and optimal
refunding scheme. Journal of Public Economics 88, 713 – 725.
the same time vote for rather inefficient instruments of Hahn, R., 1984. Market power and transferable property rights.
regulation. Incentives for the right rate and the right Quarterly Journal of Economics 99, 753 – 765.
direction of technological progress are important. But, Innes, R., Bial, J.J., 2002. Inducing innovation in the environmental
the potential of traditional technology should also be technology of oligopolistic firms. Journal of Industrial Eco-
nomics 50 (3), 265 – 287.
efficiently exploited. Hence both the long and the
Jaffe, A.B., Newell, R.G., Stavins, R.N., 2002. Environment policy
short term incentives of environmental policy instru- and technological change. Environmental and Resource Eco-
ments must be considered. nomics 22 (1–2), 41 – 69.
Jung, Ch., Krutilla, K., Boyd, R., 1996. Incentives for advanced
pollution abatement technology at the industry level: an
evaluation of policy alternatives. Journal of Environmental
References Economics and Management 30, 95 – 111.
Katsoulacos, Y., Xepapadeas, A., 1996. Environmental innovation,
Amacher, G.S., Malik, A.S., 2002. Pollution taxes when firms choose spillovers and optimal policy rules. In: Carraro, Katsoulacos
technologies. Southern Economic Journal 68 (4), 891 – 906. C.Y., Xepapadeas, A. (Eds.), Environmental Policy and Market
Biglaiser, G., Horowitz, J.K., 1995. Pollution regulation and Structure. Kluwer Academic Publishers, pp. 143 – 150.
incentives for pollution-control research. Journal of Economics Kennedy, P.W., 1999. Learning about environment damage:
and Management Strategy 3 (4), 663 – 684. implications for emissions trading. Canadian Journal of Eco-
Bruneau, J.F., 2004. A note on permits, standards, and technological nomics 32 (5), 1313 – 1327.
innovation. Journal of Environmental Economics and Manage- Kennedy, P., Laplante, B., 1999. Environmental policy and time
ment 46, 1192 – 1199. consistency: emission taxes and emissions trading. In:
Cadot, O., Sinclair-Desgagne, B., 1996. Innovation under the Petrakis, E., Sartzetakis, E.S., Xepapadeas, A. (Eds.), Environ-
threat of stricter environmental standards. In: Carraro, C., mental Regulation and Market Power. Edward Elgar, Chelten-
Katsoulacos, Y., Xepa-Padeas, A. (Eds.), Environmental Policy ham, U.K.
 T. Requate / Ecological Economics 54 (2005) 175–195 195

Kneese, A.V., Schulze, C.L., 1975. Pollution, Prices and Public Petrakis, E., Xepapadeas, A., 2003. To Commit or not to Commit:
Policy. Brookings, Washington, D.C. Environmental Policy in Imperfect Markets. University of Crete,
Laffont, J.-J., Tirole, J., 1996a. Pollution permits and compliance mimeo.
strategies. Journal of Public Economics 62 (1–2), 85 – 125. Phaneuf, D.J., Requate, T., 2002. Incentives for investment in
Laffont, J.-J., Tirole, J., 1996b. Pollution permits and environmental advanced pollution abatement technology in emission permit
innovation. Journal of Public Economics 62 (1–2), 127 – 140. markets with banking. Environmental and Resource Economics
Lanjouw, J.O., Mody, A., 1996. Innovation and the international 22 (3), 369 – 390.
discussion of environmental responsive technology. Research Requate, T., 1995. Incentives to adopt new technologies under
Policy 25, 549 – 571. different pollution-control policies. International Tax and Public
Malueg, D.A., 1989. Emission credit trading and the incentive to Finance 2, 295 – 317.
adopt new pollution abatement technology. Journal of Environ- Requate, T., 1997. Green taxes in oligopoly if the number of firms is
mental Economics and Management 18, 52 – 57. endogenous. Finanzarchiv 54, 261 – 280.
Milliman, S.R., Prince, R., 1989. Firms incentives to promote Requate, T., Unold, W., 2001. On the incentives of policy
technological change in pollution control. Journal of Environ- instruments to adopt advanced abatement technology if firms
mental Economics and Management 17, 247 – 265. are asymmetric. Journal of Institutional and Theoretical Eco-
Montero, P., 2002a. Permits, standards and technology innova- nomics 157, 536 – 554.
tion. Journal of Environmental Economics and Management 44, Requate, T., Unold, W., 2003. Environmental policy incentives to
23 – 44. adopt advanced abatement technology—will the true ranking
Montero, P., 2002b. Market structure and environmental innovation. please stand up? European Economic Review 47, 125 – 146.
Journal of Applied Economics 5 (2), 293 – 325. Requate, T., in press. Commitment and timing of environmental
Parry, I., 1995. Optimal pollution taxes and endogenous techno- policy, adoption of new technology, and repercussions on R&D.
logical progress. Resource and Energy Economics 17, 69 – 85. Environmental and Resource Economics.
Parry, I., 1998. Pollution regulation and the efficiency gains from Spulber, D., 1985. Effluent regulation and long run optimality.
technological innovation. Journal of Regulatory Economics 14 Journal of Environmental Economics and Management 12,
(3), 229 – 254. 103 – 116.
Parry, I.W.H., Pizer, W.A., Fischer, C., 2003. How large are the van Soest, D.P., Bulte, E.H., 2001. Does the energy-efficiency
welfare gains from technological innovation induced by paradox exist? Technological progress and uncertainty. Environ-
environmental policies? Journal of Regulatory Economics 23 mental and Resource Economics 18 (1), 101 – 112.
(3), 237 – 255. van Soest, D.P., in press. The impact of policy instruments on the
Petrakis, E., Xepapadeas, A., 1999. Does governmental precommit- adoption of energy saving technologies under uncertainty.
ment promote environmental innovation? In: Petrakis, E., Resource and Energy Economics.
Sartzetakis, E.S., Xepapadeas, A. (Eds.), Environmental Regu- Weitzman, M., 1974. Prices vs. quantities. Review of Economic
lation and Market Power: Competition, Time Consistency and Studies 41, 477 – 491.
International Trade. Edward Elgar, pp. 145 – 161.