Journal of Regulatory Economics; 21:3 305±316, 2002 # 2002 Kluwer Academic Publishers. Manufactured in The Netherlands.

The Timing of Environmental Policy: A Note on the Role of Product Differentiation* JOANNA POYAGO-THEOTOKY

School of Economics, University of Nottingham,

Nottingham NG7 2RD, England, U.K. E-mail: [email protected]

KHEMARAT TEERASUWANNAJAK

School of Economics, University of Nottingham,

Nottingham NG7 2RD, England, U.K. E-mail: [email protected]

Abstract We examine the role that product differentiation can play in the design of environmental policy under full commitment and no commitment on the part of the environmental regulator. We consider a setting with two ®rms selling a differentiated product which generates pollution through emissions. Firms can reduce their emissions by undertaking abatement activities while an environmental regulator taxes emissions. The main results are: (1) When products are highly differentiated, the optimal time-consistent (no commitment) tax is always lower than the optimal pre-commitment tax. As the degree of product differentiation decreases, for relatively ef®cient abatement technology and high damages, the timeconsistent emission tax exceeds the optimal pre-commitment one. (2) Abatement when product differentiation is extensive is higher under the time-consistent regime unless the abatement technology is extremely ef®cient. The same ranking applies to social welfare. However, as products become more and more similar, these results are ( partially) reversed and pre-commitment could lead to both higher levels of abatement and welfare.

1. Introduction In recent years increased interest has been directed towards examining the role of commitment in a number of different policies pursued by government, e.g., Maskin and * This paper has bene®ted from unpublished work with E. Petrakis where some preliminary results have been derived. We are also grateful to an anonymous referee, Katrin Millock and Claudio Piga for very helpful comments. The usual disclaimer applies.

306

JOANNA POYAGO-THEOTOKY AND KHEMARAT TEERASUWANNAJAK

Newbery (1990) examine this for the case of international trade policy, Leahy and Neary (1996, 1997) for the case of R&D and output subsidies and Petrakis and Xepapadeas (1999a,b) for the case of environmental taxation. It is now well understood that the ability of a government or regulator to commit credibly to a particular policy has important implications for various aspects of the relevant economic activity. In the context of environmental policy, recent work by Petrakis and Xepapadeas (1999a,b), examines the issue of credibility when a government can tax the emissions of pollution-generating ®rms within (i) a monopoly and (ii) a homogeneous-product Cournot oligopoly. They show that, contrary to the generally held view about the superior welfare properties associated with a committed policy maker,1 there are cases where a non-credible policy (in the sense that it lacks commitment) results in higher welfare and induces ®rms to undertake increased abatement activities generating less pollution. In the present paper, we complement and elucidate this work by examining the role that product differentiation can play in the design of environmental policy. We consider a simple setting with two ®rms selling a differentiated product which generates pollution through emissions. Firms can reduce their emissions by undertaking abatement activities while an environmental regulator taxes emissions. The ability of the environmental regulator to commit credibly or not to a policy (in this case an emission tax) can be related to the positive and normative implications of two policy regimes. In the ®rst regime, the regulator precommits credibly to the emission tax, then the two ®rms decide on abatement given the tax and ®nally, they compete in the product market by choosing quantities. In the second policy regime, the regulator is unable to commit to a policy in a credible manner. In this case, when an emission tax is chosen ®rms would expect the regulator to change it after they have determined their investment in abatement. This is equivalent to having ®rms set abatement ®rst, then the regulator setting the emission tax and ®nally ®rms choosing their output. In this instance, the regulator follows a time-consistent policy. The main results we obtain are as follows: (1) When products are highly differentiated, the optimal time-consistent tax is always lower than the optimal pre-commitment tax. As the degree of product differentiation decreases, for relatively ef®cient abatement technology and high damages, the time-consistent emission tax exceeds the optimal precommitment one. (2) Abatement when product differentiation is extensive is higher under the time-consistent regime unless the abatement technology is extremely ef®cient. The same ranking applies to social welfare. However, as products become more and more similar, these results are ( partially) reversed and pre-commitment could lead to both higher levels of abatement and welfare. These results hint to the potentially important role that the extent of product differentiation can have in the design of environmental policy; for example, discretion (as opposed to commitment) on the part of the regulator/taxing agency seems to be indicated for industries where products are more differentiated. In relation to Petrakis and Xepapadeas (1999a,b), our results differ in that we concentrate on a ®xed market structure, a duopoly, and carefully examine how the degree 1 For example, see Leahy and Neary (1996, 1997) and Herguera et al. (1998).

307

THE TIMING OF ENVIRONMENTAL POLICY

of product differentiation interacts with the timing of the environmental policy.2 In contrast, Petrakis and Xepapadeas (1999b) use a homogeneous Cournot oligopoly and examine the effect of the size of the industry on environmental policy. Thus, when the industry is relatively small their results are qualitatively similar to the results we obtain for the case of a high degree of product differentiation while when the industry is relatively large these results get reversed.3 Hence, if we interpret a high degree of product differentiation as indicating a low degree of competition and vice versa it is clear that our results are complementary to theirs and that, in addition to the number of ®rms the degree of product differentiation is an equally important factor that needs to be taken account of in the design of environmental policy. The paper is structured as follows. In section 2 the model used is laid out. Sections 3±5 contain the results of the analysis while section 6 provides some concluding remarks.

2. The Model We examine a duopoly where ®rms sell differentiated goods and compete by setting quantities. In doing so we are extending recent work by Petrakis and Xepapadeas (1999a,b), [PX in what follows] in that we analyze in detail the effect that product differentiation has on the (uniform) emission tax, the abatement level chosen by ®rms and social welfare when (i) the regulator can precommit to the setting of the emission tax, or (ii) the regulator is unable to precommit. In line with Dixit (1979), we postulate that there is a representative consumer whose preferences for consumption of the two goods, q1 and q2 , and the numeraire good, m, are described by U…q1 ; q2 † ‡ m with U…q1 ; q2 † ˆ a…q1 ‡ q2 †

…q21 ‡ 2yq1 q2 ‡ q22 †=2

…1†

where qi is the output of ®rm i; i ˆ 1; 2; a > 0 and y [ …0; 1† captures the degree of product differentiation; the higher y, the lower is the degree of product differentiation. When y?0, the two ®rms are effectively local monopolists while y?1 implies that the two goods are virtually identical (or homogeneous).4 Thus a low y corresponds to a situation of rather limited competition and a higher y captures intensi®ed competition. The assumption of linear preferences in the numeraire good guarantees the absence of income effects and thus allows us to concentrate on a partial equilibrium analysis. The speci®cation for the utility function, although restrictive, results in a linear demand structure which is very tractable, pi ˆ a

qi

yqj ;

i; j ˆ 1; 2;

i 6ˆ j

…2†

2 Our results could be extended to the case of a ®xed number n-®rm oligopoly with no further insights gained. 3 Their simulation results are based on the comparison of two cases: n ˆ 3Ðwhich is identi®ed with a low degree of competition and n ˆ 10Ðwhich implies intense competition. 4 We note that the extreme cases of y?0 and y?1 correspond to the monopoly and duopoly with homogeneous product cases of PX. However, PX do not consider the duopoly case explicitly.

308

JOANNA POYAGO-THEOTOKY AND KHEMARAT TEERASUWANNAJAK

where pi is the price for good i and a is a measure of market size. The production of these differentiated goods results in pollution which is taxed at the rate t on emissions5 while ®rm i can reduce its tax burden by undertaking abatement, zi , to reduce its emissions. The cost function for ®rm i is given by c…qi ; zi † ˆ cqi ‡ g

z2i 2

where c is the unit cost of production, i.e., there are constant returns to scale and g captures the ef®ciency of the emission-reducing technology. Notice that abatement is characterized by decreasing returns, g > 0. In what follows, without loss of generality, we normalize the constant cost of production at c ˆ 0. Firm i's emissions are given by ei …qi ; zi † ˆ qi zi and it is obvious that ®rm i can reduce its emissions by zi by investing an amount g…z2i =2† in abatement.6 Given pollution, the extent of damage is captured via a quadratic damage function, " #2 2 1 X D…e1 ; e2 † ˆ d …e † ; 2 iˆ1 i where d is proportional to marginal damage. To guarantee an interior solution for abatement we assume that d > 1=2. In the sequel we compare two alternative policy regimes: (i) in the ®rst regime, the regulator sets the emission tax (i.e., precommits to the tax), then the duopolists choose their investment in abatement conditional on the emission tax and, ®nally, they set their output; (ii) in the second regime, ®rms move ®rst by choosing their abatement, then the regulator sets the emission tax and, ®nally, ®rms choose output. As is standard, we use subgame perfection as our solution concept.

3. Environmental Policy Under Regulator's Precommitment In the third stage, ®rm i chooses output to maximize pro®t pi ˆ …a

qi

yqj †qi

t…qi

zi †

1 2 gz 2 i

yielding the ®rst-order condition a

2qi

yqj

t ˆ 0:

…3†

5 We consider a uniform emission tax only. One could also consider the case of differentiated emission taxes, but this would burden the analysis without adding any additional insight to the issues addressed in this paper. 6 The particular choice for the speci®cation of the pollution generation process is made for the sake of simplifying the analysis. We conjecture that even with a non-linear function between output and abatement the results would be similar, notwithstanding the fact that a ®rm would have a strategic effect in the commitment case too. This effect would be more pronounced in the time-consistent case.

309

THE TIMING OF ENVIRONMENTAL POLICY

Imposing symmetry, qi ˆ qj :q, we obtain equilibrium output per ®rm qˆ

a t 2‡y

…4†

and pro®ts 1 2 gz : 2 i

2

pi ˆ q ‡ tzi

…5†

Notice that, from (4), as y?0, equilibrium output tends to the monopoly output while as y?1, equilibrium output approaches the homogeneous good Cournot output. In the second stage, ®rm i chooses abatement, zi , to maximize pro®ts as given by (5). The ®rst-order conditions yield z1 ˆ z2 ˆ

t g

…6†

and ®rm pro®ts are then pˆ

…a

t†

2

…2 ‡ y†

2

‡

t2 : 2g

…7†

In the ®rst stage, the regulator chooses the emission tax taking into account how ®rms will respond to it. Welfare is de®ned as the sum of consumer and producer surpluses, or, as the difference between the representative consumers' utility minus abatement costs and damage: !2 2 1 X d …q zi † 2 iˆ1 ! 2 2 1 X 1 X 2 g d zi …q 2 2 iˆ1 iˆ1

TW ˆ U…q; q†

2 1 X g z2i 2 iˆ1

ˆ 2aq

2

…1 ‡ y†q

!

!2 zi †

so the regulator chooses t to maximize the expression above. This maximization yields the optimal emission tax ~t ˆ 2

ag‰…2d 2

1†g ‡ 2d…2 ‡ y†Š D

…8†

where D ˆ 2d…g ‡ 2 ‡ y† ‡ g…2 ‡ y† ‡ g2 …1 ‡ y† > 0. Then we can calculate the equilibrium values for the remaining variables:

310

JOANNA POYAGO-THEOTOKY AND KHEMARAT TEERASUWANNAJAK

1†g ‡ 2d…2 ‡ y†Š D a‰…2d ‡ g†…2 ‡ g ‡ y†Š q~ ˆ D zˆ ~

a‰…2d

…9† …10†

and given that (by assumption)7 d > 1=2; ~z > 0 and ~t > 0. Moreover, q~ equilibrium pro®ts and welfare are:

~z > 0. Finally,

2 ~ ˆ q~2 ‡ ~t =2g p

…11†

2

f ˆ a …2d ‡ g†…3 ‡ y ‡ g† : TW D

…12†

This completes the analysis of this policy regime.

4. Time-Consistent Environmental Taxation (Non-Committed Regulator) The last stage (output choice) is the same as for the previous case. Output and pro®ts are given by (4) and (5) respectively. In the second stage, the regulator chooses the emission tax that maximizes total welfare, anticipating the choice of output by ®rms in the third stage. Welfare is de®ned as in the previous section: TW ˆ U…q; q† ˆ 2aq

!2 2 1 X d …q zi † 2 iˆ1 ! 2 2 1 X 1 X 2 g d zi …q 2 2 iˆ1 iˆ1

2 1 X g z2i 2 iˆ1

…1 ‡ y†q

!

!2 zi †

:

The ®rst-order condition is given by: 2 a

…1 ‡ y†

2 X …q d iˆ1

with dq dt ˆ …2 ‡ y† emission tax is

1

!! zi †

dq ˆ0 dt

…13†

< 0. Using the expression for output (4), into (13) the optimal

t* ˆ

a…2d

1† d…z1 ‡ z2 †…2 ‡ y† : 1 ‡ 2d ‡ y

7 Note that d > 1=2 is suf®cient to guarantee that ~z > 0.

…14†

311

THE TIMING OF ENVIRONMENTAL POLICY

It is evident from (14) that an increase in abatement by either ®rm leads to a reduced emission tax, …dt=dzi † < 0 (this is a strategic effect); as identi®ed by PX a ®rm will strategically use its abatement to lower the tax on its emissions. Firms can in¯uence the tax on emissions: by increasing investment in emission-reducing activities they can expect a lower emission tax. Of course, this strategic aspect is absent from the case of precommitment to the emission tax. Using (14) in (4) and (5) we obtain q* ˆ

a ‡ d…z1 ‡ z2 † 1 ‡ 2d ‡ y

p*i ˆ q* 2 ‡ t* zi

…15†

1 2 gz : 2 i

…16†

In the ®rst stage, ®rms choose their abatement level, anticipating the regulator's choice of emission tax and how this is affected by their (strategic) choice of emission-reducing abatement activities. Notice that from the f.o.c. in this stage we can derive the slope of the reaction functions in …zi ; zj † space, i.e., qzi ˆ qzj

d…1 ‡ y†…2 ‡ 2d ‡ y† 2

g…1 ‡ y† ‡ 2d…1 ‡ y†…2 ‡ 2g ‡ y† ‡ d 2 …6 ‡ 4g ‡ 4y†

< 0;

…17†

in other words abatement is a strategic substitute. This is in contrast to the commitment case where …qzi =qzj † ˆ 0. This observation will be useful when explaining the results in the next section. Firm i maximizes pro®t, as in (16), yielding the (symmetric) equilibrium abatement level z* ˆ

a‰4d2 ‡ …2d 1†…1 ‡ y†Š G

…18†

2

where G ˆ g…1 ‡ y† ‡ 2d 2 …4 ‡ 2g ‡ 3y† ‡ d…1 ‡ y†…6 ‡ 4g ‡ 3y† > 0 and given that d > 12 it is evident that z* > 0. Using (18) we obtain the remaining equilibrium values for this policy regime: t* ˆ

a‰…2d

q* ˆ

1†…1 ‡ 2d ‡ y†g ‡ d……2d G

1†y

2†Š

a‰g ‡ 2d…2 ‡ 2d ‡ g† ‡ …3d ‡ g†yŠ G

p* ˆ q * 2 ‡ t * z * TW * ˆ …2a

…1 ‡ y††q*

…20†

1 *2 gz 2 2d…q*

…19†

…21† z* †

2

2

gz* :

…22†

In this case too it can be checked that q* z* > 0, i.e., there net emissions are positive in equilibrium. Having completed the analysis of the two different policy regimes we now proceed in comparing them.

312

JOANNA POYAGO-THEOTOKY AND KHEMARAT TEERASUWANNAJAK

5. Comparing Policy Regimes In this section we provide a comparison in terms of emission tax, abatement and welfare.8 We summarize our ®ndings in a number of propositions. Assumption: Let 1=2 < d < d …y†. This is a technical assumption, needed in the proof of Proposition 1. Essentially it excludes very high values of d, i.e., it restricts the degree of convexity of the damage function. b such that for all d < d…y†, b e Proposition 1: For any ®xed g there is a d…y† t > t* , i.e., the optimal pre-commitment tax is higher than the optimal time-consistent tax, and for all b b e d  d…y†, t  t* , i.e., the opposite is true. The critical value d…y† is decreasing in the product differentiation parameter, y. Proof: From (8) and (19) we obtain e t t* ˆ

a…2 ‡ y†…T0 ‡ gT1 ‡ g2 T2 † GD

…23†

where T0 ˆ 2d 2 …2 ‡ y†‰2 ‡ y…1 2d†Š, T1 ˆ d‰…2 ‡ y†…4 ‡ 3y 10d† 4d 2 yŠ, and T2 ˆ 2 ‡ 4d…1 ‡ d† ‡ y…3 ‡ 2d ‡ y† > 0. Note that sign… e t t* † ˆ sign…T0 ‡ gT1 ‡ > 2 < g T2 † as GD > 0. Furthermore, T0 < 0 as d> d…y†, where d…y† ˆ …2 ‡ y=2y† and …d…y††0 < 0. T1 > 0 as d < d  …y†, where



d …y† ˆ

10 ‡ 5y ‡

p …2 ‡ y†‰50 ‡ y…41 ‡ 12y†Š 4y

(we discard the other solution to T1  0 being negative) and …d  …y††0 < 0. Note also that b d…y† < d …y†. Then, for g?0, sign… e t t* † ˆ sign…T0 †, and d…y†:d…y† so that e t > t* for b b * d < d…y† and e t  t for d  d…y†. Further, since T1 > 0 and T2 > 0, the critical value 2 b d…y†:fdjT 0 ‡ gT1 ‡ g T2 ˆ 0g is higher than d…y† for positive values of g and 0 9 b …d…y†† < 0. The result then follows. & Corollary: For y?0, i.e., when products are very different, the optimal pre-commitment tax is always higher than the optimal time-consistent emission tax, e t > t* . Proof: Straightforward. 8 A detailed comparison of pro®t levels can be found in Teerasuwannajak (1999). b is available from the authors on request. 9 The exact solution for d…y†

&

THE TIMING OF ENVIRONMENTAL POLICY

313

The intuition behind Proposition 1 is as follows: In the case of no-commitment, each ®rm has a strategic incentive to increase abatement in order to induce the regulator to impose a lower emission tax subsequently (see (14)). This aspect is absent when the regulator precommits  to an emission tax. From (17), q=qy qzi =qzj < 0.10 Combining this with the observation that the strategic effect encourages over-investment, q=qy‰dt=dzi Š < 0, explains formally why as the degree of product differentiation diminishes (and hence competition intensi®es) the incentive to overinvest in emission reduction decreases.11 In other words, when y is high and hence competition is intense, by exploiting the strategic aspect of investing in abatement a ®rm stands to gain relatively more through a lower emission tax than in the case of a low y. However, each ®rm will reason in this way, and given that the emission tax affects both of them it will prefer the rival ®rm to undertake the relevant investment (free-riding) so that, in equilibrium, both ®rms will have less incentive to overinvest in emission reduction and thus the emission tax will be higher relative to the pre-commitment case. As products become more differentiated, this latter mechanism becomes less and less important and in the limit, when y?0, we have that the pre-commitment tax exceeds the time-consistent tax. Next we consider the comparison of abatement. Proposition 2: For ®xed d, there is a b g…y† such that for g > b g…y† each ®rm's abatement is lower in the case of commitment, e z < z* while for g < b g…y† the opposite holds, e z > z* . The critical value b g…y† is increasing in the product differentiation parameter, y. Proof: From (9) and (18) we obtain e z

z* ˆ

a…W0 ‡ gW1 ‡ g2 W2 † GD

…24†

where W0 ˆ 2d…2 ‡ y†‰2d2 y ‡ …1 ‡ y†…2 ‡ y†…1 ‡ d†Š > 0, W1 ˆ 4d 3 y d…1 ‡ y† 2 …2 ‡ y† ‡ …1 ‡ y†…2 ‡ y† ‡ 2d 2 ‰y…y 2† 6Š and W2 ˆ 2d…1 ‡ 2d ‡ y† < 0. Note * that sign…e z z † ˆ sign…W0 ‡ gW1 ‡ g2 W2 † as GD > 0. For g?0, sign…e z z* † ˆ 2 sign…W0 † > 0. For g > 0, note that q=qg…W0 ‡ gW1 ‡ g W2 † ˆ W1 ‡ 2gW2 > < 0 as g< W1 =2W2 . Given W2 < 0 and the non-negativity of g, g 6< 0, we can discard the > case of g < W1 =2W2 as some values of y would lead to a negative value for g. Hence we are left with g > W1 =2W2 and it is thus the case that W1 ‡ 2gW2 < 0. De®ning by b g…y† the positive solution to W0 ‡ gW1 ‡ g2 W2 ˆ 0 (there are two solutions; one positive12 and one negative) we establish that for g > b g…y†, e z < z* and for all 0  g < b g…y†, e z > z* . Further, q=qy‰b g…y†Š > 0. The result then follows immediately. &

  10 Note that q=qy qzi =qzj ! ‰2d2 …1 ‡ 2g†…2d ‡ 3 ‡ 2y† ‡ g…1 ‡ y†2 …2d 1†Š. 11 Note that q=qy‰dt=dzi Š ˆ q=qy‰ …2 ‡ y†d=1 ‡ y ‡ 2dŠ ˆ d…1 2d†=…1 ‡ 2d ‡ y†2 < 0, given that d > 1=2 (by assumption). 12 The exact solution for b g…y† is available from the authors upon request.

Figure 1. Total welfare under time-consistent and precommitment policies.

314 JOANNA POYAGO-THEOTOKY AND KHEMARAT TEERASUWANNAJAK

THE TIMING OF ENVIRONMENTAL POLICY

315

For given y, when the abatement technology is relatively ef®cient (low g) ®rms tend to do more so that they face a lower tax in the time-consistent case but, in equilibrium, given the explanation given for Proposition 1, e z > z* . Further, as y increases, i.e., products become less differentiated and competition intensi®es it is more likely that e z < z* . Thus, in this instance no-commitment promotes abatement. Next we examine the welfare ranking of the two policy regimes. Claim For each d, there is a e g…y† such that for g < e g…y†, social welfare in the case of f > TW * , commitment is higher than in the case of no-commitment (time-consistency), TW f * e e while for g > g…y† the opposite is true, TW < TW . g…y† is increasing in y. f TW * . Proof: From (12) and (22) we obtain the difference in social welfare, TW Plotting this difference for various values of y, y [ …0; 1†, yields the result. Figure 1 illustrates. Note that the dark area represents combinations of g and d such that the f TW * is negative, while the white areas indicate relevant combinations difference TW for a positive difference. & Our intuition is as follows: For low g, abatement is relatively effective so that ®rms have a greater incentive in doing more of it in the time-consistent policy regime; however, in the equilibrium, ®rms invest relatively less as we know from Proposition 2, e z > z* . This results in higher emissions and reduced welfare for the time-consistent scenario. For large g, abatement is less effective and ®rms are affected by the free-rider problem to a lesser extent so that, again from Proposition 2, z* > e z which leads to lower emissions and hence f < TW * . In other words, a regulator who cannot commit credibly can generate higher TW welfare in the case of relatively expensive (i.e., less effective) emission reduction activities (high g) for a given degree of product differentiation. For the case of low product substitutability, y?0, the mechanism leading to reduced investment in emission reduction (see discussion following Proposition 1) is very dampened but as y increases and products become more similar it becomes more pronounced and it is likely that welfare in the precommitment scenario is higher. Figure 1 provides a graphic illustration for the above Claim. Note that in the white areas in ®gure 1 social welfare is higher under precommitment while in the shaded areas the opposite is true.

6. Concluding Remarks In this paper, we have considered the case of a duopoly producing a differentiated product and identi®ed the extent of product differentiation as the driving force of our results. A high degree of product differentiation coincides with a low degree of competition while a low degree of product differentiation implies intense competition. We have shown that, in the limited context of the model used, when products are highly differentiated, the optimal emission tax under no-commitment (time-consistent tax) is always lower than the optimal emission tax under commitment. Moreover, abatement and social welfare are higher under the no-commitment policy regime, unless the abatement technology is extremely ef®cient. As products become less differentiated, commitment on the part of the regulator could

316

JOANNA POYAGO-THEOTOKY AND KHEMARAT TEERASUWANNAJAK

result in increased abatement and welfare. These results show that product differentiation can play a signi®cant role in the design and implementation of environmental policy when taking into account the prevalence of differentiated commodities. The role of product differentiation needs to be further examined in a number of alternative settings to check the robustness of the results obtained in the present paper. This has to be left for further research.

References Dixit, A. 1979. ``A Model of Duopoly suggesting a Theory of Entry Barriers.'' Bell Journal of Economics 10: 20±32. Herguera, I., P. Kujal, and E. Petrakis. 1998. ``Non-credible Policies and Leap-frogging in Vertically Differentiated Industries.'' Mimeo, Universidad Carlos III de Madrid, Madrid. Leahy, D., and J. P. Neary. 1996. ``International R&D Rivalry and Industrial Strategy without Government Commitment.'' Review of International Economics 4: 322±338. Leahy, D., and J. P. Neary. 1997. ``Public Policy towards R&D in Oligopolistic Industries.'' American Economic Review 87: 642±662. Maskin, E., and D. Newbury. 1990. ``Disadvantageous Oil Tariffs and Dynamic Consistency.'' American Economic Review 80: 143±156. Petrakis, E., and A. Xepapadeas. 1999a. ``Does Government Precommitment Promote Environmental Innovation?'' In Environmental Regulation and Market Power, edited by E. Petrakis, E. S. Sartzetakis and A. Xepapadeas, Cheltenham: Edward Elgar Publishing. Petrakis, E., and A. Xepapadeas. 1999b. ``Environmental Policy, Government Commitment and Market Power.'' Mimeo, University of Crete, Rethymnon, Crete. Teerasuwannajak, K. 1999. Product Differentiation in Determining the Environmental Policy Regime. MSc thesis, University of Nottingham, School of Economics.