Environmental Tax Competition in the Presence of Multinational Firms

Transcription

1 Environmental Tax Competition in the Presence of Multinational Firms Essi Eerola FPPE, University of Helsinki February 2002 Abstract We study the design of environmental policy in a two country model with an imperfectly competitive polluting industry. We Þrst consider policy design in a setting where polluting Þrms are national and serve foreign markets by exports anhenasituationwhereþrms are multinational and have production plants in both countries. We show that while the non-cooperative policy design leads to too lax emission taxation when polluting Þrms are national, it leads to too severe taxation when they are multinational. The reason is twofold. When Þrms are multinational, environmental policy is not useful in shifting rents to domestic shareholders. In addition, when designing their policies, national governments ignore proþts accruing to foreign shareholders of multinational Þrms. I wish to thank Markus Haavio, Erkki Koskela, Ville Mälkönen, Niku Määttänen, Markku Ollikainen and the participants in the ESF conference International Dimension of Environmental Policy in Acquafredda for useful comments. The usual disclaimer applies. This paper is part of the project Studies in Environmental and Resource Economics Þnanced by the Academy of Finland. The funding is gratefully acknowledged. Department of Economics, P.O. Box 54, FIN University of Helsinki, Finland; Fax: ; essi.eerola@helsinki.þ 1

2 Keywords: Environmental policy, multinational Þrms, tax competition JEL ClassiÞcation: D62, F23, H77 1 Introduction Multinational Þrms play an increasingly important role in international competition. In 1997 the gross product of all multinational Þrms accounted for roughly one quarter of the world s gross product. In addition, global sales of foreign affiliates of multinational Þrms were twice as high as global exports indicating that to a large extent multinational Þrms spread horizontally, that is, exports are at least partially replaced by foreign production. 1 An often expressed concern is that the increasing dominance of multinational Þrms will undermine the efficiency of national environmental regulation. When faced with tightening environmental regulation, multinational Þrms, unlike national Þrms, are able to shift production from one country to another so as to escape higher tax rates. The fear is that this may induce national governments to relax environmental protection in order to prevent domestic production from ßeeing abroad. Whether this concern of detrimental regulatory competition between national governments is valid, is an open question, as the problem of regulating multinational polluting Þrms has not been considered in the economics literature. In contrast, the possible inefficiency of national environmental policies when polluting Þrms are national and compete internationally has been studied by several authors. This literature provides a background and starting point for the present study. It is well established that when the international market is perfectly competitive and environmental damage caused by emissions is local, there is no reason to expect environmental policy of a small country to be distorted. 2 However, when the polluting Þrms have market power, matters are different. In a framework of strategic trade policy introduced by Brander and Spencer (1985), Barrett (1994) and Ulph (1996) show that national govern- 1 See UNCTAD (2000). 2 See e.g. Rauscher (1994). 2

3 ments set emission standards that lead to marginal cost of abatement being lower than marginal damage caused by emissions. In a similar setting, Conrad (1993) and Kennedy (1994) show that non-cooperative emission tax rates are lower than the tax rate that would prevail if the national governments coordinated policies so as to maximize joint welfare of the countries. The intuition is that relaxing environmental standards induces domestic producers to expand output and their foreign rivals to reduce output, and the positive effect of increased market share and thus higher proþts for domestic shareholders dominates the negative welfare-effect caused by increased emissions. 3 This paper focuses on the design of national environmental policy in the presence of multinational Þrms. The main questions to be addressed are: First, how does emergence of multinational Þrms change the incentives of national governments for environmental regulation? Second, when the polluting Þrms are multinational, would coordination of environmental policies imply more lax or more severe regulation? Following the literature on strategic trade policy, we build a two country model of regulatory competition and extend it to multinational Þrms. Environmental regulation is imposed on an existing industry after negative welfare-effects of emissions have been identiþed. 4 Each national government controls the level of emissions by levying a tax on emissions generated in its country. The governments are constrained to set a uniform tax rate that applies to all polluting plants within their territory. 5 We consider both non- 3 These results may change if producers compete in prices instead of quantities (Barrett 1994) or if there are general equilibrium effects in factor markets (Rauscher 1994). See also Nannerup (2001) where domestic production consists of several industries. 4 Markusen et al. (1995), Rauscher (1995), and Hoel (1997) analyze a situation where national governments Þrst choose emission taxation and a Þrm or Þrms then choose where to locate. Unless transportation costs are high enough, the Þrms produce in one country only. In general, there may be several reasons to become multinational. Main conclusions of the literature on multinational production are that foreign production is preferred to exports if Þrm-level scale economies or transportation costs are high and plant-level scale economies and barriers to direct investments are low. Multinational production is also more likely to arise if the countries are relatively similar in factor endowment or in size. For a survey on the literature see Markusen (1995). 5 This would be the case for instance within the EU single market. 3

4 cooperative and coordinated policy design. Policy design is analyzed under two different industry structures. In the Þrst case, the two countries are occupied by national Þrms that serve the foreign market by exports. In the second case, the countries are occupied by multinational Þrms that have production plants in both countries and are thus able to shift production from one country to another. This reßects the assumption stated above that being multinational is advantageous in the sense that it allows the Þrm to respond more efficiently to changes in national regulatory environment. We conþrm the previous Þnding that setting emission taxation in a non-cooperative manner leads to too lax emission regulation when the polluting Þrms are national. However, when the polluting Þrms are multinational, we demonstrate a strikingly different result: competition between the governments then leads to too severe emission taxation. Thus, relaxing emission taxation from the non-cooperative level increases welfare in both countries. The reason is that the incentives of national governments for environmental regulation change in two ways when Þrms are multinational compared to when they are national. First, when one government tightens emission taxation, both multinational Þrms reduce output in that country and shift production to the other country while cutting back supply in both markets. As multinational Þrms react in the same way to any policy change, changes in environmental policy do not inßuence the market share of the domestic Þrm. Environmental policy is thus not useful in shifting rents to domestic shareholders. Second, when polluting Þrms are multinational, in each country there is a production plant that is largely or entirely in foreign ownership. When setting policies, national governments ignore proþts that accrue to foreign shareholders of these plants and are inclined to set too severe emission taxation. The remaining of the paper is organized as follows. In section 2 we present the set-up. In sections 3 and 4 we analyze emission tax policies in the case of national Þrms and multinational Þrms. Section 5 concludes. 4

5 2 The Set-Up We consider a model of two identical countries, home and foreign, and a polluting industry consisting of one Þrm in each country. Consumers in the two countries beneþt from consumption of the good produced by the Þrms, but experience a welfare loss due to harmful emissions generated by production. Both Þrms may supply both foreign and home market. The aggregate supply in home market, X, isthus X = x H + x F, where x H is supply by home Þrm and x F supplybyforeignþrm. In the same manner, aggregate supply in foreign market, X,is X = x H + x F, where x H is supply by home Þrm and x F Þrm j = H, F is then is supply by foreign Þrm. The revenue of p (X) x j + p (X ) x j, where p ( ) is the inverse demand function in the two countries. We will assume that Assumption 1 Demand function satisþes: p 0 ( ) < 0 and p 00 ( ) h + p 0 ( ) < 0 for h = x H,x F,x H,x F. The plant level production cost function for each Þrm is c( ). ThetwoÞrms possess technology to abate emissions generated by production and the plant level abatement cost function for both Þrms is g ( ). Throughout the paper we assume that Assumption 2 Production and abatement cost functions satisfy: c 0 ( ) > 0, g 0 ( ) > 0, and c 00 ( ) > 0, g 00 ( ) > 0. 5

6 Convex production costs reßect the idea that expanding production in one plant becomes increasingly costly as the level of production increases. Constant marginal costs, besides implausible, would imply the undesirable feature that production in one plant could be increased without limit and thus even a marginal increase in emission tax rate would induce multinational Þrms to shift all production from one country to another. 6 Together Assumptions 1 and 2 are sufficient to guarantee that the second-order conditions for proþt maximization for national and multinational Þrms are satisþed and that the reaction functions of the Þrms are downward sloping. Let y j denote the plant level production of Þrm j. We choose units of emission so that when the plant level abatement of Þrm j is a j, emissions generated by the production of that plant, e j,are e j = y j a j. National governments control the level of emissions by levying an emission tax. Each government levies a tax on emissions generated in its own country. Tax revenues are distributed back to consumers as lump sum transfers. Let t =(t, t ) denote the vector of emission tax rates where t is tax rate in home country and t tax rate in foreign country. Emissions generated by production are local, and the overall level of emissions in home country is E and in foreign country E. The negative welfare-effect of emissions is d (E) in home country and d (E ) in foreign country with d 0 ( ) > 0 and d 00 ( ) 0. Using the framework presented above, we will analyze the design of environmental policy under two different industry structures. In the Þrstcasetobeanalyzed,Þrms are national and may serve foreign market by exports. Each Þrm has one production plant and thus plant level output and emissions correspond to total output and emissions in 6 Plant level Þxed set-up costs are not explicitly considered here as the number of plants is exogenous to the analysis. We assume that these costs are such that only two Þrms exist in the market despite the convex production costs. 6

7 each country. Total cost of production for home and foreign Þrm are then c y H + g a H + te H and c y F + g a F + t e F. As both Þrms have only one production plant, overall emissions in home country are E = e H and emissions in foreign country E = e F. In the second case, the two countries are occupied by two multinational Þrms establishedinbothcountries. EachÞrm has one production plant in each country. In order to distinguish between the two plants of one Þrmlocatedindifferent countries, we let a star refer to a production plant located in foreign country. Thus total cost of production for Þrm j is c y j + c y j + g a j + g a j + te j + t e j. Consequently, when Þrms are multinational total emissions in home country are E = e H + e F and total emissions in foreign country E = e H + e F. 7 Timing of the events is as follows. The national governments of the two countries Þrst simultaneously set their emission tax rates. After the policies have been chosen, the Þrms decide how much to produce and how much to supply in each market. The multinational Þrms also decide where to produce. We solve the problem in the usual manner by backward induction. 3 Tax Policy and National Firms We begin by solving the proþt maximization problems of the national Þrms given the tax policies chosen by the two governments. After that we consider Þrst non-cooperative policy design and then policy design under coordination. 7 Policy design is analyzed separately under the two industry structures. Therefore, the overall number of production plants being different in the two cases does not inßuence the results. One could equally well consider a situation where also in the case of national Þrms the total number of production plants is four, that is, each national Þrm has two production plants located in one country. 7

8 3.1 National Firms Let π H denote proþts of home Þrm and π F proþts of foreign Þrm. Given any pair of tax rates (t, t ) set by the national governments, proþt maximization problems of the Þrms are and max π H = p (X) x H + p (X ) x H c y H g a H te H y H,a H,x H max π F = p (X ) x F + p (X) x F c y F g a F t e F, y F,a F,x F where X and X are aggregate supplies in home and foreign market and x H = y H x H and x F = y F x F. Assuming interior solutions, Þrst-order conditions for proþt maximization for home Þrm are π H y = H p0 (X) x H + p (X) c 0 y H t =0, (1) π H a = t H g0 a H =0, (2) π H x = H p0 (X ) x H + p (X ) p 0 (X) x H p (X) =0. (3) The Þrst-order conditions of foreign Þrm are derived in an analogous manner. 8 The conditions imply that decisions on output, abatement, and supply are optimal when marginal production cost equals marginal revenue, marginal abatement cost equals emission tax rate, and marginal revenues in the two markets are equalized. We denote the proþt maximizing output, supply, and emission levels of Þrm j as a function of the tax rates by y j (t), x j (t), x j (t), ande j (t). As demand conditions are identical in the two countries, the Þrms divide their output equally between the two markets, i.e. x j (t) =x j (t) for j = H, F. Consequently,X (t) =X (t). Given Assumptions 1 and 2, from the Þrst-order conditions it follows that 8 Derivates of functions of one argument are denoted by primes. Subscripts refer to partial derivatives of functions of several arguments. 8

9 Characterization 1 When home government tightens emission control, home Þrm cuts down supply in both markets and foreign Þrm expands supply in both markets. Emissions decrease in home country and increase in foreign country. Aggregate supply in both markets and overall emissions decrease. Proof. See Appendix. By using conditions (1)-(3), the effect of a marginal policy change by home government on the proþt ofhomeþrm can be expressed as π H t (t) = e H (t)+p 0 (X (t)) x F t (t) y H (t), (4) where x F t (t) is the marginal change in proþt maximizing supply of foreign Þrm in each market. The direct effect of a higher tax rate is the increased tax burden on each unit of emissions captured by the Þrst term in equation (4). The second term in turn reßects the effect of a tax change on the proþt ofhomeþrm through supply decisions of its rival. Since by Characterization 1 x F t (t) > 0, thiseffect is negative in both markets. Thus, π H t (t) < 0. Inthesamemanner,theeffect of a marginal policy change by home government on the proþt of foreign Þrm is π F t (t) =p0 (X (t)) x H t (t) y F (t), (5) where x H t (t) is the marginal change in proþt maximizing supply of home Þrm in home and foreign market. As by Characterization 1 x H t (t) < 0, it directly follows that π F t (t) > 0. Equations (4) and (5) imply that by relaxing environmental standards home government increases proþts of home Þrm and reduces proþts of its rival. These asymmetric effects of taxation enable national governments to use environmental policy to shift proþts for domestic producers. For notational clarity we adopt the following notation: z H,t = p 0 (X (t)) x F t (t) yh (t), (6) z F,t = p 0 (X (t)) x H t (t) y F (t), (7) z H,t = p 0 (X (t)) x F t (t) yh (t), (8) 9

10 and z F,t = p 0 (X (t)) x H t (t) yf (t). (9) From the previous discussion it follows that z H,t < 0 and z F,t > 0, whereasz H,t > 0 and z F,t < Emission Tax Competition Consider now the problem of the national governments when they choose their tax policies non-cooperatively taking the tax policy of the other government and the equilibrium behavior of the Þrmsasgiven. Wedenoteaggregatewelfareinhomecountry by w andinforeigncountrybyw. Since the countries are identical in what follows we will mainly focus on home government. The objective of each government is to maximize domestic aggregate welfare which consists of consumer surplus, proþts of the national Þrm, tax revenue, and disutility from pollution. The objective of home government is then to choose t so as to maximize w = Z X 0 p (h) dh p (X (t)) X (t)+π H (t)+te (t) d (E (t)), and the objective of foreign government to choose t so as to maximize w = Z X 0 p (h) dh p (X (t)) X (t)+π F (t)+t E (t) d (E (t)). The Þrst-order conditions for welfare maximization, w t =0and wt =0, implicitly determine equilibrium tax rates. By using equations (4) and (6), the Þrst-order condition for home government can be written as p 0 (X (t)) X t (t) X (t)+z H,t +(t d 0 (E (t))) E t (t) =0, where the Þrst term is the effect of a policy change on domestic consumer surplus, the middle term is the effect on the proþt of the domestic Þrm through supply decisions of its rival, and the third term is the welfare-effect of a change in emissions. 10

11 Under the assumption that the second-order conditions for welfare maximization are satisþed, that is, w tt 0 and wt t 0, theþrst-order conditions determine a unique equilibrium which we denote by t n =(t n,t n ), where superscript n refers to non-cooperative outcome. Because the Þrms and countries are identical, t n = t n. As a result, in the non-cooperative equilibrium, the level of output and emissions are the same in the two countries, i.e. y H (t n )=y F (t n ) and E (t n )=E (t n ).TheÞrst-order condition for home government can thus be rewritten as (t n d 0 (E (t n ))) E t (t n )=p 0 (X (t n )) x H t (t n ) X (t n ), (10) where in the left hand side we have the welfare-effect through a change in emissions and in the right hand side the welfare-effect through a change in market price due to the supply reduction by home Þrm. As Characterization 1 shows, in addition to inßuencing the supply decisions of home Þrm, a higher tax rate in home country induces foreign Þrm to increase supply which beneþts consumers in home country, but reduces the proþt of home Þrm. However, in a symmetric equilibrium where x H (t n )=x F (t n ),these effects cancel each other. 9 The right hand side of equation (10) is thus strictly positive. In addition, by Characterization 1 E t (t) < 0. Thus in a symmetric non-cooperative equilibrium t n = t n <d 0 (E (t n )) = d 0 (E (t n )), (11) i.e. equilibrium tax rates are lower than the marginal damage caused by emissions. 9 Rewriting p 0 (X (t)) X t (t) X (t)+z H,t gives p 0 (X (t)) X t (t) X (t)+p 0 (X (t)) x F t (t) x H (t)+p 0 (X (t)) x F t (t) x H (t) = p 0 (X (t)) x H t (t) X (t)+x F t (t) X (t) x F t (t) x H (t) x F t (t) x H (t) = p 0 (X (t)) x H t (t) X (t)+x F t (t) x H (t)+x F t (t) x F (t) x F t (t) x H (t) x F t (t) x H (t) = p 0 (X (t)) x H t (t) X (t)+x F t (t) x F (t) x H (t). 11

12 3.3 Coordinated Tax Policies We now turn to the problem of the national governments when they coordinate policies and choose emission tax rates so as to maximize joint welfare of the two countries denoted by W. Joint welfare consists of proþts of the two national Þrms and consumer surplus, tax revenue, and disutility of pollution in the two countries. Thus the objective of the governments is to choose (t, t ) so as to maximize W = π H (t)+π F (t)+cs (t)+cs (t)+te (t)+t E (t) d (E (t)) d (E (t)), where cs (t) = Z X 0 p (h) dh p (X (t)) X (t) and cs (t) =cs (t) since X (t) =X (t) for all tax rates. The Þrst-order conditions for maximization of joint welfare, W t =0and W t =0, implicitly determine the equilibrium tax rates under coordination. By using equations (4)-(7), the Þrst-order condition for home government can be written as cs t (t)+cs t (t)+z H,t + z F,t +(t d 0 (E (t))) E t (t)+(t d 0 (E (t))) Et (t) =0, where the Þrst two terms are the effects of taxation on consumer surplus in the two countries, z H,t and z F,t are the indirect effects of a higher tax rate in home country on proþts through market price, and the last two terms are the welfare-effects through changes in emissions. Again, under the assumption that W tt 0 and W t t 0, theþrst-order conditions determine a unique equilibrium denoted by t o =(t o,t o ), where superscript o refers to the cooperative outcome. Due to the symmetry of the countries t o = t o, and thus y H (t o )=y F (t o ) and E (t o )=E (t o ).TheÞrst-order condition for home government can therefore be simpliþed to (t o d 0 (E (t o ))) (E t (t o )+E t (to )) = p 0 (X (t o )) y H t (t o )+y F t (t o ) x H (t o ), (12) 12

13 where in the left hand side we have the effect of a tax increase through change in emissions and in the right hand side the effect of a tax increase through market price. Characterization 1 implies that the right hand side is strictly positive and E t (t) + Et (t) < 0. Thus t o = t o <d 0 (E (t o )) = d 0 (E (t o )). (13) Not surprisingly, in the cooperative equilibrium, marginal damage of emissions is higher than emission tax rates. Complete internalization of the damage caused by emissions is not socially desirable as tightening emission control induces polluting Þrms with market power to further reduce suboptimally low supply. 10 Direct comparison of the Þrst-order conditions (10) and (12) does not reveal in what way and why the non-cooperative outcome is inefficient. In order to establish what kind of policy change would increase joint welfare of the two countries from the non-cooperative level, we evaluate W t and W t at the non-cooperative equilibrium. If W t > 0 and W t > 0 at (t n,t n ), we conclude that a policy change introducing a marginal increase in tax rates from the non-cooperative level would increase welfare in both countries. Countries being identical we concentrate on home country alone. By using equation (10) we have that at (t n,t n ) W t =(t n d 0 (E (t n ))) Et (t n ) p 0 (X (t n )) x F t (t n ) X (t n ), (14) whichconsistsofthreedifferent welfare-effects that home government imposes on foreign consumers when it tightens emission taxation. First, a higher tax rate in home country induces foreign Þrm to expand output which increases emissions and tax revenue in foreign country. Since t n <d 0 (E (t n )), this increase in emissions reduces welfare of foreign consumers. Second, an increase in t induces home Þrm to cut down production and reduce supply in both markets. This increases proþts for foreign Þrm but reduces consumersurplusinforeigncountry. Asnotedinconnectionwithequation(10),in 10 This is the same mechanism, unrelated to regulatory competition, as in Barnett (1980). See also Kennedy (1994). 13

14 a symmetric equilibrium these two effects cancel each other. Third, supply expansion by foreign Þrm increases consumer surplus in foreign country. This positive effect is captured by the second term in equation (14). Consequently, an increase in t has a negative effect on foreign welfare through increased emissions but a positive effect through larger consumer surplus. However, it can be shown that the relative size of these two effects in the non-cooperative equilibrium is such that Proposition 1 When the two countries are occupied by national Þrms, setting policies non-cooperatively leads to too lax emission regulation. Proof. See Appendix. The non-cooperative tax rates are lower than cooperative tax rates that maximize the joint welfare of the two countries. From Characterization 1 it then also follows that level of emissions in both countries and supply in both markets are higher in the non-cooperative equilibrium than when decisions on taxation are taken in a coordinated manner. 4 Tax Policy and Multinational Firms In this section, we consider multinational Þrms that have production plants in both countries. The crucial difference with the previous analysis is that multinational Þrms are able to react to policy changes not only by adjusting supply and abatement levels, but also by shifting production from one country to another. In order to study tax policies when polluting Þrms are multinational, we Þrst have to clarify ownership structure of the multinational Þrms. By assumption fraction α of the shares of home Þrm is owned by consumers in home country and fraction (1 α) is owned by foreign consumers. Similarly, fraction (1 β) of the shares of foreign Þrm is owned by consumers in home country and the rest by foreign consumers. These ownership shares are exogenous to the analysis. We concentrate on a symmetric situation 14

15 where α = β. In the extreme situation where α = β =1, each multinational Þrm is entirely owned by consumers in one country. We proceed as in section 3 by Þrst solving the proþt maximization problems of the Þrms and then considering the problem of the national governments. 4.1 Multinational Firms Givenemissiontaxes(t, t ) set by the national governments, proþt maximization problems of Þrm j = H, F is max = p (X) x j + p (X ) x j c y j g a j te j c y j g a j t e j, y j,a j,y j,a j,i jπj where i j denotes net imports by Þrm j to home market. So, for Þrm j supply in home market is x j = y j + i j and supply in foreign market is x j = y j i j. Costs in each plant consist of production cost, abatement cost, and emission tax. Assuming interior solutions, the Þrst-order conditions for proþt maximization for home Þrm are π H y H = p0 (X) x H + p (X) c 0 y H t =0, (15) π H a H = t g0 a H =0, (16) π H y H = p0 (X ) x H + p (X ) c 0 y H t =0, (17) π H a H = t g 0 a H =0, (18) π H i H = p0 (X) x H + p (X) p 0 (X ) x H p (X )=0. (19) The Þrst-order conditions for foreign Þrm are again derived in an analogous manner. Together these conditions imply that multinational Þrms choose levels of output, abatement, and imports so that marginal production cost equals marginal revenue, marginal abatement cost equals emission tax rate, and marginal revenues from the two markets are equalized. 15

16 As in the case of national Þrms, we denote the optimal output, supply, and abatement levels as a function of the emission tax rates for Þrm j by y j (t), y j (t), x j (t), etc. Due to identical demand conditions in the two countries, we again have that x j (t) =x j (t) for j = H, F and consequently X (t) =X (t). From the Þrst-order conditions we can deduce that Characterization 2 When home government tightens emission control, both multinational Þrms cut back on production in home country and increase production in foreign country. Both Þrms reduce supply in both markets. Emissions decrease in home country and increase in foreign country. Overall emissions decrease. Proof. See Appendix. Comparison with Characterization 1 shows that regardless of whether Þrms are national or multinational aggregate supply in both markets and overall emissions decrease when either of the governments tightens emission control. The difference is that national Þrms react asymmetrically whereas multinational Þrms react symmetrically to any policy change. Assume, for instance, that home government tightens pollution control. If the Þrms are national, foreign Þrm increases and home Þrm reduces supply in both markets. If, in contrast, Þrms are multinational, both Þrms reduce supply in both markets. Characterization 2 implies that the two multinational Þrmsalwaysproduceequal output in their plants located in the same country, i.e. y H (t) =y F (t) and y H (t) = y F (t). Since we know that aggregate supply in the two markets is always equal, it follows that x H (t) =x F (t). Given how Þrms react to changes in taxation we have that π H t (t) = e H (t)+p 0 (X (t)) x F t (t) y H (t)+y H (t) (20) and π F t (t) = e F (t)+p 0 (X (t)) x H t (t) y F (t)+y F (t), (21) 16

17 where the Þrst term is the increased tax burden on each unit of emissions. The second term is the indirect effect of a policy change that comes through supply decisions of the rival Þrm. Whenemissiontaxationbecomesstricterinhome country,both Þrms increase production in their plant located in foreign country but reduce supply in both markets. This implies that using environmental policy to increase market share of the domestic producer is not feasible when Þrms are multinational. When the Þrms are multinational, in each country there is a production plant that belongs to a Þrm that may be largely or even entirely owned by foreign shareholders. Had the national governments the opportunity, they would tax more severely emissions generated by the plant in foreign ownership. Thus the restriction that national governments cannot discriminate between the Þrms, but are instead constrained to set a uniform emission tax rate that applies to all Þrms operating within the country, plays a crucial role. AsinthecaseofnationalÞrms, we adopt the following notation: z H,t = p 0 (X (t)) x F t (t) y H (t)+y H (t), (22) z F,t = p 0 (X (t)) x H t (t) y F (t)+y F (t), (23) z H,t = p 0 (X (t)) x F t (t) y H (t)+y H (t), (24) and z F,t = p 0 (X (t)) x H t (t) y F (t)+y F (t). (25) From the previous discussion it follows that z H,t > 0, z F,t > 0, z H,t > 0, andz F,t > Emission Tax Competition We now analyze the design of emission taxation when the national governments act noncooperatively taking the policy of the other government and the equilibrium behavior of the polluting Þrms as given. Due to the assumption that α = β, thecountriesare again symmetric, and we will mainly focus on the home government. Each government chooses its tax policy to maximize domestic aggregate welfare taking the policy of the 17

18 other government as given. The objective of home government is thus to choose t so as to maximize w = cs (t)+απ H (t)+(1 β) π F (t)+te (t) d (E (t)), and the objective of foreign government is to choose t so as to maximize w = cs (t)+(1 α) π H (t)+βπ F (t)+t E (t) d (E (t)). Taking into account equations (20)-(23), we can write the Þrst-order condition for home government as cs t (t)+αz H,t +(1 β) z F,t +(t d 0 (E (t))) E t (t)+(1 α) e H (t)+βe F (t) =0, where the Þrst term denotes the effect of a tax increase on consumer surplus. The two following terms denote the indirect effect of policy change on the proþts of the two Þrms through supply decisions of the rival Þrm. The fourth term is the welfare-effect of a change in emissions. Finally, (1 α) e H (t)+βe F (t) denotes the increased tax revenue from foreign owners of multinational Þrms. Under the assumption that w tt 0 and wt t 0, theþrst-order conditions again determine a unique non-cooperative equilibrium t n =(t n,t n ). Since the Þrms and countries are identical and the ownership structure of the multinational Þrmsissym- metric, t n = t n. Thus, in equilibrium the output of each Þrmisthesameinthe two plants and emission levels in the two countries are equal, i.e. y H (t n )=y H (t n ) and e H (t n )=e H (t n ). Then by using (22) and (23) and taking into account that x H t (t) =x F t (t) and e H (t) =e F (t), the equilibrium condition for home government can be simpliþed to (t n d 0 (E (t n ))) E t (t n ) p 0 (X (t n )) x F t (tn ) X (t n )+e H (t n )=0. (26) The second term reßects the welfare-effect of a tax increase through market price. This term is negative as the negative effect of a reduction in output on consumer surplus dominates the positive effect on proþts. The last term reßects the effect of 18

19 increased tax revenue from foreign owners of the multinational Þrms and is strictly positive. Condition (26) shows that as long as α = β, the policy problems of the two governments are identical and the non-cooperative outcome does not depend on α and β. As E t (t) < 0, t n = t n T d 0 (E (t n )) = d 0 (E (t n )). (27) Thus we have that Characterization 3 For multinational Þrms non-cooperative tax rates may be higher or lower than the marginal damage of emissions depending on whether the positive effect of increased tax revenue from foreign owners of the multinational Þrms or thenegativeeffect of reduced consumer surplus dominates. Proof. Follows directly from equation (26) and Characterization 2. Equilibrium tax rates may exceed marginal damage of emissions because setting a higher tax rate yields increased tax revenue from foreign shareholders of the multinational Þrms. This effect is not operative when Þrms are national. In that case non-cooperative tax rates are always lower than marginal damage of emissions. 4.3 Coordinated Tax Policies We now turn to the situation where the governments coordinate tax policies in order to maximize joint welfare of the two countries. Under coordination, the governments choose t and t so as to maximize W = cs (t)+cs (t)+π H (t)+π F (t)+te (t) d (E (t)) + t E (t) d (E (t)). Taking into account equations (20)-(23), the Þrst-order condition for maximization of joint welfare for home government can be written as cs t (t)+cs t (t)+ωh,t + ω F,t +(t d 0 (E (t))) E t (t)+(t d 0 (E (t))) E t (t) =0, 19

20 whereagainwehavetheeffect of a higher tax rate in home country on consumer surplus in the two countries, on proþts of the two Þrms, and on tax revenue net of environmental damage in the two countries. As in the previous cases studied, under the assumption that W tt 0 and W t t 0, theþrst-order conditions determine a unique cooperative equilibrium t o =(t o,t o ) with t o = t o. Therefore, y H (t o )=y H (t o ) and e H (t o )=e H (t o ).Thus,theÞrst-order condition for home government can be rewritten as (t o d 0 (E (t o ))) (E t (t o )+E t (to )) = p 0 (X (t o )) x F t (to )+x H t (t o ) X (t o ). (28) Since x H t (t) < 0 and x F t (t) < 0, the right hand side is strictly positive. Furthermore, by Characterization 2, (E t (t)+et (t)) < 0. Therefore, it readily follows that t o = t o <d 0 (E (t o )) = d 0 (E (t o )). (29) While non-cooperative tax rates may be higher than the marginal damage of emissions, coordinated tax rates are always lower than the marginal damage. The reason is thesameaswhenþrms are national. Completely internalizing the negative effects of emissions is not socially beneþcial as it would lead to too large reductions in supply. As in the case of national Þrms, we wish to analyze the relationship of the noncooperative and coordinated outcome. In order to compare the outcome when the governments set their policies in a non-cooperative manner and when they coordinate policies, we again evaluate W t at the non-cooperative equilibrium. By using equation (26) we have that W t evaluated at t n =(t n,t n ) is W t = p 0 (X (t n )) X t (t n ) (1 α) x F t (t n ) βx H t (t n ) X (t n ) (30) +(t n d 0 (E (t n ))) E t (tn ) e H (t n ), where the right hand side reßects all the effects of tighter emission taxation in home country on the welfare of foreign consumers. First, when home government increases emission tax both Þrms reduce supply in foreign market. This increases market price reducing consumer surplus and increasing proþts. The negative effect on consumer 20

21 surplus dominates the positive effect on proþts. This consumer price effect is captured by the Þrst term in (30). Second, when home government increases its emission tax, emissions in foreign country increase. The second term in equation (30) reßects the welfare-effect of increased emissions in foreign country. If t n <d 0 (E (t n )) this effect is negative. Finally, the last term captures the effect of a higher tax rate on foreign owners of multinational Þrms in the form of increased tax burden on each unit of emissions generated in their plant located in home country. This shareholder effect is always negative. Hence, if t n <d 0 (E (t n )), allthesewelfare-effects are negative. However, it can be shown that even if t n >d 0 (E (t n )) the positive effect of increased tax revenue net of environmental damage never dominates the negative effects that a higher tax rate in home country has on welfare in foreign country. Thus we have that Proposition 2 When the two countries are occupied by multinational Þrms, setting emission taxation non-cooperatively leads to excessively severe taxation. Proof. See Appendix. The non-cooperative tax rates are higher than cooperative tax rates that maximize the joint welfare of the two countries. Characterization 2 then implies that also level of emissions in both countries is lower and supply in both markets is smaller in the non-cooperative equilibrium than when decisions on taxation are taken in a coordinated manner. As the multinational Þrms react in the same way to any policy change relaxing environmental control does not increase market share of the domestic Þrm. Furthermore, the existence of a foreign production plant in each country creates an incentive to tax foreign owners of multinational Þrms. To highlight the role of consumer price effect and shareholder effect, let us assume for a moment that both Þrms sell their production to a third country and that proþts of the multinational Þrms are not taken into account by the governments when they choose tax policies. We have that Corollary 1 When proþts of multinational Þrms are not taken into account by national 21

22 governments and Þrms sell their product to a third market, non-cooperative outcome induces too lax emission taxation. Proof. See Appendix. Clearly, when the product is sold to a third market, consumer price effect plays no role. In addition, the shareholder effect is not present when proþts of the multinational Þrms are not taken into account. In this situation national governments of the host countries have an incentive to capture part of the proþts of the multinational Þrms by imposing high emission tax rates. Consequently, the non-cooperative tax rates exceed the marginal damage of emissions. Thus a tax increase by one government has a positive effect on welfare in the other country: Emissions increase in foreign country but the net effect is positive as the negative effect of increased environmental damage is outweighed by the positive effect of increased tax revenue. 5 Conclusions We analyzed environmental tax competition in a two country world with a single polluting industry. We Þrst considered a case of national Þrms where Þrms produce only inonecountryandserveforeignmarketbyexports. Wethenconsideredacasewhere Þrms are multinational and have production plants in both countries. National polluting Þrms react asymmetrically to any policy change by national governments. For instance, tightening emission control in home country induces foreign Þrm to expand supply and home Þrmtocutdownsupplyinbothmarkets. These asymmetric reactions to policy changes enable governments to use environmental policy as a means to increase the market share of domestic Þrm and proþts of domestic shareholders. In order to determine in what way non-cooperative policy design may be inefficient, we studied the effect of an increase in tax rates from the non-cooperative level on welfare in the two countries. Tightening emission taxation in home country affects the welfare of the consumers in foreign country in three ways. First, as home Þrm reduces supply in 22

23 both markets the proþt of foreign Þrm increases but consumer surplus in foreign country decreases. In a symmetric equilibrium these two effects cancel each other. Second, as the foreign Þrm expands output, emissions increase in foreign country. This reduces welfare of foreign consumers. Third, supply expansion by foreign Þrm has a positive effect on consumer surplus in foreign country. We showed that the positive effect on consumer surplus dominates the negative effect of increased pollution in foreign country. Welfare in the two countries could thus be increased by tightening emission taxation in both countries. Non-cooperative design of emission taxation leads to too low emission tax rates and consequently, to too high level of emissions in the two countries. The incentives of national governments for emission taxation change in two ways when Þrms are multinational compared to when they are national. First, unlike national Þrms, the multinational Þrmsreactinthesamewaytoany policychangebythe national governments. Therefore, relaxing emission taxation is not useful in shifting proþts to domestic shareholders as changes in taxation do not inßuence market shares of the Þrms. Second, when polluting Þrms are multinational, in each country there is a production plant that may be largely or entirely in foreign ownership. When setting policies, national governments ignore proþts that accrue to foreign shareholders of these plants and are thus inclined to set too severe emission taxation. As a result, when Þrms are multinational non-cooperative policy design leads to too severe taxation. Relaxing emission taxation from the non-cooperative level in both countries would unambiguously increase welfare in the two countries. References [1] Barnett, A. H. (1980), The Pigouvian Tax Rule under Monopoly, American Economic Review 70(5), [2] Barrett, Scott (1994), Strategic environmental policy and international trade, Journal of Public Economics 54(3),

24 [3] Brander, James A. and Barbara J. Spencer (1985), Export subsidies and international market share rivalry, Journal of International Economics 18(1-2), [4] Conrad, Klaus (1993), Taxes and Subsidies for Pollution Intensive Industries as Trade Policy, Journal of Environmental Economics and Management 25(2), [5] Hoel, Michael (1997), Environmental Policy with Endogenous Plant Location, Scandinavian Journal of Economics 99(2), [6] Kennedy, Peter W. (1994), Equilibrium Pollution Taxes in Open Economies with Imperfect Competition, Journal of Environmental Economics and Management 27(1), [7] Markusen, James R. (1995), The Boundaries of Multinational Enterprises and the Theory of International Trade, Journal of Economic Perspectives 9(2), [8] Markusen, James R., Edward R. Morey, and Nancy Olewiler (1995), Competition in regional environmental policies when the plant locations are endogenous, Journal of Public Economics 56(1), [9] Nannerup, Niels (2001), Equilibrium pollution taxes in a two industry open economy, European Economic Review 45(3), [10] Rauscher, Michael (1994), On Ecological Dumping, Oxford Economic Papers 46(0), [11] Rauscher, Michael (1995), Environmental Regulation and the Location of Polluting Industries, International Tax and Public Finance 2(2), [12] Ulph, Alistair (1996), Environmental Policy and International Trade when Governments and Producers Act Strategically, Journal of Environmental Economics and Management 30(3),

25 [13] UNCTAD (2000), World Investment Report 2000: Cross-border Mergers and Acquisitions and Development, UnitedNations,NewYork. A Appendix Proof of Characterization 1. dy H = 1 p 00 (X) x H +2p 0 (X) c 00 (y H ) dx H dx = p00 (X ) x H +2p 0 (X ) H p 00 (X) x H +2p 0 (X). Totally differentiating (1)-(3) yields < 0, dah = 1 g 00 (a H ) > 0, and Since x H (t) =x H (t) and x F (t) =x F dx (t), H =1. From the above relations and dx H the corresponding equalities for foreign Þrm it follows that de H < dyh < 0, dx H = dxh < 0, and def = dyf > 0, dxf = dx F > 0. Proof of Proposition 1. written as The Þrst-order condition for foreign government can be (t n d 0 (E (t n ))) = p0 (X (t n )) x F t (tn ) X (t n ) Et. (tn ) Substituting this expression for (t n d 0 (E (t n ))) in (14) yields W t = p0 (X (t n )) x F t (tn ) X (t n ) Et (t n ) Et p 0 (X (t n )) x F (tn t ) (tn ) X (t n ) = p 0 (X (t n )) X (t n ) Et xf (tn ) t (t n ) Et (tn ) xf t (t n ) Et, (t n ) where the part before the square brackets is negative as p 0 (X (t n )) < 0 and Et (t n ) > 0. When Þrms are national, E (t) =e F (t) =y F (t) a F (t). Furthermore,sincey F (t) = 2x F (t) it follows that E t (t n ) = 2x F t (t n ) a F t (t) and E t (tn ) = 2x F t (tn ) a F t (t), 25

26 where a F t (t) =0and a F t (t) > 0. Consequently, 0 < xf t (tn ) E t (tn ) < xf t (t n ) E t (t n ) Hence, evaluated at the non- and the term inside the square brackets is negative. cooperative equilibrium W t > 0. Proof of Characterization 2. By totally differentiating (15)-(19) and taking into account Assumption 1 and 2 we have that dy H 1 dah 1 da H = < 0, = > 0, p 00 (X) x H +2p 0 (X) c 00 (y H ) g 00 (a H ) di H dy = p 00 (X) x H +2p 0 (X) < 0, and H p 00 (X) x H +2p 0 (X)+p 00 (X ) x H +2p 0 (X ) dx H = p00 (X ) x H +2p 0 (X ) dx H p 00 (X) x H +2p 0 (X). =0, Since X (t) =X (t) dx H dih =1and dx H dy = 1 H 2. From the above relations it follows that de H < dxh < 0, dy H = de H > 0, dih > 0, and d y H + y H < 0 and d e H + e H < 0. Similar relations hold for foreign Þrm. In particular, we have that de F < dyf < 0, dy F = de F > 0 and d y F + y F < 0 and d e F + e F < 0. As the Þrms are identical y F (t) = y H (t) and y F (t) = y H (t). relations for home and foreign Þrm implies that d (E + E ) < 0, dx < 0, anddx < 0. Using the above 26

27 Proof of Proposition 2. written as The Þrst-order condition for foreign government can be t n d 0 (E (t n )) = p0 (X (t n )) X t (t n ) (1 α) x F t (tn ) βx H t (tn ) X (t n ) e H (t n ) Et (tn ) = p0 (X (t n )) x F t (tn ) X (t n ) e H (t n ) (A1) Et (tn ) where the second equality uses x F t (t) =xh t (t). Let κ = E t (t n ) Et (tn ) = y H t = y H t yt H yt F (tn ) a H t where the last equality follows from a H t that a H t (t n ) a H t (tn ) a H t (t n )+y H (tn )+y F t (t n )+yt F (tn )+yt F t (t n ) (tn ) a F t (tn ), (t n )=a F t (t n ) a F t (t n ) (tn ) a F t (tn ) (tn ) > 0, a F t (tn ) > 0, yt H (t n )=yt F (t n ) > 0, andyt H Furthermore, since the production cost function is convex yt H Consequently, (t n )=0. From Remark 2 it follows (tn )=yt F (tn ) < 0. (t n )+yt H (tn ) < 0. 1 < κ < 0. (A2) Substituting (A1) for t n d 0 (E (t n )) in (30) yields W t = p 0 (X (t n )) x F t (tn ) X (t n )+ p0 (X (t n )) x F t (tn ) X (t n ) e H (t n ) E Et (tn t ) (tn ) e H (t n ) = p 0 (X (t n )) X (t n ) x F t (tn ) x F t (tn ) κ e H (t n )[κ +1], where the second equality uses e H (t n )=e H (t n ). Since p 0 (X (t)) < 0, x F t (t) < 0, and x F t (t) < 0, takingintoaccount(a2)showsthatw t < 0 in the non-cooperative equilibrium. Proof of Corollary 2. When proþts of multinational Þrms are not taken into account and their product is sold to a third market, aggregate welfare in the two countries is w = te (t) d (E (t)) and w = t E (t) d (E (t)). 27

28 The non-cooperative tax rates are determined by Rewriting these equations yields (t n d 0 (E (t n ))) E t (t n )+E (t n ) = 0 and (t n d 0 (E (t n ))) E t (tn )+E (t n ) = 0. (t n d 0 (E (t n ))) = E (tn ) E t (t n ) and (t n d 0 (E (t n ))) = E (t n ) E t (tn ), where the right hand side in both equations is strictly positive since E t (t) < 0 and Et (t) < 0. Thustn >d 0 (E (t n )). As dw = w (tn ) t + w (tn ) t and w(tn ) t =0we have that dw = w (tn ) t =(t n d 0 (E (t n ))) E t (t n ). Since E t (t) > 0 it follows that dw > 0. 28