Foreign Direct Investment and the Choice of Environmental Policy

Transcription

1 Foreign Direct Investment and te Coice o Environmental Policy Ida Ferrara, Paul Missios y and Halis Murat Yildiz z Marc 6, 011 Abstract We use an oligopoly model o intra-industry trade to amine te implications o oreign direct investment or te pollution aven ypotesis and environmental policy. Countries wic lower environmental standards to be more competitive in world markets generate pollution avens i environmental policy is xed. However, i is a viable option as a mode o entry, pro t-siting considerations weaken in avour o environmental considerations and -recipients tigten environmental policy, weakening te pollution aven ypotesis by reducing incentives to relocate production. Interestingly, may still occur in spite o te stricter standards in order to level te playing eld. We derive conditions under wic te -receiving country as an incentive to manipulate its environmental standard to prevent or attract, potentially eliminating or creating pollution avens. Witout manipulation o standards, leads to improvements in world pollution levels. However, wen countries manipulate teir standards, can lead to a dirtier environment i te two countries are substantially di erent in teir valuation o environmental damages. Keywords: Environmental policy, Foreign Direct Investment, Pollution Heaven Hypotesis. JEL Classi cations: F18, Q8, F1 York University, 4700 Keele Street, Nort York, Ontario, Canada M3J1P3. ierrara@yorku.ca. y Ryerson University, 350 Victoria Street, Toronto, ON, Canada M5B K3. pmissios@ryerson.ca. z Ryerson University, 350 Victoria Street, Toronto, ON, Canada M5B K3. yildiz@ryerson.ca. 1

2 1 Introduction As te global economy as become more integrated, ows o oreign direct investment () ave increased signi cantly. In 003, 64,000 multinationals controlled more tan 870,000 oreign a liates worldwide (UNCTAD, 008) and teir sales ceeded $18 trillion (compared to world ports o $8 trillion). Cognizant o tis trend, policymakers and researcers ave ocused on te welare implications o and on identiying economic variables tat are instrumental in decisions. Opponents to international trade and investment ows requently argue tat globalization and te presence o multinationals cause too lax environmental policies and pollution avens to emerge (Newell, 001; Cole et al., 006). According to te Pollution Haven Hypotesis (PHH), di erences in pollution regulation across countries constitute a signiicant determinant o trade patterns and /capital ows as rms in igly pollution-intensive industries ave incentives to relocate teir operations in countries wit less stringent environmental standards. In Eskeland and Harrison (003), te PHH is best seen as a corollary to te teory o comparative advantage: i it is more costly to conorm to more stringent environmental standards at ome, pro t-maximizing rms would want to relocate teir production activities. As noted in Taylor (004), a necessary, altoug not su cient, condition or te PHH is te presence o a pollution aven e ect wic results wen a tigtening o environmental regulation deters ports (or stimulates imports) o dirty goods. Wile support or te latter is provided in several empirical studies, te evidence or or against te ormer (tat is, te PHH) is rater limited. Altoug tere ists a growing body o evidence in support o a signi cant link between te stringency o pollution regulations and te location o oreign direct investment and te size o net trade ows in U.S. manuacturing industries (List and Co, 000; Keller and Levinson, 00; Ederington and Minier, 003), tus suggesting a airly strong response by rms to di erences in environmental regulation, tere is little evidence tat regulatory di erences constitute te most relevant determinant o trade ows as te PHH predicts. Various reasons wy tis is te case rom an empirical viewpoint ave been proposed (e.g., data, measurement o environmental stringency); owever, reasons wy te PHH may ail teoretically ave not received muc attention. Te ocus o isting teoretical and empirical literatures is mostly on te e ects o local environmental policies on investment ows (List and Co, 000; Keller and Levinson, 00; Javorcik and Wei, 004; Xing and Kolstad, 00). Most closely related to our paper is Markusen et al. (1993, 1995). In te 1993 article, a single active regional government in uences te plant location o a single rm wit increasing returns to scale and local pollution; in te 1995 article, te plant location problem is tended to te case in wic bot regional governments are active in policy setting. Several tensions to tis ramework

3 ave been amined, including zero transportation costs (Hoel, 1997). Oter modi cations include policy commitment/time consistency (Ulp and Valentini, 001; Petrakis and Xepapadeas, 003) and asymmetric relocation inormation (Greaker, 003). 1 Our analytical ramework di ers rom tese models in several important ways. First, rater tan amining te impact o environmental policy on te location decision o production, we ocus on te impact o te option or (potential relocation o production) on te coice o environmental policy. Second, instead o a single rm coosing to locate in bot regions (multiplant), one region, or no region, we assume two independent rms producing or (and competing in) te two markets. Tird, we do not rely on increasing returns to scale or sipping costs to in uence te location () decision. Finally, instead o two symmetric regions selecting environmental policy, we rely on te two countries placing di erent weigts on environmental damages to generate environmental policy di erences in te absence o (tat is, to induce a PHH incentive or ). Anoter related paper is Cole et al. (006), in wic a model o political economy wit lobbying and government corruption is employed to plicitly amine te relationsip between and environmental policy. Wile te e ect on environmental policy o an additional (oreign) producer is considered, te entry decision o te oreign rm is ogenous; we, on te oter and, are interested in te coice o and ow tis coice is manipulated troug environmental policy, depending on te ternal bene ts o. Wile te key proposition in Cole et al. tat oreign entry results in stricter environmental policy (wen corruption is low) is con rmed in most o te cases we cover, we encounter cases in wic does not occur but environmental policy is still a ected. In te present paper, we aim at lling te gap in te literature by addressing te ollowing questions: (1) Under wat conditions do di erences in environmental regulation across countries give rms in countries wit more stringent standards incentives to engage in in countries wit less stringent standards (tis question is directly related to te PHH)? () How do tese incentives a ect local environmental policy and welare in recipient countries? (3) Under wat circumstances would an -recipient country coose to tigten its environmental policy in response to? (4) I tis tigtening occurs, would te recipient country ever select an environmental standard tat is tigter tan te standard o te source country (tat is, would te recipient ever become more green tan te originally green country)? (5) I reduces te welare o te recipient country, would te recipient country manipulate its standard to prevent? (6) More generally, under wat conditions would te recipient country adjust its emission standard to attract wen te source country would preer porting and to prevent wen te source country would 1 Altoug te treat is present, relocation never occurs in Greaker (003). 3

4 preer engaging in? (7) Wat is te impact o on te state o te environment worldwide and in te -recipient country wen is eiter induced or prevented and wen it is not? To answer te above questions, we use a two-country oligopoly model o intra-industry trade. As in te perectly competitive model employed in Copeland and Taylor (1994), we assume pollution to be purely local and allow or a tecnology tat abates emissions o pollution. Bot countries ave te same production and abatement tecnologies. In order to amine te implications o, we consider a treestage game. In te rst stage, countries simultaneously decide teir environmental policy, coosing te emission standards or local rms tat maximize welare de ned as te sum o consumer surplus and producer surplus less environmental damages. Eac country takes into account environmental damages wen setting environmental policy, but te two countries di er in te weigt tey assign to environmental damage. Te second stage o te game involves te rm s decision about weter to serve oreign markets troug port or. In tis setup, te rm in te country wit more stringent environmental standards could move production to te country wit more lax standards, depending on te bene ts o suc a move relative to te cost o setting up a oreign plant. 3 Finally, rms engage in Cournot competition in te product markets. In order to maintain te ocus o te analysis on te implications o asymmetric emission standards or oreign direct investment, we ignore trade policy. In te absence o, te country wic places less empasis on environmental damage as an incentive to lower environmental standards to become more competitive in world markets; owever, once is available as a mode o entry, te country is conronted wit two con icting e ects o. On one and, as a positive e ect troug greater local production/consumption associated wit lower domestic prices (iger consumer surplus) and may generate ternal bene ts. On te oter and, as a negative e ect troug lower pro ts or te domestic rm due to a loss in competitive advantage (lower producer surplus) and additional environmental damages rom greater local production. I te ome country assigns a iger weigt on environmental damage tan te oreign country does, te ormer can be considered, in te absence o an option, as te more environmentally riendly or green country wile te latter is te less environmental riendly or grey country. As te ome country becomes more environmentally sensitive (as its weigt on environmental damage increases), it cooses more stringent emission standards wile te oreign country cooses less stringent standards. We rst consider a traditional PHH case in wic te oreign country does not alter its emission standard We consider two similar countries to avoid non-environmental policy related incentives or. Te model is tus best suited or te analysis o te impact o on policy setting among equally industrialized countries rater tan between Nort and Sout. 3 We ignore te possibility o reciprocal and ocus on te decision o te ome rm. 4

5 in response to. Not surprisingly, we sow tat, as long as te xed cost o aving an additional plant is su ciently low and abatement is costly, te ome rm acing a less stringent environmental standard abroad as incentives to relocate its production to te oreign country (te traditional PHH case). We ten allow or te possibility tat te ost country is able to respond to and amine te Subgame Perect Nas Equilibria (SPNE) under various conditions. Wen occurs, te oreign country as two active producers witin its borders, and te relocated ome rm aces te same standard as te oreign rm. Tus, te pro t-siting motive disappears wile te environmental damage e ect widens and te oreign country tigtens its emission standard. Wen te two countries are su ciently similar in weiging environmental damages, te grey country (oreign country) can become greener tan te originally green country (ome country). More interestingly, tere are cases in wic te ome rm cooses to engage in (rater tan to port) in te oreign country even toug te oreign emission standard is stricter tan its own standard under te port case. Intuitively, as countries become more asymmetric in teir environmental riendliness, te gap between teir standards increases under port; as long as te asymmetry is not too large, te ome rm would preer to relocate to te oreign country in order to level te playing eld even i relocation entails acing a tigter standard. 4 Wen allowing te oreign country to respond to te ome rm s, we later consider te question o weter te oreign country can induce port () via preventing (attracting) by adjusting its standard and, i so, weter tis move leads to iger welare in te oreign country and a cleaner environment worldwide. We obtain tat, wen te tra weigt te ome country assigns to environmental damage is low and te bene t te oreign country derives rom te ome country s is ig, te oreign country is better o inducing te ome rm to engage in by increasing its emission standard above te level prevailing under port; wen te tra weigt is ig and te bene t is low, te oreign country is better o inducing te ome rm to port by lowering its standard below te level prevailing under. Relative to te case in wic environmental policy is not adjusted, pollution avens tat would ave isted do not come about wen is prevented wile oter pollution avens tat would ave not isted are generated wen is attracted. From a purely environmental perspective, we sow tat, relative to te port case, te ome rm s results in two counteracting e ects in te oreign country: an emission standard e ect wic amounts to a reduction in emission standards and a scale e ect wic amounts to an increase in output. As te ormer e ect dominates te latter (given te convity o costs/damages), leads to a cleaner environment in 4 Tis is similar to strategic investment to raise a rival s costs, as in Salop and Sce man (1983). Here, te intention is not to increase te rival s abatement cost (it may in act lower it) but to eliminate te gap between te costs o te two rms. 5

6 te oreign country (and tus worldwide) relative to te port case. However, wen te oreign country manipulates standards to eiter prevent or induce, a dirtier world environment can result wen is induced i te two countries do not di er substantially in teir valuation o environmental damage. Relative to te case, a cleaner environment obtains wen port is induced i te two countries di er substantially in teir valuation o environmental damage. Model We develop an oligopoly model o trade wit two countries ( or ome and or oreign) and two goods (x and y). Good y is te numeraire good produced under perect competition wit a constant-returns-toscale tecnology. Tere is no pollution associated wit te production o good y. Good x, te polluting (dirty) good, is produced by a single pro t-maximizing rm in eac country at zero marginal cost. For convenience, we reer to ome (oreign) country s monopolist as rm (). We assume tat preerences over te two goods are quasi-linear and te inverse demand or good x in eac country is linear, tat is, X p i (x i ) = x zi ; (1) z=; were x i denotes te total quantity o good x sold in country i : x i = P z=; x zi., p i denotes te price o good x in country i, and x zi denotes te output sold by country z s rm in country i: Firm i s total production is made o its sales in te domestic market denoted by x ii and in te oreign market denoted by x ij wit i 6= j. Firms compete in quantities (Cournot) in eac market. For simplicity, we assume tat eac unit o x produced generates one unit o pollution and tat, as in Copeland and Taylor (1994), pollution is purely local. Moreover, abatement is possible but costly. Speci cally, i a government imposes a cap on te emissions o rm i, denoted e i, te cost o meeting tis target is C i (a i ) = a i ; () were a represents abatement wic is equal to te di erence between production and te appropriate emission standard or 0 a i = X j=; Environmental damages are quadratic in unabated local emissions and equal to x ij e i 1 A. (3) i = 1 e i : (4) 6

7 To amine te implications o access, we consider a tree-stage game. In te rst stage, countries simultaneously decide over teir environmental policy, coosing te welare-maximizing emission standards or local rms. Welare is de ned as te sum o consumer surplus (CS) and producer surplus (P S) less environmental damages ( ). To di erentiate between te two countries in terms o teir environmental attitude so tat tey select distinct emission standards, we assume tat te ome country places eavier empasis on environmental damages in its welare. 5 Hence, te welare o te ome country is W (e) CS (e) + X j (e) (1 + w) (e) (5) j=; wereas te welare o te oreign country is W (e) CS (e) + X j (e) (e); (6) j=; were e = [e ; e ] is te vector o emission standards, w > 0 captures te additional value tat te ome country places on te environment, and ij denotes te pro t o rm i in country j. Te initial game setup is illustrated in Figure 1a. In te second stage o te game, rm decides weter to serve te oreign market troug port or. In tis setup, te rm in te country wit a more stringent environmental standard could move production to te country wit a more lax standard, depending on te bene t o suc a move relative to te cost o setting up a oreign plant, wic we assume to be xed at F. As pre-commitment and time consistent policies ave previously been compared in Ulp and Valentini (001) and Petrakis and Xepapadeas (003), we assume or te most part te latter (cept wen amining te bencmark o no policy reaction/adjustment). Finally, in te tird stage, rms engage in Cournot competition in te two product markets. We obtain te subgame perect Nas equilibrium (SPNE) by backward induction. 3 Environmental Policy and Welare under Export To nd te Subgame Perect Nas Equilibria (SPNE), it is necessary to determine te payo s to te countries and rms rom and port. To begin, we consider te case were te ome rm ports rater tan relocates. Tis case urter serves as a bencmark to study te implications o or strategic environmental policy and social welare. Since eac rm produces witin its own country, it is subjected to te local emission standard. Hence, rm i aces an endogenously determined emission standard e i and 5 We would ave qualitatively similar results by assuming tat te two countries ave di erent weigts on producer surplus. However, as our ocus is on environmental policy, ow it responds to, and ow it can be manipulated to attract or prevent, we maintain te assumption tat te two countries di er in teir environmental awareness. 7

8 te amount o pollution it abates is so tat its pro t is i = X j=; a i = X j=; 0 1 p j x X x ij j=; x ij e i ; (7) e i 1 A ; i = ; : (8) It is immediate tat te marginal cost o abatement is equal to abatement. We note tat, in order to maintain te ocus o our analysis on te implications o asymmetric emission standards or oreign direct investment, we ignore trade policy. 6 We also ignore local taxation o pro ts, unlike Greaker (003), so pro ts return to country o rm ownersip. Given te emission standards e and e, te pro t-maximizing output coices must ii = 3x ii x ji x ij + e i = 0 ij = 3x ij x jj x ii + e i = 0; i; j = ; : We simultaneously solve te above conditions to obtain te Cournot-Nas equilibrium in te port scenario, namely, x ii = x ij = 3 + 4e i 15 e j and p i = 3 e i e j ; (10) 5 and te ollowing comparative i i = 4 15 > 0 j ij = 1 < 0; j 15 or i; j = ;. Tus, rm i s total output increases wit its own emission standard wile it decreases wit its rival s emission standard. Moreover, te e ect on own output dominates te e ect on rival output so tat total output sold (price) in a country rises (alls) as eiter country weakens its standard, tat is, X i = X zj = j 5 > 0 i i = 1 < 0: j 5 Wit te equilibrium beavior o rms as above described, we nt amine te rst-stage welare maximization problem governments ace to determine te non-cooperative emission standards, tat is, max W (e) = x (e) + X p j (e)x j (e) e j=; 0 X x j (e) j=; e 1 A (1 + w)e 6 Inclusion o tari s would provide countries wit an additional incentive or (i.e., tari jumping) tat would cloud te analysis o decisions resulting rom di erences in environmental policies. (13) 8

9 and max W (e) = x (e) + X p j (e)x j (e) e j=; 0 X x j (e) j=; e 1 A e : (14) Te rst-order conditions or te above problems i = + e W i + e W j ; i; j = ; and i 6= j; (15) were = > 0, and te second-order conditions are satis ed w = w < 0: (16) We tus ave tat e i and e j are strategic substitutes W i = 3 < 0: j 5 Combining te above two rst-order conditions, we obtain te negatively sloped reaction unctions in emission standards, tat is, e i W i + e ; i; j = ; and i 6= j; (18) W i j W W i < 0; i; j = ; and i 6= j: (19) Te negative relationsip between ome environmental policy and oreign environmental policy stems rom te nature o Cournot competition and provides support or te presence o a pro t-siting motive. We also note is independent o w, te absolute value o te slope o te ome country s alls @e < 0; implying tat te coice o te ome country s emission standard becomes less sensitive to te oreign country s coice as te ome country becomes more environmentally conscious (or te additional weigt it places on environmental damages increases). 7 Simultaneously solving te two conditions in (15), we can 7 See appendix. 9

10 press te optimal emission standards in te port scenario (e i ) as! e i W W W W W j ; i; j = ; and i 6= j: (1) W = w < 0, we ave tat e e W W W W W < so tat standards are tigter in te ome country tan in te oreign country. Tus, in te absence o an option, te ome country can be considered as te more environmentally riendly or green country wile te oreign country is te less environmentally riendly or te grey country. Accordingly, as te ome country becomes more environmentally sensitive (as w increases), it cooses more stringent emission standards wile te oreign country cooses less stringent standards, 8 > : (3) Using te optimal emission standards, we can write te pro t unctions o te two rms in terms o emission standards as 6 + 8e (e ; e ) = e (e ; e ) = 15 e e e e : (4) Similarly, welare levels are W (e ; e ) = 1 W (e ; e + e 5 + e {z } CS + e ) = 0 + e {z } CS 6 + 8e e ( + w)e + 15 {z } 6 + 8e + 15 P S e e {z } P S : (5) In te section tat ollows, we consider te case in wic te ome rm (acing a more stringent standard under port) is ree to coose between port and as a mode o entry into te oreign 8 See appendix or more details on emission standards under port, wit accomodation, and environmental policy adjustment to induce or prevent. 10

11 country. For te SPNE, we ten endogenize te ome rm s coice over te mode o entry and te coice o environmental policy o te oreign country. We assume tat te ome country is passive in setting its environmental policy (setting an optimal policy as i tere was a rm present) and receives no ternal costs or bene ts rom te presence or absence o its rm. Te ome rm ten cooses weter to port or engage in given te relocation cost it aces and te environmental policy it will ace in te oreign country. [Insert Figure 1a ere] 4 Foreign Direct Investment occurs i it is pro table or te ome rm to move production to te oreign country in order to take advantage o te iger emission standard in tat country (or to level te playing eld, as we discuss below). By relocating production to te oreign country, te ome rm as to pay an ogenous plantlevel xed cost equal to F. We ignore te possibility o reciprocal and ocus clusively on te decision o te ome rm. In addition, we allow or te possibility tat te -recipient or ost country (i.e., te oreign country) bene ts or su ers rom te ome rm s. We denote te bene t (loss i negative) as B and, or simplicity, assume tat it is ogenously given. In essence, B captures spillover e ects o in te ost country. Altoug standard teory points to -generated ternalities wic raise te productivity o ost actors o production (Glass and Saggi, 1999 and 00), te evidence about te presence o productivity spillovers is rater mixed. Wile a positive industry-level correlation between and productivity is detected in Caves (1974), Blomström (1986), and Dri eld (000), te incidence o spillovers is ound to be in uenced by ost industry s and ost country s caracteristics. At te micro-level, no evidence o iger levels o total actor productivity is ound in sectors wit iger oreign participation in Morocco (Haddad and Harrison, 1993), or Venezuelan manuacturing companies (Aitken and Harrison, 1999), and or low-tecnology Indian companies (Katuria, 1998 and 000). Wen te ome rm engages in, te ome and oreign rms pro ts are = X j=; 0 1 p j x X x j j=; e 1 A F (6) and = X j=; 0 1 p j x X x j j=; e 1 A : (7) 11

12 Given te oreign country s emission standard e, te pro t-maximizing output coices must ii = 3x ii x ji x ij + e = 0 ij = 3x ij x jj x ii + e = 0; i; j = ; and i 6= j: Hence, in equilibrium, rm i s output levels are x ii = x ij = + e 5 ; i; j = ; and i 6= j: (9) To determine te payo s under, we need to amine several cases wic di er in ow te recipient country (oreign country) reacts to. First, we consider te case o no reaction to : te oreign country selects te emission standard above derived or te port scenario regardless o weter occurs. Tis corresponds to a traditional pollution aven ypotesis case (i.e. rms move to countries wit weaker environmental policy witout regard or wat appens to policy i tey move). Ten, we amine te case in wic te oreign country endogenously determines its emission standard in response to ; we tus obtain conditions under wic te ome rm undertakes and discuss te implications o or te oreign country s welare. Finally, we tend te game o Figure 1a to consider weter te oreign country as incentives to manipulate its standard to prevent (attract) wen yields iger (lower) pro ts to te ome rm tan porting and discuss te welare implications o suc a strategy. 4.1 No response to In a typical pollution aven ypotesis case, rms acing weaker environmental standards in oreign countries sit production to tose countries (in our model troug ) witout in uencing local standards. Wit te oreign country coosing te emission standard prevailing under port, pro t maximization by eac rm yields identical output levels as x nr ii (e ) = xnr ij (e ) = + e ; i; j = ; and i 6= j; (30) 5 were te superscript nr reers to no response levels. Eac rm s pro t is nr nr e 8 + e = 50 e 8 = + e 50 17e 17e F (31) and oreign welare is W nr e 1 + e = 50 63e + B: (3) 1

13 We note tat only occurs i te ome rm s pro ts are greater troug tan troug port. Speci cally, i te xed cost o relocating production to te oreign country is su ciently low, te ome rm preers to port as a mode o entry, tat is, nr e e ; e 0 i F F nr ; (33) were wit 8 + e F nr = e 15 e + e ; (34) > 0; (35) or tat te critical value o F (below wic te ome rm engages in ) increases in w. In oter words, as te emission standards o te two countries become more asymmetric (as w increases), te ome rm as greater incentives to engage in in te oreign country. I tere ists no bene t rom, te oreign country strictly preers te ome rm to port rater tan to perorm. Tere are tree distinct e ects o on te oreign country: (i) a decrease in te oreign rm s pro ts rom a loss in competitive advantage; (ii) an increase in environmental damages rom a rise in local production; (iii) an increase in consumer surplus rom a decrease in te price. Te rst two e ects o outweig te last e ect and oreign welare alls i tere ists no ternal bene t rom receiving. Hence, we can always identiy a (positive) critical bene t level, denoted by B nr, above wic te oreign country preers te ome rm to engage in rater tan port, tat is, W nr e W e ; e 0 i B B nr ; (36) were B nr = + e 0 + e 6 + 8e + 15 e 1 + e 50 13e : (37) From an environmental perspective, wenever te ome rm engages in and te oreign country does not cange its environmental policy, environmental damages are iger in te oreign country (and tereore worldwide) relative to te port case Optimal Response to Nt, we consider te case in wic te oreign country, in response to, adjusts its emission standard. I occurs, te oreign country as two active producers witin its borders and te ome rm aces te 9 See appendix or details on critical F and B values. 13

14 same standard set by te oreign country or te oreign rm. Tus, te pro t-siting motive disappears wile te environmental damage e ect widens. As a result, te oreign country as an incentive to lower its emission standard wen aced wit, so tat acts as a disciplining device or governments wising to ploit environmental standards to gain competitive advantage. Stage tree o te game remains te same as in te no response case, so tat output levels are given by (9). Te oreign country, owever, aces twice te environmental damages as in te port case; accordingly, it adjusts its emission standard to = 1 5 = 0; (38) wic yields e F DI = 4 1 (39) as te optimal emission standard in te oreign country under. Upon comparison o te above wit (1), we obtain Proposition 1: Te optimum emission standard in te oreign country is always more stringent wen te oreign country responds to optimally relative to te port and no response cases. Furtermore, te gap between te port and standards in te oreign country widens as te ome country becomes more environmentally conscious: e F DI e < 0 e F DI > 0: (40) Since pollution is local, te oreign country not only takes into account te environmental damages generated by its own rm but also te damages generated by te ome rm. In te absence o any adjustment in te emissions standard, consumer surplus is iger wile producer surplus is lower since te protection o te domestic industry a orded by a weaker emission standard is lost as te ome rm enters and receives te same protection. Moreover, emissions (tus environmental damages) rise signi cantly. Since te oreign country s bene t rom is ogenous, it does not a ect te emission standard. In suc a case, te negative e ects o on producer surplus and environmental damage dominate te positive e ect on consumer surplus. Tus, as we illustrate in Figure, te oreign country as an incentive to reduce its emission standard relative to te port case. More interesting is te comparison o te emission standard in te oreign country (originally te grey country) under, wic bot te oreign and ome rms ace, wit te emission standard in te ome country (originally te green country) in te port scenario, tat is, e F DI e 0 i w w; (41) 14

15 were w = :08. Hence, we obtain Proposition : Wen te oreign country (originally te grey country) responds optimally to te ome rm s, its emission standard alls below te ome country s (originally te green country) standard under port i w is su ciently low. Te above proposition implies tat, wen te ome rm engages in and te two countries weigts on environmental damage are very di erent, te grey country (te oreign country) selects an environmental standard tat is weaker tan te standard o te originally green country (te ome country) under port. However, as we also sow in Figure, te oreign country can become greener tan te green country (te ome country) wen te two countries are su ciently similar in weiging environmental damages. [Insert Figure ere] At ig values o w, te standard in te oreign country is muc iger tan tat in te ome country. Wile generates additional environmental damages and takes away pro t-siting motives, tus inducing te oreign country to lower its standard, tese e ects are not su ciently large to eliminate te policy gap resulting rom a divergence in te two countries environmental positions. However, wen te two countries are similar (i.e., w is su ciently low), te two countries emission standards are not very di erent in te port scenario so tat te damage e ect and weakened pro t-siting e ect under outweig te environmental weigt di erential e ect. From a purely environmental perspective, te ome rm s results in two counteracting e ects in te oreign country: (i) an emission standard e ect, according to wic, wen is accommodated, te oreign country s emission standard alls (even below te green country s standard under port wen w < w); (ii) a scale e ect, according to wic raises te production level in te oreign country relative to te port case and tis, in turn, increases environmental damages or a given standard. Te ormer e ect dominates te latter and leads to a cleaner environment in te oreign country (and tus worldwide) relative to te port case wen te oreign country responds to te ome rm s by lowering its emission standard. Te environmental improvement does depend on w; speci cally, as w increases, te improvement gets larger F DI e F DI i > 0; (4) were e F DI e F DI e = e F DI e + ef DI > 0: (43) 15

16 Using te optimum emission standard e F DI, we nd te rms pro t levels to be F DI F DI e F DI e F DI = 3 F (44) 1 = 3 1 and te oreign welare level to be W F DI e F DI = + B: (45) 7 Again, i te xed cost o relocating production to oreign country is su ciently low, te ome rm preers to port as a mode o entry, tat is, F DI e F DI e ; e 0 i F F F DI ; (46) were F F DI = e 15 e + e 0 i w w; (47) were w 0:43. We tus ave Proposition 3: Te incentive to perorm is increasing in w. For w < w, te ome rm does not engage in. For w > w, te ome rm engages in even i it entails acing tigter standards tan under port at values o w ranging between w and w. As illustrated in Figure 3, te critical F value below wic te ome rm engages in, F F DI, is increasing in w. In oter words, as te emission standards become relatively more asymmetric (as w increases), te ome rm as greater incentives to perorm in te oreign country. Also, as e F DI < e olds, F F DI < F nr always obtains. For su ciently small values o w (w < w), te ome rm does not ave any incentive to perorm in a country wic adjusts its standard optimally and tus F F DI < 0 wen w < w. However, tere are instances in wic te ome rm cooses to engage in even toug te oreign emission standard alls below its own standard under port (since w > w). As w increases, te gap between te two countries standards widens under port. Provided tat w is not too large (w < w), te ome rm preers relocating to te oreign country, even i it ends up acing tigter standards, in order to level te playing eld. Intuitively, i te ome rm does not engage in, its cost o abatement is iger tan tat o its competitor. By relocating to te oreign country, te ome rm aces te same marginal abatement cost as te oreign rm so tat it is no longer at a competitive disadvantage. Wen w is su ciently ig (w > w), te ome rm aces a lower marginal abatement cost in te oreign country under relative to port. Hence, te ome rm as two reasons to engage in : (i) to take advantage 16

17 o lower abatement costs and (ii) to remove te competitive advantage o te oreign rm resulting rom di erences in standards. [Insert Figure 3 ere] Comparing te regions o no response and optimal response in Figure 3, we can see tat some wic would ave occurred in te absence o accommodation is deterred wen environmental standards are optimally adjusted (dark grey saded area). Tis suggests tat optimal response weakens te incentive to relocate production, tereby weakening te PHH (at least in te absence o manipulation o standards to induce, as we sow below). We can easily sow tat, i tere ists no bene t rom, te oreign country strictly preers te ome rm to port rater tan to engage in. Te (positive) critical bene t level, denoted by B F DI, above wic te oreign country is better o under relative to port satis es W F DI e F DI W e ; e 0 i B B F DI ; (48) were + e B F DI = e 6 + 8e + 15 e e 7 : (49) Since te oreign country determines its emission standard optimally, te critical bene t level is smaller tan te one obtained under no response, tat is, B F DI < B nr, as we illustrate in Figure 4. [Insert Figure 4 ere] Tere are two possible SPNE o te game described to tis point, depending on te cost o relocation. I te cost o relocation F is iger tan te critical xed cost or, F F DI (wic takes into account wat te oreign country does wit its standard i te rm relocates), te SPNE consists o te ome rm not engaging in and te oreign country setting its optimal port standard e. However, i te cost o relocation is lower tan te critical F, te SPNE consists o te rm engaging in and te oreign country coosing te optimal standard e F DI. Tese two potential outcomes are saded in Figure 1a. To enric te game, we nt consider te question o weter te oreign country can induce port () via preventing (attracting) by adjusting its standard (to in uence te critical xed cost o te ome rm above or below te actual relocation cost), and, i so, weter tis can lead to iger oreign welare. 10 As we sow in te section tat ollows, tese possibilities arise as SPNE only wen te ternal 10 Industrial policy could be used ere rater tan environmental policy. Since te ternality is environmental, it is not unreasonable to adjust environmental policy to induce or prevent. Furter, national treatment rules o te WTO (Article III) may prevent te use o industrial policy against oreign entrants. 17

18 bene ts or costs o take treme values. In oter words, te SPNE described to tis point are also SPNE o te tended game unless te ternal bene t is below a critical tresold wen F < F F DI (so tat SPNE entails te oreign country tigtening its policy to induce port) or wen te ternal bene t o is above a di erent (iger) tresold wen F > F F DI (so te SPNE entails te oreign country weakening its standard to induce ). 4.3 Adjustment in Emission Standards (Manipulation) In te preceding analysis, te initial xed cost o relocation restricts te oreign country to setting an optimal response to or port but te setup does not permit te country to be more proactive and strategically in uence te mode o entry o te ome rm. Altoug te oreign country sets its policy beore te mode o entry decision is made, te ternal bene ts or costs (being xed) cannot directly in uence te coice o policy and tereore cannot impact te mode o entry. As suc, te game to tis point best describes outcomes wen tere are limited costs or bene ts to or te oreign country. Tis strategic option adds a layer to te game, wic is illustrated in Figure 1b. Here, te oreign country is not required to manipulate its standard but as te option to do so i te bene ts or costs o are suc tat its welare increases. [Insert Figure 1b ere] I te oreign country can tigten or relax its standard to eliminate or generate incentives or te ome rm to engage in, it can select its emission standard suc tat te ome rm is (at most) indi erent between port and. For a given e, te ome rm s pro t under is F DI (e ) = 8 ( + e ) 17e 50 F (50) wile its pro t under port is 6 + (e 8e (e ) ) = 15 e e (e ) : (51) From (15), te ome country s best response to a cange in te oreign emission standard under port is e (e ) = 1 + e ; (5) were 1 = w and = w : (53) 18

19 Using te pressions in (53), we obtain te emission standard in te oreign country (denoted by e ) tat makes te ome rm indi erent between port and, tat is, suc tat F DI e = q (1 ) (1 )[50F ( ) ] (1 )( ) (e ) = (e ), as (1 ) : (54) (1 )( ) Hence, te ome rm strictly preers port to below e wile it preers to port above e, tat is, + F DI (e ) < (e ) wen e < e (55) F DI (e ) > (e ) wen e > e > ; (56) tat is, e is increasing in F and decreasing in w. Intuitively, as w increases (te two countries become more asymmetric in teir valuation o environmental damages), te ome rm s incentive to engage in increases as te gap between te emission standards o te two countries increases, so tat te oreign country can induce or prevent wit a lower standard. To understand te intuition beind te positive e ect o a cange in F on te tresold standard e, we note tat e F DI < e < e or combinations o F and w values at wic te oreign country may opt or environmental policy manipulation to induce or prevent. 11 I te ome rm as no incentive to engage in under optimum standards (F > F F DI ) wile te oreign country preers to port (B > B F DI ), te oreign country must increase its standard above e F DI to make attractive (or, equally, to make port unattractive); ence, e > e F DI. I te ome rm as an incentive to engage in under optimum standards (F < F F DI ) wile te oreign country preers port to (B < B F DI ), te oreign country must lower its standard below e to make port attractive or unattractive; ence, e < e. We tus ave tat, as F increases, te ome rm s incentive to perorm decreases so tat te oreign country as to increase its emission standard above e F DI by more to induce in te rst instance (tat is, wen te ome rm does not want to engage in but te oreign country preers to port) wile it as to decrease it below e by less to induce port in te second instance (tat is, wen te ome rm wants to engage in but te oreign country preers port to ); in bot cases, te tresold standard e increases as F increases. 11 See appendix or more details on te relationsips between e F DI and e and between e and e. 19

20 Wen te oreign country prevents by lowering its standard below e and te ome rm decides to port, te ome country adjusts its emission standard as well. Given tat te emission standards o te two countries are strategic substitutes, a lower emission standard in te oreign country implies a iger emission standard in te ome country, tat is, we tus ave tat e < e =) e = 1 + e > e ; (57) e < e < e < e =) (e e ) < e e ; (58) so tat te gap in environmental policy between te two countries narrows, troug a tigtening o te oreign standard and a weakening o te ome standard, wen port is induced as opposed to being optimally cosen by te ome rm. 1 In te ollowing subsection, we derive te conditions under wic adjusting te emission standard to induce port or is in te interest o te oreign country. To tis end, we compare te oreign country s welare levels in te and port scenarios using te adjusted emission standard above given. We nally amine te environmental implications o adjusting emission standards to induce or prevent rom a global perspective as well as rom te oreign country s perspective Attracting We rst consider te case in wic te ome rm as no incentive to engage in under optimum standards (F > F F DI ). In suc a case, te oreign country can be better o by inducing wit e i te bene t rom is su ciently large, tat is, W F DI (e ) > W e ; e i B > B = + e 0 + e 6 + 8e + 15 e e 3 ( + 7e ) ( 3e ) : (59) 50 It ollows immediately tat te above critical bene t level (B) rises wit F and alls wit w since te oreign country as to make larger adjustments in its emission standard (tis is more costly rom a welare perspective) to induce wen F gets larger or w gets smaller. Also, as W F DI (e ) W F DI e F DI, we always obtain tat B B F DI. In ligt o te above, we ave Proposition 4: Suppose tat F > F F DI olds. Ten, te ollowing SPNE result: (i) wen B < B, te ome rm ports and te ome and oreign countries coose teir optimum emission standards e and e ; 1 See section on emission standards in te appendix. 0

21 (ii) wen B > B, te oreign country induces te ome rm to engage in by adjusting its emission standard to e were e F DI < e < e. Te rst o tese two SPNE corresponds to te SPNE o te previous no manipulation game wen F > F F DI : as long as te ternal bene ts are su ciently small, te oreign country simply accepts tat te ome rm wants to port and cooses te optimal policy or its own rm. Te new SPNE occurs wen tese bene ts are large, so tat wile te xed cost would normally induce te ome rm to port, te oreign country nds iger welare rom strategically weakening its policy to induce te ome rm to relocate. We illustrate te equilibria or F = 0:01 and = 1 in Figure 5. For a given F, we use (47) to obtain te value o w ( bw) suc tat te ome rm preers port or w < bw and or w > bw. As F increases, bw increases so tat te ome rm preers port or a wider range o w values. Hence, in terms o Figure 5, te ome rm preers port or w < 0:7159. However, or B > B, te oreign country preers and tus induces te ome rm to engage in by increasing its standard above e F DI ; or B < B, te oreign country preers port and te optimum port emission standards (e and e ) prevail. Induced results in a Pareto improvement over port: te ome country is better o rom iger producer surplus and consumer surplus and lower environmental damages; te oreign country is better o rom te ternal bene ts o and iger consumer surplus tat o set te additional environmental damages and reduced producer surplus. [Insert Figure 5] From an environmental perspective, wen we compare worldwide environmental damages under port and under induced, we obtain Proposition 5: For a given F, tere ists a w l suc tat F DI (e ) > P and F DI (e ) < P i=; w l is increasing in F. i=; i e ; e or w < w l i e ; e or w > w l, were w l = 0 i F = 0 and w l > 0 i F > 0; urtermore, Wen te ome rm does not ave any incentive to engage in but te oreign country adjusts its standard to induce, te environment is dirtier under relative to port wen te two countries are not substantially di erent in teir valuation o environmental damages unless te xed cost is zero. As te two countries become more asymmetric (w increases), te oreign country can induce wit a smaller upward adjustment in its standard (rom e ) so tat it becomes less likely or te environment to be dirtier under. Conversely, as te cost o te ome rm aces increases, te oreign country 1

22 as to adjust its standard upward by a larger amount in order to induce te ome rm to engage in ; ence, te range o w values or wic te environment is dirtier under widens. In Figure 6, we plot te critical F (F F DI ), as a unction o w, below wic te ome rm preers over port and two isovalues curves. Te isovalue curve labelled D F DI sows te combinations o w and F suc tat te world s environmental damage under induced is actly te same as tat under port. Above te D F DI curve, damages are iger under induced ; below te curve, tey are iger under port. For a given F, te F F DI curve gives te critical w ( bw) suc tat te ome country preers port over or w < bw. I B > B, te oreign country induces by raising its standard above e F DI to e ; te resulting worldwide environmental quality is worse or w < w l and better or w l < w < bw. [Insert Figure 6] Te positive environmental implications o inducing old true, altoug or a smaller range o w values, wen we only consider te oreign country s pollution level. In Figure 7, we ave isovalue curves or oreign pollution level di erences between induced and port (D F DI ) and between and induced port (D ), in addition to te isovalue curves given in Figure 6 (DF DI and D ) and te tresold level o F above wic te ome rm preers port (F F DI ). In Figure 7, we also sow te contour o te set o easible F values or given w values (F F ) as derived in te Appendix. Hence, te pollution level in te oreign country is lower under induced tan under port or w l < w < bw, wit w l < w l implying tat, or w l < w < w l, environmental quality worsens in te oreign country but improves worldwide Preventing [Insert Figure 7 ere] We nt consider te case in wic te ome rm as an incentive to engage in under optimum standards (F < F F DI ). Te oreign country can ten be better o by inducing port wit e i te bene t rom is su ciently small or negative, tat is, W (e ; e ) > W F DI e F DI i + e + e B < B = e 15 e e 7 : (60) Since te oreign country is able to induce port wit a smaller adjustment in its emission standard (tis is less costly rom a welare perspective) as F increases or w decreases, B rises wit F and alls wit w. Also, as W (e ; e ) W e ; e, we always obtain tat B B F DI. We tus ave 13 For F = 0:0, te ome rm may be induced to engage in or w < bw 1:09. I is induced, te pollution level increases in te oreign country or w < w l 1:01 and worldwide or w < w l 0:81. Hence, environmental quality improves in te oreign country or 1:01 < w < 1:09 and worldwide or 0:81 < w < 1:09.

23 Proposition 6: Suppose tat F < F F DI olds. Ten, te ollowing SPNE obtain: (i) wen B > B, te ome rm engages in and te oreign country cooses its optimum emission standard e F DI ; (ii) wen B < B, te oreign country induces te ome rm to port by adjusting its emission standard to e wile te ome country uses e were e < e < e < e F DI < e. Te rst o tese two SPNE corresponds to te SPNE o te no manipulation game rom above wen F < F F DI : as long as te ternal costs o are su ciently small, te oreign country accepts and cooses te optimal policy or te two rms present. Te new SPNE occurs wen tese costs are signi cant, so tat wile te xed cost would normally induce te ome rm to relocate, te oreign country nds iger welare rom strategically strengtening its policy to eliminate te incentive to move. We illustrate te above equilibria in Figure 5 as well or a given positive F and. Te ome rm preers or w > 0:7159. However, or B < B, te oreign country preers port and tus induces te ome rm to port by decreasing its standard below e F DI ; or B > B, te oreign country preers and tus te optimum emission standard under (e F DI ) prevails. Te our possible SPNE are saded in Figure 1b; tere is a unique SPNE or eac pairing o F and B. From an environmental perspective, wen we compare worldwide environmental damages under and induced port, we ave Proposition 7: For a given F, tere ists a w u suc tat F DI > P and F DI e F DI i=; e F DI < P i=; i (e ; e ) or w > w u ; urter, w u is increasing in F. i (e ; e ) or w < w u Wen te ome rm does ave an incentive to engage in but te oreign country adjusts its standard to induce port, te environment is dirtier under relative to port wen te two countries are substantially di erent in teir valuation o environmental damages. As te two countries become more symmetric (w decreases), te oreign country can induce port wit a smaller downward adjustment in its standard (rom e ) so tat it becomes less likely or te environment to be dirtier under. As te cost o te ome rm aces increases, te oreign country as to adjust its standard downward by a smaller amount in order to induce te ome rm to port; ence, te range o w values or wic te environment is dirtier under narrows. In Figure 6, te isovalue curve labelled D gives te combinations o w and F suc tat te world s environmental damage under is actly te same as tat under induced port. Above te D curve, damages are lower under ; below te curve, tey are iger under induced port. For a given F (e.g., 3

24 F = 0:01), te F F DI curve gives te critical w ( bw) suc tat te ome rm preers over port or w > bw. I B < B, te oreign country induces port by lowering its standard below e worldwide environmental quality is worse or bw < w < w u and better or w > w u. to e ; te resulting As in te induced case, we nd tat it is possible or te oreign pollution level under to be lower tan te pollution level under induced port. In terms o Figure 7, we ave tat environmental quality improves bot in te oreign country and worldwide or bw < w < w u but improves only worldwide or w u < w < w u Conclusion An important question tat as largely been ignored in te literature on te relationsip between and environmental policy is about strategic considerations countries entertain in setting teir environmental standards under te treat o increased production rom, particularly wen pollution is local. According to te standard PHH argument, as it relates to, production sits rom countries wit stringent standards to countries wit weaker standards. Altoug tere are many oter actors determining plant location, te idea tat rms would case lower standards is not unreasonable and could lead to a race to te bottom in environmental policy. Surprisingly, te empirical evidence on te PHH is limited and, wile empirical reasons ave been suggested wy te PHH may not old (e.g., data, measurement o environmental stringency), teoretical planations or te possible ailure o te PHH ave not been torougly plored. In tis paper, we attempt to provide one o suc teoretical planations by amining te relationsip between and endogenous standards, and leave te empirical investigation o te implications o or environmental policy to uture work. We also consider only relocation as te preerred mode o entry and do not amine oter metods o entry like joint ventures (see Javorcik and Saggi, 010). Aside rom considering te question o weter and wen di erences in environmental policies trigger rms in countries wit stringent standards to move production to countries wit less stringent standards, we also amine te impact tat as on te global pollution level as well as on te state o te environment in -recipient countries. We sow tat standards can become tigter in te ace o and, altoug unlikely, may even become stricter in te originally grey country tan in te originally green country (tis would appen in instances in wic te two countries are not very di erent in ow tey value environmental damages). Tus, wen 14 For F = 0:0, te ome rm may be induced to engage in port or w > bw 1:09. I port is induced, te pollution level increases in te oreign country or w > w u 1:6 and worldwide or w > w u 1:71. Hence, environmental quality improves in te oreign country or 1:09 < w < 1:6 and worldwide or 1:09 < w < 1:71. 4

25 environmental policy is endogenous, te PHH is weakened by te act tat te tigtening o standards in -recipient countries creates weaker incentives or. Te standard in te grey (-recipient) country is in act always stricter under, wen te country responds optimally to, tan under port. Wen te grey country adjusts its environmental policy in response to, te di erential in environmental standards between te two countries is smaller, or a given gap in environmental riendliness between te two countries, tan it would be witout a reaction to increased domestic production and pollution, and decreases as te two countries become more divergent. In tis way, acts as a disciplining device or countries considering weakening teir environmental standards or competitive gain. At te same time, te strengtening o standards under serves to reduce incentives or relocation o production, so tat less relocation occurs due to di erences in environmental standards. Tis amounts to a weakening o te traditional pollution aven ypotesis: rms may not coose to relocate in response to di erences in environmental policy as, by so doing, tey would ace stricter standards in te new location. We ten derive conditions under wic te oreign country as incentives to manipulate its standard to induce wen te ome rm preers port (wen te xed cost o is above its tresold level) or to induce port wen te ome rm preers (wen te xed cost o is below its tresold level). Wenever manipulation is optimal, te oreign standard is weaker wen is induced tan wen it is not, and stricter wen port is induced tan under port. Te possible manipulation o environmental policy by te oreign country to induce or prevent, depending on weter te bene t rom ceeds or alls sort o its tresold level, as important implications or te overall e ect o on te state o te environment. In act, wile always results in a cleaner environment in te oreign country (and tus worldwide) in te absence o manipulation as te emission standard e ect (tigtening o oreign standard) always dominates te scale e ect (increase in production in te oreign country) due to te convity o te damage unction, weter yields a cleaner or dirtier environment in te presence o manipulation depends on ow di erent te two countries are in teir valuation o te environment. Speci cally, worsens (improves) te environment wen te two countries are quite similar (di erent) and is induced or wen te two countries are quite di erent (similar) and is not induced. Tat di erences in environmental awareness play an important role in ow a ects te environment stems rom te act tat incentives to engage in increase wit te ome country s environmental awareness over and above tat o te oreign country; te more di erent te two countries are, te larger te gap in teir environmental standards and te greater te bene ts te ome rm can derive rom. Wen is induced, te oreign standard is less stringent tan in te absence o manipulation but te gap 5

26 between te two standards decreases as te ome rm s incentives or increase (tat is, as te ome country becomes more environmentally conscious); ence, te more similar te two countries are, te larger te gap and te dirtier te environment. On te oter and, wen is not induced, te oreign standard is more stringent tan under port but te gap between te oreign standard under and te oreign standard under induced port decreases as te ome country becomes more environmentally conscious (or te incentives or increase); ence, te more di erent te two countries are, te smaller te gap and te dirtier te environment. Overall, wenever occurs wit or witout inducement, pollution is likely to decrease in -recipient countries and worldwide wenever tere ist neiter trivial nor substantial di erences in environmental attitude between te -recipient and te -source countries. As te cost o engaging in increases, pollution is less likely to decrease under, particularly in less divergent countries; in oter words, te more costly is, te less similar te two countries ave to be in teir environmental awareness or pollution to decrease under. Te presence o ternal bene ts o as tus additional implications or te pollution aven ypotesis: i provides signi cant ternal bene ts to te recipient nation, we would pect to see more PHH-induced relocation o production as countries weaken standards to attract investment (ligt grey area in Figure 5); i te spillover e ects o are small or negative, countries may tigten standards to prevent PHH-driven relocation o production (dark grey area in Figure 5). Equilibrium outcomes ultimately depend on te xed costs o relocation and ow di erently countries weig environmental damages. Noneteless, we sow tat, in most cases, results in a tigtening o environmental policy and is likely to improve te quality o te world environment (at least among countries tat are similar in production and abatement tecnology tat account or environmental damages wen setting policy). 6 Appendix 6.1 Emission standards Using te rst-order conditions or welare maximization wit respect to te standards, we can write te two countries reaction unctions as and e = 114 3e w e = 114 3e ; 313 6

27 we tus ave tat te absolute value o te slope o te ome country s reaction unction is decreasing in w, = 3 (5) ( w) < 0; wile te slope o te oreign country s reaction unction is independent o w. 1 Letting = 11+75w, = w, and = w, we can press te port standards as e = 04 and e = 38( w); so tat e e = 1710w > 0; and te marginal e ects o w on te = 4695e < = 345e > 0: Te oreign standard prevailing under in te absence o any response by te oreign country is te same as tat under port, tat is, e. Wit accommodation, tat is, wen te oreign country adjusts its standard in response to, e F DI = 4 1 < e, wit te di erence in standards increasing in w. We in act ave tat e F DI 90 e = w < 0 and Furtermore, wic is positive or w > w = 1015 e e F e F DI = > 0: 145 e = w Under environmental policy adjustment by te oreign country to induce or prevent, we ave tat e = w w 3 5( w)M and e = [ ( w) + 5M] ; were M = p r 450 ( w) 3 F > 0; 7

28 wic gives te range o easible F values or te system to yield a real solution as F < F = ( w) ; ence, we ave tat We can ten readily see tat e e = 15 [( w) M] > = 5 3 (313 > as < @F < 0. Te e ect o w on e is not as straigtorward; owever, noting = ( w) + 115( w)F 1 M 3 < = 850 < 0; F =0 we can Wen comparing e wit e < 0 or F > 0. and e wit e, we ave tat e e = e e = 0 or F = F ind, were e e < 0 or F < F ind F ind = 34 ( w) w = F nr ; e > 0, and e e > 0 or F < F e ) < 0: We ten obtain tat, or F < F F DI < Find, tat is, wenever port may ave to be induced as te ome rm would engage in, e e < 0 and e e > 0. As e < e < e < e, te gap in standards under port is narrower wen port is induced tan wen it is optimally cosen by te ome rm, tat is, e e < e e or F < F F DI. Wen we compare e wit e F DI, we ave tat e e F DI = 0 or F = Find F DI; were and e obtain tat F F DI e F DI Find F DI = 50 ( w) ( 7 + 6w) 441 ( w) ; > 0 or F > Find F DI given e F DI Find F DI = w > 0. Upon comparison o F F DI ind w w 3 " e wit F F DI, we e F DI ( w) # > 0; ence, or any F > F F DI > Find F DI, tat is, wenever may ave to be induced as te ome rm would not engage in, e e F DI > 0. 8

29 6. Critical F and B values Upon comparison o F F DI wit F nr, tat is, te tresold levels o F below wic te ome rm engages in wen te oreign country adjusts its standard in response to and wen it does not, we obtain tat 90 F nr F F DI = w w > 0; 7 were F nr = 34 ( w) w > 0 i w 0 and F F DI = w w > 0 i w > 0:43; 49 urtermore, we know tat F F nr = w w > 0; were F is te largest easible value o F, as above de ned, wen we consider te possibility o policy manipulation to induce or prevent. We tus derive te marginal e ects o w on te two critical F values as well as teir di erence nr F = 18 ( w) e > 0; = 60 ( w) e > 0; and F nr F F = 138 ( w) e > 0; Similarly, upon comparison o B F DI wit B nr, tat is, te tresold levels o B above wic te oreign country is better o wit wen it adjusts its standard in response to and wen it does not, we ave tat B nr B F DI = 7 ( w) > 0; were B nr = w + 43w > 0 and B F DI = w w > 0: 9

30 Hence, we can press te marginal e ects o w on te two critical B values and teir di erence = 6 ( w) e > 0; and B nr B F = ( w) e > 0; = 414 ( w) e > 0; Reerences [1] Aitken, B. and A. Harrison (1999), Do Domestic Firms Bene t rom Foreign Investment? Evidence rom Venezuela, American Economic Review, 89(3), [] Antweiler, W., B. R. Copeland, and M. S. Taylor (001), Is Free Trade Good or te Environment?, American Economic Review 91(4), [3] Blomström, M. (1986), Foreign Investment and Productive E ciency: Te Case o Mico, Journal o Industrial Economics, 35(1), [4] Bommer, R. (1999), Environmental Policy and Industrial Competitiveness: Te Pollution Haven Hypotesis Reconsidered, Review o International Economics, 7(), [5] Caves, R.E. (1974), Multinational Firms, Competition and Productivity in Host-Country Markets, Economica, 41(16), [6] Cole, Mattew A., Robert J. R. Elliott, Per G. Fredriksson (006), Endogenous Pollution Havens: Does In uence Environmental Regulations?, Scandinavian Journal o Economics, 108(1), [7] Copeland, B. R., and M. S. Taylor (1994), Nort-Sout Trade and te Environment, Quarterly Journal o Economics, 109(3), [8] Donneneld, S. (003), Regional Blocs and Foreign Direct Investment, Review o International Economics, 11(5), [9] Dri eld, Nigel (001), Te Impact o Domestic Productivity o Inward Investment in te UK, Mancester Scool, University o Mancester, 69(1),

31 [10] Ederington, J., and J. Minier (003), Is Environmental Policy a Secondary Trade Barrier? An Empirical Analysis, Canadian Journal o Economics, 36(1), [11] Ederington, J., A. Levinson, and J. Minier (004), Trade Liberalization and Pollution Havens, Advances in Economic Analysis and Policy, 4(), Article 6. [1] Eskeland, G., and A. Harrison (003), Moving to Greener Pastures? Multinationals and te Pollution Haven Hypotesis, Journal o Development Economics, 70(1), 1-3. [13] Glass, Amy, and Kamal Saggi (1999), Policies under Sared Factor Markets, Journal o International Economics, 49(), [14] Glass, Amy, and Kamal Saggi (00), Multinational Firms and Tecnology Transer, Scandinavian Journal o Economics, 104(4), [15] Greaker, M. (003), Strategic environmental policy wen te governments are treatened by relocation, Resource and Energy Economics 5, [16] Haddad, M., and A. Harrison (1993), Are tere Positive Spillovers rom Direct Foreign Investment? Evidence rom Panel Data or Morocco, Journal o Development Economics, 4(1), [17] Henderson, D. J., and D. L. Millimet (007), Pollution Abatement Costs and Foreign Direct Investment In ows to U.S. States: A Nonparametric Reassessment, Review o Economics and Statistics, 89(1), [18] Hoel, M. (1997), Environmental policy wit endogenous plant locations, Scandinavian Journal o Economics 99, [19] Ja e, A. B., S. R. Peterson, P. R. Portney, and R. N. Stavins (1995), Environmental Regulations and te Competitiveness o U.S. Manuacturing: Wat does te Evidence Tell Us?, Journal o Economic Literature, XXXIII, [0] Jaumotte, F. (004), Foreign direct investment and regional trade agreements: Te market size e ect revisited, International Monetary Fund Working Paper 04/06. [1] Javorcik, B. S. and K. Saggi (010), Tecnological Asymmetry Among Foreign Investors And Mode O Entry, Economic Inquiry 48(),

32 [] Javorcik, B. S., and S-J. Wei (005), Pollution Havens and Foreign Direct Investment: Dirty Secret or Popular Myt?, Contributions in Economic Analysis and Policy, 3(), Article 8. [3] Kan, M. E., and Y. Yosino (004), Testing or Pollution Havens Inside and Outside o Regional Trading Blocs, Advances in Economic Analysis and Policy, 4(), Article 4. [4] Kayalica, M. O., and S. Lairi (005), Strategic Environmental Policies in te Presence o Foreign Direct Investment, Environmental and Resource Economics, 30, 1-1. [5] Katuria, V. (1998), Foreign Firms and Tecnology Transer Knowledge Spillovers to Indian Manuacturing Firm, INTECH Discussion Paper Series No. 9804, United Nations University, January. [6] Katuria, V. (000), Productivity Spillovers rom Tecnology Transer to Indian Manuacturing Firms, Journal o International Development, 1(3), [7] Keller, W., and A. Levinson (00), Pollution Abatement Costs and Foreign Direct Investment In ows to U.S. States, Review o Economics and Statistics, 84(4), [8] Levinson, A. (000), Te missing pollution aven e ect: amining some common planations, Environmental and Resource Economics, 15, [9] List, J. A., and C. Co (000), Te E ects o Environmental Regulations on Foreign Direct Investment, Journal o Environmental Economics and Management, 40(1), 1-0. [30] Markusen, J. R., E. R. Morey, and N. D. Olewiler (1993), Environmental Policy wen Market Structure and Plant Locations are Endogenous, Journal o Environmental Economics and Management, 4(1), [31] Markusen, J. R., E. R. Morey, and N. D. Olewiler (1995), Competition in regional environmental policies wen plant locations are endogenous, Journal o Public Economics, 56(1), [3] Newell, P. (001), Managing Multinationals: Te Governance o Investment or te Environment, Journal o International Development, 13, [33] Ornelas, E. (005), Rent Destruction and te Political Viability o Free Trade Agreements, Quarterly Journal o Economics, 10(4), [34] Petrakis, E. and A. Xepapadeas (003), Location decisions o a polluting rm and te time consistency o environmental policy, Resource and Energy Economics 5,

33 [35] Salop, C. S. and T. D. Sce man (1983), Raising rivals cost: Recent advances in te teory o industrial structure, American Economic Review 73, [36] Taylor, M. S. (004), Unbundling te Pollution Haven Hypotesis, Advances in Economic Analysis and Policy, 4(), Article 8. [37] Ulp, A., and L. Valentini (1997), Plant location and strategic environmental policy wit intersectoral linkages, Resources and Energy Economics, 19, [38] UNCTAD (008), World Investment Report 008: Transnational Corporations and te Inrastructure Callenge, New York and Geneva, United Nations. [39] Wu, X. (004), Pollution Havens and te Regulation o Multinationals wit Asymmetric Inormation, Contributions in Economic Analysis and Policy, 3(), Article 1. [40] Xing, Y., and C. Kolstad (00), Do Lax Environmental Regulations Attract Foreign Investment?, Environmental and Resource Economics, 1(1), 1-. [41] Yu, Z. (00), A Teory o Strategic Vertical DFI and te Missing Pollution-Haven E ect, FEEM Working Paper

34 Nature F F F F Home Firm Foreign Country port response no response π e e W e e π e W e Home Firm Home Firm port port π e e W e e π e W e π e e W e e π e W e Figure 1a: Tree diagram o te game between te oreign country and te ome irm under te assumption tat te oreign country does not manipulate its standard to induce irms to perorm wen te cost o relocation is ig or induce irms to port wen te cost o relocation is low. Saded boxes represent SPNE depending on.

35 Nature F F F F Foreign Country Foreign Country manipulation no manipulation manipulation no manipulation Home Firm Home Firm Home Firm Foreign Country port port port response no response π e e W e e π e W e π e e W e e π e W e π e e W e e π e W e Home Firm Home Firm port SPNE depending on (oreign country preers wen π e e W e e π e W e port SPNE depending on (oreign country preers port wen π e e W e e π e W e Figure 1b: Tree diagram o te game between te oreign country and te ome irm wen oreign country as te option to manipulate standards to induce or deter. Te oreign country manipulates its standard to induce wen and and to induce port wen and, but does not manipulate standards wen and or wen and.

36 e i p e e p e w w e p e Figure : Emission standards under port and or te ome and oreign irms ( F reduction in PHH area F nr F e e but ome irm preers to port w w w Figure 3: Critical values and mode o entry o te ome irm (

37 B B nr B w Figure 4: Critical beneit levels ( B increase in PHH area F HF: FC: e F B F HF: ; FC: e F B F HF: port FC: e ; HC: e F F HF: p; FC: e; HC: e F reduction in PHH area w Figure 5: Possible equilibria ( w 7 59