International Agreements on Product Standards under Consumption Externalities: National Treatment versus Mutual Recognition

Transcription

1 International Agreements on Product Standards under Consumption Externalities: National Treatment versus Mutual Recognition Difei Geng April, 2018 Abstract This paper provides a comparative analysis of product standards agreements between countries with di erent levels of economic development. A simple North-South model of product standards is developed where countries have heterogeneous preferences for a negative or positive consumption externality. I compare two major types of standards agreements, those based on national treatment (NT) and mutual recognition (MR). Unlike NT, MR can induce a mismatch of standards between developed and developing countries, a problem that tends to get worse as country preferences diverge. Due to this mismatch problem NT tends to become relatively more welfare-enhancing than MR for countries with more dissimilar preferences. These ndings explain why the WTO, the TPP and the EU choose di erent types of standards agreements. The paper also sheds new light on the desirability of international harmonization of product standards between countries at di erent stages of development. Keywords: Product standards, North-South model, Trade agreements, National treatment, Mutual recognition, Welfare. JEL Classi cations: F13, F18, O24. Department of Economics, Sam M. Walton College of Business, University of Arkansas - Fayetteville, dg017@uark.edu. 1

2 1 Introduction Over the past six decades or so, signi cant tari reductions worldwide have brought increased attention to various types of non-tari barriers to international trade. 1 Particularly, much attention has been centered around product standards, which arise from products having to meet speci c requirements in order to be authorized for sale in a given country. important feature of international agreements on product standards is that they can be reached between countries of varying levels of economic development. For instance, there are in uential product standards agreements signed between developed and developing countries, such as the Agreement on Technical Barriers to Trade under the World Trade Organization (WTO) and the rules about product standards laid down by the Trans-Paci c Partnership (TPP). 2 On the other hand, product standards agreements such as the one enforced by the European Union (EU) was formulated predominantly between developed countries. 3 It turns out that these product standards agreements also have distinct foundations. While the WTO and the TPP s agreements are based on the principle of national treatment (NT), the approach adopted by the EU rests on the rule of mutual recognition (MR). In a nutshell, NT mandates that product standards imposed on foreign rms should be no stricter than those on domestic rms i.e. NT is essentially a non-discrimination requirement. On the other hand, MR requires that a country s standards on foreign rms be the same as what those rms have to comply with in their home countries. By design NT and MR are not compatible with each other: a country under NT has full control over the standards applied in its domestic market, while under MR standards on foreign rms are determined by their own countries. This salient distinction sparks several important questions: Why have the WTO, the TPP and the EU chosen di erent principles for their product standards agreements? Are these choices well-founded from a welfare perspective? More broadly, 1 As Baldwin (1970) vividly notes: The lowering of tari s has, in e ect, been like draining a swamp. The lower water level has revealed all the snags and stumps of nontari barriers that still have to be cleared away. For a most recent comprehensive discussion of non-tari barriers, see Ederington and Ruta (2016). 2 It is worth noting that the U.S. has pulled out of the TPP under the new Presidential administration so the future of the trade agreement is uncertain at best. 3 See the Council Resolution 85/C 136/01 by the EU on a new approach to technical harmonization and standardization. The document is available online at An 2

3 what are the determinants of the relative e cacy of alternative policy regimes on product standards? This paper shows that heterogeneity in levels of development between countries can be key in answering these questions. It is well-known that the WTO and the TPP involve both developed and developing countries whose economic characteristics such as per capita income varies substantially. Hence their member countries may exhibit quite di erent preferences for product standards, which is exempli ed by the observation that standards are typically stricter in more developed countries. On the other hand, the EU consists predominantly of developed countries with relatively homogeneous preferences for product standards. Motivated by these observations, I show that as di erences in levels of national development increase NT tends to perform relatively better than MR from the welfare point of view. This nding o ers a novel explanation for why the WTO and the TPP with more heterogeneous member states choose NT while the EU prefers MR as the foundations of their product standards agreements. Moreover, it suggests that the choices made by these trade blocs regarding the type of standards agreement indeed aim to maximize welfare. In a broader sense, this paper demonstrates how country heterogeneity can have important implications for the optimal design of international trade agreements. Speci cally, I consider a North-South model where countries at di erent stages of development have heterogeneous consumer preferences for a negative or positive consumption externality. An example of negative consumption externality is pollution generated by the use of goods such as motor vehicles and pesticides. On the other hand, consumption of pharmaceutical products that deal with contagious diseases can generate a positive externality by making other people less likely to be infected. I assume that the North is a ected by such consumption externalities to a larger degree than the South. This assumption is empirically relevant. For instance, in the case of negative consumption externality the well-known environmental Kuznets curve suggests that developing countries tend to attach less importance to the negative impacts of pollution as compared to developed countries. When the level of externality represents product quality, the country heterogeneity I examine also re ects non-homothetic preferences such that consumers in developed countries with high per capita incomes prefer better quality products, a pattern that has received substantial empirical 3

4 support from recent trade literature. 4 I consider a model of vertical (or quality) standard where a standard stipulates the magnitude of consumption externality that a good can produce. A higher or stricter standard allows for less negative or greater positive externality. The timing of decision making in the model is as follows. First, countries enter into a product standards agreement based on either NT or MR. Second, countries set their product standards abiding by NT or MR. Third, rms produce under the mandatory standards, and trade and consumption take place. Two scenarios are considered depending on the presence of strategic incentives. In the rst scenario countries coordinate their product standards to maximize joint welfare, while in the second they non-cooperatively choose their standards to maximize each s own welfare. The rst scenario is important for two reasons. First, by shutting down the e ects of strategic interactions, I am able to identify a possible mismatch of product standards between the North and the South that arises from the implementation of MR, and such a mismatch problem would get worse as country heterogeneity rises. This turns out to be the key driving force for NT to perform relatively better than MR between more heterogeneous countries. Second, it allows me to establish the socially e cient outcome constrained by NT or MR, which is the benchmark needed for analyzing the e ciency loss caused by the strategic interactions between countries. To see this, note that the e ects of strategic interactions can be identi ed by comparing the non-cooperative Nash equilibrium with the coordination outcome. Since the former is constrained by NT or MR, it is necessary to compare it with the e cient outcome that is also constrained by these two principles. That said, I also consider coordination under no constraints to study the potential e ects of imposing NT and MR. Under policy coordination, I show that both the North and the South choose the high (low) standard when externality is su ciently large (small) regardless of NT or MR. When externality is moderate, the South chooses the low standard while the North enforces the high standard in the presence of NT. However, this is not the case under MR: by mandating countries to recognize each other s standards, MR can give rise to a possible mismatch of standards between countries when they actually favor di erent standards. As a result, 4 See, for example, Hallak (2006), Feenstra and Romalis (2014) and Caron et al. (2015). Speci cally, these and other empirical studies have identi ed per-capita income as an important factor determining the quality of goods imported and exported by countries. 4

5 the North and the South always choose identical standards despite heterogeneous country preferences. Such a mismatch problem does not occur under NT because countries have full control over the standards implemented in their own markets. It follows that joint welfare under NT is weakly higher than that under MR for all levels of externality but is strictly higher when externality is moderate. Furthermore, as preference heterogeneity increases between the North and the South NT strictly dominates MR over a larger range of externality, indicating that the mismatch problem under MR tends to get worse as the gap in development between countries increases. The next main nding is that NT would become relatively more e ective than MR with greater heterogeneity in the level of economic development between the North and the South. I de ne the e ectiveness of NT relative to MR as the ratio between the ranges of externality in which joint welfare is higher under NT and MR respectively. Intuitively, a higher value of this ratio indicates that NT dominates MR over a larger portion of the range of externality and thus should be considered as relatively more preferable from the welfare point of view. I rst show that when countries non-cooperatively set their product standards NT does not always weakly dominate MR. Particularly, with strategic incentives countries tend to overuse the high standard under NT as compared to the social optimum. This is because NT mandates the same standard to be implemented on all rms, so if a country lowers the standard for domestic rms it has to extend it to foreign rms as well. As a result, lowering the domestic standard would generate a positive pro t externality on foreign rms and this is why the high standard is overly used under NT. On the other hand, countries use too much of the low standard under MR since lowering the domestic standard would confer a negative pro t externality on foreign rms by allowing home rms to enjoy reduced production cost in both countries. Given this observation, I show that the e ectiveness of NT relative to MR increases as countries are at more di erential stages of development. The intuition is straightforward: as country preference heterogeneity rises the mismatch of standards under MR tends to induce a larger welfare loss relative to NT. One interesting implication of the above ndings is that they provide an empirically relevant explanation about why the WTO, the TPP and the EU prefer di erent types of product standards agreements. It is well-known that MR may help reduce adaptation cost 5

6 for rms that sell to countries implementing di erent standards, so why do the WTO and the TPP not adopt such a policy regime? The explanation suggested by my analysis is that levels of development within the WTO and the TPP are considerably heterogeneous such that the welfare loss caused by MR due to the mismatch of standards may outweigh the potential bene ts. On the other hand, the reduction in adaptation cost can prevail for the EU of which countries are have similar preferences for product standards. This is indeed consistent with the observed choices made by these organizations regarding the types of their product standards agreements. In this sense, my analysis o ers a new insight about the role country heterogeneity plays in the design of international trade agreement. Importantly, the paper also provides new insights into international harmonization of product standards which is a pillar of the WTO s Agreement on Technical Barriers to Trade. Speci cally, the paper shows that for goods with su ciently large or small levels of consumption externality, harmonizing standards even across heterogeneous countries is necessary for global optimality regardless of NT or MR. In other words, the North and the South could always share common interests in setting product standards and such policy alignment can indeed be socially desirable. Moreover, in such cases countries could harmonize their standards even without policy coordination, indicating that Nash equilibrium and e cient outcome may coincide. This is quite surprising given that setting national product standards creates an international pro t externality on foreign countries. On the other hand, harmonization over all types of goods can be suboptimal, since when externality is moderate di erential standards should be implemented in light of heterogeneous country preferences. Overall, my analysis provides a more subtle view about the e ciency implications of harmonizing product standards across countries. Finally, I show that a central nding in Costinot (2008) may not hold under heterogeneous countries. By comparing NT and MR between identical countries, Costinot (2008) shows that NT (MR) induces higher global welfare for relatively high (low) consumption externality. However, I show that in a North-South world the reverse of the pattern as in Costinot (2008) can arise: NT (MR) can dominate for relatively low (high) levels of externality. The intuition is that as externality reduces, the North and the South will lower their standards sequentially, so that the underuse (overuse) of the low standard under NT (MR) occurs twice, once for 6

7 the South and the North respectively. This may duplicate the region over which NT and MR dominates each other. As a result MR could dominate NT for higher levels of externality. An important policy implication is that the degree of country heterogeneity should be taken into account if application of NT and MR is contingent on the magnitude of consumption externality. The rest of the paper proceeds as follows. Section 2 reviews related economic literature. Section 3 describes the model. In Section 4, I compare NT and MR assuming countries can coordinate policies on product standards, while in Section 5 I investigate how the comparative assessment may be altered when countries strategically choose their standards. Section 6 provides further discussions and Section 7 concludes. I collect all proofs in the appendix. 2 Related Literature To my best knowledge this is the rst study that analyzes product standards agreements in a North-South world. The paper that is most related to my work is Costinot (2008), who compares alternative types of product standards agreements between identical countries and therefore does not investigate the role of country heterogeneity. Another related study is Markusen (2017) who takes the perspective of asymmetric country preferences to examine government s incentives for setting environmental standards. Similar to the current study, the paper analyzes both cooperative and non-cooperative policies albeit the focus is not institutional arrangements such as NT and MR. Sara and Saggi (2008) examine the implications of country heterogeneity for trade agreement but their focus is NT in the context of government taxation. 5 This paper also relates to the theoretical studies that examine the welfare implications of a particular type of product standards agreement. Battigalli and Maggi (2003) provide an incomplete contract explanation about why NT has been taken as a rigid rule in the WTO. 5 Sara and Saggi (2008) relates to a small literature that examines the principle of NT in di erent policy contexts. Their paper extends Horn (2006) who studies the welfare implications of the NT clause in the GATT under symmetric countries. Ferrara et al. (2015) examine NT in the context of government taxation and goods with consumption externality. Geng and Saggi (2015) evaluate the case for implementing NT in the international protection of intellectual property. While these studies compare scenarios with and without NT, I compare NT with MR as these are extant regulatory regimes of product standards that obligate a dominant share of the countries in the world. 7

8 More recently, Staiger and Sykes (2011) apply the terms-of-trade framework to analyze the e ect of NT on the regulation of product standards. The paper shows that the terms-of-trade e ect gives rise to incentives for large countries to impose discriminatory standards that are overly stringent on foreign rms. Moreover, prohibiting such discrimination through NT may still lead to ine ciently stringent standards under certain circumstances. Toulemonde (2013) provides a welfare analysis of MR under di erential learning ability of consumers about foreign norms. The analysis identi es the winners and the losers (i.e. rms versus consumers) due to the adoption of an MR agreement. The current paper di ers from these studies by highlighting countries of heterogeneous economic development and by comparing NT and MR as alternative principles of product standards agreements. 6 There also exists an established literature that examines the strategic incentives of countries in choosing product standards. For quality standards, Barrett (1994) analyzes and emphasizes the strategic use of environmental standards between countries whose rms compete on the world market. Boom (1995) examines the e ect of asymmetric standards on market outcomes in a vertical di erentiation model. Fischer and Serra (2000) study the incentives for a country to impose minimum standards when its own rm faces foreign competition in the home market. For compatibility standards, Gandal and Shy (2001) examine a country s decision to recognize foreign standards and characterize conditions under which countries are willing to form a standards union. Klimenko (2009) models strategic interaction of national product standards in the context of technical compatibility. While my analysis also captures strategic incentives, it di ers from the above studies by placing a clear-cut focus on both country heterogeneity and the institutional aspect of product standards. 3 Model Consider a world of two countries: the North (N) and the South (S). Each country has one rm that sells a homogeneous good with two possible versions: H and L. Consumption of each unit of version L generates a negative externality of a magnitude > 0, while this 6 Other related studies include Bagwell and Staiger (2001) and Suwa-Eisenmann and Verdier (2002). For a comprehensive treatment of the e ect of TBT liberalization, see Baldwin (2000). 8

9 externality for version H is zero. 7 Section 6 discusses how the analysis also applies to positive consumption externality. The unit costs of version H and L are denoted with c > 0 and zero respectively so that version H is more expensive to produce. 8 One example of the negative consumption externality is pollution so that H (L) can be considered as the clean (dirty) version of a good. For ease of exposition I will refer to the negative externality as pollution when necessary. Firms are assumed to share identical technologies and compete à la Cournot in both countries. 9 A representative consumer in either country can buy at most one unit of the good. Consumer s utility in country i is given as: 8 < u p i s i ' i if she buys either version U i = : s i ' i if she buys nothing, i = S; N (1) where u represents consumer s willingness to pay for the good, p i is the market price, ' i denotes the magnitude of pollution incurred by the country, and s i 2 [0; 1] is a parameter capturing the degree to which consumers are a ected by pollution. 10 de ned as: In particular, ' i is ' i = i q ii + j q ji, i; j = S; N (2) where q ii and q ji represent quantities of the good sold by the domestic and the foreign rm, i and j can equal or zero depending on the version being sold. Thus, the amount of pollution generated in a country depends on the quantity as well as the version of the good being consumed in its market. Note that even if a consumer does not buy the good, she is 7 Normalizing to zero for version H is without loss of generality. The analysis will go through as long as version L generates a larger negative externality than version H. The same is true for the treatment of the production costs of the two versions of the good. 8 Following Costinot (2008) I assume the regularity condition that c < 1=4. 9 Following previous studies I assume that markets are segmented, so that rms can charge prices independently across countries. This is helpful to isolating the e ect of country heterogeneity by controlling for the potential impact of market structure. A notable observation is that the EU actually involves a more integrated market than the WTO. Exploring how di erence in market structure may a ect the performance of product standards agreement is an interesting topic for future research. 10 Instead, it is possible to model heterogeneous country preferences by assuming S < N where i is the upper-bound of the distribution of u for country i, i = S, N. However, this formulation is less intuitive since it implies that the North has a stronger preference for both the high and the low pollution version than the South. 9

10 still a ected by the pollution due to the consumption of the dirty version by other buyers. 11 As with previous literature I assume u is uniformly distributed over [0; 1]. The market price can then be written as: p i = 1 (q ii + q ji ). (3) Consumers in the two countries have heterogeneous preferences for pollution. Without loss of generality, it is assumed that: s S < s N. (4) Condition (4) says that consumers in the South are relatively less a ected by pollution than those in the North. 12 This may capture the feature that developing countries are more willing to sacri ce environmental conditions in exchange for industrialization, which in turn arises from their comparative advantage in pollution-intensive productions. I further normalize s N to 1 and denote s S with s. Hence s is our key parameter that represents the degree of heterogeneity in country preferences for pollution, which further stands for the di erence in the level of development between countries. Next let denote a product standard that stipulates the version of the good which can be sold on the market. In particular can be H or L. I refer to = H as a high or strict standard under which only the clean version can be produced, and = L as a low or lenient standard which permits the sales of the dirty version. Each country sets standards on the domestic and the foreign rm. Let ij be country i s standard on the rm from country j where i; j = S; N. 13 product standards. 14 Countries are subject to the rule of NT or MR when setting their Under NT countries have to treatment every rm the same way so 11 It is worth noting that the formulation of ' implies that the consumption externality is local in the sense that it only a ects domestic consumers. Local externality is a useful benchmark and also holds widely in reality. For example, the environmental impacts of many goods such as automobile, pesticide and product package tend to be local. Despite the prevalence of local externality we also discuss the case of global externality in Section Hence there exist two types of preference heterogeneity in the model. Consumers between countries have asymmetric preferences for pollution which is captured by the parameter s; consumers within a country share a common preference for pollution, but may have heterogeneous tastes for the good per se which is captured by the random variable u. 13 In principle countries may allow rms to produce more than one versions, but in that case rms would simply produce version L as the production cost is lower. Therefore, without loss of generality we simply assume the choice of standard to be a singleton. 14 Focusing on NT and MR re ects the fact that the WTO member states, which account for around 85% of world countries and more than 95% of global trade, follow either NT or MR when enforcing their 10

11 that ii = ij. On the other hand, given MR a country s standard on the foreign rm is aligned with what the rm follows in its home country and this implies ij = jj. As a useful observation, note that under NT a product standard applies to both rms selling in a given country, whereas it applies to a given rm selling in both countries under MR. It is worth noting that under either policy regime each country only needs to determine the standard for its own rm: ii. Under NT this standard will automatically extend to the foreign rm selling in the host country, while under MR the host country s standard for the foreign rm is determined by its own country. To economize notations I simply denote country i s standard under both policy regimes as i, and use ( i ; j ) to represent a generic policy combination for the two countries. As a useful illustration, let (f SS ; SN g; f NN ; NS g) be a full- edged policy combination with f SS, SN g and f NN ; NS g denoting standards in the South and the North respectively. Then we have the following correspondences: (i) (H; H) = (fh, Hg; fh; Hg) under both NT and MR, (ii) (L; L) = (fl; Lg; fl; Lg) under both NT and MR, (iii) (H; L) = (fh, Hg; fl; Lg) under NT and (fh, Lg; fl; Hg) under MR, (iv) (L; H) = (fl, Lg; fh; Hg) under NT and (fl, Hg; fh; Lg) under MR. Next, let us de ne country i s national welfare under policy regime r as: w r i ( i ; j ; ) = cs r i ( i ; j ; ) + r ii( i ; j ; ) + r ij( i ; j ; ), i = S; N, (5) where r = NT or MR, cs i () is consumer surplus, ii () and ij () represent rm s domestic and foreign pro ts. World welfare is de ned as the sum of each country s welfare: ww r ( i ; j ; ) = w r i ( i ; j ; ) + w r j( i ; j ; ). (6) I consider two scenarios depending on the presence of policy coordination between counpolicies of product standards. It is interesting to examine the case where countries choose standards under no institutional constraint. While I consider this possibility under policy coordination between countries, the case of Nash equilibrium is rather complicated and is beyond the scope of this paper. 11

12 tries. In the rst scenario countries coordinate their product standards to maximize their joint welfare under NT or MR; in the second countries non-cooperatively choose product standards to maximize their own welfare. The interaction between countries and rms proceeds as follows: Stage 1: countries simultaneously choose their product standards, either cooperatively or non-cooperatively, subject to the NT or the MR constraint. Stage 2: rms compete à la Cournot in both countries following the standards set in the rst stage. I use backward induction to solve this game. Before analyzing the open economy, it is helpful as a benchmark to examine rm s behavior and optimal product standard under autarky. To this end, note that the domestic rm under autarky maximizes the pro t as: max q pq I( = H)cq, (7) where p = 1 q given that u is uniformly distributed over [0; 1], I() is an indicator function equal to one if the standard is strict (i.e. = H) and zero otherwise. Solving the problem in equation (7) yields optimal outputs as q( = H) = (1 c)=2 and q( = L) = 1=2. The rm s pro t is calculated as ( = k) = p( = k)q( = k) where k = H; L. Given the equilibrium market price, it is easy to see that only consumers with u > bu( = k) would buy the product. It follows that consumer surplus and national welfare can be calculated as: cs( = k) = Z 1 bu(=k) (u p( = k))du s( = k)q( = k), w( = k) = cs( = k) + ( = k). (8) To maximize national welfare the country simply compares w( = H) with w( = L). It follows that optimal product standard would depend on the magnitude of pollution. In particular, we have: w( = H) > w( = L) if and only if > AU, (9) 12

13 where AU is a threshold level of the pollution. 15 Condition (9) indicates that the strict standard is preferred if the pollution is su ciently large such that it exceeds AU, otherwise the lenient standard maximizes national welfare. Intuitively, a fundamental trade-o faced by the government is between lowering the production cost and curbing the negative impact of pollution. When is large the latter consideration is dominant, so implementing the strict standard is desirable; otherwise the lenient standard is pursued so as to exploit cheaper production technology. Moreover, it is straightforward to show < > 0. (10) The rst inequality of condition (10) says that a country would be more likely to impose the lenient standard as it becomes more tolerant of the pollution. The second inequality indicates that as it is costlier for the rm to produce under the strict standard, the lenient standard becomes more appealing and will be chosen at a higher level of the pollution. As we shall see, these mechanisms continue to play a fundamental role when countries determine their product standards in an open economy setting. 4 Coordination over product standards 4.1 E cient product standards under NT and MR I start by analyzing coordination over product standards between countries under NT or MR. The purpose of this analysis is two-fold. First, by eliminating strategic incentives I can focus on the welfare impacts of imposing NT and MR. This allows me to unveil a possible mismatch of product standards between countries that is induced by MR. This mismatch problem is the key reason why NT performs relatively better than MR for countries with more divergent levels of development. The second goal of this section is to establish the constrained e cient outcomes that will serve as the benchmark for studying the welfare implications of strategic interactions in the non-cooperative case. Since the Nash equilibrium to be analyzed 15 To preserve space, we collect expressions of all the thresholds of in the Appendix. 13

14 is subject to NT or MR, it is necessary to compare it with the coordination outcome that is also constrained by NT or MR so as to properly identify the e ects of strategic interactions. 16 Assume rst that both countries follow NT which mandates equal treatment of rms of all country origins. Then countries maximize joint welfare as: max ww NT ( i ; j ; ) s.t. i = ii = ij i; j = S; N (11) i ; j 2fH;Lg Since each country can choose high (H) or low (L) standard, there are four possible policy combinations under NT: (H; H), (L; L), (H; L) and (L; H). As will be seen, the constrained e cient policy combination that maximizes joint welfare depends crucially on the magnitude of pollution. As a useful starting point it can be shown that: ww NT (L; H; ) > ww NT (H; L; ) for all. (12) Importantly, Eq. (12) says that (H; L) is dominated by (L; H) for all levels of and therefore is never chosen. The intuition for this result is straightforward: given the South is more tolerant of pollution, it should not implement a stricter standard than the North from the e ciency point of view. 17 Now given su ciently large, it is optimal to impose the strict standard in both countries to avoid high pollution. As declines it becomes desirable to have countries switch to the lenient standard. In particular, it is jointly optimal to have the South rst loosen its standard, simply because its consumer welfare is less a ected by pollution. It can be shown that there exists a unique threshold NT cu such that: ww NT (L; H; ) > ww NT (H; H; ) if and only if < NT cu. (13) That is, for below NT cu the South should switch to the low standard while the North maintains the high standard. As further decreases it is optimal to also have the North 16 Of course it is obvious that when countries engage in policy coordination, the joint welfare they can achieve under no institutional constraint must be weakly higher than that under NT or MR. Hence the goal of my analysis of policy coordination is not to emphasize this point. 17 While I have suppressed the details, raising the standards in either country increases local prices and reduces the intensive margin of trade. This is consistent with the empirical ndings in Fontagné et al. (2015). 14

15 lower its standard. One can nd the unique threshold NT cl below which this is the case: ww NT (L; L; ) > ww NT (L; H; ) if and only if < NT cl. (14) Moreover, it can be veri ed that: NT cl < NT cu if and only if s < 1, (15) which implies that (L; H) must be the unique e cient policy combination for NT cl < < NT cu. To verify this is indeed the case, simply note that Eqs. (13) and (14) together imply that (L; H) dominates both (H; H) and (L; L) for NT cl < < NT cu. Furthermore, to see how preference heterogeneity a ects the e cient standards under NT, one may readily derive the following NT NT cl ) < 0. Recall that ( NT cl ; NT cu ) is the region over which (L; H) is e cient. Therefore, as preference asymmetry between the North and the South increases (i.e. s falls), (L; H) will remain e cient over a larger range of pollution. I summarize the above ndings in the following proposition: Proposition 1. Suppose countries coordinate their product standards under NT. Then e cient product standards are as follows: (L; L); (i) when pollution is high, i.e. > NT cu, both countries choose the strict standard (H; H); (ii) when pollution is low, i.e. (iii) when pollution is moderate, i.e. < NT cl, both countries choose the lenient standard: NT cl standard while the North enforces the strict standard: (L; H); < < NT cu, the South chooses the lenient (iv) as heterogeneity in development increases between countries, the range of pollution [ NT cl ; NT cu ] over which (L; H) is e cient expands. Now let us establish the e cient standards under MR. Particularly, countries maximize 15

16 joint welfare subject to the MR constraint: max ww MR ( i ; j ; ) s.t. i = ii = ji i; j = S; N. (16) i ; j 2fH;Lg Essentially, the MR constraint ensures that each rm is subject to a unique standard in every country. As a result, there are four policy combinations that can arise under MR: (H; H), (L; L), (H; L) and (L; H). Note that despite the same notations, the latter two policy combinations represent policy pro les (fh, Lg; fl; Hg) and (fl, Hg; fh; Lg) which are di erent from those under NT, i.e. (fh, Hg; fl; Lg) and (fl, Lg; fh; Hg). To begin with, it can be shown that the following condition holds: ww MR (L; H; ) = ww MR (H; L; ) for all, (17) that is, when standards are asymmetric across countries under MR, world welfare remains unchanged regardless of the country that implements the high or low standard. The reason is that in either case each country always has both versions produced, and since rms have identical cost functions it does not matter which rm produces version H or L. Market prices and outputs remain the same in both cases, which leads to the same aggregate pro ts (i.e. the sum of rm s global pro t) and consumer surplus. Without loss of generality I just assume countries choose (L; H) over (H; L) since it is natural for the South to implement the low standard. Next one can show that it is e cient for both countries to enforce the strict standard if pollution is high. In particular, it can be established that there exists a threshold MR cu that: such ww MR (L; H; ) > ww MR (H; H; ) if and only if < MR cu. (18) Hence both countries choose the high standard for > MR cu while the South should switch to the low standard for < MR cu. Similarly, one can nd another threshold MR cl such that: ww MR (L; L; ) > ww MR (L; H; ) if and only if < MR cl, (19) 16

17 i.e. it is e cient for both countries to set the low standard when the level of pollution falls below MR cl. Importantly, it can be shown that: MR cu < MR cl. (20) The implications of Eq. (20) are illustrated in Figure 1. As can be seen from the gure, when MR cu < MR cu < < MR cl we have (L; H) dominated by either (H; H) or (L; L). Moreover, when and > MR cl, Eqs. (18) and (19) together indicate that (L; H) remains dominated by (H; H) and (L; L). It follows that (L; H) is never optimal under MR for all values of. This implies that only (H; H) and (L; L) can possibly maximize joint welfare. [Figure 1 here] Now comparing (H; H) and (L; L), one can establish a new threshold cm such that: ww MR (H; H; ) > ww MR (L; L; ) if and only if > cm. (21) Moreover, it can be readily checked that: MR cu < cm < MR cl. (22) Eqs. (21) and (22) together imply that countries always choose identical standards under MR for joint optimality. In particular, they both enforce the high (low) standard when pollution is higher (lower) than cm. The following proposition summarizes the above ndings: Proposition 2. Under MR, e cient standards are symmetric regardless of national di erence in development. Speci cally, (i) when pollution is high, i.e. > cm, both countries choose the strict standard (H; H); (ii) when pollution is low, i.e. < cm, both countries choose the lenient standard (L; L). Importantly, Proposition 2 shows that under coordination countries would never choose asymmetric standards (L; H) when following MR. The key intuition is that with country preference heterogeneity, (L; H) under MR induces a mismatch of product standards between 17

18 countries. In particular, when it is e cient to enforce a lenient standard only in the South, MR requires the North to also apply the same standard to the Southern rm. This can reduce Northern welfare since its consumers are more averse to pollution. On the other hand, the South also needs to recognize and apply the high standard to the Northern rm, which is again welfare-reducing since Southern consumers can tolerate higher pollution. Hence the mismatch of standards induced by MR is in fact mutual. As a result, countries seek to avoid the mismatch problem by simply choosing symmetric standards for all levels of pollution. It is important to note that the mismatch of standards does not occur under NT. The reason is that each country under NT has full control over the standards enforced in its market and will not be a ected by what standards are implemented in the foreign country. 4.2 Comparing NT and MR It has been shown that the mismatch of standards between countries can occur under MR but not NT. It is then natural to ask when the mismatch problem would a ect the e ciency of MR relative to NT. This question can be answered by comparing the coordination outcomes under NT and MR. Figure 2 illustrates the comparison. When pollution is high or low (i.e. > NT cu or < NT cl ), both policy regimes induce the same coordination outcome (H; H) or (L; L). This implies that the mismatch problem under MR does not take its toll. Intuitively, when externality is either signi cant or minor, both countries would like to avoid or exploit it regardless of policy regime. As a result e cient product standards under NT and MR coincide. [Figure 2 here] When there is moderate pollution (i.e. NT cl < < NT cu ), however, the two policy regimes yield di erent coordination outcomes: under NT countries choose (L; H) while under MR either (H; H) or (L; L) is chosen. This immediately implies that joint welfare is lower under MR since (H; H) and (L; L) are options available under NT but are not chosen. It follows that the mismatch problem under MR gives rise to an e ciency loss relative to NT for intermediate levels of pollution. Intuitively, when the externality is moderate it is e cient to allocate product standards di erentially across countries depending on their preferences, 18

19 with the South implementing lower standards. However this is not feasible under MR because countries have to recognize each other s standards even if the standards are incompatible with their domestic needs. To prevent this problem from occurring countries would choose symmetric standards (H; H) or (L; L) which allow them to attain higher joint welfare. But this is still less e cient than (L; H) under NT that respects country preference heterogeneity while not giving rise to the mismatch problem. It is easy to verify that the mismatch of standards under MR tends to get worse as country heterogeneity increases. In particular, one can show the following holds: d( NT cu ds NT cl ) < 0 for 0 s 1, that is, the range of pollution where NT yields higher joint welfare than MR expands as country heterogeneity rises. This indicates that the mismatch problem under MR is more likely to cause an e ciency loss relative to NT as country heterogeneity rises. As will be seen, in the non-cooperative case this turns out to be the key driving force for NT to fare relatively better than MR between countries with more divergent levels of development. Proposition 3 summarizes the e ciency implications of the two policy regimes. Proposition 3. Suppose countries coordinate their product standards to maximize joint welfare. Then: (i) The mismatch of standards induced by MR reduces e ciency relative to NT for intermediate levels of pollution: NT cl joint welfare. < < NT cu. Otherwise both policy regimes yield identical (ii) The range of pollution [ NT cl ; NT cu ] expands as s decreases, that is, the mismatch problem under MR becomes more e ciency-reducing with higher national di erence in development Note that the reasoning is robust to how preference di erence across countries is modeled. Suppose instead s S = 1 and s N rises above 1, i.e. the North now cares more about pollution in an absolute sense. Then it is straightforward to check that NT will dominate MR over some new range [ e NT l increases it can be shown that e NT l falls while e NT u ; e NT u ], and as s N remains unchanged. As a result e NT NT e u l continues to expand so that NT would still dominate MR for a larger range of the pollution. This further implies that our results would remain valid if both s N and s S vary and diverge from each other. The intuition is that the relative performance of NT to MR depends on the degree of asymmetry in country preference rather than its magnitude. 19

20 It can be readily seen that heterogeneity in national economic development is essential for the mismatch problem under MR to arise. If countries are identical (i.e. s = 1) then it is straightforward to show that NT cu = NT cl = cm. That is, e cient standards under NT become symmetric for all and coincide with those under MR. It follows that NT and MR always yield identical e ciency implications. Intuitively, when countries are identical they simply prefer the same standards so that MR would not induce the mismatch problem. Lemma 1. Suppose countries are identical, i.e. s = 1. Then coordination under NT or MR yields the same e cient product standards and welfare so that the mismatch of standards under MR does not arise. Figure 3 depicts the welfare comparison of NT and MR over the two-dimensional parameter space (; s). [Figure 3 here] Finally, it is useful to examine the socially e cient outcome that is attained under no institutional constraints. By comparing this unconstrained e cient outcome with the socially optimal standards under NT and MR, I can identify the potential impacts of imposing the constraint of NT or MR. The reasoning for the unconstrained case is similar as before: starting with (H; H) for high pollution, it is e cient to switch to the lenient standard rst in the North and then in the South as pollution declines. Moreover, I show in the appendix that it is e cient for either country to switch to the lenient standard on both rms simultaneously. In other words, unconstrained e cient outcome entails non-discriminatory product standards that follow NT. This result is stated in the following proposition. Proposition 4. Coordination under no institutional constraints induces product standards in each country that are consistent with NT. Importantly, Proposition 4 implies that the e cient standards under NT are in fact rst-best in the sense that they are also optimal under no institutional constraints. More importantly this is true regardless of development heterogeneity. It follows that imposing NT does not undermine e ciency under coordination, and therefore it is not even necessary to mandate NT when coordination is possible as it is exactly what optimality entails. 20

21 4.3 Global externality Consumption externality can also travel beyond national borders and become global. In this section I investigate the welfare implications of NT and MR under coordination with global consumption externality. Recall that the mismatch of standards induced by MR is the key driving force that gives NT an edge with local externality. Therefore the goal of this subsection is to examine whether this mechanism is altered in the presence of global externality. To this end, I assume that the pollution incurred by a country is caused by both domestic and foreign consumptions, so that consumer utility in country i is modi ed as follows: 8 < u p i s i (' i + ' j ) if she buys either version U i = : s i (' i + ' j ) if she buys nothing, where measures the impact of the pollution arising from foreign consumption. I assume that < 1, i.e. pollution from the foreign country has a smaller e ect on domestic consumer utility. This is reasonable as pollution typically dies down with distance. Given the speci - cation in (23), one can apply similar reasoning as before to solve for various thresholds of and characterize the optimal standards under the two types of policy regimes. In particular, starting with uniform strict standards across countries, one can nd NT 0 u (23) under NT such that the South switches to the lenient standard for < NT 0 u, and NT 0 l such that the North also does so for < N0 l. Moreover, it can be shown that: NT 0 u > NT 0 l if and only if < 1. (24) Hence for NT 0 l < < NT 0 u similar lines, we can nd MR0 u > M0 u the di erential standards (L; H) are optimal under NT. Along and MR0 l under MR such that (H; H) dominates (L; H) for and (L; L) dominates (L; H) for < MR0 l. Moreover, it can be checked that: MR0 l > MR0 u. (25) As with the case of local externality, condition (25) says that the di erential standards (L; H) are never optimal under MR. This leaves (H; H) and (L; L) as the only possible candidate 21

22 policy options. One can further show that there exists 0 cm such that (H; H) and (L; L) are optimal for > 0 cm and < 0 cm respectively. Therefore global consumption externality does not qualitatively change the pattern of constrained optimal standards. It follows that NT still weakly dominates MR for all and such domination is strict when the externality is moderate (i.e. NT 0 l < < NT 0 u ). Intuitively, the key driving force remains unchanged given global externality: the problem of mismatched standards continues to exist under MR and the incentive for avoiding this problem leads countries to stick with symmetric standards. Moreover, it is easy to check that xing, the range of in which NT strictly dominates MR (i.e. ( NT 0 u NT 0 l )) expands as s falls, indicating that the mismatch becomes more prevalent as preference heterogeneity rises between the North and the South. 5 Non-cooperative product standards 5.1 Nash equilibrium In this section, I derive non-cooperative Nash equilibrium where countries simultaneously and independently set their product standards to maximize each s own welfare. Let us rst consider the equilibrium outcome under NT. To start with, it is worth noting that policy decisions under NT are independent across countries, that is, one country s optimal policy does not rely on the policy decision of the other country. 19 This is because the non-discrimination constraint imposed by NT eliminates the pro t-shifting incentives as in Brander and Spencer (1985): countries are not able to shift pro ts to the domestic rm via setting a higher standard on the foreign rm since doing so will violate NT. On the other hand, the policy decisions under MR are interdependent between countries because the pro t-shifting incentives are present, as will be seen in the ensuing analysis. It turns out that Nash equilibrium outcome also depends on the level of pollution. First, for large uniform strict standards must be the unique Nash equilibrium. To see this, note that it is the South that always has a stronger incentive to loosen the standard 19 An implication is that maximizing national welfare is equivalent to maximizing the sum of consumer surplus and rm s domestic pro t, since rm s foreign pro t will be determined by the other country s standards. 22

23 as pollution reduces. Particularly, one can show there exists a unique threshold NT u which the South would adopt the lenient standard: below w NT S (L; H; ) > ws NT (H; H; ) if and only if < NT u. (26) It follows that the South would choose the strict standard when > NT u. Moreover, the North would maintain the strict standard whenever the South does so because it is less tolerant of the pollution. Therefore no country has an incentive to deviate to a lenient standard for > NT u, which veri es that (H; H) is the unique Nash equilibrium. Next one can show that (L; L) is the unique Nash equilibrium if pollution is su ciently low. Note that starting from (L; L) it is the North that is more willing to raise the standard as continues to rise. This implies there exists NT l high standard when > NT l : such that the North would switch to the w NT N (L; H; ) > wn NT (L; L; ) if and only if > NT l. (27) On the other hand, the North would adopt the lenient standard as long as < NT l. But whenever this is the case the South would also implement the lenient standard since it is less a ected the pollution relative to the North. equilibrium for < NT l. Moreover, it can be shown that: Thus (L; L) must be the unique Nash NT l < NT u if and only if s < 1, (28) thus one needs to solve for the equilibrium for NT l < < NT u. But over this range neither (H; H) nor (L; L) can be an equilibrium because condition (26) and (27) together indicate that either country would have an incentive to deviate. This leaves (H; L) and (L; H) as the only possible Nash equilibrium. Next, note that given the South being less averse to pollution it would never impose a higher standard relative to the North. This rules out (H; L) as an equilibrium for all. It follows that (L; H) must be the only possible Nash equilibrium. To see this is indeed the case, simply note that condition (26) indicates that the 23