Domestic Product Standards and Free Trade Areas

Domestc Product Standards and Free Trade Areas Akhko Yanase a, Hrosh Kurata b a Graduate School of Economcs, Nagoya Unversty b Faculty of Economcs, Tohoku Gakun Unversty May, 201 Abstract Ths study consders endogenous determnaton of domestc standards on products that cause negatve consumpton externaltes n the presence of a possble free trade area (FTA) n a three-country world, and examnes how an FTA affects the optmal levels of (external) tarffs and standards chosen by each country and natonal welfare. We demonstrate that under an FTA, member countres standards become more strngent than under a tarff-war equlbrum based on the most-favored-naton (MFN) clause. Member countres welfare may or may not be hgher under the FTA than under the MFN equlbrum, whereas the nonmember country unambguously become better off after the FTA formaton. We also compare two regmes, natonal standards and harmonzaton, regardng the choce of standards by the FTA members and show that n comparson wth the natonal standards, harmonzaton of standards wthn an FTA wll lead the member countres to choose less strngent standards and make the formaton of the FTA more favorable. Keywords: Regonalsm; Optmal tarffs; Standards; Free trade areas; Internatonal olgopoly JEL classfcatons: F12; F13; F15; F18 The authors would lke thank Ncola Conglo, Marco Martorana, Mnoru Kunzak, Kesaku Hgashda, Morhro Yomogda, Nortsugu Nakansh, Laxun Zhao, Cheng-Hau Peng, Kazunobu Hayakawa, Toshhro Ichda, and Tayo Yoshm for ther comments on earler drafts. Correspondng Author; Address: Graduate School of Economcs, Nagoya Unversty, Furo-cho, Chkusa-ku, Nagoya 6-8601, Japan; E-mal: yanase@soec.nagoya-u.ac.jp hkurata@mal.tohoku-gakun.ac.jp 1

1 Introducton Preferental trade agreements (PTAs) has dramatcally been ncreased n the last two decades. For example, n 2012, WTO receves 511 notfcatons of PTAs and among them 319 n force, whch s more than fve tmes from the correspondng number n 1992. 1 Under the PTAs, tarffs and other trade barrers are removed n most of goods trade between ts member countres. Even though tarffs on trade n commodtes are elmnated va trade agreements, nontarff barrers may stll exst among the natons that conclude the agreements. Among others, techncal regulatons and product standards, the man purpose of whch s to keep safety or envronmental condton for the country, may vary from country to country, and such dfferent regulatons and standards may be obstacles for foregn exporters. Moreover, f regulatons and standards are set arbtrarly, they could be used as an excuse for protectonsm. These Techncal Barrers to Trade (TBT) have become a bg concern not only for today s WTO but also for PTAs. For example, when the European Unon (EU) forms a PTA, the agreement often requests the partner country to harmonze ts natonal standards and conformty assessment procedures wth those of the EU (Stoler, 2011). 2 Of another example s the Trans-Pacfc Partnershp (TPP) agreement, negotatons for whch have recently been takng place as an expanded verson of the 2005 Trans-Pacfc Strategc Economc Partnershp Agreement (TPSEP). The TPP s ntended to be a hgh-standard agreement specfcally amed at emergng trade ssues n the 21st century, and the key features of broad outlnes of the TPP ncludes regulatory coherence; commtments wll promote trade between the countres by makng trade among them more seamless and effcent. 3 1 See WTO webste: http://www.wto.org/. 2 Chapter of the EU South Korea FTA ncludes specfc undertakngs on good regulatory practce: transparency n makng rules, use where possble of nternatonal standards, provdng the other Party wth an opportunty to dscuss rules before they are made, and allowng suffcent tme for the other Party to comment on then and to take account of ther adopton. 3 Chapter 8 of the TPSEP agreement s allotted for TBT. Artcle 8.7 states that the Partes shall use nternatonal standards, or the relevant parts of nternatonal standards, as a bass for ther techncal regulatons and related conformty assessment procedures where relevant nternatonal standards exst or ther completon s mmnent, except when such nternatonal standards or ther relevant parts are 2

In lght of the growng mportance of standards or regulatons that may act as nontarff barrers n PTAs, several questons wll arse. Do standards become more or less strngent under a PTA than n the absence of t? After a formaton of the PTA, do member or nonmember countres become better off? Do potental PTA members have an ncentve to harmonze ther standards? Ths paper addresses these ssues theoretcally by usng a smple three-country olgopolstc trade model. In ths paper, among the several forms of PTAs, we focus on free trade areas (FTAs), where each member country chooses ts external tarffs ndependently. The structure of the model employed n ths paper s based on those n the studes on welfare effects of PTAs or the possbltes for PTAs to be multlateral free trade under nternatonal olgopoly (Y, 1996, 2000; Krshna, 1998; Freund, 2000; Orneras, 2005a, b; Sagg, 2006). We extend the basc three-country model of mperfect competton by ncorporatng the endogenous determnaton of standards by natonal governments. In ths paper, we consder standards for controllng negatve externaltes generated by consumpton of goods. In order to enter the mporter country s market, foregn exporters must produce goods that meet the mport s standard, and thus the standard can be a nontarff barrer. Governments are assumed to be benevolent, wthout any poltcal ncentves to set ther respectve standards. Under these assumptons, we consder endogenous determnaton of standards as well as (external) tarffs. The man results obtaned n ths paper are as follows. Compared wth the polcy game n the absence of FTAs, an FTA makes the member countres to choose more strngent standards. Regardng the natonal welfare n each country, the FTA member countres may or may be better off under the formaton of an FTA, whle the nonmember country becomes better off for the case of low degree of transboundary externaltes. By comparng the case n whch FTA members ndependently determne ther respecneffectve or napproprate to fulfl legtmate objectves. In vew of the fact that most exstng PTA arrangements take the form of FTAs and less than 10% can be consdered to be fully fledged customs unons, Facchn et al. (2012) develop a poltcal economy model of trade polcy under mperfect competton to provde a postve explanaton for the prevalence of FTAs. 3

tve natonal standards wth the case n whch the FTA member countres harmonze ther standards wthn the FTA, such harmonzaton of standards wll lead the member countres to choose less strngent standards and make the formaton of the FTA more favorable, provded the degree of transboundary externaltes s not so hgh. Fnally, we brefly state about the related studes n order to clarfy our paper s standpont. Recently, we often fnd theoretcal studes on trade and standard. 5 Fscher and Serra (2000) pont out the possblty of strategc use of standards under nternatonal olgopoly;.e., governments can exclude foregn frms by settng the lowest standard though t ncreases costs of domestc frms. Costnot (2008) compares performance between two dfferent agreements on product standards; WTO s approach based on a Natonal Treatment (NT) prncple and EU s approach that reles on a prncple of Mutual Recognton (MR). More recently, Takarada et al. (201) nvestgate regonal and multlateral agreements on standards n a three-country olgopolstc trade model to shed lght on how country/regon specfc regulatons affect multlateralsm. Notce that n these studes focus on standards as nontarff barrer by assumng no tarff-based protecton. In contrast to these studes, we consder both tarffs and nontarff barrers to take account of the nteractons between these tradtonal and modern forms of trade protecton. The rest of ths paper s organzed as follows. In secton 2, we set up a three-country model of nternatonal olgopoly wth standards mposed by the natonal governments. In secton 3, we derve the equlbrum of the polcy game n the absence of FTAs. In secton, we derve the equlbrum of the polcy game under an FTA. When governments n the member countres determne ther respectve standards, they may act ndependently or jontly. We consder both cases. We examne how a formaton of the FTA affects the standards chosen by the member countres and natonal welfare n member and nonmember countres. We also compare the outcomes under natonal standards wth 5 In the text, we focus on theoretcal studes on trade and standard, whch have qute smlar perspectve. As well as theoretcal studes, we can fnd emprcal studes on ths ssue. For example, Baghdad et al. (2013) emprcally test the relatonshp wth RTAs wth envronmental provsons and polluton levels.

those under harmonzaton of standards wthn the FTA. In secton 5, we extend the basc model by ncorporatng transboundary externaltes and dscuss whether or not the results derved from the basc model are mantaned. In secton 6, we provde bref concludng remarks. 2 Model We consder three symmetrc countres, A, B, and C, each wth one frm that produces a homogeneous product. Consumpton of the product generates negatve externaltes, whch s assumed to be local, and the level of the externaltes depends on a standard mposed by the natonal government. Let us denote the standard mposed n country by s ( = A, B, C), and the externaltes per unt of consumpton s gven by b(s ), whch s assumed to be decreasng n s. There s also another homogeneous good, whch serves as the numerare and s assumed to be freely traded and produced under perfect competton wth constant returns to scale technology, and generate no externaltes. There s a contnuum of homogeneous consumers of measure one. Each consumer n country has the followng quas-lnear preference: U(Q, Y ; s, Q ) = u(q ) + Y b(s ) Q, where Q and Y are hs consumpton of the externalty-generatng product and the numerare, respectvely, and Q s the aggregate consumpton of the product. As explaned above, b(s ) Q denotes the negatve externaltes n ths country. Throughout the paper, we assume that u(q ) s quadratc and hence the consumers utlty maxmzaton derves the lnear nverse demand functon P (Q ) = α Q, α > 0. The three frms compete n quanttes n each of the natonal markets, whch are assumed to be segmented. We assume that these frms have dentcal technologes and each frm s unt producton cost s a functon of standard mposed n the country of consumpton. Let us denote the unt producton cost by c(s ), whch s assumed to be ncreasng n s. That s, t s more expensve to produce at a hgher standard. 6 Ths 6 For example, f the use of a certan food addtve s legally prohbted n a country, the frms 5

s because the frms are banned from sellng goods that do not meet ndvdual natonal standards, and thus the frms produce goods whch exactly fulfll the requrements of standards n each country s market. We also assume that the governments mposes tarffs on mports. Because the markets are segmented, we can consder the Cournot Nash equlbrum n each market separately. Let us denote the output of the domestc frm by q and the outputs of foregn exporters by q j, and q k,, j, k = A, B, C, j k, thus Q = q +q j +q k. The foregn frms face a specfc tarff when exportng to country. Let us denote the tarff rates on mports that the natonal government n country mposes by t j and t k, respectvely. Assumng that the frms do not ncur fxed costs, profts of the respectve frms supplyng to the market n country are gven by π = [P (Q ) c(s )]q = [a(s ) Q ]q, π j = [P (Q ) c(s ) t j ]q j = [a(s ) Q t j ]q j, π k = [P (Q ) c(s ) t k ]q k = [a(s ) Q t k ]q k, where a(s ) α c(s ). From the frst-order condtons for proft maxmzaton, Cournot Nash equlbrum output of each frm s derved as follows: q = a(s ) + t j + t k, q j = a(s ) 3t j + t k The total output sold n country s therefore derved as, q k = a(s ) + t j 3t k. (1) Q = 3a(s ) t j t k. (2) These equlbrum outputs are a functon of polcy varables mposed by the government n country. The total proft of the frm producng the externaltes-generatng product s the sum of profts from domestc and export sales: π = π + π j + π k,, j, k = A, B, C, j k. Natonal welfare n each country s defned as the sum of consumer surplus supplyng food products n that country s market should avod usng that materal n ther producton process. 6

CS = Q 0 P (x)dx P (Q )Q, total proft π, and tarff revenue t j q j + t k q k, mnus the socal cost from negatve externaltes b(s )Q. Gven the lnear demand functon, t holds that CS = (Q ) 2 /2 and π = (q ) 2 +(q j ) 2 +(q k ) 2. In lght of (1) and (2), natonal welfare s therefore gven by W = [3a(s ) t j t k ] 2 32 + t j a(s ) 3t j + t k + [a(s ) + t j + t k ] 2 + t k a(s ) + t j 3t k + [a(s j) 3t j + t jk ] 2 + [a(s k) 3t k + t kj ] 2 b(s ) 3a(s ) t j t k. (3) In the followng analyss, we assume that there are two standards avalable, s H and s L, wth s H > s L. If the natonal government choose a hgh standard, frms ncur a unt cost c(s H ) = γ > 0, but consumpton of the good does not generate negatve externaltes,.e., b(s H ) = 0. If the government choose a low standard, frms do not ncur costs,.e., c(s L ) = 0, but consumpton generates negatve externaltes b(s L ) = β > 0. Due to these smplfcatons of unt cost and externalty functons, we can compare the welfare levels under dfferent regmes of polcy games. In our analyss, we consder the followng order: n the frst stage, each government determnes tarffs and standards, and n the second stage the frms compete n the Cournot way. We derve the subgame perfect Nash equlbrum of ths game by backward nducton. Although we mplctly consder the stuaton n whch the governments determne these polcy nstrument smultaneously, under our settng the same solutons are obtaned even f the governments frst determne the standards, and then chooses the optmal tarffs that maxmzes natonal welfare takng the standard as gven (see Appendx). We thus formulate the governments behavor as f ther polcy games are played sequentally. 3 Polcy Game wthout FTAs In ths secton, we consder a polcy game n the absence of FTAs as a benchmark. It s natural to assume that the determnaton of tarffs follows the prncple of nondscrmnaton, known as the most-favored-naton (MFN) clause. That s, the government n 7

each country mposes a sngle nondscrmnatory tarff on ts tradng partners and all countres smultaneously choose ther respectve tarffs to maxmze ther own welfare. Substtutng t j = t k = t nto (3), the natonal welfare s rewrtten as W = [3a(s ) 2t ] 2 32 + t a(s ) 2t 2 + [a(s ) + 2t ] 2 + [a(s j) 2t j ] 2 + [a(s k) 2t k ] 2 b(s ) 3a(s ) 2t. () Takng the tarffs n other countres t j and t k as gven, the government n country determnes t so as to maxmze welfare (). From the frst-order condton W / t = 0, the MFN tarff for a gven standard s derved as t = 3a(s ) + b(s ) 10 t M (s ). (5) Notce that because of the assumptons of segmented markets and a constant unt cost (for a gven s ), a country s MFN tarff does not depend on the other countres tarffs. Ths also means that the MFN tarffs n the tradng partners are t M (s j ) and t M (s k ). The strngency of standards affects the MFN tarff rate n the followng manner: t M (s H ) = 3(α γ) 10 < 3α + β 10 = t M (s L ). (6) Because more strngent standards lead a hgher cost of complance, whch reduces outputs of both domestc and foregn frms and rases the prce. In order to compensate for the resultng losses n consumer surplus and tarff revenue, the natonal government wll reduce the tarffs on mports. We turn to the determnaton of standards, whch are made noncooperatvely n ths benchmark stuaton. In lght of (5), the natonal welfare () can be rewrtten as W M = w M (s ) + π M j (s j ) + π M k (s k), where w M (s ) [11a(s ) 10t M (s )][a(s ) + 2t M (s )] 32 b(s )[3a(s ) 2t M (s )] (7) s the domestc surplus and πj M (s j ) [a(s j) 2t M (s j )] 2, πk M (s k ) [a(s k) 2t M (s k )] 2 (8) 8

are the export profts. Takng the standards n other countres s j and s k as gven, the government n country determnes s {s L, s H } so as to maxmze the natonal welfare W M. Snce W M s addtvely separable, the problem s equvalent to choose s that maxmzes the domestc surplus (7). A drect calculaton yelds that w M (s H ) = 2(α γ) 2 /5 and w M (s L ) = (α 2 6αβ + β 2 )/10. Hence, f w M w M (s H ) w M (s L ) = 8αγ + γ2 + 6αβ β 2 10 (9) s postve (negatve), the government fnd t optmal to choose s H (s L ). In order to exclude cases where governments set prohbtve tarffs, we restrct our attenton to the parameter values that satsfy α > γ and α > 2β. 7 Gven these parameter restrctons, t s clear from (9) that w M > 0 ( w M < 0) holds f and only f γ < (>) γ M (β) α α 2 3 β2 αβ + 2. It s easly verfed that γ M (β) s ncreasng and convex n β: d γ M (β) dβ = 3α β (α 2 ) 1/2 32 β2 αβ + > 0, d 2 γ M (β) dβ 2 = 5α2 (α 2 ) 3/2 32 β2 αβ + > 0. In addton, t holds that γ M (α/2) = (1 5/)α 0.10α. Therefore, the government chooses a hgh standard f (β, γ) S M H, whereas t chooses a low standard f (β, γ) SM L, where S M H {(β, γ) [0, α/2] [0, α] γ < γm (β)} and S M L {(β, γ) [0, α/2] [0, α] γ > γ M (β)}. Lemma 1. Under the MFN, each government s domnant strategy for standards s to choose s H f 0 < γ < γ M (β) and s L f γ M (β) < γ < α. The ntuton behnd Lemma 1 s straghtforward. In tghtenng standards, natonal governments face a trade-off between a rse n the frms unt cost γ and a mtgaton of negatve externaltes β. If γ s relatvely hgh compared wth β, the former effect domnates the latter, and thus the governments wll choose a less strngent standard. If β s relatvely hgh, the government wll make an opposte choce. 7 Substtutng (5) nto the export outputs, we obtan q j = q k = [a(s ) 2b(s )]/10, whch becomes postve f α > γ (n the case of hgh standard) and α > 2β (n the case of low standard). 9

Polcy Games under an FTA Let us now suppose that countres A and B form a free trade area (FTA), whch reduces tarffs between these country to zero: t AB = t BA = 0. By defnton of an FTA, the member countres set ther respectve external tarffs ndependently, and let us denote the external tarff rate by t C = t, = A, B. Substtutng these nto (3), the natonal welfare of a member country s gven by W = [3a(s ) t ] 2 32 + t a(s ) 3t + [a(s ) + t ] 2 + [a(s j) + t j ] 2 + [a(s k) 2t C ] 2 b(s ) 3a(s ) t,, j = A, B, j. (10) Because tarffs are elmnated wthn the FTA partners, total sales and the export to the FTA partner ncrease, whch lead to ncreases n consumer surplus and export proft n the FTA partner s market. At the same tme, the elmnaton of tarffs reduce the domestc proft and tarff revenue. In addton, the ncrease n the domestc consumpton leads to larger negatve externaltes. Takng the external tarff rate n the FTA partner t j and the tarff rate n the nonmember country t C as gven, the government n country determnes t so as to maxmze welfare (10). The optmal external tarff for a gven standard s derved as t = 3a(s ) + b(s ) 21 t F (s ), (11) whch s less than the optmal tarff level under the MFN, t M (s ). Lemma 2. For a gven level of standards, the optmal external tarff under an FTA s lower than the MFN tarff: t F (s ) < t M (s ). Lemma 2 s a well-known tarff complementarty effect (Bagwell and Stager, 1998). Intutvely, because of a change n compettve advantage n the member countres markets, a formaton of FTAs reduces mports from the nonmember country, whch n turn reduces tarff revenue and consumer surplus, and n order to offset these negatve effects the member countres encourage the mport from the nonmember country by reducng external tarffs. 10

The government s behavor n the nonmember country s the same as that under MFN, and thus, for the frms n the FTA members, the export proft n the nonmember country s π M C (s C), = A, B. Then, n lght of (11), the member country s natonal welfare (10) can be rewrtten as W F = w F (s ) + π F j(s j ) + π M C (s C), where w F (s ) 11[a(s )] 2 + 6a(s )t F (s ) 21[t F (s )] 2 32 s the domestc surplus and b(s )[3a(s ) t F (s )] (12) π F j(s j ) [a(s j) + t F (s j )] 2 (13) s the export proft n the FTA partner s market. In the followng analyss, we consder two scenaros regardng the determnaton of the FTA members standards. One s a natonal standards regme, where the government n each member country ndependently determnes the level of standard so as to maxmze ts own welfare. The other s a harmonzaton regme, where the governments n member countres harmonze ther standards; they jontly determne the standards so as to maxmze the jont welfare wthn the FTA..1 Natonal standards In the natonal-standards regme, takng the standards n other countres s j and s C as gven, the government n country determnes s {s L, s H } so as to maxmze the natonal welfare W F, or equvalently, the domestc surplus w F (s ). From (11) and (12), t holds that w F (s H ) = 5(α γ) 2 /1 and w F (s L ) = (15α 2 30αβ + β 2 )/2. Therefore, the government fnd t optmal to choose s H (s L ) f w F w F (s H ) w F (s L ) = 30αγ + 15γ2 + 30αβ β 2 2 (1) s postve (negatve), or equvalently, f γ < (>) γ F (β) α α 2 2αβ + β2 15. It s easly verfed that γ F (β) s ncreasng and convex n β. As n the polcy game equlbrum n the absence of the FTAs, we can partton the set { (β, γ) R 2 + α > γ, α > 2β } 11

nto two regons, one n whch the member countres choose s H and the other n whch they choose s L. Lemma 3. Suppose that countres A and B form an FTA. If the governments n member countres choose ther respectve natonal standards, each government s domnant strategy s to choose s H f 0 < γ < γ F (β) and s L f γ F (β) < γ < α..1.1 Comparson of optmal standards Comparng Lemmas 1 and 3, we see that a formaton of the FTA changes the regons of optmal standards. More specfcally, we obtan the followng lemma. Lemma. It holds that γ M (β) < γ F (β) β (0, α/2). (Proof) Comparng the slopes at the orgn, we have d γ M (0)/dβ = 3/ < 1 = d γ F (0)/dβ. In addton, t holds that γ F (α/2) = (1 15/30)α 0.8709α > γ M (α/2). Snce both γ M (β) and γ F (β) are ncreasng and convex n β, the statement of the lemma holds. Lemma mples that the FTA further parttons the regon S M L defned n the prevous secton, where the governments choose a low standard under the MFN, nto two subsets. In lght of Lemmas 1 and 3, f γ M (β) < γ < γ F (β), the member countres change ther behavor n such a way that they choose a low standard under the MFN but they choose a hgh standard after the formaton of the FTA. Ths s llustrated n Fgure 1. If (β, γ) s n regon I, the member countres choose s L both under the MFN and FTA. If (β, γ) s n regon II, the member countres choose s L under the MFN but they choose s H under the FTA. If (β, γ) s n regon III, whch s the same as S M H defned n the prevous secton, the member countres choose s H both under the MFN and FTA. Proposton 1. In comparson wth the MFN, an FTA makes the member countres to choose more strngent standards n the sense that there exsts a set of parameters (β, γ) n whch the member countres choose a low standard under the MFN but choose a hgh standard under an FTA and f (β, γ) s not an element of ths set, the member countres do not change ther standards before and after an FTA formaton. 12

Fgure 1: Comparson of optmal standards before and after an FTA The ntuton behnd Proposton 1 s as follows. Elmnatng tarffs between FTA partners ncreases consumpton and producton n the member countres, but at the same tme, the member countres suffer from reduced tarff revenue. However, because of the tarff complementarty effect demonstrated n Lemma 2, the mport from the nonmember country ncreases, whch mtgates the reducton of tarff revenue. Therefore, the postve effect of the FTA domnates the negatve one, and the member countres wll not be worse off even they rase the unt costs of frms supplyng to the market by adoptng more strngent standards. Moreover, the ncrease n consumpton wll cause expanson of negatve externaltes, and thus the member countres should actually adopt more strngent standards..1.2 Comparson of natonal welfare We are now n a poston to examne whether or not an FTA makes each country better off. As shown n Fgure 1, there are three possbltes regardng the combnaton of opt- 13

mal standards under MFN and FTA. Therefore, we make comparsons of each country s natonal welfare n the respectve regons. Member countres We begn wth the member countres. In regon I, the natonal welfare s gven by W M = w M (s L ) + π M j (s L ) + π M C (s L) under the MFN and W F = w F (s L ) + π F j(s L ) + π M C (s L) under the FTA,, j = A, B, j. Then, from (5), (7), (8), (11), (12), and (13), we have W M W F = w M (s L ) + πj M (s L ) [w F (s L ) + πj(s F L )] (23α 1256β)(3α + β) =. 100 Then, W M > W F (W M < W F ) f and only f β > (<) 23α/1256 0.3368α. 8 In regon II, the natonal welfare under the MFN s gven by W M = w M (s L ) + π M j (s L ) + π M C (s L) and under the FTA, W F that = w F (s H ) + π F j(s H ) + π M C (s L). Then, t holds W M W F = w M (s L ) + π M j (s L ) [w F (s H ) + π F j(s H )] = 11α2 3136αβ + 686β 2 + 300αγ 2150γ 2. 900 Gven the parameter restrctons α > γ and α > 2β, t can be verfed that W M (W M < W F ) holds f and only f > W F γ > (<) γ(β) α 7 5 1 86 α2 32 3 αβ + 7 3 β2. It also holds that γ(0) 0.0333α and γ(α/2) 0.663α ( γ M (α/2), γ F (α/2) ). In regon III, the natonal welfare s gven by W M = w M (s H ) + π M j (s H ) + π M C (s H) under the MFN and W F = w F (s H ) + π F j(s H ) + π M C (s H) under the FTA. Then, t follows that W M W F = w M (s H ) + πj M (s H ) [w F (s H ) + πj(s F 11(α γ)2 H )] = 900 < 0. Therefore, the FTA acheves hgher welfare than MFN for the member countres. To sum up, we have the followng lemma (see also Fgure 2). 8 It holds that γ F (23 α /1256) 0.221α < γ M (α/2). 1

Lemma 5. For the FTA member countres = A, B, the natonal welfare under the FTA and the natonal standards compared wth the MFN s as follows. () If γ F (β) < γ < α, W M () If γ(β) < γ < γ F (β), W M () If 0 < γ < γ(β), W M > W F < W F. (W M > W F. < W F ) for β > (<) 23α/1256. Fgure 2: Comparson of the FTA members welfare before and after an FTA Lemma 5 can be nterpreted as follows. In regon I, where the member countres choose s L both under the MFN and FTA. Because of the elmnaton of tarffs wthn the FTA and the tarff complementarty effect, output and consumpton become larger under the FTA. Therefore, the member countres gans from the formaton of the FTA unless the negatve externaltes per unt of consumpton ndcated by the parameter β 15

are large enough. In regon III, where the member countres choose s H both under MFN and the FTA. In ths case, the negatve externaltes are nternalzed by the mposton of a hgh standard, and thus the FTA unambguously acheves hgher welfare to the member countres. In regon II, where the member countres choose s L under MFN but they choose s H under the FTA, the member countres face the frms hgher complance costs but the negatve externaltes wll be reduced under the FTA. If γ s relatvely hgh, the hgh complance costs harm the member countres, but f β s hgh, the escape from the negatve externaltes benefts the member countres. The possblty of welfare reducton n the member countres mples that may reduce the ncentve for the FTA. If we explctly consder the process of the formng FTA before the polcy game, the governments may not choose to form the FTA. Nonmember country In the nonmember country, the natonal government always chooses the MFN tarff and does not change the standard before and after the formaton of FTA. Therefore, comparsons of the nonmember s welfare are equvalent to those of ts export profts: W M C W F C = 2[πM C (s ) π F C (s )], = A, B, where π F C(s ) = [a(s ) 3t F (s )] 2 s the export proft n each member country s market. In both regons I and III, W M C < W F C holds because πm C (s L) π F C (s L) = (17α 2β)(3α + β)/900 < 0 and π M C (s H) π F C (s H) = 51(α γ) 2 /900 < 0. In regon II, where the members choose s L under the MFN and s H under the FTA, t holds that π M C(s L ) π F C(s H ) = (17α 1β 10γ)(3α + 1β 10γ). 900 Snce α > 2β, t holds that 17α 1β 10γ > 10(α γ) > 0. Therefore, the sgn of the above expresson s postve (negatve) f and only f γ > (<) (3α + 1β)/10 holds. Regon II s contaned n the set { (β, γ) R 2 + γ < (3α + 1β)/10, α > 2β }. Therefore, also n regon II t holds that W M C < W C F. To sum up, we have the followng proposton. Proposton 2. In comparson wth the MFN, the nonmember country of the FTA unambguously becomes better off under the formaton of an FTA.

Intutvely, Proposton 2 can be nterpreted as follows. When (β, γ) s n regon I or regon III, the nonmember s welfare mprovement stems from the tarff complementarty effect shown n Lemma 2. Because the members adopt the same standards under the FTA as under the MFN, the nonmember country wll face a lower tarff rate under the FTA than under the MFN, whch ncreases ts exports to the member countres and hence export profts of the domestc frm. If (β, γ) s n regon II, the member countres wll tghten up ther standards under the FTA, whch rases the unt producton cost of the frm n the nonmember country. At the same tme, as we have shown n (6) and smlar result holds for the comparson of t F (s H ) and t F (s L ), a more strngent standard lowers the optmal (external) tarff. Therefore, n regon II, the tarff rate that the nonmember country faces under the FTA becomes lower than the MFN tarff n two channels: one s the tarff complementarty effect and the other s a more strngent standards chosen by the FTA members. Because the unt cost s at a moderate level n regon II, the reducton n the tarff offsets the ncrease n the unt cost, and thus the nonmember country can ncrease ts export under the FTA..2 Harmonzaton of standards We now consder the second scenaro: Suppose that the FTA member countres harmonze ther standards n addton to elmnate tarffs between them. More specfcally, the member countres A and B determne the common standards s A = s B = s {s L, s H }, takng the standards n the nonmember country s C as gven, so as to maxmze the sum of the natonal welfare WA F + W B F. Notce that n advance to the determnaton of standards, the member countres choose ther respectve external tarffs ndependently. Snce W F s addtvely separable, the member countres determne the par of standards that maxmzes,j=a,b [wf (s) + π F j(s)] ω(s). Lemma 6. Suppose that countres A and B form an FTA and harmonze ther standards. Then, t s optmal for both member countres to choose s H f 0 < γ < γ F (β) and s L f γ F (β) < γ < α, where γ F (β) α α 2 202 23 αβ + 129 387 β2. 17

(Proof) In lght of (11), (12), and (13), we have ω(s H ) ω(s L ) = 387γ2 77αγ + 606αβ 23β 2. (15) 1 Therefore, ω(s H ) > ω(s L ) holds f and only f 387γ 2 77αγ + 606αβ 23β 2 > 0, or equvalently, gven the parameter restrctons, γ < γ F (β). Conversely, f γ > γ F (β), t holds that ω(s H ) < ω(s L ). Snce d γ F (0)/dβ = 101/129 0.7829 and γ F (α/2) = (1 359/6 3)α 0.518α, t follows that γ M (β) < γ F (β) < γ F (β) for all β (0, α/2). Ths mples that, n lght of Lemmas 3 and 6, regon II n Fgure 1 s further parttoned nto two subregons: one n whch s H s chosen by member countres both under natonal standards and harmonzaton (when γ M (β) < γ < γ F (β)), and the other n whch s H s chosen under the natonal standards but s L s chosen under harmonzaton (when γ F (β) < γ < γ F (β)). See also Fgure 3. Fgure 3: Harmonzaton of standards under the FTA 18

Proposton 3. The FTA member countres choose less strngent standards when they harmonze ther standards than when they choose ther respectve natonal standards n the sense that there exsts a set of parameters (β, γ) n whch the member countres choose a low standard under harmonzaton but choose a hgh standard under the natonal standards. Intutvely, Proposton 3 can be nterpreted as follows. If the FTA member countres determne ther respectve standards ndependently, ther objectve s to maxmze the domestc surplus w F (s ). By contrast, f the member countres harmonze ther standards, they mutually take the export proft of the FTA partner πj(s F ). Because an ncrease n the standard reduces the FTA partner s export proft, compared to the natonal standards, each member country should adopt less strngent standards under harmonzaton. Let us conclude ths secton by analyzng the welfare effects of the FTA wth harmonzaton of standards. For the nonmember country, t s verfed that Proposton 2 remans vald; the nonmember country unambguously gans from a formaton of FTAs. For the member countres, because γ M (α/2) < γ(α/2) < γ F (α/2) holds, we obtan the followng lemma. Lemma 7. For the FTA member countres = A, B, the natonal welfare under the FTA and harmonzaton of standards compared wth the MFN s as follows. () If γ F (β) < γ < α, W M () If γ(β) < γ < γ F (β), W M () If 0 < γ < γ(β), W M > W F < W F. (W M > W F. < W F ) for β > (<) 23α/1256. From Lemmas 5 and 7, we fnd that f β and γ are of moderate sze, the member countres welfare under the FTA becomes hgher than the MFN level when they jontly determne the standards even though the opposte holds when they determne ther respectve standards ndependently, as llustrated by the shaded area n Fgure. In other regons, such a reversal of welfare rankng between the MFN and the FTA does 19

not occur. The reversal of the welfare rankng mples that harmonzaton of standards wthn an FTA makes the formaton of the FTA more favorable. Fgure : Comparson of the welfare effects of FTA between natonal standards and harmonzaton Proposton. Regonal harmonzaton of standards ncreases the gans from an FTA formaton n the sense that there exsts a set of parameters (β, γ) n whch, for = A, B, W M W F > W F holds when the member countres seek ther natonal standards but W M < holds under harmonzaton of standards. Proposton mples that the FTA member countres have ncentve to harmonze ther standards. It may explan the recent ncrease n the FTA wth harmonzaton of standards. Notce that the less strngent standards are chosen by the members for the sake of the larger gans of the FTA. 20

5 Transboundary Externaltes In ths secton, we consder an extenson by ncorporatng transboundary consumpton externaltes nto our basc model. When consderng regulatons to lmt automoble exhaust emssons, such transboundary externaltes wll be an mportant aspect to deal wth. Let us denote the degree of transboundary externaltes by δ [0, 1]. Then, the representatve consumer s utlty functon s rewrtten as U(Q, Y ; s, Q ) = u(q ) + Y b(s ) Q j δb(s j) Q j and hence the natonal welfare (3) s rewrtten as W = [3a(s ) t j t k ] 2 32 a(s ) 3t j + t k + t j [ δ b(s j ) 3a(s j) t j t jk + [a(s ) + t j + t k ] 2 + t k a(s ) + t j 3t k + [a(s j) 3t j + t jk ] 2 + b(s k ) 3a(s k) t k t kj b(s ) 3a(s ) t j t k + [a(s k) 3t k + t kj ] 2 ]. () The above expresson mples that the presence of transboundary externaltes does not affect the frst-order condtons for optmal unlateral tarffs W / t j = 0, and thus the optmal tarff formulas,.e., (5) under the MFN and (11) under an FTA, are stll vald when transboundary externaltes exst. 5.1 MFN In the presence of transboundary externaltes, the natonal welfare under the MFN can now be rewrtten as W M = w M (s )+η M j (s j )+η M k (s k), where the domestc surplus w M (s ) s defned by (7) and η M l (s l ) π M l (s l ) δb(s l )Q l = [a(s l) 2t M (s l )] 2 δb(s l ) 3a(s l) 2t M (s l ), l = j, k (17) s the net export profts deducted the costs of transboundary externaltes. Notce that because the optmal standards depend on the comparson between w M (s H ) and w M (s L ), the replacement of π M l (s l) by η M l (s l) due to the presence of transboundary externaltes does not affect the condton under whch whether s H or s L s chosen. In other words, Lemma 1 s stll vald. 21

5.2 FTA wth natonal standards After the formaton of an FTA between countres A and B, each member country s natonal welfare can be rewrtten as W F = w F (s ) + η F j(s j ) + η M C (s C), = A, B, where the domestc surplus w F (s ) s the same as (12) and η F j(s j ) π F j(s j ) δb(s j )Q j = [a(s j) + t F (s j )] 2 δb(s j ) 3a(s j) t F (s j ) (18) s the export profts net of transboundary externalty costs assocated wth the FTA member country. Agan, the presence of transboundary externaltes does not affect the optmal choce of standards characterzed n Lemma 3. Therefore, Lemma also holds, and so does Proposton 1. That s, n comparson wth the MFN, an FTA makes the member countres to choose more strngent standards. The welfare effects of the FTA compared wth the MFN are obtaned by a comparson between W M and W F, as mplemented n secton.1. Gven the components of natonal welfare explaned above, the welfare effects can be dependent on the degree of transboundary externaltes, δ, as well as the other parameters β and γ. Let us begn wth the welfare comparson n the member countres. As n the case wthout transboundary externaltes, there are three possbltes,.e., the case where the member countres choose s L both under the MFN and FTA (regon I n Fgure 1), the case where the member countres choose s L under the MFN but they choose s H under the FTA (regon II), and the case where the member countres choose s H both under the MFN and FTA (regon III), dependng on the values of β and γ. In regon I, the dfference between W M and W F s W M W F = w M (s L ) + ηj M (s L ) [w F (s L ) + ηj(s F L )] (3α + β)[23α 8(157 + 210δ)] =, 100 from whch t holds that W M > W F (W M < W F ) f and only f β > (<) 23α/[8(157 + 210δ)]. The cutoff value for β becomes smaller f δ becomes hgher, mplyng that the 22

transboundary externaltes make the formaton of an FTA less benefcal for the member countres when these countres choose a low standard both under the MFN and FTA. In regon II, the dfference between W M and W F s W M W F = w M (s L ) + η M j (s L ) [w F (s H ) + η F j(s H )] = 2150γ2 + 300αγ 11α 2 196( + 15δ)αβ + 98(7 + 10δ)β 2, 900 whch becomes postve (negatve) f and only f γ > (<) γ(β, δ) α 7 1 5 86 α2 Note that γ(β, δ) s ncreasng n δ: 32 + 30δ 3 αβ + 7 + 10δ β 3 2. γ(β, δ) δ = 7β(3α β) > 0. 1 3 86 α2 32+30δ αβ + 7+10δ β 3 3 2 In other words, an ncrease n the degree of transboundary externaltes make the formaton of an FTA more benefcal for the member countres n the case where the member countres choose more strngent standard under the FTA than under the MFN. Fnally, n regon III, all countres choose s H both under the MFN and FTA and the analyss n the absence of transboundary externaltes remans vald snce b(s H ) = 0 n the present model. Whle transboundary externaltes narrow the regon n whch W M < W F holds for the member countres when these countres choose the low standard both under the MFN and FTA, the regon wth W M < W F expands when the member countres choose more strngent standard under the FTA than under the MFN. Intutvely, these fndngs can be nterpreted as follows. A formaton of an FTA ncreases consumpton n member countres, and so does the consumpton externaltes for a gven standard. In the presence of transboundary externaltes, the socal costs of externaltes become hgher f the member countres choose the same standard both under the MFN and FTA. Because of ths, transboundary externaltes wll have a pressure to reduce welfare under the FTA n regon I. In regon II, however, the member countres choose more strngent standards under the FTA than the MFN equlbrum, and n ths partcular 23

model, b(s H ) = 0 and hence the socal costs of externaltes dsappear completely under the FTA. Therefore, the advantage of the FTA n terms of the socal costs of externaltes ncreases as δ becomes hgher. Let us proceed to the nonmember country s welfare. Comparsons of the nonmember s welfare are now equvalent to those of ts export profts net of the socal costs of transboundary externaltes snce W M C W F C = 2[ηM C (s ) η F C (s )], = A, B, where η F C(s ) = [a(s ) 3t F (s )] 2 δb(s ) 3a(s ) t F (s ). If member countres choose s L both before and after the formaton of the FTA (.e., regon I n Fgure 1), t holds that η M C(s L ) η F C(s L ) = (3α + β)[51α 8(9 + 70δ)β]. 1700 As shown n secton.1, gven the condton α > 2β, the sgn of the above equaton s unambguously negatve f δ = 0. However, even f the condton α > 2β s satsfed, the sgn of the above equaton can be postve when δ > (51α 72β)/560β. The ntuton behnd ths result s smlar to the narrowng of the regon n whch W M < W F holds n regon I for the member countres: snce an FTA ncreases consumpton n member countres, the consumpton externaltes n these countres become larger, and f the degree of transboundary externaltes s suffcently hgh, the socal costs of transboundary externaltes may outwegh the beneft of the FTA whch the nonmember country can earn. In the case where the member countres choose s L under the MFN and s H under the FTA (.e., regon II), t follows that η M C(s L ) η F C(s H ) = (17α 1β 10γ)(3α + 1β 10γ) + 980δβ(3α β), 900 the sgn of whch s, n vew of the dscusson n secton.1, unambguously negatve for any δ [0, 1]. Therefore, as n the case of no transboundary externaltes, the nonmember country becomes better off under the FTA. Fnally, f the member countres choose s H both under the MFN and FTA, π M C (s H) π F C (s H) = 51(α γ) 2 /900 < 0 holds, as n the case of no transboundary externaltes, because b(s H ) = 0. 2

Proposton 5. In comparson wth the MFN, the nonmember country of the FTA may be worse off under the formaton of an FTA f γ F (β) < γ < α and f the degree of transboundary externaltes s suffcently hgh. 5.3 FTA wth harmonzaton of standards As dscussed n secton.2, the FTA member countres jontly determne the common standards s A = s B = s so as to maxmze the sum of ther natonal welfare W F A + W F B. In the presence of transboundary externaltes, the problem s equvalent to maxmze,j=a,b [wf (s) + η F j(s)] ω(s). It s verfed that ω(s H ) ω(s L ) = ω(s H ) ω(s L ) + δb(s L)[3a(s L ) t F (s L )] 2 = 387γ2 77αγ + (606 + 630δ)αβ (23 + 2δ)β 2 1 (19) hold, where ω(s) =,j=a,b [wf (s) + π F j(s)] as s defned n secton.2. Therefore, n vew of Lemma 6, t follows that the cutoff value for γ s γ F 202 + 210δ 23 + 2δ (β, δ) α α 2 αβ + β 129 387 2 and f γ s hgher than ths value, t s optmal for both member countres to choose s L, whle the member countres choose s H f γ s lower than the cutoff value. It s easly verfed that ths cutoff value s ncreasng n δ. Ths mples that, as expected, the exstence of transboundary externaltes motvates the FTA member countres to choose more strngent standards f they harmonze ther standards. Compared wth the cutoff value n the case of natonal standards under the FTA, t follows that γ F (β) γ F (β, δ) = α 2 202 + 210δ αβ + 129 23 + 2δ β 387 2 α 2 2αβ + β2 15 > 0 (< 0) 15α( 15δ) β(1 15δ) > 0 (< 0). Therefore, n contrast to the case wthout transboundary externaltes, γ F (β, δ) can be 25

larger than γ F (β) f the degree of transboundary externaltes satsfes 9 60α β 15(15α β) < δ < 387α2 606αβ + 23β 2. 2(15α β)β That s, for suffcently hgh values of δ, there can exst a set of parameters (β, γ) n whch the member countres choose s H under harmonzaton but choose s L under the natonal standards, or n other words, Proposton 3 may not hold. The possblty that γ F (β, δ) can be larger than γ F (β) mples that the rankng of welfare gans from an FTA between natonal standards and harmonzaton, llustrated n Fgure, can be reversed; f γ F (β, δ) > γ F (β), there exsts a set of parameters (β, γ) n whch, for = A, B, W M < W F holds under natonal standards but W M > W F holds under harmonzaton. Proposton 6. If the degree of transboundary externaltes s suffcently hgh, regonal harmonzaton of standards may nduce the FTA member countres to choose more strngent standards than those under natonal standards and may reduce the gans from an FTA formaton. 6 Concludng Remarks In ths paper we consdered a three-country model of nternatonal olgopoly wth endogenous determnaton of tarffs and standards on products that cause negatve consumpton externaltes n the presence of possble FTA. The man results that we obtaned are as follows: Compared wth the MFN, an FTA makes the member countres to choose more strngent standards. Compared wth the MFN, the FTA member countres may or may be better off under the formaton of an FTA, whle the nonmember country becomes better off for the case of low degree of transboundary externaltes. 9 The left-hand sde of ths nequalty comes from 15α( 15δ) β(1 15δ) < 0. The rght-hand sde comes from the constrant that γ F (β, δ) should be a real number for gven parameter values. Notce that f β < α/2, t holds that 0 < (60α β)/[15(15α β)] < 1. 26

Harmonzaton of standards wthn an FTA wll lead the member countres to choose less strngent standards and make the formaton of the FTA more favorable, provded the degree of transboundary externaltes s not so hgh. These results wll have mplcatons for the recent movement towards regonalsm that takes harmonzaton of domestc polces nto consderaton as well as lberalzng trade n goods. Throughout ths paper we have focused on FTAs as a form of preferental trade agreement, and have not consdered the case of a customs unon (CU), where member countres set a common external tarff, harmonzng ther external trade polcy. Our next task s to examne the effects of CU wth endogenous determnaton of standards. Of another nterest s the analyss of dynamc tme-path problem that consders whether regonal standards are stumblng blocks or buldng blocks toward multlateral harmonzaton. There are also a number of possble extensons of the basc model, e.g., dfference n preferences and technologes among countres, generalzaton of demand functons, ntroducton of qualty mprovng R&D, and sequental determnaton of trade and domestc polces. Appendx A.1 Smultaneous determnaton of tarffs and standards In ths Appendx, we show that when the governments smultaneously determne ther tarffs and standards, the optmal polcy mx s the same as that under the sequental polcy choce. Let us consder the polcy game under MFN. The government n country chooses t and s so as to maxmze (). If s = s H, () s rewrtten as W = [3(α γ) 2t ] 2 32 + (α γ + 2t ) 2 + t α γ 2t 2 + Π, (A.1) whereas t s rewrtten as W = (3α 2t ) 2 32 + (α + 2t ) 2 + t α 2t 2 β 3α 2t + Π (A.2) 27

f s = s L, where Π {[a(s j ) 2t j ] 2 + [a(s k ) 2t k ] 2 }/. A drect comparson between (A.1) and (A.2) reveals that t s optmal for country to choose s H (s L ) f and only f t < (>) γ(2α + 11γ)/[(3γ + β)]. In lght of ths fact, the relaton between t and W can typcally be llustrated as a curve wth twn peaks (see Fgure A.1). Fgure A.1: Relaton between t and W From (A.1) and (A.2), the local maxma are attaned at t = 3(α γ)/10 (when s = s H ) and t = (3α + β)/10 (when s = s L ). Substtutng t = 3(α γ)/10 nto (A.1) and t = (3α + β)/10 nto (A.2) respectvely yeld 2(α γ) 2 + Π f s = s H, max W = 5 t α 2 6αβ + β 2 + Π f s = s L. 10 (A.3) It s clear from (A.3) and Fgure A.1 that f 8αγ +γ 2 +6αβ β 2 s postve (negatve), s H (s L ) acheves the maxmum level of W. Ths means that the optmal polcy mx, ( ) whch s the domnant strategy, for each country under MFN s (t M, s M ) = 3(α γ, s 10 H f γ < α α 2 3 β2 αβ + and 2 (tl, s L ) = ( 3α+β, s ) 10 L f γ > α α 2 3 β2 αβ +. However, 2 the condton whether γ s hgher or lower than α α 2 3 β2 αβ + s equvalent to 2 the condton mposed n Lemma 1. Therefore, the tmng of the determnaton of polcy nstruments does not matter. 28

The above-mentoned equvalence between the smultaneous determnaton of polcy nstruments (.e., tarffs and standards) and sequental determnaton of these polces also holds n the polcy games under an FTA between countres A and B. References [1] Baghdad, L., I. Martnez-Zarzoso, and H. Ztouna, Are RTA agreements wth envronmental provsons reducng emssons? Journal of Internatonal Economcs 90 (2013), 378 390. [2] Bagwell, K. and R. Stager, Regonalsm and Multlateral Tarff Cooperaton, n J. Pgott and A. Woodland (eds.), Internatonal Trade Polcy and the Pacfc Rm, London, Macmllan, 1998. [3] Costnot, A., A Comparatve Insttutonal Analyss of Agreements on Product Standards, Journal of Internatonal Economcs 75 (2008), 197 213. [] Facchn, G., P. Slva, and G. Wllmann, The customs unon ssue: Why do we observe so few of them? Journal of Internatonal Economcs, forthcomng. [5] Fscher, R. and P. Serra, Standards and Protecton, Journal of Internatonal Economcs 52 (2000), 377 00. [6] Freund, C., Multlateralsm and the Endogenous Formaton of Preferental Trade Agreements, Journal of Internatonal Economcs 52 (2000), 359 76. [7] Krshna, P., Regonalsm and Multlateralsm: A Poltcal Economy Approach, Quarterly Journal of Economcs 113 (1998), 227 51. [8] Ornelas, E., Trade Creatng Free Trade Areas and the Undermnng of Multlateralsm, European Economc Revew 9 (2005a), 1717 1735. [9] Ornelas, E., Endogenous Free Trade Agreements and the Multlateral Tradng System, Journal of Internatonal Economcs 67 (2005b), 71 97. 29

[10] Sagg, K., Preferental Trade Agreements And Multlateral Tarff Cooperaton, Internatonal Economc Revew 7 (2006), 29 57. [11] Stoler, A.L., TBT and SPS Measures n Practce, n J.-P. Chauffour and J.C. Maur (eds.), Preferental Trade Agreement Polces for Development: A Handbook, 2011, World Bank. [12] Takarada, Y., Y. Kawabata, A. Yanase, and H. Kurata, Standards and Trade: Multlateralsm versus Regonalsm, mmeograph (201). [13] Y, S.-S., Endogenous Formaton of Customs Unons under Imperfect Competton: Open Regonalsm s Good, Journal of Internatonal Economcs 1 (1996), 153 177. [1] Y, S.-S., Free-Trade Areas and Welfare: An Equlbrum Analyss, Revew of Internatonal Economcs 8 (2000), 336 7. 30