Common Markets and Trade Liberalization

Common Markets and Trade LiberalizationAuthor(s): Costas HadjiyiannisSource: The Canadian Journal of Economics / Revue canadienne d'Economique, Vol. 37, No. 2(May, 2004), pp. 484-508Published by: Wiley on behalf of the Canadian Economics AssociationStable URL: http://www.jstor.org/stable/3696157 .

Accessed: 17/06/2014 22:55

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Wiley and Canadian Economics Association are collaborating with JSTOR to digitize, preserve and extendaccess to The Canadian Journal of Economics / Revue canadienne d'Economique.

http://www.jstor.org

This content downloaded from 195.78.108.60 on Tue, 17 Jun 2014 22:55:02 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=black

http://www.jstor.org/action/showPublisher?publisherCode=cea

http://www.jstor.org/stable/3696157?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


Common markets and trade liberalization Costas Hadjiyiannis Department of Economics, University of

Cyprus

Abstract. The GATT Rounds show that trade liberalization is essentially a cooperative non-stationary dynamic process. Therefore, the impact of Regionalism on trade liberalization possibly changes over time. I adapt the trade liberalization model of Devereux (1997) to examine how this impact varies. Common markets lead to a one-time shock in immediate tariffs, as well as to a change in their rate of decline. I find that common markets that happen late in the trade liberalization process are more likely to lead to a decline in immediate tariffs. Common markets also increase the rate of decline of tariffs after their formation. JEL Classification: F03, F15

Marches communs et libdralisation du commerce. Les diverses rondes de negociation du GATT ont montre que la liberalisation du commerce est essentiellement un processus dynamique non-stationnaire de cooperation. Donc l'impact du regionalisme sur la liberalisation du commerce peut changer dans le temps. L'auteur utilise le modele de liberalisation du commerce de Devereux (1997) pour analyser comment cet impact varie. Les march6s communs d6clenchent un choc imm6diat dans les droits de douanes en place ainsi qu'un changement dans leur taux de declin. Il semble que les marches communs qui se materialisent tard dans le processus de liberalisation du commerce sont davantage susceptibles d'entrainer un declin immediat dans les droits de douane. Les marches communs accroissent aussi le taux de reduction dans les droits de douane apr6s leur formation.

I am grateful to Robert Staiger, Scott M. Taylor, Michael S. Michael, Sofronis Clerides, Alex Michaelides, and seminar participants at the University of Wisconsin - Madison and the University of Cyprus for helpful comments. Email: [email protected].

Canadian Journal of Economics / Revue canadienne d'Economique, Vol. 37, No. 2 May / mai 2004. Printed in Canada / Imprime au Canada

0008-4085 / 04 / 484-508 / ? Canadian Economics Association



Common markets and trade liberalization 485

1. Introduction

The impact of regional trade agreements or, more accurately, preferential trade agreements (PTAs), has been one of the hottest topics in the Inter- national Trade literature in the last fifteen years. The main reason is the proliferation of these agreements during this period. As of June 2002 over 200 PTAs were registered with the World Trade Organization (WTO) and more than 150 of them are still in effect. The WTO reports that nearly all of its 146 members are involved in at least one. The WTO itself seems to have an overall positive view of PTAs. Article XXIV of GATT, paragraph 4, states: 'The contracting parties recognize the desirability of increasing freedom of trade by the development, through voluntary agreements, of closer integration between the economies of the parties to such agreements.' This article essentially grants an exception to PTAs from article I, which is commonly referred to as the Most Favored Nation (MFN) clause, as long as the PTA applies to almost all trade between the member countries and as long as tariffs on non-members are not increased. In addition, the Enabling Clause grants exceptions to PTAs between developing countries and article V of GATT exempts PTAs in the area of services.

Academics, on the other hand, are a lot more sceptical about the impact of these agreements. Panagariya (2000) (whose categorization I follow) and Krueger (1999) provide extensive reviews of the literature on PTAs, which can be divided into two major categories (see also Bhagwati and Panagariya 1996). The first deals with the welfare effects of these agreements and the second with the impact of these agreements on the multilateral trade liberalization process of the WTO. The literature on the welfare effects of PTAs was sparked by Viner (1950), who divided the effects of PTAs into trade creation and trade diversion. Most of the papers in this line of research are static, and the general conclusion is that PTAs may reduce the welfare of non-members. Important contributions in this line of research include Meade (1955), Lipsey (1957), Krugman (1991), and Richardson (1993). Some of these papers are also indirectly dealing with the impact of PTAs on trade liberalization, since they consider their effects on external tariffs. There are also a number of papers on necessarily welfare-improving PTAs, which basically establish the fact that welfare improving PTAs always exist (e.g, Kemp and Wan 1976; Bond, Syropoulos, and Winters 2001).

The second category of papers in this literature deals directly with the relationship between PTAs and multilateral trade liberalization. Bhagwati (1993) characterizes this issue as the dynamic time path issue or, in other words, the issue of the shortest path to free trade. This line of research can be classified into three categories. The first assumes no interaction between the multilateral process and PTAs and investigates whether PTA formation will eventually lead to one huge PTA covering the whole world (e.g., Baldwin 1995; Bond and Syropoulos 1996). The second category assumes



486 C. Hadjiyiannis

that the two processes, regional and multilateral, interact, and it investigates the impact of PTAs on multilateral trade liberalization. Within this line of research there are at least two distinct approaches. The first looks at political economy considerations to investigate whether PTAs render infeas- ible an initially feasible multilateral trade liberalization. Important contributions in this area include Levy (1997) and Krishna (1998).

Another approach is to concentrate on terms of trade games and the self-enforcing nature of the multilateral process. The multilateral process is modelled as an infinitely repeated tariff game, in which the threat of future punishments leads to lower tariffs. Within this framework, these authors investigate the impact of PTA formation on multilateral tariffs (e.g., Bagwell and Staiger 1997a,b). This paper belongs to this group. Finally, some authors investigate the impact of multilateral trade liberalization on PTA formation by endogenizing the decision to form a regional agreement, such as Freund (2000). This is also investigated, among other things, in Ethier (1998). (See Bhagwati, Krishna, and Panagariya 1999 for reprints of several papers mentioned.)

The key innovation of this paper is the introduction of PTA formation in the context of a non-stationary multilateral process. The observed multilateral trade liberalization process is non-stationary, with tariffs falling over time. What is missing from the literature is the analysis of the impact of regionalism over time. The only way to investigate this is within the context of a non-stationary dynamic trade liberalization model. I adapt such a model from Devereux (1997) to include more than two regions and then allow some of the regions to engage in regionalism. In Devereux (1997) human capital accumulation is the driving force behind the trade liberalization process. Human capital accumulation makes it increasingly costly to revert to autarky, and that threat leads to cooperative tariffs that decrease with time. The formation of common markets leads to a market power effect that increases both the benefit and the cost of deviating from a cooperative tariff. The latter decreases with time, while the former does not. This means that common markets that happen later, rather than earlier, are more likely to lead to a decrease in immediate tariffs. I also find that common markets increase the rate of decline of cooperative tariffs.'

The paper is organized as follows. In section 2 I present the basic model and derive the competitive equilibrium of the model given some tariffs. In section 3 I set up and solve for the trade liberalization process. These two sections extend Devereux's (1997) model to allow for more than two regions. The impact of common markets on the multilateral trade liberalization process is analysed in section 4. In section 5 I present the conclusions.

1 Another author who derives a non-stationary trade liberalization process is Staiger (1995). Hadjiyiannis (1999a) finds similar results using the same methodology applied to the trade liberalization model of Staiger (1995).




2. The model

There are two types of regions in the world, home and foreign. These regions are common markets in the sense that tariff decisions are taken at the regional level.2 In this model, governments set tariffs to maximize joined welfare.3 An asterisk denotes foreign regions. There are R such regions of each type in the world. There are only two goods in the world denoted by the subscripts 1 and 2. There is a measure (1/R) individuals in each region.4 Total consumer utility for each home region is

Ut = CtC2t (1) R

where cit is the total consumption of good i in period t. Regions of the same type are identical, so there is no trade between them in

symmetric equilibria. However, consumers trade with the residents of regions of the other type. I assume that there are no international capital markets.

The production side of the economy is given by the following Ricardian production technologies:

ylt = atilt (2)

Y2t = btl2t (3)

lit + 12t = (4) R

where ylt is the home production of good i and lit is the total labour allocated to the production of that good.5

Labour productivity changes over time depending on the current state of sector-specific technical knowledge. This technical knowledge accrues through two channels. The first is internal to the region, and it accrues because of domestic production of the good. This could be due to learning-by-doing or domestic trade in ideas. The second channel is knowledge spillovers from other regions because of international trade in ideas, reverse engineering (whenever the two regions trade) or sharing of information through trade shows, Email, site visits, and so on. The state of technical knowledge in each sector is characterized by

2 In Section 4, I explain why I choose to call these regions common markets as opposed to customs unions.

3 The term 'government' refers to the regional authorities in this paper, since individual countries are not modelled.

4 This assumption means that the total population of all regions of each type is 1. 5 Utility and technologies are defined symmetrically for foreign regions as Ut = 1/Rc*,c*t,

Ylt = btlr,, yjt = atr2, t', +t lt = 1/R.



488 C. Hadjiyiannis

at = ahat where hat = hat-l(l + Ullt-1 + eOLit-1) (5)

bt = phbt where hbt = hbt-l(l + 012t-1 + aOL2t-1) (6)

at- = ahat where hat = h*atl(1 + c?ltl +

rOLzt_1) (7)

b = hbt where hbt = hbt_(1 + al-1 +

OLt-1), (8)

where hi, denotes the sector-specific human capital accumulated in industry i at time t, for i = a, b, and at and b, denote the corresponding stocks of human capital. Lit represents the labour allocated to the production of good i by all other regions, home and foreign. Industry a represents the industry in which the region has a comparative advantage. Without loss of generality, let that industry produce good 1 in home regions and good 2 in foreign regions. Human capital accumulation depends on the share of labour devoted to the production of each good. The more a region specializes in that good the higher the growth rate of human capital is in that sector. The parameter a determines how fast human capital can grow. In addition, human capital accumulates as a result of international spillovers within the specific sector. The parameter 0 determines how fast human capital grows due to international spillovers. It is assumed that Oe(0, 1). This implies that internal spillovers increase human capital faster than external ones. Rivera-Batiz and Romer (1991) provide a formal justification for this assumption. They consider trade in ideas and suggest that national borders restrict such trade. More informal explanations can be based on the fact that the flow of information is simply much easier within a region than between regions.6

The game in every period is then played as follows: First the workers allocate themselves between industries given anticipated wage rates. Then governments take these labour allocations as given and choose their optimal tariffs. This ensures that the impact of tariffs can not be avoided by ex post factor allocation.7 Finally, firms maximize their profits given these labour allocations and tariffs. At the end of the period, good markets clear determining prices and wage rates. All agents have perfectly rational expectations, so their anticipated wages coincide with the actual wage rates at the end of the period. The objective of the governments is to maximize the discounted utility of their citizens given a discount factor be[0, 1].

2.1. Equilibrium I derive the competitive equilibrium of the one-stage game for each of the following special cases: autarky (prohibitive tariffs), free trade (zero tariffs), and positive but non-prohibitive tariffs. Throughout the analysis the following conditions are imposed on the parameters:

6 This assumption is going to prove crucial to the results of the paper. Basically, if 0 = 1, there are no growth gains from trade. I postpone the discussion of its importance until the next section, where I discuss the gains from trade.

7 Staiger and Tabellini (1987) and Lapan (1988) examine this issue. Tariff policies may not be time-consistent if factor allocations take place after tariffs are imposed. In that case, ex-post factor allocations may partly reverse the impact of tariffs.




6(1 + a)2 < 1 (9)

ha0 = h~0 and hbO = h*0 (10)

phbO < 1. (11) ahao

Equation (9) ensures that under all possible cases the sum of discounted welfare is positive and finite. Equation (10) imposes symmetry between home and foreign regions and equation (11) ensures that home regions have a comparative advantage in good 1 and foreign regions in good 2.

2.1.1. Autarky Consider a time period t. In autarky there is no trade between regions. This implies that every region is producing both goods. Let good 2 be the numeraire good and Pt be the price of good 1. Consumer utility maximization implies that

PA4Clt = C2t. (12)

Since both goods are produced in every region, real wages between the two sectors must be equal. This implies that Pt = b,/at. Market clearing implies that cit= a,lt, and c2t = btl2t. These three conditions lead to lt,= 12t = (1/2R). The per period welfare is U,= (1/4R3)atb,. The sum of discounted welfare as viewed from period 0 is

VA ahaohbo (13) 1 - 6(1 a+ 2

+ -)2

2.1.2. Free trade In this case the regions specialize in their comparative advantage good and trade with regions of the other type for the other good. In any period t, world output of each good is a,. Symmetry implies that the relative price of good 1 is 1 and that each region consumes (a,/2R) of each good. The per period welfare of each region is Ut = (1/4R3)a2, and the sum of the discounted welfare as viewed from period 0 is

1 C2h2 VA 4R3 a h(14)

1 - 6(1 +?E + aOR•1)2

Equations (13) and (14) demonstrate the gains from trade. These can be divided into static and dynamic. Comparing the numerator in the two expressions, one can see the static gains from trade. In the free trade case regions specialize in the first period, so they do not waste resources on the low-productivity sector. Comparing the denominators, one can see the dynamic gains from trade. For 0 < 1, (1 + (a/ R) + aO(R - 1)/R)2 > (1 + (a/2R) + aO(2R - 1)/2R)2. In other words, the growth



490 C. Hadjiyiannis

rate of output under free trade is higher than that under autarky. By producing more of their comparative advantage good at home, regions accumulate technical knowledge faster, leading to higher productivity growth. These dynamic gains from trade are going to be driving the trade liberalization process. If 0 = 1, the two growth rates are the same and there are no dynamic gains from trade. This is because no matter where production takes place, all regions get the full benefit of it, whether or not they specialize. In that case, opening up trade leads only to static gains.

2.1.3. Positive tariffs Let 7 be the gross tariff rate. This means that if tariffs are zero, 7 = 1. Also, assume that the government of each region distributes tariff revenues to consumers as lump-sum transfers. Let T, be these transfers for a home region. Let

yi be the production of good i. Then

Tt=(r,- 1) (C2t -Y2t) and the budget

constraint faced by consumers is

PtClt + TtC2t = PtYlt + rtY2t + Tt. (15)

Equations (12) and (15) give equilibrium consumption as

c r=

t (Ptyli + y2t) (16) (1 + ?t)Pt

1 C1 = (Ptt +2t) (17)

1

Clt = (Ptylt Y) (18) (I +vt*)P,

c2 =+) (Ptytr

+ yt). (19) (1 + 7-*)

Market clearing implies that the sum of production of each good must be equal to the sum of the consumption for the same good. Market clearing and equations (16) through (19) give

ZEj 7tj'Y2t" + ZEl • Ij yrtj 1 7+rY2t1 Pt = 1 (20) ZITh-,tY1I E I+ ZiY*t

Equilibrium labour allocation is given by

1 1 lit = if Ptat > bt, t l = if PtbtT* > at

t if Pat = btr l if Pbt = at(21)

llt = 0 if Pta,

> btr lt

= 0 if Pbtr < at.




Good 1

Price that would cause specialization

PPF1

PPF2

Good 2

FIGURE 1

The inequalities in (21) determine the tariff ranges for which specialization is possible. A competitive equilibrium is {cit,

ci, Pt, lit, li Tt, t7-, t t, for i= 1, 2 and

t= 1,..., oo, that satisfies consumer utility maximization, government budget constraints, market clearing and labour market equilibrium conditions.

Figure 1 illustrates this simple Ricardian set-up. The Production Possibility Frontiers (PPFs) for each type of region are linear and trade is advantageous for both types of regions if the price of good 1 is between (bt/at) and (a,/bt). In that case, every region completely specializes in its comparative advantage good. Otherwise, there is no trade and each region is in autarky.

3. The trade liberalization process

In this section, a trade liberalization process is derived by letting governments play a repeated tariff determination game. Governments are faced with an identical stage game in every period that differs only in the value of the state variable. The state variable is the value of human capital. Governments make their tariff decisions after workers decide on labor allocations. I examine equilibria in which the regions trade with regions of the opposite type. Note that, because of the Ricardian technologies, trade implies complete specialization in the comparative advantage good. In other words, labour allocations do not change. For this reason, there are no time-consistency problems with the tariff choices of governments.

Regions coordinate their tariff decisions, subject to the condition that any such agreement is self-enforcing. This can be thought of as a multilateral trade liberalization process, facilitated by the World Trade Organization (WTO). Strategies in this game are history dependent. In other words, regions cooperate until there is a deviation. If any region deviates, then they all switch to a



492 C. Hadjiyiannis

punishment tariff forever. In other words, regions impose the cooperative tariff Tct if all regions cooperated for all the periods up to t. Otherwise, they all revert to autarky forever. I focus on equilibria that satisfy the following conditions:

1. Equilibrium is symmetric and subgame perfect. (All regions foreign and domestic choose the same tariff).

2. If a deviation occurs then all regions revert to autarky i.e. impose prohibitive tariffs.8

3. From the equilibria satisfying the above, select the one with the lowest cooperative tariffs.

In such an equilibrium, all regions cooperate and specialize in the production of their comparative advantage good. The period welfare from cooperating is

Tct a2 Uct = t (22) (1 + ct)2 R3

If all other regions are cooperating while region X deviates by choosing TDt, the relative price of good I is

R(1 + TDt) Pt = . (23) (1 + rct)+ (1 + TDt)(R - 1)

The period welfare from deviating is

R-Dt a2

UDt - ntt (24) S(1 + Dt)(l +

ct) + (1 + nt)2(R

- 1)R3

If a region is going to deviate, it will pick rDt to maximize (24). Maximizing (24) gives TDt as9

TDt = . (25) v/R--1

The benefit from deviating is the difference between what a region makes if it deviates and what it makes if it cooperates. This is given by

8 It can be shown that autarky is a Nash equilibrium of the stage game and that it is the maximum punishment that regions can impose in a subgame perfect equilibrium.

9 Note that in this framework prohibitive tariffs are infinite. The intuition is as follows. Tariff decisions are taken after labour is allocated between sectors. With Ricardian technologies, trade implies complete specialization so all labour is always allocated to the production of the comparative advantage good. The imposition of a tariff after labour allocations are made affects only consumption, since production is fixed. For the specific utility function in equation (1) both goods are always demanded even if the price of one of them is arbitrarily large.




R7-t a2 UDt - Uct RTDt

(1 + TDr)(1 + Tct) + (1 + TDt)2(R - 1) R3

Tct a2

(1 + ct)2 R3"

(26)

Equation (26) makes sense only if it is decreasing in Tct. This ensures that an increase in the cost of deviating results in a decrease in equilibrium tariffs (which increases the benefit of deviating by an equal amount). The following condition ensures that this is always the case:

-Tc, argmin [UDt - Uct] Vt. (27) rct

The discounted welfare from cooperating from period t + 1 forever as viewed from period t is

o-p0( 1 -R

2

ct+1l+i)2

2 (28)

The discounted welfare from autarky from period t + 1 to infinity as viewed from period t is

1 babt ( + + a•OR-)(1

+ (9O) Up 4R3

1-6(1 + a+ o2

1)---)2 The cost of deviating is the difference between the future welfare lost due to

cooperation and the future autarky welfare. This is

a2

00R-" Rt+2I+i Ucoop - UP = 6 1

Tct+ t

+i)+

1 6ab,(1 + + 6BR-)(1+ AO) lR_3

R R 2 (30) 4R3 1 (1l+ +U02•

The equilibrium of the game as defined above just balances the cost of deviat-

ing with its benefit. Setting (26) equal to (30) and simplifying, we get

Dr Dt ct

(1 + Dot)(1 + ct) + + + TDt)2(R - 1) (1 + Tct)2

WR-1 Tct+l+i

=O6 6 + a+(( + + 7ct+l,+•+)2

=1 (1O R R ))(I(31) 4R3 ahao (1 + + 0-)

1 - (1+

+ 4 +

--)2 " 1



494 C. Hadjiyiannis

Let VDt be the left-hand side of (31) and Vc be the right-hand side of (31). Also, let U,, be the second term of V,.

PROPOSITION 1. Ups is decreasing in t.

Proof. Recall that 0 < 1. This implies that

(1?+o0)< (1+ -R+a oR 1

for all R. Therefore,

R <R . (I+?ao)/Q(?? + R+ 1)R<1.

This means

((I )/( R-t+

+

is decreasing in t. All other terms in Ups are independent of t which implies that

Ups is decreasing in t.

Proposition 1 implies that the cost of deviating increases with time. As time goes on, the productivity gap between the good a region is specializing in and the good it is not producing is increasing. This is because human capital in the former is accumulating faster because more of its production takes place internally. Reverting to autarky is therefore increasingly costly with time. This is demonstrated in figure 2.

PROPOSITION 2. Equilibrium cooperative tariffs fall over time.

Proof. From proposition 1, the cost of deviating increases with time. From equation (31) VDt, or the benefit of deviating is independent of time. Since VDt is decreasing in

rct and Vc is increasing in Tct, a lower rct

sets the two sides of equation (31) equal.

Proposition 2 establishes a trade liberalization process that resembles that of GATT, with equilibrium cooperative tariffs falling over time. o In other words, self- enforcing agreements lead to decreasing tariffs over time. Figure 2 illustrates the intuition behind this result. As time goes on productivity in each of the two sectors increases. However, because each region is specializing in its comparative advantage good, productivity in that sector increases faster. This, of course, is a direct consequence of the fact that internal spillovers are more efficient than external ones

10 Whether free trade will eventually be achieved depends on the parameters. If the parameters are such that free trade is never achievable, cooperative tariffs can be shown to converge to a low but positive cooperative tariff.




Good 1

time increases

Good 2

FIGURE 2

(0 < 1). This reinforces each region's comparative advantage over time. Also, this period's tariffs do not affect next period's growth, for as long as regions keep specializing and trading. So, as long as tariffs are non-prohibitive and regions specialize, the productivity gap between the two sectors increases over time.

This increases the cost of reverting back to autarky, because in that case the region will have to produce the low-productivity good as well. This means that as time goes on punishments become more and more severe. As punishments become more severe, higher levels of cooperation can be supported, leading to decreasing cooperative tariffs with time.

This same logic would apply to a Heckscher-Ohlin world. In that case, trade would increase the production of the comparative advantage good and decrease that of the other good, compared to autarky. Over time, the comparative advantage of the region would be reinforced because of knowledge spillovers, for the same reasons as above. In other words, the increased internal production of the comparative advantage good increases factor productivity faster than that of the other good. That leads to increased specialization in that good, increasing the productivity gap between the two goods. Therefore, reverting to autarky becomes increasingly costly, and cooperative tariffs decrease over time.

4. Common market formation

Within the trade liberalization process established above, regions can form common markets. In the real world, common markets differ from customs unions in that they allow free mobility of factors of production. However,



496 C. Hadjiyiannis

factor mobility is not explicitly modelled in this paper and common markets just set tariffs to maximize regional welfare. Human capital accumulation is a key feature in this model and the results depend on the assumption that internal knowledge spillovers are more efficient than external ones. In the context of this model regionalism only happens between regions of the same type. Since these regions are identical, there is no trade between them. To better justify the fact that internal knowledge spillovers are more efficient than external ones, I assume that all regionalism is in the form of common markets, and factor mobility, which is not explicitly modelled, is one of the factors that speeds up human capital accumulation.

Common market formation is represented in the model by a fall in the number of regions from Ro to R1. To keep the model symmetric at all times, I make the simplifying assumption that R is the same for foreign and home regions before and after the formation of common markets. Common markets between regions of opposite types are not investigated. This assumption allows me to isolate the impact of the market power effect of common markets. Other authors use similar methods of modelling preferential trade agreements. Bagwell and Staiger (1997a) use the same method to model the formation of Customs Unions. Bagwell and Staiger (1997b) use agreements with non-modelled countries and do not allow free trade areas between their two modelled countries. To investigate trade liberalization after the formation of regional trade pacts we need to make sure that regions do not all integrate into one big region. One of the simplest ways of doing that is to assume only two traded goods and two types of regions that can not enter into regional pacts with regions of the opposite type."

Common markets, and regionalism in general, are usually associated with a market power effect. This market power effect is a consequence of the larger size of the regions. In this model, the market power effect is linked to the impact of Common markets on the rate of human capital accumulation. A larger region has more to gain by deviating from the multilateral agreement because the benefits from increased tariffs apply to a larger market. Deviating does not affect human capital accumulation in the deviating period. At the same time, a larger region has more to lose in the punishment phase because reverting to autarky is more costly for a larger region. This is because larger regions produce a larger share of world output internally, and, therefore, the productivity gap between the import and export goods increases more rapidly.

Rivera-Batiz and Romer (1991) investigate the links between regionalism and growth. I investigate the impact of these links on the market power effect. Common markets have two effects. The first effect is a once-off level effect.

11 This seems to correspond very loosely to the pattern of regionalism observed in the real world, where we observe agreements between similar countries. One notable exception is that of NAFTA.




This arises from the fact that bigger regions have more workers and therefore produce and consume more of the good. Owing to the structure of the utility function, this leads to a once-off increase in income per capita. However, this effect has no impact on trade liberalization, because it affects VDt and V, (equation (31)) proportionately.

The second effect is an increase in the rate of growth of human capital. Larger regions produce more domestically, which increases the rate of growth of productivity. Again, this is a product of the fact that the flow of technical knowledge is more efficient internally than externally. This affects different parts of equation (31) differently and has an impact on the trade liberalization process. It is precisely this effect that I investigate.

Before proceeding to examine that issue, it is important to address the reasons why common markets exhibit this growth effect. After all, what is it that makes the spillover of technical knowledge faster after the creation of a common market? A formal explanation can be found in Rivera-Batiz and Romer (1991). They consider trade in ideas as well as trade in goods. Governments restrict trade in ideas for the same reasons they restrict trade in

goods, that is, to give domestic producers of ideas an advantage over foreign producers. In the presence of common markets such restrictions are lifted and ideas are traded freely. This leads to an increase in the production of ideas and therefore an increase in human capital accumulation.

A number of informal explanations can also be used to explain this. In a common market labour can move freely between the two regions. Hiring labour from what used to be another region increases knowledge spillovers. Labour movements, however, are not modelled here. Also, one can argue that there is a closer relationship between firms within a common market. For example, firms in the same industry probably belong to the same organizations and take part in more of the same trade shows.

For the rest of the analysis I ignore the one-time level effect because it has no impact on trade liberalization and concentrate on the growth effect of common markets. The timing and the decision to join a common market is exogenous to the model. I assume that external political factors determine the decision of if and when to join a common market. Many analysts argue that this assumption is not that far from the truth. They argue that political reasons, as opposed to economic ones, play a big role in the formation of common markets. The initial stages of what is now the European Union is the example most often cited. Most analysts agree that the main motivation behind the establishment of the European Union was to avoid another devastating war between Germany and France. Nevertheless, economic factors clearly affect these decisions, and

endogenizing this decision will be an interesting extension to this paper. I consider a very simple common market formation process in which all

players are surprised by the formation of the common markets. This is not a very realistic assumption, but it is relatively easy to investigate. More complicated (and also more realistic) common market formation processes



498 C. Hadjiyiannis

yield similar results (see Hadjiyiannis 1999b). I separate the impact of common markets into two parts. The first is an immediate one-time change in tariffs and the second is the change in the rate of decline of cooperative tariffs or the change in the future path of cooperative tariffs.

Assume that, before the beginning of a given period t, the number of regions goes down from R0 = 2 to R1 = 1.12 All the results generalize to any change in the number of regions but specifying the number of regions makes the model more tractable.

4.1. The immediate impact of common market formation To evaluate the immediate effect we need to look at the impact of common markets on each of the components of equation (31). Consider first the welfare from cooperating, Ucoop. From equation (28), lowering the number of regions increases the rate of human capital accumulation from (1 + (a/2) + (ao/2))2 to (1 + U)2. That increases the cost of deviating, Vc. In addition, lowering R increases the rate of human capital accumulation in the punishment phase from (1 + (a/4) + (3ao/4))2 to (1 + (a/2) + (oa/2))2. This increases the welfare in the punishment phase and decreases Vc.

PROPOSITION 3. Reducing R increases the immediate cost of deviating Vc iffuture equilibrium tariffs are kept constant.

Proof From equation (31) observe that (1 + U)2 - (1 + (u/2) + (/2))2> (1 + (a/2) + (ue/2))2 - (1 + (a/4) + (3ao/4))2. The right-hand side of this inequal- ity is the change in the rate of growth of human capital accumulation of Ucoop, and the left-hand side is the change in the rate of human capital accumulation of

U,. Therefore, reducing R while keeping all future tariffs unchanged increases the rate of growth of Ucoop more than that of U,. Since Uoop, > Up to begin with, reducing R increases Vc because it increases Ucoop more than U,.

Proposition 3 shows that common markets increase the cost of deviating just as the market power effect suggests. The intuition behind this result is that larger regions have more to lose if they trigger the punishment phase. This is because their larger size means that they produce internally a lot more of the export good, and, therefore, human capital accumulates faster increasing the productivity gap between the import and export sectors.

The next step is to examine what happens to the benefit of deviating, VDt. Calculations show that (0 VDt/OR) < 0. This means that a decrease in R because of common market formation increases the benefits from deviating. This is exactly what the market power effect predicts; that is, common markets are better deviators. As regions become larger, they have more to gain by deviating because that deviation now applies to a bigger import market. Common

12 This and equation (27) imply that 7 < 2.




markets, therefore, lead to an increase in the deviating tariff, TDt, which in turn leads to an increase in the benefit from deviating.

The balance between these two effects determines the immediate effect of common markets on trade liberalization. If the increase in the cost of deviating dominates the increase in the benefit of deviating, then cooperative tariffs, Tct, decrease. Otherwise they increase. This balance depends critically on the parameter values and especially the discount factor 6.

PROPOSITION 4. An increase in 6 increases the likelihood that common market formation lowers immediate tariffs, assuming future equilibrium tariff rates remain constant.

Proof. From equation (31) observe that VDt is independent of 6. This implies that

(OVDt/OR) is also independent of 6. To show that (aVc/OR6) < 0, recall from proposition 3 that (a Vc/aR) < 0 if all future tariffs are kept constant. Also, note that the term multiplied by 6 is the rate of human capital accumulation and is bigger in Ucoo is bigger than in U,. Therefore, the effect of 6 on the change on

Uoo, (i.e., (OU9oopi/R)) dominates. This implies that (i9Vc/R) is increasing in 6, or, in other words, that (OVcl/RO6)< 0, if all future tariffs are kept constant. Therefore, the cost of deviating increases as 6 increases while the benefit remains constant, leading to lower cooperative tariffs Tc.

Common markets increase the benefit and the cost of deviating. The benefit of deviating is realized right away, while the cost is incurred in the future. A higher discount factor implies that the future is more important to the regions, therefore, the increase in cost dominates. This leads to lower tariffs.

The key issue under consideration is how the impact of common market formation changes over time. Assume that common markets happen at time t just before the decisions for that period are taken. The cooperative tariffs for t are given by

1 = Z2((1 ?")2)

Tct+l+i

(1 + ct)2 i=O (1 + ct++i)

1ho 1 + o

6(1 + o0)2 42)2 2

(32) 4 ahao

+

+ + + + )2 '

Note that equation (32) is the same as equation (31), with R = 1, except the fact that the productivity gap (b,/at) was only growing at the slower rate corresponding to R = 2 up to the time period t. Define Trct as the 'immediate impact curve' defined by equation (32). Each point on this curve represents the equilibrium tariff if liberalization happened in that period.

To simplify the analysis, I assume that the two trade liberalization paths corresponding to R1 and Ro cross at most once. I also assume the same thing



500 C. Hadjiyiannis

about the immediate impact curve (given by (32)) and the trade liberalization path corresponding to Ro. Let VcI, be the right-hand side of equation (32). Then sufficient but not necessary conditions for this are13

AVct(RI) > A Vct(Ro) (33) A

VcIt A Vct(Ro). (34)

I postpone the discussion of the impact of these assumptions on the results for later on in this section.

PROPOSITION 5. The trade liberalization path Tct(RI) with R = R1 is steeper than

Tc,(Ro), the trade liberalization path with R = Ro. Also, the immediate impact curve, 7cIt is steeper than Tct(Ro).

Proof. Note that the right-hand side of equation (32) and that of equation (31) with R = R1 are the same. Also note that (OVDt(R1)/7ct,) < ( VDt(Ro)/O-ct)- This means that for both 7c,(R1) and Tit it takes a larger decrease in tariffs to increase the left-hand side of the relevant expression by the same amount as that in 7ct(Ro). From equations (33) and (34) notice that in both cases the change in the right-hand side is bigger. Therefore, both paths are steeper than

Tct(Ro).

First, consider Tct(Ri) and rct(Ro). The benefit of deviating is independent of time in both cases. On the other hand, the cost of deviating increases, owing to the increase in the productivity gap. The productivity gap is growing faster with R = R1, so cooperative tariffs need to fall more in that case to equate the two sides of equation (31). At the same time, with R = R1 the benefit of deviating is less sensitive to tariff changes. This suggests that it takes a larger change in tariffs with R = R, to increase the benefit of deviating by the same amount. Since that amount is also higher with fewer regions, the trade liberalization path with R = R, is steeper.

Now compare the immediate impact curve with the trade liberalization path with R = Ro. The change in the productivity gap is the same. This is because before common markets are formed the productivity gap grows at the slower rate. It is only after regionalism that the higher growth is achieved. As soon as common markets are formed, the benefit of deviating becomes less sensitive to changes in tariffs. Recall that for the same change in the benefit of cooperation, it takes a larger change in tariffs to adjust. Therefore, the immediate impact curve is steeper than the trade liberalization path with R = R0, but flatter than that with R= R1. Another property of the immediate impact curve is that if common markets happen at t =0, then the tariff at t =0

13 Alternatively, one could impose at most single crossing without these stronger conditions and that the tariff paths with either number of regions converge to free trade and prove the same results.




would be the same as that implied by the trade liberalization path with R = R1. The reason is that the productivity gap is the same for all three cases at time t = 0.

Let common markets be formed at time t. Then trade liberalization follows Tct (Ro) up to t and then jumps to the value of rclt at time t. Then it follows

rct (R1) from the point on 7-ct (R1) that corresponds to the same productivity gap as the one reached at time t.

PROPOSITION 6. The later a common market is established the more likely it is that it will lead to a drop in immediate tariffs.14

Proof. From proposition 5 the immediate impact curve is steeper than the trade liberalization path before common markets. Then there are three possibilities.

1. The immediate impact curve starts above the trade liberalization curve, but it is not steep enough to intersect it before they both converge to their limits. However, since the immediate impact curve is steeper, the difference between the two gets smaller with time. This means that the later common markets are formed the smaller the immediate increase in the tariffs would be.

2. The immediate impact curve starts below the trade liberalization curve. As time goes on, the immediate impact curve gets further away from the trade liberalization curve, since it is steeper. Therefore, the later the common markets are formed, the higher the immediate decrease in the tariffs would be.

3. The immediate impact curve starts higher than the trade liberalization curve, but it is steep enough to intersect it. Initially, the immediate impact curve is above the trade liberalization curve, but, since it is steeper, it gets closer and closer until eventually it crosses and starts getting further away. So the later common markets happen the more likely it is that they are going to lead to a decrease in immediate tariffs.

Figure 3 illustrates these results. The dotted line in all three panels represents the immediate impact curve.'5 Panel A shows the case where all common markets, no matter when they take place, lead to an increase in tariffs (low 6). Panel B shows the case where, no matter when common markets happen, they lead to a decrease in tariffs (high 6). Note that in this case the immediate impact curve and the trade liberalization path before liberalization never cross. The most interesting case is panel C which presents the intermediate case. The early common markets increase immediate tariffs, while the late ones decrease tariffs.

The intuition behind proposition 6 is the following. Common markets increase the cost of deviating. They also increase the benefit of deviating.

14 This does not imply that common markets will always lower immediate tariffs. 15 This is not the path that cooperative tariffs follow after jumping to that level.



502 C. Hadjiyiannis

Tc Tc

Tc(Ro)

immediate impact immediate impact

T

)c(R)(R0)

time time Panel A Panel B

tc

.-

immediate impact

rC(RO)

A

T•(Rc)

time Panel C

FIGURE 3




The latter does not vary with time, but the former does. The reason is fairly simple. In periods of deviation or cooperation, regions are still specializing. The only thing that changes with time is the productivity gap, which is relevant only in the punishment periods, since regions have to produce both goods.

The key effect is the one on autarky welfare. As time goes on, the productivity gap increases, which reduces autarky welfare. The impact of common markets on autarky welfare is increasing in this productivity gap. As the productivity gap gets wider, common markets reduce this welfare more, which in turn means they increase the benefit of cooperation more. This means that more cooperation can be supported through time. In other words, common market formation is more likely to decrease tariffs if it happens later in the process. It is precisely dynamic results like proposition 6 that can be achieved only with the use of non- stationary dynamic trade liberalization models.

These results rely heavily on the single-crossing assumptions. These assumptions make it easy to compare the three tariff paths. Crossing at most once means that before the crossing point one of the two paths has lower tariffs and after the crossing point the other has lower tariffs. Abandoning this assumption will not change the main result of the paper, which is that the impact of regionalism changes through time. In fact, it does not even change the result that in the latest section, before converging to free trade, regionalism leads to a drop in immediate tariffs and in the section before that to an increase. If, however, there are more crossings, then there are just more sections to comment on.

Point A16 in panel C of figure 3 is the point where the immediate impact is zero. Common markets before this point lead to an immediate increase in tariffs, while those after this point lead to an immediate fall. In this way the point A summarizes all the relevant information on the immediate impact of common markets for this case. It is worthwhile to perform comparative statics on A and investigate its properties.

PROPOSITION 7. For the case in panel C offigure 3:

1. An increase in 6 will shift A to the left. 2. An increase in a will shift A to the left. 3. An increase in 0 will shift A to the right. 4. An increase in (phbo/ahao) will shift A to the right.

16 Point A should not be thought of as the optimal time to have common market formation. That is because there is a trade-off between lower tariffs right now and how early we reach free trade. It is possible that free trade is achieved earlier by having common markets as soon as possible, since the rate of decline of tariffs is faster.



504 C. Hadjiyiannis

Proof

1. An increase in 6 increases the right-hand side of equation (32) (see the proof for proposition 4). That lowers the cooperative tariffs. Therefore, at the previous point A the immediate impact is to lower tariffs. This implies that the new point A will be to the left of the old point A.

2. An increase in a increases the right-hand side of equation (32) because it increases U00oo more than U,. This increases the cost of deviating, and compared with the old point A, the immediate impact curve is lower. The new A is to the left of the old A.

3. From equation (32) notice that an increase in 0 increases Up, leaving every- thing else unchanged. That decreases the cost of deviating, thus leading to less cooperation at the old point A. This implies that the new point A will be to the right of the old point A.

4. From equation (32) an increase in (phbo/aho) increases Up, leaving every- thing else unchanged. This decreases the cost of deviating, thus leading to less cooperation at the old point A. This implies that the new point A is to the right of the old point A.

The intuition behind the first statement of proposition 7 is that increasing the discount factor increases the impact of common markets on the cost of deviating. The reason is that the cost of deviating is incurred in the future, while the benefit is incurred right away. A higher discount factor makes the future more important.

An increase in a increases the rate of growth of human capital accumulation. This increases the value of future cooperation (Ucoop) and the value of autarky (Up). It increases the former more than the latter, since in cooperation regions specialize and more of the production is done internally, absorbing the full impact of the increase in a. That is not true in autarky, since in that case more of the production is done externally and the impact of a is neutralized by 0.

An increase in 0 diminishes the difference between internal and external knowledge spillovers. The productivity gap is not growing as fast, which in turn means that the decrease in autarky welfare, due to the common markets, is not growing as fast. The punishment phase is not as severe. Therefore, the increase in the cost of deviating due to the market power effect is lower.

Finally, a decrease in the initial productivity gap (represented by an increase in (/hbo/ahao) will mean that at any future time the productivity gap is smaller. A smaller productivity gap means that the increase in the welfare from autarky due to the common markets is bigger. This is because punishments are not as severe. Devoting resources to the low-productivity sector is not as bad because of this smaller productivity gap. This means that the increase in the cost of deviating due to the market power effect is now lower, leading to less cooperation.




immediate impact

ttm

Ir after Common Markets

t, t2 time

FIGURE 4

4.2. The change in the path of cooperative tariffs Common markets also change the path of tariffs after the initial one-time shock. After common market formation tariffs are determined by equation (31) with R = R1. This is exactly like starting trade liberalization from scratch with fewer regions and new initial conditions (boi/ai)= (fhbo/ahao), where t is the period that common markets are formed. In other words, the cooperative tariff path for any time t > i is determined by

(6( 2\i Tct+1+i

(1 + ct)2 i=

+ (1 + Tr+1+i)

l bi ((1 + O) 1-t

6(1 + o)(1 + O)

4a,

( + a) 1 - 6(1 + + )2 (35)

Proposition 5 establishes that the trade liberalization path is steeper for R = R1 than R = Ro. After the initial shock, tariffs fall faster according to (35). The intuition behind this is that the productivity gap grows faster after common markets are formed. The reason is that fewer regions means that a larger proportion of world production is produced internally, leading to faster growth in the comparative advantage sector. This increases the productivity gap faster and makes punishments more severe faster. Therefore, cooperative tariffs fall faster.

Figure 4 illustrates the situation for the intermediate case shown in figure 3 (panel C). The dotted line in figure 3 represents the immediate impact. A couple



506 C. Hadjiyiannis

of examples of the path of cooperative tariffs for different cases (t'= t, and t= t2) are shown. Note that these paths are steeper than the previous paths. They represent the trade liberalization process from (31) with R= R1 and initial conditions (bi/ai).

These results shed some light on whether common markets shorten the time needed to get to global free trade. The first thing to note is that with or without common markets the parameters might be such that free trade is never achievable. In that case tariffs will converge to some positive tariff. If free trade is achievable and the parameters are such that the immediate impact of common markets is to lower tariffs, then they will shorten the time needed to get to free trade, since common markets always increase the rate of decline of tariffs. It is not clear that free trade will be achieved earlier if the immediate impact is to increase tariffs. It is clear, however, that late common markets are more likely to shorten the path to free trade because they are more likely to lead to a decrease in immediate tariffs.

5. Conclusion

In this paper I examine the impact of common market formation in the context of a non-stationary dynamic model of trade liberalization and compare the results with the existing literature. The main contribution is an examination of how the impact of common markets varies with time. The key results of the model are that the later common markets happen, the more likely it is that they lead to a drop in immediate tariffs. In addition, the rate of decline of tariffs increases after their formation. These results imply that late common markets are more likely to shorten the path to free trade. It is also shown that a high discount factor, a high rate of technology transfer, a large difference between internal and external spillovers, and a large initial productivity gap increase the likelihood that common markets lower immediate tariffs.

The results support the view that it is useful to study the impact of regionalism on trade liberalization using non-stationary dynamic trade liberalization models. Only then can one investigate the full impact of regionalism. It should not be a surprise that these effects are dynamic and change according to when regionalism happens.

The scope of this paper was limited to addressing the impact of common markets on trade liberalization. To keep things simple, the formation of common markets was exogenous to the model. A natural extension would be to relax that assumption and make endogenous the decision to join a common market. Another possible extension would be to investigate what happens if common markets between regions of different types are allowed. Finally, one can relax the assumption that there are only two traded goods and allow each region to have a comparative advantage in a good and have monopolistic competition with other regions.




References

Bagwell, Kyle, and Robert Staiger (1997a) 'Multilateral tariff cooperation during the formation of customs unions,' Journal of International Economics 42, 91-123

- (1997b) 'Multilateral tariff cooperation during the formation of free trade areas,' International Economic Review 38, 291-319

Baldwin, Richard (1995) 'A domino theory of regionalism,' in Expanding Membership of the European Union, ed. Pertti Haaparanta and Jaakko Kiander (Cambridge, New York, and Melbourne: Cambridge University Press)

Bhagwati, Jagdish (1993) 'Regionalism and multilateralism: an overview,' in New Dimensions in Regional Integration, ed. Jaime de Melo and Arvind Panagariya (Cambridge, New York, and Melbourne: Cambridge University Press)

Bhagwati, Jagdish, and Arvind Panagariya (1996) 'Preferential trading areas and multilateralism-strangers, friends, or foes?' In The Economics of Preferential Trade Agreements, ed. Jagdish Bhagwati and Arvind Panagariya (College Park and Washington, DC: University of Maryland, Center for International Economics and AEI Press)

Bhagwati, Jagdish, Pravin Krishna, and Arvind Panagariya (1999) Trading Blocs: Alternative Approaches to Analyzing Preferential Trade Agreements (Cambridge and London: MIT Press)

Bond, Eric, and Constantinos Syropoulos (1996) 'The size of trading blocs: Market power and world welfare effects,' Journal of International Economics 40, 411-37

Bond, Eric, Constantinos Syropoulos, and Alan Winters (2001) 'Deepening of regional integration and multilateral trade agreements,' Journal of International Economics 53, 335-61

Devereux, Michael (1997) 'Growth, specialization, and trade liberalization,' Inter- national Economic Review 38, 565-85

Ethier, Wilfred (1998) 'Regionalism in a multilateral world,' Journal of Political Econ- omy 106, 1214-45

Freund, Caroline (2000) 'Multilateralism and the endogenous formation of preferential trade agreements.' Journal of International Economics 52, 359-76

Hadjiyiannis, Costas (1999a) 'The effects of customs unions on the process of multilateral trade liberalization: the importance of timing,' Working Paper 99-14, Depart- ment of Economics, University of Cyprus

- (1999b) 'Regionalism and trade liberalization,' PhD dissertation, University of Wisconsin

Kemp, Murray, and Henry Wan (1976) 'An elementary proposition concerning the formation of customs unions,' Journal of International Economics 6, 95-7

Krishna, Pravin (1998) 'Regionalism and multilateralism: a political economy approach,' Quarterly Journal of Economics 113, 227-51

Krueger, Anne (1999) 'Are preferential trading arrangements trade-liberalizing or pro- tectionist?' Journal of Economic Perspectives 13, 105-24

Krugman, Paul (1991) 'Is bilateralism bad?' In International Trade and Trade Policy, ed. Elhanan Helpman and Assaf Razin (Cambridge and London: MIT Press)

Lapan, Harvey E. (1988) 'The optimal tariff, production lags, and time consistency,' American Economic Review 78, 395-401

Levy, Philip (1997) 'A political-economic analysis of free-trade agreements,' American Economic Review 87, 506-19

Lipsey, Richard (1957) 'The theory of customs unions: trade diversion and welfare,' Economica 24, 40-6

Meade, James (1955) The Theory of Customs Unions (Amsterdam: North-Holland)



508 C. Hadjiyiannis

Panagariya, Arvind (2000) 'Preferrential trade liberalization: the traditional theory and new developments,' Journal of Economic Literature 38, 287-331

Richardson, Martin (1993) 'Endogenous protection and trade diversion,' Journal of International Economics 34, 309-24

Rivera-Batiz, Luis, and Paul Romer (1991) 'Economic integration and endogenous growth,' Quarterly Journal of Economics 106, 531-55

Staiger, Robert (1995) 'A theory of gradual trade liberalization,' in New Directions in Trade Theory, ed. Jim Levinsohn, Alan Deardorff, and Robert Stern (Ann Arbor: University of Michigan Press)

Staiger, Robert W., and Guido Tabellini (1987) 'Discretionary trade policy and exces- sive protection,' American Economic Review 77, 823-37

Viner, Jacob (1950) The Customs Union Issue (New York: Carnegie Endowment for International Peace)



Documents

Common Markets and Trade Liberalization