Abstract:
Regional integration (RI) leads to an increased net export in the increasing returns (IRS) good for the larger country in a customs union (CU), and to an ambiguous effect on the small country in the CU, whereas the outside country decreases its net export in that good. Intra-industry trade is likely to increase in initial phases of RI, but will decrease later. Large countries will increase their net export of the IRS-good in response to global integration (GI) when they are CU-members. RI leads to higher welfare in the member countries, but to lower welfare in outside countries. GI is likely to benefit all countries.

JEL Classifications Fl2, F13

A rapidly growing number of studies use models with imperfect competition and trade costs to examine issues such as specialisation in production and industrial concentration. In such models, the size of the market becomes of crucial importance (see Krugman, 1980, Helpman & Krugman, 1985, chapter 10, Venables, 1987, Krugman & Venables, 1990). A large home market can give rise to net exports in industries characterised by increasing returns to scale (IRS) and the international location of IRS-industries can be of considerable importance for economic welfare.¹ Thus, there may be what Krugman (1983) has defined as market-size externalities.² Countries with domestic production of IRS-industries are, other things being equal, likely to have higher welfare because of higher factor rewards and lower product prices (cf. Krugman, 1980, Venables, 1987).

If "market-size externalities" are important, analyses of trade liberalisation would gain by considering the international location of industries subject to such externalities. Lately, radical moves towards regional integration have been made through the deepening of European integration and the formation of the North American Free Trade Area (NAFTA). Although global integration has also been undertaken through the new GATT agreement, it has taken place in the context of existing regionalism and is likely to have different effects than in a world of non-discrimination. Yet, existing studies generally fail to incorporate regional integration.³ Therefore, we extend and generalise the models introduced by Krugman (1980) and Helpman & Krugman (1985, chapter 10) to account for the interplay between trade impediments and market size, allowing for the presence of a customs union.⁴

This paper adds to earlier work on market size and trade by showing: first, that global reduction in all trade impediments leads to increased (decreased) production in the IRS-industry of countries that are larger (smaller) than the world average. However, in the presence of a customs union, this is not the case, even the largest country in the world will not necessarily increase its production in the IRS-industry. Second, that regional integration leads to decreased production in the IRS-industry for the outside country. The larger of the two union members will necessarily have more of the IRS-industry located domestically while the effect on the smaller country in the CU is ambiguous. Third, that initial phases of regional integration are generally associated with an increase in the share of intra-industry trade, while later stages are associated with a reduced share of IIT. Fourth, that regional integration unambiguously benefits the member countries and hurts the outside country. The effects of global trade liberalisation are somewhat less clear-cut, although all countries experience welfare gains in the absence of a customs union.

The rest of the paper is organised as follows. In section II, we present the model. Section III examines the production effects of global integration and section IV examines these effects under regional integration. Section V analyses welfare effects and finally we offer some concluding comments in section VI.

Consider a model with one factor of production (labour) and three countries. Without loss of generality, we can choose units so that there is one unit of labour in country I. In country II, there is units of labour whereas the endowments of labour in country III equals units.There are thus two industries. Industry X is producing differentiated products under IRS and industry Y that produces homogeneous products under CRS. For the sake of simplicity, it is assumed that free trade prevails in industry Y.

We follow Helpman & Krugman (1985) in assuming trade impediments of an "iceberg type" in industry X, i.e. that only a certain proportion of each exported unit is received by the importer. They should be thought of as a composite of various man-made and natural (transportation costs and cultural differences) trade barriers that can be reduced (or increased) but not totally removed by policy measures. We define as the fraction of differentiated products that "arrive" at the importers in bilateral trade flows between regions I-II and I-III, whereas is the fraction of differentiated products that arrive at the importers in trade flows between regions II and III. A customs union (CU) is conventionally defined as an absence of tariffs and quotas among member countries that also apply a common external tariff. In this paper, there is a common external trade barrier, but positive internal trade barriers even after the formation of the CU reflecting natural barriers to trade.

We suppose that all individuals have identical utility functions given by:

where x_i is consumption of variety i in industry X whereas Y is consumption of the homogeneous product. Since individuals derive the same utility from the consumption of differentiated products produced in different countries, n is the total number of varieties available.⁵

It then follows that aggregate demand for products produced in the three regions will equal (see Helpman & Krugman, 1985, p. 206 ):

(2)-(4)

where p_j is the price of varieties produced in country j (equal for all varieties produced in each country since they enter the utility function symmetrically and since they are produced with the same technology).⁶

In the IRS-industry, the average cost function is equal to:

where w is the constant marginal cost and µw fixed cost; x_i output per firm. From the demand functions, we can derive marginal revenue (equal across markets) as . Profit-maximisation requires that marginal revenue equals marginal cost and for equilibrium with monopolistic competition, price must equal average cost. Using these conditions, we can derive equilibrium output per firm as (cf. Krugman, 1980, p. 952).

Equation (6) shows that output per firm will be equalised across countries.⁷

The Analysis
Consider the demand equations (2)-(4) and let us divide both sides of the three equations with n₁, n₂ and n₃, respectively. Furthermore, we choose units so that w=p=1 and define Rewriting then yields

Equations (7)-(9) can be solved to yield the number the varieties in each country as:⁹

(10)-(12)

There are several possible equilibria. Each country can either have zero or a positive number of firms in the IRS-industry. We restrict our analysis to the case where all countries have positive production¹⁰

¹According to a certain terminology, this could be said to be a demand-explanation of net trade. For other such explanations, see Hunter & Markusen (1988) and Torstensson (1993). The most common explanation of net trade is the Heckscher-Ohlin model. Empirical studies do, however, suggest that additional explanations of net trade are needed (see e.g. Baldwin, 1979). Torstensson (1995a) shows that technical differences may beimportant, but it is also clear that there is room for demand-explanations.

²Krugman (1993) has forcefully argued that "market-size externalities" means that it can be beneficial to have domestic production in IRS-industries.

³For a model allowing for multinational firms and in which market size matters, see Markusen & Venables (1994) and Markusen (1995). For useful surveys of the theory of regional integration (RI), see e.g. Melo, Panagariya & Rodrik (1993) and Hine (1994). For earlier studies of RI with monopolistic competition, see Ethier & Horn (1984), Krugman (1991a), (1993b). See also recent studies of economic geography, e.g. Krugman (1991b) and Krugman & Venables (1993). However, as argued by Baldwin (1993, p. 46): "The question of what regional integration does to industrial concentration is important, but as yet it is not well understood." Baldwin & Venables (1995) examine aspects of regional integration and market size.

⁴A similar extension is used by Krugman (1994) to examine cases of industrial specialisation based on transportation costs and so called transportation "hubs".

⁵In this model. the number of varieties is constant and trade liberalisation only affect the location of production.

⁶Note that there is an indirect demand for products used up through trade impediments. That is, we have to multiply the foreign demand with and , respectively.

⁷We also need equilibrium on the labour market (see Torstensson, 1994 for conditions).

⁸Before we proceed, it can be useful to note that and decrease in trade impediments and increase in the degree of scale economies (and thus degree of product differentiation).

⁹For details, see the Mathematical Appendix.

¹⁰The conditions for positive production are discussed in Torstensson (1994) as is the trade pattern in this framework.