Counsels of Imperfection
Anthony Venables

Recent years have seen the rapid development of a new approach to the theory of international trade, incorporating details of industrial organization. Whereas conventional trade theory is concerned almost exclusively with the case of perfect competition and non-increasing returns to scale, the new literature relaxes these assumptions. Attention is focussed on industries in which production is subject to increasing returns to scale, in which firms may produce differentiated products, and where competition is imperfect. The new theory helps account for a number of empirical puzzles which troubled the conventional theory; it also offers a very different perspective on trade and industrial policy. This article outlines some of the insights offered by the new approach to trade theory and describes the work of the Centre and its Research Fellows in developing the theory and applying it to questions of economic policy.

There are of course a variety of plausible theories of imperfect competition, and some of the results obtained in this area are sensitive to the nature of the competition which is assumed to take place among firms. We may take as a central case an industry which contains a relatively small number of firms, some of which are located in the home economy and the remainder in the foreign economy. Each firm has some monopoly power, which may be derived from two possible sources. The first possibility is that each firm's product is differentiated from the products of other firms; because the product is distinctive it has its own demand curve in each country. The second possibility is that although the output of every firm in the industry is identical, some firms are large enough for their supply decisions to affect the market price. If in addition international arbitrage is incomplete, then firms' demand curves in each country are independent. Each firm perceives that if it increases its supply in one economy, the price in that economy will fall, but the price in the other economy may be unchanged.

Thus, whether their products are differentiated or not, each firm has some market power in both countries. If firms seek to maximize profits, then they will each sell in both the home and foreign economies, so that intra-industry trade occurs. Where all firms produce identical products, this intra-industry trade involves what is termed 'cross-hauling', as every firm exploits its market power in each economy. Models of imperfect competition therefore offer a more straightforward explanation of such 'cross-hauling' than does conventional trade theory, which assumes perfectly competitive firms without market power.

The new theory predicts that the volume of this intra-industry trade may be large relative to the output of the good. In the simplest case, in which there are two economies whose structures are identical and there exist no barriers to trade, each economy exports half its output and imports half its consumption. Intra- industry trade is balanced, and there are no net trade flows, but cross-hauling is substantial. Barriers to trade such as tariffs or transport costs reduce the volume of this trade, but the possibility remains that the two economies may both import and export similar or even identical products, despite the fact that trade itself is costly.

In practice intra-industry trade is not balanced, and we observe that net trade flows do occur. In order to explain the actual pattern of net trade flows we also need a theory which accounts for the location of firms in the imperfectly competitive industry. Conventional trade theory can provide such an explanation in terms of factor endowments, technological differences, or differences in governments' tax and tariff policies. This accounts for the pattern of net trade flows, but these occur against a background of a large volume of intra- industry trade which is most easily explained by the newer approaches to trade theory.

Bringing imperfect competition and increasing returns to scale into trade theory generates important insights into the sources of potential gains from trade. Conventional trade theory emphasizes that these gains arise because trade allows each country to specialize in the production of those goods in which it has a comparative advantage. The new approaches suggest that in addition trade may be beneficial because it is pro- competitive, reducing monopoly power. Trade may also permit fuller realization of economies of scale and increase the variety of products available for consumption in an economy.

Since markets are assumed to be imperfectly competitive, however, the theory offers no guarantee that the market equilibrium will realize these potential gains. For example, consider a two- country model, with a single firm located in each economy. Trade between these economies changes the market structure in each from monopoly to duopoly. Price is reduced, so benefitting consumers, but at the same time producers' profits fall. Price is now closer to marginal cost, and taking both economies together, trade is certainly beneficial (in the sense that the sum of consumer and producer surplus is increased). World welfare is therefore augmented. But it is not necessary that both countries experience an increase in welfare as a result of this trade. Suppose the economies are opened to trade, and the profits earned by the domestic country's firm on its sales in the foreign market are less than the profits earned by the foreign firm on its exports (perhaps because the two markets differ in size). Trade therefore results in a net transfer of profits from the domestic economy to the foreign economy. This international redistribution of profits creates the possibility of a domestic welfare loss.

The example above took the number of firms in each economy as fixed - one in each. Assumptions concerning the entry and exit of firms are often crucial in imperfectly competitive models. Results obtained to date suggest that free entry and exit of firms will make gains from trade more likely. Trade will certainly increase the number of firms supplying each market and so reduce price and raise consumer surplus, as in the example above. With free entry, however, equilibrium profits are zero, since if profits are positive new firms will enter the market and reduce prices. Trade cannot reduce profits, as in the previous example, but instead may cause exit of firms from the industry, reducing the number of producers in each country. This increases the scale of production of each firm, and if there are increasing returns to scale, average costs are reduced.

Free entry in this example is sufficient to ensure that gains from trade are realized. This is a surprisingly general result of the new trade theory. For example, suppose there are transport costs and that all firms produce identical products. Trade therefore involves the costly cross-hauling of identical products, but even this apparently pointless trade increases welfare. Why should this be so? Remember that firms only export if doing so yields positive profits. Free entry reduces the total profits of all firms in the economy to zero, so ensuring that price (and hence average cost) is lower with trade than under autarchy.

In addition to these arguments there may be another source of gains from trade if firms in the industry produce differentiated products. We have seen that trade increases the number of firms supplying each market; if each firm produces a different product type, then trade may raise welfare by increasing the variety of products available for domestic consumption. It is clearly impossible to prove that these results concerning gains from trade are completely general. Nevertheless the new approaches to trade theory suggest that, in a wide range of cases, such gains are both positive and significant.

The new literature on trade under imperfect competition sheds important light on our understanding of trade policy and of open economy industrial policy. In the presence of imperfect competition the gains from policy intervention can arise in two ways. First, national policy may be used to change the economy's terms of trade (the ratio of export to import prices). Second, policy may correct supply imperfections which arise from imperfect competition.

The first of these arguments is of course the foundation of the conventional theory of optimal tariffs, the level of which is chosen so as to alter the economy's terms of trade. But the existence of imperfect competition obliges us to reassess such arguments. In order to illustrate this, consider an economy which has no domestic production of the good in question (so abstracting from problems of imperfections in domestic supply), and which consumes an arbitrarily small share of world output of the good. Suppose further that this output is supplied by a single foreign firm with monopoly power in the economy under consideration. Tariff policy may then be used to induce the monopoly supplier to change the import price, even though the price in the rest of the world is constant.

The optimal policy may not be an import tariff, however, as the following example illustrates. Suppose that the foreign monopolist has constant marginal costs of production and sets price equal to marginal cost plus a constant proportional mark- up. A specific import tariff is equivalent to an increase in marginal cost, and so will raise price by the amount of the tariff plus the mark-up. This increases the price of the import net of the tariff and thereby raises the unit cost of imports to the domestic economy, reducing its welfare. In order to reduce the unit cost of imports, policy must take the form of a specific import subsidy! The general point here is that policies can change the terms of trade, not from changing supply and demand on the world market, but from the interaction between government policy and the decisions of the firm supplying imports. Although there is an argument for policy intervention in trade, we cannot determine what form it should take without detailed information on how the firm chooses its output and on how the government is able to manipulate the firm's actions.

We may now turn to the case in which the domestic economy produces as well as consumes the good in question. In addition to changing the terms of trade, policy may now attempt to correct imperfections in domestic supply. The inefficiency most often cited arises from the fact that in imperfect competition domestic firms set price above marginal cost. Welfare is increased by any policy which expands a domestic activity operating at a price in excess of marginal cost: this suggests that policies should be designed to expand the output of domestic firms. The argument may be illustrated by considering the example of a small subsidy to the exports of a single domestic firm. Such a subsidy reduces the domestic firm's marginal cost of supplying the foreign market, so reducing the equilibrium price in the foreign market and expanding the sales of the domestic firm in this market. The domestic economy therefore experiences a terms of trade deterioration because the price of its exports is reduced. Nevertheless the export subsidy has expanded an activity for which price exceeds marginal cost. Under a wide range of circumstances, the net effect of the policy is to increase the total surplus extracted from the foreign market; that is, the extra profits earned on the larger volume of domestic exports outweighs the loss of profits due to the terms of trade deterioration. Essentially the policy shifts profits from foreign firms to the domestic firm, and hence raises domestic welfare.

This argument has been the subject of vigorous criticism, two strands of which are particularly important. First, notice that the objectives in the foreign market of the domestic firm (in the absence of the subsidy) and of the government (using the subsidy) are identical; they both wish to maximize producer surplus on sales in this market. Clearly then the export subsidy can be effective only if the government has some power that the firm does not. In this instance the government's power arises from the assumption that it decides upon its actions before firms do, and is thereby able to commit itself to actions in a way that firms cannot. The export subsidy is therefore taken as given, a parameter in the game between domestic and foreign firms which alters the equilibrium in favour of the domestic firm. The domestic firm could not achieve this outcome by itself, because if it threatened to act as if there were a subsidy, it would not be believed by its rivals and therefore could not change the equilibrium. Arguments for policy intervention hinge crucially on the assumed interactions among firms and between firms and government. If, for example, we change the nature of this interaction so that the domestic firm chooses - and commits itself to - its output level before foreign firms act (i.e., the domestic firm becomes a 'Stackelberg leader'), then we have given the domestic firm the same power as government. There is then no case for the government to intervene with the export subsidy. Assumptions concerning interactions between firms, or between firms and government, are often not made explicit in arguments for particular policy interventions, but their validity is crucial to the effectiveness of the policy.

The second criticism levelled against the argument for export subsidies is that it takes the number of domestic firms as constant. If there is free entry and exit then the subsidization of domestic firms will increase their number. With increasing returns to scale this replication of firms is inefficient, because it tends to increase the unit cost of production in the economy. This weakens, but does not overturn, the case for export subsidies. Even with free entry it remains true that the export subsidy increases the scale of each domestic firm (though not the scale of its foreign rivals), so that average costs are reduced.

The arguments above were conducted in the context of an export subsidy. Similar arguments can be constructed for other policies designed to encourage the expansion of domestic firms. Such policies include import tariffs and subsidies to marginal production costs or to some component of these costs such as research and development expenditure. In all these cases the essential message of the theory is clear. The design of policy is sensitive to the nature of the competitive interaction between firms, the ease of entry and exit in the industry, and the extent to which government can manipulate the actions of firms.

The models outlined above demonstrate that rapid progress has been made in developing the way in which we think about trade and trade policy. However, the new literature does not give unambiguous answers to many questions of interest. For example, the desirability of positive tariffs is extremely sensitive to assumptions about the behaviour of firms and the ease of entry into markets. Perhaps this should not surprise us. We observe many different market structures and should not expect policy recommendations to be universally applicable. The variety of possible cases does however suggest directions for future research.

At the theoretical level, the new approach has generally dealt with the 'extreme' or polar cases: firms are either completely free to enter the industry or no entry is possible at all; markets are either unified through international arbitrage or segmented nationally. Further research on the factors which may give rise to these polar cases is required. We also need more exploration of behaviour in 'intermediate' cases. It seems likely that both these problems will lead to greater recognition and modelling of asymmetries between firms. For example, asymmetries between incumbents and potential entrants in particular markets may generate barriers to entry and to trade. Recognition of asymmetries between firms which produce in a country and those that merely export to it may also lead to new approaches to the analysis of multinational corporations.

If the new insights into policy described above are to have practical relevance, empirical work is also needed. For example, arguments supporting a tariff in a particular industry require detailed study of that industry, in order to judge the speed at which entry and exit would occur, the extent to which markets are unified or segmented, and the intensity of competition. These are difficult empirical matters to assess, but the new theoretical literature outlined above provides a consistent structure within which to analyze the problem and focusses attention on the essential empirical issues. In addition to industry-level studies it is important to attempt to assess these arguments for particular policies in a general equilibrium framework. This approach prevents us from ignoring the effects of interactions between markets and allows us to assess the quantitative importance of the arguments outlined above and their sensitivity to alternative assumptions concerning market structures. It also provides a consistent framework within which the consequences of different types of policy for different industries can be investigated.

Anthony Venables is a Lecturer in Economics at Sussex University and a Research Fellow in the Centre's International Trade programme. Further details of the research described in this article can be found in CEPR Discussion Papers Nos. 9, 38, 41, 42, 80 and in particular No. 74. His joint paper with Alasdair Smith, examining trade and industrial policy in open economies, will be presented to the April meeting of the Economic Policy Panel in Paris. The analysis of international trade under imperfect competition was the focus of a September 1984 conference organized by CEPR and the International Economics Study Group, a report of which can be found in CEPR Bulletin No. 5. CEPR has recently launched a research project on Strategic Trade Policy jointly with the National Bureau of Economic Research.