Economic Integration 

2 - Modifications to basic Customs Union Theory

There are a number of problems with the basic model which we considered in Lecture One. Principal among these are:

(a) The assumptions are very strict, ignoring such factors as monopoly power, transport costs, economies of scale, unemployment and the distribution of income, information deficiencies, adjustment costs and non-tariff barriers, the balance of payments and exchange rates.

(b) It takes no account of intra-industry trade and technological influences on trade.

(c) It assumes that the removal of tariffs will have a significant effect on trade. This is open to doubt given the large tariff reductions which have taken place under the GATT and since international investment has to some extent replaced international trade.

(d) The model is static.

The assumptions can, however, be relaxed. We have already done this in considering the large country case. Let us return to this case briefly before going on to other issues.

2.1 Customs Unions versus other forms of integration

We have seen in Lecture One that, in the small country case, welfare arguments do not support the establishment of customs unions. The optimal policy for the small country, rather, is the unilateral declaration of free trade. In the large country case (where the supply curves of P and W slope up), if both P and W have the same elasticity of supply, the optimum policy for H becomes the application of a non-discriminating tariff to the goods of both P and W.

This is the basis of the argument by Cooper and Massell (1965a) that there is always a non-preferential tariff policy for country H that is superior to joining a customs union. They consider the reduction in price in two stages:

(i) H reduces the tariff level for both P and W such that the price in H falls to the level in P. This gives the same union price and production, consumption and import changes as does the customs union.

(ii) Now we introduce the customs union starting from the new price. In (i), gains from trade creation still exist, but imports will continue to come from W and thus there is no trade diversion. Further, the new imports due to trade creation generate tariff revenue. All the losses arise from step (ii) since it is this which causes trade diversion. Thus, a policy of unilateral tariff reduction (UTR) is superior to customs union formation. The Cooper and Massell case is shown in a diagram by Södersten and Reed (1994) on page 328. Things change if the elasticities of supply differ - in this case, a discriminating tariff is called for, but it remains unlikely that a customs union is the best option.

All of this assumes that the customs union resulting from the integration of H and P is small relative to total world trade. If the union is large, the formation of the union introduces the possibility of welfare gains to H and P at the expense of W (a beggar-thy-neighbour gain) as the large union is able to change the terms of trade in its favour. However, if W is also a large trading bloc, attempts by (H + P) to change the terms of trade in its favour is likely to lead to retaliation, causing both sides to be worse off.

All of this explains the quote from Heffernan and Sinclair:

"A country may be better off in a CU (customs union) than if it applies non-discriminatory tariffs against all other countries....If the country and the union are small, it must be better off trading freely with everyone than retaining barriers against non-union imports."

i.e. (i) small countries joining to form a small union would be better off not forming the union but practising unilateral free trade;

(ii) small countries joining to form a large union may make beggar-thy-neighbour gains at the expense of outside countries, but may not do so.

(iii) large countries may gain from joining a customs union but it is unlikely.

Remember that `a small country' here is one which is not large enough for changes in its demand or supply to affect world prices.

A variant of the above arguments was produced by James Meade (1955) and by Ethier and Horne in the form of an argument for incomplete tariff reduction. Further, it has been argued that free trade areas are preferable to customs unions.

2.2.1 Meade and the argument for incomplete tariff reduction

Meade, provided an argument that welfare is more likely to be increased if tariffs are partly removed between H and P, rather than being completely removed in a discriminatory way as in a customs union. There are two reasons for this: one relating to trade creation and the other to trade diversion:

(i) Successive reductions in tariffs contribute less and less to gains from trade expansion within the customs union. This is because, as tariffs are reduced, the least efficient firms in H (i.e. those where the efficiency gap between firms in H and those in P is greatest) will be replaced first. As the tariff continues to be reduced, the more efficient firms in H will begin to be replaced. It follows that the production gain from each successive reduction in tariffs will be smaller.

(ii) On the other hand, losses from trade diversion increase at least in proportion as the degree of trade discrimination (against country W) rises. This is because the least efficient firms in W will be excluded first.

Thus, it is possible to find the theoretical point at which welfare is maximised (where the gains from trade creation from a further small reduction in tariffs are just equal to the losses from trade diversion resulting from the same tariff decrease). This will be at a point short of a full customs union.

According to this view, a customs union becomes only the fourth-best possibility. We have, in order of desirability:

1. free trade

2. non-discriminatory tariff reduction (here there would be no TD)

3. trade preference area (with incomplete tariff reduction against P)

4. customs union

2.2.2 Ethier and Horne's argument against customs unions

Ethier and Horne (1984) produced a variation on Meade's proposition, arguing that a full customs union with no internal tariffs is necessarily worse than partial union with non-zero internal tariffs. They start with full union and then consider what happens when the internal tariff is raised slightly. In other words, their approach is the reverse of Meade's but their conclusion is the same.

They also put forward the notion of trade modification e.g. if the formation of a customs union affects the quantity of cars demanded, it will also affect the quantity of petrol imported (that is, a full model must consider the impact of integration on complementary goods also).

Winters (1988) provides support for customs unions. He analyses the consequences of prohibiting members from taxing or controlling inter-member trade and the potential dangers of members resorting to subsidy-based protection. The Ethier-Horn argument for internal tariffs in a CU is shown to be inapplicable to most of the EU's existing, cost-increasing barriers to trade. The implication of abolishing both Article 115, which prevents trade deflection, and MCAs on agricultural trade are examined. National production subsidies are examined in the context of the free intra-EU mobility of capital and skilled labour. Subsidies are doubly harmful to a small open economy in the presence of factor mobility. Moreover, even where rent-snatching subsidies may be profitable for the EU as a whole, subsidy wars could erupt between member states as they compete for rent-generating industries.

2.2.3 Free Trade Areas versus Customs Union

Various points may be made here. Firstly, consider the case where countries are small but efficient producers of a limited number of products and have relatively low initial tariffs (e.g. EFTA). For such countries, it is argued that a free trade area is better than a customs union because it does not involve a common external tariff.

Suppose, within a customs union, the tariff is fixed at the average of the pre-union tariffs of the member countries. For each good, for some members the external tariff will rise and for others it will fall. The outcome will be inefficient trade diversion to within the area but no external trade creation through expansion of demand within the group. For example, suppose that, before the formation of the customs union, H had a tariff of 20% against both P and W while P's tariff against H and W was only 8% and that, after the union, both H and P levy a common external tariff of 14% against W. Trade creation comes about from the removal of H's tariff against P - it is not influenced by the size of the tariff against W. However, trade diversion will have increased. Prior to the union, W will have been able to make exports to P which will disappear as a result of P's tariff against W rising from 8 to 14%. Thus, the rest of the world will be worse off and some countries within the group will be worse off with a customs union than with a free trade area.

On the other hand, it may be argued that free trade areas result in trade deflection as well as the deflection of production and investment. Trade deflection occurs if imports enter high-tariff members of the FTA through low-tariff members. With no rules of origin, and ignoring transport costs, imports would enter the FTA via the country which applies the lowest tariff for the relevant commodity. The FTA may also produce an uneconomic structure of production with detrimental effects on world efficiency to the extent that the manufacture of products with a high percentage of foreign-made materials and semi-finished products shifts to low-tariff countries.

Deflection of investment may follow deflection of production as foreign investors move funds to countries with lower tariffs on raw materials and semi-finished products. Similarly, factories will be set up to assemble parts produced in W with low labour costs if tariff advantages make this operation possible.

El Agraa's (1994) summary of the position is as follows:

1. There are no fundamental differences between FTAs and CUs with regard to trade creation. If there is any difference it is likely to be one of magnitude.

2. It is possible that the adoption of a common external tariff may lead to trade creation in circumstances which might produce trade diversion in an FTA. This requires an assumption of irrational tariff levels, but this is far from a remote possibility. To this extent a CU is preferable to an FTA.

3. Trade deflection is peculiar to FTAs. If there exist no rules of origin, an FTA will produce a common external tariff equivalent to the lowest tariff among members for each commodity and hence is preferable to a CU. However, FTAs in practice do have rules of origin and since there is no evidence that tariffs are "made to measure", it seems that customs unions are preferable to free trade areas. This conclusion is reinforced by a consideration of deflection in production and investment.

2.3 Possible Reasons for the Formation of Customs Unions

Given some of the doubts raised about the benefits to be obtained from customs unions, we need to consider briefly why so many have been formed. Possible arguments include:

(i) The large union case mentioned above;

(ii) the real rationale is protectionism: a CU may offer the cheapest means of protecting industries threatened from outside (e.g. by the newly industrializing countries). Further, because the cost of protecting industry is reduced, there is an income and substitution effect which allows more of it to be protected.

(iii) economic integration is a prerequisite for and stimulus to political integration rather than an aim in itself; thus a CU may be thought to be just one step on the way to political union.

(iv) Nielsen et al (p.36) put forward the argument that a customs union should be seen as a means of obtaining common tariff reductions. Thus, the first-best policy for a country may involve unilateral decreases in tariffs (with the ultimate aim of unilateral free trade) but this may not be possible for internal political reasons.

(v) The basic trade creation/trade diversion model treats H only as an importer. The importing country clearly loses from trade diversion because it results in a worsening of its terms of trade. However, P which is treated as an exporter gains from trade diversion. Since each member of a customs union will, in practice, be both an exporter and an importer, the losses through trade diversion on imports might be matched by gains through trade diversion on exports. If this was indeed the case, it could be argued that the formation of a customs union would be a way of obtaining trade creation gains without running into balance-of-payments difficulties. Further, the customs union may help to reduce the short-term balance of payments problems which might arise because resources are not in reality fully mobile. A non-discriminating tariff reduction increases imports through trade creation. But this produces a balance of payments deficit unless exports can be increased at the same time. The customs union may achieve this through trade creation in partner countries. Also, in a customs union, exporters can be more confident of continued unrestricted access to markets than in international trade generally and this may make export adjustment easier.

(vi) The basic model is too simple because it assumes that W does not impose tariffs. If W does impose tariffs, as is likely, H and P (prior to the formation of the customs union) will face two world prices - one at which they import from W; another at which they can export to W. This distinction between the world export and import prices would be strengthened if it were assumed that there were transport costs between H/P and W but no such costs (or negligible ones) between H and P. Using the assumption of a tariff-ridden world, Wonnacott and Wonnacott (1981) show that there are some welfare gains from customs union membership that cannot be secured through unilateral non-discriminating tariff reductions. These arise where the exporting member of the customs union is a low-cost producer, but prior to the customs union could not fully exploit its comparative advantage because of the tariffs imposed by other countries. Södersten and Reed develop this argument on pages 332 and 333. This argument can be linked to point (iv) above since it may be possible to use customs union formation to bargain down tariffs with W.

(vii) Individual countries may have different reasons for joining regional trade groupings. They may do so, for example, to underpin firmly controversial domestic reforms (as in the case of Mexico). The great enthusiasm for the USA for regional groupings reflects their desire to force open export markets around the world. Other countries join existing groupings simply out of fears of the effects of exclusion from them.

There are three other, more theoretical, arguments.

2.3.1 Redistribution among members; gains & losses for the union as a whole

In our comparative static example of gains and losses from customs union formation, we concluded that the union as a whole would be better off from integration if:

b + d > e

where b occurs because resources are released as the inefficient industry in H contracts; d is a consumption gain resulting from a better allocation of consumer expenditures; e represents the production loss resulting from trade diversion. However, e is a gain to P's producers and is thus a gain within the union. This is an example of a member gaining while another loses. In general, it is very likely that gains from integration will be unevenly spread among countries.

Assume that there is a costless redistribution of gains among partners. In this case, it is possible to prove that the formation of a CU need never harm any country (Kemp and Wan, 1976). Assume that before union there was a full (tariff-ridden) equilibrium. Then, after union, it will be possible to find a CET structure which would make W's trade with the union just equal to its trade with H and P combined before the union. Thus, W would not be worse off as a result of the union between H and P. H and P's original consumption bundles would still be available, because their combined production possibilities and net trade with W would be unchanged.

At worst, then, the union would have no effect on H and P. However, it is very likely that production and consumption could be re-arranged to make both H and P better off.

This model is unrealistic, but it shows that the harmful effects of customs unions derive not from integration per se but from the absence of optimal trade policy and redistribution of gains. It also shows that, in theory, it is possible to shift the world economy from tariff-ridden trade to free trade through a series of Pareto-improving customs unions.

2.3.2 The Public Goods defence of Customs Unions

This argument was developed by Cooper and Massell (1965a) and by Harry Johnson (1965). Let us consider Johnson's version. He assumed:

(i) governments use tariffs to achieve certain non-economic objectives;

(ii) actions taken by governments are aimed at offsetting the differences between private and social costs;

(iii) government policy is a rational response to the demands of the electorate;

(iv) countries have a preference for industrial production

He distinguished between private and public consumption goods; real income (utility enjoyed from both public and private consumption, where consumption is the sum of planned consumption expenditure and planned investment expenditure); and real product (defined as total production of privately appropriable goods and services).

Competition among political parties will make the government adopt policies that will maximize consumer satisfaction from both private and collective consumption goods. Satisfaction is maximized when the rate of satisfaction per unit of resources is the same in both types of consumption goods. Collective preference for industrial production implies that consumers are willing to expand industrial production (and employment) beyond what it would be under free trade. The policy will be mainly financed by tariffs and protection will be carried to the point where the value of the marginal utility derived from collective consumption of domestic and industrial activity is just equal to the marginal excess private cost of protected industrial production.

This marginal excess cost of protected industrial production consists of two parts: (a) the marginal production cost which is equal to the proportion by which domestic cost exceeds world market cost (in a very simple model this is equal to the tariff rate); and (b) the marginal private consumption cost which is equal to the loss of consumer surplus resulting from the fall in consumption brought about by the tariff. In equilibrium, the proportional marginal excess private cost of protected production measures the marginal degree of preference for industrial protection.

See Diagram 2.1

It can be argued that in the above case the optimal intervention strategy is a direct subsidy to the industries whose production the government wishes to promote. Tariffs are only a second-best policy and an economic rationale for CUs can only be established on public goods grounds if for political or other such reasons governments are denied the use of direct production subsidies. But GATT does prohibit the use of subsidies in international trade and so most countries use tariffs even though subsidies are superior to them.

Nielsen et al argue (p 35) that this still only provides a case for an optimal non-discriminating tariff rate at the level at which marginal net welfare gain in the production of the socially desired goods is exactly equal to the marginal welfare loss in the consumption of these goods. They suggest that an economic rationale for customs unions requires further restrictions on the model such as the existence of collective preferences for trade with specific partner countries and the prohibition by the GATT of discriminating tariffs except in connection with a customs union.

Johnson develops his model by assuming that industrial production consists of a variety of products in which countries have differing degrees of comparative advantage, that countries differ in their overall comparative advantage in industry compared with non-industrial production, that no country has monopoly/monopsony power, and that no export subsidies are allowed. Countries may be both importers and exporters of industrial products and this, in combination with the preference for industrial production, will motivate each country to practice some degree of protection. A country can gratify its preference for industrial production only by protecting import-competing industries. The condition for equilibrium remains the same, but S_H+U changes because the protection of import-competing industries will reduce exports of both industrial and non-industrial products (for balance of payments reasons). Hence, in order to increase total industrial production by one unit, it is necessary to increase protected industrial production by more than one unit to compensate for the induced loss of industrial exports. The protection of import-competing industries also reduces industrial exports by raising their production costs (due to perfect factor mobility). The stronger this effect, ceteris paribus, the higher the marginal excess cost of industrial production. This marginal excess cost will be greater, the larger the industrial sector relative to the non-industrial sector and the larger the protected industrial sector relative to the exporting industrial sector.

Now, compare:

(a) reciprocal tariff reductions on a most-favoured-nation basis: the reduction in tariff rates will increase imports from all other countries;

with

(b) discriminatory tariff reductions (starting from a position of non-discrimination).

There are two advantages in (b):-

(i) each country can offer its partner an increase in exports of industrial products without any loss of its own production by diverting imports from third countries (TD);

(ii) when TD is exhausted, any increase in partner industrial exports to this country is exactly equal to the reduction in industrial production in the same country (TC), hence eliminating the gain to third countries.

Thus, discriminatory reciprocal tariff reduction costs each partner less, in terms of reduction in domestic industrial production (if any) incurred per unit increase in partner industrial production, than does non-discriminatory reciprocal tariff reduction.

On the other hand, preferential tariff reduction imposes an additional cost on the tariff reducing country: the excess of the costs of imports from the partner country over their cost in the world market.

The implications of this analysis are:

(i) both TC and TD yield a gain to the CU partners;

(ii) TD is preferable to TC for the preference granting country since a sacrifice of domestic industrial production is not required; and

(iii) both TC and TD may lead to increased efficiency through economies of scale.

2.3.3 Domestic distortions and customs unions

A substantial literature on the question of whether formation of a CU is economically desirable when there are domestic distortions e.g. effect of trade unions, minimum wage legislation. The usual assumption is that domestic distortions result in a social average cost curve below the private one. This shows that a CU can produce gains. However, it is not an absolute argument for a CU since it can also be shown that better results still would have been obtained from the use of first-best instruments to eliminate the distinction between private and social costs in P. This being so, the argument for a CU depends on the first-best instruments not being available for some reason.

2.3.4 El Agraa's summary of economic arguments for customs unions

1. The rationale for regional economic integration rests on the existence of constraints on the use of first-best policy instruments.

2. Even when the existence of constraints on superior policy instruments is acknowledged, it is misleading to identify the results of regional economic integration by comparing an arbitrarily chosen common policy with an arbitrarily chosen national policy. Ignorance and inertia may provide sufficient reason for existing national policies being sub-optimal but it is wrong to attribute gains which could have been obtained from appropriate unilateral action to a policy of regional economic integration. Again, although it is reasonable to use the optimal common policy as a point of reference, this may overstate the gains to be achieved if constraints and inefficiencies in the policitical processes by which policies are agreed upon prove to be greater among a group of countries than within any individual country.

3. Despite 1 and 2, there is a strong general case, in principle, for regional integration. Where economies of scale are in part external to national industries, the rationale for economic integration rests essentially upon the recognition of the externalities and market imperfections which extend beyond national boundaries. Under such circumstances, unilateral national action will not be optimal whilst integrated action offers scope for gain.

4. The achievement of the potential gains from economic integration will be limited to countries able and willing to cooperate to distribute these gains so that all partners may benefit compared to the results achievable by independent action.

2.3.5 The views of the World Trade Organisation

The question of whether regionalism undermines globalism has been much discussed in recent years. A recent study of regionalism from the IMF by de la Torre and Kelly (1992) argues that `beyond a certain threshold an undue emphasis on regionalism would undercut the multilateral trading system and render it inoperative'. If the authors are right, regionalism makes globalism more essential but simultaneously enfeebles it. This is no cause for celebration.

But the WTO argues in a report published in 1995 that the rapid growth worldwide of regional economic groupings has not so far impeded the development of freer world trade and has sometimes helped to promote it. The report says that there have been no fortress-type regional arrangements among WTO members. They find no evidence that discriminatory trade blocs may be emerging in North America, Western Europe and the Asia-Pacific region. However, it suggests that, as the world trade policy agenda advances, the WTO may need stronger procedures for deciding whether regional groups are fully compatible with multilateral rules.

The report also expresses doubts about the economic value of non-reciprocal agreements, such as the EU's preferential trade links with the ACP countries. Such arrangements provide limited economic gains and can have protectionist consequences, because they enable strong countries to restrict imports from weaker ones in `sensitive' sectors, such as agriculture and textiles.

By the end of 1994, the GATT had been notified of 108 regional arrangements, 33 of them in the last five years. Almost all of the more than 120 members were linked to at least one such grouping. The proportion of world trade conducted within regions had risen from 40.6% in 1958 to 50.4% in 1993. However, there was no evidence that reciprocal regional agreements had distorted trade and investment flows at the expense of the world economy.

Only in the EU had intra-regional trade increased faster than trade with third parties. Nonetheless, EU trade with third parties had continued to grow in line with its members' economic growth.

Some regional agreements have enabled members to accept obligations to liberalise faster than required by the GATT and have helped stimulate progress at the multilateral level. However, it is emphasized that the Uruguay round agreement embodies commitments on agriculture, goods, services and intellectual property which go further than most regional arrangements.

2.4 Likelihood of gains and losses from customs unions

It is possible to develop some general propositions regarding gains and losses in particular cases (although there are general second-best difficulties with the existence of other distortions). The principal propositions are:

(a) It is likely that TC will be greater than TD if countries joining together in a CU are similar in the range of products they produce before the formation of the union. This is because trade creation occurs through the replacement of domestic production by more efficient production within the union. If future members produce essentially different goods, there will be little scope for such replacement and, hence, little trade creation.

It follows that the potential gain from trade creation will be greater if the range of efficiencies among future partners in comparable activities is large. The problem, however, is that if this is so, integration may well lead to substantial frictional and structural unemployment.

(b) TC is more likely in the long run (a dynamic gain), the more different the economies of partners are capable of becoming in production after the formation of the union. This will ease any problem of frictional and structural unemployment under (a).

(c) TC is more likely the greater are the initial tariff rates among future partners. However, such differences can be thought of as being just a proxy for differences in production costs among countries and so this is making much the same point as the second part of (a).

(d) TC is more likely, the higher is the elasticity of demand for imports on which duties are removed since the higher this elasticity is, the greater will be the gains to consumers through integration.

(e) TC is more likely, the higher is the elasticity of supply of exports from partners since this will increase the extent to which inefficient domestic industry will be replaced by more efficient production in fellow members.

(f) The gains from integration are likely to be greater, the higher is the proportion of the total pre-integration trade of a country made up by trade with future partners i.e. the greater is the ratio of intra-trade to total trade (before integration).

(g) Net TC is more likely, the lower the common external tariff (CET) on those products for which embers remove internal tariffs. Losses from TD will be minimized if the CET is not greater than the average of the previously existing independent national tariffs.

(h) Net TC is more likely, the larger is the preferential area. The extreme case in which there could, by definition, be no trade diversion will be where the customs union includes all countries (this would, of course, be the case of universal free trade).

Appleyard, Conway and Field (1989) examined the effects of a customs union on the pattern and terms of trade in a Ricardian model with three countries and a continuum of goods. The pattern and terms of trade are determined endogenously within the model. Trade creation and trade diversion impacts are shown to be determined by the position of countries' traded goods on the continuum. In addition, the desirability of economic integration is found to be dependent on the level of country development. In particular, there are gains for mid-continuum countries (NICs).

2.5 Customs Unions and Developing Countries

Customs unions for developing countries are part of an overall import-substitution development strategy. The argument for such a strategy depends on infant industry and balanced growth ideas as well as on the unreliability of foreign markets for surplus product. However, there are often great costs in terms of allocational efficiency e.g. the need for complicated import-control systems, exchange controls, multiple exchange rates etc.. Also it is possible for the choice of industries to develop to be wrong (for example, it was often argued in the past that the prestige associated with heavy industries led many developing countries to concentrate too much on their development).

Still, if import-replacement is accepted as a development strategy, then:

2.5.1 Advantages of Integration for Developing Countries

(a) the size of the market is important: the formation of a CU allows the expansion of the domestic market for customs union production;

(b) trade diversion from lower-cost sources in developed countries is desirable but the sharply-rising marginal cost of protection emphasizes the importance of concentrating on least-cost industries;

(c) the effect on world welfare is hardly a problem since the international trade flows involved are bound to be only a very small proportion of total world trade flows;

(d) domestic labour drawn into trade-diverting activities may have previously been unemployed or under-employed and may thus have a low opportunity cost;

(e) the customs union is likely to provide increased security for local entrepreneurs. Local firms in developing countries (faced by a high degree of market imperfection, insecure factor supplies, uncertainties associated with world market prospects and risks associated with unstable political regimes) typically favour low outputs and high profit margins rather than attempting to build optimal scale plants for meeting estimated future demand.

(f) studies have also shown gains from saving of scarce foreign exchange.

2.5.2 Problems associated with Customs Unions for Developing Countries

(a) the problem of the distribution of gains since the integrating area is bound to consist of countries at different stages of economic development. There will be a tendency for new industries and other economic activity generated by the enlarged market to gravitate towards the most highly-developed centres. Consequently, several customs unions have adopted schemes for the allocation of new industries among the countries making up the customs union in the hope of securing an "equitable" distribution of the industry gain;

(b) the high dependence of developing countries on indirect tax revenues, especially on levies on foreign trade produces a need for the acceptance of a formula for the distribution of tariff revenues among members;

(c) there are often inadequate transport facilities to make the enlarged market economically meaningful. There is not much point in having a free trade area in South America if member countries find it cheaper to ship to North America and Europe than across the Andes. The trade orientation of developing countries historically towards the industrial nations will often influence relative transport costs;

(d) meaningful integration requires the lifting of exchange restrictions as well as trade restrictions. This is often not done.

2.5.3 Broad Approaches to Customs Unions among Developing Countries

In their `new view of customs unions', Cooper and Massell proposed two broad approaches to customs unions for developing countries:

(a) Specify the set of efficient external tariffs which lead to the desired amount of industrial production in the customs union as a whole.

(b) Allow each country to produce any specified level of industrial goods in its own territory and use the CET for revenue-raising purposes with the aim of maximizing joint income of the customs union.

The problem with (b) is that of regional imbalance - with industries unequally distributed amongst partners. Thus it must be supported by compensation payments among partners but these are a poor substitute for real industrial development.

Thus, (a) is more likely in practice.

An alternative is to have a partial CU i.e. with a CET but not internal free trade. Members decide on the number of industries needed to achieve industrial output goals. These are allocated among members on the basis of comparative advantage. Some industries will be established by one member while protecting the home market but selling freely in other members' markets.

2.6 Trade Creation and Trade Diversion in the EC - empirical investigations

We have seen that there are considerable theoretical doubts as to whether it is desirable for countries to form customs unions. All that can be said is that it is necessary to look at particular cases. We have seen that it may be possible to say something about the likelihood of success of a given customs union on the basis of the pre-union characteristics of the member countries e.g. imports as a percentage of GDP, import demand elasticities, the direction of trade pre-union. However, this can at best be very approximate. Ex-post studies which attempt to look at actual effects of the formation of a union seem on the surface to be more promising and many attempts have indeed been made to measure the gains and losses resulting from the formation and then the expansion of the EC. There are, however, a number of problems associated with such empirical studies.

2.6.1 Problems of Measurement

A basic problem in attempting to measure the effects of customs union formation is the counterfactual problem that changes in production are influenced by many factors other than integration itself. This problem is made worse by the fact that there is little point in undertaking an ex-post study until the customs union in question has been in existence for some time as several of the effects may take a number of years to become apparent. The problem then is to compare the world as it has developed with the world as it would have been like if the customs union had not been formed, and this is ultimately unknowable. As Nielsen et al express it on page 29, the task is to create an anti-monde - a representation of the world as it would have been without the CU - and then to compare the actual world with it. Nielsen et al refer to this as `residual imputation'.

Another possibility is to compare developments in the customs union with developments in similar third countries over the relevant time period. This, however, requires the assumption that these countries have themselves been unaffected by the formation of the customs union. This is no problem in the case of small customs unions, but very awkward in the case of large customs unions such as the EC.

Still, there have been several attempts at both kinds of study. They have, for the most part, concentrated on TC and TD and hence on the impact of the CU on trade volumes. Trade figures are, however, expressed in money terms and can only be converted into volumes on the assumption of unchanged relative prices. An extra difficulty is that it is very likely that integration will change relative prices, producing an index number problem.

The standard approach to measurement is to calculate net TC: Total increase in imports by H (adjusted to exclude non-integration influences) - the fall in imports from non-members caused by integration.

The first difficulty here (other than the index number problem mentioned above) lies in the adjustment of figures to exclude non-integration influences. This is extremely difficult and dependent on the assumptions and interpretation of the researcher.

Another problem lies in the choice of dates, before and after integration, for the comparison. We have seen that it is essential to make the post-integration date sufficiently after the occurrence of integration to allow time for the effects of integration to have worked through but it is also necessary to make the pre-integration date sufficiently before integration to take account of changes occurring during the movement to full integration. In addition, it is desirable to ensure that the years chosen have not been influenced by one-off shocks which may cause them to provide a distorted view of longer-term economic developments.

An alternative approach involves the calculation of separate measures of TC and TD as changes in imports as a proportion of consumption. Standard formulas for these are:

TC = Change in Total Imports/Consumption (between two years before and after integration).

TD = Change in imports from third countries/C (between two years before and after integration).

The same problems apply as above.

2.6.2 Ex-post studies of the EC

Verdoorn (quoted in Scitovsky, 1958) saw the gains from customs union formation as relatively small. He estimated that trade between the countries of the then EEC would increase by 17 per cent, producing an estimated welfare effect of about 0.05 percent of the sum of the national incomes of the original six. Subsequent studies have given larger estimates, although with considerable variation. For example, Truman (1969) estimated that by the time the reduction of internal tariffs was sixty per cent complete, intra-EEC trade was approximately thirty per cent higher than it would have been and that this had been achieved with no net trade diversion from the rest of the world.

Williamson and Bottrill (1971) concluded that by 1969 intra-EEC trade was approximately 50 per cent higher than it would have been. Hine (1985) notes that there is some empirical support for the argument that member countries that had high tariffs before joining the EEC tended to experience trade creation, while those which had low tariffs suffered mainly trade diversion.

Other studies include Balassa (1967); Truman (1972); Major and Hays (1970); Balassa (1974); Prewo (1970) - for summary figures see Vickerman (1992). Also see Mayes (1978); Miller and Spencer (1977).

Nielsen et al (pp 29-33) examine some of the problems of early studies e.g. the unrealistic assumption that the shares of imports of member states and third-party countries would remain the same) and the attempts to eliminate the effects of income growth from analyses of foreign trade patterns. Reproduced here is the table taken from page 30 of Nielsen et al which shows estimates made of TC and TD in five studies (it is in turn taken from Balassa (1975).

--------------------------------------------------------------------------

Trade creation and trade diversion in the EC

Author Year TC %* TD %+

US$bn US$bn

-------------------------------------------------------------------------

Balassa 1970 11.3 13 0.3 1

Truman# 1968 9.2 26 -1.0 7

Kreinin# 1970 6.7 14 1.7 9

Williamson & Bottrill# 1969 11.2 25 0.0 0

Aitken 1967 9.2 14 0.6 2

# manufactured products only

* in per cent of total EC imports

+ in per cent of extra-EC imports

As well as looking at the figures, you should read the comments in Nielsen et al (pages 30-33). This will allow you to adopt a proper scepticism towards these and similar calculations. It will also remove any surprise that you might have felt over the differences in the estimates made. It remains that all five studies suggest that the trade-creating effects of the formation of the original six-country CU exceed the trade-diverting effects, although the exclusion of agricultural products from three of the studies removes what has always been thought of as the most likely source of trade diversion in the formation of the EC. Since the studies employ different methods, it is probably reasonable to attach some weight to this general conclusion. However, these studies do only relate to the original six member states and, according to the general rules we outlined above, we should expect there to have been less TD resulting from their integration than with the subsequent expansions of the union.

Even if one accepts the view that TC has comfortably exceeded TD, the estimates of welfare gains resulting from this are extremely small (around 0.15 % of GDP). Further, in these studies, the static welfare gains from the union are very likely over-estimated. In our diagram showing the net gains from the formation of a customs union (see Lecture 1), we showed that there would be a welfare gain if, and only if, the area (b + d) was greater than e, where b and d represented the production and consumption gains of trade creation while e represented the loss from trade diversion. However, it can be demonstrated (see Nielsen et al p. 33-4) that TC and TD, as measured in the various studies, bear little relationship to TC and TD in welfare terms.

Concentration on trade creation and trade diversion can mean that many other possible gains of integration are ignored. It has often been argued that the low estimates of gains from the formation of the EC have arisen because of the failure to consider such gains.

2.7 Additional possible gains from the formation of a customs union

The comparative static model used by Viner simply compares two equilibrium positions (before and after) while ignoring the process by which the new equilibrium comes about. That is, the model does not consider issues such as changing capital stocks, levels of industrial concentration or technical innovation. But all of these may be influenced by the formation of the customs union. The gains additional to net trade creation include:

(f) reduced uncertainty ® increased efficiency

(j) possible responses by firms in third countries who lose market share as

a result of the CU and may seek to fight back by, for example,

restructuring in order to reduce costs, accepting lower profit margins,

or investing in new technology. All of these carry with them gains for

the consumer.

We have already considered improvements in the terms of trade in the `beggar-thy-neighbour large CU case. Of the others, the most important are generally held to be (d), (a) and (e). Let us consider these.

2.7.1 Competition effects of customs unions

Nielsen et al break these into theoretical effects (pp37-40) and dynamic effects (pp 44-45). In the theoretical case, we return to our basic partial equilibrium diagram but now assume that prior to the formation of the CU, the home industry was operating technically inefficiently and thus that the CU has the effect of forcing an increase in efficiency which is indicated by a lowering of the domestic supply curve. This adds a cost reduction gain to the gain from trade creation. The problem with this is that our initial Vinerian model assumed perfect competition in both factor and product markets! With perfect competition, firms are by definition operating technically efficiently.

Thus, the cost-reduction argument needs to be set within a much more complex model in which imperfect competition is assumed and in which firms are free to choose to reject cost minimization as an objective. Then it can be argued that increased external pressure as a result of the formation of the CU may force firms to increase their efforts to reduce costs. In other words, the CU leads to a reduction in the degree of X-inefficiency.

The dynamic arguments rest on the assumption that increased competition leads to increased innovation. Firms seem only likely to engage in R&D activities if they are able in one form or another (through patents, copyrights etc) to maintain for some time at least a quasi-monopoly position which allows them to earn super-normal profits from their innovations. In any case, only large countries may have the resources necessary to carry out significant R&D programmes. Putting these arguments together with the argument on incentives from competition leads Nielsen et al to propose that the most innovative firms are probably found under conditions of monopolistic competition or oligopoly. They also point out that the causal direction between market form and R&D is not clear. It may be that changes in market form affect innovation, but the reverse may also be true, as in the product life cycle hypothesis.

Again, the extent to which integration produces a more competitive environment depends on the nature of competition policies within the customs union. Certainly, it is not clear what effect a change in market form resulting from the formation of a customs union will have on total R&D and hence on innovation. Empirical research does not help in this matter.

2.7.2 Internal economies of Scale

Again Nielsen et al split the argument into two. The first section (pp 40-2) deals with internal economies of scale arising, for example, within trades or industries with high fixed costs and lower constant marginal costs. Thus the firm's average costs are diminishing, a situation incompatible with perfect competition. Assume then that, prior to the CU, both prospective partners were protecting their home industries, keeping out imports and growing only to the size allowed by the home market. The formation of the CU will then cause the whole market to be taken over by the more efficient of the two firms, lowering average costs of production. Price will fall and consumption will increase in both countries. The production and consumption gains in the home country will be increased and there will be an additional cost reduction gain in the partner country.

However, other assumptions may be made which will produce different results e.g. that pre-CU the home country did not protect its home industry, choosing to import from the third-country instead. Then the formation of the CU will, as usual, cause a trade diversion effect to set against the cost reduction effect in the partner country which now reaps the benefit from economies of scale. A more interesting case is that in which, prior to CU, the high-cost home country protected and produced at home but the medium-cost partner country did not do so and imported from the third country. Now the CU produces TC for the home country but trade suppression for the partner country which now replaces lower cost imports with higher cost domestic production (compare this with the definition of trade diversion).

The argument may be complicated in other ways. For example, assume that prior to the formation of the CU, both H and Pprotect, and that H is (as always) the least efficient. Nonetheless, its pre-union production may be higher than that in the more efficient P (because the home market of H is larger than that of P). Then, it may be the case that actual unit costs in H are lower than in P because H has been able to exploit more of the potential internal economies of scale. It may follow then that, after the formation of the CU, production will be concentrated in H, the country with the higher average cost curve. This might also happen in the trade suppression case above if the industry cannot easily be established in the potentially lower-cost partner country after the formation of the CU. Both of these arguments depend on imperfect information about demand and production conditions everywhere in the customs union.

It is also possible that potential gains from economies of scale may be thwarted by consumer tastes. If these differ across the market so much as to encourage the maintenance of much differentiation in production lines. That is, if each member's citizens prefer the products of their own country (irrational though this may be in economic terms), integration will not lead to economies of scale. This result may also come about through imperfect information.

Dynamic economies of scale arise from learning effects associated with large scale production. These cause the firm's average cost curve to shift downwards. Thus, it is argued, that the formation of the CU will cause increased specialization and this will be reflected in an increase in accumulated production (an expression of the gains accruing to a firm from increased experience in production, marketing and distribution of its products). The firm's average cost curve will shift downwards at a higher speed than it would have done in the absence of the CU. However, the difficulty with the argument that integration produces increased specialization is that the relocation of production depends on the stage of industrial development in each country, the specificity and mobility of factors of production, and on the inherited stocks of industrial capital.

As well as doubts existing regarding the theoretical benefits resulting from the formation of a CU through economies of scale, there are measurement difficulties. For example, there are problems associated with the measurement of average plant size. Much evidence is inferential, relying on comparisons with US production and consumption patterns and plant-size. Hence there are doubts as to the extent to which there are unreaped economies of scale available in countries which have quite large domestic markets even without integration.

Further, there is a question as to the extent to which economies of scale are linked with higher rates of economic growth, even in those areas where economies of scale are generally held to be most likely e.g. aerospace, vehicles, computers, finance, marketing, and research and development.

El Agraa in The Theory and Measurement of International Economic Integration provides a detailed analysis of potential gains and losses from economies of scale, proposing that even if the overall result is a net gain, the distribution of these gains is important consideration. Alternatively, if economies of scale accrue to an integrated industry, then the locational distribution of the production units is crucial. He provides a mathematical treatment of economies of scale from which he derives seven propositions:

(i) if the elasticity of returns to scale of the exportable industry is equal to or greater than that of the importable industry, TC I (the replacement of H's consumption of the domestically produced high cost good by imports of the same product from W) always leads to improvement in welfare, given a positive price-output response;

(ii) if the elasticity of returns to scale of the exportable industry is smaller than that of the importable industry, TC I may lead to deterioration in welfare.

(iii) In the presence of variable returns to scale, TD I (the replacement of H's import of the good from W by imports of the same product from P because of the discriminatory tariff abolition by H on imports from P only) may lead to improvement in welfare;

(iv) If the elasticity of returns to scale of the exportable industry is equal to or greater than that of the importable industry, TCII (the replacement of H's imports of the good from producers in P by producers in W) leads to improvements in welfare when there is a positive price-output response;

(v) If the elasticity of returns to scale of the exportable industry is smaller than that of the importable industry, TC II may lead to a reduction in welfare;

(vi) If the elasticity of returns to scale of the exportable industry is equal to or greater than that of the importable industry, TD II (the replacement of H's imports of the good from producers in W by producers in P due to the discriminatory tariff on imports from W) leads to a deterioration in welfare when there is a positive price-output response;

(vii) If the elasticity of returns to scale in the exportable industry is smaller than that of the importable industry, TD II may lead to improvement in welfare.

El-Agraa, however, goes on to say that the mathematical treatment of TC and TD under conditions of economies of scale is either unnecessary or useless and is made purely for their mathematical convenience rather than to advance our knowledge in this field of economics!!!

2.7.3 Effects on growth rates

The standard argument is that gains from economies of scale, increased specialization and enhanced competitiveness (that is, market expansion) lead to a faster rate of growth and this induces an increase in investment (investment creation), further increasing the growth rate.

However, it is also possible that integration will produce investment diversion with investment being diverted from its most rational location in the world economy to the integrating region because of the tariff discrimination produced by integration.

2.8 Likelihood of Dynamic Gains

Many studies have been done to investigate the likelihood of dynamic gains occurring in practice. For example, Venables (1987) constructed a model which focused on an industry in which all firms produce an identical product and follow Cournot behavior. Thus there is intra-industry trade. The number of firms in each country is endogenously determined by free entry and increasing returns to scale which ensure thet the number is finite. Taxation and tariff policy changes the profitability and and hence the number of firms in each country. This in turn changes the degree of competition and thus the price-marginal cost mark-ups, scale and average costs. Thus, there is scope for an active tariff policy on the part of a single economy although this will damage trading partners. Co-ordinated policy and the formation and enlargement of a customs union are considered. The analysis confirms the view that customs unions permit both an increase in firm scale and an intensification of competition in member countries. Tariff reductions cause firms to expand production and free entry causes a reduction in price-marginal cost margins. However, the gains to either customs union membership or enlargement are not unambiguous. Countries which are net importers of the products of imperfectly competitive industries may suffer welfare losses from membership; and expansion of the customs union may benefit those joining at the expense of existing members.

2.9 Effects of the widening of the EU

All of the studies quoted earlier related to the early days of European integration and dealt only with effects on the original six members. It is, of course, equally important to consider the impact of the widening of the area. We have looked above at some of the principles which might determine whether a particular widening would be successful or not but answers to such questions also require empirical studies. One such study was that of Yannopoulos (1987) which considered possible trade effects arising from Spanish accession to the EU. He concluded that there was strong evidence for potential trade displacement effects against the exports of the GMP as a result of Spain's accession to the EC. But it was not certain that these would emerge in practice? He argued that this would depend on, among other things, the supply responsiveness of both indigenous Spanish firms and `footloose' international production to the changes in cost competitiveness. This, in turn, would depends on the structural characteristics of the industries in question. There was a possibility, he suggested, of MNEs in `footloose' industries moving to Spain to take advantage of its improved competitive position. There was less certainty regarding indigenous firms due to structural weaknesses (sub-optimal plant sizes, use of outmoded equipment, undercapitalization, neglect of R&D). But these reservations re the expected size of trade displacement effects did not necessarily apply to agriculture. As the experience of Ireland in the 1970s and of Greece in the 1980s had demonstrated, the application of the CAP in low income/low price economies could be substantial.

Notes

1. This, however, is not necessarily the case. If, prior to the customs union, P imported the good from W but after the customs union was formed it exported the good to H, it is possible that both H and P could lose through trade diversion. See Södersten and Reed p. 330.

Bibliography

Appleyard D R, P J Conway and A J Field (1989), `Effects on pattern and terms

of trade in a Ricardian model with a continuum of goods', Journal of

International Economics, Aug

Balassa B (1967) `Trade Creation and Trade Diversion in the European Common

Market', Economic Journal, 78, 971-8

Balassa B (1974), `Trade creation and trade diversion in the European Common

Market: an appraisal of the evidence', Manchester School, 93-135,

reprinted in in B Balassa (ed) (1975)

Balassa B (ed) (1975), European Economic Integration, Amsterdam: North Holland

Cooper R C and B F Massell (1965), `A New Look at Customs Union Theory',

Economic Journal, 75, 742-7

de la Torre a and M R Kelly (1992), Regional Trade Arrangements, Occasional

Paper 93, Washington DC: International Monetary Fund

El-Agraa A M (ed) (1994), The Economics of the European Community, 4th edn,

Hemel Hempstead: Harvester Wheatsheaf

Ethier W and H Horn (1984), `A New Look at Economic Integration' in H Kierzkowski (1987), Protection and Competition in International Trade,

Oxford: Oxford University Press

Hine R C (1985), The Political Economy of European Trade, Brighton: Harvester

Wheatsheaf

Johnson H G (1965) `An economic theory of protectionisn, tariff bargaining and

the formation of customs unions", Journal of Political Economy, 73, 256 -83

Kemp M C and H Wan (1976), `An elementary proposition concerning the formation

of customs unions', Journal of International Economics, 6, 95-7

Major R L and S Hays (1970), `Another look at the Common Market', National

Institute Economic Review, Nov

Mayes D (1978), `The effects of economic integration on trade', Journal of

Common Market Studies, XVII, Sept, 1-25

Meade J E (1955), The Theory of Customs Unions, Amsterdam: North Holland

Miller M H and J E Spencer (1977), `The static economic effects of the UK

joining the EEC, a general equilibrium approach', Review of Economic

Studies, 44, 71-93

Nielsen G U-M, H Heinrich and J D Hansen (1991), An Economic Analysis of the

EC, London: McGraw-Hill

Prewo W E (1974), `Integration effects in the EEC: an attempt at

quantification in a general equilibrium framework', European Economic

Review, 5, 379-405

Scitovsky T (1958), Economic Theory and Western European Integration, London:

George Allen and Unwin

Södersten B and G Reed (1994), International Economics, 3rd edn. London:

MacMillan, ch 16

Truman E M (1969), `The European Community: trade creation and trade

diversion', Yale Economic Essays, 9, Spring, 201-57

Truman E M (1972), `The production and trade of manufactured products in the

EEC and EFTa: a comparison', European Economic Review, 3, 271-90

Venables A J (1987), `Customs Unions and Tariff Reform Under Imperfect

Competition', European Economic Review, 31 no 1/2, 103-10

Vickerman R W (1992), The Single European Market, London: Harvester Wheatsheaf

Williamson J and A Bottrill (1971), `The impact of customs unions on trade in

manufactures', Oxford Economic Papers, 23, 323-51

Winters L Alan (1988), `Completing the European Internal Market: Some Notes

on Trade Policy', European Economic Review, 32/7, 1477-99

Wonnacott G P and R J (1981), `Is unilateral tariff reduction preferable to

a customs union? The curious case of the missing foreign tariffs',

American Economic Review, 71

Yannopoulos G N (1987), `Trade effects from the extension of customs unions

on third countries: a case study of the Spanish accession to the EEC',

Applied Economics, 19/1, 39-50