The Future of Preferential Trade Arrangements
for Developing Countries
and the Current Round of
WTO Negotiations on Agriculture

Stefan Tangermann
Institute of Agricultural Economics
University of Göttingen
stanger@gwdg.de

Paper prepared for FAO/ESCP

April 2001

Comments made by participants in an Informal Expert Meeting on 1 February 2000, organized by FAO, and written comments by Jim Greenfield and Alan Matthews are gratefully acknowledged. The author alone is responsible for any errors and omissions

Contents

0 Executive Summary and Recommendations

1 Introduction

2 The Nature of Major Preferential Arrangements

3 Conceptualising Trade Preferences As an Element of Economic Relations Between Developing and Developed Countries

4 The Benefits of Trade Preferences

5 The Costs of Trade Preferences

6 The Status of Trade Preferences in the WTO

7 Alternative Options for the Future of Trade Preferences in the WTO

8 When Preferences Get Lost: Is there a Case for Compensation?

9 Conclusions

References

Appendix: Price and Quantity Effects of Trade Preferences and MFN Tariff Reductions

Executive Summary and Recommendations

1. Trade preferences for developing countries have been a feature of industrialised countries’ commercial policies since nearly forty years. However, with overall trade liberalisation, tariff preferences are gradually losing importance. In agriculture, on the other hand, tariff preferences can still be potentially valuable because MFN tariffs are extremely high in many cases, though they are also in the process of being reduced. Yet, because of the ‘sensitive’ nature of their agricultural policies, developed countries have usually been rather reluctant to provide deep preferences for agricultural products. At the same time, some special preferential regimes have provided significant tariff preferences for selected agricultural products, for constrained groups of developing countries. The EU’s preferences for sugar imports from selected ACP countries are a case in point. Hence the picture is very diverse when it comes to preferential treatment of agricultural exports from developing countries.
In this situation, one can ask a number of questions regarding the future of trade preferences in the ongoing round of WTO negotiations. Should developing countries strongly defend their trade preferences, and try to improve them? What are the benefits and costs of trade preferences? How do trade preferences compare to other forms of assistance for economic development? Are developing countries going to lose a lot when agricultural trade is further liberalised and preference margins are eroded? If so, do they have a right to compensation? Which form could such compensation take? Which role should trade preferences play in the next round of WTO negotiations on agriculture? The present paper makes an attempt at providing some tentative responses to such questions. Moreover, with the decline in MFN tariffs, other measures affecting trade such as standards are assuming a growing importance and the treatment of developing countries regarding such measures may also become of greater significance in the future. This subject is, however, outside the scope of this paper although it merits further study.

2. Regarding the nature of preferential arrangements for developing countries, three major forms can be distinguished, i.e. the Generalized System of Preferences, special preferential regimes for sub-sets of developing countries (such as Lomé/Cotonou or the Caribbean Basin Initiative), and regional free-trade areas between developed and developing countries. However, the last form, involving reciprocal preferences, does not belong to the category of trade preferences for developing countries in a strict sense.

3. When conceptualising trade preferences as an element of economic relations between developing and developed countries, the “trade rather than aid” perspective has some economic appeal. Trade preferences have the potential of helping developing countries to foster self-sustained economic development. They can substitute for, but probably also add to, economic transfers from developed to developing countries in the form of financial assistance. However, there are also drawbacks. The most obvious problem is resistance from the side of producers in developed countries. Less obvious, but still relevant is the problem of leading the production structure in developing countries in a direction which is not sustainable when overall trade liberalisation makes progress. In such cases, policies should be considered to “capture” a part of the rents from prefential schemes for use in programmes of benefit to farmers, rather than creating production patterns that would be unsustainable at future world prices following further liberalization of trade. Finally, there is the possibility of a loss in world welfare resulting from trade diversion.

4. Trade preferences can have various benefits for the exporting countries concerned. Empirical quantitative estimates of the overall size of these benefits are difficult, and hence rarely found. However, a relatively easily calculated indicator of potential benefits is the preference margin. Available estimates of preference margins show that they can amount to significant shares of the value of exports from the developing countries concerned. However, preference margins are a rather unreliable measure of economic benefits. Welfare gains for the exporting countries concerned are usually much smaller than the preference margin. Moreover, under certain conditions the preference margin flows to agents in the importing country, rather than to the exporting countries. In the absence of comprehensive analyses of preference benefits in individual beneficiary countries, the basis is relatively weak for judging which groups of developing countries ‘deserve’ preferences most. However, there are good intuitive grounds to argue that trade preferences are particularly important for the poorest countries and other vulnerable developing countries, such as small, island and land-locked countries.

5. However, trade preferences can also involve costs. Improving and expanding preferences requires ‘negotiating capital’. With successive rounds of MFN tariff reductions, the value of preferences is bound to decline, and it is important to assess carefully how much ‘negotiating capital’ should be invested in a business that may not be very profitable in the long run. Insistence on non-reciprocity of preferences can undermine the overall influence of developing countries in multilateral trade negotiations. Specific and deep preferences can potentially lead the production structure in the beneficiary countries in a direction that is not sustainable when MFN tariffs come down. Preferences tend to result in trade diversion, and the consequent costs for other exporting countries. Finally, countries benefiting from preferences may lose interest in MFN tariff reductions, and this is a cost to the multilateral trade regime overall.

6. Regarding the status of trade preferences in the WTO, universal trade preferences for imports from all developing countries, as extended under the GSP, are consistent with the GATT under the Enabling Clause. The same holds for specifically generous preferences granted to all least-developed countries. However, developed countries are not legally committed to providing such preferences. They can, therefore, decide unilaterally on preference margins, and also withdraw preferences without violating GATT/WTO commitments. Specific trade preferences for constrained sub-sets of developing countries, such as those provided under the Lomé Convention or under the Caribbean Basin Economic Recovery Act, however, are not consistent with the GATT. Yet, the WTO has in the past granted waivers that allowed the countries concerned to maintain these specific preferences.

7. When it comes to considering options for the future of trade preferences in the WTO, a number of issues come to mind. Rather than working towards an expansion of ‘shallow’ preferences for all developed countries under GSP regimes, an attractive alternative might be to aim at ‘deep’ preferences for the least-developed and vulnerable countries. In this context, the Enabling Clause could be amended by including small and other vulnerable countries, in addition to the least-developed countries, in the category of developing countries that can receive preferences deeper than GSP preferences. The functioning of the existing GSP preferences, which should certainly be maintained, can be improved by binding them in the WTO; removing conditionalities; setting preferential tariffs relative to MFN tariffs (rather than defining them as absolute tariffs); expanding TRQ volumes; simplifying rules of origin; and by providing better preferences where MFN tariffs exhibit peaks and tariff escalation.

8. The issue of compensation for the erosion of preference margins is highly complex. It is not necessarily clear that all reductions of MFN tariffs for products where preferences exist actually result in an erosion of (economically meaningful) preference margins. There are cases where the erosion of preference margins is (partly or wholly) outweighed by the favourable market effects of trade liberalisation. Moreover, where preference erosion clearly results in an economic loss to the exporting countries concerned, there are arguments both in favour and against compensation. In addition, if the case for compensation is accepted, it is not unequivocally clear who should ‘pay’, and who should ‘receive’ compensation. Various forms of compensation can be considered, without any of them having a clear-cut priority. And finally, in many cases it will be difficult to provide a reliable estimate of economic effect of preference erosion, and hence of the magnitude of compensation that may be justified.
All this is not to say that (i) preference erosion is a non-issue, and that (ii) compensation for preference erosion has no role to play in a multilateral round of trade negotiations. However, the issues discussed here should warn against suggesting simple solutions. In the end, compensation will be a matter of negotiation. As a rough guideline in discussing the issue of compensation, it may be useful to distinguish between two categories of preferences, i.e. GSP regimes on the one hand, and specific deep preferences for limited groups of developing countries on the other hand. Where GSP preferences erode as a result of multilateral negotiations on tariff cuts, the most natural way to negotiate on compensation may be to seek a structure of extra cuts of MFN tariffs that benefits developing country exporters. On the other hand, where very specific and deep preferences for individual countries and commodities are concerned, like under the EU sugar regime for the ACP countries, a relatively strong case can be made for cash compensation.

9. The analysis presented in this paper suggests a number of recommendations regarding the future of trade preferences in the current round of WTO negotiations on agriculture.

• Rather than working towards an expansion of ‘shallow’ preferences for all developed countries under GSP regimes, it may be more attractive to aim at ‘deep’ preferences for the least-developed and vulnerable countries.

• The Enabling Clause should be amended by including small and other vulnerable countries, in addition to the least-developed countries, in the category of developing countries that can receive preferences deeper than GSP preferences.

• Existing GSP preferences should be maintained, and legally bound in the WTO.

• Conditionalities attached to trade preferences (such as labour rights or environmental standards) should be removed where they are not in line with overall standards agreed multilaterally in the WTO.

• Preferential tariffs should not be defined as absolute tariffs, but should be set relative to MFN tariffs, i.e. as given monetary units below MFN tariffs (where MFN tariffs are specific tariffs) or given percentages of the MFN tariffs (where MFN tariffs are ad valorem tariffs).

• Where preferential tariffs under GSP regimes are constrained by tariff rate quotas, quota volumes should be expanded.

• Rules of origin should be simplified.

• Better preferences should be provided under GSP regimes where MFN tariffs exhibit peaks and tariff escalation.

• Such improvements in the regime of GSP preferences can be regarded as an element of compensation for the erosion of GSP preferences resulting from multilateral negotiations on MFN tariff reductions. In addition, compensation can also come in the form of extra cuts of MFN tariffs that benefit developing country exporters.

• Where very specific and deep preferences for individual developing countries and commodities are concerned, like under the EU sugar regime for the ACP countries, a relatively strong case can be made for cash compensation once preference margins erode as a result of multilaterally agreed MFN tariff reductions or as a consequence of domestic policy changes in the developed countries concerned.

• Studies are needed to assess more accurately the value of preferential arrangements to the recipient countries, including case studies of selected countries and commodities. There is also a need for studies on options for assistance, including legal support, that helps developing country exporters to cope with technical standards affecting trade, including under the SPS and TBT Agreements, and in markets of growing interest such as organic products.

0

Introduction

Trade preferences for developing countries have been a feature of industrialised countries’ commercial policies since nearly forty years now. They are considered an important element of the overall set of policies that help to improve the prospects for economic growth in developing countries, and an instrument for achieving a better balance of economic conditions between the richer and the poorer nations of the world. However, with overall trade liberalisation, tariff preferences are gradually losing importance. At the same time, some preferential schemes for special groups of developing countries, such as the Lomé Convention between the EU and countries in Africa, the Caribbean and the Pacific (ACP), have come under serious stress in the WTO. In this situation one may well ask what the future of trade preferences should be in the coming round of WTO negotiations. Should developing countries strongly defend their trade preferences, and try to improve them? What are the benefits and costs of trade preferences? How do trade preferences compare to other forms of assistance for economic development?

In agricultural trade, tariff preferences for developing countries have a somewhat special status. On the one hand, many tariffs are still extremely high in agriculture, and hence preferential arrangements can be potentially valuable. On the other hand, because of the ‘sensitive’ nature of their agricultural policies, developed countries have usually been rather reluctant to provide deep preferences for agricultural products. At the same time, some special preferential regimes have provided significant tariff preferences for selected agricultural products, for constrained groups of developing countries. The EU’s preferences for sugar imports from selected ACP countries are a case in point. Hence the picture is very diverse when it comes to preferential treatment of agricultural exports from developing countries. At the same time, agricultural trade has also been brought on the road towards liberalisation since the Uruguay Round has fundamentally changed the rules of the game in agriculture. Are developing countries going to lose a lot when agricultural trade is further liberalised and preference margins are eroded? If so, do they have a right to compensation? Which form could such compensation take? Which role should trade preferences play in the next round of WTO negotiations on agriculture?

The current study makes an attempt at responding to some of these questions. Given its limited scope, it cannot do this on a comprehensive basis, but has to concentrate on a number of selected issues. The literature on trade preferences, and in particular on trade preferences in agriculture, is surprisingly limited. In this situation, some of the discussion presented here is necessarily of a tentative nature, but may stimulate a more intensive discussion of a broader set of issues.

The study starts by providing a short impression of the nature of major preferential arrangements (Section 2). It then proceeds to discuss the place of trade preferences in the overall framework of economic relations between developed and developing countries (Section 3). On this basis, major benefits and costs of trade preferences are discussed (Sections 4 and 5). The changing role of trade preferences in the GATT/WTO legal framework is outlined (in Section 6) before the future of trade preferences in the WTO is considered (Section 7). The study ends with a discussion of the issue of preference erosion and potential compensation (Section 8) and some conclusions (Section 9).

1 The Nature of Major Preferential Arrangements

Trade preferences have long been an instrument of foreign and commercial policy, employed to establish closer relations, both economic and political, between the countries concerned. Preferential treatment by developed countries of imports from developing countries have, in addition, become a firm element of development-oriented policies. The various existing regimes of trade preferences for developing countries can be classified in three major categories, i.e. (i) the Generalized System of Preferences, (ii) special preferential regimes for sub-sets of developing countries, and (iii) regional free-trade arrangements between individual developed countries and given groups of developing countries.

(i) The origins of the Generalized System of Preferences (GSP) date back to the 1960s, when the need to improve trading conditions for developing countries was discussed in Geneva at UNCTAD I (1964). The general decision to establish a GSP was then taken in New Delhi at UNCTAD II (1968), and put in concrete form in an UNCTAD understanding in October 1970.1 The GSP was to provide tariff preferences for all developing countries, on a non-discriminatory and non-reciprocal basis. Preferential tariffs were foreseen mainly for manufactures and semi-manufactures, in order to overcome the dependence of developing countries on exports of raw materials with their unfavourable long-term price trends and pronounced short-term quantity and price fluctuations. At the time, the violation of the most-favoured nation (MFN) principle through the provision of tariff preferences for developing countries was inconsistent with fundamental GATT rules, but this problem was eventually healed in a general form when the “Enabling Clause” was adopted in the GATT in 1979 (see below, Section 6). The EU was the first grouping of developed countries to introduce its GSP in July 1971, but other developed countries followed soon.

Though tariff preferences provided under the GSP follow an approach that was agreed at the multilateral level and all developed countries provide preferences under this regime, the concrete form of preferential treatment provided is decided individually by each developed country granting preferences. Hence the nature of GSP preferences differs among the individual developed countries. Differences relate to all aspects of the preferential regimes, i.e. product coverage, margins of preference, rules of origin, specific preferences for the least-developed countries, criteria for graduation of developing countries (or some of their exports) out of the regime when they reach given levels of economic development or export performance, and additional clauses such as commitment of the beneficiary countries to certain labour rights or environmental standards.

The GSP regimes of the individual developed countries also differ with respect to the extent to which they cover agricultural products. All GSP regimes include some agricultural products, but the number and type of products included differ widely, as does the size of preference margins. For example, in 1992 the EU’s GSP applied to 168 beneficiary countries and included 530 agricultural products, while that of the USA covered 133 developing countries and 467 agricultural products, and Japan’s GSP applied to 151 beneficiary countries and 289 agricultural commodities (Yamazaki, 1996, p. 412). On aggregate for all agricultural products covered and across all beneficiary countries, the preference margin amounted to 14 per cent of the value of imports in the EU, six per cent in the USA, and 16 per cent in Japan (calculated from Yamazaki, 1996, p. 414).

(ii) In addition to GSP, some developed countries provide special and more favourable tariff preferences to constrained sub-sets of developing countries, usually linked to them through previous colonial ties or regional political relationships. The most prominent special regimes are those maintained by the EU under the Lomé Convention (now the Cotonou Agreement) for given countries in Africa, the Caribbean and the Pacific (ACP), and the US system in favour of Caribbean countries, the Caribbean Basin Initiative (CBI), recently extended to the Sub-Saharan African countries by the African Growth and Opportunity Act. Under these special regimes, preference margins are usually larger than under the respective GSP, and more products are covered. In some cases, preferences even cover some ‘sensitive’ agricultural products excluded from GSP treatment, such as sugar both in the EU regime for given ACP countries and in the CBI of the US. In some cases, preference margins for selected agricultural products under these special regimes are rather large. For example, the Lomé Convention’s sugar protocol for selected ACP countries guarantees these countries the same sugar price that EU producers receive. Very specific are also the preferences the EU grants to banana imports from the ACP countries, and these preferences, and the related restrictions to imports from non-ACP countries, have caused significant difficulties in the GATT/WTO, as became obvious in the various successive banana disputes.

The special preferential regimes for constrained sub-sets of developing countries have recently run into difficulties in the GATT/WTO because they are not in line with the “Enabling Clause” which covers only preferential treatment for all developing countries as provided under the GSP (see below, Section 6). In response to the resulting need for a restructuring of the special preferential regimes, the EU has decided, on 26 February 2001, to provide duty-free treatment of essentially all imports from the least-developed countries (LDC), under a scheme dubbed the “Everything but Arms” (EBA) Initiative.2 Under this arrangement, the EU will provide duty and quota free access to EU markets for nearly all goods to all 48 LDC on the list of the United Nations.3 Duty-free access will not be provided for arms (25 tariff lines). For three agricultural products (bananas, sugar and rice), duty-free access will be implemented in three respectively four (bananas) progressive steps within three/four years.

Inside the EU, strong resistance was voiced against zero-duty treatment of some agricultural products under this initiative. In particular, the EU’s sugar and rice lobbies argued vigorously against the EBA plan, fearing that the initiative would fatally undermine the sustainability of the EU’s highly protective market regimes for sugar and rice. The problem these lobbies see is that the LDC covered by the EBA initiative might export, in line with the respective rules of origin, their total domestic sugar and rice production to the EU, while importing their domestic consumption requirements from the world market. In addition, it was feared that the LDC might import raw rice, process it and then export it to the EU, adding sufficient value so as to meet the rules of origin requirements.4 The resulting large imports of sugar and rice, it is feared, might then force the EU to liberalise its market regimes for these products, not the least because the EU cannot export these quantities again under the existing WTO constraints on subsidised exports. It would be difficult to overcome such fears by tightening the rules of origin. After all, for homogeneous products like sugar it is not only hard to control the actual origin of a given shipment, but it is also nearly impossible to prevent the preferred country from exporting all its domestic production, while consuming imported produce.5 An alternative solution might have been to set quotas for preferential imports, and this was exactly what was considered in the EU in the context of the debate about the EBA initiative and ‘sensitive’ products such as sugar. However, in the end no quotas will limit the duty-free access under the EBA scheme, though safeguard measures can be brought to bear if “serious difficulties” should arise on EU markets.

(iii) Not the least because of the GATT problems caused by the special preferential regimes for sub-sets of developing countries, some developed countries have begun to convert some of these regimes into GATT-consistent regional free-trade areas with reciprocal trade liberalisation between the developed and developing countries concerned. An example of this type are the New Euro-Mediterranean Agreements that the EU has concluded with several countries in the Mediterranean basis, and which are still being negotiated with other countries in that region.6 Reciprocal regimes of a similar nature are supposed to replace in the future the EU’s Lomé preferences for those ACP countries that do not belong to the group of LDC (see below, Section 6). Though such regional free-trade regimes have historically originated from non-reciprocal trade preferences for developing countries, they do no longer belong to the category of tariff preferences for developing countries in a strict sense. These reciprocal regimes will, therefore, also not be covered any further in this study.

In summary, preferences for developing countries come in three major forms, i.e. the Generalized System of Preferences, special preferential regimes for sub-sets of developing countries (such as Lomé/Cotonou or the Caribbean Basin Initiative), and regional free-trade areas between developed and developing countries. However, the last form, involving reciprocal preferences, does not belong to the category of trade preferences for developing countries in a strict sense.

2 Conceptualising Trade Preferences As an Element of Economic Relations Between Developing and Developed Countries

Preferential treatment by developed countries of their imports from developing countries clearly is an important element of the overall web of economic (and political) relations between North and South. Trade preferences aim at fostering economic growth in the South, through making it possible for developing countries to engage in economically rewarding activities. The fundamental purpose of trade preferences, therefore, is to provide developing countries with better opportunities to achieve, by themselves, a self-sustained improvement of their economic, and hence hopefully social and political, fate. For some time in the past, trade preferences have clearly been seen as a possible alternative to financial and technical assistance, as reflected in the slogan “trade rather than aid”. This slogan, and the underlying economic concept, has a lot of intuitive appeal, and has probably helped to convince politicians in developed countries that they should agree to opening markets in their countries more widely to imports from developing countries.

It is simply a very persuasive argument that the South should be given better opportunities for improving its economic situation through its own activities, with a hopefully lasting positive effect, rather than being dependent on aid provided by the North, often without a perspective for achieving self-sustained economic growth. If combined with critical comments on developed countries’ trade policies, which arguably do no more than provide protection for outdated industries, the argument in favour of “trade rather than aid” is even stronger. It becomes nearly irresistible if the point is added that the world is anyhow moving in the direction of trade liberalisation, through both multilateral negotiations in the GATT/WTO and unilateral policy reforms in many parts of the world. Trade preferences for developing countries, then, appear to achieve merely an anticipation of the more general liberalisation of global trade that will (hopefully) come in the near future, in one sub-sector of international trade where the exporting countries concerned are in particular need of that early harvest.

Even though occasionally exaggerated in the political domain, the “trade rather than aid” perspective indeed makes some economic sense. However, it also has some drawbacks. Let us briefly consider some of the direct economic implications of trade preferences in order to better understand which role they can play in the overall framework of economic relations between developing and developed countries. Based on these economic implications we can also identify some important factors in the political economy of trade preferences.

At a fundamental level, and in a way taking the “trade rather than aid” perspective literally, trade preferences can be seen as a mechanism for making an economic transfer from the North to the South. Developed countries waive some part of their tariff revenue, providing for larger export revenues in the developing countries concerned. In other words, through trade preferences money is transferred from the budget of developed countries to the economies of developing countries, in addition to the expansion of quantities exported from the developing countries and the resulting domestic effects in these countries. From this perspective, trade preferences can in some way potentially be seen as a direct substitute for financial assistance that is channelled into development projects, be it in the area of infrastructure, in sectoral programs or anywhere else as part of bilateral or multilateral development assistance. Donor and recipient countries must then ask themselves whether trade preferences are a better form of support for development than financial assistance. This will depend on a number of considerations, including the size of the transfers that can be generated in either way, and which form of support promises to do better in terms of fostering self-sustained economic growth in the developing countries. This is not the place to dig deep into such considerations, but at least some of the more important factors can be discussed.

In this context, it is useful to take a brief look at the economic mechanics involved in trade preferences, among others to see what the implications are for market development in developing countries, and to understand precisely who gains and who loses from trade preferences. As we shall see, some results depend on the market situation and a number of other factors, and hence the economic implications of “trade rather than aid” are not unequivocal.

Let us start with a simple situation in which a given developed country (say, country A) begins to extend a trade preference to (all or some) developing countries (country group B) for one given product (say maize).7 Country A has in the past imported maize also from other developed countries, and continues to charge the (higher) MFN tariff on those imports. World market supply of maize from the other developed countries is available at a given price, irrespective of the quantity imported by country A.8 The quantity of maize shipped from developing country group B, though, is the larger the higher the price received on exports. Maize exported from country group B is of the same quality as maize exported from other developed countries, and receives the same price in the domestic market of country A.

Once country A grants a trade preference to country group B, some part of the maize originally imported from other developed countries is now sourced in country group B, without any effect on the domestic market price in country A.9 In this case, as suggested above, the most obvious effect of the trade preference, in addition to the expansion of exports, is that country A waives some of its tariff revenue (that it used to earn on maize imports now coming from country group B), while country group B now earns a higher price on exporting maize to country A. In other words, there is now a transfer from the budget of country A to the economies of country group B. The “trade rather than aid” element of preferential tariffs can, in this case, to some extent be described as country A providing financial assistance to country group B, where the amount of money transferred does not flow to some development project but to country group B producers of the product concerned (and possibly to some market intermediaries).

In terms of development strategy, both country A and country group B will have to consider the question of whether this form of support provided to country group B maize producers, and to the overall economies of country group B, is more useful for longer-term self-sustained economic growth in country group B than an equivalent amount of money that might have been transferred, as financial assistance, directly from the exchequer of country A to development projects improving, say, infrastructure in country group B. The answer to this question will tend to be the more positive the more likely it is that country group B would export the larger post-preference quantity of maize to country A anyhow under free trade conditions. After all, if the trade preference allows country group B to gear up its production structure early to what one day, when world trade has (hopefully) been liberalised completely, will anyhow be optimal, then it should really be in the interest of country group B to get the chance of receiving this trade preference, and thus to be able to provide its maize producers with market incentives which point in the right direction. The alternative, i.e. receiving financial assistance e.g. for infrastructure projects, may also have significant benefits, but may contribute less to self-sustained growth than allowing producers to develop their production capacities on a private basis, through ‘proper’ price incentives.

Alternatively, if it should turn out that the trade preference ‘sucks’ production structure in country group B into a direction which is not sustainable in the long run, then the situation is more problematic. This is the case if the trade preference leads producers in country group B to invest human and physical capital into a sector which promises attractive immediate returns as a result of preferential market opening in country A, which though has no long run prospects as a sector for profitable exports. This more problematic situation prevails if the volume of exports from country group B triggered by preferential treatment exceeds the quantity that country group B will be able to ship to country A once trade has been completely liberalised. Where this is the case, the trade preference leads production structure in the developing countries in a wrong direction as new adjustment costs will have to be born once exports go down again as liberalisation of MFN tariffs is achieved. In a situation like that, country group B might have been better off had country A transferred the same amount of money to them not through a trade preference, but in the form of direct financial assistance for the development of infrastructure or for other development projects with long lasting positive effects for economic growth.10 Alternatively, the government of the exporting country could skim off (some part of) the preference margin and use the revenue for development purposes not directly related to the preferential export activity. If this is done, then the danger of seeing the production structure being ‘sucked’ in a non-sustainable direction can be averted by the developing country concerned.

In the case discussed so far, there is not the slightest doubt that country group B exports of maize to country A after the introduction of the trade preference are larger than the amount of maize that country group B would ship to country A under free trade. The reason is that even after introduction of the preferential tariff the domestic market price in country A is still the given world market price plus the MFN tariff (as some imports still come in from other developed countries). In other words, on its maize exports to country A, country group B earns the going world market price plus the MFN tariff, minus any remaining preferential tariff. As that remaining preferential tariff is less than the MFN tariff, the net price earned by country group B is higher than the world market price. However, once country A eliminates its MFN maize tariff altogether, country group B earns no more than the going world market price. As that price is less than the price earned under the preferential tariff (and with the MFN tariff still in place), complete elimination of country A’s maize tariff will result in a reduction of country group B maize exports, and hence a new downward adjustment of maize production in country group B, after its maize output was originally expanded after the introduction of the preferential tariff.11

In this case, it is particularly clear that country group B would have been better off had it received the equivalent sum as a pure monetary transfer in the form of financial assistance. However, this may not be a realistic alternative. Trade policies, including trade preferences, on the one hand and development assistance on the other hand are decided by different policy makers, influenced by different lobbies and considering different criteria. Hence the rate of substitution among transfers through trade preferences and financial assistance may be far less than one. Giving up on one dollar worth of a trade preference that developing countries could have received does not at all mean that one dollar more will be made available in the form of financial assistance. The ‘development lobby’ in developed countries will argue for both trade preferences and financial assistance. However, its influence on trade policies is not directly comparable with its influence on development aid. This is the more the case as there is a typical asymmetry in the political assessment of revenue forgone and expenditure made. Revenue forgone is invisible, as it is not accounted for in the public budget. Expenditure, on the other hand, is very visible in the respective budget line. Hence, tariff revenue forgone because of a trade preference is likely to count far less in the political debate than expenditure actually made on development assistance. As a consequence, developing countries probably don’t have to fear that they lose (significant amounts of) financial assistance if they manage to obtain trade preferences.

On the other hand, developing countries have only a given amount of ‘negotiating capital’ that they can bring to bear on developed country governments, and they must consider where the last ‘dollar of negotiating capital’ achieves most. From this perspective it is of course relevant that attempts at achieving trade preferences potentially run against the interests of those producer groups in the developed countries that benefit from trade protection. In the case discussed so far, maize producers in country A (if they understand the situation correctly) have no reason to be concerned, as the only effect of the trade preference is to replace imports from other developed countries by imports from developing countries. The trade preference does not affect the domestic price in country A, and hence does not result in a loss to domestic maize producers in country A. Opposition to the trade preference could, then, come only from maize producers in other developed countries who lose market shares in country A. It can also come from other developing countries not benefiting from the same tariff preference. Opposition might, though, also come from economists who can well argue that in this case the world overall clearly loses welfare as low-cost imports from other developed countries are replaced by higher-cost imports from country group B.12

Starting from this simple case, other more complex cases can easily be derived, though not all constellations of possible assumptions need to be discussed here. If export supply from other developed countries is less than infinitely elastic, the trade preference will depress the domestic market price in country A. As a result, imports into country A will rise, making a loss in world welfare somewhat less likely. However, in addition to opposition from producers in other developed countries, there will also be opposition from maize producers in country A. This strengthens the relative advantage of using developing countries’ ‘negotiating capital’ for arguing in favour of financial assistance. On the other hand, it is less certain that the production structure in the developing countries is ‘sucked’ in a sub-optimal direction that may not be sustainable after complete trade liberalisation.13 Somewhat similar is the case in which the trade preference for country group B has the effect of squeezing all imports from other developed countries out of the market of country A. In this case, too, the domestic market price in country A will fall as the trade preference is introduced, and opposition will come from producers in both country A and in other developed countries.

In summary, the “trade rather than aid” perspective has some economic appeal. Trade preferences have the potential of helping developing countries to foster self-sustained economic development. They can substitute for, but probably also add to, economic transfers from developed to developing countries in the form of financial assistance. However, there are also drawbacks. The most obvious problem is resistance from the side of producers in developed countries. Less obvious, but still relevant is the problem of leading the production structure in developing countries in a direction which is not sustainable when overall trade liberalisation makes progress. Finally, there is the possibility of a loss in world welfare resulting from trade diversion.

3 The Benefits of Trade Preferences

Whatever one may conclude from the more conceptual discussion presented in the preceding section, in practical terms trade preferences are granted to developing countries because they demand better access to developed countries’ markets, and because developed countries are convinced by the view that this is a particularly useful way of providing poorer countries with better opportunities for economic growth. On this basis, in the developing countries the expected benefits of trade preferences are ‘hard’ economic advantages such as better access to developed country markets, larger volumes of exports, higher prices earned on exports, improved economic welfare, more jobs, and more rapid economic growth. In addition, there can be benefits to developing countries in the area of ‘soft’ (but nevertheless important) factors such as better familiarity with sophisticated markets in developed countries, growing awareness of the need to produce high qualities, a more outward-oriented attitude of the economy, new business alliances, etc. .

While all these benefits may exist in reality, unfortunately they are very hard to identify and measure empirically. The major reason is that it is very difficult to tell which concrete development in a given developing country was actually caused by trade preferences, and would not have occurred had the trade preference not been granted. The observed growth rate of exports to developed countries granting trade preferences is not a reliable indicator as it is not clear how exports would have developed in the absence of trade preferences. A somewhat better impression might be gained by comparing the growth of developing country exports to developed countries granting preferences with the growth of exports to developed countries not granting preferences, as done e.g. by Weston, Cable and Hewitt (1980). However, it is not necessarily clear that exports to preference granting countries would have, in the hypothetical case of an absence of preferences, developed in the same way as the exports that actually went to developed countries not granting preferences.

This fundamental problem of the lack of a with-and-without comparison is probably the major reason why only relatively few empirical studies appear to have looked into the benefits of trade preferences. It is only through the use of quantitative trade models based on assumptions regarding elasticities that actual trade flows under preferential treatment can be compared with the hypothetical trade structure that would have resulted had these trade preferences not been extended. Given the fine commodity detail of most preferential schemes, and the difficulties of estimating elasticities at that level of disaggregation, it is a major task to construct such quantitative models. However, there are a few studies that have used such models to estimate the effects of preferences for developing countries on the volume and value of trade flows, though in most cases these studies have looked at relatively broad categories of products. Baldwin (1984) reports on some such studies. The findings regarding the increase of developing country exports due to preferential treatment under various schemes analysed in these studies are in the order of magnitude of 25 per cent.

One other rough quantitative indicator of (potential) benefits derived from trade preferences, that can be estimated without making any assumptions on elasticities, is the size of preference margins. Conceptually this indicator is relatively straightforward, though there are several limitations to its economic interpretation, as will be shown below. Under certain (rarely realistic) conditions the preference margin is equal to the static welfare gain that the preferred exporting country can derive from a trade preference. Empirical calculation of the preference margin can be done without major difficulties, though it requires a lot of time-consuming data processing.14 Fundamentally, the preference margin per unit of product exported to a given importing country is the difference between the MFN tariff and the preferential tariff for that product, both tariffs being expressed as specific tariffs.15 The total value of the preference margin for a given product is that margin multiplied by the quantity exported to the importing country granting the preference.16 Of course, total values of preference margins can then also be aggregated across products. To normalise across countries and products, preference margins can be expressed as a percentage of the value of exports of the products concerned.

Yamazaki (1996) has estimated preference margins for agricultural exports from developing countries resulting from various preferential schemes operated by the EU, the USA and Japan. The study finds that the aggregate preference margin for all preference granting countries covered, all beneficiary countries and all agricultural products was 1,853 million US$ in 1992, equivalent to 12 per cent of the total value of preferential trade. 73 per cent of that margin was provided by the EU, while Japan and the USA shared the remainder about equally. As far as exporting regions are concerned, Africa received slightly more than one third of that aggregate margin, South America nearly one fourth, and each of the Far East, Central America and the Caribbean about one sixth. With regard to products, it turned out that 46 per cent of the preference margin provided by the EU came for sugar, though sugar accounted for no more than ten per cent of all exports to the EU covered in the analysis. The study also estimated the loss in preference margins resulting from the MFN tariff reductions agreed in the Uruguay Round, under the assumption of no improvement in the preferential schemes. It was found that the preference margin would decrease by 632 million US$, or one third. The Far East was expected to lose the largest share of its pre-Uruguay Round preference margin, i.e. 63 per cent, while Africa was found to lose 25 per cent. For several small countries it was found that their total preference margin was eliminated, as post-Uruguay Round MFN tariffs were reduced to the preferential tariffs existing in 1992.

The preference margin for agricultural exports from the ACP countries to the EU was estimated in an FAO study by Sharma (1997). For 1996, i.e. with MFN tariffs at the beginning of the post-Uruguay Round period, the aggregate preference margin for all ACP countries and all agricultural products was estimated at 710 million ECU (840 million US$), about 14 per cent of the value of trade covered. 52 per cent of that preference margin resulted from sugar, followed by 21 per cent from beef. The study expected the aggregate preference margin to fall by 16 per cent by the year 2000, due to the tariff reductions during the Uruguay Round implementation period. In this analysis, the preference margin for the two Protocol products sugar and bananas, accounting for 60 per cent of the aggregate in 1996, was assumed to remain unaffected by the Uruguay Round tariff reductions.

Estimates of preference margins for selected agricultural exports from the African ACP countries (AACP) to the EU were provided in a study for UNCTAD by Tangermann and Josling (1999). Some of the results of that study are reproduced here in Tables 1 to 3.

For the aggregate of the selected agricultural products covered in the study, the total preference margin for the AACP was estimated to be around 630 million ECU, based on 1997 trade data and 1999 tariffs. The share of the preference margin in the value of exports differed very much from product to product. It was highest in those cases where specified ACP countries had received specific preferences for given quantities of beef and sugar, laid down in Protocols attached to the Lomé Convention.17 In the case of Protocol beef it was estimated that the preference margin amounted to as much as 75 per cent of the export value. On the other hand, in the cereals sector the preference margin enjoyed by the AACP was estimated to be no more than 0.5 per cent.

Table 1: Preference Margins for Selected Groups of Agricultural Products Exported from the AACP to the EU Under Lomé Provisions,
1999 EU Tariffs (1997 Trade Data)

Source: Tangermann and Josling (1999, p. 50).

Table 2: Preference Margins for Beef and Sugar Exported from the AACP to the EU Under Lomé Provisions, 1999 EU Tariffs (1997 Trade Data)

Source: Tangermann and Josling (1999, p. 53).

Table 3: Preference Margins for Protocol Beef and Sugar Exported from Individual AACP to the EU Under Lomé Provisions, 1999 EU Tariffs (1997 Trade Data)

Source: Tangermann and Josling (1999, p. 54).

In that study, preference margins were also calculated for MFN tariffs that would result if it should be agreed in the next round of WTO negotiations on agriculture that all tariffs have to be brought down to 28 per cent of their pre-Uruguay Round base level.18 The results of that calculation are reproduced here in Table 4. On aggregate across the products covered in the study, the hypothetical tariff reduction looked at would reduce the preference margin by more than a half.

Table 4: Preference Margins for Selected Groups of Agricultural Products Exported from the AACP to the EU Under Lomé Provisions,
Hypothetical EU Tariffs After the Next WTO Round (1997 Trade Data)

Source: Tangermann and Josling (1999, p. 56).

Of course, such estimates of preference margins can also be repeated for alternative sets of (assumed) preferential tariffs. For example, in the Tangermann/Josling study it was analysed how the AACP would fare if the Lomé preferences were to be replaced by the EU’s GSP.19 The result is reproduced here in Table 5. It was found that for the least-developed countries among the AACP the aggregate preference margin under GSP would be only 7 per cent less than under the Lomé preferences, while the remaining AACP would lose around 85 per cent of the preference margin if they had to rely on GSP treatment rather than Lomé preferences. In a more recent update of that study, Tangermann (2000) also estimated the changes in preference margins brought about by moving from the Lomé Convention to the Cotonou Agreement in the year 2000, and preference margins that will result from the new set of (zero duty) preferences for the LDC established by the EU under the “Everything but Arms” (EBA) regime.

Table 5: Preference Margins for Selected Groups of Agricultural Products Exported from the AACP to the EU, Lomé Preferences and GSP Preferences Compared, 1999 EU Tariffs (1997 Trade Data)

Source: Tangermann and Josling (1999, p. 59).
Note: in this table, 0.0 denotes a positive value less than 0.1, while 0 denotes a zero value.

Such estimates of preference margins are relatively easily done, and they may also have some intuitive appeal. However, their results should be interpreted with much caution. In particular, the actual welfare effects of trade preferences for the recipient countries may deviate significantly from such mechanically calculated preference margins, and usually they will be far smaller. For the same reason the actual economic effects of MFN tariff reductions for preference-receiving countries may be considerably smaller than the erosion of preference margins estimated in this mechanical way. Some of the major factors causing differences between estimated preference margins and ‘true’ economic effects will be discussed below in Section 8, in the context of asking whether there is a case for compensation when preferences erode.

The actual welfare effects for the recipient countries also depend on how any benefits from trade preferences are distributed among the various groups of market participants. In particular, are the benefits likely to accrue to economic agents in the exporting developing country, or to agents in the importing developed country? The answer to this question depends partly on the competitive structure of the market concerned, and partly on how the trade preference is administered. The relationship between market structure and distribution of gains resulting from trade preferences is complex and cannot be discussed here in detail. However, intuition suggests that the more of the potential benefit resulting from a trade preference will tend to end up in the hands of the import trade in the importing country (i) the more concentrated that trade is, and (ii) the closer supply from the exporting developing countries is to a situation of full competition. In any case, with monopsonistic structures and behaviour in the import trade in the importing developed countries, some part of the preference margin will end up in the hands of agents in the importing country, rather than in the exporting countries.

A similar situation is often created by government policy if the preference is constrained by a given maximum quota of imports, i.e. by a tariff rate quota (TRQ). Where that quota is binding, a quota rent results, and in the case of trade preferences that quota rent can be as high as the preference margin. Where quotas bind, licenses to preferential trade must be issued in order to assure that imports at the preferential rate do not exceed the TRQ. Economic theory and practical experience show that in such cases the quota rent, i.e. in this case the preference margin, flows to the holder of the license. For most TRQs, and in particular where the preferential quota is not allocated to specified individual exporting countries, governments of the importing countries have a tendency to allocate licenses only to trading companies registered in their territory. In other words, where tariff preferences are constrained by TRQs that are not specifically allocated to individual exporting countries, there is a strong tendency that the preference margin ends up not in the exporting countries but in the importing country.20

As suggested above, any static economic welfare benefits estimated in quantitative studies, e.g. through assessing preference margins, can at best reflect one element of the various potential positive effect of trade preferences. Many other benefits, in particular the dynamic advantages and the ‘soft’ benefits referred to above largely escape quantification on the basis of aggregate data. It would be a matter of detailed case studies to assess the specific benefits that individual beneficiary countries have derived from tariff preferences. In the absence of such studies it is also difficult to say how the overall benefits of tariff preferences differ between different groups of developing countries, and to make any general statements on the question of which categories of developing countries ‘deserve’ tariff preferences most. However, a few hypotheses are intuitively appealing.

First, if trade preferences are seen as one type of economic assistance that rich countries can provide to poor countries, then it appears logical to suggest that preferential treatment should be the more important, and should be provided the more generously, the lower the level of economic development is in the potential beneficiary country. Political economy considerations would point in the same direction. For the preference-granting country, the political ‘cost’ of providing preferences results from the resistance of domestic producers of competing products, while the political ‘benefit’ stems from the preference that citizens have for helping people in poorer countries. The poorer the potential beneficiary country, the lower will the ‘cost’ element tend to be, because the threat of competition is probably felt the less the lower the level of development in the potential source of imports. At the same time the ‘benefit’ is high in the case of poorer potential recipients as the need for assistance is then felt particularly strongly. At the same time, the dynamic and ‘soft’ advantages of trade preferences are the more important for a country the less its level of economic development.

Second, it can be argued that trade preferences are the more important the larger the role is that exports can play in economic development of a given recipient country. In this context, the size of recipient countries can play a role. It is a well established fact that trade is the more important, as a percentage of GDP and as a source of welfare improvement, the smaller the country concerned. Trade theory suggests, and empirical observations confirm that large countries tend to trade less, in GDP percentage terms, than small countries because they exhibit a wider variety of products with different comparative advantages inside their own territories, and because they can make better use of economies to scale on their domestic market. In a way, large countries conduct ‘trade’ between the regions, and companies, in their own territories, while small countries need to trade with the rest of the world in order to exploit comparative advantages and economies of scale. On the basis of this argument, trade preferences are particularly important for small developing countries. Along the same lines, preferences are particularly important for ‘vulnerable’ developing countries such as island and land-locked countries, as export expansion is particularly important, but also often particularly difficult for them.

In summary, trade preferences can have various benefits for the exporting countries concerned. Empirical quantitative estimates of the overall size of these benefits are difficult, and hence rarely found. However, a relatively easily calculated indicator of potential benefits is the preference margin. Available estimates of preference margins show that they can amount to significant shares of the value of exports from the developing countries concerned. However, preference margins are a rather unreliable measure of economic benefits. Welfare gains for the exporting countries concerned are usually much smaller than the preference margin. Moreover, under certain conditions the preference margin flows to agents in the importing country, rather than to the exporting countries. In the absence of comprehensive analyses of preference benefits in individual beneficiary countries, the basis is relatively weak for judging which groups of developing countries ‘deserve’ preferences most. However, there are good intuitive grounds to argue that trade preferences are particularly important for the poorest countries and other vulnerable developing countries, such as small, island and land-locked countries.

4 The Costs of Trade Preferences

Like in most aspects of life, trade preferences have not only benefits, but also costs. While most of the benefits come as direct economic advantages, the costs tend to be more disguised, and in part come only in the longer term. However, they are nevertheless real and have to be weighed against the benefits. Moreover, the benefits of preferences may decline over time, while the costs may remain constant, or can even increase.

Industrial tariffs in developed countries have very much declined in the past. This has greatly reduced the importance of trade preferences as an instrument to overcome cost disadvantages of countries that are still at an early stage of industrial development. Moreover, much of the benefits of trade preferences was concentrated on a small number of strongly export-oriented developing countries. In addition, as preferences under the GSP are unilaterally determined by the developed countries, they can and do attach strings to them, such as the adoption of given labour standards, cooperation in drug control, and other policy conditions.21 Some observers argue that the relevance of tariff preferences is strongly declining. It has been noted that “trade preferences are fading away as part of multilateral trade liberalisation, but they remain as part of development folklore” (Robertson, 1999, p. 59). Quite apart from the economic effects of trade preferences, many commentators argue that insistence on non-reciprocity may not be in the long-run interest of developing countries as it tends to undermine their influence in the multilateral trade regime.

On the other hand, contrary to the situation in industrial products, in agriculture many tariffs are still extremely high in the developed countries, not the least on products of particular export interest to developing countries. There are still potential economic benefits that can be derived from preferential access to developed countries’ agricultural markets. However, the same protectionist sentiments that lead developed countries to erect the high trade barriers also get in the way of them granting generous preferences for agricultural imports from developing countries. As a result, agricultural products have not been important elements of preferential treatments under the GSP schemes of most industrialised countries. In particular, temperate zone agricultural products have been largely excluded from preferential treatment, or received preferential treatment only within tight quotas. For tropical products, on the other hand, developed countries’ MFN tariffs are zero or relatively low anyhow, and therefore preferences do not help very much.

In a situation like that, a strategic question for developing countries is whether the ‘negotiating capital’ they have is better used in WTO negotiations on further reductions of MFN tariffs in agriculture, or in attempts at deepening tariff preferences under GSP schemes, expanding their product coverage, and increasing TRQ volumes where preferences are constrained by quotas. Though the appropriate response to this question may differ from case to case, overall the MFN route should be considered carefully. GSP schemes will probably continue to be unilateral policies, extended on a ‘voluntary’ basis by the developed countries concerned, and differing among the importing countries. Negotiating better agricultural preferences in these schemes is bound to run directly against the agricultural lobbies in the individual developed countries, who can always argue that their respective country should not do what other developed countries fail to provide. In WTO negotiations on MFN tariff reductions, on the other hand, all countries are in the same boat, and developing country exporters can join forces with agricultural exporters from developed countries, as has happened in the Cairns Group. Tariff reductions are therefore more likely to be achieved in multilateral WTO negotiations, rather than in country-to-country negotiations on GSP schemes.

At the same time, the international trading regime for agricultural products is gradually moving towards a situation of lower tariffs. It will take a long time to reach relatively free trade in this area. But if the Millennium round should agree on another 36 percent reduction of agricultural tariffs (from their pre-UR level), then only 28 percent of the original tariffs will be left after the next implementation period, and one more round of WTO negotiations with the same result could suffice to eliminate many agricultural tariffs altogether. It may well be that the process of bringing down agricultural tariffs takes considerably longer. However, the direction of change is relatively certain, and it is a move towards lower MFN tariffs. In this process, agricultural preferences are bound to lose their value increasingly, as most industrial preferences have lost their value already. Hence, efforts to improve agricultural preferences should be seen as investing limited negotiating capital in a business that will not be very profitable in the long run.

Moreover, making use of any significant new agricultural preferences that could potentially be obtained in the short run requires adjustments in domestic production and market structures in the developing exporter countries concerned, and these adjustments may well turn out to have been in sub-optimal directions in the long run when preference margins evaporate because of MFN tariff reductions. This will be the case where the long run world market price is below the price that can be earned, through preferential exports, in the short run on highly protected markets of industrialised countries (see above, Section 3). From this perspective the more promising approach may well be to work towards larger MFN tariff reductions in the next round of WTO negotiations on agriculture. This is not at all to say that existing agricultural preferences under GSP schemes should be eliminated. However, major efforts to improve them may involve more costs than benefits.

At a completely different level, preferences cause costs resulting from their technical implementation. In particular, preferences go along with rules of origin, to make sure that only, or primarily, domestic production in the beneficiary countries benefits from the trade preferences. Typically, rules of origin are rather complex, and the texts specifying them are usually much longer than the texts laying out the trade preferences as such. In the beneficiary countries, all sorts of administrative efforts have to be made to comply with the rules of origin, and to demonstrate that this is the case. Depending on the size of the preference margin, the costs of implementing the rules of origin may actually be larger than the value of the preference. In the reality of trading activities, it therefore sometimes happens that preferences remain unused because meeting the rules of origin was simply too burdensome.

Preferences also involve costs that are borne by countries other than the beneficiaries. For the importing countries granting preferences, there are direct economic costs in the form of tariff revenue foregone. From an overall economic perspective, these fiscal costs may well be smaller than the benefits (to domestic consumers) resulting from larger volumes of imports and lower domestic prices. However, this is the case only if the preferences concerned result mainly in trade creation rather than trade diversion. On the other hand, where trade diversion results, costs are borne also by those other exporting countries that would have supplied the goods which now come from the beneficiary countries. Trade diversion may well be justified, on the grounds of positive effects for economic development in the beneficiary countries, if the losing countries are industrialised exporters. However, the losing countries may also be developing economies that have graduated out of the GSP, or those not belonging to the special group of developing countries benefiting from constrained preferential regimes. In these cases it is more difficult to disregard such costs in the form of trade diversion.

Finally, there are political economy costs of preferences in the sense that the countries receiving preferences become secondary beneficiaries of MFN protection in the importing countries. Beneficiary countries strongly interested in a given set of tariff preferences may lose interest in pushing for MFN tariff reductions, or may even actively oppose liberalisation of given policy regimes in the importing countries. For example, the ACP countries benefiting from the EU’s preferences for sugar and bananas have no interest in seeing the EU liberalise its sugar and banana regimes. If it comes to discussing reform of these regimes, the beneficiary countries are likely to request compensation, like EU farmers request compensation in such cases. To the extent that such compensation is considered in the form of cash payments, the budget expenditure required for that compensation may get into the way of policy reform. For example, in its October 2000 proposal for (extremely modest) modifications of the EU sugar regime, the European Commission used the argument of an expected income loss of 250 million Euro in the ACP countries concerned to explain why a price cut of (a hypothetical) 25 per cent for EU sugar was infeasible (European Commission, 2000b).

In summary, trade preferences provide not only benefits, but can also involve costs. Improving and expanding preferences requires ‘negotiating capital’. With successive rounds of MFN tariff reductions, the value of preferences is bound to decline, and it should be carefully assessed how much ‘negotiating capital’ should be invested in a business that may not be very profitable in the long run. Insistence on non-reciprocity of preferences can undermine the overall influence of developing countries in multilateral trade negotiations. Specific and deep preferences can potentially lead the production structure in the beneficiary countries in a direction that is not sustainable when MFN tariffs come down. Preferences tend to result in trade diversion, and the consequent costs for other exporting countries. Finally, countries benefiting from preferences may lose interest in MFN tariff reductions, and this is a cost to the multilateral trade regime.

5 The Status of Trade Preferences in the WTO

Trade preferences for given groups of countries run against one of the central pillars of the GATT, the principle of non-discrimination expressed in the MFN clause which requires (among others) importers to provide all suppliers the same treatment as the most-favoured nation among the suppliers, laid down in Article I of the GATT. Because of the fundamental importance of this principle for the multilateral trading system, exceptions from MFN treatment are closely circumscribed in the GATT, and require specific GATT legislation.

The original text of the General Agreement did not allow for preferences in favour of developing countries. The only exception to the MFN principle built into the GATT legal framework from the beginning was the provision for reciprocal free trade within customs unions and free-trade areas (GATT Article XXIV). This provision could not be applied to preferential imports from developing countries because no reciprocity was involved in these development-oriented trade preferences. Moreover, trade preferences for developing countries were not intended to cover “substantially all the trade” as required for free-trade areas in GATT Article XXIV. Since no other GATT provision could possibly provide shelter, trade preferences for developing countries were simply illegal under the original GATT. When Part IV of the GATT, on Trade and Development, was negotiated in 1964, a number of developing countries suggested that GATT Article I should be amended to allow for trade preferences for developing countries.22 However, no agreement on preferential trade provisions could be reached at the time, and Part IV, therefore, only drops the reciprocity requirement for developing countries when developed countries negotiate (non-preferential) concessions with them.

Thus, when developed countries began to provide preferential treatment to imports from developing countries in the 1960s, they needed to find a specific exemption from GATT rules for this purpose. At the time this was achieved through GATT waivers (under GATT Article XXV:5) for the individual developed countries providing preferences. The issue of trade preferences came up in a more general form again in the GATT when the GSP approach was adopted in an UNCTAD understanding in October 1970. Among the various options considered for solving the resulting legal problem in the GATT, the GATT Contracting Parties opted again for a waiver approach, in a decision adopted in May 1971. However, during the Tokyo Round negotiations, a Framework Group was established on request of developing countries, with the objective of finding a more permanent legal solution for trade preferences. As a result of the negotiations, an agreement was found, as one part of the Tokyo Round results, which became known as the ‘Enabling Clause’.23 This agreement did not amend the text of the GATT, but as a decision by the GATT Contracting Parties it had an essentially equivalent legal effect.

Under the Enabling Clause, “notwithstanding the provisions of Article I of the General Agreement, contracting parties may accord differential and more favourable treatment to developing countries, without according such treatment to other contracting parties”. More specifically, the clause allows for “preferential tariff treatment accorded by developed contracting parties to products originating in developing countries in accordance with the Generalized System of Preferences”; “differential and more favourable treatment ... concerning non-tariff measures”; “regional or global arrangements ... amongst less-developed contracting parties for the mutual reduction or elimination of tariffs ... [and] non-tariff measures, on products imported from one another”; and “special treatment of the least-developed among the developing countries in the context of any general or specific measures in favour of developing countries”. Under the Enabling Clause, “developed countries do not expect reciprocity for commitments made by them in trade negotiations”, and “shall exercise the utmost restraint in seeking any concessions or contributions for commitments made by them to reduce or remove tariffs and other barriers to the trade of [least-developed countries], and the least-developed countries shall not be expected to make concessions or contributions that are inconsistent with the recognition of their particular situation and problems”.

The Enabling Clause has, thus, created a permanent legal basis for trade preferences provided by developed countries, both generally for all developing countries under GSP regimes, and at the same time for specific more preferential treatment of the least-developed countries. On the other hand, there is no absolute legal requirement set in the GATT for providing any given trade preferences. In other words, developed countries can and should provide trade preferences for developing countries, but they are not legally forced to do so. As a result, trade preferences under the GSP continue to be unilateral concessions provided by the developed countries concerned, which can always be changed, and even withdrawn completely, without violating GATT/WTO commitments. While MFN tariffs, where bound in the Schedules of the importing countries concerned, are legal commitments that cannot be changed without following the respective GATT rules, trade preferences for developing countries are not bound in the GATT/WTO and hence can, as far as GATT/WTO rules are concerned, be changed unilaterally by the developed countries granting them.

The Enabling Clause relates only to preferences provided by developed countries, and to mutual, i.e. reciprocal, trade preferences among developing countries. It does, though, not establish a legal basis for the provision of trade preferences by developing countries for imports from the least-developed countries. For such trade preferences, a specific waiver is required. A waiver to that effect was agreed by the WTO Members on 15 June 1999.24 Under this decision, “the provisions of paragraph 1 of Article I of the GATT 1994 shall be waived until 30 June 2009, to the extent necessary to allow developing country Members to provide preferential tariff treatment to products of least-developed countries, designated as such by the United Nations, without being required to extend the same tariff rates to like products of any other Member, ... on a generalized, non-reciprocal and non-discriminatory basis.”

Under the Enabling Clause, tariff preferences granted by developed countries must not discriminate among developing countries, except for the possibility of providing more generous preferences to all least-developed countries. Specific preferences for constrained sub-sets of developing countries granted by individual developed countries, such as those granted by the EU to ACP countries under the Lomé Convention, are not covered by the Enabling Clause. For a long time nobody in the GATT protested against these specific preferential schemes, and it was not completely clear whether they were (implicitly) considered to be legal. It was not before the first GATT panel on the EU import regime for bananas in 1993 that an explicit decision was taken on the issue. The panel found the EU’s preferences for banana imports from the ACP countries inconsistent with GATT Article I, also not in line with GATT rules on free-trade areas, and that Part IV of the GATT did not provide a justification for non-reciprocity in free-trade areas involving developing countries. The Enabling clause did not provide a shelter for these ACP preferences as they were not extended to all developing countries. By implication that meant that all specific EU preferences for the ACP countries, and all other similar schemes of other developed countries were to be considered illegal under the GATT. In response to this panel outcome, rather than eliminating its trade preferences under the Lomé Convention, the EU requested a GATT waiver that would allow it to continue to provide the special trade preferences to the ACP countries. That waiver was indeed granted in 1994, for a period lasting until the expiration of Lomé IV, i.e. 29 February 2000.25

Given the GATT inconsistency of special preferences for sub-sets of developing countries, the EU looked for alternative approaches in the negotiations with the ACP countries on a successor agreement to the Lomé Convention. In the end, these attempts failed in the sense that a completely new and WTO compatible framework, substituting for the Lomé trade arrangements already in 2000, could not yet be agreed. However, a new agreement between the EU and the ACP countries was signed in Cotonou, Benin, in June 2000. Like Lomé IV, the new Cotonou Agreement contains, among others, trade preferences for the ACP countries. The trade arrangements under the Cotonou Agreement are of a temporary nature, but outline the path towards a new future trading regime between the ACP countries and the EU which is hoped to establish conditions which will eventually be WTO compatible. The new arrangements will come in essentially two forms.

First, new economic partnership agreements (EPA) will be negotiated which aim at a progressive removal of trade barriers between the EU and the ACP concerned. It is intended that the EPA will establish reciprocal free trade between the EU and the ACP concerned, in line with WTO rules on regional free trade arrangements. The intention is to establish the new EPA with all ACP countries which are not least-developed countries. Hence the second element of the trading arrangements foreseen in the Cotonou Agreement relates to the least-developed countries. In this regard, the EU committed itself to “start by the year 2000, a process which by the end of the multilateral trade negotiations and at the latest 2005 will allow duty-free access for essentially all products from all least-developed countries building on the level of the existing trade provisions of the Fourth ACP-EC Convention”.26 In other words, this second part of the future trade arrangements seeks to avoid the problem of WTO inconsistency by extending unilateral EU trade preferences to all least-developed countries, including those that are not members of the ACP group (at this time there are nine least-developed countries on the UN list that are not ACP countries). The European Council has meanwhile adopted a new regime that will provide zero-duty access for all products except arms from the least-developed countries, under a regime dubbed “Everything but Arms” (EBA) (see above, Section 2).

During the period in which these new arrangements are negotiated, scheduled to last until 2008 at the latest, the ACP countries and the EU will need another WTO waiver for the arrangements established under the Cotonou Agreement. The EU, and Tanzania and Jamaica on behalf of the ACP countries, have approached the WTO about a new waiver on 29 February 2000.27 However, at the time of writing a decision on that waiver has not yet been taken by the WTO.28

Equivalent to the EU’s problem with trade preferences for the ACP countries is the situation regarding the special trade preferences granted by the USA under the Caribbean Basin Economic Recovery Act. Like the EU, the USA received a WTO waiver in 1995, valid until 31 December 2005.29 It appears that there is wide agreement in the WTO that such waivers for specific preferential schemes should not be extended for ever. The developing and developed countries engaged in such schemes will therefore have to work towards new solutions, possibly in the form currently considered between the EU and the ACP countries.

In summary, universal trade preferences for imports from all developing countries, as extended under the GSP, are consistent with the GATT under the Enabling Clause. The same holds for specifically generous preferences granted to all least-developed countries. However, developed countries are not legally committed to providing such preferences. They can, therefore, decide unilaterally on preference margins, and also withdraw preferences without violating GATT/WTO commitments. Specific trade preferences for constrained sub-sets of developing countries, such as those provided under the Lomé Convention or under the Caribbean Basin Economic Recovery Act, however, are not consistent with the GATT. Yet, the WTO has in the past granted waivers that allowed the countries concerned to maintain these specific preferences.

6 Alternative Options for the Future of Trade Preferences in the WTO

Given that the next round of WTO negotiations is supposed to be a “development round”, the interests of developing countries should be given priority in all areas of the negotiations, including the setting of tariffs. Irrespective of the drawbacks of preferential treatment, discussed above, developing countries are likely to continue to emphasise the need for better preferential access to developed country markets, including those for agricultural products. Thus the issue of trade preferences will definitely be on the negotiating table. However, rather than requesting, in a wholesale fashion, the largest possible trade preferences for all developing countries, in all developed countries, for all products, more differentiated negotiating approaches should also be examined. Some of them will be discussed in this section.

At the most fundamental level, a strategic question that should be considered first is whether the negotiating aim should be to achieve what could be called ‘shallow’ preferences for all developing countries, or whether ‘deep’ preferences for the least-developed and vulnerable countries are a more promising aim. In this distinction, ‘shallow’ preferences stand for limited preference margins for selected products, sometimes constrained by TRQs, while ‘deep’ preferences would, in the extreme case, amount to zero preferential tariffs for all products. Why would a choice probably have to be made between these two options, rather than requesting ‘deep’ preferences for all developing countries? The hypothesis behind this suggestion is that developed countries may be willing to ‘give’ only a somehow limited amount of preferences, in the form of either ‘shallow’ preferences for all developing countries, or ‘deep’ preferences for a smaller group of developing countries. At the same time, developing countries have only a finite amount of ‘negotiating capital’ in the negotiations, and may therefore have to make a choice as to where to invest their negotiating efforts.

Several factors will have to be considered in making this choice. One of the most important aspects may be that ‘shallow’ preferences are likely to lose their value relatively soon as MFN tariffs are reduced further, while ‘deep’ preferences may last longer, in the sense of not being eroded so quickly. Another factor is the issue of which developing countries ‘deserve’ preferences most. As suggested above, a good point can be made that the least-developed and vulnerable countries indeed are most in need of support for their efforts to expand exports. This view is probably also prevalent among most developed countries. Hence, it may indeed be easier in the negotiations to achieve ‘deep’ preferences for the least-developed and vulnerable countries, than to obtain marginal improvements of ‘shallow’ preferences for all developing countries. An indication of the willingness of (some) developed countries to provide ‘deep’ preferences for the least-developed countries is the recent “Everything but Arms” initiative of the EU Commission (see above, Section 2). If indeed accepted by the EU Council of Ministers, these ‘deep’ preferences would not even have to be negotiated in the WTO, they would be extended unilaterally by the EU.

One option that should be considered in this context is to amend in the WTO the definition of “least-developed” countries in the Enabling Clause, to also include vulnerable developing countries such as small, island and land-locked countries in the category of developing countries to which preferences deeper than GSP preferences can be granted. As argued above (Section 4), preferences are particularly important not only for poor, but also for small and other vulnerable countries.

Of course, if such ‘deep’ preferences were to be granted, and finally not only by the EU but also by other developed countries, this would and should not do away with the existing preferences extended under GSP regimes. Is there anything that should be done about these GSP preferences in the coming negotiations? Several options come to mind. One would be to bind these preferences in the WTO. For the time being, GSP preferences are extended unilaterally by the developed countries concerned, based on their national legislation. In legal terms, developing countries have no ‘rights’ on them. This could be changed in the future, by binding these preferences in the WTO.30 At the same time, conditionalities attached to trade preferences (such as labour rights or environmental standards) should be removed where they are not in line with overall standards agreed multilaterally in the WTO.

Another step that could be suggested is to move from preferential tariffs set in absolute terms (as given ad valorem or specific tariffs) to preferences defined in terms of preference margins. This would mean that the preferential tariffs are defined relative to MFN tariffs, being given monetary units below MFN tariffs (where MFN tariffs are specific tariffs) or given percentages of the MFN tariffs (where MFN tariffs are ad valorem tariffs). Defining tariff preferences in this way would tend to guard against preference erosion when MFN tariffs are reduced in the future.31 Of course, ideally these preference margins would then be bound in the WTO.

An option certainly worth pursuing is to increase TRQ volumes under GSP regimes. Of course, this is most useful where the preferential quotas are currently fully utilised. However, even quotas that are not fully used so far can prove to be constraining in the future, and it may therefore be worthwhile expanding them as well. Another area where improvements can be made is the definition of rules of origin. In many cases, these rules are too complex and get in the way of a sensible utilisation of tariff preferences.

Beyond such general approaches, some more product-specific improvements of preferences could also be suggested, and may have some chance of being considered in the negotiations. For example, in agriculture the phenomenon of tariff peaks is still prevalent. Attempts will certainly be made, and have already been suggested by some countries in their initial negotiating proposals, to adopt some formula for tariff reductions which reduce high tariffs by larger percentages than low tariffs. The so-called Swiss formula, used for industrial tariff reductions in the Tokyo Round, is one formula which would achieve this. However, it is far from certain that an approach like that will actually be adopted in the current round of negotiations. Even if that should be the case, significant tariff peaks are still likely to remain. In this context, developing countries could suggest that for products with tariff peaks there should at least be preferential tariff reductions of significant magnitude, such that at least developing countries can gain better access to the markets concerned.

Along similar lines, tariff escalation still is a problem in some developed countries’ tariff schedules for agricultural products and processed foods. Tariff escalation is a serious barrier to developing countries’ efforts to diversify their export structures and in particular to expand their exports of high-value processed products. Again, efforts will anyhow be made in the current negotiations to do something about tariff escalation. However, in this area, too, chances of success are at least uncertain. In this situation, there may be a point for developing countries to suggest that at least they should be allowed better access to developed countries’ markets for processed products. Hence specific new trade preferences should be considered where tariff escalation still prevails even in the schedules of preferential tariffs.

In summary, a number of options can be considered regarding the future of trade preferences in the WTO. Rather than working towards an expansion of ‘shallow’ preferences for all developed countries under GSP regimes, an attractive alternative might be to aim at ‘deep’ preferences for the least-developed and vulnerable countries. In this context, the Enabling Clause could be amended by including small and other vulnerable countries, in addition to the least-developed countries, in the category of developing countries that can receive preferences deeper than GSP preferences. The functioning of the existing GSP preferences, which should certainly be maintained, can be improved by binding them in the WTO; removing conditionalities; setting preferential tariffs relative to MFN tariffs (rather than defining them as absolute tariffs); expanding TRQ volumes; simplifying rules of origin; and by providing better preferences where MFN tariffs exhibit peaks and tariff escalation.

7 When Preferences Get Lost: Is there a Case for Compensation?

One central issue that has plagued the policy debate about development-oriented trade preferences since they were first extended is the erosion of preference margins when MFN tariffs are reduced. It has often been said that a general tariff reduction, as agreed for example in a multilateral round of trade negotiations, is a two-edged sword for developing countries as it results not only in benefits, arising out of improved access of their exports to world markets, but at the same time also in costs, in the form of an erosion of the preference margins enjoyed by them on exports to developed countries’ markets. In some cases, it has been suggested, these costs may actually be larger than the benefits derived from general trade liberalisation, such that some developing countries may actually lose overall from multilateral rounds of trade negotiations. Under such conditions it could potentially be discussed whether some form of compensation should be provided to the losers.

Along similar lines, if a developed country unilaterally reforms its policy regime for a product which is of export interest to developing countries and where trade preferences had been extended to them in the past, that policy reform may result in a reduction, if not elimination, of the preference margin previously enjoyed in exporting that product to the developed country concerned. A possible future reform of the EU’s market regime for sugar is a particularly relevant case in point. Those ACP countries (and India) that currently benefit from preferential access to the EU’s sugar market would lose significant economic benefits if the EU were to cut the level of its support price for sugar. Similar losses could one day occur to developing countries currently benefiting from preferential access to the U.S. sugar market, and to ACP countries exporting bananas to the EU. In all of these cases, the issue of compensation will immediately arise.

The basic economics of preference erosion appear to be relatively simple. Clearly, when the MFN tariff on a given product is eliminated altogether there is no longer any scope for preferential treatment, and hence the economic benefit that may have resulted from the trade preference in the past is eliminated with the MFN tariff. With an MFN tariff of zero the preference margin is necessarily zero too. Similarly, when MFN tariffs are reduced (but not eliminated) then there is a tendency for preference margins to decline, and hence for economic benefits derived from preferential treatment to be eroded.

However, these statements cannot simply be translated into the statement that all reductions (or elimination) of MFN tariffs unequivocally result in an economic loss for those countries that used to benefit from trade preferences in the past. Hence, before discussing the matter of compensation for the erosion of preference margins we need to deal at least briefly with the question of how the cases can be identified in which a reduction of MFN tariffs does not result in an economic loss for the recipients of trade preferences, or even improves their economic situation. Two categories of such cases should be distinguished, each with several different cases. The first category comprises those cases in which preferences are set in a way that guards against erosion, or make erosion a non-issue. The second category is that of cases where market effects, i.e. price formation and the response of trade flows, are such that the overall net outcome for preference recipients is positive, even though preference margins as such may be eroded. Unfortunately it is not always exactly easy to measure all these cases precisely in quantitative empirical research, and within the scope of this study it is not possible to go beyond a more general discussion of the qualitative nature of these cases.

7.1 Non-Erosion of Preferences Resulting from the Setting of Preferences

To start, it should be remembered that it is not necessarily the case that a reduction in MFN tariffs, as possibly resulting from a multilateral round of tariff cuts, always leads to an erosion of the preference margin for all products. The behaviour of the preference margin, measured as MFN tariff minus preferential tariff, both tariffs being expressed as specific tariffs (or the specific tariff equivalents of the actual ad valorem tariffs), clearly depends on the way preferences are defined. In some cases, preferential tariffs are expressed as the MFN tariff applicable minus a given margin. In such cases, and where the MFN tariff is a specific tariff and the reduction margin is also expressed in specific tariff form (i.e. the preferential tariff is equal to the MFN tariff minus a given number of monetary units per quantity), the absolute size of the preference margin is not affected by a reduction in the MFN tariff. This is at least the case where the reduced MFN tariff is still higher than the reduction margin.

More frequent are cases where the preferential tariff is defined as a given percentage reduction of the MFN tariff. In these cases, the absolute size of the preference margin declines by the same percentage as the MFN tariff, though of course the share of the preference margin in the MFN tariff remains constant. In these relatively frequent cases, erosion of the preference margin is a relevant phenomenon, but its incidence is much smaller than where the preferential tariff is set as a given rate irrespective of the applicable MFN tariff, or where the preference comes in the form of a zero preferential tariff.

On the other hand, there have been cases where after a general round of multilateral tariff reductions developed importing countries have established new lists of (lower than before) preferential tariffs, in an attempt to maintain the size of preference margins and thus to avoid the phenomenon of preference erosion. This is, of course, not possible under the most generous form of a trade preference, i.e. a zero preferential tariff. In other words, where preferences were most generous in the past, the preference margin is also most vulnerable to erosion, because any reduction of the MFN tariff has the inescapable implication that the preference margin declines.

However, in some cases even zero preferential tariffs may not be vulnerable to preference erosion, simply because in practice no effective preference existed in the first place. Indeed, there are cases in which importing developed countries have established a trade “preference” in the form of a percentage reduction of the MFN tariff or a zero preferential tariff, but at the same time apply a zero MFN tariff. Where such “empty” preferences exist, the preference margin is of course effectively zero, and hence cannot erode with general tariff reductions. Such cases occur, for example, in the EU’s preferences for agricultural imports from the ACP countries. One eighth of the value of all exports from African ACP countries to the EU of agricultural products covered by the list of Lomé IV preferences consists of products with such “empty” preferences (Tangermann and Josling, 1999, p. 46).

Finally, there are cases where the actual monetary value of a trade preference, and hence the economic benefit to the exporting country, is less than the apparent preference margin (MFN tariff minus preferential tariff). In such cases, a reduction of the MFN tariff, and the resulting apparent erosion of the preference margin, does not lead to a corresponding loss of benefits for the exporting countries. The most obvious case of this nature is that of a prohibitive MFN tariff containing “water”, in the sense of being higher than necessary to result in zero imports. In such cases, some part or all of the MFN tariff reduction does no more than reduce the water in the tariff. To the extent that this happens, the economically valuable part of the preference margin (which of course does not include the water in the tariff) is not affected by the reduction in the MFN tariff. In agriculture, prohibitive tariffs are a relatively frequent phenomenon, as indicated by the large number of tariff lines where imports at MFN tariffs are (practically) zero. For that reason it is probably fair to say that a significant part of MFN tariff reductions, as possibly agreed in the next round of WTO negotiations on agriculture, will not result in an erosion of (economically valuable) preference margins.

Unfortunately, a quantitative assessment of the extent to which a given set of MFN tariff reductions does no more than squeeze water out of prohibitive tariffs, and hence an estimate of the extent to which MFN tariff reductions do not result in any erosion of economically valuable preferences, is extremely difficult to provide as it would require an amount of price information that is usually not available in published statistics.

At the same time (and often in the same cases), even preferential tariffs have not helped developing countries to export significant quantities to the markets of industrialised countries. In these cases, too, the effective erosion of preference margins has only limited implications in practice. However, conventional estimates of preference margins, and of their erosion through cuts in MFN tariffs, would pick these cases up automatically as the total value of the preference margins calculated (per unit preference margin multiplied by the quantity traded) would be small anyhow in the first place, because the quantity traded was small or even zero in the first place. As a result, in these cases any given percentage reduction in the total value of the preference margin is also small by implication.

7.2 Market Effects as Balance for Preference Erosion

Trade liberalisation in importing countries, through tariff reduction, is generally a good thing for exporting countries, as it improves the chances of exporting more and earning higher prices. This is, of course, also true for developing country exporters. It should, therefore, not be surprising to find that reductions of MFN tariffs in developed countries may well improve the overall economic situation of developing countries, even though there may be partial losses in the form of preference erosion. In other words, beneficial market effects of trade liberalisation may in some cases well be larger than the isolated implications of preference erosions.

The most obvious chance of overall gains for developing countries is, of course, that where reductions of MFN tariffs occur for products on which developing country exporters had not received any preferential treatment in the past, though they did already export the products concerned to developed countries, or where they can at least begin to export these products once MFN tariffs are reduced. Depending on the particular composition of their exports, and on the product structure of trade preferences, the gains on products without preferences may well be larger than the losses on products subject to preference erosion.

Somewhat similar is the case where trade preferences are constrained to given quantities of exports only, i.e. where tariff rate quotas (TRQs) apply. In such cases it may well happen that the developing countries concerned also export beyond the TRQs, at MFN tariffs. Indeed, where preferences are constrained by TRQs, cases of exports at MFN tariffs have been observed even where the TRQs were not fully utilised. Several reasons can be behind such cases, but most frequently it is the administrative hassle of meeting the rules of origin which explains this outcome. In any case, where exports from developing to developed countries occur simultaneously at both preferential and MFN rates, a reduction of MFN tariffs benefits those exports which occurred under these non-preferential conditions, and this may well outweigh the erosion of preference margins on exports under preferential conditions. Whether this happens, of course depends on the relative size of the within-TRQ and above-TRQ exports, and on the magnitude of the reduction in the MFN tariff and the pre-liberalisation size and definition of the preference margin. Moreover, in multilateral negotiations on tariff cuts, it is well conceivable that exports to countries that had not granted preferences benefit more from tariff reductions than exports to preference granting countries suffer from preference erosion.

Somewhat more analytically complex are cases where overall gains due to market effects of MFN tariff reductions may occur even for those exports that take place under preferential conditions only. The easiest case of this nature is where the preference margin is defined in specific tariff form (the preferential tariff is defined as the MFN tariff minus a given number of monetary units per quantity of import). In this case the preference margin in absolute terms is not affected by a reduction in the MFN tariff (as long as the post-liberalisation MFN tariff is still larger than the preference margin). On the other hand, due to the parallel reduction of both the MFN tariff and the preferential tariff the quantity of imports rises, and the price received on exports goes up as well.32 In other words, under these conditions MFN tariff reductions not only occur without a loss in the preference margin per unit of export. Instead, there is actually a gain to preferred developing country exporters, because of larger export volumes and higher prices received.

However, even where preferences are defined as percentage reductions of the MFN tariffs, there can be overall gains. In these cases, the absolute preference margin per unit of export is eroded as a consequence of the reduction in the MFN tariff. However, due to market effects resulting from trade liberalisation, it may under certain conditions still be the case that the price received on exports rises and the export quantity expands. These cases of positive overall effects are less likely than those that can be expected where preferences are defined in specific tariff reduction form. Yet, they are still possible.33 Where this is the case, the overall economic effect of the MFN tariff reduction can still be positive for the preferred exporters, even though the preference margin is eroded.

In summary, it is not at all the case that MFN tariff reductions always result in economic losses for exporting developing countries that used to benefit from trade preferences in the past. It is a matter of empirical quantitative analysis whether erosion of preferences really is a problem in the sense of resulting in overall losses for the exporting developing countries concerned. Unfortunately, however, that quantitative analysis is not always exactly easy. Conventional quantitative analysis of preference margins is done by calculating the preference margin per unit of export (MFN tariff minus preferential tariff, both expressed as specific tariffs), and then multiplying the per unit preference margin by the quantity exported. If this calculation is done first for pre-liberalisation MFN tariffs and then for post-liberalisation MFN tariffs, both at a given quantity of exports (usually taken from a historical reference period, i.e. before MFN tariff reduction), one gets a first impression of the magnitude of preference erosion. However, as indicated here, an estimate of this kind may well get things wrong, for several reasons. In particular, an estimate done in this way will not pick up cases where the MFN tariff reduction does no more than reduce the water in the tariff. Moreover, such estimates do not take into account the market effects which may result in rising export prices and expanding export quantities. A proper estimate of the overall economic effects would require a full market model with responsive prices and quantities, for all tariff lines, and ideally including all cross-product linkages. Such models are rarely available in sufficient detail. Hence an appropriate quantitative estimate of the full effects of MFN tariff reductions for developing countries receiving trade preferences is a rather elusive task.

7.3 The Issue of Compensation for Preference Erosion

Even though it is difficult to measure the overall effects empirically, there is no doubt that it is perfectly possible that multilateral or unilateral trade liberalisation can result in an erosion of tariff preferences that had economic value for the developing country exporters concerned, and that the effect of this preference erosion can be larger than any economic benefits resulting from market effects. Hence, in the following let us assume that the finding has been established unequivocally that a given group of developing countries faces an overall economic loss from a given set of MFN tariff reductions. Where this is the case, the issue of compensation can be discussed. Four questions are particularly relevant in this context. First, can a case for compensation be made at all? Second, if compensation is considered, who should ‘pay’ for that compensation, and how are the recipients determined? Third, which instruments could be used for compensation? Fourth, how should the magnitude of compensation be determined?

(1) Whether there is a case for compensation is a complex economic and even more political question. Several arguments can be advanced for and against compensation. The most obvious argument in favour of compensation is that there is an economic loss for the developing countries concerned. Trade liberalisation is generally thought to generate economic benefits, and these economic benefits are the major reason why governments go through the political trouble of liberalising trade at all. However, if some countries lose from trade liberalisation, then they should be compensated, because such losses go against the fundamental reason for trade liberalisation. Also, the overall economic gains resulting from trade liberalisation make it possible to compensate the losers, and still leave the gainers better off. This is the more compelling as the developing countries losing from preference erosion are the poorer countries of this world. It would run against global equity to accept a situation where, as a result of all countries jointly agreeing on multilateral trade liberalisation, the rich countries get richer, while the poor countries get poorer.

Moreover, as suggested above, trade preferences for developing countries can be interpreted as a substitute for financial and technical assistance. A simple straightforward cut of financial and technical assistance would not be easily accepted. Why then should a reduction of benefits effected in a different, but comparable way, be acceptable?

A further argument in favour of compensation is that there are precedents. In the Marrakesh Ministerial “Decision on Measures Concerning the Possible Negative Effects of the Reform Programme on Least-Developed and Net Food-Importing Developing Countries”, it was acknowledged, very much in line with the argument presented above, “that the progressive implementation of the results of the Uruguay Round as a whole will generate increasing opportunities for trade expansion and economic growth to the benefit of all participants”, while “during the reform programme leading to greater liberalisation of trade in agriculture least-developed and net food-importing developing countries may experience negative effects”. On these grounds, certain measures in favour of the developing countries concerned were agreed. The negative effects considered in this Ministerial Decision did not explicitly (and probably not even implicitly) include the erosion of tariff preferences. They obviously concentrated on the import, rather than the export side of the developing countries concerned. Moreover, very little in the form of tangible action has so far been implemented as a result of the Decision. However, it can be argued that the Decision has established the principle that overall trade liberalisation can result in losses for certain groups of poor countries, and that there is a reason to do something about these losses.

More specific arguments may apply to preferences for limited sets of developing countries and individual products. For example, in the specific case of the EU’s sugar preferences for ACP countries (and India), which might significantly lose value once the EU liberalises its sugar regime unilaterally, the point could be made that domestically in the EU the principle is now firmly established that farmers losing from cuts in support prices receive compensation in the form of direct payments. Sugar producers in the ACP countries concerned, it could be argued, are much poorer than sugar beet growers in the EU, and hence need compensation at least as much as they. Moreover, while sugar beet growers in the EU can relatively easily switch to other products such as cereals or oilseeds, the ACP countries concerned will find it much more difficult to develop new economic activities providing for jobs and income.

However, critics will also raise a number of arguments which speak against compensation. Trade liberalisation, it can be said, is an overall positive process which in the long run improves economic opportunities for all countries. More specifically, preference receiving developing countries may have been secondary beneficiaries of protection in the developed countries, but they must not take this position to the point where they begin to cause, through compensation claims, difficulties for the process of trade liberalisation. Since those losing from tariff cuts in developed countries usually don’t get compensated, why should producers in other countries have a claim on compensation? Moreover, preferential tariff reductions can be interpreted, as suggested above, as an anticipation of trade liberalisation to come in the future, for a specific group of countries which should be allowed an earlier harvest because of their specific needs. From this perspective it can be argued that the beneficiaries should not benefit twice, once when trade liberalisation was brought forward specifically for them, and the second time through compensation when the rest of trade is liberalised. Perhaps more to the point, it was always clear that one day the process of general trade liberalisation would go on, thereby squeezing preference margins. As this perspective was always an element of the overall framework of economic relations between developing and developed countries, why should there now be claims for compensation once the process of trade liberalisation actually materialises?

(2) Regarding the question of who should ‘pay’ if compensation is considered necessary and justified, several different suggestions could be made. If the erosion of preference margins is a result of the process of multilateral trade liberalisation, as agreed in the WTO, it could be argued that the community of developed countries on aggregate should ‘pay’. After all, decisions on tariff cuts were taken jointly by all WTO Members. Hence developing countries suffering from preference erosion should receive compensation from the developed countries as a group, through some multilateral agency (possibly to be created for this purpose). The contributions to be made by the individual developed countries would be determined on the basis of some general indicator, such as the magnitude of trade concessions made in the first place, or on the basis of GDP in the individual developed countries.

A different suggestion would be that each importing developed country ‘pays’ individually. GSP preferences were set individually by the developed countries, with marked differences from country to country. Moreover, tariff cuts of the developed countries will differ from country to country, even with a generally agreed formula, as base levels of tariffs diverge very much among developed countries, in particular for agricultural products. Also, the commodity structure of imports varies considerably among developed countries, and hence the concrete effects of tariff cuts in developed country A on the economic wellbeing of developing countries will be rather different from those of tariff cuts in country B.

A third suggestion could be that exporting developed countries should ‘pay’ the compensation bill. Trade liberalisation is in their favour, and essentially implemented because they have pressed for it. The most significant economic gains from cuts in MFN tariffs accrue to the exporting developed countries which now get better market access. As they are the main gainers, why should they not compensate the losers?

When it comes to determining the ‘recipients’ of compensation, or rather which ‘recipient’ country should get how much compensation, the suggestions that can be made differ in a similar way. All developing countries as a group, it could be said, benefited from preferential treatment under GSP – hence compensation should not be targeted to individual developing countries, but be ‘paid’ to the aggregate of developing countries as a group. For an approach that is specific from developing country to developing country, however, speaks the fact that no two developing countries will be affected the same way, because of differences in their export structures, with regard to both products and destinations.

(3) Several instruments could be used to implement some form of compensation. The most obvious choice is a direct cash transfer in lump sum form (say a given annual payment for an agreed number of years, irrespective of market developments and other events). Economists have a preference for this form of compensation as it involves less distortions of resource allocation than any other form of compensation. In agricultural policy the concept of ‘decoupled payments’ has gained considerable ground in the last three decades or so, first in academic writings and then, to some extent, in the practice of agricultural policy making. Hence in agriculture there is a precedent for how to compensate losing producers for the negative effects on them of policy reforms. Why should this concept be limited to the domestic arena, and not also be used in international relations?

The most obvious case for compensation in the form of cash payments to developing countries is that where a domestic agricultural policy reform in a developed country has significant negative effects on them, and where domestic producers in the developed country concerned also receive compensation in the form of cash payments. Again, a possible reform of the EU’s sugar regime is an obvious case in point. The EU Commission has now suggested that a “more fundamental sugar reform” will be considered in 2002. One option for reform is to go in the same direction in which the EU’s cereals regime has moved since 1992, i.e. a significant cut in support prices and compensation to farmers in the form of direct cash payments.34 If this option were chosen by the EU, then a good case could be made that ACP countries (and India) should be treated the same way as EU sugar beet growers. After all, in this specific case the ACP countries (and India) not only ‘possess’ a tariff preference in the conventional form. They are also guaranteed the same sugar price as EU sugar producers (for a given quantity, like EU sugar producers who are also subject to a quota). As a result they are treated in essentially the same way as EU sugar producers. Why should they then not also be treated the same way as EU sugar producers when it comes to policy reform?

A slight variation on the theme of compensation through cash payments is additional financial or technical assistance for development projects, over and above the financial flows currently implemented. As suggested above, trade preferences can to some extent be seen as a substitute for financial and technical assistance. Hence, when preferences get lost and compensation is considered, why not return to the nearest alternative, i.e. larger volumes of development assistance? One form of assistance that is particularly relevant in a trade context is support for developing countries to help them comply with technical, phytosanitary and sanitary standards established by developed countries. It would be counter-productive to request lower standards specifically for imports from developing countries, as this is likely to have a negative impact on the image of products originating in developing countries. However, given the complexities of many standards in developed countries, developing countries often find it difficult to comply with them. Larger and more easily accessible assistance for developing countries in their attempts at living up to these standards can, therefore, be a very useful compensation for the erosion of trade preferences.

A different type of compensation could come in the form of additional tariff cuts for products of special export interest to developing countries. Two sub-categories of this form of compensation could be considered. First, tariff preferences for developing countries could be improved. For products where preferences already existed and where MFN tariff reductions result in preference erosion, new lower preferential tariffs can be set. However, this is possible only where the preferential tariff (after the MFN tariff reduction, which may also bring preferential tariffs down if they are set in relation to MFN tariffs) is still above zero. For products where tariff preferences were limited by TRQs, larger quotas can be agreed. For products not yet subject to tariff preferences in the past, preferential tariffs can be introduced. In a way, though, new and better tariff preferences only postpone the problem of preference erosion, because in future rounds of tariff cuts these newly established preferences will also be subject to erosion.

For this reason, a second sub-category of tariff cuts of interest to developing countries could be considered. This would be extra cuts of MFN tariffs for those products where most of world exports come from developing countries. While this type of compensation may appear attractive in principle, it may not have much potential in practice, as it may turn out that for the products concerned (i.e. those where most world exports come from developing countries), most developed countries have already extended tariff preferences, so that cuts in MFN tariffs for these products don’t help developing countries very much (if anything at all, given that MFN tariff cuts then also squeeze existing preference margins).

(4) Determining the magnitude of compensation is a most difficult business. As discussed above it is extremely difficult to provide sound empirical estimates of the quantitative economic implications of MFN tariff reductions for the exporting developing countries that used to benefit from tariff preferences. A simple calculation of preference margins, and of their decline when MFN tariffs are reduced, will not provide an appropriate measurement in all cases where quantities traded and prices received are likely to change as a result of the reduction in MFN tariffs. Moreover, in the many cases where MFN tariffs contain water, it is not even (easily) possible to estimate the change in the preference margin per unit of exports.

This said, the only approximation that is empirically feasible may well be such a mechanical estimate of the change in the total value of preference margins. Starting from that rough base, some standard reduction could be made, to account for the fact that in a number of cases the economic loss resulting from preference erosion (if there is any economic loss at all) will be smaller than the mechanically calculated decline in the total value of the preference margin. For example, it could be suggested that compensation should be no more than a given fraction (say, two thirds) of the mechanically calculated decline in the total value of the preference margin.

There are, however, some specific cases in which the magnitude of the loss, and hence of the potential claim for compensation, can be estimated with reasonable precision. These are the cases where the preference applies to a given quantity of exports only, i.e. where tariff quotas exist, where no exports from the developing countries concerned occurred at MFN tariffs, and where it is unlikely that beyond-quota exports will be shipped even after the reduction of the MFN tariff. In these cases one is essentially faced with pure quota rents, and the change in these quota rents can be estimated reasonably well. Again the EU sugar preferences for ACP countries and India are a case in point. In this case a relatively reliable indicator of the loss resulting from a reduction in the EU support price for sugar is the change of the EU price multiplied by the quantity of sugar exported to the EU from the developing country concerned. However, if the price reduction for EU sugar should go as far as making it no longer attractive for the ACP countries (and India) to utilize the whole preferential quota for sugar exports to the EU (or should the whole sugar Protocol of the Lomé/Cotonou Agreement be replaced by a different regime which no longer provides for preferential access to the EU sugar market), then it can be argued that the economic loss for the ACP countries concerned and India is somewhat smaller than the mechanically calculated current preference margin. This is because a reduction in the amount of sugar actually exported to the EU, and hence a cut in sugar production in the exporting countries, saves production costs in these countries (Wissenschaftlicher Beirat, 1994, p. 34-36). Because of these cost implications, the actual economic net benefit to the ACP countries concerned and to India from exporting preferential sugar to the EU is smaller than the mechanically calculated preference margin.35 However, even in cases like this, the change in the preference margin resulting from a reduction of the domestic price in the importing country is still a relatively good starting point for considering the magnitude of potential compensation, and appropriate downward adjustments can then be made from that sum.

In summary, the issue of compensation for the erosion of preference margins is highly complex. It is not necessarily clear that all reductions of MFN tariffs for products where preferences exist actually result in an erosion of (economically meaningful) preference margins. There are cases where the erosion of preference margins is (partly or wholly) outweighed by the favourable market effects of trade liberalisation. Moreover, where preference erosion clearly results in an economic loss to the exporting countries concerned, there are arguments both in favour and against compensation. In addition, if the case for compensation is accepted, it is not unequivocally clear who should ‘pay’, and who should ‘receive’ compensation. Various forms of compensation can be considered, without any of them having a clear-cut priority. And finally, in many cases it will be difficult to provide a reliable estimate of economic effect of preference erosion, and hence of the magnitude of compensation that may be justified.

All this is not to say that (i) preference erosion is a non-issue, and that (ii) compensation for preference erosion has no role to play in a multilateral round of trade negotiations. However, the issues discussed here should warn against suggesting simple solutions. In the end, compensation will be a matter of negotiation. As a rough guideline in discussing the issue of compensation, it may be useful to distinguish between two categories of preferences, i.e. GSP regimes on the one hand and specific deep preferences for limited groups of developing countries on the other hand. Where GSP preferences erode as a result of multilateral negotiations on tariff cuts, the most natural way to negotiate on compensation may be to seek a structure of extra cuts of MFN tariffs that benefits developing country exporters. On the other hand, where very specific and deep preferences for individual countries and commodities are concerned, like under the EU sugar regime for the ACP countries, a relatively strong case can be made for cash compensation. The reasoning for suggesting this distinction is as follows.

The erosion of GSP preferences under multilaterally agreed tariff cuts affects a large number of countries, on both the recipient and the granting side. In this case the question of who should receive and pay compensation is particularly acute. Also, the magnitude of actual preference erosion and hence of compensation claims is difficult to determine with so many tariff changes and participants involved. Cash compensation would be extremely difficult to implement, because it would essentially require an international fund into which all developed countries pay, and which then makes compensation payments to all developing countries. This does not appear to be an attractive proposition. For all these reasons, extra cuts of MFN tariffs on goods of particular export interest for developing countries, over and above the generally agreed tariff reductions, may be the most appropriate form of compensation for the erosion of GSP preferences.

Where specific deep preferences, granted by individual developed countries, to selected developing countries, are concerned, like in the EU/ACP sugar case, the opposite reasoning applies to all these criteria. It is fully clear who the recipients are and who the donor is. The magnitude of preference erosion can be estimated with reasonable precision, not the least because these preferences typically come in the form of TRQs, and quotas tend to be fully used. Cash compensation is a conceivable option, not the least because domestic producers in the developed countries tend to be also compensated that way when significant liberalisation occurs. For all these reasons, explicit compensation, in cash form, may be the appropriate solution in these cases.

8 Conclusions

Trade preferences have played an important role in the commercial relations between developing and developed countries since UNCTAD has called for them in the mid-1960s. They violate one of the fundamental principles of the GATT, i.e. non-discrimination on the basis of MFN treatment. It has, therefore, taken the GATT a while to come to grips with trade preferences. However, based on the ‘development factor’, the GATT has finally acknowledged trade preferences for developing countries and agreed the “Enabling Clause”. Yet, the role of trade preferences has kept changing in the GATT. After special preferential schemes for constrained groups of developing countries, such as the EU’s preferences for ACP countries under the Lomé Convention, were silently tolerated for quite some time, more recently it was made clear that such regimes are inconsistent with the requirement to provide non-reciprocal preferences to all developing countries on the same basis. This is one of the reasons why trade preferences are currently in a state of flux. The deeper preferences that used to be provided to selected groups of developing countries in the past may in the future be extended to all least-developed, and possibly also to vulnerable developing countries, and may actually be improved to become completely duty-free access to (some) developed countries. The non-reciprocal preferences for other sub-groups of developing countries, on the other hand, are likely to be transformed into reciprocal free-trade arrangements between the developed and developing countries concerned, adding to the anyhow growing coverage of regional trading arrangements in the modern world of international trading relations.

This leaves the remaining ‘traditional’ generalised preferences for developing countries in a somewhat uncertain future. Moreover, with progressive overall liberalisation of trade, since the Uruguay Round including agricultural trade, it becomes less and less clear what the role is that trade preferences can play in the future. In addition, the benefits of trade preferences may occasionally have been overestimated in the past, and the costs involved may not always have been fully appreciated. Against this background one may well suggest that the high time of trade preferences is over, at least as a general element of commercial relations among all developed and developing countries.

This is not to say that trade preferences should no longer continue to play a role in international trade negotiations. Zero-duty access for the least-developed and vulnerable countries is a policy that is certainly worth pursuing. In agriculture, where tariffs are often still extremely high, this policy can be rather useful until further reductions have brought tariffs down to negligible levels. One should, though, note that such zero-duty preferences can accelerate the movement towards lower overall levels of protection in agriculture, by undermining the sustainability of the protective regimes currently maintained in many developed countries. This is only welcome from an overall perspective of trade liberalisation, while for the least-developed countries it means that any success in achieving zero-duty access contains the seed of making the very preferences useless because it tends to help bringing MFN tariffs down more rapidly.

However, in the final analysis, overall trade liberalisation is in the interest of all countries, including the developing countries so far benefiting from trade preferences. While there may be short-run losses because of preference erosion, there are long-run benefits of improved access to all markets for all products. This is also why the preferable form of compensation for preference erosion probably is a further reduction of tariffs on products of special export interest to the developing countries concerned. On the other hand, where specific deep preferences on agricultural products for selected groups of developing countries have essentially put these countries in the same camp as domestic producers of these products in the developed countries, like under the sugar protocol of the Lomé Convention, cash compensation is an option worth considering, like such compensation is often offered to farmers in developed countries. It may well be advisable to go for such compensation, rather than trying to fight a losing battle against liberalisation of the respective market regimes in the developed countries concerned.

Finally, even though the high time of trade preferences may be over, more research should be done on the effects of trade preferences while they still play some role. It is surprising how little is known about the actual effects of tariff preferences. Some limited research has been done on the overall size of preference margins, and a few studies have looked into the effects on trade flows. However, much more research should be done in these areas. Also, very little is known about the concrete effects that given trade preferences for given products have had in individual developing countries, a matter on which case studies could throw much more light. The limited research available is not the firm basis one would like to have for making more concrete suggestions regarding the future role of trade preferences in the multilateral trading system.

References

Baldwin, R.E. (1984), Trade Policies in Developed Countries. In: R.W. Jones and P.B. Kenen (eds.), Handbook of International Economics. Volume I: International Trade. Amsterdam etc.: North-Holland.

Borrmann, A., C. Borrmann, C. Langer and K.-W Menck (1985), The Significance of the EEC’s Generalised System of Preferences. Hamburg: Verlag Weltarchiv.

European Commission (2000a), Everything But Arms (EBA) Council Regulation amending the GSP so as to extend duty and quota free access to least-developed countries (LDCs). Brussels,1 March 2001. http://www.europa.eu.int/comm/trade/miti/devel/eba4_sum.htm.

European Commission (2000b), Commission proposes overhaul of sugar market. Brussels, 4 October 2000. IP/00/1109.

European Commission (2001), EU Trade Concession to Least Developed Countries: Everything But Arms Proposal. Possible Impacts on the Agricultural Sector. http://europa.eu.int/comm/commissioners/fischler/eba_en.pdf.

Grethe, H. and S. Tangermann (1999), The New Euro-Mediterranean Agreements. An Analysis of Trade Preferences. Paper prepared for Commodities and Trade Division, FAO Economic and Social Department. Diskussionsbeitrag 9902. Göttingen: Institute of Agricultural Economics.

Josling, T.E. (1997). Implications of Regional Trade Arrangements for Agricultural Trade. Economic and Social Development Paper No. 133 (Rome: Food and Agriculture Organisation).

Long, O. (1985), Law and its Limitations in the GATT Multilateral Trade System. Dordrecht, Boston, Lancaster: Martinus Nijhoff.

Oyejide, T. Ademola (1997), “Africa’s Participation in the Post-Uruguay Round World Trading System”. Program for the Study of International Organization(s) (PSIO) Occasional Paper, WTO Series Number 6. Geneva: PSIO.

Robertson, D. (1999), “Trade Challenges for Developing Countries: Preparations for the New WTO Round”. A report prepared for the Commonwealth Secretariat. London: Commonwealth Secretariat

Roessler, F. (1998), “Domestic Policy Objectives and the Multilateral Trade Order: Lessons from the Past”. In: A.O. Krueger (ed.), The WTO as an International Organization. Chicago and London: The University of Chicago Press.

Senti, R. (1986), GATT - Allgemeines Zoll- und Handelsabkommen als System der Welthandelsordnung. Zürich: Schulthess Polygraphischer Verlag.

Sharma, R. (1997), The Impact of the Marrakesh Agreement on Trade of Agricultural Products in ACP Countries. FAO Contribution to the ACP Secretariat. Rome: FAO Commodities and Trade Division.

Tangermann, S. (2000), The Cotonou Agreement and the Value of Preferences in Agricultural Markets for the African ACP. Paper prepared for UNCTAD. Göttingen: Institute of Agricultural Economics. November 2000.

Tangermann, S., u. T. Josling (1999), The Interests of Developing Countries in the Next Round of WTO Agricultural Negotiations. Study prepared for UNCTAD. Göttingen: Institute of Agricultural Economics.

UNCTAD (1997), The Post-Uruguay Round Tariff Environment for Developing Countries. Document TD/B/COM/1/14. Geneva: UNCTAD.

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

Weston, A. (1995), The Uruguay Round – Costs and Compensation for Developing Countries. Geneva: UNCTAD.

Weston, A., V. Cable and A. Hewitt (1980), The EEC’s Generalized System of Preferences. London: Overseas Development Institute.

Wissenschaftlicher Beirat beim Bundesministerium für Ernährung, Landwirtschaft und Forsten (1994), Vorschläge für eine grundlegende Reform der EG-Zuckermartpolitik. Münster: Landwirtschaftsverlag.

Yamazaki, F. (1996), Potential Erosion of Trade Preferences in Agricultural Products. “Food Policy”, Vol. 21, pp. 409-18.

Yeats, A.J. (1994), What Are OECD Trade Preferences worth to Sub-Saharan Africa? Policy Research Working Paper No. 1254. Washington D.C.: World Bank.

Appendix: Price and Quantity Effects of Trade Preferences and MFN Tariff Reductions

The algebraic model presented here is a very simple static three country model, very much along the lines of Viner’s (1950) theory of customs unions, though the model presented here will not reflect a customs union, but a unilateral tariff preference. The three countries (or country groups) considered are an importing country granting a trade preference (country A), the group of exporting developing countries receiving the trade preference (country B), and other (developed) exporting countries (country C) facing the MFN tariff on exporting to country A. Only one product is considered. For simplicity, import demand and export supply curves are assumed to be linear, and both the MFN and the preferential tariff are assumed to be specific tariffs.

Import demand in country A, MA, and export supply in countries B and C, XB and XC, depend on domestic prices Pi.

(1) MA = mA - nAPA

(2) XB = mB + nBPB

(3) XC = mC + nCPC

Note that these equations are written such that all slope parameter ni are positive.

Export prices received in countries B and C are the domestic market price in country A minus the preferential tariff TB respectively the MFN tariff TMFN (transport cost and trading margins are neglected).

(4) PB = PA - TB

(5) PC = PA – TMFN

Equations (4) and (5) imply that the price received by preferred exporters from country B exceeds the world market price, i.e. the price received by MFN exporters from country A, by the preference margin TMFN – TB ,

In equilibrium, aggregate export supply equals import demand.

(6) XB + XC = MA

This system of six equations can be solved for the six unknowns, i.e. three quantities and three prices.

If export supply of the non-preferred country C is assumed to be infinitely elastic, price in country C, equivalent to the world market price, is exogenously fixed, PC* . The price in importing country A is equally fixed, at

This is true at least as some part of country A’s imports are sourced in country C, i.e. as long as the trade preference does not result in a situation where country B takes over all of the import market in country A. At the same time, as long as that is the case, price in country A is not affected by the trade preference.

The price received by the preferred exporters is determined by the fixed world market price plus the preference margin,

In this simple case, it is clear that the price received by the preferred exporters responds inversely to the level of the preferential tariff. Hence introduction of a trade preference (i.e. a reduction of TB from the original situation where TB = TMFN) raises the price received by the preferred exporters, and hence their export quantity and welfare. At the same time there is no doubt that complete liberalisation, i.e. setting both TMFN and TB to zero, results in a price to preferred exporters which is less than the price received under a positive MFN tariff and a trade preference. In other words, the tariff-cum-preference scenario ‘sucks’ exports from the preferred exporters B in a direction which is not sustainable under free trade.

On the other hand, if the MFN tariff is reduced without being eliminated, the price received by the preferred exporters, the quantity exported by them, and their economic welfare directly depend on what happens to the preference margin while the MFN tariff declines. If the preferential tariff is defined as the MFN tariff minus a given number of monetary units per quantity (i.e. if the preference margin is kept constant), and if the reduced MFN tariff is still at least as high as the preferential tariff, then reduction of the MFN tariff has no effect on the price received by the preferred exporters, and hence on their export quantity and welfare. Conversely, if the preference margin declines as a result of the reduction of the MFN tariff, then the preferred exporters clearly lose on price, quantity and welfare.

In the more general case where export supply of the MFN exporter C is less than infinitely elastic, prices in all three countries depend on the tariffs set. Solving the system of equations for the prices, it can easily be shown that price in country A then is

while the price received by preferred exporters is

(7) PB = [nCTMFN – (nA + nC)TB + m] / n ,

where m and n are defined as m ≡ mA – mB – mC and n ≡ nA + nB + nC .

In this case it is clear that, with a given MFN tariff, price in country A is the lower the more generous the trade preference is, i.e. the lower TB is. Conversely, the price received by preferred exporters is the higher the lower the preferential tariff is set. In this case it is not unequivocally clear whether the trade preference has the effect of ‘sucking’ country B’s exports in a direction that is not sustainable under free trade. This depends on the sign of the term nCTMFN – (nA + nC)TB . If that term is positive, then complete trade liberalisation will make PB decline, and vice versa. Which sign prevails depends on both tariff rates and response parameters. The larger the preference margin is (i.e. the lower TB with a given TMFN), the more likely it is that the export structure resulting from the trade preference is not sustainable after complete trade liberalisation. However, with a zero preferential tariff it is unequivocally clear that reduction of the MFN tariff to zero reduces the price received by preferred exporters, and hence triggers downward adjustment of the export volume.

What happens to the price received by preferred exporters (and hence to their export quantity and welfare) if the MFN tariff is reduced again depends on how the preference margin behaves in this case. This is best seen if equation (7) is rewritten such that the preference margin ( PM ≡ TMFN – TB ) can bee seen,

(7a) PB = [(nA + nC)PM – nATMFN + m] / n .

If the MFN tariff reduction leaves the size of the preference margin unaffected, then PB definitely rises and hence preferred exporters gain. If the preference margin goes down by the same amount of monetary units as the MFN tariff (i.e. the preferential tariff remains constant when the MFN tariff is reduced), then preferred exporters definitely lose. In between these two bordering cases, however, there are situations where the preference margin declines somewhat with the MFN tariff reduction, but the price received by preferred exporters still rises, and hence their exports grow and welfare improves.

1 On the origins of the GSP, see for example Borrmann et al. (1985, pp. 23-27), Long (1985, p. 99 ff), and Senti (1986, p. 112 ff), and the literature cited there.

2 Specifically generous preferences for the leas-developed countries are legal under the Enabling Clause, see below, Section 6.

3 For a summary of the “Everything but Arms” initiative, see European Commission (2000a).

4 Such fears were supported by an analysis done by the European Commission (2001).

5 See Josling (1997).

6 For an analysis of agricultural trading conditions under these agreements, see Grethe and Tangermann (1999).

7 The algebra representing the argument advanced here is presented in the Appendix.

8 In standard terms of trade theory, country A is ‘small’ vis-ā-vis the other developed countries, facing an infinitely elastic supply of imports. In reality, country A does not necessarily have to be ‘small’ in conventional terms. The reason why the product considered here (‘maize’) is supplied with infinite price elasticity can also have to do with production or consumption conditions in other developed countries. For example, it may be the case that the product can be produced at constant marginal cost for all quantities relevant to the problem discussed here.

9 Lack of any price effect in country A follows from the assumption of infinitely elastic supply from other developed countries, and prevails as long as the increase in imports from country group B is not sufficient to wipe out all imports from other developed countries.

10 This case of a production structure ‘sucked’ in the wrong direction may sound somewhat artificial. However, it is probably very real where specific deep preferences have been provided, with the effect of making the developing countries concerned largely dependent on the preferential exports considered. This is likely to be the case in a number of ACP countries supplying sugar and bananas to the EU under preferential conditions. Once the EU begins to liberalise its sugar and banana regimes, these countries will have a difficult time to adjust their production structures.

11 In a dynamic setting, it is of course possible that the world market price of maize has increased meanwhile, because of changes in supply/demand conditions. However, this would have also happened in the absence of the trade preference, and it does not affect the conclusion that in the moment country A eliminates its tariff on maize country group B experiences a price reduction on its exports to country A.

12 In terms of trade theory, in the case discussed so far the trade preference results only in (welfare-reducing) trade diversion and not in (welfare-enhancing) trade creation.

13 With a preferential tariff of zero, however, it is still true even in this case that preferred developing countries still receive a price on exports to country A that is higher than the price prevailing after complete tariff removal. Hence the problem of a ‘wrong’ production structure is still relevant under those conditions. It is only where the preferential tariff is close to the MFN tariff that complete liberalisation does not result in a price loss for preferred developing countries, and hence does not require a re-adjustment of their production structure. See the Appendix.

14 The technical difficulty requiring a lot of work is that trade statistics have to be combined with tariff information, both often not coming in the same product disaggregation.

15 In empirical research, it is often more convenient to work with ad valorem tariffs (or their equivalents) and trade values rather than quantities.

16 When calculating preference margins for several exporting countries shipping to one preference-giving importing country, it is convenient to use import rather than export statistics, because one data source can be used for all exporting countries, and because of data consistency.

17 For the two remaining Protocol products, bananas and rum, preference margins were not estimated in the study. An estimate of preference margins for bananas is difficult because of the complex structure of the EU banana regime. Preferences for rum are no longer relevant as the EU’s MFN tariff on rum is now zero anyhow.

18 The idea behind this assumption was that participants in the next round of WTO negotiations might agree to another reduction of tariffs by 36 per cent, from their pre-Uruguay Round base level (so as to avoid the effect of a thinning base with successive rounds of reductions), applied as a flat rate to all products (without less reduction for ‘sensitive’ products), and making up for below-average tariff reductions on ‘sensitive’ products in the Uruguay Round.

19 The background to that particular part of the analysis is the WTO legal problem regarding EU preferences for the ACP countries. One theoretical way of solving that problem would be to eliminate the ACP preferences, and treat ACP countries in the same way as other developing countries.

20 For a somewhat more extensive discussion of this issue, see Grethe and Tangermann (1999).

21 See Roessler (1998).

22 For a short history of the GATT’s dealings with trade preferences for developing countries, see Long (1985, p. 99 ff), and Senti (1986, p. 112 ff).

23 GATT, Basic Instruments and Selected Documents (BISD), 26th Supplement (1980), pp. 203-5.

24 WTO document WT/L/304, 17 June 1999.

25 Under the respective Marrakesh Understanding, the EU’s Lomé waiver would have expired automatically after the Uruguay Round, on 31 December 1996. However, on 14 October 1996 it was again extended by the WTO Members until 29 February 2000. See WTO document WT/L/186, 18 October 1996.

26 Cotonou Agreement, Article 37:9. This provision of the Cotonou Agreement also foresees that the new regime for LDC “will simplify and review the rules of origin, including cumulation provisions that apply to [LDC] exports”.

27 WTO document G/C/W/187, 2 March 2000.

28 According to press reports, there is a general inclination among the WTO Members to grant the waiver. However, some WTO Members are reluctant to go ahead with a decision as long as the EU has not yet brought its banana regime into line with the WTO requirements.

29 WTO document WT/L/104, 24 November 1995.

30 See the proposal for a WTO Preference Scheme by Oyejide (1997).

31 Preference erosion could, though, still occur to some extent even under such a regime. Where preferential tariffs are defined in specific-tariff form, the preference margin would erode when the MFN tariff becomes less than the preference margin set. Where preferential tariffs are defined as given percentages of MFN ad valorem tariffs, the absolute preference margin declines in parallel with the MFN tariff.

32 The algebra showing this result is presented in the Appendix.

33 The algebra showing this result is presented in the Appendix.

34 When submitting its proposal for a much more limited immediate amendment of the EU’s sugar regime in October 2000, the EU Commission already mentioned the option of going in this direction. However, it also suggested that “the very significant budgetary consequences preclude this option” (document IP/00/1109, Brussels, 4 October 2000, available on the website of the EU Commission, http://europa.eu.int/comm/dg06/
newsroom/en/32.htm). In any case, the mere fact that this option is mentioned by the Commission indicates that it will be considered seriously when it comes to discussing EU sugar reform in 2002.

35 In a graphical representation of the market situation in the exporting country concerned, the cost saving is equivalent to the triangular area below the supply curve and above the world market price, between the TRQ quantity and the quantity produced at the world market price.