Trade reforms are being increasingly featured in the design of adjustment programs supported by Fund resources. This paper reviews the trade policy content of Fund-supported programs approved in the period 1990–93.2

Trade reforms are being increasingly featured in the design of adjustment programs supported by Fund resources. This paper reviews the trade policy content of Fund-supported programs approved in the period 1990–93.2

Conceptual Framework and Methodology

Open markets are essential to attaining high-quality growth. A well-established and widely accepted rationale for trade reform is that an outward-oriented strategy is more likely to assist in achieving sustainable long-term growth than an inward-oriented one.

Trade reform encompasses improving transparency, increasing predictability, and liberalizing the trade policy regime. Transparency is improved by reducing the complexity of regulations and shifting from quantity-to-price-based measures; predictability is gained by adhering to a clear and consistent direction for trade policy and avoiding stop-go measures; liberalization is achieved by reducing protection levels and dispersion thereby increasing the role of competitive market forces at the expense of discretionary state intervention. Trade reform helps economic agents to base their production and investment decisions on market price signals, improves resource allocation, reduces anti-export bias, spurs domestic firms to restructure in the face of foreign competition, and increases national and international welfare through efficiency gains. In economies in transition, it is an even more potent instrument in the transformation to market economies.3

The more difficult question is, How much trade reform is necessary and over what periods should it be achieved? While there is no one magic level of protection to target, the experience of successful reformers illustrates what is achievable. Trade liberalization does entail dislocations for the hitherto protected industries and the transition has to be managed in an orderly manner. A further complicating feature is that in practice trade policy instruments are used not solely for protection and resource allocation objectives (henceforth referred to as “trade policy” objectives), but are often used to serve other objectives—such as fiscal revenue, balance of payments management, and income distribution goals (henceforth referred to as “nontrade policy” objectives).

The conceptual framework for trade reform must, therefore, take account of two principles. First, trade policy must be cast in terms of medium-term efficiency goals—deviations to meet short-term nontrade policy objectives should be kept to a minimum to meet emergency situations and should be strictly temporary. Second, alternative policy instruments must be instituted to cater to nontrade policy objectives. Translating some aspects of this conceptual framework into practical policy guidelines is discussed later.

When trade reform is undertaken in the context of a Fund-supported program, it becomes an element of a comprehensive, integrated policy package aimed at achieving noninflationary growth and a viable external position with a high level of resource use in the medium term. Such an integrated approach implies that the trade reform is likely to have a greater chance of success, as it is likely to be accompanied by complementary macro-economic and structural measures. Such measures would aim to establish financial stability, to deregulate domestic product and factor markets, and to improve the responsiveness of economic agents to relative price changes, as well as to improve the efficiency of the public sector and the climate for private sector investment through public expenditure and tax program improvements, public enterprise and marketing board reforms, and so on. Trade reform under Fund-supported programs often provides additional benefits to the member in terms of improving confidence and credibility, catalyzing other forms of external financing, and encouraging capital inflows, including foreign direct investment.

This survey is based on a review of the trade policy content of programs supported by Fund arrangements under the structural adjustment facility (SAF), enhanced structural adjustment facility (ESAF), extended Fund facility (EFF), and stand-by arrangements. It covers all SAFs, ESAFs, EFFs, and stand-by arrangements approved by the Executive Board of the Fund between 1991 and 1993 plus the three arrangements under the ESAF approved in 1990 (Table 1). This yields 78 arrangements with 59 countries, covering 47 stand-by arrangements, 20 ESAFs, 8 EFFs, and 3 SAFs. A multi-year program is counted as a single arrangement. The 59 countries are classified into the following regional groups: 14 as economies in transition, of which 5 are grouped the rest under “Other”; the remaining 45 are classified as developing countries, of which 21 are in Africa, 6 in Asia, 3 in the Middle East, and 15 in the Western Hemisphere.4

Table 1.

Stand-By Arrangements and SAFs Approved 1991–SM and EFFs and ESAFs Approved 1990–93

(Date of Approval)

article image
article image
Source: IMF, International Financial Statistics and Transactions of the Fund, various issues.

Czechoslovakia was replaced by the Czech Republic and the Slovak Republic on January 1, 1993.

In this paper, trade reform efforts are assessed by classifying the 59 countries according to the restrictiveness of their trade regimes at the beginning and the end of the period under review, that is, the end of 1990 and the end of 1993. Since different measures of openness only partly capture the protective effect of tariff and nontariff trade barriers, the characterization of trade regimes as “open” or “restrictive” needs to be viewed in the light of the standards used here, rather than in any absolute sense. In general, the standards used here are stricter than those used in previous similar trade surveys. This partly reflects the decline in average protection levels in general as developing and transition economies liberalize their trade regimes. But it also reflects an attempt to take better account of average tariff levels inclusive of the various other charges that add to the cost of imports but that are not evident from data on statutory tariffs alone. Eliminating quantitative import restrictions alone does not justify characterization of the trade regime as open or “liberal,” if in fact there exists a variety of import charges and tariffs that could add up to relatively high levels of protection. In this paper, the stance of trade policy is evaluated on the basis of the coverage and intensity of quantitative restrictions and the average level of (the all-inclusive) statutory tariffs.5

The tariff regime is classified into three categories (“restrictive,” “moderate.” and “open”), based on the experience of the successful trade reformers (e.g., in Latin America) and taking into account the average tariff of less than 5 percent prevailing in major industrial countries:

Classification of Tariff Regime

article image

Since the classification of the restrictive tariff category is open ended, countries with very high tariffs in 1990 might make substantial progress in reducing tariffs during the period under review, but still not achieve the average tariff levels of the “moderate” category. Hence, the nature of reforms in countries that remained in the restrictive tariff category at the end of 1993 are examined further; in particular, those countries whose average tariffs remained above 40 percent are distinguished from those that were able to reduce average tariffs to a range of 25 to 40 percent.

The difficulties of measuring QRs are well documented.7 While tariff equivalents of QRs are conceptually appealing, they pose immense data problems, particularly on an economy-wide comparative basis for a large group of countries. Hence, the assessment here is based on the import and export coverage and the intensity of the respective QRs.8 Coverage is measured as the proportion of tariff lines covered by QRs; where this information is not available, the share of imports affected by the QR is used as a proxy.9 Measuring the coverage alone is, however, not sufficient, because various QRs differ in their protective effects; hence, the intensity of the QR needs also to be taken into account. For example, an automatic authorization procedure applied to all imports is less restrictive than outright import bans. QRs are classified as high-, medium-, or low-intensity items. High-intensity practices include, for example, import bans, binding quotas, nonautomatic licensing, and discretionary foreign exchange allocation procedures. Transparent and automatic licensing procedures and nonbinding quotas, for example, are considered low-intensity QRs. Without complete and reliable information in some cases, judgement is used for the QR classification, yielding the following matrix:

Classification of QR Regime

article image

The next step is to consider the combined effect of QRs and tariffs, in order to measure the overall stance of the trade regime. Liberalization in one area may still leave major impediments to trade. For example, lowering tariffs will not increase import competition (though it may reduce costs) if quotas remain binding. Similarly, eliminating QRs might not induce further imports if tariffs remain prohibitive (although this does improve transparency). The approach used here is to consider the more restrictive measure, whether tariffs or QRs, as the determinant of the overall stance of the trade regime. Thus, a “restrictive” tariff regime (with average tariffs above 25 percent) combined with an “open” QR regime would yield a characterization of the overall trade regime as being “restrictive.” Accordingly, the following matrix guides the determination of the overall trade policy stance:

Combined Trade Policy Stance

article image

The above approach to the classification of trade regimes implies that relative progress in trade reforms is not measured in terms of the comparative effort to reduce trade restrictions, but by the levels of protection attained. For example, a country may sharply reduce its average tariffs—say from 100 percent to 50 percent—in the period under consideration, and still be classified as restrictive; in contrast, another country may reduce its average tariffs by a small amount—say, from 11 percent to 9 percent—and shift from the classification “moderate” to “open.” To identify the most restrictive trade regimes remaining at the end of 1993 among the countries in the “restrictive” category, a further benchmark, as indicated above, is used, namely, whether average statutory tariffs (inclusive of other charges) remained above 40 percent.

The analysis of the nature and extent of the trade policy content of Fund-supported programs, and the characterization of the degree of openness of the trade regimes, is made for all program countries, taking account of all the arrangements in the period (i.e., the “population”). Assessing the contribution of trade reform to program progress in terms of macroeconomic or structural adjustment, or establishing empirical causal links is beyond the scope of this paper. A simple characterization of the group of slower and faster reformers among the program countries is attempted in terms of major macroeconomic variables. The discussion of specific design issues, which is based on a more in-depth scrutiny of a sample group of countries, identifies some macroeconomic developments that have helped shape the design of trade reform.

Trade Policy Content

There was substantial progress toward market opening under Fund-supported programs in the period under consideration, as developing countries and economies in transition continued trade liberalization begun in the second half of the 1980s or initiated in the 1990s. All 59 program countries liberalized at least some parts of their trade regimes in at least one of their arrangements. Nearly all countries (56) lowered tariffs, and all but six reduced QRs, with most implementing a combination of both.

Liberalization of Quantitative Restrictions

The trend toward reduced reliance on QRs observed in the 1980s intensified further in the 1990s. Many countries narrowed the scope and reduced the intensity of QRs. The liberalization of QRs was sufficiently substantive to generate a major shift in the classification of their regimes into the “open” category (Table 2). In particular, the number of restrictive QR regimes fell from 35 to 14 between the end of 1990 and the end of 1993. By the end of 1993, over two thirds of program countries could claim to have achieved “open” QR regimes, compared with less than one fourth only three years earlier.

Table 2.

Summary of Trade Regimes

article image
Sources: IMF; GATT; and World Bank.

Progress in liberalizing QRs was observed across all geographical regions. Particularly in Africa, exchange system reforms enabled many countries in a short period of time to eliminate complex quantitative controls that had been maintained for many years to ensure the administrative allocation of foreign exchange. By 1993, over 60 percent of African program countries had open QR regimes, while less than 30 percent maintained restrictive ones. Ethiopia, Mauritania, and Tanzania, for example, all moved from a restrictive to a relatively open QR regime during 1990–93 by replacing arbitrary administrative foreign exchange allocation procedures with foreign exchange auctions. Tanzania also limited the negative list under the open general license system to goods controlled for health and security reasons and eight luxury items. Mauritania eliminated permits, while Tanzania removed licensing as well as minimum prices for exports. In addition, a number of countries initially in the moderately restrictive QR-regime category, where licensing covered certain products only (e.g., Cameroon and Morocco), made further progress in reducing the list of products subject to licensing and easing authorization procedures, thereby attaining an open regime by the end of 1993. African countries that continued restrictive QR regimes during the period under review included Côte d’Ivoire, Nigeria, and Zimbabwe. The first two made little progress in liberalizing their extensive import prohibitions, licenses, and prior authorizations. Nigeria, for example, justified its QRs on balance of payments grounds in the GATT, but these also constituted an important instrument in encouraging domestic production; it reversed its initial reforms of the foreign exchange allocation system. Zimbabwe, on the other hand, made gradual progress toward a market-based system of foreign exchange allocation, by increasing the ratio under the export retention scheme and, initially, also expanding the open general import license system; nevertheless, by the end of 1993, imports not covered by the more liberal foreign exchange allocation procedures still constituted more than a fourth of total imports.10

In the Western Hemisphere, more countries started the period with open QR regimes than in Africa. Most of the remainder quickly liberalized in line with a shift in development strategy away from reliance on import substitution toward exported growth. By the end of 1993, all but two countries maintained “open” QR regimes. Examples of fast reformers were Ecuador and Honduras, which, respectively, eliminated extensive import prohibitions and prior authorization procedures or licensing requirements. In some cases (e.g., Brazil and Jamaica), restrictive licensing procedures for selective imports were discontinued as part of ongoing trade reforms, moving the countries from the “moderate” to the “open” QR-regime category.

The economies in transition had the most impressive record in eliminating QRs. In 1990, all of these countries (except Poland) relied heavily on QRs as part of the system of central planning. As state involvement in international trade was reduced dramatically, QRs were eliminated across the board. By the end of 1993, trade regimes in Eastern Europe and the Baltics were largely free from QRs. The Asian economies in transition have been slower in eliminating QRs, largely reflecting continued state involvement in foreign trade and slower progress in establishing market-based economies.

In most countries outside the economies in transition, QRs were chiefly imposed on imports, and this was also the area where most of the liberalization occurred. But there was progress also in eliminating QRs on exports, particularly as part of the liberalization of the exchange systems aimed at improving export incentives. For example, Burundi substituted export licenses for the less restrictive system of export declarations; Egypt eliminated export bans; and India and Tanzania reduced export licensing, quotas, permits, and the use of minimum export prices. In the economies in transition, where exports were tightly controlled at the beginning of the period under review, significant progress was made in liberalizing export regimes, in close association with reforms in other areas, notably price liberalization and monetary independence. By the end of 1993, the Baltic countries and most East European program countries had phased out QRs on exports. In other economies in transition, the continued use of restrictions reflected mainly a desire to maintain domestic prices below world prices for key inputs and some consumer goods.

Tariff Reform

In contrast to the striking results on QR liberalization, progress on tariff reform was modest. In the initial period, $4 percent of all developing program countries had average statutory tariff rates above 25 percent. Only seven developing countries (six of them in the Western Hemisphere) had moderately restrictive tariff regimes (with average rates between 11 percent and 25 percent) and none had relatively open tariff regimes (i.e., with average tariffs 10 percent or lower). By the end of 1993, 12 shifted from the restrictive to the moderately restrictive tariff category, 26 (nearly 60 percent of the total) remained in the restrictive category, and I achieved an open tariff regime.

Many developing countries in Africa, Asia, and the Middle East lowered tariffs in the 1991–93 period, but the reductions were not large enough to bring the average statutory rate to 25 percent or less.11 Tariffs in several cases were sharply reduced from prohibitive levels during this period. For example, the average statutory rate (including other charges) fell in Burundi from 51.5 percent to 40.4 percent, in India from 125 percent to 71 percent, in Kenya from 42 percent to 34 percent, and in Pakistan from 85 percent to 48 percent. By the end of 1993, nine program developing countries maintained very restrictive tariff regimes with average statutory rates above 40 percent (Bangladesh, Burkina Faso, Burundi, Equatorial Guinea, Gabon, India, Mauritania, Pakistan, and Rwanda).

In the Western Hemisphere, countries generally had achieved lower tariff levels compared with other developing countries by the end of 1990. In the early 1990s, six more countries in this region moved from the restrictive to a moderately restrictive tariff regime (Brazil, Honduras, Jamaica, Nicaragua, Panama, and Uruguay) and one country (Ecuador) achieved an open tariff regime. Significant tariff reductions were generally undertaken as part of ambitious liberalization programs to restructure and open the economy, often lowering the maximum tariff rate to 20 percent. Being concerned with competitiveness and export performance, many of these countries tried to imitate the Mexican tariff liberalization.12

In economies in transition, the most important aims of tariff reform were to reorient tariff structures to the needs of market-based economies and to eliminate anomalies inherited from the centrally planned system. The latter included the removal of discretionary exemptions through minimum tariffs and the reduction of excessively high rates on selective goods. While the overall picture does not appear to have changed much from the end of 1990 to the end of 1993, the country composition did change as the tariff regimes of half of the program countries changed category. Tariffs in a number of countries with moderately restrictive regimes at the end of 1990, such as Estonia and Lithuania, were reduced significantly so that they had open tariff regimes by the end of 1993. In contrast, in several countries where tariffs were already low at the end of 1990 (for example, Albania, Bulgaria, Mongolia, and Poland) revenue and protection considerations motivated increases in the statutory average tariff such that these tariff regimes became moderately restrictive by 1993.13

Export taxes were rather insignificant and only a minor contributor to total revenue from taxes on international trade in most developing countries (with the exception of some primary goods exporters in Africa). Hence, some countries were able to eliminate them in the 1990–93 period (e.g., Burundi, Ethiopia, except on coffee, and Peru). The importance of export taxes has generally also decreased in economies in transition. In a few cases, new export taxes temporarily replaced QRs and were subsequently eliminated (e.g., Latvia and the Kyrgyz Republic). In addition, several program countries reduced the anti-export bias of their trade regimes by streamlining and widening the scope of the duty-drawback system (Ethiopia, Nepal, and Papua New Guinea) or improving temporary admission schemes (Honduras).

Overall Stance of Trade Regimes

Combining the stance of tariff and QR regimes provides an indicator of the overall trade policy stance; as mentioned earlier, if one of the components is restrictive, the overall trade regime is considered to be restrictive. On this basis, most of the program countries had restrictive regimes at the end of 1990, five had moderately restrictive trade regimes, and only one (Poland) maintained an open trade regime (Table 2).

During the period under review, the picture changed significantly. By the end of 1993, trade liberalization in 16 countries moved them up to the moderately restrictive category, while 4 even achieved open trade regimes. Only Poland increased its trade restrictions such as to switch from the open to the moderately restrictive category.

Significant trade reforming countries can be found in every region, but economies in transition and developing countries in the Western Hemisphere made the most visible progress. In most economies in transition, the overall trade regime had become less restrictive by 1993, as QRs were lifted following the deregulation of the economy and tariffs remained comparatively low (albeit increasing in some cases from their even lower levels in 1990). While many countries in the Western Hemisphere started with high tariffs and binding quantitative restrictions, they were able to reduce both, making the overall trade regime less restrictive than in 1990; notwithstanding this progress, only one (Ecuador) moved to the open category.

In Africa, much was achieved in phasing out quantitative restrictions but tariffs remained high, often because of revenue reasons; hence, overall trade regimes continued to be largely restrictive. In Asia and the Middle East, trade liberalization was gradual; notwithstanding progress in liberalization during 1991–93, program developing countries in Asia and the Middle East remained largely in the restrictive category.

An interesting question is whether countries that moved up into a less restrictive trade category faced initial conditions that were better than those that did not make such a move. A full-fledged exercise to investigate this issue would need to look into the range of macroeconomic and structural particularities of the slower and the faster reformers. Such a comprehensive review is beyond the scope of this paper. The illustrative exercise below provides information on the macroeconomic indicators associated with slower and faster reformers without establishing causal links. Of particular interest is whether the macroeconomic situation of the faster trade reformers was noticeably better in the initial period compared with that of the slower reformers; and whether faster reformers faced a deterioration in their macroeconomic situation relative to the slower reformers.

The identification of slow and fast reformers is based strictly on the classification of trade regimes in Table 2. Fast reformers are defined as those that liberalized their trade regimes sufficiently to move the combined (QR plus tariff) trade regime from the restrictive category to (at least) one category above, between 1990 and 1993.14 Slow reformers are defined as those where both the QR and the tariff regimes remained in the restrictive category in both 1990 and 1993.15

Selected macroeconomic indicators (covering external current account and fiscal balances, gross reserves, inflation, and trade taxes as a percent of total tax revenue) are presented for the group of slow and fast reformers (as defined above) in Table 3. It appears that in the initial period, slow reformers (as a group) faced larger fiscal and current account deficits than fast reformers, a lower level of gross reserves and, of particular note, greater reliance on trade taxes for fiscal revenues. Only with respect to inflation was their initial position more favorable than that of the fast reformers—the latter group included many Latin American countries that were experiencing triple-digit inflation in 1990. Overall, slow reformers faced more difficult initial macroeconomic conditions relative to fast reformers.

Table 3.

Macroeconomic Indicators in Selected Program Countries

article image
Sources: National sources: and IMF staff estimates.

Based on the classification in Table 2, these countries remained in the restrictive category for both QRs and tariffs.

Based on the classification in Table 2, these countries liberalized trade regimes such as to move up at least one category in the combined (QR and tariff) trade stance 10 the open or moderately restrictive category.

Over the period under review, the current account positions of the fast reformers deteriorated somewhat, whereas that of the slow reformers improved. However, fast reformers improved their gross reserve positions, in contrast to the slow reformers, where reserves remained unchanged (in terms of months of imports) in 1993 compared with 1990. A striking feature was that fast reformers were able to reduce considerably their fiscal deficits (as a proportion of GDP) at the same time as their reliance on trade taxes for fiscal revenue also declined significantly. In contrast, the fiscal positions of the slow reformers showed a marginal deterioration while their reliance on trade taxes declined somewhat. Inflation decelerated for both groups.

While causal links should not be drawn from the above exercise, and exogenous factors would certainly have played a role, the macroeconomic characteristics associated with the two groups suggest that fast reformers were willing or able to undertake both faster trade reform and fiscal reform, to contain current account deficits, and to attract capital inflows—an approach typically associated with bold, comprehensive macroeconomic-cum-structural reform packages. The slow reformers started off with more difficult initial macroeconomic conditions and were willing or able to proceed gradually with both trade and macroeconomic reform. Adverse initial macroeconomic conditions need not unduly slow down initiation of trade reform as long as the country is willing or able to adopt a broad-based package of macroeconomic and structural measures. In particular, complementary reforms of domestic tax systems will be necessary so as to reduce reliance on trade taxes. The examples of some fast reformers (Honduras, Jamaica, and Uruguay) show that dependence on trade taxes can be reduced significantly even within a relatively short (three-year) period.


The extent to which trade liberalization induces a lasting reorientation of production, consumption, and investment patterns depends in part on the sustained implementation of a credible reform program. This implies adherence to a clear and consistent direction of trade policy and to the preannounced targets, as well as a coherent program of complementary macroeconomic and structural policies that ensures that trade reform is sustainable. If governments partially or fully reverse earlier liberalization, the ripple effects (unless reversals prove to be very short-lived) could go beyond the additional protection provided through the new measures. The credibility and predictability of trade reform is jeopardized and further lobbying for additional protection is encouraged. The extent and nature of reversals in trade reform in the 59 program countries in the period under review is discussed below.

In about a fourth (13) of the program countries, initial trade reforms were partially reversed.16 All but 3 of these 13 countries increased tariffs or imposed new surcharges on imports, while 4 widened the scope of quantitative restrictions. But only in Poland were the reversals strong enough to induce a shift in the classification of the overall trade regime.17

The significance of trade policy reversals varied considerably across countries. For example, in the cases of Argentina (the statistical tax)18 and Algeria and Kenya (reimposition of QRs), reversals raised protection for that part of the trade regime by the end of 1993 to levels similar to those at the end of 1990, thereby largely offsetting the progress achieved in between. In most other countries (e.g., the Philippines, Sri Lanka, and Tanzania) the scope of reversals was more limited. This reflects the fact that, in nearly all of these countries, imports of specific goods or from specific sources were targeted discriminatorily, thereby limiting the average increase. Exceptions were the broader-based, non-discriminatory increases in tariffs and other charges in Argentina (statistical tax) and in Bulgaria and Poland (import surcharges).

The reasons for the reversals varied. For some, the rationale was to safeguard temporarily the balance of payments and fiscal revenues, in the face of lagging domestic tax reforms or appreciating real exchange rates. However, for most, protection motives played some role, as governments appeared to at least partially accommodate rising domestic pressures for protection. In some cases, external developments contributed to the reversals. For example, Burkina Faso experienced a large terms of trade shock that led to a substantial weakening of the fiscal and balance of payments position in the presence of a fixed exchange rate. In Kenya, a sharp drop in external financing weakened the balance of payments dramatically.

It is notable that many of the countries concerned had significantly liberalized their trade regimes just prior to the reversals but were unable to fully sustain the liberalization. While reversals occurred in program countries in nearly all regions, nearly half of all cases involved economies in transition. Many economies in transition (particularly in the former Soviet Union) had initially low tariffs as their trade relations were governed by QRs and state-trading arrangements. After having eliminated QRs rapidly and comprehensively, some of them faced increasing protectionist pressures or were unable to fully sustain open markets in light of inadequate complementary reforms in other (nontrade) areas. In some of these cases, reversals at least partly reflected the desire to support the balance of payments and the budget.

Collaboration with the World Bank

Trade policy is an integral part of both Fund and World Bank lending operations. Consultation and collaboration between the two staffs is particularly necessary and desirable in this area. The procedural requirements of SAFs and ESAFs (in terms of participation with national authorities in the formulation of policy framework papers) provide a vehicle for coordination of Fund and Bank staff advice on trade policies for low-income countries. But in other cases, too, cooperation in the trade area between the two staffs—at headquarters and in the field—reflects the crucial role played by sustained trade reforms in achieving macroeconomic and structural adjustment. Thus, trade reforms supported by Bank operations need to be consistent with the macroeconomic framework of Fund-supported programs; similarly, the success of trade liberalization measures initiated by the authorities in cooperation with the Fund is often also dependent on other structural reforms contained in Bank-supported programs.

The review indicates that collaboration in the trade area between Fund and Bank staffs has been close. In most cases, Fund staff have consulted Bank staff or have relied on them for the design of trade reforms incorporated in Fund-supported programs. In many instances, the design of Fund-supported adjustment programs has focused on liberalization of QRs and the prerequisite modifications of exchange systems, rather than on the design of tariff reform. Particularly where the latter required comprehensive, resource-intensive studies, Bank staff often took the lead. With the decline in reliance on QRs, Fund staff are paying greater attention to tariffs; in some recent cases, Fund-supported programs have included far-reaching tariff reforms as their initiation was considered necessary in the early stages of the program. In many cases of longer-run, ongoing reforms, where complex systems of both tariffs and QRs are being gradually phased out, the World Bank Group has supported trade reforms by International Bank for Reconstruction and Development (IBRD) or International Development Association (IDA) lending operations; Bank staff have consulted Fund staff on the fiscal and balance of payments effects of the envisaged trade reforms.

Trade Policy Conditionality

In order to investigate various aspects of conditionality on trade policy included in Fund-supported adjustment programs, arrangements were classified into three groups: (1) whether Fund disbursements under the program were made contingent on the implementation of specific trade measures; (2) whether the government’s commitment to trade reform was expressed in the form of a general or specific intent of policy but not tied to disbursements; (3) whether trade policy measures were implemented without recourse to the approach in (1) or (2). Below is a review of the nature, determinants, and implementation of conditionality on trade policy.19 The review indicates that the nature of conditionality depended mainly on the initial trade policy stance and the existence of ongoing unilateral (or “autonomous”) trade liberalization programs.

Over 85 percent of the 78 arrangements in this review envisaged some trade policy action ex ante (Table 4). The increased focus on trade reform in Fund-supported adjustment programs reflects the general recognition of the importance of structural adjustment as a contributor to sustainable growth. In the small proportion of cases where ex ante trade policy measures were not included as part of program design (most were stand-by arrangements), trade reforms did take place and were reported ex post.

Table 4.

Summary of Conditionality Applied

article image
Sources: National sources; and IMF staff estimates.Note: EFF = extended fund facility; ESAF = enhanced structural adjustment facility; SAF = structural adjustment facility.

In over two fifths of the arrangements, disbursements were linked to implementation of specific trade reforms. In another two fifths or so, governments committed themselves to undertake trade reforms without the use of binding conditionality. In nearly three fourths of all ESAFs, SAFs, and EFFs, trade reform measures were tied to Fund disbursements. Similar conditionality was applicable in only one fourth of the stand-by arrangements (this is, of course, not surprising given that ESAFs, SAFs, and EFFs by definition pay greater attention to structural elements).

A disaggregation of trade policy measures by type of conditionality shows that binding conditionality was more often applied to the liberalization of quantitative restrictions than to tariff reform (Table 5). This reflects the view that QRs are the more pernicious form of trade restriction, and their removal is often accorded higher priority (particularly at earlier stages of trade reform) than price-based restrictions. Implemention of reductions in quantitative restrictions was subject to specific conditionality in 40 cases compared with 33 cases related to tariff measures. Nonetheless, program design did pay considerable attention to price-based restrictions, albeit most often in the form of nonbinding government intentions. Where reform of the tariff structure and the system of QRs was considered especially crucial to the success of the structural adjustment component of the program, the measure was often implemented at the start of the program, in many cases as part of the central government’s budget. In many program developing countries, the authorities undertook trade reform measures in addition to those envisaged under Fund-supported adjustment programs. This explains why arrangements with developing countries accounted for about 70 percent (12 on tariffs, 15 on QRs) of those trade reform measures taken outside Fund programs, even though developing countries had only four arrangements that did not contain ex ante trade liberalization measures (Table 5).

Table 5.

Summary of trade Reform Measures by Type of Conditionality1

article image
Sources: National sources; and IMF staff estimates.

Single arrangements are counted several times if they involved several trade measures at different levels of conditionality.

As Tables 4 and 5 show, the conditionality applied to trade reforms varied across regions and arrangements. These differences were attributable to several factors, including the initial stance of the trade regime, the strength of the member’s ongoing unilateral trade liberalization program, and the track record on implementation of reforms. Arrangements with countries with initially restrictive trade regimes were more likely to incorporate specific and binding trade policy conditionality compared with those with open trade regimes. For example, countries with initially restrictive QR regimes accounted for 49 percent of all program developing countries, but their share in arrangements with binding trade conditionality was significantly higher (63 percent). Similarly, all but one of the countries with initially restrictive tariff regimes was subject to some form of binding trade conditionality. In some cases, trade reform was already sufficiently well covered in the context of Bank-supported lending. In other cases, developing countries initiated their own (and often ambitious) trade reforms, independent of the use of Fund resources, and were willing to continue with trade liberalization without outside persuasion. In still other cases, specific and binding conditionality on trade policy was often used to strengthen the hands of the authorities to carry out difficult reforms and to resist pressures from domestic special interests. In the economies in transition, it was recognized early on that trade reform would play a crucial role in the requisite radical transformation of their economies; hence, many of the arrangements with these countries contained binding conditionality, in addition to nonbinding trade policy measures intended to be taken by the authorities.

In two thirds of the arrangements, trade reforms were implemented as envisaged in the programs. In the remaining one third featuring specific and binding conditionality, there was nonobservance of the timing element of the envisaged trade policy measures; but this was generally not sufficiently serious to warrant disruption in disbursements under the arrangements. In some cases, delays in implementation were not a failure of policy, but rather reflected limitations in administrative capacity. In some instances, conditional trade policy measures were judged as having been implemented on the basis of available information at the time, but subsequent data revealed that they had been only partially implemented. This indicates the importance of formulating clear and easily monitorable measures.

Issues in the Design of Trade Reform

In formulating trade reforms, a number of design issues need to be addressed. For example, the extent and sequencing of trade reforms have to be determined by the authorities in the light of their economic necessity, political feasibility, and administrative capacity. The policy measures need to be prioritized consistent with the entire package of macroeconomic and structural measures; the prerequisites and complementary measures must be already in place or implemented along with the envisaged trade policy measures; and thought needs to be given on how to deal with the transitional costs of the measures (such as possible temporary rise in unemployment in affected industries) and their acceptance by the public through improved social safety nets. In countries where the administrative capacity is a constraint, this needs to be taken into account in the design of the reform, at the same time as efforts are made to improve such capacities through technical assistance. Within the trade policy area, the relative emphasis on the different elements needs to be determined. For example, should QRs be eliminated prior to tariff reforms or along with them? Which QRs are the most harmful? Should some sectors be liberalized before others? Should the program target average tariffs or only maximum and minimum rates? Should maximum tariffs on luxury items be left out of the reforms for income distribution reasons? Are tariffs in neighboring countries a reasonable target? Which trade measures should be subject to binding conditionality? How should the conditionality be defined?

These and similar questions often come up during program negotiations and monitoring of program implementation.20 Many of the issues have been addressed in varying degrees in past trade policy surveys, and in general and country papers prepared within and outside the Fund.21 Thus, there is consensus that the overall trade reform effort is best addressed in the context of a medium-term strategy with clearly established and announced immediate and medium-term objectives. Such objectives provide correct signals and help avoid uncertainty caused by frequent changes in the trade regime. Firm adherence to a timetable of trade and related liberalization establishes the credibility of the reform effort from the outset and ensures that the relative price structure is actually factored into the decisions of economic agents. Indeed, experience shows that the credibility and the completeness of a reform program are the most important factors for its success. Sustained liberalization of the trade regime has been most often attained when the timing and scope of reforms is preannounced, quantitative restrictions are removed at the outset of the program, tariff levels and dispersion are reduced, and complementary domestic price, tax, and public enterprise reforms are implemented.

Among other things, this paper reviews selected design issues emanating from the review of trade reforms in Fund-supported adjustment programs in 1990–93. The selected issues cover the interaction of trade reforms with the fiscal revenue aspects of programs, and with exchange rate policy; selected aspects in the formulation of QR liberalization; and tariff reform issues. To investigate these issues, the approach will now shift from a review of the population of programs in 1990–93 to a sample based on the review of the experience with trade liberalization in selected countries.22 Where relevant, examples from the experience of countries outside the sample are also provided to illustrate design issues.

An important lesson emanating from the review of design issues is that trade reform should be geared to medium-term efficiency goals. The use of trade policy instruments (tariffs, QRs) for fiscal, balance of payments, or income distribution reasons is a second-best solution, and should be resorted to only sparingly and temporarily to meet emergency situations where better alternatives are not available.

Interaction with Other Program Policies

Trade reforms in Fund-supported adjustment programs have formed part of a larger macroeconomic and structural adjustment effort to correct external and internal imbalances and to achieve sustainable growth. This effort often includes fiscal adjustment and exchange rate policy measures that, in turn, have important implications for the design and sequencing of trade reforms.

Fiscal Considerations

Trade taxes are not optimal instruments to raise revenues because they distort production and consumption. Domestic measures, such as lump-sum taxes, income taxes, or commodity taxes (excise, value-added tax (VAT), etc.), applied neutrally to domestically produced and imported goods should be the preferred instruments to raise revenue. In practice, however, governments rely heavily on trade taxes to generate revenue because administrative capacity constraints or insufficiently developed institutions do not permit efficient collection of nontrade taxes.

Generating or maintaining fiscal revenues from trade taxes has been a major determinant in shaping tariff policy in many Fund-supported adjustment programs, and particularly in the sample countries. Fiscal considerations often played a key role in limiting the magnitude of tariff reforms and the speed of their implementation. Indeed, it appears that fiscal objectives have overridden trade policy considerations in most cases where conflicts arose between the two. In many cases, limited progress in trade reforms is attributable to shortcomings in fiscal policy implementation. In some of the sample countries (e.g., Bangladesh, Sri Lanka, and Zimbabwe) where trade taxes account for a high proportion (e.g., 20–35 percent) of total tax revenue, fiscal considerations resulted in adoption of a “gradual” approach to tariff reform, and in some instances even to temporary tariff increases for revenue reasons. Since Fund-supported adjustment programs generally target the central government budget—and trade taxes tend to accrue to the central government alone—revenue considerations can drive tariff policy even in countries where trade taxes account for less than 10 percent of total tax revenue, as in Argentina. The crucial link between reform of the domestic tax system and tariff reduction is also illustrated by the experience of the sample countries. For example, Poland introduced an import surcharge and Sri Lanka delayed reduction of the maximum tariff, in part due to the significant public sector deficits and difficulties in timely implementation of planned alternative revenue instruments.

The lesson from this experience is that program design needs to explicitly recognize the linkage between fiscal and trade policy. Chances that tariff reforms are implemented—and sustained—can be improved by restructuring the central government’s budget early in the reform process with a view to decreasing the relative importance of trade taxes in total revenue. In some cases, this might imply expanding or introducing new instruments, such as mass consumption taxes, particularly where structural adjustment programs aim to reduce already high income and excise taxes. In other cases, this might require increased attention to the establishment of efficient tax administration.

If program design makes tariff reforms contingent on the development of alternative revenue sources, a clear and binding timetable for implementation of the latter is desirable. Corrective domestic revenue and expenditure measures taken early on in the program period minimize the risk of slippage in tariff reforms. Furthermore, contingency revenue measures outside the trade area should be formulated for use in the event that revenue gaps were to emerge. The particular measures would, of course, need to be geared to the individual country context. The fact that many of the fast trade reformers in Latin America have not resorted to tariff increases in spite of budgetary problems might serve as an indication that such revenue measures can be found outside the trade area, provided the necessary fiscal administrative capacity is developed.

This is not to deny that there could arise instances where the viability of the overall program could be jeopardized by slippages in the fiscal area, and that it may be infeasible to meet this emergency situation in the short run without resort to trade measures such as import surcharges. In such cases, it would be important to ensure that the deviation from medium-term trade reform objectives is strictly temporary by including a preannounced and binding timetable for a phased elimination of the surcharge. Adherence to the programmed elimination of the import surcharge would be convincing proof of the authorities’ medium-term reform path and that the temporary deviation was a measure of last resort. The frequent, and sometimes prolonged, use of import surcharges in a number of program countries unfortunately suggests that there continues to be too much of a tendency to view this instrument as the first line of defense to contain budgetary deficits in the face of revenue shortfalls or expenditure overruns. Such a perception may itself prove to be a serious impediment to bringing about the domestic political consensus for implementing the necessary fundamental restructuring of the domestic tax base and of expenditures.

It is important to emphasize that a well-designed trade reform need not be inimical to revenue generation. The substitution of tariffs for binding QRs, for example, serves the objectives of transparency and efficiency at the same time as revenues are enhanced (examples are Latvia and the Philippines). In many program countries, tax collection rates are well below statutory tariffs, partly reflecting widespread use of exemptions.23 In addition, very high tariffs often encourage smuggling, tax evasion, misclassification, corruption, and so on, all of which act to deprive the government of the intended revenue. A well-designed tariff reform would aim to reduce the scope of discretionary exemptions; this in turn, directly improves revenue. The choice of the particular method of reducing tariffs (collapsing the top rate to the next-highest level, or proportional reductions in all rates) also influences revenue. Indeed, it is feasible in some cases that a well-designed trade reform has a “Laffer” effect so that as statutory rates come down, the import tax collection rate increases.24 Finally, in a number of comprehensive tariff reforms, minimum rates have been increased at the same time as maximum rates are brought down, in order to both reduce dispersion and safeguard revenues. While many countries have room to raise minimum rates, care must be taken that these are not raised too much (double-digit levels should be avoided), as this not only increases the cost of imported inputs but may also prove a disincentive to exports in the absence of compensatory schemes, such as well-functioning duty drawback systems, temporary admission schemes, or export-processing zones.

Exchange rates and Exchange Restrictions

Appropriate exchange rate policies are essential to ensure that trade reforms are consistent with balance of payments objectives. Such policies, supported by firm macroeconomic policies, reduce the need for maintaining restrictions to ration foreign exchange, and also stimulate exports. The effectiveness of exchange rate policies, in turn, is enhanced when complemented by the liberalization of trade and payments regimes.

In a number of countries, trade restrictions have been used to facilitate the administration of restrictive foreign exchange allocation systems and other exchange restrictions. Indeed, exchange and trade restrictions often act as substitutes, and the benefits of liberalizing one may not be fully realized in the absence of liberalization of the other. Effective trade liberalization needs as a prerequisite (or at least as complementary measures) adoption of liberal exchange systems. Part of the success of Fund-supported adjustment programs in reducing reliance on QRs is attributable to liberalization of exchange restrictions. The experience of the sample countries indicates that in Ethiopia and Zimbabwe, for example, reliance on quantitative import restrictions resulted from disequilibria in the foreign exchange market and the system of administrative allocation of foreign exchange. Progress in removing quantitative import restrictions depended on exchange reforms that helped alleviate foreign exchange shortages.

Where trade restrictions are pervasive, providing high levels of protection across significant parts of the economy, the equilibrium exchange rate is likely to be well above the level that would have prevailed in the absence of restrictions. Under these circumstances, trade liberalization may lead to a deterioration in the current account, as the boost to imports (including private consumer demand) is likely to be felt more rapidly than the impact on exports. In the absence of adequate reserves, the trade liberalization would normally need to be accompanied by a devaluation to safeguard the balance of payments—the extent of the exchange rate change would depend on the tightness of fiscal and monetary policies. If the trade reform improved confidence so as to generate capital inflows in sufficient amounts to offset the deterioration in the current account, a devaluation may not be necessary, at least initially, for immediate balance of payments reasons. However, as domestic industries will generally need time to adjust to the new realities after a major trade reform, a devaluation might still be necessary to avert future balance of payments problems. Where the initial trade restrictions are not pervasive, or the trade liberalization is confined to specific input sectors, a devaluation may not be necessary, provided that the improvement in competitiveness deriving from cheaper inputs is translated sufficiently rapidly into higher exports.

Even if quantitative import restrictions are the binding constraint at the beginning of the trade reform, tariff policy needs to be coordinated with progress in exchange rate reform. High tariffs may not even provide much protection if duties are calculated on the basis of grossly overvalued exchange rates, as occurred in Ethiopia, for example. With devaluation of the exchange rate, maintenance of high tariffs may unduly compress imports. In Ethiopia, the failure to reduce tariffs with the devaluation in 1992 led to sharp increases in import costs; as an interim measure, import duties were calculated on the basis of the old exchange rate, but this jeopardized the fiscal revenue targets.

Pressures for protection are likely to rise in countries experiencing a real exchange rate appreciation due to surges in capital inflows, or where the nominal exchange rate serves as an anchor but inflation remains higher than programmed. Ideally, the real appreciation would be tackled with appropriate monetary and fiscal policies.25 In practice, as pressures for protection from domestic producers rise, and trade deficits widen, across-the-board import surcharges, taxes on imports, or other trade restrictions are sometimes used to approximate a real depreciation on the import side. Among the sample countries, Argentina and Poland illustrate this experience. Across-the-board import surcharges or taxes for balance of payments reasons avoid interindustry distortions in the import-substituting sector but penalize exports. Such measures need to be confined to emergency situations and be kept strictly temporary.

Tariff Policy Design

As QRs are dismantled, the focus of trade reforms is turning increasingly to tariff policy, including the appropriate levels and dispersion of tariffs, the relative merits of different averaging techniques from the view-point of monitoring, and the issue of sequencing. Below are some considerations in the design of tariff reforms, based in part on a review of the different approaches to such reforms pursued in the sample countries.

Monitorable Targets

A typical consideration in undertaking tariff reform is to reduce effective protection for final goods, which in many developing countries tends to be high owing to the cascading structure of nominal tariff rates. The concept of effective protection, however, is less useful as a monitorable target; it is not easy to measure as its computation is information intensive, data on input-output coefficients being required for the entire range of products in a tariff schedule. Hence, for all practical purposes, monitorable targets for reform need to be formulated in terms of nominal tariffs. This has typically been the case in Fund-supported adjustment programs in recent years. Some Fund-supported programs have targeted average nominal tariff rates (e.g., Brazil, India, and Moldova).

A key objective of tariff reform is to reduce dispersion by reducing the spread in nominal tariff rates, at the same time as tariff levels are brought down. However, a given average target is consistent with different combinations of the spread of nominal tariffs, and hence undesirably large spreads cannot be ruled out through reliance on a target for the average. The problem is addressed most directly by setting targets for maximum tariffs and the number and level of tariff bands. These targets would naturally also provide bounds for the average tariff level, which would be determined in the process of allocating products to the different bands in the tariff schedule. Under the circumstances, a target for the average tariff could serve as a supplemental guide in designing the program, to help ensure that product allocation among different bands is not skewed in favor of the higher tariff bands.

The average tariff level might be measured in three different ways: the simple average of statutory rates; the import-weighted average of statutory rates; and the average tariff collection rate (which is by definition an import-weighted measure). All these measures contain useful information, but are not equally easily monitorable or equally revealing about the stance of protection. For example, the average collection rate depends not only on the tariff rate, but also on the nature and magnitude of exemptions granted, on preferential trade arrangements, and on other instruments such as variable levies, antidumping and countervailing duties, etc., which serve different policy objectives. A low collection rate could thus reflect high de jure exemptions, or an inefficient and dishonest customs administration if there are de facto exemptions (e.g., tax evasion).26 The average collection rate is, therefore, not directly indicative of the stance of protection.27

Statutory tariff rates should therefore be the preferred measure. The import-weighted measure may underestimate the extent of protection because products facing high tariffs could enter the calculation with low weights, so that a low import-weighted average statutory rate may itself be the result of high protection.28 Thus, it would seem preferable to measure the extent of tariff protection in terms of the simple average of statutory tariff rates, notwithstanding the usual drawback in averaging associated with outliers. If the import-weighted average is to be used, the base year should be fixed on a recent year. Finally, tariffs need to be measured inclusive of all other charges (e.g., fees, statistical taxes, surcharges, etc.) so that targeted reductions capture amalgamated rates.

Tariff Levels and Dispersion

For most developing countries that cannot affect the foreign currency prices of their imports, theoretical considerations call for zero tariffs across the board. However, the optimal tariff structure is nonzero and differentiated if governments want to pursue objectives other than pure welfare maximization.29 Under these circumstances, the theoretically optimal tariff structure should be differentiated according to the price elasticity of demand for revenue objectives; according to the stage of processing under a protection objective; and according to the income elasticity of demand under an income distribution objective. Only under a balance of payments objective would theory call for a uniform tariff structure. With regard to the protection objective, however, the difficulties and pitfalls of targeting would suggest that it may be preferable to adopt the alternative approach of broader-based protection with relatively narrow differentiation among sectors.

With regard to other objectives, considerations of political economy, administrative convenience, and lack of information also provide strong arguments against complex and differentiated tariff structures. If interest groups perceive that the authorities are unwilling to provide too much differentiation in protection, they may refrain from further lobbying to secure greater protection, thereby minimizing rent-seeking costs. Less complex tariff structures can be administered more easily, avoiding cumbersome paperwork, and reducing the incentive to misclassify products. Furthermore, inadequate information about the relevant economic variables suggests that simply designed tariff structures are the preferred option. Simplicity is served by avoiding too many bands; while there is no firm theoretical basis for precisely choosing the number of bands, rules of thumb based on the experience of successful reformers suggest that a few bands (in the range of 3–5) would be appropriate.30

Tariff structures vary considerably among countries (Table 6). A question frequently encountered relates to the choice of targets for tariff levels and dispersion in the short- to medium-term. The experience of the relatively successful Latin American tariff reformers is particularly illustrative in this respect. It suggests that rapid reform—over a period of 2–5 years—is economically and politically feasible. Many Latin American countries had very restrictive tariff regimes in the 1980s but were able to modify their regimes rapidly in the late 1980s and early 1990s. While there are exceptions, the broad pattern that emerges from the experience of successful reformers is that they were able to bring down maximum tariffs from very high (sometimes triple-digit) levels to a range of 30–35 percent within three years or so, and subsequently brought these down further to around 20 percent. Several Latin American program31 countries considered in this paper (e.g., Costa Rica, Ecuador, Guatemala, Peru, Uruguay, and Argentina prior to the increase in its statistical tax) lowered tariffs generally to the 0–25 percent range, usually with average statutory tariffs between 10 percent and 15 percent.32 They also simplified their tariff structures to sharply reduce the number of tariff bands, in many cases retaining only a few (1-5). Similarly, several other (nonprogram) Latin American countries (e.g., Bolivia, Chile, Colombia, Mexico, and Venezuela) also have achieved simple tariff structures and low or moderate tariff levels over a relatively short period of time.

Table 6.

Comparative Tariff Regimes in Selected Countries1

(In percent)

article image
article image
Sources: National sources; IMF Annual Report on Exchange Arrangements and Exchange restrictions. GATT; and World Bank.

Status based on information available in May 1994.

When two rates are given, the first refers to the normal maximum rate and the second refers to the maximum rate applicable to a few selected items. Rates shown without a slash (/) refer to the normal maximum rate. A plus (+) sign implies that the maximum rate can be higher for some items.

Simple average rates except for Ethiopia and South Africa where weighted average rates are given.

Includes charges levied exclusively or discriminatorily on imports.

Excluding zero rates.

There are 35 ad valorem rates but 200 ad valorem equivalents if specific and formula duties are taken into account.

In many East European countries (a group that could also be considered as fast trade reformers) average tariff levels are low or moderate, but tariff dispersion is larger than in most Latin American countries, with maximum tariffs (including surcharges) above 40 percent and minimum tariffs between zero and 8 percent. Agricultural imports carry, on average, significantly higher tariffs than industrial goods (in part to insulate domestic producers from the effects of subsidized agricultural exports of many industrial countries). Adopting a more gradual approach, a number of (nonprogram) Asian countries have also achieved (over a decade or so) low-to-moderate average tariff levels but, as in the case of Eastern Europe, tariff dispersion continues to be larger than in Latin America. For example, average tariff levels are 10–15 percent in Korea, Malaysia, and Thailand,33 but maximum tariffs are in the range of from 30 percent to 60 percent. A number of program Asian and Middle Eastern countries that have adopted a gradual approach (e.g., Bangladesh, Egypt, and Sri Lanka) or whose major tariff reforms have only recently been initiated (e.g., India, Pakistan, and the Philippines) had restrictive tariff regimes with high average and maximum tariffs at the end of 1993.

Most African countries have made only limited progress in reducing tariffs and maintain high average statutory rates (inclusive of other charges). A number of them have moderate average customs tariffs but apply significant other charges (e.g., Malawi, Mali, Morocco, South Africa, and Zimbabwe). Slow progress in tariff reforms has often been attributed to revenue constraints. However, some program countries in Africa and in other regions with similar constraints have made significantly greater progress in lowering tariffs (Benin or Nepal, for example).

Rapid progress in tariff reform is possible even under difficult initial situations, as shown by a comparison of the experience of Bangladesh and Ethiopia. Both countries had initially maximum tariffs well above 100 percent, faced vulnerable external positions, and relied heavily on trade taxes for fiscal revenues. Ethiopia undertook a major tariff reform in June 1993, lowering the maximum rate to 80 percent and greatly reducing the number of bands and zero-rated items; the average statutory tariff was reduced from 41 percent to 29 percent. In contrast, Bangladesh’s gradual tariff reform, initiated in 1986, yielded an average (unweighted) nominal rate of protection of 42 percent by 1993/94. In addition to fear of loss of revenues, the factors cited by the authorities for not progressing faster included concerns about the social and economic costs associated with restructuring domestic industries and the use of the trade regime to achieve a more equitable income distribution in the absence of alternative direct methods of resource transfers.

The extent and speed of future tariff reforms would naturally need to be tailored to the particular circumstances of each country. Nevertheless, the experience of tariff reform in the successful Latin American countries indicates what is feasible if the reform is designed well, accompanied by a package of complementary domestic measures, and if its value is explained to the public. For countries currently facing highly distorted trade regimes, tariff reform would probably need to proceed in stages within a medium-term framework; the less distorted cases could perhaps proceed faster to reach the level attained by Latin America. A medium-term tariff reform could consider the following elements: (1) all import taxes and charges should be incorporated into the tariff structure; (2) discretionary exemptions should be removed; (3) the tariff structure should be simplified into a few (preferably within 3–5) bands; (4) the all-inclusive maximum tariff should be significantly lowered with only a short list of exceptions; (5) this should be accompanied by a reduction in the average statutory tariff rate; (6) the tariff reform should be preannounced as a minimum in terms of its medium-term target, and preferably its annual targets too (like the multiyear tariff reforms in Brazil, Colombia, and Pakistan). Annual targets would need to be worked out with particular attention to the revenue and balance of payments impact of the reform. As mentioned earlier, care has to be taken to contain minimum tariffs to relatively low levels (preferably below double digits) in order to avoid raising costs of raw materials and intermediate inputs and to avoid hurting exports. Minimum tariffs at the higher end (e.g., around 10 percent) would require an effective duty drawback system to ameliorate the effects on exports (such drawback systems need to be kept administratively simple if they are to be effective).

The experience of Latin America suggests that it would be feasible in the first stage to bring down maximum tariffs to levels in the region of 30 percent to 35 percent and reduce average tariffs to a range of 15 percent to 20 percent. Countries that have already simplified and rationalized their tariff structures and reached moderate levels of tariffs could gear their medium-term tariff policy toward a further reduction of average tariffs (e.g., to about 10 percent) and of maximum tariffs (e.g., to about 20 percent with no exceptions).

In practice, some countries have proceeded much faster than the above scenarios, while others have failed to reach even moderate tariff levels despite many years of reform.

Implications of Regional Integration

A number of countries have increasingly liberalized trade in the context of regional arrangements. This may have implications for their external trade regimes (i.e., vis-à-vis third-country trading partners). Among the sample group of countries, for example, Argentina and Poland have entered into regional arrangements with some of their major trading partners, aimed at establishing free trade among regional members. Regional trade liberalization may affect tariff policy in a number of ways. Members of a customs union will be bound by the common external tariff, and liberalization with respect to third countries will need to be agreed upon among all the partner countries. Even under a free trade area without a common external tariff, the room for further liberalization on a most-favored-nation (MFN) basis may be limited in the face of revenue constraints and domestic pressures for protection.34 In fact, tariff increases for protection or revenue purposes might have to be larger vis-à-vis third countries as it may not be possible to raise tariffs on imports from members of the regional arrangement.

The above factors need not, however, inhibit further MFN trade liberalization, as exemplified by recent developments in the Caribbean Community (CARICOM) and the Central American Common Market, if regional partners are convinced of the beneficial effects of the reform for their countries. In the design of tariff reforms, regional perspectives need to be taken into account in the case of countries belonging to a regional trading arrangement.

Quantitative Restrictions

In the sample countries reviewed, QRs were generally maintained for one of two reasons: to ensure the administrative allocation of foreign exchange and to protect domestic industries. It is generally accepted that QRs are less transparent and more restrictive than tariffs. Hence, they should in principle be liberalized at an early stage of trade reform. A number of countries have considered a one-step removal infeasible and have opted for a phased reduction. Numerical targets for QR reduction need to be precisely defined and monitorable. This is especially important where progress is to be monitored through coverage ratios (such as proportion of imports or domestic value added). Care must be taken to avoid a bunching of sensitive items for liberalization at the end of the period, as this is often not credible. The exceptions for liberalization must also be precisely defined and limited to a short list for GATT-sanctioned reasons such as health, national security, or public morals.

An emerging trend in some program (and nonprogram) countries that is disquieting is that as QRs are liberalized (and high tariffs brought down), new forms of protection are introduced, such as variable import levies or increased recourse to antidumping measures; this tends to occur especially in the agricultural sector, partially offsetting the impact of the QR liberalization. Faster progress on liberalization can sometimes be obtained if the QRs are replaced initially by tariff equivalents, which are subsequently reduced according to a predetermined schedule.

In countries where QRs are the result of the foreign exchange allocation process, their removal is intimately tied to reforms of the exchange system. Frequently used strategies include the introduction of an auction market for foreign exchange with a gradually declining negative list (for which the foreign exchange must not be used); gradually expanding an export retention scheme (ERS) by shortening the negative list and increasing the retention ratio; or gradually adding more products to an open general import license system (OGIL). Especially in an early stage of the reform, when the OGIL is rather limited, the ERS has the advantage of limiting discretionary selection of imports for liberalization; under the ERS, importers determine what they want to import from retained export earnings (although there is usually a negative list). However, if the retained export earnings are nonsalable or the foreign exchange market is relatively thin, the ERS is of little help to importers who do not have their own sources of foreign exchange earnings.

GATT Consistency

The design of trade reforms needs to be consistent with the member’s obligations under the GATT (and in future with the new World Trade Organization). An area meriting particular attention is to ensure that GATT tariff “bindings” are not violated.35 This could potentially occur, for example, if increases in minimum tariffs are contemplated. If the tariff increases violate GATT bindings, the country needs to obtain a temporary waiver of its obligations under the GATT, giving it time to consult with trading partners about appropriate compensation (the latter can be provided, for example, by reducing bound rates on other items of interest to its trading partners). In practice, during the review period (1990-93), GATT consistency has not become an issue in the design of Fund-supported adjustment programs. An exception was the case of Egypt.36 Since this occurrence, World Bank and Fund staff have heightened their awareness of potential GATT-inconsistent measures arising in connection with reform programs and have strengthened their respective institutional review mechanisms so as to better identify such potential problems at an early stage. With the broader reach of trade obligations undertaken in the Uruguay Round agreement, including a general increase in the level of bindings, the need to remain alert to this issue is greater than before. Ultimately, national authorities bear the responsibility to ensure that their adoption of policy measures is consistent with their treaty obligations.

Other potential areas meriting attention include the GATT’s prohibition on import surcharges on bound items (unless the relevant GATT waiver is obtained), on export subsidies on manufactures, and on trade restrictions for environmental purposes. It is also important to avoid trade remedy laws and practices (antidumping and countervailing) that are inconsistent with GATT rules.


The main conclusions are listed below.

• Significant progress was made toward trade reform under Fund-supported adjustment programs. Between 1990 and 1993, out of 59 program countries, those with restrictive trade regimes fell from 53 to 33.

• Greater progress was made in liberalizating quantitative restrictions compared with tariff reforms, but the latter are beginning to get increasing attention under Fund-supported adjustment programs. The slower progress on tariffs was due in part to continued reliance on international trade taxes as a source of budgetary revenues.

• The greatest progress in trade reform took place in East European economies in transition and in Latin America.

• The macroeconomic characteristics associated with slow and fast trade reformers suggest that the former faced initial conditions that were more difficult than those of the latter. But fast trade reformers also made faster progress in the macroeconomic (fiscal) area, suggesting probably more willingness or ability to take bolder, comprehensive reforms.

• Reversals in trade reform (partial or full) were limited (13 cases out of 59) and often reflected competitive pressures due to appreciating real exchange rates, lagging domestic tax reforms, or political difficulty in resisting domestic pressures for protection.

• Over two fifths of the arrangements contained specific and binding conditionality on trade policy, while another two fifths contained nonbinding commitments or intentions on trade reforms. Specific and binding conditionality was more often applied to liberalizing QRs than to tariff reforms. The incidence of such conditionality depended on a number of factors including the degree of restrictiveness of the initial trade regime and the presence of ongoing unilateral trade liberalization programs being pursued by the authorities independent of Fund programs.

The lessons from the review of design issues indicate the following:

• Trade policy must be geared toward medium-term efficiency goals. The use of trade instruments for non-trade objectives is a second-best solution and should be resorted to only sparingly and temporarily to meet emergency situations where better alternatives are not available. The phased elimination of such temporary restrictions should be made a binding condition in programs. If trade restrictions are inevitable, across-the-board import surcharges are the least distortive measure, but they do penalize exports in the absence of efficient compensatory schemes.

• There have often been conflicts between short-term fiscal objectives and medium-term trade reform goals. However, a well-designed tariff reform may actually improve the import tax collection rate at the same time as statutory tariff levels come down, at least in the initial stages. Tariff reforms will be sustainable if they are accompanied by a restructuring of the domestic tax base early in the reform process so as to decrease reliance on trade taxes.

• Sustainable trade reforms need as a prerequisite—or at least as complementary measures—liberalization of exchange systems and exchange rate flexibility.

• It is essential to devise simple and straightforward monitoring devices on trade policy, and the measures should be consistent with the GATT.

• The extent and speed of trade reform needs, of course, to be tailored to the individual circumstances of the program country. Nevertheless, the experiences of the successful trade reformers, particularly in Latin America, offer lessons for the design of tariff reforms. Trade reform would involve replacing QRs with tariffs and reducing the latter in phases. Elements of a medium-term tariff reform, preferably preannounced, could be to amalgamate various charges into the tariff structure, eliminate exemptions, simplify the tariff structure by reducing the number of tariff bands to a few (e.g., 3–5), reduce average tariffs to moderate levels, and bring down maximum tariff’s significantly. Countries that have already simplified tariff structures and achieved moderate tariff levels could gear their medium-term tariff policy toward further reductions in maximum and average tariffs. The experience of Latin America suggests that it is feasible in the first stage to reduce maximum tariffs to a range of between 30 percent and 35 percent, and average tariffs to a range of between 15 percent and 20 percent. And in the subsequent stage, it is feasible to reduce maximum and average tariffs further to 20 percent and about 10 percent, respectively. For countries with high tariff protection (mainly in Asia, Africa, and the Middle East) these scenarios imply reaching the current tariff levels of Latin America by the turn of the century.


  • General Agreement on Tariffs and Trade, Trade Policy Review Mechanism (Geneva: GATT Secretariat, various issues).

  • International Monetary Fund, Annual Report on Exchange Arrangements and Exchange Restrictions (Washington: International Monetary Fund, various issues).

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, International Financial Statistics (Washington: International Monetary Fund, various issues).

  • International Monetary Fund, Trade Policy Reform in the Countries of the Former Soviet Union, IMF Economic Reviews, No. 2 (Washington: International Monetary Fund, 1994).

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, Transactions of the Fund (Washington: International Monetary Fund, various issues).

  • Kelly, M., A. McGuirk, and others, Issues and Developments in International Trade Policy, World Economic and Financial Surveys (Washington: International Monetary Fund, 1992).

    • Search Google Scholar
    • Export Citation
  • Loser, C. and E. Kalter, eds., Mexico: The Strategy to Achieve Sustained Economic Growth, IMF Occasional Paper, No. 99 (Washington: International Monetary Fund, September 1992).

    • Search Google Scholar
    • Export Citation
  • Pritchett, L., and G. Sethi, “Tariff Rates, Tariff Revenue, and Tariff Reform: Some New Facts,” World Bank Economic Review, Vol. 8, (January), pp. 116.

    • Search Google Scholar
    • Export Citation
  • Schadler, S. and others, Recent Experiences with Surges in Capital Inflows, IMF Occasional Paper, No. 108 (Washington: International Monetary Fund, December 1993).

    • Search Google Scholar
    • Export Citation
  • Subramanian, A., A. Ibrahim, and L. Torres-Castro, “Optimal Tariffs: Theory and Practice,” IMF Working Paper, WP/93/50 (Washington: International Monetary Fund, June 1993).

    • Search Google Scholar
    • Export Citation
  • Thomas, V., J. Nash, and others, Best Practices in Trade Policy Reform (Washington: Oxford University Press for the World Bank, 1991).


The principal authors of this paper.


For a review of trade reforms in Fund-supported programs in the second half of the 1980s, see Kelly, McGuirk, and others (1992).


References to “program” countries in this paper pertain to countries listed in Table 1.


Ideally an assessment of tariff regimes would include average tariff levels as well as dispersion. However, in view of the difficulties of measuring the latter across the population of countries this study relies only on the average statutory tariff.


Including other import taxes, such as surcharges, statistical taxes, fiscal duties, stamp taxes, service fees, or consumption taxes levied only on imports.


For a discussion of various concepts to measure nontariff barriers, see Kelly, McGuirk, and others (1992). The commitments to tariffication of QRs under the agricultural agreement in the Uruguay Round have helped to focus on the importance of developing a common methodology to measure such barriers.


QRs considered here are import and export prohibitions, quotas, licensing, discretionary foreign exchange allocation, and state-trading monopolies. Other nontariff measures, such as standards, minimum import prices, or antidumping duties, are not considered in the classification because information on a comparable basis for the 59 countries is not available. While the excluded measures may in some cases operate as important barriers to trade, it is unlikely that this has generally affected the classification for most of the countries reviewed.


Weighing by import values may underestimate protection levels as the import levels may themselves be low because of the restrictive impact of prohibitions and quotas.


At the beginning of 1994, Zimbabwe introduced a number of reforms in its exchange and payments system, including the retention of export proceeds in foreign currency accounts, and significantly reduced the coverage of its negative list for imports.


In connection with the CFA franc devaluation, all CFA franc zone countries reduced tariffs considerably in early 1994.


For a description of Mexico’s trade liberalization, see Loser and Kalter (1992).


In some of these cases import duties were reduced in 1994.


Economies in transition were excluded from this illustrative exercise as most of them engaged in far-reaching changes in their structural and macroeconomic policies in order to transform from centrally planned to market economies. They are thus less comparable with developing countries, which were not engaged in such fundamental transformation.


As mentioned earlier, the classification of trade regimes in this paper does not fully take into account the degree of reform effort. Measuring the relative speed of the trade reform would require deriving a single cardinal indicator of trade reform in each country in the population—an exercise beyond the scope of this paper. The identification above of “slow” and “fast” reformers must be viewed against these limitations.


In some cases, new liberalization measures were adopted or planned for 1994 or later, but these are not considered here as they fall outside the period under review.


This shift was from the open to the moderately restrictive trade category; the reversal has to be viewed against the initial sharp liberalization in 1990, which lowered average tariff levels to 5.5 percent, as part of a major transformation to a market economy.


The Argentine authorities have announced that the statistical tax on imports will be eliminated on January 1, 1995.


All Fund-supported programs contain a standard “standstill” trade clause that is binding, namely, the avoidance of introduction of new, or intensification of existing restrictions on imports for balance of payments purposes. The discussion here goes beyond the standstill provisions to investigate trade liberalization measures.


This paper does not deal with design issues that could arise from tied aid and donor procurement requirements that may result in some complexities and inefficiencies in import sourcing; this issue could be important in the design of trade reform in some countries, for example, in Africa.


The sample countries chosen were Argentina, Bangladesh, Ethiopia, Poland, Sri Lanka, and Zimbabwe. They represent active trade reformers in different regions that faced a variety of trade policy issues as their trade regimes underwent significant changes in the 1990s.


In Egypt, exemptions represented nearly 49 percent of potential tariff revenue in 1986; in Bangladesh, the average statutory tariff rate in 1992 was about 120 percent; whereas the average level of duty collected was less than 40 percent mostly due to exemptions; and for the same reason, in Kenya, in 1991, import duty collections as a proportion of import value were less than a third of the average import tariff.


For example, simulations show that revenues would actually increase or remain stable if maximum rates are lowered in Kenya and Pakistan, respectively. In addition, collection rates are generally much higher for lower tariffs than for higher tariffs, so that a combination of a small increase in the minimum and large reduction in the maximum might be revenue neutral. See Pritchett and Sethi (1994).


For a discussion of real appreciation following capital inflows, see Schadler, and others (1993).


A low collection rate could thus call for removal of exemptions and for improving customs administration, rather than raising statutory tariff levels.


The stance of protection should be measured by the marginal price that an importer would have to pay. If there are exemptions, with prohibitions on resale, the marginal price would be the nominal rate rather than the collected rate.


The average tariff would preferably be calculated by weighting the statutory tariffs on a product by its share in total domestic production. However, due to data limitations in most developing countries, this is not possible to calculate for many of the program countries.


Tariffs could be used, for example, for revenues, income distribution, or balance of payments management.


The above issues are covered in more detail in Subramanian, Ibrahim, and Torres-Castro (1993)


“Program” countries refer to those identified in Table 1 as having entered into a stand-by, EFF, SAF, or ESAF arrangement with the Fund in 1990–93.


Generally, raw materials and capital goods not available domestically mostly carry low tariffs (0–5 percent), while semifinished goods and finished goods are subjected to tariffs of 10–15 percent and 20–25 percent, respectively. Some sensitive items carry maximum tariffs of 30–35 percent.


Hong Kong has no import tariffs, and in Singapore 91 percent of the tariff lines are duty free.


Further MFN liberalization might be less difficult if the share of a country’s trade within the regional arrangement is very large. However, for most program developing countries this is not yet the case.


A tariff is “bound” in the GATT when a member undertakes not to increase it above its bound level without compensation to its GATT trading partners; a “binding” of the statutory import tariff involves also binding various other charges and duties levied on the particular import item.


Under the World Bank’s structural adjustment loan with Egypt, the authorities undertook to raise minimum tariffs to 5 percent, to partially offset revenue losses from the reduction of peak tariffs and to reduce dispersion. This tariff reform was also incorporated in the Fund-supported program.