POLICY RESEARCH WORKING PAPER 1218 Implementation < * of Tradea ReformS*cs w* in Sub-Saharan Africa How Much Heat - and How Much Light? ii.e:.ai Mea. ni,... John Nash b st1t r ( : ^mesr he :otunu ,., ... ..w. ... ¢. ..... tWIftac*e,can f'n I: . I I. ....& ' : .. ,'' The World Banklc:* th, nQn4 ...,s Policy Research Department ~~~ ~ 0ut?i i ip~eiefl~1 Trade Poky Division ~ ~ ~ ~ ~ ~ ~ tae ~iy fiddft November 1993~ ~ ~ ~ ~~~~~~~~untledintrm Polic ReerhDprnt : ' , , p:... .,',,''' '," ." .'"' Trd Polc Diiin ,,,,, , ,,,,, POLICY RESEARCH WORKING PAPER 1218 Summary findings Adjustment programs in Sub-Saharan Africa have been Using a new method for quantifying nontariff somewhat less intensive in trade reform than programs protection in terms of 'tariff equivalence," Nash in other countries have been. Implementation of trade argues that, in gen.eral, countries are not in danger of reform overall, however (but not the most important "de-industrialization" from the rapid disprotection of reforms), has been better in Sub-Saharan Africn. import-substituting industry. However, franc-zone Retrogression has also been more frequent. countries showed greater declines in incentives for As a group, the intensive adjustment lending import substitution because of their lower rate of real countries made significant progress in the 1980s and devaluation. One implication may be that their ability early 1990s, but there was significant variation a,nong to reduce tariffs and nontariff barriers is impeded by them. For Sub-Saharan Africa, progress has been more their inability to offset them with devaluations as other impressive in recent years than in earlier years. In countries did. Non-franc-zone countries reduced tariff- many countries, adjustment did not begin until the equivalent protection in recent years by 15 to 49 mid-1980s and relatively tew measures were percentage points more than franc-zone countries, implemented up front. For the franc-zone countries, while incentives declined by 15 to 20 percentage points underimplementation rates are lower in the most more in the franc-zone countries. recent data, and by some - but not all - measures How open are the trade regimes at this point? The their openness has improved more in recent years. By decline in tariff-equivalent protection, although not virtually all measures, however, improvements over trivial, is insufficient to reduce the protection to earlier periods have not been as great for non-franc- moderate levels reiative to deep reformers in East Asia zone countries. and Latin America. Reduced protection was largely offset by real The biggest pi oblem is with foreign exchange devaluation in most country groups and, by most allocation. Mauri,us may be the only non-franc zone meastures, incentives to produce import substitutes Sub-Saharan country in which the currency is actually improved in the years immediately after the essentially convertible and has been for some time. first adjustmert loan. In more recent years, the This basic reform has not begun in most countries or incentives have fallen modestly. has only recently been completed (Ghana). This paper - a product of the UNDP/World Bank Trade Expansion Program and the Trade Policy Division, Policy Research Department - is part of larger efforts in the department to focus on adjustment problems in Africa and to develop quantitative measures of economic policy. Copies of the paper are availablle free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Dawn Ballantyne, room N1O-023, e:.tension 37947 (44 pages). November 1993. The Policy Research Working Paper Series disseminates the findings of uvork in progress to encourage the exchange of ideas about development issue. An objective of the series is to get the findings out quickly, even if thc presentations are kss than fully polished The papers carry the names of the authors and should be used and cited accordingly The fir, ings, interpretations, and conclusions are the authors' ouwn and should not be attributed to the World Bank, its Executive Board of. irectors, or any of its member countries. Produced by the Policy Research Dissemination Center Trade Policy Reform Implementation in Sub-Saharan Africa: How Much Heat and How Much Light? John Nash (PRDTP, The World Bank) This paper is a product of the UNDP/World Bank Trade Expansion Program. The views expressed here are not necessarily those of the World Bank, UNDP, or affiliated organizations. I am grateful for comments on earlier drafts from Elliot Berg, Christine Jones, and Miguel Kiguel. Some of the country-specific information was provided by authors of background papers on this project, including Gabriel Castillo, Faezeh Foroutan, Eric Fournairon, and Coulibaly Kalamogo. Francis Ng provided excellent research assistance. Trade Policy Reform Implementation in Sub-Saharan Africa: How Much Heat and How Much Light? 1. Introduction There is a widespread perception that the performance of the tradable goods sectors in miost Sub Saharan (SSA) countries in the 1980's has been disappointing. After more than a decade of adjustment, in some cases, export sectors in most of these countries remain dependent on a few primary commnodities, whose prices are currently very low by historical standuards. Import substitute producing sectors also remain weak. There is some talk about the "de-industrialization" of Africa, implying that the situation of the industrial sector, at least, may have deteriorated in some countries, though this is clearly not a universal phenomenon. Two competing explanations have been advanced for the putative poor response to adjustment of the tradable goods sectors. One is that in reality there has been very little adjustment in policy because of a lack of commitment on the part of African governments, at least with respect to trade policy, in spite of the fact that virtually all SSA countries have adopted "adjustment programs."' Another explanation is that in spite of having made significant progress in implementing trade policy reforms, the response has been limited by one or more of the following factors: inappropriateness of the reforms to the African context; poor design, pacing, or sequencing of the reforms; unsupportive exte;nal environment; or inconsistency of trade and other policies. In countries where trade policy reforms have not been implemented well, a number of factors may have played a role in the failure to do so. First, there may not be a very strong predisposition to trade reform among many policy-makers in the region. Policy-makers may remain unconvinced of the I. Oxford Analytica summed up this view in its December 15,- 1992 article on "Me Uruguay Round antd the Developing World': 'African countries have shown less enthusiasm for free trade (than other developing countries). Notable exceptions, however, are Gambia, Ghana, Ivory Coast, Kenya, Senegal, Zambia and Zaire. Pressure from creditors and do-ors, both bilaterally and through the Paris club, has again been an important motivation in prompting this change of heart. benefits of reform, perhaps because they do not perceive that the evidence indicates that reforms have led to successful outcomes in countries where they have been tried. This in turn, may interfere with the development of a feeling of "ownership" of the reform program by the local authorities, a factor which has been identified as important in determining the success of adjustment programs in many countries (World Bank, 1988, 1990, 1992). Second, the widely recogniz d problem of insufficient institutional cipacity may limit the ability of SSA countries to organize counterpart teams, to enter into a broadly-based policy dialogue, and to implement whatever measures are proposed. Finally, the political economy of trade reform in the SSA region, as well as the fact that the vast majority of the countries concerned are among the least-developed, may require a somewhat different approach to reform. An additional issue that may magnify the importance of all the above problems is that the trade regimes of many African countries in the early 1980s were among the most interventionist in the world, with strong elements of dirigisme, tight controls on foreign exchange and large overvaluation, and heavy direct taxation of export sectors. Th.is starting position may have required even stronger political decisions and better administrative capacity, and led to fears of heavier short-term costs than in other regions. All these factors may have made reformn more difficult. In light of these concerns, the purpose of this paper is to examine how trade policy reforms were designed and to what extent they have been implemented in SSA. The data on which the analysis is based is of three types, each with its own strengths and weaknesses (Box 1). The paper will try to synthesize information from these different sources of data and exploit their complementarity to get a clearer and more accurate picture of implementation experience than could be obtained from any one source alone. "Trade policy" here will be defined broadly to include measures whose major impact is to change the incentives for producing or consuming either tradable goods in general (versus nontradables) or some tradable goods relative to others. -2 - This paper comprises four sections including this introduction. Section II sets the stage by describing in general terms the initial conditions of the trade regimes in SSA before the adjustment programs began, to indicate what needed to be done to achieve a reduction in the anti-trade bias. Section III examines what trade policy actions were proposed in the adjustment programs, the degree to which the :roposals were implemented, and the effect this had on the anti-trade bias. Section IV concludes. II. Initial conditions: the problem of redundancy of instruments In a stylized description, the trade regime in a "typical" non-CFA zone African country pre- adjustment could be characterized by a fixed official exchange rate and rigid controls, with multiple rates and a large black-market premium; a restrictive licensing system for imports, operated in conjunction with the exchange controls; a fairly high and escalated import tariff structure comprised of several layers2; extensive exemptions from duties, especially for imported inputs used by local producers; reference price schemes for imports; government-controlled marketing channels for major commodity exports (and often imports), with large explicit or implicit ta^.es on these exports; a lack of institutions to support non-traditional exports; and cumbersome export procedures for private exporters, designed to ensure repatriation of export proceeds. Each of these mechanisms created a high bias against trade,3 and they were to a considerable extent redundant. That is, exchange controls, import licenses, and tariffs all act to restrict imports and any one of them could be used to provide any level of protection for domestic production. However, although these instruments were 2. Typicat , the level of tAriffs Mr a ar bound by obligations under a regional agreement. But these agreements generally put tio limit on fiscal and other surcharges, which can therefore be used not only for fiscal, but also for protective purposes. 3. .Ine 1983 World Develooment Re_ot ra tes pricing policies in the 1970s of a sample of 8 SSA and 23 non-SSA de-eloping economies. There are 7 different pricing categories, of which 3 -- exchange rate, protection of manufacturing, and protection or taxation of agriculture - - relate directly to trade policy. Of the SSA sample,0hana is rated highly distoned in all 3 categories; Nigerian and Tanzania in 2; Senegal, Cameroon and Kenya in 1. Only Malawi and Cote d'lvoire ame not rated highly distorted in any of the 3. -3- clearly redundant in .fect, they were not always redundant in purpose. African countries were (and are) highly reliant on tariffs for revenue, and in many of them, this was the primary motive for setting rates at high levels. (The revenue motive does not, of course, explain why rates were also very disperse.) Exchange controls and nontariff import controls were often used as substitutes for macroeconomic discipline when balance of payments crises arose, though they were applied discriminatorily against imports that competed with domestic production. Exemptions and low rates on inputs, of course, were generally intended to be protective of domestic industry. For countries in the CFA zone4 many of the same instruments were used to restrict trade, but with one important difference. These countries have convertible currencies, pegged at a fixed rate to the French franc. Their monetary arrangements among themselves and with the French Treasury virtually ensure against systematic, large losses of reserves and shortages of foreign exchange of the same magnitude that have plagued other African economies. (At times, however, reserve losses have been avoided only with the use of stringent import licensing). Thus, foreign exchange controls or rationing as practiced in non-CFA countries are less of a proolem, and black market premia tend to be very low. The black market premium has in recent years gen.rally averaged less than 5 percent (see Table 5), though there have been exceptions, including 1978 (when it reached 16.5 percent), 1983 (7.8 percent), and 1989 (5.8 percent) Likewise, many restrictions on the export side were redundant. Requirements to repatriate export proceeds at a below-market exchange rate, low official procurement prices (with a procurement monopoly held by a parastatal agency), export bans and licensing, and explicit taxes have all been used to impose taxes on export sectors. Although not every mechanism was applied simultaneously to every export, in some countries, several were used simultaneously. In Ghana, for example, the 4. Actually two 7--es f countries, each of which has its own monetary union, but sirmilar arrangements. These countries are Benin, Burkinm Faso. C, D'lvoiz, Mali, Niger, and Senegal in the West African Monetary Union, which uses the Banque Centrale des Etats de ['Afrique de l'Ou:st (BCEA-) as cent;al bank; aid Cameroon, the Central African Republic, Chad, the Congo, Equatorial Guinea, and Gabon, which use the Banque des Etsts de l'Afrique Centrale (BEAC) as central bank. - 4 - government cocoa marketing parastatal controlled exports, was run inefficiently at high cost, was taxed by the central government, and received an extremely low exchange rate for exports. Private exports were banned. Ir addition (or, one could argue, as a consequence of these other measures), the official procurement price to producers wvas set at a level far below border parity. One implication of the redundancy is that "trade policy" in the Afiican context -even more than in other developing regions - must be understood in broad terms that include commercial policy, ex..ange rate policy, and (especially for primary commodities) marketing issues as well. Another implication is that piecemeal trade policy reforms do not work - as they might in other contexts - to gradually reduce protection and reverse the anti-trade bias. The first reforms may have little or no effect on the bias. With no change in foreign exchange rationing or import licensing, reduction of tariffs will not reduce the anti-trade bias, and may even increase it indirectly by creating greater escalation in the structure or by reducing government revenue, thereby increasing the deficit and contributing to the overvaluaeion of the exchange rate. But then, as the final redundant "layer" of protection is peeled off, there may be a very abrupt drop in protection as the previously reduced tariffs beco. ie binding (as in Ghana). The multiplicity of trade policy instruments also makes it relatively simple to offset the effects of reforms involving one type _f measure with changes in another. Tariff reductions can be accompanied or followed with increases in fiscal duties or other import charges (such as Ghana's "super sales tax") to avoid reduction in protection. Or other instruments can be substituted, as Senegal has done with its reference price scheme for imports. Also, what appears to be reform may not in fact be a step toward reducing protection and the anti- trade bias. For example, in almost all SSA countries where the reform involved a staged shifting of imported products from more to less restrictive licensing (for access to foreign exchange and/or import licensing per ), the first products to be shifted were imported inputs, and the last were final -5- products that corrpeted with local production. Reducing the effective pricc (or equivalently, ir.creasing the availability) of inputs may well have be, an appropriate measure in some cases to increase the low rates of nroductive capacity utilization. but the immediate effect of such measures is in fact to increase - not reduce - effective protection.5 What this means is that to be effective in reducing the anti-trade bias, trade policy reform in the African context must be quite "deep," involving aI the layers of instruments, not just the superficial ones. III. What was planned and what was done? Thle design and implementation of trade policy reform programs The experience of African countries in designing and implementing reforms has been extraordinarily diverse. This diversity, as well as some of the issues that arose in different countries, is discussed in section A. This is done by examining examples of what was planned and done with respect to eac'i of the major categories of trade policy. But accurate generalizations cannot be made, nor trends detected, based on examples. Sections B and C, therefore, use two kinds of aggregate statistical information to draw inferencc regarding design and implementation issues for larger sub- groups of African countries. A. Anecdotal evidence, country by country 1) Exchange rate As noted earlier, the real exchange rate in most non-CFA SSA countries pre-adjustment was fixed at a significantly overvalued disequilibrium level, even for the existing trade regimes, as indicated by chronic balance of payments problems. Furthermore, reducing trade restrictions has the 5. Thit would not be the case in countries where exports are more intensive users of imported inputs than are import substitutes. This does not appear to be the case in SSA countries. -6- effect (ceteris paribus) of depreciating the hypothetical equilibrium rate. And external shoc'.s (failing commodity prices, increasing oil prices, and rising interest rates on external debt) for many countries would have also called for real depreciation. Consequently, virtually all adjustment programs6 were planned to include a depreciation in the real exchange -ate. In non-CFA countries, this took the form of devaluation of the nominal rate, comhined with supporting fiscal and monetary measures. In CFA countries, the nominal exchange rate remained fixed, so the effort to a^hieve a r.-ad depreciation concentrated on stringent fiscal and monetary policy to try to achieve either a nominal deflation in the domestic price level or at least deflation relative to trading partners. With few exceptions (most notably Zambia), countries did achieve some real devaluation. For the non-CFA countries, this amounted to 26 percent comparing the years before to the 3 3 ears after the beginning of the adjustment program, or 30 percent comparing the 3 years before the program to the 3 most recent years (Table 5). (For CFA countries, the real devaluation was much smaller.) Often, howeve,, the potential beneficial effects of the devaluation for export sectors was attenuated by marketing arrangements. State monopolies of export marketing allowed the government to keep producer prices depressed. Thus, devaluations were accompanied by a fall in the ratio of producer prices to border prices, as in Madagascar and Tanzania (Jacquemont and Assidoii, 1988). In some CFA countrit. (notably Senegal and Cot; d'Ivoire), there was also an attempt to mimic some of the effects of a nominal devaluation by putting in place a subsidy for selected exports and simultaneously raising import tariff rates. (This was also the first step in Ghana's program, but was soon followed by nominal devaluations). Although these episodes have sometimes been called "mock devaluations," since they were designed to raise the prices of both imports and exports relative to nontradables, their effects on prices differed considerably from those of a nominal exchange rate 6. Hee, we do not draw a distinction between stabilization and adjustment programs. Real exchange rate adjustment can be considered part of either one. -7- devaluation. One reason for this was the selective nature of their product coverage; another was that the rates varied, not only between exports and imports, but among the various imports. Thus, unlike a nominal devaluation, these schemes shifted relat.,ve prices among tradables. Another major difference was that they relied on the ability of customs to coll wt the taritfs and of other agencies to distribute the subsidies. These schemes never funct.oned well (particularly the export subsidies) and were eventually abandoned. More recent efforts to achieve a real aevaluation in the CFA countries (under the rubric of "increasing competitiveness') have focused on lower; ig the effective price of non- tradable inputs such as water, electr;city, and especially labor through regulatory and labor market reforms. In non-CFA countries, the adjustment programs generally included significant changes in the mechanisms by which foreign exchange was allocated. These usually involved establishment of a pot of foreign exchange that would be available to fund the import of certain products automatically (i.e. without licensing). These schemes went by different names, including "Liberalized import Regime" (Madagascar),"Open General Licensing" (Uganda, Madagascar) or exchange auctions (Ghana, Nigeria, Uganda, Zambia). The group of products eligible for such treatment was intended to be expanded over time. Sometimes this was done by beginning with a few products on a "positive" list and later switching to a "negative" list, whereby all products were eligible except those specifically excluded. The foreign exchange to eligible enterprises was sometimes allocated at a fixed rate (as in Uganda early on), but this usually was intendeu to evolve into a foreign exchange auction, with allocation at a variable rate depending on fluctuations in supply and demand at each auction. The auction and official market would gradually merge as more products were made eligibia for auction funding. Private sector foreign exchange operations were also legalized in some countries (Uganda, Ghana, and to a lesser extent, Nigeria). -8 - While all of these seem to be steps toward currency convertibility, only one Pon-Ct7A country (Mauritius) seenis to have established (and met) de facto convertibility as an explicit goal. In Nigeria, where there has been no formal foreign exchange allocation since 1986, the "auction" did not actually serve as a market-based allocation system, as evidenced by persistent unmet demand and parallel market permia. As of August 1992, there existed a parallel market with a premium of 50 percent. Some others are close; Ghana and Uganda ha o reasonably well functioning exchange rate auctions, which are sometimes perceived to be equivalent to a well-functioning free markeL in foreign exchange. And importers who do not wish to use the auction can go to private fore:gn exchange bureaus. However, conceptually, these auction mechanisms differ in a very important way from a true market-based system. The auctions are of the Dutch variety, where successful bidders pay their own bid price, not the price of the marginal bid. Thus, unlike in a market system, different users (bidders) pay different prices for the same commodity, with the bidders who can put the foreign exchange to its highest valued use (and are thus the higilest bidders) paying the highest implicit tax. Thus, some of the potential benefits of the market mechanism-especially with respect to the efficient intertemporal resource allocation that comes about through attracting resources to the most profitable activities-is vitiated. This also raises domestic prices above border prices for imported goods financed by exchange purchased at the infra-mirginal rates, imposing a new, variable, and non-transparent protect.on mechanism if these are goods that compete with domestic production. There are also empirical indications of problems with these auctions. In Ghana, shortages have been reported, again calling into question whether the auction is really being allowed to determine a market-clearing rate. In Uganda, importers pay a significant premium (reportedly around 8 percent) at the private bureaus. apparently indicating high transaction costs in dealing with the auctions. In other countries, progress in moving to a market-based system has been virtually nil (Kenya), or has been reversed (Madagascar suspended its OGL system in late 1992, Nigeria has recently -9- retrogressed), or is partial and recent. Zambia started another OGL and auction system in 1991, but some imports are excluded and disbursements are slow. Tanzania's latest OGL system (there was an earlier one in the 1970's) was started in 1988, but is neither open-only certain imports are eligible-nor genera!, since importers need a separate license (Lyakurwa). Whether the OGL systems actually constitute a step towards reducing protection depends very mucih on whether they are "open" to goods that compete with domestic production. Another important exchange rate reform was the legalization of "own-funds" or "no forex" imports, whereby no license was required (or was automatically granted) for imports made using the foreign exchange from the importer's own sources, whatever they were. In Tanzania, the plan was initially (in 1984) to allow only certain products to participate in the scheme, but the government later decided to expand this to virtually all imports. "Own Funds" now account for around 30 percent of all imports in Tanzania, with open general licensing accounting for a similar proportion. But the system still administratively allocates 20-25 percent of the foreign exchange used for imports. Kenya in 1991 introduced a "foreign exchange bearer's certificate" scheme, which superficially resembled the "own funds" regime. Under Kenya's plan, a recipient of exchange remittances from abroad could exchange them for a certificate which was a claim on the foreign exchange whenever the bearer chose to withdraw it. The certificates were tradeable, and traded at a premium of 34-50 percent over the official rate. However, unlike some other "own funds" schemes, this one did not eliminate the requirement of a license for imports made using fcreign exchange obtained through the scheme. The effect was not, therefore, significant liberalization or reduction of protection, though it did have other salutary effects. 2). Import policy a). Non-tariff barriers (NTBs) to imports - 10 - All of the reform programs included reduction of t}e coverage of non-tariff barriers to imports. In some of the non-CFA countries, the exchange allocation and import licensing arrangements were inextricably linked, so that the measures that liberalized the former automatically liberalized the latter (Nigeria, Tanzania, Ghana). In countries like Nigeria, Ghana, and Uganda, virtually all imports could be imported without license by the late 1980's, using funds that could be purchased at auctions or from foreign exchange bureaus. In others, the plan was to remove the NTBs without significantly changing the foreign exchange allocation system. In Kenya, for example, the import bans and "non-objection certificates"' were removed very early (1980) while the lifting of licensing requirements (initially by reducing coverage by 20 percent of imports) began in 1982. There was, however, reimposition of licensing requirements soon thereafter, with no further progress until 1988. Removal of licensing requirements in the CFA countries (like exchange restrictions in non-CFA countries) was selective, i.e., applied only to certain products. This meant that the process, particularly in early stages, did not necessarily reduce protection. In Senegal, the first quantitative restrictions to be lifted were those on non-competing imports (1986-87). These measures did not by themselves reduce the effective protection (probably the opposite), though they were followed within a year by the elimination of virtually all quantitative restrictions on competing imports (except rice, sugar, and cement) as well. In Cote d'Ivoire, the program envisioned removal of all quantitative restrictions (and replacement with surcharges), but progress was slow and was reversed completely in 1989 when restrictions were reimposed on virtually all competing imports. Apart from licensing, trade regimes included other protectionist instruments. One of these was the practice of assigning a minimum "reference price" (unconnected with the actual price paid on world markets) for purposes of 7. These required that a prospective importer essentially obtain the permission of competing local producers as a condition of getting impor license. - 11 - levying the import tariff, or directly establishing a minimum import tax. Both have been used in Senegal, for example. The use of reference prices was abolished in 1986, but reintroduced (along with minimum taxes on different products) in 1989-90. Import prohibitions have been used for some particularly sensitive products, either "luxuries" or those competing with domestic production, ana have proven difficult to eliminate. In its 1986 program, Zambia planned to eliminate import licensing and all protective import prohibitions, but this was never done. Nigeria, which eliminated other licensing requirements in 1986, still bans a number of agricultural items to protect domestic production. b). Tariffs All of the trade policy reform programs included some actions on tariffs, but the timing and type of action varied significantly. There was not universal adherence to the now-standard recommendation of raising tariffs on final products as NTBs are removed, then reducing them in phases, while increasing rates on inputs. In Senegal, tariff rates were first raised on a few products in the early 1980's (as part of the "mock devaluation"), then the structure was rationalized somewhat by reducing the spread between rates on final goods and inputs from 40 percent in 1985 to 25 percent in 1988. Most of this was done before protective NTBs were significantly affected. In Senegal, the tariff reforms were to some extent compromised by reclassification of some goods through the "Codes de Precision." Similarly, the first tariff reduction in Kenya, which reduced the average level from 58 percent to 50 percent, came in 1983, long before significant action was taken to reduce NTB coverage. As of 1986, rates in Kenya were still high and dispersed with ERPs averaging 90 percent and ranging from -167 percent to 1,019 percent (Mosley). In 1989, tariff reform continued with a rationalization of the structure (reducing number of different rates from 25 to 12), and some modest reductions in rates in 1990. - 12 - Ghana is an extreme case of implementing tariff reforms before NTB reforms. The tariff structure was rationalized to three rates (10, 20 and 30 percent) in the early 1980's. The 2xchange allocation system and import licensing were not significantly reformed until much later. When that happened in 1987-89, the relatively low tariff rates suddenly became the binding constraints on imports, and effective protection fell precipitously. This led to cries of distress from domestic producers and then to imposition of a "super sales tax" on some imports of 75-500 percent in January 1989. This was lowered to a range of 10-100 percent in 1990. Overall progress in tariff reduction in SSA has varied greatly from country to country. None, however, have approached (or announced intentions to approach) levels or rate structure that would be classified as low in comparison to, say, the more open economies of Latin America and the Caribbean. There a number of countries have (or are in the process of putting in place) rate structures of around 5-20 percent8. Ghana came closest to this, when its rates were in the 10-30 percent range, before introduction of the super sales tax on imports. Nigeria's rate structure was also close to this range after the 1988 reductions; between 1989 and 1991, however, a number of rates were raised, some to 300 percent. Maximum rates in most SSA countries with relatively liberal foreign exchange and import licensing regimes are generally over 100 percent. Meaningful comparisons across countries are difficult to make, however, because the reported "ranges' and "maximum rates" are very dependent on the level of aggregation at which the data is compiled and reported. Another tariff-related instrument that is often t3ed to increase effective protection is duty exemption9. Exemptions given on imported inputs to import-competing production have the effect of 8. Chile has for many years had a low, uniform tariff, though the rate has changed from time to time. It is currently 12%. Bolivia's range is 5-10%, Mexico and Argentina each have tariffs that rmnge from 5-20%. Cosua Rica's range will be 10-20% by mid-1994, and the mst of Central America has announced its intention to follow shortly. 9. Exemptions on impons of non-competing final good impons, as some countries (e.g. Uganda) grant to non-govemmental organizations, are not directly protective of import-substituting production. Nor are exemptions on imponed input. granted to exporters. Indirectly, all exemptions are potentially protective in the sense that by reduciog revenues they may increase the fiscal deficit or require that tariffs on other (competing) impons be set at higher levels for fiscal reasons. - 13- increasing effective protection. The exemptions may be granted as part of some formal scheme (e.g. an investment incentive package) or on an ad hoc basis. The exemption system may be very complex. Mauritius, for example, has over 400 categories of exemptions, of which over 200 are commonly used. Other than food, most are on imported inputs for manufacturing. Even leaving aside exemptions of inputs for export production, exemptions amount to 36 percent of the total import charges that would otherwise be chargeable in Mauritius. In the UDEAC countries10, the actual collection rates as a fraction of the duties legally chargeable ranged from 15 percent (Chad) to 46 percent (Gabon). The share of goods that paid the full rate was very low in all these countries ranging from 9 percent (Chad) to 39 percent (Gabon). A third common tariff-related instrument is a requirement that importers deposit domestic currency some period of time in advance of receiving foreign exchange. This is best analyzed as a tariff, since it raises the cost of importing by some proportion of the value of the shipment, but does not act as a quantitative restraint. The amount of additional cost depends on the length of the delay in receiving the exchange, the opportunity cost of the funds, and the amount of the deposit required. For example, an advance deposit requirement of 100 percent value, with a delay of 3 months, and an opportunity cost of funds of 20 percent per year would be equivalent to a tariff surcharge of 5 percent. Some countries with relatively liberal regimes maintain these requirements, though at reduced levels. Nigeria reduced the requirement from 100 percent to 25 percent in 1988. And in Kenya, where reforms have not advanced very far, the requirement is currently 40 percent. In Tanzania, the requirement is 100 percent and in an environment of credit scarcity, this has discouraged utilization of the otherwise relatively liberal access to foreign exchange (Lyakurwa). 10. World Bank study, cited in EC (1992). - 14 - c) Protection rates Judging the extent to which protection has been reduced by the trade policy reforms is tricky. One problem is the redundancy of instruments. This tends to make true effective rates of protection (ERPs) higher than they would appear to be from tariff rates alone. On the other hand, smuggling tends to make domestic prices lower than tariff-inclusive border prices". IThus, it is impossible to quantify protective effects of NTBs or tariffs without a product-by-product comparison of domestic and border prices. Another problem is that even in regimes where tariffs are the only or main protective instrument, and when smuggling is not an issue, summary measures of tariffs are meaningless or misleading as indicators of ERP. The legal tariff schedule is misleading because of prohibitively high ratesI2 and because of the prevalence of exemptions. But actual collection rates (duty collections as a fraction of import value), which differ from the legal schedule by the average percentage of exemptions, can be misleading as well. A low collection rate is sometimes taken as an indicator of low protection, but it can mean just the opposite if the rate is low because of exemptions on non-competing imported inputs or because inputs carry low legal tariffs relative to final products ("escalated tariff structure"). And these are, in fact, the most common causes of low collection rates. A few studies have estimated effective rates of protection, taking account of the escalation in the tariff structure (i.e., the higher rate on final products). Mosley estimated sectoral ERP's in Kenya before the latest reforms to range from -167 percent to 1,019 percent. averaging 90 percent. (However, such range estimates are very sensitive to the level of aggregation of products into sectors). Rouis traced the evolution of the "average" ERP in Senegal, based on changes in the I1. One study in Cameroon concluded that even though imports of textiles were controlled by licensing and carried a tariff rate of 135% domeatic prices only exceeded border prices by 35 % This was attributed to illegal imports. 12. Once the tariff rate on a given item reaches a certain point, imports fall to nil. Rates higher than this do not have any effect on protection. Many tariff regimes in relatively closed economies have a number of examples of such 'water' in the tariff, which makes protection appear to be higher than it actually is. - 15 - nominal tariff rates and certain assumptions about the average value added, and found the ERP fell from 165 percent in 1985 (pre-reform) to 89 percent in 1988, then increased, reaching 98 percent in 1990. A study of protection based on surveys of import-substituting manufactured firms in the UDEAC countries indicated that 23 of the firms (out of 76 useable surveys) showed negative value added (infinite ERPs) at world prices, while 8 others had ERPs greater than 600 percent. In Tanzania in the early 1980's, the average rate of effective protection in manufacturing was estimated to be 500 percent, declining to 150 percent in 1985 due to increased imports through the "own funds" scheme. 3) Export policy An important distinction must be drawn between traditional exports (mostly agricultural or mineral) and non-traditional exports. For the former, the major issues have been direct taxes or indirect taxes administered via state-owned marketing enterprises (sometimes called marketing boards), exchange policies, and sometimes legal barriers to exports. Non-traditional exports hame not generally been explicitly taxed nor marketed by state monopolies. The major issues for them have been how to put in place mechanisms to compensate for the overall anti-trade bias inherent in the import regime, and in some cases licensing procedures that are cumbersome, though not intended to be restrictive. a). Traditional Several countries have gone a long way toward eliminating taxes and allowing producer prices to be set by border prices. The most successful in this was Nigeria, where the marketing boards for cocoa, cotton, groundnuts, rubber, and palm kernels were simply abolished, thereby eliminating them as a vehicle for taxing producers. However there has been some back-tracking; bans have been imposed on exports of skins, wood, palm kernels, and some grains to encourage domestic processing. There has also been some discussion of a ban on cocoa exports, but no action so far. Madagascar moved slowly, but eventually (1988) turned over internal and external marketing of export crops - 16 - (except vanilla) to the private sector and in 1987 abolished export taxes, except those on cloves, vanilla and coffee. In countries where one or a few crops provide most export earnings - coffee and cocoa in Cote d'lvoire, cocoa in Ghana, coffee in Uganda - reforms have been especially slow, at least partially out of concern over government revenue losses. The government in Cote d'Ivoire planned (though it did not really carry through) the elimination of taxes on all exports, e coffee and cocoa. In Ghana, there was no real effort to privatize cocoa marketing until a pilot program to license private purchases was launched in the early 1990's. Price reforms in Ghana have taken the form of raising cocoa producer prices, rather than institutionalizi.g a X .!eiai.ism to link domestic and border prices. Plans to divest some plantations of the marketing board were announced in 1984 and again in 1986- 87, but with limited actions. Uganda began late, but has made some progress, beginning with increases in coffee producer prices in 1987, then continuing in 1991 with replacement of the coffee pricing system with a system of floor price and variable levy (to link prices more closely with border prices), and (in 1992) with reforms to allow private exporters to compete with the marketing board. It is noteworthy, however, that other (nontraditional) exports were given exchange retention privileges in 1989 and had licensing requirements eliminated in 1990, with coffee specifically excluded. Similarly, non-traditional exporters in Tanzania were allowed to retain 50 percent of their foreign exchange earnings, while traditional exporters were restricted to 10 percent. Another problem in Tanzania has been the reluctance to abolish or significantly reform the traditional export marketing boards, with the result that very little of the benefit of the devaluations has been passed on to producers. A number of countries have in place export bans on certain products. Some of these are linked to environmental concerns (e.g., Uganda's ban on ivory exports) or are used to prevent the export of food or raw materials that are subsidized domestically. Others are simply attempts to protect domestic - 17 - processors by keeping their raw materials priced lower than their world market values (e.g., logs in Ghana). b.) Non-traditional Attempts to promote non-traditional exports have taken several forms. One of the early forms was direct export subsidies based on the value of exports. Kenya and Tanzania operated such systems in the 1970s and early 1980's, as did Senegal and Cote d'Ivoire, as part of the mock devaluation schemes in the latter two countries. These systems encountered problems such as slow disbursement, high fiscal cost and fraud (e.g., fish were smuggled into Senegal, then re-exported to collect the subsidy), and have by now been abandoned. Other countries have given exporters special subsidies, such as tax holidays, income tax refunds, and depreciation allowances (Ghana, Mauritius, Nigeria). Most of the SSA countries have attempted to implement some type of system to allow exporters to be either exempt ex ante from taxes on imported inputs (in-bond or temporary admissions schemes or export processing zones) or to have these taxes rebated ex post (drawback mechanisms). The effective operation of these schemes is critically important to allow exporters to compete on world markets, where other firms have access to imported inputs on a duty- and restriction-free basis. Their importance is magnified by the high protection accorded to local manufacturers of inputs in SSA countries. In Cote d'lvoire, empty metal boxes used for canning are 20-40 percent more costly than ini France or Thailand, with local production a virtual monopoly protected by a high tariff. Polystyrene boxes used to package exported fish are five times as expensive in Cote d'Ivoire as in TLunisia (World Bank 1992 b). In spite of all the attempts to simplify life for exporters, they still face lengthy delays and regulatory burdens in many countries, or at best, reforms to change this situation are very recent. Examples of drawback schemes that are new or recently improved include Senegal (1990-91); Malawi (1991); Tanzania (1988); Uganda (1990); Zambia (1991) and Nigeria (1991). In Madagascar and - 18 - Tanzania, delays have been related to export licensing or certification. In Tanzania, as a result of establishment of a Trade Facilitation Council, the average waiting time for each export license has recently been reduced from 6 months to one week, though to make the system operational required 8 years of effort and a Presidential decree (Lyakurwa). In other countries, even where licenses are not a major problem (e.g., Ghana and Uganda), there are lengthy delays and rnuch "red tape" required to obtain rebates on imported duties. This has been a very common complaint among exporters, and is so serious that many firms simply do not find it is worth the bother to use the schemes. The complexity of the systems is related to the lack of trust the governments have in exporters (World Bank 1992b). Other than Mauritius, which has implemented an EPZ system with spectacular success, none of the SSA countries has managed to put in place a system of exemption or rebate that is successful in attracting wide-spread participation by exporters. (Madagascar has also had some moderate success, largely as a spill-over effect from Mauritius). One reason for the disbursement delays is that the mechanisms have not relied on "standard" input-output coefficients like the East Asian schemes. Thus, importers must document their input use for each shipment. Another major problem has been that these schemes have not been accorded priority in allocating the budget, so shortages have prevented disbursement of the rebates. Also, some have been designed to rebate only part of the duties paid on imputs used for exports (half in Nigeria, raised to 100 percent in 1991). As a result, as of 1992, exporters in Tanzania were still waiting over a year for drawback disbursements because of budget shortages. Kenya began an ex ante exemption scheme in 1992, though it has had a drawback schemp for some time. The drawback scheme, however, was never heavily used, because of the slow disbursements and reluctance of the government to provide adequate funding for the drawbacks. The latest improvements in the Nigerian system appear promising, as they make half of the drawback available upon application, thereby mitigating the problem of delays. (This was combined with an - 19 - increased budgetary allocation.) The Nigerian system is also promising in that it relies on standard coefficients, with exporters able to apply for individual coefficients if they wish. New Export Processing Zones (EPZs) or in-bond schemes have been established in Nigeria (in- bond, 1991); Kenya (EPZ, 1990; in bond, 1989); Togo (EPZ, 1989); and Zambia (in-bond, 1991- 92). The schemes in Togo and Madagascar follow the Mauritius example of allowing EPZ status for individual factories. These EPZ and in-bond schemes have the advantage for exporters of providing ex ante exemption; that is, import duties are not paid, so there is no need to rely on the vagaries of a rebate system. However, they have the inherent limitation that they can be used only by manufacturers that export all (or virtually all) their production. (A few countries, !ike Mauritius, allow limited sales in the local market.) They tend to be used only by a few large firms-especially in-bond schemes (World Bank, 1992b). To some extent, the regulatory burden on exporters is also connected to the foreign exchange system. Exporters do not generally have automatic access to foreign exchange (and import licenses), so they are forced to deal with the red tape of the allocation system (World Bank, 1992b). Since exporters are required to repatriate exchange and often are allowed to retain only part of their earnings (and in some cases cannot sell even the retained portion), governments have felt it necessary to closely regulate activities. This has changed recently in some countries. But in most countries, allowable retention rates are relatively low; the idea is to allow exporters to retain little more than the foreign exchange they will need for their imported inputs. In some countries, restrictions on foreign exchange retention by exporters have been relaxed somewhat; in others they have been tightened. In Uganda and Zambia in 1991-92, exporters (except coffee in Uganda) can retain and sell all their earnings. This has been true in Nigeria since 1986, though there is still a repatriation requirement and funds must be deposited in special accounts. In Ghana, the goal was to allow 50 percent retention; retention rates are now 5-35 percent, depending on the product. When Tanzania introduced its - 20 - scheme, the retention rate was 50 percent, but this was subsequently reduced to 35 percent (Lyakurwa). In Kenya, the retention system has had a very stop and go history. The rate was at first (in August 1992) set at 100 percent for nontraditional goods exporters, and 0 for all others, but wac expanded to include traditional products at a 50 percent rate in late 1992. Shortly thereafter, the whole scheme was suspended and the retention accounts nationalized (at a low official exchang- rate) when the government's relations with the IMF and World Bank broke down. Very recentiy, the scheme has been reinstated, with a 50 percent rate applicable to all exports (including services). Retention schemes have on balance been helpful, but in some cases have had unintended consequences when other distortions remain. In Ghana, wood processors encountered shortages of high-quality timber because loggers had a high incentive to export the timber to be eligible for the 20 percent foreign e. ehange retention scheme. This problem might have been resolved by making indirect exporters eligible for the retention. Unfortunately, the government tried to resolve the problem by lowering the retention rate for log exporters and banning certain types of exports altogether. Other reforms in export-related policies have also come only recently, or are seriously compromised by macroeconomic policies. Investment codes in many African countries have been liberalized to encourage foreign direct investment (Zimbabwe in 1989; Tanzania and Malawi in 1990, Zambia and Burundi more recently). But foreign exchange shortages and regulations necessitate restrictions on royalty payments, profit repatriation, and payment of expatriate staff (World Bank 1992b). Macroeconomic imbalances also lead to high government borrowing requirements, which crowd out export finance. Infrastructural problems and associated high cost of transportation, energy, and water are also high on the list of exporters' problems, but are beyond the scope of this report. - 21 - B. An overview from the ALCID. comparison of SSA and other countries The Adjustment Lending Conditionality and Implementation Database (ALCID) contains information on policy actions taken in conjunction with adjustment loans of the World bank (Box 1). Virtually all countries in SSA, and many developing wountries in other regions, received such loans in the 1980s. Generally, any policy reforms undertaken as part of an adjustment prograrn were discussed in the loan documents and were thus included in the database. Thus the ALCID contains arguably the most comprehensive catalog available ot policy reforms in the developing world in the 1980s. The level of generality used in classifying these r licy actions in the ALCID does not allow in-depth aggregate analysis. But the breadth of coverage makes possible meaningful comparisons of general patterns across time and across regions. Below, I first discuss evidence regarding the design of reform programs, then regarding their imp!ementation. I) Wnat was planned Policy actions supposed to be taken as part of adjustment programs in SSA countries have consistently emphasized trade policy and exchange rate (TPER) measures less than has been the case in non-SSA countries. For the whole period 1980-92, 16.8 percent of the actions specified in non- SSA adjustment loans were TPER, compared to 12.2 percent in SSA adjustment loans (Table 1). This pattern holds for every year in this period, except 1980 and 1988. It also holds for critical actions, i.e., those that are judged by the task manager of a World Bank adjustment loan to have been critical to the success of the adjustment program: 22.7 percent were TPER in non-SSA, compared to 16.4 percent in SSA. As in non-SSA countries, TPER actions in SSA have received diminished emphasis since 1986-87, when they represented 17.6 percent of all actions in SSA and 24.3 percent in non-SSA (28.8 percent and 31.8 percent of critical actions). In subsequent years, they have represented between 5.7 percent and 12.5 percent of all actions in SSA and between 10.2 percent and 18.0 percent in non-SSA. TPER actions have represented a somewhat smaller proportion of conditions in CFA countries (8.5 percent from 1980-92) than in.other SSA countries (13.9 percent). - 22 - Table 1. Conditions related to trade policy-' (percent) Other countnes Sub-Saharan All counines Year Non-7TER TPER Non-7'PEP! TPER Non-TPC TPER 1980 71.. 28.8 60.0 40.0 67.9 32.1 1981 72.2 27.8 80.8 19.2 7.5.8 24.2 1982 85.0 15.0 95.0 5.0 87.0 13.0 1983 72.9 27.1 82.9 17.1 76.4 23.6 1984 61.8 38.2 86.5 13.5 73.9 26.1 1985 87.2 12.8 90.3 9.7 88.3 11.7 1986 76.3 '3.7 80.9 19.1 78.8 21.2 1987 76,2 24.8 84.6 16.4 79.5 20.5 1988 89.8 13.2 88.1 11.9 89.0 11.0 1989 82.0 18.0 88.0 12.0 84.5 15.5 1990 89.7 10.3 94.3 5.7 92.3 7.7 1991 87.0 13.0 88.8 11.2 87.7 12.3 1992 88.6 11.4 89.8 10.2 89.2 10.8 ALL 83.2 16.8 87.8 12.2 85.3 14.7 -'TPER=Trade Policy Conditions Source: World Bank ALCID Table 2. Sub-Saharan Africa trade policies (pe.cent) Poli(y C. Exchange MIX M dunesl X di. "esl MIX Other Other Other rate Quantitative sub. sub. ftnance X regime X insn, trade Year rest. & promo. policies All 1980 10.0 20.0 20.0 0.0 0.0 10.0 10.0 10.0 100.0 1981 20.7 3.4 10.3 10.3 13.8 20.7 10.3 10.3 100.0 1982 75.0 0.0 25.0 0.0 0.0 0.0 0.0 0.0 100.0 1983 21.4 7.1 10.7 0.0 32.1 21.4 7.1 0.0 100.0 1984 8.8 14.7 14.7 5.9 2.9 29.4 20.6 2.9 100.0 1985 19.0 33.3 9.5 4.8 4.8 9.5 9.5 9.5 100.0 1986 17.9 17.9 17.9 8.5 7.5 14.2 10.4 5.7 100.0 1987 16.7 10.8 34.2 6.7 2.5 9.2 6.7 13.3 100.0 1988 13.3 15.9 17.7 2.7 7.1 12.4 15.0 1 .9 100.0 1989 23.1 13.5 32.7 3.8 0.0 12.5 2.9 11.5 100.0 1990 14.3 19.6 10.7 10.7 0.0 8.9 21.4 14.3 100.0 1991 8.8 25.0 26.5 1.5 4.4 1.5 17.6 14.7 100.0 1992 15.9 25.6 17.1 6.1 3.7 18.3 4.9 8.5 100.0 ALL 16.5 16.8 21.7 5.4 5.2 12.8 10.8 10.8 100.0 Source: World Bank ALCID In SSA loans, actions related to import duties have been the most common (21.7 percent of all TPER actions), followed by quantitative restrictions on imports and exports (16.8 percent), exchange rate (16.5 percent), "other" export incentives and regime'3 (12.8 percent), "other trade policies""4 13. Oththr export incentives and regime' means not related to credit or direct subsidies. 14. 'Other trade policy' measures were often studies or agreement on a plan to take action later. - 23 - and "other export institutions and promotion" (each with 10.8 percent), export duties/subsidies (5.4 percent) and import/export finance (5.2 percent)(Table 2). The patterns in CFA and other SSA countries was similar, with two exceptions: CFA countries, as expected, had virtually no conditions related to the exchange rate, and they had relatively more conditions related to export duties and subsidies. This latter finding was apparently related to the "mock devaluation" episodes (see below) CFA countries also differed from other SSA countries in an important aspect of sequencing. In the latter countries, conditions related to removal of quantitative controls were more concentrated in the pre-loan period than were conditions related to duties and subsidies. This pattern was reversed in CFA countries. With respect to timing of actions, adjustment loans to Africa have placed somewhat less emphasis on planned implementation prior to effectiveness'5 of the loar. than h.ave non-SSA loans. In SSA, 25.7 percent of TPER actions were supposed to be pre-effectiveness, compared to 35.1 percent in non-SSA. In some cases there was a conscious decision by the World Bank not to delay loans until actions had been taken, due to the urgent need for foreign exchange. This was the case, for exarnple, in Ghana, where the economy was virtually grinding to a halt. SSA loans, however, placed more emphasis on legal conditions than did non-SSA (30.0 percent of TPER actions in SSA, compared to 26.3 percent in other countries). 2) What was done The three implementation categories - canceled, non- and partial implementation - together are referred to here as "under-compliance" or "under-implementation". Among all TPER actions called for in SSA loans, 64.6 percent were fully or over-implemented; 19 percent were substantially implemented; and 16.4 percent were under-implemented as of the release of the last tranche of the 15. The datc of 'effectiveness' is the date the loan (or parn of it) becomes available to the borrower. - 24 - loan (Table 3). Comparable figures for non-SSA loans were 55.1 percent, 18.) percent and 21.4 percent. 16 This under-compliance rate for the CFA countries was 26 percent, compared to 14.1 percer.t for other SSA countries. In SSA, the types of actions with minimal under-compliance were exchange rate, import/export quantitative restrictions, and import duties/subsidies, with under- compliance rates of 6-13 percent. Those with greatest under-compliance were export duties/subsidies, other trade policies, and other export institutions/promotion. (However, all of the under-compliance with "export duties/subsidies", and much of the problem with "other trade policies", was in the CFA countries.) This pattern was slightly different in non-SSA loans, where other export institutions/promotion, exchange rate, and other trade policies were clearly the greatest problems, and export duties/subsidies, import/export quantitative restrictions, and other export regime the least serious problems. With respect to the type of reforms most directly associated with institutions - "other export institutions and promotions" - implementation in SSA was somewhat better than in other countries as of last tranche release (27.6 percent under-implementation versus 38.3 percent), but slightly worse than in other countries as of the most recent implementation information. Table 3. IT implementation for Sub-Saharan Africa-' (percent) Poolicy code Canceled Non Partial Substanrial Full Over ALL Exchange rate 0.0 6.3 0.0 22.9 68.8 2.1 100.0 M/X Quantitative rest. 0.0 2.4 7.1 19.0 69.0 2.4 100.0 M duties/sub 4.4 0.0 8.9 15.6 66.7 4.4 100.0 X duties/sub 14.3 0.0 14.3 14.3 57.1 0.0 100.0 M/X finance 0.0 9.1 13.6 22.7 54.5 0.0 100.0 Other X regime 0.0 7.3 9.8 17.1 65.9 0.0 100.0 Other X inst. & promo 0.0 13.8 13.8 24.1 48.3 0.0 100.0 Other trade policies 0.0 4.5 13.6 9.1 13.6 59.1 100.0 ALL I .9 6.1 8.4 19.0 63.1 1.5 100.0 - IT is implementation as of last tranche release. Source: World Bank ALCID 16. Figures do not add to 100% because 4.6% were coded *futther monitoring irrelevant.- - 25 - In both SSA and non-SSA, under-compliance rates for TPER actions overall were lower based on most recent information17 than on last tranche release -- 11.8 percent versus 16.4 percent in SSA (Table 4), and 11.4 percent versus 21.4 percent in non-SSA. In non-CFA Africa, the most recent under-compliance rates for exchange rate actions and for quantitative restrictions were higher than the last tranche rate. All other conditions showed lower or the same under-compliance rates in the most recent sample. In th. CFA sample, all categories showed improvement in the most recent rates. In SSA, implementation of "critical" TPER conditions was worse than for all TPER conditions (17.7 percent under-compliance versus 11.8 percent). This pattern was true for CFA and non-CFA Africa, though the difference was greater in the CFA countries. In non-SSA, in contrast, implementation rates were better for critical than for all TPER conditions (6.5 percent under- compliance versus 11.4 percent). The SSA under-compliance rate for critical conditions was thus considerably higher than non-SSA (17.7 percent versus 6.5 percent). In SSA, TPER actions that were to be implemented before loan effectiveness had an under- compliance rate of 12.5 percent compared to 11.6 percent for other actions. For non-SSA, the figures are 4.5 percent for pre-effectiveness conditions and 12.7 percent for others. For legal TPER conditions in SSA. the under-compliance rate was 7.5 percent, compared to 14.6 percent for other actions. In non-SSA, under-compliance was 11.9 percent for legal conditions, and 11.1 percent for other actions. 17. This paragraph is based on implementation information as of the most recent information coded in the ALCID. In some cases, the most recent information was as of last tranche release. In others, more recent information (either 1988 or 1989) was available, indicating how the implementationstatus of each condition changed afler the loan was fully disbursed. - 26 - Table 4. Most reent implementation for Sub-Saharan Africa (percent) Policy code Canceled Non ParIal Substannal Full Over Monitonne AILL irrelevant Exchange rate 0.0 10.4 0.0 14.6 62.5 6.3 6.3 100.0 M/X Quantitative rest. 0.0 7.1 2.4 9.5 69.0 2.4 9.5 100.0 M duties/sub 4.4 0 0 2.2 11.1 66.7 4.4 11.1 100.0 X duties/sub 14.3 0., 7.1 14.3 28.6 0.0 35.7 100.0 M/X finance 0.0 9.1 0.0 22.7 36.4 0.0 31.8 100.0 Other X regime 0.0 7.3 2.4 12.2 61.0 2.4 14.6 100.0 Other X inst. & promo 0.0 12.8 25.5 14.9 34.0 6.4 6.4 100.0 Other trade policies 4.5 9.1 0.0 13.6 59.1 0.0 13.6 100.0 ALL 1.9 7.2 2.7 14.1 57.4 3.0 13.7 100.0 Source: World Banl ALCID 3) Hypotheses from the ALCID data From this general picture of reforms in SSA and comparisons with other regions, it is not possible to reach firm conclusions, but the evidence does at least suggest hypotheses. Trese are discussed here with suggested evidence. a) Non-CFA African countries had a very good implementation of minor reforms, but much worse implementation than other countries of the really key ("critical") TPER actions. In SSA, implementation of "critical" TPER actions was substantially worse than for all TPER actions. And, since the converse was true in other countries, implementation of critical TPER actions in SSA was much worse than in other countries (17.7 percent under-compliance in SSA, versus 6.5 percent in others). The discrepancy was greater in the CFA countries. This finding could account for much of the perception of implementation problems in SSA. b) Because they concentrated on tariff measures in the early stages, some of the adjustment programs did not significantly reduce the anti-export bias, which came primarily from exchange controls, import licensing, and direct or indirect export taxes. Given the redundant restraints on trade, it might be expected that top priority would have been given to exchange rate reforms, since as long as multiple and overvalued rates and administrative allocation remained in place, imports would continue to be restricted (regardless of import licensing - 27 - or tariffs) and exports repressed. After this, removal of quantitative restrictions on imports and exports would be meaningful in the sense that protection or taxation would then be provided by relatively transparent means (import tariffs and export taxes), which could then be reduced. While administrative allocation of foreign exchange is in place, reforms related to quantitative import controls or tariffs may have little effect on protection rates. In some respects, the observed pattern of planned actions followed this pattern. For example, some exchange rate adjustment featured prominently in the planned programs in the non-CFA countries, though in the early 1980s, implementation was another matter. But in other ways there are discrepancies. In the CFA countries (and some non-CFA), the relatively light emphasis on reducing quantitative controls before tariff and tax reform is somewhat puzzling. In some countries, Uganda being one, the early emphasis on tariff reforms can be understood as attempts to raise revenue, rather than reduce protection, but in others (Ghana) this was not so. Given the prevalence of the heavy taxation of exports, it is not clear why these reforms were a relatively small part of the program. To some extent, this may reflect the fact that the taxation was often not direct; thus reducing it may have required reforms in parastatal marketing institutions, rather than tax reduction per se. But in other cases, direct taxation reduction was explicitly planned, but not implemented as scheduled (Cote d'lvoire, Senegal). Another explanation may be concern for the revenue implications of lowering taxes, given the high dependency on taxes on mineral and agricultural exports. Dependency on cocoa revenues made Ghana's reforms in the cocoa sector (including reduction of taxation levels) very difficult. These observations suggest that early SSA reforms may not have really reduced the anti-export bias very much, since they were concentrated more on only one of the redundant layers of protection, tariffs. Other, more binding protection measures, such as exchange controls, quantitative restraints and export taxes, were addressed, but perhaps not sufficiently to allow the tariffs to become the binding constraint. c) Actions related to exchange rates and controls were important and well-implemented in SSA initially, but there was some backsliding in SSA (versus continued progress in other countries), undercutting the overall reform process in SSA. In actions not related to the exchange rate, SSA countries continued to make some progress in implementation after final tranche release, but progress was much slower than in other countries. Overall, SSA countries had a somewhat better implementation record of planned TPER actions than did other countries, as of the last disbursement of the adjustment loans. Furthermore, the best short-run implementation in SSA was in the area of perhaps the most serious problems, exchange rates (under-implementation rate of 6.3 percent for SSA, versus 31.3 percent for others). These patterns changed, however, when the criterion was implementation as of the most recent information. By this measure, implementation rates of TPER actions overall was virtually identical in SSA and other countries. And with respect to exchange rate actions, whereas under-implementation in non-SSA fell from 31.3 percent as of last tranche release to 6.3 percent, the rate in SSA rose from 6.3 percent to 10.4 percent'8. The implication is that at least some SSA countries made much slower progress than countries in other regions after last tranche release in continuing to implement previously planned actions, and in the important area of exchange rates, there was some backsliding. The underimplementation rate for CFA countries (26 percent) was considerably higher than that for non- CFA (14 percent). The most recent rate for the CFA (18 percent) was also higher than for the non- CFA. 18. Note that these implementation figures are for non-CFA countries, since nominal exchange rate movement was never part of the planned adjustment programs in the CFA zone. - 29 - d) Trade policy reform programs in SSA were slower than those in non-SSA, partly by design and even more so because of low implementation of 'up front' conditions. The implementation record of actions not legally required was poor. The relatively low reliance on reforms planned to be quickly implemented and the poor record on reforms that were not legal conditions make it appear that TPER actions in SSA may have been implemented reluctantly. With respect to timing, SSA reform programs were less "front-loaded" than reform programs in other countries, in both TPER and non-TPER actions. In SSA, 25.7 percent of TPER actions were planned for implementation before effectiveness of the associated loans, compared to 35.1 percent in non-SSA. In Ghana, and perhaps other SSA countries, there was an explicit effort to avoid up front conditions that might be difficult to implement, so as not to interfere with the rapid disbursement of urgently needed funds. Furthermore, the under-compliance rate for these up-front actions was relatively high in SSA (12.5 percent).19 Thus, in SSA a relatively small fraction of the policy actions were both planned to be done quickly and actually implemented on schedule. TPER adjustment programs in SSA placed more emphasis on legal conditions, as opposed to actions that were taken before the loan or were intended to be taken, but were not legally required for release of funds. Legal conditions were 30.0 percent of TPER actions in SSA adjustment programs, compared to 26.3 percent in non-SSA. And in SSA, the under-implementation of legal conditions was much lower than under-implementation of non-legal actions (7.5 percent versus 14.6 percent). In non-SSA, the rate of under-implementation for legal conditions and that for other actions were almost identical (11.9 percent and 11.1 percent). The evidence on timing of actions (relatively few up-front actions in SSA) and on implementation patterns (much worse in SSA for actions not legally required to get access to the loan funds) provides some support for the hypothesis that these 19. This compares to 4.5 percent in non-SSA for up-front actions, and 11.6 percent for actions in SSA not intended to be taken up-front. - 30 - TPER adjustment programs in SSA were planned and implemented more reluctantly and with less conviction of their benefits than in other countries. C. Evidence on reforms from performance of trade-related variables The ultimate test of whether an economic policy change has been effectively implemented, of course, is whether it has an effect on real economic activity. In this section, I examine the behavior of a number of economic variables that would be expected to change with trade policy changes. The movements in these variables are reported in Tables 5 and 6. For each variable, the "After" column shows the change from the three years before the beginning of structural adjustment (the year of the first World Bank Structural Adjustment Loan or IMF Structural Adjustment Facility) to the three years after. (The year the loan or facility was made available is excluded.) The "Recent" column shows the change from three years before to the three most recent years. For non-trade policy adjusting countries (including those that began adjusting in 1991), The "Recent" column is based on a comparison of 1980-82 to 1989-91. The rows report the averages of several groupings of countries, the members of which are listed in the table. 1. Indirect indicators of trade policy changes The first variable, the real effective exchange (REER), can be considered a direct policy variable in the short run. In the medium term, however, the sustainable equilibrium level of the REER is determined by (among other things) trade policy; the more open the regime (ceteris paribus), the more depreciated the equilibrium REER. The level of the REER index is not meaningful, but its direction of movement is. Taken as given that the pre-adjustment level was substantially -31 - overvalued20, depreciation is either a policy of greater openness pr se or an indirect indicator of such a policy shift. The results in Table 5 indicate that on average, SSA countries did devalue their currencies in real terms. (The means and medians tell similar stories.) The only sub-group whose currencies did not devalue in real terms is the non-trade-adjusting group. As expected, the CFA countries devalued (in real terms) only modestly, while the most intensive adjusting countries achieved the largest devaluation on average. The countries showing the largest real devaluations were Ghana, Nigeria, Tanzania, Mozarnbique, Madagascar, and Sierra Leone. The second variable in Table 5 is the change in the black-market premium (BMP). The BMP is an indicator of several trade-policy-related variables (among other things), including restrictiveness of foreign exchange licensing and mis-alignment of the REER. The mean BMP fell, Uut there were substantial differences across groups and even within groups. As the small median figures show, there were about as many countries in each group where the BMP increased as decreased. But there were several examples of spectacular declines, such as Ghana (a change of 2,059 percentage points from pre-adjustment to the most recent years), Guinea (1,042), Nigeria (247), Uganda (186), and Tanzania (158). 20. This is supported by a number of indicators, including chronic balance of payments difficulties and the large real appreciation of SSA currencies in the 1970s and early 1980s, The IMF World Economic Outlook, 1982 estinuted on average SSA currencies appreciated in real terms by 44 percent between 1973 and 1981. - 32 - Table S. Performance of trade policy-related economic variables: Mean (lines 1-7) and Median (lines 8-14) (1) 2) (3) (4) (S) (6) (7) REER BMP MIGDP CM/M NFCM8ANFM CM/Can Mfcw/Con Afier Recent AJier Recent Afier Recent After Recent Afier Recent Afier Rece After ReceNt I. SSA -22.0 -24.9 -33.4 -55.6 -.7 -0.4 0.2 0.5 II 2.6 -1.6 -1.9 -0.6 -0.4 2. CFA i -11.1 -10.S 1.4 2.0 -5.0 -6.9 -2.4 -1 .3 -0.9 0.0 -1.7 -2.6 -0.8 -1.2 3. Non-CFA -25.9 -30.2 -48.5 -82.0 0.3 3.3 0.6 0.4 0.6 2.4 -1.6 -1.6 -0.8 -0.3 4 IAL -38.1 -41.6 -162.6 -192.5 2.3 3.5 3 7 3.6 4.3 6.6 135 0.2 1 9 2.2 5. NonTAL N.A 0.4 NA. 149.4 N.A. -14.0 NA. -2.6 NA. -0.6 N.A. -11.8 N.A- -6.1 6. Other AL -26.3 -35.1 -31.9 -59.3 2.6 2.3 -131 -0.3 -0.5 1.0 0.1 0.6 -0.2 0.0 7. Non-SSA IAU -9.4 -24.6 1336 -15.8 - .0 3.2 0.3 2.2 1.6 3.2 -0. I 1.0 0.3 0.9 9. SSA -11.9 -15.6 25.0 20.7 -3.1 -1.6 -0.2 0.5 -0.2 0.5 -04 -0.3 -02 0.1 9. CFA -10.0 -9.2 2.3 2.7 -5.4 -7.0 -3.0 -2.2 -1.5 0.9 -08 -06 -06 -0.5 10. Non CFA -13.0 -22.6 2.8 0.7 -0.6 45 0.1 3.1 -0.2 0.5 .01 07 01 136 11. IAL -16.8 -21.7 -0.8 -0.4 1.7 -1.0 2.3 2.6 2.0 7.5 05 09 07 2.6 12. Non TAL N.A, -3.9 N.A. 4.3 N.A. -6.8 N.A. -2.0 N.A -0S N.A -4.0 N.A -1.3 13. OtherAL -12.2 -19.6 2.2 2.5 -2.4 -0.6 -2.4 -2.2 07 04 -0S -04 -04 -01 14. Non-SSA IAL -33.6 -22.9 2.9 -2.6 -1.2 0.3 0.9 4.1 22 3 I 00 3.1 04 1.0 All ligures denote changc in th Eil. cxcept RI-)ER figurcs ,whic percentage change. 'Aftcr column figurcs show clngeinvariaile for avrageof years the first structural adjustmcnt or tradecrelated sectoral adjustmcnt loan from the World Bank or structural adjustment facility from the IMF. compared to the average of the 3 years before the loan. 'Rcent' column figures show change in variable for the most recent 3 years. compared to the 3 years before. For thc non TAL countries, the comparison is 3 mobt reccnt years versus 1980-82 REER = Real Effcctive Exchange Rate CM = Consumption imports Con=Total consumption (decrease indicates dcpreciation) M = Total imports N.A-= not applicable BMP = Blaik market premium NFCM = Non-food consumption imports The classification of countries by adjustmcnt status is as follows: The intensive adjustmcnt lending (IAL) countries arc: Cote d'lvoirm Ghana, Guioira-lissats, Kenya. Madagascar. Malawi, Mauritania. Mauritius, Nigeria. Scnegal. Tanzania, Togo. Zambia. The non-trade adjustment lending (NTAL) countries are: Botswana. Burkina Faso, Ethiopia. Lcsotho, Liberia, Rwanda, Seychelks. Sudan, Swaziland. Thc others (OAL) are: Angola. Benin, Burundi, Cameroon, Central African Republic Chad, Congo, Gabon, Gambis Guinea Mali, Mozanibique, Niger. Sierra Leone. Somali, Uganda, Zaire. Zimbabwc The non-SSA intensiVe adjustment lending countrics (non-SSA [AL) are: Bolivia, Brazil, Chile, Colombia, Coas Rica, Jaumica, Korea. Mcxico, Morocco, Pakistan, Philippines, Thailand, Turkey, and Uruguay. This classification foilows the schemfe in the third report on adjustment lending (World Bank, 1992). except that Burkina Faso and Sudan were classificd as 'Other AL' in tbe report on adjuitment lending, but 'Non- TAL' in the classification above The adjustment loans to both these countrics in the early to mid-1980s wer fertilizer or agriculture loans, with few or no accompanying trade policy reforms. (Burkina Fasc received a SAL in 1991, but our data serics cnds in 1990 or 1991.) Sources: BESD, IMF, and World Dcvelopment Report. 1990 data. - 33 - The next 5 variables are different measures of import penetration in the domestic market. In general, it would be expected that as trade barriers are reduced, imports would increase as a fraction of (iDP.2 Column (3), the import-GDP ratio, is a commonly used measure of openness based on this idea. This measure has one serious flaw, particularly in the African context. The problem is that the first -- and sometimes only -- step in liberalization is relaxing restraints on imported capital goods, inputs, and selected consumer goods that do not compete with domestic production. Since production of import substitutes tends to be intensive in imported inputs, the step may increase imports while increasing protection.22 One way to minimize this problem is to look at the most heavily protected sub-sectors of imports. Because of the typical escalated structure of protection, these tend to be consumption goods, and especially non-food consumption goods. Columns (4) through (7) are measures of the relative composition of imports and consumption. Increasing shares of consumer imports (or non-food consumer imports) in total imports (or total non-food imports) or in consumption (or non-food consumption) would indicate reduced protection. This measure is not a perfect measure of protection: some protected import substitutes are not consumer goods and some consumer goods do not compete with domestic production and so are not protected. But in principle, this should be a better indicator of protection than are total imports. The results for SSA as a whole are quite mixed, with some indicators showing increased and some decreased openness. Within some groups, however, the results are more consistent. By almost all these measures, the CFA economies and the non-trade-adjusting economies have become more 21. How much they would increase, however, would be determined by the extent and design of the liberalization and by the degree (and speed) by which domestic producers could increase efficiency to compete with the imports. 22. In the long run, meps that increase protection will generally reduce the total value of imports, since the value of the final consumer goods imports displaced by the increased domestic production will exceed the value of the increased imports of inputs and capital gooda. This need not hold in the short run, however, since capital good imports are 'lumpy.' Even in the long run, it may not be true of observed imports, since the marginal impors displaced may be those in the black market. This is also not true of negative value added production, which is not a trivial qualification in the African context. One study concluded that 70 percent of Madagascar's industrial sector was producing negative value added, e.g., the value at border prices of inputs exceeded that of outputs. (See Jacquemont and Ass;don 1988.) - 34 - protectionist, and the intensive adjustment lending countries have become less protectionist. And, to put the numbers in the table in perspective, it should be borne in mind that the average size across all SSA countries in the sample of the M/GDP rat1io is around 40 percent; that of CM/M about 33 percent; that of CM/CON about 10 percent; and that of NFCM/CON about 6 percent. So, a I percentage point change in NFCM/CON is more significant than a similar change in M/GDP. Viewed in this light, the increases for the IAL countries in the indicators based on consumer imports look substantial. And by most measures, the openness in these countJies has continued to increase in recent years. For the "other adjustment lending" (OAL) countries, the indicators are again mixed about whether there has been a reduction or an increase in openness since the pre-adjustment period. There is, however, fairly consistent evidence that openness has increased somcwhat in recent years compared to the years immediately after the beginning of adjustment. That is, for eight of the ten import-penetration mea5 1res, the "Recent" column is larger -- less negative or more positive -- than the "After" column. 2. Quantitative measures of movements in protection rates It is in principle possible to derive a numerical estimate of changes in the tariff equivalent of all restrictions on imports. Tiiis can be done using the import demand function as follows: Let the demand for imports be M = a + bY + c[PME (I + t)]; where M is imports (in quantity, not value, terms), Y is income, PM is import price in dollars, E is the real exchange rate, and (1 + t) is the "tariff equivalent" of import restrictions, that is, a measure of the increase in domestic prices that would be needed to reduce import demand to the same degree as the import restrictions.23 If 23. In principle P. E (I +t) is the 'full' domestic price of the imported good. If it is rationed by a non price mechanism, (I +t) includes the marginal value of waiting time, bribery, or other costs incurred to purchase the good. - 35 - all variables are in natural logaiithms, then the above equation can be differentiated and re-arranged to show %A (1+t)=-{%AM-b%AY-c[%APM+ %AE]1/c, where %A is the percentage change in a variable, and b and c are the import elasticities with respect to income and price. Data is available for imports, income, import prices, and the exchange rate; and the elasticities can be estimated (or assumed on the basis of previous estimates for other developing countries),24 so the change in (I + t) can be estimated. Essentially, it is the residual change in imports that cannot be explained by changes in income or the major variables that affect domestic prices (international price and exchange rate fluctuations). The results of these calculations, using three different definitions of imports and two definitions of the exchange rate, are reported in Table 6 for the seven country groupings. As in Table 5, the 'after" column reports the change for the three years after the initiation of the structural adjustment program, relative to the three years before the program. (The year the program began is excluded.) The "recent" column reports the change for the most recent years, relative to the three years before the program. Since the focus here is on measuring changes in protection, the three measures of imports are those which exclude imports that are not produced locally, or at least are not heavily protected in most countries--food, fuel, and inputs used in domestic production of final goods. Ideally, the calculations would have been based only on imports competing (actually or potentially) with domestic production, but it was not possible to identify these. In interpreting the results of Table 6, it should be kept in mind that a decrease indicates a reduction in protection. 24. Coefficients estimated country by country were quite erratic, even in sign, probably because of the short data series available Consequently, I used coefricient. of b = 1.25 and c = -I, which are consistent with the range of values for these parameters that Pritchett (1987) estimated for a broad sample of developing countries. The demand elasticity is also consistent with those found in de Rosa's (1992) survey of other authors' estimates. - 36 - Table 6: Change in tariff equivalent of Import restrictions: mean (lines 1-7) and median (lines 8-14) (percentage change) Non-Food Non-Fuei Cons. M Cons, M M i______________________ A,lcr Recent After Recent After Recent 1. SSA -15.45 -31.2 -24.7 49.7 -27.3 -40.4 2. CFA -9.1 -16.4 -8.6 -17.5 -26.0 -30.8 3. Non-CFA -18.9 -39.5 -33.2 -66.7 -28.1 -45.7 4. IAL -43.2 -56.1 -68.1 -94.5 -50.0 -65.6 5. Non-TAL N.A. 40.4 N.A. 35.3 N.A. 13.9 6. Other AL -11.5 -38.3 -12.5 47.5 -27.2 -44.0 7. non-SSA IAL 4.4 -26.6 -9.4 -23.3 11.0 -6.2 8. SSA -12.7 -31.7 -25.8 40.3 -30.5 -36.0 9. CFA -1.2 -9.1 -2.5 -2.5 -22.4 -22.6 10. non-CFA -27.2 -37.2 -37.5 -74.3 -32.1 -37.8 11. LAL -38.0 -37.3 -54.8 -104.4 -46.7 41.2 12. non-TAL N.A. 0.9 N.A. -2.5 N.A. -2.8 13. other AL I -10.5 -37.4 -29.5 -43.3 -31.1 -36.8 14. non-SSA IAL 5.4 -23.4 2.8 -39.1 15.9 -16.0 Source: B£SD and IMF data. The results here indicate that in SSA as a whole, there was rather modest reduction in (1 +t) in the years after beginning adjustment, and protection continued to decline thereafter. The reductions were more modest for the CFA countries than the non-CFA. Protection fell the most in the intensive adjustment lending countries, while declining less in other adjusting countries and actually increasing in the non-trade-adjustment lending (non-TAI ) countries. - 37 - Table 7 displays estimates of the changes in the ratio of domestic to border prices, derived by a 4 IL adding the percentage changes in the real exchange rate and that of the estimated (I +t).25 This, in effect, measures how much relative incentive for production of importable production has changed, factoring in exchange rate and comnmercial policy. Since (1 +t) was itself estimated as a residual, the numbers in the table should not be considered independent of the numbers in the previous table, but they do show something different. What these numbers indicate is how the incentive to produce import-substitutes has changed, taking into account the effects of both "protection" and the real exchange rate on domestic prices. Table 7 Change ir incentives' for importable production: Mean (lines 1-7) and Median (lines 8-14) Non-Food Non-Fuel Cons. M Cons. M M l_____________________ ,After Recent After Recent After Recent SSA 13.1 5.4 4.0 -11.4 -0.1 -5.6 CFA 2.9 -4.3 3.8 -5.1 -13.6 -18.5 Non-CFA 18.7 10.8 4.1 -14.7 7.3 1.4 |AL 14.7 6.1 .6.6 -27.6 9.9 -2.8 Non-TAL N.A. 28.1 N.A. 23.0 N.A. 5.1 Other AL 5.3 -5.2 5.0 -12.2 -10.2 -12.4 non-SSA IAL -7.5 -55.4 -21.3 -52.1 -0.9 -35.0 SSA 9.2 -0.7 10.4 -11.1 -8.6 -11.7 CFA 7.5 -0.7 11.9 3.3 -20.9 -15.8 non-C,-A 22.9 -0.7 9.4 -16.6 1.7 -6.5 IAL 10.7 1.5 5.7 -12.9 8.4 -11.7 non-TAL N.A. 22.9 N.A. 19.4 N.A. -4.8 other AL 7.1 -4.9 11.9 -30.7 -14.9 -9.2 non-SSA IAL 8.1 -34.8 -5.5 -56.1 10.8 -33.6 a. 5%ARlER+%A(1 +t) Source: BrSD and IMF data. 25. If PM is import border price, ER the exchange rate and PD the domestic price, then P0 = (PM) (ER) (I +l) and the ratio of the domestic to border price is (ER) (I +t). Thus, % A(PD/PM) % A (ER) + A(I +t) - 38 - Together, the evidence in Tables 5-7 seems to point to the following tentative conclusions: Protection of impott substitutes by tariffs and non-tariff barriers (NTBs) in Sub-Saharan Africa asa whole has declined. Relative to what were apparently very high levels of protection before the reforms began, the original reductions were moderate, though they have become more significant in recent years. Both the initial and subsequent reductions in the intensive adjusters were much greater than in other countries. Concurrently with the reduction of tariffs and NTBs, however, most countries were devaluing currencies in real terms. This was even true of CFA countries, though the extent of devaluation was much less than in others. Consequently, the total incentives to produce import substitutes relative to nontradables did not decline nearly as much as the protection rates would indicate, and may have even increased in some groups. This may partially explain why the import penetration-based policy indicators (Table 5) do not show more improvement. IV. Conclusions and implications Adjustment programs in SSA have been somewhat less intensive in trade policy reforms than have the programs in other countries. Implementation of trade policy reforms overall has been better in SSA than in other countries, though the opposite is true of the most important reforms. There was also a higher frequency of retrogression in SSA and slower continued progress after adjustment loans were fully disbursed. As noted at the outset, the experience of different African countries in implementing trade policy reforms has been quite diverse. As a group, the intensive adjustment lending countries made significant progress in the 1980s and early 1990s by almost any measure. Even within this group, however, there was significant variation, with countries like Mauritius and Ghana making much greater strides than others, and some making progress that was later reversed, either dramatically (Madagascar, Zambia) or incrementally (Kenya, Senegal, Nigeria). - 39 - By virtually every measure, the CFA countries as a group have not been as successful as the non- CFA ccuntries in implSi ntinrtradepolicy reforms. This was indicated directly by their higher rates of underimplementation of policy actions (particularly "critical" actions) and indirectly by the lower values of the measures of increased openness for these countries. For SSA as a whole, progress is more impressive in recent times than in earlier years. In many countries, adjustment did not begin until the mid-1980s (as measured by the first IMF or World Bank adjustment operation) and relatively few measures were implemented up front. Underimplementation rates of policy actions are lower based on the most recent information than on information as of the last tranche release of adjustment loans. The measures of changes in openness almost all show greater improvement (compared to pre-adjustment values) in 1989-91 than in the years following the beginning of the adjustment period in each country. Anecdotal evidence supports this conclusion, and indicates more progress ir 1991 and 1992 in some countries that would not be picked up in our data (e.g., Zambia, Tanzania). For the CFA countries, however, the evidence on this is mixed. Their underimplimentation rates are lower in the most recent data, and by some measures-but not all-their openness has improved more in recent years. But by virtually all measures, the improvements relative to earlier periods have not been as great as for non-CFA countries. Reduction in trade regime-based protection was largely offset by real devaluation in most country groups, and by most measures, incentives to produce import substitutes actually improved in the years immediately after the first adjustment loan relative to the pre-adjustment period. In more recent years, the incentives have fallen, relative to the post-loan period, though the fall has been modest. This has some relevance to the debate on whether trade policy reform has proceeded so quickly that some countries are in danger of "de-industrialization" from the rapid disprotection of import substituting industry. The evidence here argues that in general this is not the case: total incentives have fallen only - 40 - modestly, if at all. As with all averages, of course, these may obscure somespecific cases (countries or industries) where the experience differs. Tne evidence on changes in protection and in overall incentives may also illustrate why the CFA has a relatively poor record in implementing reforms. In spite of the fact that the CFA countries showed poorer implementation of reforms and smaller reduction than non-CFA countries in trade regime-based protection, they showed larger declines in overall incentives for import-substitute production. This, of course, was due to the fact that their rate of real devaluation was much lower. There may be a direct connection here: their ability to reduce tariffs and non-tariff barriers to imports may have been impeded by their inability to offset these reductions by devaluations, as other countries did. This general principle is well understood. But the quantitative difference between the CFA and non-CFA countries is interesting. Non-CFA countries reduced tariff-equivalent protection in "recent" years by 15-49 percentage points (depending on the measure used) more than CFA countries, while incentives declined by 15-20 percentage points more in the CFA countries. An important issue, distinct from the question of how much progress has been made, is how open the trade regimes are at this point in the adjustment process. Put another way, the question is how much more remains to be done. All of the quantitative indicators in this paper are measures of changes, so by themselves they are not capable of answering this question. To draw any inferences on this, we have to rely on other information as well. The indicators of change in tariff equivalent protection tell us that this fell between 30 and 50 percentage points in SSA as a whole between the beginning of adjustment in each country and the period 1989-91. This is not a trivial decline. However, given what we know about the stringency of import controls in the pre-adjustment period-especially on the consumer imports upon which these estimates of protection reduction are based-it would not appear that this is sufficient to reduce the protection to moderate levels, relative to deep reformers in East Asia o. Latin America. - 41 - This conclusion is reinforced by what we know about specific trade regimes. Perh'ps the most A 4' important problem is with foreign exchange allocation. Mauritius may be the only non-CFA country in SSA where foreign exchange allocation has been market driven, (i.e., the currency is essentially convertible at a rate mimicing a free-market rate) for a substantial period of time. In other countries, this very basic reform has not begun, or has been only recently completed (Ghana). Several countries, most notably Ghana and Uganda, have reasonably well-functioning auctions. However, since they are of the "Dutch" variety, they impose implicit taxes on inframarginal bidders; thus, different users of foreign exchange pay different prices, unlike a convertible currency market, though the size of the tax is limited by other legal channels for foreign exchange purchase. In countries undertaking phased currency reforms, the first step (liberalizing non-competitive imported inputs and capital goods) increased effective protection, though eventually the liberalization began to affect competing imports. Likewise, only a few countries are without import licensing requirements for a substantial portion of competing imports . CFA countries have a (more or less) convertible currency, but have moved slower than others in reducing other non-tariff and tariff barriers. On the export side, progress has been made in producer pricing for traditional crops, but parastatal marketing enterprises have generally kept producer prices de-linked from border prices and prevented producers from realizing the full benefits of exchange rate adjustments. For non-traditional exports, retention schemes have been helpful, but licensing requirements connected to foreign exchange repatriation regulations remain a problem. And there has been little progress on establishing efficient systems to give exporters access to inputs at internationally competitive prices, either through export processing zones, which are still in their infancy, or through duty drawback or exemption schemes, which have long exisited, but remain plagued with disbursement delays and consequent low rates of use. - 42 - R%ferences Commission of the European Communities, 1992. Trade Policy Reform in Sub-Saharan Africa (draft- October). De Rosa, D.A. 1992. "Protection and the Own-Funds Window in Tanzania: an Analytical Framework and Estimates of the Effects of Trade Liberalization," Journal of African Economies 2 (1):24-47. Jacquemont, P., E. Assidon, and A. Honvo Akanni, 1988. Exchange Rate Policy and Adjustment in Africa. Paris: Ministry of Cooperation and Development. Lyakurwa, W.M., 1991. Trade Policy and Promotion in Sub-Saharan Africa, (May). Nairobi: African Economic Research Consortium. Mosely, P., J. Harrigan and J. Toye. 1991. Aid and Power: The World Bank and Policy-based Lending. Pritchett, L. 1987. "Import Demand Elasticities in LDCs: Estimates and Determinants," mimeo, World Bank Trade Policy Division (November). World Bank. 1988b. Adiustment Lending: An Evaluation of Ten Years of Experience, Policy and Research Series 1, World Bank, Washington, D.C. World Bank, 1990. Adjustment Lending: Policies for Sustainable Growth, Policy and Research Series 14, World Bank, Washington, D.C. World Bank, 1992. Adjustment Lending and Mobilization of Private and Public Resources for Growth. Policy and Research Series 22, World Bank, Washington, D.C. World Bank. 1992b. "Building a Competitive Edge in Sub-Saharan African Countries," Industry Series Paper No. 57, Industry and Energy Department (April). - 43 - b¶ox 1. Reform Measurement: Caveat Lector The trade policy reform indicators on which this paper relies are of three types, each with its own strengths and shortcomings. One indicator (section III A) is anecdotal evidence of the experience of individual countries, drawn from various sources, including the case studies carried out in connection with this study These examples are illustrative and make certain abstract points more concrete. They cannot, of course, prove or disprove any general propositions. The second, used in Section III B, is derived from data in the World Bank's Adjustment Lending ConditionaLity and Implementation Database (ALCID). The ALCID contains information on most of the adjustment lending operations of the World Bank since 1980. Each policy action associated with each loan is classified as to the type of policy it involves, its purpose, its legal status with respect to the loan, its planned timing relative to the loans disbursements, its importance (whether it was "critical" to the success of the adjustment effort), and (for a smaLler sample) its implementation status. The main weakness of measures based on ALCID data is that summary statistics - for example, percentage of planned actions that were actually well-implemented - cannot control for the relative importance of some actions. This problem is potentially important, especially when restrictions are redundant; it is in principle possible to remove 99% of all trade restrictions but still have a very closed regimc if the remaining S% (e.g., exchange rationing) are the ones that are the most "binding." This problem is mitigated by the ALCID's distinction of "critical" versus "non- critical" actions, but, at the least, some nuances are lost; some "non-critical" actions are more important than others, and this is not captured. Other problems with the ALCID are the limited number of actions coded for implementation, especialy in recent years, and the rather broad (and sometimes ambiguous) categories used for classifying different types of actions. Its advantage, however, is its comprehensive coverage, which enables construction of a broader picture of reforms all over SSA, as well as a comparison with other regions. A third type of trade policy reformn implementation indicator used in Section III C is the openness of the trade regime. This can be measured using several types of data. The ones that are available for a large sample of countries relate to real exchange rates, black market premia, and measures of import penetration. These are to a large extent indirect measures of the openness of the trade regime, since they reflect the results of trade policy, rather than the policy per se. Black market premia, for example, are affected by poLitical and capital market conditions as well as by exchange rationing for imports. Likewise, imports are affected by relative prices and aggregate demand variables as well as by exchange controls and commercial policy. Import measures are also sometimes criticized because changes in these may reflect exogenous changes in foreign exchange availability, rather than in underlying policy. This has some validity, but to the extent foreign exchange is used to import goods competing with domestic products, changes in its availability do measure changes in protection. At P rate, in the absence of systematic time-series data on the policy instruments per e, these are useful measures of the extent to which protection of import-substitutes has been reduced. Thus, they complemcnt the ALCID data, which can measure the extent of reform, but not (directly) its effectiveness in reducing the bias. Furthermore, some of the problems mentioned above can be partially resolved by the techniques described in Section III C. For example, the estimation of "tariff equivalent" incorporates a correction-albeit somewhat crude-for the effects of import prices and aggregate demands on the demand for imports. -44 - A Policy e"sarch Working Paper Serles v Contact Title Author Date for paper WPS1 194 How Fast Has Chinese Industry Tom Rawski September 1993 E. Khine Grown? 37471 WPS1195 The Enterprise Sector and Mark Schaffer September 1993 E. Khine Emergence of the Polish Fiscal 37471 Crisis, 1990-91 WPS1 196 Corporate Tax Structure and Jeffrey Bernstein September 1993 C.Jones Production Anwar Shah 37754 WPS1 197 Determinants of Inflation among Bruno Boc3ara September 1993 C. Jones Franc Zone Countries in Africa Shantayanan Devarajan 37754 WPS1 198 Enterprise Reform in China: The Natalie Lichtenstein September 1993 M. RFngua Evolving Legal Framework 81710 WPS1199 Public Pension Governance and Olivia Mitchell October 1993 D. Evans Perlormance: Lessons for Developing 37496 Countries WPS1200 The Lite-Cycle Distributional Jane Falkingham October 1993 D. Evans Consequences of Pay-As-You-Go Paul Johnson 37496 and Funded Pension Systems WPS1201 Five Criteria for Choosing among Margaret E. Grosh October 1993 M. Oufteo Poverty Programs 37792 WPS1202 Privatization and Foreign Investment Frank Sader October 1993 Rose Vo in the Developing World, 1988-92 33722 WPS1203 Determinants of Value-Added Tax Zeljko Bogetic October 1993 F. Smih Revenue: A Cross-Section Analysis Fareed Hassan 36072 WPS1204 Structural Adjustment, Economic Nisha Agrawal October 1993 K. Rivera Performance, and Aid Dependency Zafar Ahmed 34141 in Tanzania Michael Mered Roger No,d WPS1205 Wage and Employment Decisions Simon Commander October 1993 0. del Cid in the Russian Economy: An Analysis Leonid Liberman 35195 of Developments in 1992 Ruslan Yemtsov WPS1206 Empirical Perspectives on National Ishac Diwan October 1993 A. Yideru Index Funds Vihang Errunza 36067 Lemma W. Senbet WPS1207 Characteristics and Performance Bill H. Kinsey October 1993 H. Binawanger of Settlement Programs: A Review Hans P. Binswanger 31871 10 Pollcy Research Working Paper Series Contact Ttle Author Date for paper WPS1208 Primary School Achievement in Levi M. Nyagura October 1993 I. Conachy English and Mathematics in Abby Riddell 33669 Zimbabwe: A Mufti-Level Analysis WPS1209 Should East Asia Go Regional? Arvind Panagariya October 1993 D. Ballantyne No, No, and Maybe 37947 WPS1210 The Taxation of Natural Resources: Robin Boadway October 1993 C. Jones Principles and Policy Issues Frank Flatters 37699 WPS1211 Savings-lnvestment Correlations Nlandu Mamingi October 1993 R. Vo and Capital Mobility in Developing Countries 31047 WPS1212 The Unks between Eoonomic Policy Ravi Kanbur October 1993 P. Attipoe and Research: Three Examples from 526-3003 Ghana and Some General Thoughts WPS1213 Japanese Foreign Direct Investment: Kwang W. Jun November 1993 S. King-Watsor Recent Trends, Determinants, and Frank Sader 33730 Prospects Haruo Horaguchi Hyuntai Kwak WPS1214 Trade, Aid, and Investment in Sub- Ishrat Husain November 1993 M. Youssef Saharan Africa 34637 WPS1215 How Much Do Distortions Affect William Easterly November 1993 R. Martin Growth? 39065 WPS1216 Regulation, Institutions, and Alice Hill November 1993 D. Evans Commitment: Privatization and Manuel Angel Abdala 38526 Regulation in the Argentine Telecommunications Sector WPS1217 Unitary versus Collective Models Pierre-Andre Chiapxori November 1993 P. Attipos of the Household: Time to Shift the Lawrence Haddad 526-3002 Burden of Proof? John Hoddinott Ravi Kanbur WPS1218 Implementation of Trade Reform in John Nash November 1993 D. Ballantyne Sub-Saharan Africa: How Much Heat 37947 and How Much Light? WPS1219 Decentralizing Water Resource K. William Easter November 1993 M. Wu Management: Economic Incentives, Robert R. Hearne 30480 Accountability, and Assurance WPS1220 Developing Countries and the Bernard Hoekman November 1993 L. O'Connor Uruguay Round: Negotiations on 37009 Services