How to identify trade liberalization episodes: an empirical study on Kenya

Although conceptually fairly simple, it is far from being straightforward to identify trade liberalization episodes empirically. Trade policy reform is typically a complex exercise, involving removal of, or reduction in quantitative or administrative import restrictions,' temporarily replaced by tariff surcharges, unification and subsequent reduction of tariffs, and introduction of direct incentives to exporters (Thomas and Nash 1991a). Despite the difficulty, unambiguous identification of liberalization episodes is important if we wish to study the economic performance of liberalizers vis-a-vis sustaining protection. In the economies where import restrictions abound, the net effect of a reform package on relative prices is likely to be uncertain ex ante as the level of protection of various industries may be unknown and may vary considerably from sector to sector. Furthermore, depending on changes in other variables over time, such as production costs, world prices, or domestic demand, import restrictions become more or less binding. Therefore, the government may have a great difficulty in matching the removal of non-tariff restrictions with the desired level of temporary tariff surcharges, or devaluation. Ex post the effect of the reform on relative prices may be as intended (that is, an increase in the price of exportables relative to both importables and nontradables), perverse, or it may have no effect at all, if other factors override the effects of changes in trade policy instruments. In the literature on trade liberalization and economic performance it is often disputed whether liberalization should be defined as a move to neutrality of relative prices, or as a move towards free trade with less government involvement than before. In the former case, introduction of export subsidies would qualify as liberalization, whereas according to the latter definition, the trade regime would be considered to have become less liberal. Another issue of disagreement is whether or not to include devaluation in the definition of trade liberalization. Devaluation often accompanies trade reforms to make import liberalization macro compatible, safeguard external reserves, and to provide improved incentives to the export sector. If devaluation is carried out without any changes in trade policy instruments, its effect on the average relative price ratio of importables to exportables, or the ratio of domestic price of importables to their world price is ambiguous. The effect depends on the type of trade policy instruments being used, such as value or quantity quotas. In Africa, foreign exchange rationing for balance of payments purposes is closely linked to import controls which provide protection for domestic industries. When devaluation is combined with a change in the mechanism of allocating scarce foreign exchange, such as introduction of an auction, the subsidy which imports of capital and intermediate goods often enjoy due to preferential access to foreign exchange, combined with an over-valued currency, will be removed, and imports of final goods will no longer be discriminated against through non-issuance of licenses. This clearly qualifies as a move towards free trade. A recent comparative study on trade liberalization, which was undertaken by the World Bank, and which covers 19 countries, does not employ a quantitative or objective measure for identifying episodes but uses instead a subjective liberalization index (Michaely et al. 1991). Although no unambiguous criteria common to all countries was used to determine when a liberalization episode began, or when an episode can be considered to have been completed, stopped or reversed, the study, nevertheless, claims to include all significant liberalization episodes of a minimum duration of two years, implemented by developing countries from World War II until 1984. The subjective criterion, which is an annual liberalization index, ranging from one (highest possible degree of trade intervention) to 20 (complete liberalization), was calculated for each country, based partly on various quantitative criteria, such as effective rate of protection, actual tariff rates, real exchange rate, degree of openness (sum of imports and exports over GDP), gap between foreign and domestic terms of trade, and so forth, and partly on judgment by the individual researcher in charge of a particular country study. It is obvious that the indices are not at all comparable across the sample. Country studies also lack consistency in the way in which various policy measures are assumed to impact on changes in the index. As pointed out in two reviews of the World Bank study (Collier 1993; Greenaway 1993), the use of a subjective index makes interpretation of empirical findings much more arbitrary than if a common quantitative criterion had been used. The country study on Chile, for example, based the subjective liberalization index on five indicators of trade controls: an import restriction index (assigned qualitatively on the basis of changes in QRs over time), the ratio of effective exchange rate to nominal exchange rate, the ratio of black market rate to official exchange rate, an export quota index, and the implicit tariff rate. An iterative process, including judgment when in doubt, was then used to derive the liberalization index based on these indicators. Although the authors of the study consider the implicit tariff as the ideal index of trade liberalization, they chose not to use it alone as information on its actual levels is available with significant coverage for a few years only. As will be shown below for Kenya, it is possible to derive the average implicit tariff as an index, without full information on its actual levels over time, by using average domestic and world price indices for importables. This is by no means an easy task, however, as the available price indices are prone to biases which are difficult to control and sometimes even to detect. The World Bank study completely excludes Sub-Saharan Africa. As there have been attempts to liberalize trade regimes in Africa in the 1970s and particularly in the 1980s, application of common quantitative criteria to the identification of liberalization episodes to all developing countries could have resulted in inclusion of a few of these episodes in the study. This would have been useful from the viewpoint of policy recommendations as more attention would certainly have been devoted to the negative effects of incredibility suffered by many of these reforms. By anticipating a reversal, the private sector not only inflicts additional welfare costs arising, for example, from excessive accumulation of stocks of importables but speculation may also deplete external reserves, endangering the entire process of liberalization. The main purpose of the rest of the paper is to derive the implicit tariff index for Kenya, and to assess its validity and reliability as an objective criterion for identifying trade liberalization episodes in African economies, including an assessment of the quality of the available price data.` To place the chosen index into context, Section 2 will provide a review of quantitative measures of trade policy. In Section 3 we will construct our base case implicit tariff index using the most relevant domestic price deflator and price index for the corresponding categories of imports. This index will then be compared to (i) a narrative of changes in trade policy in Kenya since the early-1970s, and to (ii) a measure of import compression developed by Narasimhan and Prichett (1993). A few snapshots of the actual levels of protection, both the nominal rate of protection ( implicit tariff) and the effective rate of protection are available for Kenya over time. The former assesses protection on the price of the good, while the latter measures the protection given to the value added, or the domestic factors of production. Section 4 will examine whether these snapshots concur with the changes in trade policy as indicated by the implicit tariff index, the narrative, or by the index of import compression. Section 5 contains an assessment of the quality of domestic price deflators that are available for constructing the implicit tariff index, while Section 6 examines the reliability of corresponding import price indices. Section 7 concludes.