Export diversification by developing countries into manufactures is one of the more important developments of the past two decades. A fundamental question is why this change has occurred in some developing countries and not in others. Based on a review and an extension of recent economic literature on export diversification, this study addresses the question by examining the role of both the internal and external economic environment.

Export diversification by developing countries into manufactures is one of the more important developments of the past two decades. A fundamental question is why this change has occurred in some developing countries and not in others. Based on a review and an extension of recent economic literature on export diversification, this study addresses the question by examining the role of both the internal and external economic environment.

For many years, there has been considerable interest in the role that export diversification can play in reducing the variability of export earnings of developing countries. More recently, with the emergence of the debt crisis in the early 1980s, the focus of discussion has shifted to the contribution of export diversification to raising the growth rates of both exports and domestic output. The renewed interest in export diversification also stems from the erratic swings and secular decline in primary commodity prices since the early 1970s.

The paper has five main sections. The first section gives evidence on recent trends in export diversification into manufactured goods in developing countries. The second section examines, in three subsections, the role of diversification in the growth of export earnings, the stability of these earnings, and growth of domestic output. The role of diversification—particularly into manufactures—in raising the trend of export earnings is reviewed in the first subsection and the validity of the underlying demand and supply assumptions is investigated; in the second subsection, diversification as a means to reduce earning instability is reviewed, based on recent empirical literature, analysis of the underlying causes of instability, and country experience; the final subsection investigates the interdependence among export diversification, export growth, and domestic growth, as well as the relationship between export diversification and the efficiency of resource allocation. The third section examines the effect of macroeconomic policy and supply-side incentives on export diversification, as well as on the growth of exports and output. The discussion draws upon selected macroeconomic indicators for groups of developing countries and on case studies of four countries—Malaysia, Côte d’Ivoire, Ghana, and Argentina. The final section presents the paper’s main conclusions.

Patterns of Export Diversification

Export diversification can be described as an increase in the number of distinct products in the export base, combined with a reduction in dependence on any one product as a source of foreign exchange earnings. Such diversification in many countries can be described as a three-stage process coinciding with the stages of economic development itself.1 First, export production is diversified from a few primary commodities to a wider range of commodities. This process is typical of a country that does not have a well developed manufacturing sector. For example, Guyana—traditionally heavily dependent upon exports of sugar, rice, and bauxite for its foreign exchange receipts—has diversified into fish, shellfish, and timber during the past two decades.

Second, export production is diversified from primary products into a wider range of goods that includes the production of manufactures for export. This process generally takes considerable time, and has two distinct phases. The first phase consists of the processing of primary commodities and light assembly or rudimentary manufacturing industries, while in the second phase, production shifts, on a larger scale, to more sophisticated manufactured goods. Côte d’Ivoire is an example of a country that had completed the first phase by the mid-1970s and that is now beginning to move into the second. It has achieved high growth rates in its coffee, cocoa, and log exports and has successfully moved into processing these products as well as into new natural-resource based exports, including palm products, cotton, and fruit. It has also initiated the production of such manufactured exports as chemicals, plastics, and textiles. As the development process continues into this second phase, diversification is accompanied by improved technology, economies of scale, and surpluses for reinvestment.

In the third stage of the diversification process, exports become still more diversified, to include services such as financial services, insurance, commission and agency fees, construction services, communications services, royalties, and management fees. While this third stage of diversification is more relevant to economically developed countries than to developing countries, a few of these—such as Brazil, India, Korea, and Singapore—have begun to move into this stage. (It should be noted that earnings from some services accrue to countries at all stages of development. For instance, foreign exchange earnings from tourism and workers’ remittances are recorded by countries at all three stages of export diversification, as are freight and passenger transportation services, which are particularly significant for countries that earn foreign exchange from port services or flags of convenience, such as Panama and Liberia.)

The process of export diversification may seem, at first, to contradict the Ricardian concept of comparative advantage, where maximum gains from trade can be attained by specialization. However, when the theory of comparative advantage is set in the context of a dynamic world economy—where erratic swings in world market prices, trade barriers in foreign markets, and lack of perfect foresight are present—producing and exporting a variety of products may well in fact maximize domestic growth and foreign exchange earnings. Moreover, the comparative advantage of countries has rarely remained the same over time. Changes in world prices, in relative domestic prices and costs, and in technology can affect both the structure of the domestic economy and the competitive position of domestic producers. For example, comparative advantage in natural rubber production shifted from Brazil to Asian countries in the early 1900s as new seed varieties and plantation cropping evolved. Similarly, the development of synthetic rubber has again altered the comparative advantage previously enjoyed by Asian natural rubber producers.

Over the past 25 years, a major shift in the production of developing countries as a group has taken place; the share of primary products in total output has declined, while that of manufactures, particularly manufactures for export, has expanded. The shift into manufactures has been much greater for the medium-income and newly industrialized developing countries, although the share of manufactures has also risen in some of the least developed countries (Table 1).2 In virtually all of the countries for which data are available there has been an increase in the share of manufactures in total output between 1965 and 1981.3

Table 1.

Selected Developing Countries: Manufactured Exports, Manufacturing Output, Export and GDP Growth, 1965–81

(In percent)

article image
Source: United Nations SITC data, from World Bank, World Bank Development Report (Washington, 1986); and International Monetary Fund World Economic Outlook (Washington, 1986). Regional aggregates are weighted averages where the weights used are the country share in the relevant variables.Note: Data limitations prevent strict comparability of statistics. The definition of manufactures for the purpose of classifying exports is often broader than that used for the sectoral classification of output. Manufactured exports include re-exports and gross exports from free trade zones.


During the period 1965 to 1984 these countries diversified out of manufacturing into services.

All countries have, therefore, had a structural change in the source of their export earnings over the last two decades, albeit in varying degrees, as diversification from primary products into manufactures—as well as diversification among manufactures—has taken place. Export volume growth, however, has been the fastest in regions and countries for which manufactures constitute a high proportion of exports. Asian countries, which have been the most successful in exporting manufactures, have also had the fastest growth of export volume, while low export volume growth has been characteristic of African countries, which tend to be highly dependent on primary commodities. Outside the oil exporting countries, diversification has also occurred within primary product exports, with the proportion of countries dependent upon a single commodity for more than half of their primary export earnings falling. Despite this, the increase in the volume of primary product exports has been much slower than that for manufactured exports, particularly for African countries, whose experiences can be characterized, to varying degrees, by: (1) increased concentration in products such as coffee, cocoa, and tea, for which world demand expanded slowly; (2) policies that effectively transferred resources from the agricultural sector to the rest of the economy; (3) lack of investment, research, and agricultural extension services; (4) limitations on essential imported intermediate inputs precipitated by foreign exchange constraints; and (5) loss of traditional European markets as agricultural output expanded in those countries. There are, of course, some individual countries whose experience has differed from this pattern in one or more respects.

Effects of Export Diversification on Export Earnings, Earning Stability, and Economic Growth

Diversification and Growth of Export Earnings

While export diversification has been generally found to contribute to the growth of export earnings, to the reduction of instability in these earnings and to increased domestic economic activity, the causal linkages are both more complex and less certain than might initially be supposed.

Trends in Export Earnings

Trade in primary commodity exports, especially agricultural commodities, declined in relative importance for developing countries from 1965 to 1980 as market shares were lost to industrial countries, reflecting both increased price and non-price competition (such as trade credits offered by exporting industrial countries). The shift in some industrial countries from importing to exporting food paralleled the mounting agricultural surpluses that resulted from domestic pricing policies. However, the failure of developing countries to diversify both agricultural exports and markets also contributed to their shrinking share of world trade. Demand for world foodstuffs and beverages increased significantly after 1973, reflecting the rapid increase in demand for food by the Organization of Petroleum Exporting Countries (OPEC), the newly industrializing countries, and nonmarket economies. This higher demand, however, was increasingly satisfied by industrial countries. Moreover, these countries, on the whole, also met the rising demand for high-value exports such as meat, poultry, dairy produce, fruit, and vegetables, with the biggest relative increase in exports coming from the European Community. The performance of developing countries in these growing export markets for agricultural goods was quite mixed and varied widely across regions and countries. The successes recorded by developing countries were for exports of nontraditional crops, such as soybeans.

Manufactured exports from developing countries, however, grew slightly faster than those from industrial countries, although from a much lower base (Table 2). This shift in trade partly reflected the growing trend in developing countries to export processed food and raw materials that were formerly exported without processing. Nevertheless, despite this gain by developing countries, total exports of the industrial countries grew somewhat faster than those from developing countries, reflecting the relatively slower growth of primary product exports of developing countries.

Table 2.

Exports from Developing Country Regions, Industrial Countries, and OPEC to the World, 1965 and 1980

article image
Source: Bond (1987).Note: Data in parentheses are annual average growth rates.

Excluding trade by non-market-oriented countries that are not members of the International Monetary Fund.

These aggregate results, however, conceal substantial regional differences. Export earnings from primary commodities in the developing world grew, on average, fastest in Asia, while the slowest rates of growth took place in Africa and the Western Hemisphere. (These data exclude OPEC as a group.) Furthermore, the growth of Asian earnings from exports of manufactured goods also outpaced that of developing countries as a group.

Role of Diversification in Growth of Export Earnings

A country in the process of diversification will find its export growth affected not only by the growth of activity in the industrial countries but also by other exogenous variables, such as changes in international prices of commodities relative to those of manufactures, the composition of its exports, the income elasticity of demand for its exports, its geographical location, and the export prices of its competitors. A country’s success in realizing diversified export growth depends fundamentally upon the types of commodity exported and how world demand for each type evolves. For example, prices of manufactures have had a higher trend than prices for food and raw materials, while the income elasticity of demand for minerals and metals tends to be greater than that for foodstuffs.

In addition, the growth of exports also depends upon the location of the exporter. For example, the European Community is the principal market for most of Africa’s commodity exports because of Africa’s location (and historical trade links). Partly because of the expansion of European agriculture in the last decade, Africa’s exports of certain products to this region declined. By comparison, most of the markets for Asian exports have been growing rapidly, reflecting both increased regional trade between Asian countries with rising per capita incomes as well as the diversification of Asian exports into manufactured goods for which demand has increased faster than for primary products.

Finally, the trade and agricultural policies of industrial countries influence the export options open to developing countries. As a result of these policies, developing countries have either lost market shares in certain products or have been prevented from expanding into traditional markets in the industrial world. For example, sugar quotas in the United States and the European Community have sharply limited market access, and the Multifiber Agreement has constrained the expansion of trade in textiles. The incidence of such policies varies among products, and policies differ among importing countries. For this reason, both the composition and direction of export trade exert a crucial influence on the potential growth rate of exports.

Long-run trends in net barter terms of trade. It has been argued, based on an observed decline in the prices of primary products relative to those of manufactures, that diversification from exports of primary products into manufactured exports can raise the long-term trend of a country’s export earnings. A consistent trend in the net barter terms of trade, however, depends upon three assumptions: (1) that the supply elasticity for primary products is lower than that for manufactured goods;4 (2) that the income elasticity of demand is higher for manufactured goods; and (3) that each developing country is a price taker for its main exports. (It is important to distinguish between price takers and countries with dominant market positions, such as Chile in copper or Zaire in cobalt, where the terms of trade decline as a consequence of higher export volume.)5

The empirical question of whether primary commodity prices, in fact, are in long-term decline remains a subject of controversy. The early findings of Prebisch (1950)6 and Singer (1950), that prices of primary products were falling secularly relative to those of manufactured goods, were attributed to different market structures in developing countries (assumed to be primary product exporters) and developed countries (taken to be manufactured goods exporters). The observed long-term deterioration in the net barter terms of trade implied a deterioration in the purchasing power of developing countries’ exports and suggested—together with the trade pessimism arising from the experiences of the Depression—the adoption of development strategies based on import substitution. Lewis (1969), however, found no such long-term deterioration in primary product prices. For tropical products, which constituted the bulk of developing countries’ exports, he found no discernible trend for the period 1871–1965. Lewis was, nonetheless, concerned about the poor prospects for tropical product prices in the 1960s. Significant technological advance in food production in the industrial countries had raised agricultural productivity and supply, thus depressing food prices. The pessimistic prognosis for developing countries’ tropical exports reflected the widening gap in factor productivity between developed and developing countries.7 Diversification of a country’s production and export base from tropical products to manufactured goods was proposed as a means by which developing countries could raise the average productivity of their factors of production.

Although efforts have been made recently to refine the price series on manufactured and primary exports, they do not show conclusive evidence of a secular decline in the net barter terms of trade of developing countries. Kravis and Lipsey (1971, 1981) demonstrated an upward bias in the United Nations’ manufactured price series since the index did not adequately reflect changes in product quality or shifts to lighter products. Spraos (1980) pointed out that quality changes in primary products are also not taken into account when goods exported shift from low-quality products, such as coffee beans, tobacco, or short-staple cotton, to varieties that command a higher price. If quality adjustments are made to both primary and manufactured goods, the bias in the net barter terms of trade was shown to be indeterminate. More recently. Saps-ford (1985) found significant statistical evidence of a long-term declining trend in the net barter terms of trade using the aggregate price index of the International Monetary Fund, which is based on 39 market transaction prices for 34 non-fuel primary commodities. Once account was taken of the disruptions to trade during the First and Second World Wars, primary commodity prices were shown to exhibit a declining trend relative to those of manufactures between 1900 and 1982, providing strong support for the Prebisch-Singer hypothesis. Since 1982, the net barter terms of trade between non-oil primary commodities and manufactured goods have continued to fluctuate around a declining trend (see chart 1 and Appendix III, Tables 17 and 18).

Chart 1.
Chart 1.

Real Prices of Primary Commodities, 1957–85


Source: International Monetary Fund, International Financial Statistics.Note: All prices are deflated by the manufacturing price index for industrial countries.

Focusing on the long-term trend of relative prices between primary products, as a group, and manufactured goods suggests that countries seeking to raise the long-term growth of export earnings should shift out of primary products and into manufactured goods. In fact, there is no statistically significant evidence to support the premise that prices of all primary products have declined secularly relative to those of manufactured goods. Based on disaggregated data for selected commodities, only prices of food products and metals showed a significant secular decline relative to manufactured goods prices; the trend path of the real prices of beverages and agricultural raw materials was not significantly different from zero at the 5 percent level of confidence (see Appendix III, Table 18 for regression results).

Policy decisions to encourage diversification, however, must take account not just of past trends but even more so of prospects for future changes in underlying supply and demand conditions. The volume demanded of developing country exports will be directly related to the level of real income in importing countries; however, income growth differs across export markets. Growth in the export market also depends on the income elasticities of demand for the product. Earnings growth, consequently, depends upon the country the export is going to and on the product being exported.

Differential income demand elasticities. There is extensive literature on the income elasticity of demand for imports by the industrial countries8. Consensus estimates for the industrial countries are that income elasticities of demand for total imports range between 1.25 and 2.25 (Goldstein and Khan (1982)). Estimates of income elasticities of demand for total imports from developing countries range between 1.5 and 2.0 and thus are consistent with those for the industrial countries (see Hamilton and Kreinin (1980). Cline (1984). and Bond (1986, 1987)).

However, these ranges refer to elasticities for total imports. In growing markets, diversification of exports can change the long-term trend of export earnings of a developing country because income elasticities of demand vary according to the commodity composition as well as the country composition of import demand. Preliminary econometric results for manufactured exports suggest that, on average, income elasticity of demand is higher for these goods than for exports of non-oil primary products (Table 3). Thus, an increase in the rate of growth of world income will increase the volume of manufactured exports proportionately more than non-oil primary products. However, these preliminary estimates show that income elasticities for exports of manufactures are not consistently larger than the income elasticities for primary products. For example. African and Middle Eastern countries face lower global income elasticities of demand for manufactured exports than those for food. Manufactured exports from these regions are generally concentrated in semi-processed primary products that seem to face an income elasticity of demand that is lower than that for their unprocessed counterparts.

Table 3.

Estimated Income Elasticities of Demand for Volumes of Exports from Developing Countries, by Commodity and Region

article image
Sources: Goldstein and Khan (1980); and Bond (1986, 1987).

There is also a considerable literature estimating the income elasticities of demand by industrial countries for specific imports of manufactures and primary products (Table 4). Demand elasticities for specific manufactured goods are higher for products such as engineering equipment, sophisticated machinery, and luxury goods than they are for products such as textiles and leather goods. This supports the view that, in general, income elasticities are higher for relatively capital-intensive products than for relatively labor-intensive products. Moreover, the income elasticity of demand in industrial countries for manufactures from developing countries appears to be growing over time, as developing country manufactures become more capital intensive.

Table 4.

Selected Industrial Countries: Import Demand Elasticities

article image
Sources: Donges, J.B. and J. Riedel (1977).Note: * Indicates the elasticity is not significantly different from zero at the 10 percent level of significance.

These findings support the hypothesis that when the proportion of manufactured goods in total exports increases (ceteris paribus), the trend path of export earnings is raised. However, the income elasticities for some manufactured goods, particularly processed raw materials, do not necessarily differ significantly from those of some primary products.

Rate of growth of market demand. The market orientation and the composition of exports have been of central importance in determining a developing country’s export performance and diversification policy. There are disparities in the evolution of demand for exports from developing countries both among major importing areas and among major categories of exports, and the demand for traded goods has been unevenly distributed among the developing countries. Europe’s main sources of imports are developing countries in Africa and developing and industrial European countries; almost half of African exports and two fifths of European exports are imported by Europe. The United States and Japan are the principal industrial country markets for Asian developing countries, and the United States is also the main export market for Western Hemisphere countries. Because of differences in the growth of demand, during the three-year period 1983–85, developing countries in the Western Hemi-Sphere faced an average industrial country import market that grew by some 3–6 percentage points per annum faster than did the average industrial country import market faced by African, European, or Middle Eastern developing countries (Table 5).

Table 5.

Industrial and Developing Countries: Real GNP and Total Domestic Demand, 1966–86

(Percentage change)

article image
Source: International Monetary Fund, World Economic Outlook (Washington, April 1986 and April 1987).

Intra-regional trade among developing countries has expanded rapidly and has also played a role in developing countries’ diversification policies. However, such trade has been very unevenly distributed, partly as a result of differences in the rates of economic growth among regions. For example, during the recent economic recovery, intra-regional trade in South and East Asia grew by 5 percent and 10 percent in 1983 and 1984 respectively, while trade among Latin American countries fell more than 10 percent during 1983 and remained stagnant in 1984, partly as a consequence of a sharp curtailment in expenditures.

The pattern of growth in the composition of exports from developing countries has also been quite uneven. Industries such as plastic products, chemicals, machinery, and petroleum products grew fast between 1977 and 1984, compared with slower growth in such industries as farm equipment, construction and mining equipment, railroad equipment, melal mining, primary metals, and leather products. Volumes of manufactured exports have also grown much faster than those of primary products, particularly recently. During the recent expansion of imports by the United States, most of the increase consisted of manufactured products, in particular machinery, petroleum-based products, and paper and allied products, although the United States is also an important market for primary products. Therefore, the magnitude of the trade effects for developing countries has depended during the latest business cycle upturn on both the weight of manufactures and the weight of the United States in developing country trade. These results would, consequently, suggest the need to diversify not only into products for which the market is growing but also to diversify export trade into those markets which are growing the most rapidly.

Changes in competitiveness. As developing countries continue to diversify their exports and to increase the shares of products facing high income elasticity of demand and high-growth markets, much of their success in export expansion still depends on accurate knowledge of price competitiveness of their goods. Price elasticities of demand for primary product exports from developing countries as a group are quite inelastic and have been estimated to be less than one (Goldstein and Khan (1984). and Bond (1986 and 1987). Thus, if all developing countries exported sharply higher volumes of primary commodities, a large price fall would be necessary before the market would clear. It is only when one developing country that is a small exporter changes its exports independently of other developing countries that the market remains unaffected.

The demand for manufactured exports from developing countries, however, has been shown to be more price elastic in aggregate than the demand for their primary commodity exports; the evidence suggests that both domestic producers of manufactures in the industrial countries and suppliers of manufactured goods in developing countries compete with developing country exporters of manufactures. Both Khan (1974) and Grossman (1982) provide evidence of how the prices of developing countries’ exports affect imports from them by industrial countries. Khan estimates an average price elasticity of demand of 0.94 for manufactures from nine non-oil developing countries over the period 1951–69. Grossman estimates a mean price elasticity of 1.7 for U.S. imports of manufactures from the non-oil developing countries over the period 1968–78: his estimates for individual commodity groups range from 0.52 for chinaware and earthenware to 3.38 for television receivers. While, as a group, developing countries may not have much scope for increasing primary goods exports without adversely affecting prices, it seems that the volume of selected manufactured exports may be raised without a compensating fall in prices.

Changes in supply. The rapid growth in the volume of manufactured exports from developing countries also reflects supply changes in these countries themselves. Four relevant explanations have been identified (Chenery (1979)); (1) the industrial capabilities of the leading developing country exporters of manufactures have increased, contributing to their capacity to move into related export products; (2) policies in developing countries have shifted significantly away from inward-looking industrialization based on import-substituting production; (3) improvements in transport and communications have facilitated growth of trade and an international division of labor even over long distances; and (4) the rapid growth in manufactured exports has been most successful in countries that have macro-economic policies consistent with their efforts to pro-mole such exports.

The much slower growth in the volume of primary product exports of developing countries also has explanations based on the performance of their productive sectors reflecting increased food production by large food-importing developing countries, such as India and China, as well as by industrial countries. While, in the short run, the supply response of primary products to price changes has been quite small in both developed and developing countries, there is a marked increase in the price elasticity of supply over the longer run, although responsiveness varies by region and product.

Short-run changes in relative prices often result in cobweb-type swings in the production of primary commodities, especially where the cost of shifting resources is small or the products are close substitutes in production. Among developing countries, African and Asian food producers have exhibited relatively less sensitivity to short-run changes in relative world market prices. This reflects, to a large degree, producer pricing policies that, in Africa, have kept prices low—to increase government revenues from marketing boards that benefit from higher world market prices, or to subsidize urban consumption.9 Asian countries, by contrast, have generally maintained domestic producer prices above world market prices to guarantee adequate domestic staple food supplies.

While the exportable supply of some primary products can be quickly adjusted to changing external factors, this is not the case where production can only be shifted with significant capital investment (sugar to rice production, for instance, could involve extensive irrigation) or increased over long gestation periods (as in the case with tree crops). In these cases, an assessment of the long-term outlook would be required. There have been instances where producers have perceived a strong upward trend in demand, have invested to expand production in excess of the actual increase in demand, and thereby have depressed prices over the long term. A recent example of this phenomenon was the expansion in bauxite mining capacity based on a favorable outlook for aluminum demand. As a result, raw material output outpaced refining capacity, and, in the event, demand for the final product did not prove as buoyant as had been expected. Thus, the raw material price fell much more sharply than the semi-processed price, adversely affecting bauxite producers that were not vertically integrated with a major aluminum refiner.

The successful adoption of policies to promote diversification, particularly in primary products, requires awareness of what others are doing—if all countries are shifting resources into the same commodities, a global oversupply can result. There is, therefore, a need to increase the information on global demand and supply availabilities to countries making decisions on stimulating particular exports.

Trade barriers. The role of protectionism in limiting the potential growth of manufactured exports cannot be ignored. The very developing countries that successfully diversified their export base in the 1970s—the newly industrialized developing countries—are those confronting the greatest pressure to limit the inroads they are making into the markets of the industrial countries. It is the threat of increased barriers to trade that underlies the new “trade pessimism” (Bhagwati (1986)).

It is beyond the scope of this paper to explore the adverse consequences of protectionism fully.10 Barriers to trade are, however, often “porous”; that is, slight shifts in the product (such as from steel to specialty steel products), the development of new technologies (so that a new fiber, such as ramie, can be marketed) or improved quality can ameliorate the negative impact of protectionism on earnings.11 To this extent, diversification can be an important, albeit second best, tool in maintaining or increasing export earnings. Furthermore, the lack of trade barriers in itself does not guarantee that diversification and export growth will take place. For example, while Korea’s export expansion has been partly ascribed to its initial preferential market access, the absence of trade barriers alone cannot fully account for its success.12 To take another example, preferential access to the United States market for certain products was granted three years ago under the Caribbean Basin Initiative, to selected Caribbean and Central American countries; however, trade in these products has not yet increased appreciably. This suggests that factors other than free market access, such as the overall bias of the incentive structure for or against exports—as measured by the real effective exchange rate and the rate of effective protection—may be equally important in realizing a higher rate of export growth.

Trade barriers imposed by industrial countries distort the pattern of their imports and reduce the efficiency of resource allocation on a global scale. Inadvertently, however, the resulting trade distortions may favor selected developing countries. For example, the bilaterally agreed limitations on the number of Japanese cars exported to the United States stimulated a shift in their exports from relatively cheap, small-sized “basic” cars to larger, higher-priced, automobiles. This shift left a void at the lower end of the spectrum in the United States car market that has been quickly filled by Korea and Yugoslavia. Similarly, Korea, with the impetus provided by the limitations on volume growth under the Multifiber Agreement, has been switching from low-priced textiles and clothing to higher-valued items, leaving room for other developing countries at the lower end of the market.

Finally, it may be noted that developing countries themselves maintain relatively high levels of domestic protection.13 Barriers to trade imposed by these countries include high tariffs, with levels varying across product groups, and quotas, as well as restrictions arising from balance of payments difficulties. Such trade barriers contribute to distortions and reduce the efficient allocation of resources by raising the rate of return on production for home sales relative to production for export; in highly protected domestic markets increased production costs can be passed on to the consumer and producers become less able to compete abroad. Highly cascaded tariff structures, where final goods are granted greater protection than intermediate inputs and raw materials, can prevent the development of backward linkages and, in fact, over the long term may hinder the development of products that may be eventually exported. Further, such barriers limit the scope for regional trade and prevent the realization of economies of scale that access to larger markets would provide.

Diversification and Instability of Export Earnings

It has been suggested14 that producing and exporting a variety of products can offset the adverse consequences for domestic output of short-run exogenous changes in supply of or demand for individual products. By reducing the variability of export receipts on producer incomes, uncertainty could be reduced directly. Indirectly, uncertainty could be reduced by curtailing the stop-go implementation of monetary and fiscal policies, in cases where commodity boom years are associated with an expansionary domestic policy stance, and export downturns precipitate adjustment programs. The decline in uncertainty would encourage the shift from subsistence farming, reduce the need for large inventory holdings, increase investment and, consequently, raise the trend path of economic growth.

Instability of both developing and developed countries’ export earnings has varied over time, with the greatest earning instability measured in the period following World War II, when normal trade flows were not yet re-established and commodity prices were rising rapidly. The average instability of earnings, measured for each period by the mean annual variance of export earnings, fell for both developed and developing countries between 1946–58 and 1965–70 but increased thereafter (Table 6). Such fluctuations suggest that underlying global factors loom large in the determination of instability. Economic developments in the period between the late 1950s and 1971 contrast sharply with those of subsequent years (Chu and Morrison (1984)): (1) growth in industrial countries slowed substantially in 1972–82; (2) inflation was sharply higher in the later period; and (3) fluctuations in commodity prices and exchange rates were greater after 1971.15

Table 6.

Measures of Export Earning Instability

(In percent)

article image

Based on Coppock’s log variance measure of export instability.

Calculated as the average annual percentage deviation from an exponential trend.

Data from the Fund’s International Financial Statistics were used for a sample of 128 countries, including the four oil-exporting developing countries. Instability estimation based on formulation suggested by Taya (1980).

Global factors alone are not sufficient to explain the shifting differential of export earning instability between developed and developing countries. Recent evidence suggests that countries with a higher proportion of primary products in total exports exhibit greater earning instability. For example, in the 1972–85 period, such instability is estimated to have been highest for African countries whose exports remain heavily concentrated in primary commodities; by contrast, earning instability for European developing countries, where manufactures account for a large portion of total exports, was less than half that of African countries (Table 7). Nevertheless, since the 1970s, developing countries as a group have increased their exports of manufactures, while developed countries now export more primary products. On the basis of this shift, a narrowing—rather than the observed widening—gap between developing and developed countries’ group measures of earning instability might have been expected.

Table 7.

Developing Countries: Sources of Export Instability1

(In percent)

article image

Standard deviation given in parentheses.

Developing country mean earning instability calculated on the basis of 75 countries using the Fund’s International Financial Statistics data; instability was measured as deviations from an exponential trend. Covariance analysis based on a subset of 25 developing countries.

Based on the Fund’s World Economic Outlook data for 69 developing countries; mean instability based on Taya’s (1980) formulation for individual countries; aggregates are weighted averages where the weights are determined by the country’s share of exports in total export earnings.

Causes of Instability

One of the causes of increased relative instability of developing country earnings may lie in the increased price instability of primary products relative to that of manufactured goods. Between 1957–71 and 1972–82. instability of primary product prices rose relative to that of manufactured goods (Table 8). Moreover, beverage prices exhibited the widest fluctuation in the 1972–82 period and this is the product group in which developing countries have retained over 95 percent of world trade. While prices of manufactured goods are. according to these estimates, more stable than those of primary products, volumes of manufactured goods exported by developing countries exhibit greater variability than manufactures exported by developed countries. This relatively greater variability reflects: (1) the relatively higher income elasticities of demand for the manufactured products into which many of the developing countries have diversified (such as electronics assembly industries); (2) domestic input supply fluctuations, such as intermittent energy availabilities; (3) the fluctuations in access to imported inputs for manufactured exports on account of periodic lack of adequate foreign exchange; and (4) the greater instability of income growth since the 1970s.

Table 8.

Instability of World Prices of Selected Groups of Traded Goods

(In percent)

article image

Based on log variance index of instability; United Nations export unit value data used. When Coppock calculated instability index based on price data, measured instability increased. For the 21 products which are also included in the Chu and Morrison study, average instability was 19.8 percent.

Deviation from a five-year moving average; United Nations unit values are used. Figures in parentheses are the average instability measures of 22 products which also appear in the Chu and Morrison study.

Instability calculated as percentage deviation from exponential trend; quarterly commodity price data from the International Monetary Fund and the United Nations index of manufacture export prices are used.

Earning instability can also reflect underlying exogenous factors determining supply or demand. Fluctuations in product supply can result from: (1) short-run changes in production conditions; (2) fluctuations in the price of inputs; (3) changing price expectations, leading to a “cobweb” pattern of price and volume changes; and (4) changes in technology. Fluctuations in demand can occur from (1) fluctuations in trading partners’ income over the business cycle; (2) changes in relative prices of substitutes; (3) changes in taste; and (4) fluctuations in demand for products for which the export is an intermediate input or a complementary good. Similarly, policies can also influence the stability of export earnings in various ways; exchange rate and pricing policies can result in a shift of resources out of the export sector, even when the policy intention was to stabilize earnings. While initially establishing export marketing boards or fixing producer prices to stabilize farm income, countries have often been reluctant to adjust producer prices to reflect changes in costs of production, shifts in domestic prices of traded goods relative to non-traded goods, or long-term changes in world prices. Changes in taxation or trade regulations can affect export supply as well as domestic demand for traded goods. Macroeconomic policies can, by reducing domestic demand, increase export supply, or, alternatively, by limiting the export sector’s access to domestic credit or to foreign exchange for imported inputs, reduce the supply of exports.

Theoretical and Empirical Evidence

The role of export diversification in reducing earning variability and the question of whether a reduction in this variability leads to higher economic growth have been the subject of intense investigation in the economic literature since the early 1960s. There has been, however, little or weak existing empirical support for the hypothesis that an increase in export diversification, by itself, reduces earning variability.16 This counterintuitive result must be interpreted cautiously, since many of the studies were based on cross-section data, correlating export concentration indices for a representative year (or averaged for a number of years) with export earning instability. Concentration indices relate the number of export products and earnings from these products; instability is usually measured by a deviation from trend-adjusted average earnings over a given period.17 The measure of concentration generally used, the Gini-Hirschman coefficient, is subject to a number of limitations, including sensitivity to the degree of data disaggregation and the time period chosen (see Appendix I, for a more detailed discussion). The lack of statistical correlation in cross-section studies may reflect these measurement problems rather than the absence of an underlying relationship. More recently, Love (1986), in shifting from cross-section to time series estimation for 24 developing countries, found support for the hypothesis that there is a causal relationship between export concentration and earning instability.

Recent evidence from country studies also suggests that diversification into exports of manufactured goods may reduce earning instability.18 Early studies, however, largely did not support this conclusion because an observed greater stability in the aggregate volume of primary products traded more than offset their higher price instability.19 These findings led some authors to postulate that earning instability in developing countries stemmed from country-specific fluctuations in supply. For most primary products, individual developing countries are marginal world suppliers, so that reductions in exportable surpluses affect neither the aggregate world volume traded nor world prices. Export earning instability of a given country resulted from volume instability that would not be observable by measuring instability of aggregate trade in primary products and measures of instability for groups of countries would reflect the total instability caused by country-specific factors.

To determine indirectly whether earnings variability was due to supply or demand factors. Murray (1978) studied the sign of the covariance of price and quantity for 25 developing countries (Table 7). When prices move inversely with quantity, the sign of the covariance is negative and the underlying causes are said to be supply determined (demand constant). When prices and quantities move together (the covariance is positive), demand factors are more important (supply constant). Between 1952 and 1971, supply factors were found to be the major cause of earning instability in these 25 countries. Updating this study for 69 non-oil developing countries, demand factors were found to be the more important cause of earning instability in the 1972–85 period, with the notable exception of African and oil-exporting developing countries, where earning instability remains supply determined. The shift from supply-determined to demand-determined variability reflects the increase in the share of manufactures in the production and exports of many developing countries over the last 20 years, Hanson (1983) found that as countries became more developed, the contribution of volume instability to earning instability increased, while that of price instability declined.20 Supply-stimulated instability was less common in industrialized and highly diversified economies than in countries that still heavily concentrated on primary product exports. Furthermore, technological advances and improved infrastructure have helped to dampen the adverse effects of weather on agricultural production for many developing countries.

Diversification that reduces the covariance of supply responses to exogenous shocks (such as weather) or that lowers aggregate price instability can also reduce export earning instability (Brainard and Cooper (1970)). For example, rice prices tend to move in tandem with other food prices (that are close substitutes in production and consumption) and some agricultural raw materials (close substitutes in production) but tend to have little relationship with changes in prices of tree crops or minerals (Appendix III, Table 19). Prices of manufactured goods generally exhibit relatively low correlation with primary product prices, further supporting the role that shifting into manufactures can play in lowering earning instability; however, the composition of manufactured exports must also be taken into account. Diversifying from primary goods to processing these same primary products would not reduce the vulnerability of earnings to supply factors, although by increasing the domestic value added, both export earnings and economic growth could be raised. Shifting to the export of luxury manufactured goods, facing highly income-elastic demand, from food exports, facing low income-elastic demand, could raise aggregate earning instability because of the greater possible variation in export volumes. Optimal diversification, moreover, has to take account not only of short-run fluctuations but also the long-term trend path of the product’s earnings.

Finally, reduction in variability, even around an upward trend in export earnings, may or may not lead to higher real growth. That a reduction in instability raises the trend path of income growth is contested by a number of empirical studies taking a variety of approaches. MacBean (1966) and Kenen and Voivodas (1972) find no significant correlation between export earning instability and growth. Rangarajan and Sundararajan (1976) find a statistically significant relationship but no unidirectional causation; that is, for some countries, a higher growth of income was associated with more unstable export earnings. Knudsen and Parnes (1975) hypothesize that consumption and investment decisions are based on a notion of “permanent export income”; earning instability serves to reduce consumption and increase savings and investment and, as a consequence, growth. Newbery and Stiglitz (1981) illustrate that the impact of reduced export instability on investment depends upon how urban wages are determined and how changes in nominal wages affect profits.

In conclusion, the evidence suggests that earning instability is less for those countries with a lower proportion of export earnings derived from primary products. Further, for those countries that are marginal world suppliers and are heavily dependent upon primary products, the source of earning instability tends to be supply-determined. These findings suggest that diversification from primary to manufactured goods could reduce earning variability as well as raise the trend path of export earnings (as discussed in the preceding section). Identification of the source of the greatest earning instability as supply determined in African countries, however, suggests that diversification of exports may be only a partial solution; other factors, such as adequate infrastructure, extension services, access to necessary inputs, and disease and pest control, initially may be equally—-if not more—important.

Diversification and Domestic Growth

The weak empirical support for the hypothesis that less variability of export earnings is related to higher economic growth is not surprising. Often these studies failed to take account of both the influence of domestic policy and other economic variables that affect domestic growth. It is to these two topics that this paper now turns.

Export Diversification, Export Growth, and Domestic Growth

The relationship between export diversification and domestic growth is complicated, but it is generally observed that the more diversified an economy, the more developed it is, and export diversification is generally associated with the process of output diversification. It is clear that export diversification, combined with access to export markets, aids the growth process. Most developing countries have fairly little resource diversification and small domestic markets compared with the larger industrial countries, and benefit especially from export trade by gaining access to larger markets. While import substitution may increase domestic output in the short run by employing unused resources, longer-term growth would be constrained by the rate of growth of domestic demand. In the long run, output growth in excess of that of domestic demand depends upon expanding export volume. Finally, industries producing exports have been found generally to be more efficient than those producing domestic substitutes for imports, which tend to require greater diversification of production and resources than export-oriented production, and often rely upon many more imported inputs. Therefore, expanding the share of exports in real gross domestic product (GDP) is likely to be associated with more efficient allocation of resources and more rapid growth. In turn, this type of production is often accompanied by economies of scale, rising productivity and profits, rising real domestic investment, and enhanced efficiency in marketing.

However, export diversification may not necessarily be accompanied by an expansion in domestic growth; it may, in fact, increase export growth but reduce output growth if. for instance, resources that are switched from production for domestic consumption to production for export (in response to export subsidies, for example) are used less efficiently. Conversely, export diversification and output growth may occur without a concomitant increase in the growth of exports; for example, real growth in the domestic economy induced by an exogenous rise in consumer demand for exportables and nontradable goods, may, over this period, lead to a decline in export growth. Where foreign capital is used to diversify into inefficient export industries, the domestic growth process may also be hindered by the subsequent outflow of profit remittances or debt service not covered by adequate export earnings. Alternatively, fluctuations in export demand may not feed through to the domestic economy, for example, where inventories are used in a countercyclical manner.

In some cases, exports can become more concentrated and domestic output can also grow: for example, economic activity in Mexico and Indonesia expanded following the rise in export revenues as oil prices surged in the 1970s. However, such a growth path is likely to be unsustainable. Concentration in these cases is usually in one commodity, and if the price of this commodity falls, growth can slacken unless adequate foreign exchange reserves or access to foreign funds allow for countercyclical demand management or unless the country has diversified its export base in the meantime.

Empirical Evidence

Numerous empirical studies have investigated the relationships among export diversification, export growth and growth in domestic output.21 The direction of causation between export growth and domestic growth has been explored by Jung and Marshall (1985) who challenge the hypothesis that export growth causes growth in domestic output. They argue that in many cases overall economic growth causes export growth rather than vice versa, pointing out that unbalanced growth in the economy can create boom industries, which then grow more rapidly than domestic demand and encourage producers to turn to foreign markets to sell their goods. In this case, export growth is a residual of domestic economic growth. Jung and Marshall also suggest that real economic growth, when it is induced by an exogenous increase in consumer demand for exportable and nontradable goods, can reduce export growth. Jung and Marshall use the Granger notion of causality to test the hypothesis that export promotion leads real GDP growth. Results, based on time series data for 37 countries for the period 1950–81, provide evidence that exports led economic growth in only four instances: Indonesia, Egypt. Costa Rica, and Ecuador.

Similarly, Kormendi and Meguire (1985) found only weak evidence that an increase in the share of exports in total output raises real growth. In a cross-section study of 47 industrial and developing countries, the authors test various macroeconomic hypotheses relating selected economic variables to domestic growth. To measure the contribution of each variable to growth, the authors recalculated the estimation equation omitting the selected regressor. In aggregate, an increase in the share of exports in total output was calculated to contribute only 4 percent to the observed variation in economic growth. The largest contribution to growth was estimated to be made by money variance (a 22 percent positive contribution), followed by the initial per capita income level (an 11 percent negative contribution), the rate of population growth (a 10 percent positive contribution), the variance of real output-taken as a proxy for the riskiness of the technology adopted (an 8 percent positive contribution), and the variance of inflation (a 6 percent negative contribution). The contribution of each of these variables, however, was found to vary across countries; for example, the increase in the share of exports in Korea’s total output was estimated to have been the single most important determinant. While these studies suggest that the evidence in favor of export promotion as the initiator of domestic growth is not as strong as previous studies have indicated, they do not negate the strong interdependence between export expansion and GDP growth nor the importance of export growth in raising real GDP for selected countries.

In a new assessment of the relationships among export diversification, export and economic growth for individual countries, this study calculated annual Gini-Hirschman coefficients for a sample of 59 developing countries, for the period 1964–81 (Table 9).22 These coefficients reflect both the number of export product groups and the share of each product group in total exports; the larger the number the more concentrated, or less diverse, a country’s export earnings are said to be. The calculations were based on the three-digit SITC classification. Primary goods and mineral exports were included in primary goods exports; manufactured goods included both final goods and semi-processed agricultural goods; energy products were considered separately. Gini-Hirschman coefficients for nonenergy primary products and manufactured goods, calculated to illustrate the source of increased concentration or diversification, are presented in Appendix III, Table 20.

Table 9.

Selected Developing Countries: Diversification, Export Growth, and Real GDP, 1964–81

article image
Sources: International Monetary Fund, International Financial Statistics; UN SITC data—3-digit level.

Indicates increased concentration (negative number).

1981 data not available; latest available data 1979.

1981 data not available; latest available data 1980.

1981 data not available; latest available data 1978.

Average annual increases in export earnings and in real GDI’ were measured for the periods 1964–71 and 1972–81. The choice of periods allowed for the increase in global inflation, and for the price and exchange rate instability that has occurred since 1971. Separation into two periods also permits a comparison of the trend path of economic growth in 1964–71 corresponding to a given level of export diversification in 1964. with that attained in 1972–81 corresponding to a different degree of export diversification. The rate of growth of the volume of world trade differed only slightly between the two periods, falling from an annual average of 6 percent in 1964–71 to 5.6 percent in 1972–81. Countries were classified into four categories: (1) Group A—increased diversification of the export base between 1964 and 1981 associated with relatively high and sustained real growth and/or increased real growth: (2) Group B—increased export diversification associated with a slowdown of economic growth between 1964–71 and 1972–81; (3) Group C—increased concentration of the export base on a narrower range of products associated with relatively high and/or increasing real growth; and (4) Group D—increased concentration and declining real growth. The final two groups are dominated by countries that increased their dependence on energy products, reflecting the substantial increases in petroleum prices in the 1970s.23

Twenty-four countries out of the 40 recording higher export diversification had lower average annual rates of growth in the period 1972–81 than in the period 1965–71, reflecting the adverse domestic consequences of the deterioration in the terms of trade that took place in most developing countries in the later period. There were, however, 16 countries—half of which are Asian—where export diversification was accompanied by increased or sustained high levels of economic growth. Greater export concentration was almost equally distributed among high and/or increasing growth countries—generally by those where the terms of trade improved in the later period—and countries with falling growth. Of the countries sampled, Mauritius had the least diversified export base in 1964 with export earnings almost entirely derived from sugar, but since then its export base has expanded to include textiles. Yugoslavia, in both 1964 and 1981. had the most diversified export base, although the composition of its exports shifted; in 1964 manufactured goods accounted for only 64 percent of total export earnings, but by 1981 this share had risen to 87 percent. Brazil recorded the greatest increase in export diversification between 1964 and 1981, as measured by the fall in the Gini-Hirschman coefficient from 55 to 19 over the period; manufactured goods became an increasingly important source of foreign exchange. The greatest reduction in a country’s export base was recorded by Nigeria over the period 1964–79. reflecting the increased importance of oil along with the stagnation in palm oil exports, ground nuts, and cocoa; by 1979 Nigeria had the least diversified export base of the countries sampled.

Role of Policy in Export Diversification

General Policy Framework

Official economic policies can aid or hinder the process of diversification. The government can provide a climate attractive to investors and a framework in which diversification of production—including production for export—can give rise to a higher GDP growth path by implementing appropriate monetary and fiscal policies, and supply-side policies that allow exports to maintain competitiveness abroad and provide adequate incentives for domestic output.

Macroeconomic Policy

Macroeconomic policies have as common objectives: maintaining internal balance between aggregate demand and supply to avoid inflationary pressures; avoiding unsustainable external imbalances; and maintaining or strengthening the conditions for achieving a satisfactory rate of long-term output growth. The primary macroeconomic instruments—monetary, fiscal, and exchange rate policy—play a major role in demand management and in facilitating adjustment to exogenous shocks. Ideally, these policies should be carried out in such a way as to meet the criterion of efficient resource allocation.

Consistency over time and coherence among policies are a sine qua non for creating an environment in which long-term decisions about productive investment can take place and export diversification can occur, in line with a country’s evolving comparative advantage. When macroeconomic policies, however, are subject to frequent modification, long-term investment decisions can be distorted, encouraging savings to move into inventories of goods, financial investments, or abroad. Lack of coherence among policy instruments invariably produces results that are counter to those intended. For example, excess demand pressures stemming from fiscal deficits can contribute to both internal and external imbalance, notwithstanding the stance of monetary or exchange rate policy. If all or part of the fiscal deficit is monetized and the exchange rate is not fully flexible, fiscal deficits can result in balance of payments deficits.24 Similarly, devaluations of the exchange rate that are not supported by appropriate fiscal and monetary policy, would only provide short-term stimulus to the export sector. Successive devaluations of the nominal rate to isolate the external sector from unchecked excess domestic demand pressures can induce a vicious cycle of rising inflation and devaluation and an indexing of the money supply to the price level (Adams and Gros (1986)). Ultimately, the loss of control over the inflationary process would distort new investment decisions. When policy shifts direction frequently, long-term investment and export-marketing commitments are hindered by the attendant uncertainties.

Supply-Side Policies

Policy can also be used to influence the distribution of resources among productive sectors—whether import substituting, exportables, or nontraded goods. Structural policies affecting, for example, pricing, incomes, and trade policies and specific aspects of taxation, subsidies, or spending policies can affect the overall level of savings and investment, and, therefore, domestic growth. Exchange rate policy, in addition to its demand management functions, also has implications for output. In this regard, policies that are not significantly biased against exports or in favor of imports are considered to be neutral and contribute to the elimination of distortions. In turn, such a policy stance would result in the development of production along the lines dictated by comparative advantage. Policies which are biased in favor of exports—called by Bhagwati (1986) “ultra-export-promotion policies”—may lead to new distortions and the inefficient allocation of resources.

A subgroup of supply-side policies—called in this study sector-specific policies—can change the rate of return among products within the export sector and, thus, have a direct impact on export diversification. Sector-specific policies, however, have often been used to offset distortions that have arisen from macroeconomic policy variables. Further, such policies have often favored manufactured goods exports and have been biased against traditional exports. Depending upon the degree of bias, such policies can shift resources from efficient production, in which the country enjoys a comparative advantage, to inefficient production that depends on the existence of domestic subsidies in order to compete internationally.

Country Experience

General Economic Environment for Individual Countries

In reviewing the selected indicators for the 59 countries sampled, a number of characteristics seem to be shared by countries that have diversified their export bases, expanded export earnings, and realized high rates of domestic growth (Group A countries). High-growth countries, independent of the degree of export diversification, however, tend to exhibit higher rates of national saving and investment as a proportion of GDP. Those countries with high savings and investment ratios to GDP and increased export concentration, however, were predominantly oil-exporting countries that benefited by the sharp improvement in their terms of trade over the period studied. Those countries that diversified their export base and realized higher domestic growth also tended to have lower inflation rates and a greater improvement in international competitiveness, as measured by the real effective exchange rate (Table 10). There are, however, notable exceptions; for example, while most of the Group A countries had relatively low inflation during the 1964–81 period. Brazil did not. In fact, by the early 1980s, Brazil’s annual inflation rate had reached triple digits. Yet Brazil achieved the greatest export diversification of the countries studied, increased manufactured exports, and realized high rates of growth of export earnings and real GDP. Nor does improving international competitiveness alone guarantee greater diversification and higher growth. Malawi’s external competitiveness, measured by its real effective exchange rate, improved over the period but its export base became more concentrated, and while the growth of export earnings accelerated, real domestic growth declined. These examples underscore the need for caution in interpreting the role of individual policies for any particular country.

Table 10.

Selected Developing Countries: Macroeconomic Indicators, 1964–811

(Period averages unless otherwise indicated)

article image
Sources: International Monetary Fund, International Financial Statistics; and Fund staff estimates.Note: CPI is the consumer price index.

The countries and the ordering of the countries presented in this table are the same as those presented in Table 9.

Calculations based on changes in relative consumer prices; an increase implies a real effective appreciation of the exchange rate.

Data for domestic credit expansion are not available for 1964 to 1971.

Data for investment and savings for 1981 are estimates.

Data for investment, savings, and credit were calculated over the period 1969 to 1971.

Data for credit were calculated over the period 1966–71.

Data for credit were calculated over the period 1967–71.

Data for investment and savings are estimates.