III Non-Oil Developing Countries’ Export Growth and Real Income Growth
- Mohsin Khan, and Morris Goldstein
- Published Date:
- August 1982
This section focuses on the relationship between the export growth of non-oil developing countries and their economic growth rates. In brief, an attempt is made to identify the channels by which exports affect economic growth; whether the relationship between the two growth rates is rigid or subject to change; whether the relationship has strengthened or weakened over time; and how the relationship differs among the subgroups of non-oil developing countries.
General Aspects of Trade and Economic Development
Before considering how exports themselves affect a country’s growth rate, it is useful to first adopt a somewhat broader perspective by considering how international trade can contribute to economic development.51
First, trade allows a country to follow the route indicated by the theory of comparative advantage, permitting it to specialize in those areas of production which use relatively more of its abundant factors. The country can concentrate on the production of those goods that utilize domestically available resources, and import goods and services that would be expensive to produce at home. In other words, a wider set of options is made available to the country via trade, thereby implying an increase in welfare.
Second, developing countries, being generally small, offer only a limited market so that unit costs of production are higher if production is restricted largely to the domestic market. International trade offers firms the opportunity to exploit economies of scale by producing also for foreign markets. At the same time, the country can import those goods that can be produced on a large scale, and therefore cheaply, abroad.
Third, trade increases incomes and thereby the general level of demand in the economy. This has multiplier effects throughout the economic system, resulting in higher levels of savings, investment, and employment. The accumulation of capital and increases in employment play a central role in the development process, and trade not only triggers such a process, but also tends to sustain it.
Several other examples of the varied benefits of trade can be added to those above. The supply capacity of the economy is increased through imports of capital goods, raw materials, and other inputs for the production process. Through trade there is a transfer of technology from the industrial countries which can be advantageous to developing countries. Foreign trade, by providing international competition for tradable goods, is a source of stimulus and pressure for domestic production. This results not only in improved quality and generally lower prices, but also in increased efficiency as domestic firms have to meet standards set abroad, rather than deal only with the limited competition in the smaller local market. Finally, foreign trade provides revenues for the government if exports and imports are taxed directly or indirectly; these additional revenues can, in turn, be channeled into productive investment.
Exports and Economic Growth: A Framework
The preceding discussion reveals that the gains from trade rest on importing as well as exporting. Nevertheless, the literature on the sources of growth in developing countries has long given a special place to exports, hence the familiar concept of “export-led growth.”52 How exports influence economic growth and development in non-oil developing countries can perhaps best be discussed within the framework of “linkages” as outlined by Hirschman (1977). Such linkages are classified into two main categories as detailed below.
This is the direct physical linkage that results from investment-generating forces and from input-output relationships when there is an increase in the production of the export good. Historically, in a number of developing countries, it has been the growth of exports that has led to the development of infrastructure, transport and communication systems, etc., which in turn have facilitated the production of other goods and services. Investment opportunities are opened up in areas far removed from the actual export activity as the need to supply inputs rises, and productive facilities are created utilizing inputs and outputs that were nonexistent prior to the expansion in exports. Also related to the production linkage is the transfer of factors from the rest of the economy to the export sector, which is typically the most productive sector, thereby raising the overall rate of growth. These processes basically represent the “secular” or “trend” effects of export growth on output.
The increase in income that comes directly from exports leads in time to a rise in demand for a wide range of products, including nontradables. These demand pressures are reflected in an expansion in domestic supply and, therefore, involve investment in facilities providing such products. Increased demand for importables also contributes to the expansion of domestic production. In short, overall output naturally increases in response to the export-induced rise in demand. This demand linkage can be thought of as the “cyclical” effect of export growth on real output.
In combination, the production and demand linkages should ensure that an expansion in exports is accompanied by a rise in investment, an increase in and greater utilization of capacity, and an increase in employment.53 In the longer run, additional effects would emerge, such as the exploitation of economies of scale and technological improvements. Further, these linkages represent only the direct effects of export expansion. Indeed, the indirect effects on economic growth may be just as important.
The first such indirect effect is the positive impact of exports on aggregate savings. Domestic savings rise as a consequence of the general increase in incomes that results from the initial expansion of exports. Furthermore, as argued by Maizels (1968), the marginal propensity to save in the export sector could be larger than elsewhere, in which case a rise in incomes generated by exports would raise savings by more than would a general rise in incomes.54 Since the effective constraint on investment in developing countries is typically the availability of resources, both physical and financial, rather than the cost of investment funds, domestic savings tend to be directly complementary to investment; as such any rise in savings would therefore result in increased investment.
Second, the level and growth of exports can have a significant influence on the supply of foreign capital. While the relationship is not easy to quantify since it involves aspects of country risk and market evaluation of the overall economic and political prospects of a country, foreign investors may make investment decisions partly on the basis of the country’s ability to repay out of export receipts. Direct foreign investment and long-term capital inflows in turn add to domestic investment, and even short-term financial flows may serve a useful purpose by providing funds for working capital.
Finally, many developing countries are heavily dependent on imports of capital and intermediate goods as inputs into production, and exports provide the necessary foreign exchange to purchase such imports. Any decline in exports may cause these imports to be reduced, with a consequent adverse effect on production. This is the main rationale underlying the familiar “foreign exchange constraint” on growth.55
Empirical Evidence on Exports and Economic Growth
The channels by which export growth can lead or induce overall economic growth can be easily outlined. However, measuring or estimating the independent contribution of exports to economic growth is not as clear-cut, especially given the multiple ways, direct and indirect, in which the two variables are interrelated.56 In this respect, there are two types of empirical evidence that are relevant. First, there are cross-country studies that test whether those developing countries with unusually rapid (slow) growth of exports have also experienced unusually rapid (slow) economic growth. Second, there is the time-series evidence that indicates, for a single country or country group, whether periods of unusually rapid export growth have also been periods of unusually rapid economic growth.
Perhaps the simplest of the cross-country studies is that of Michaely (1977) who calculated the simple correlation between growth of per capita GNP and growth of exports for a sample of 41 developing countries. The relevant variables were defined as average values for the period 1950–73, and a positive association between the two variables was clearly evident in the data.57 Because Michaely (1977) used a nonparametric test in his study, it is not possible to determine directly the size of the effect of export growth on economic growth from his results. Fortunately, such an estimate can be obtained by regressing the growth rate of per capita GNP on the change in the ratio of exports to GNP for Michaely’s (1977) 41-country sample over the 1950–73 period.58 Such a procedure yields the result that a 1.0 percentage point increase in the growth of exports (volume) is associated with a 0.23 percentage point rise in the growth rate of real per capita GNP.
Of course, one serious failing of the Michaely (1977) test is that it assumes that the effect of other factors influencing developing country growth is zero. A better conceptual approach is to relate the gross domestic product (GDP) growth rate to all potential sources of growth, with exports being merely one such source. If correctly specified, such a procedure would uncover the influence of export growth on economic growth, other factors held constant. Such an approach has in fact been adopted by Robinson (1971), Michalopoulos and Jay (1973), Balassa (1978), and most recently, by Tyler (1981). In these studies, a production function approach is taken for estimating real income growth, and exports are included under the rationale that they raise total factor productivity, and thus deserve to be treated as a separate factor of production.
The study by Robinson (1971) specifies a growth model based on an aggregate production function for a cross-section sample of 39 developing countries. In addition to the usual factors of production, namely, capital formation and growth in the labor force, a variable measuring the foreign exchange constraint is added.59 Since this latter variable only picks up the effect of export performance on the growth rate in an indirect fashion, for the analysis here it was replaced by the corresponding growth of the volume of exports. Estimating this revised formulation yields results similar to those reported by Robinson (1971), but at the same time gives a direct estimate of the effect of export growth on output growth. The value for the export effect turned out to be quite similar to estimates obtained from the Michaely (1977) test, that is, approximately of the order of 0.2.60 Also, as in the case of the Michaely test, this estimate was statistically significant.
Michalopoulos and Jay (1973), using a sample similar to Robinson’s (1971), extended the analysis by introducing a measure of foreign investment into the relationship, along with export growth, domestic capital formation, and the growth in the labor force. They found that in this expanded framework the direct influence of export growth is much reduced, though it does remain a significant factor in explaining variations in growth rates among the countries surveyed. A 1.0 percentage point increase in the rate of growth of exports resulted in only a 0.04 percentage point increase in the rate of growth of real GNP. The difference between this result and that of Robinson (1971) probably stems from Michalopoulos and Jay’s procedure of dividing total investment into its foreign and domestic components. This introduces the foreign exchange constraint directly, thereby allowing calculation of the independent effect of export growth.
Balassa (1978) has also tested the relationship between export growth and economic growth, but using a smaller sample of 11 developing countries that had already established an industrial base. The sample period was 1960–73. Adopting essentially the framework of Michalopoulos and Jay (1973), Balassa (1978) reported a coefficient measuring the effect of export growth that varies between 0.04 and 0.05, depending on whether exports are defined in nominal or real terms. While the sample of countries differs substantially, it is interesting to note that this estimate is almost identical to the one reported by Michalopoulos and Jay (1973).
Finally, in the most recent study, Tyler (1981) examined the link between export performance and economic growth for 55 middle-income developing countries (those with GNP per capita equal to or greater than $300, in 1977 U.S. dollars) over the 1960–77 period. Both the simple correlation and production function approaches were utilized. Under the former approach, Tyler (1981) found that there was a significant positive association between real GDP growth and export volume growth (whether measured as total exports or manufactured exports only). Further, the link between GDP growth and export growth was stronger than that between GDP growth and either the growth rate of direct foreign private investment or the change in the net barter terms of trade; on the other hand, export growth was less closely linked to GDP growth for these middle-income developing countries than was either the growth rate of gross domestic investment or the growth rate of manufacturing output. Under the production-function approach, the growth rate of exports was again statistically significant after allowing for the influences of capital formation and labor force on GDP growth. Specifically, Tyler (1981) reported that each 1.0 per cent increase in the rate of growth of total exports was associated with an increase of 0.06 per cent in GDP growth—an estimate that again is close to the other production-function based estimates discussed earlier.
Table 17 summarizes the range of estimates emerging from the cross-section studies on the effect of export growth on economic growth. The estimates that come from the bivariate regressions (e.g., Michaely (1977)) are probably biased upward because they attribute to exports some of the contribution of other sources of growth that were simultaneously increasing. On the other hand, the production-function estimates (e.g., Tyler (1981)) are probably biased downward because they ignore the effects of exports on the other sources of real income growth. As previously suggested, exports can have a positive effect on savings and investment, can assist in the technological progress of the economy, and can induce factor movements toward the more productive sector. Thus, although the range is quite wide (0.04 to 0.23), it is probable that the true effect lies somewhere in the lower middle part of that range, say 0.10.
|Michalopoulos and Jay (1973)||0.04|
Another conclusion that emerges from the cross-country studies is that the significant positive association between exports and economic growth is not invariant to the degree of development of the country or to the commodity composition of its exports. Michaely (1977), for example, found that the relationship was quite different when the 41 countries were divided into two groups on the basis of per capita income. The association was particularly strong among the more developed countries (those with per capita income in 1972 of over US$200), and almost nonexistent for the least developed. This result led Michaely (1977) to conclude that “…growth is affected by export performance only once countries achieve some minimal level of development” (p. 52). Similarly, Tyler (1981) rationalizes the selection of only middle-income countries for his sample by arguing that “…some basic level of development is necessary for a country to most benefit from export oriented growth…” (p. 124). Countries that export relatively more finished and manufactured goods also seem to be the ones where export-led growth is stronger. Balassa (1978) provides some indirect evidence on this point by calculating hypothetical gains and losses in growth rates of GNP if the country in question had the average export growth rate (and incremental export-GNP ratio) of the sample. The countries with higher than expected GNP growth were the ones in which manufactures represented a relatively large share of exports.
The cross-country studies just discussed are useful for obtaining an estimate of the size of the link between export growth and real income growth for a “representative” developing country. However, these studies do not provide information on two other questions that are particularly relevant for this study, namely, has the link between export growth and economic growth become stronger or weaker over the past 15–20 years, and how does this link vary across the four analytical subgroups of non-oil developing countries used elsewhere in this paper.
As a first step toward answering these questions, Chart 1 shows the respective growth rates of real GDP and of export volume for the four analytical subgroups of non-oil developing countries during 1965–80. This chart suggests not only that export volume growth is typically considerably more variable than real income growth but also that the association between the two growth rates is relatively weak. Table 18, which presents the average growth rates for the same four subgroups, as well as for the aggregate of non-oil developing countries, over both 1965–72 and 1973–80 tells basically the same story. The subgroups with the highest export growth rates (major exporters of manufactures and net oil exporters) also have the highest GDP growth rates but, for a given subgroup, the GDP growth rate seems to bear little relationship to the export growth rate, except perhaps for major exporters of manufactures.
Chart 1.Non-Oil Developing Countries: Growth Rates of Real GDP and Export Volume, 1965–80
Source: IMF, World Economic Outlook (1981).
|All non-oil developing countries|
|Major exporters of manufactures|
|Other net oil importers|
|Net oil exporters|
The export volume data are based on a slightly different deflator than is used in IMF, World Economic Outlook (1981). As such, the figures differ somewhat from those reported in Table 1. The export volume data in this table were used because they offer a time series extending back to 1965.
In an attempt to provide a more precise description of the time-series relationship between the two variables, the real GDP growth rates were regressed on the export volume growth rates, again disaggregating the results both by analytical subgroup and by time period. The conclusions emerging from that exercise can be summarized as follows. First, taking all non-oil developing countries together, there was no significant relationship between the two variables, even when the growth rates were expressed as three-year moving averages to purge the data of erratic year-to-year movements. Specifically, growth rates of export volumes explained only 1 per cent of the variance of real GDP growth rates for all the countries over the entire 1965–80 period. Second, there is some indication that the export-GDP growth relationship was stronger in 1973–80 than in 1965–72. In the latter period, export growth rates explained about 37 per cent of the variance of real non-oil developing country GDP, with an elasticity of 0.14. Overall, there seems to be little doubt that the time-series relationship between economic growth and export volume growth in the non-oil developing countries is much weaker than that between their export volume growth and industrial country economic growth (explored in Section II). Third, among the four analytical subgroups, the link between GDP growth and export volume growth appeared strongest for major exporters of manufactures and other net oil importers and weakest for low-income countries and net oil exporters.61
This third finding is plausible on a number of counts. To begin with, such a result is consistent with the cross-country studies that found a significant export-income growth link only for middle-income developing countries. Also, manufacturing is widely regarded as the activity that generates the largest trade-induced benefits for developing countries via economies of scale and learning-by-doing (e.g., see Keesing (1979)). Further, the contribution of export growth to GDP growth can be expected to be larger, ceteris paribus, the greater the share of exports in GDP. The export share was more than twice as high for major exporters of manufactures and other net oil importers as for low-income countries.62 The export share of net oil exporters was not much below that for the middle-income oil importers but it is not surprising that export volume growth was not highly correlated with real GDP growth for the former group because of the importance of the positive movements in their terms of trade, at least over the 1973–80 period.
Other Factors Affecting Growth in Non-Oil Developing Countries
Since exports are an important but by no means dominant source of growth in non-oil developing countries, it is necessary to discuss other factors affecting growth in order to explain the 1973–80 experience; this is particularly true for the low-income subgroup where the empirical work revealed that economic growth was less dependent on exports than in the middle-income subgroup. In considering how non-oil developing countries could obtain a 5.2 per cent average growth rate in real GDP during 1973–80 despite a significant falloff in industrial country average real income growth, special attention needs to be directed to factors that could compensate for the observed reduction (vis-à-vis the period 1968–72) in export growth. In this respect, an examination of migrants’ remittances, capital flows, the structure of production, and the overall orientation of non-oil developing countries’ economic policies is in order.
Non-oil developing countries not only export merchandise to industrial countries and oil exporting developing countries, but they have increasingly exported labor to these countries as well. These workers in turn typically remit a portion of the wages they earn in the host country back to their home country, with the level of these remittances generally being a positive function of the level of economic activity in the host country.
Three aspects of workers’ remittances to non-oil developing countries are important for this paper: (i) remittances have been growing rapidly over the past decade; (ii) the absolute amount of such remittances is now large as a proportion of merchandise exports in quite a few of the labor exporting countries and hence represents a significant source of foreign exchange; and (iii) the level of and change in the flow of remittances seems to be highly correlated with the level and change in real economic activity in the host country.
Each of these three conclusions on migrant remittances is documented in a recent study by Swamy (1981). Table 19 shows that inflows of workers’ remittances grew from $2.6 billion in 1968–69 to about $23.8 billion in 1978–79, with particularly large percentage increases taking place in the Middle East, Asia, Central America, and Africa. While European countries were still receiving by far the dominant share (59 per cent) of all such remittances in 1978–79, Middle East countries saw their share increase from 4 per cent in 1968–69 to over 20 per cent in 1978–79.
Includes only Mexico, Jamaica, Haiti, El Salvador, and Guatemala.
Includes only Colombia, Paraguay, and Bolivia.
Table 20 illustrates the well-known fact that the labor exporting regions (i.e., the regions receiving the remittances) are predominantly non-oil developing countries, the exceptions being Italy and Spain, which are industrial countries, and Algeria, which is an oil exporting developing country.
|Labor Receiving Countries||Labor Sending Countries|
|Europe||Federal Republic of Germany, Switzerland, France, Belgium, United Kingdom, Austria||Turkey, Yugoslavia, Greece, Portugal, Spain, Italy, Morocco, Tunisia, Algeria|
|Middle East||Saudi Arabia, Kuwait, Qatar, United Arab Emirates, Libya||Egypt, Jordan, Yemen Arab Republic, People’s Democratic Republic of Yemen, Sudan, Syrian Arab Republic, India, Pakistan, Korea, Bangladesh|
|Southern Africa||South Africa||Lesotho, Botswana, Swaziland, Mozambique, Malawi (until recently)|
|Western Africa||Ivory Coast||Upper Volta, Mali, Benin, Guinea|
|North America||United States, Canada||Almost all the countries in the world. In particular, workers come from Mexico, Jamaica, the Bahamas, and South America.|
|Latin America||Argentina, Venezuela||Colombia, Ecuador, Peru, Bolivia, Paraguay, Uruguay, Chile|
The second point, namely, that migrants’ remittances are now an important source of foreign exchange for many non-oil developing countries is illustrated in Table 21, which shows the ratios of workers’ remittances to merchandise exports in 1978–79. Table 21 reveals that there are many non-oil developing countries where workers’ remittances accounted for more than 20 per cent of the value of merchandise exports, and some (Egypt, Jordan, Morocco, Pakistan, Portugal, Turkey, Upper Volta, the Yemen Arab Republic, the People’s Democratic Republic of Yemen) where this percentage exceeded 50 per cent in 1978–79. Special mention should be made of those non-oil developing countries that, although being unable to reap large export gains to the oil exporting developing countries because of an unsuitable commodity composition of exports, were able to share in the oil boom by receiving sharply increased remittances from their export of labor to these countries. This group includes Bangladesh, Egypt, Jordan, Pakistan, Sudan, the Syrian Arab Republic, the Yemen Arab Republic, and the People’s Democratic Republic of Yemen.
|Syrian Arab Republic||0.032||0.104||0.088|
|Yemen, People’s Dem. Rep. of||0.837||1.340||5.638|
|Yemen Arab Rep.||…||13.737||70.913|
|Korea, Rep. of||0.104||0.008||0.007|
The Swamy (1981) study also found that the level of, and cyclical fluctuations in, economic activity in the host countries explained 70–90 per cent of the variation in remittances flowing into labor exporting countries. The estimates of the elasticity of remittance inflows with respect to economic activity in the host country are given in Table 22. An implication of these estimates is that workers’ remittances would indeed be adversely affected by an industrial country slowdown but that this effect would be confined to the relatively small number of non-oil developing countries (e.g., Algeria, Greece, Jamaica, Mexico, Morocco, Portugal, Tunisia, Turkey, and Yugoslavia) that export labor heavily to industrial countries. Other labor exporters among the non-oil developing countries have much more to lose from a growth slowdown in oil exporting developing countries or in selected labor importing non-oil developing countries such as Ivory Coast and South Africa.
|Labor Exporting Country||Labor Importing Country||Elasticity with Respect to GDP, Government Expenditures, or Exports|
|Yugoslavia||Germany, Fed. Rep. of||1.82|
|Greece||Germany, Fed. Rep. of||1.17|
|Turkey||Germany, Fed. Rep. of||1.67|
|Italy||Germany, Fed. Rep. of, France, Switzerland||0.88|
|Spain||Germany, Fed. Rep. of, France, Switzerland||1.36|
|Cyprus||Germany, Fed. Rep. of||–|
|Egypt||Saudi Arabia, Libya, Kuwait||1.30|
|Syrian Arab Republic||Saudi Arabia, Libya, Kuwait||0.54|
|Yemen, People’s Dem. Rep. of||Saudi Arabia||0.65|
|Jordan||Saudi Arabia, Libya, Kuwait||1.22|
|India||Qatar, Oman, Kuwait||1.01|
|Pakistan||Qatar, Oman, Kuwait||1.39|
|Korea, Rep. of||Saudi Arabia||0.48|
|Upper Volta||Ivory Coast||0.72|
|Jamaica||United States, United Kingdom||3.19|
As is well known, non-oil developing countries experienced large current account deficits during 1973–80. Their current account deficit grew from $11.5 billion in 1973 to a peak of $46.5 billion in 1975, then fell steadily until 1979 and 1980 when it increased sharply to $57.6 billion and $82.1 billion, respectively.63 The same pattern is evident for the ratio of the non-oil developing countries’ current account deficit to GNP, which peaked at 5.4 per cent in 1974–75, declined to an average of 3.3 per cent an 1976–78, and then rose again to an average of 4.4 per cent in 1979–80.64
The extent to which current account deficits of this magnitude affect the rate of economic growth depends, in non-oil developing countries as elsewhere, on how much of the deficit can be sustainably financed. For if financing is not available (at acceptable terms), and if both exports cannot be increased much in the short run and imports are relatively insensitive to relative price movements, then the only outlet is a reduction in imports induced by a reduction in real income. And if such imports in turn are important inputs to domestic and export production, the effect on the rate of growth can be severe indeed.
One of the primary reasons why the current account deficits of the non-oil developing countries during 1973—80 did not produce a more pronounced slowdown in their growth was that such financing was available, and on terms (until 1979) that were relatively favorable to them. The low-income countries, which saw their current account deficit rise from 3.1 per cent of GDP in 1973 to 4.7 per cent of GDP in 197565 were able to secure a 60 per cent increase in official development assistance (ODA) in real terms, as industrial countries, oil exporting countries, and multilateral institutions all substantially increased their commitments. After 1975, ODA received by low-income countries declined in real terms but other official capital flows were large enough to enable these countries to increase their nominal holdings of international reserves in all years during 1973–79 except 1975;66 to keep their debt service ratios on long-term external debt below the level of 1973;67 and most important, to increase their volume of imports during 1973–80 at an average annual rate of 3.3 per cent despite an average increase in the purchasing power of their exports of only 0.6 per cent.68
In the case of the middle-income non-oil developing countries, the predominant mode of financing was private long-term capital, most of which represented loans from financial institutions.69 Since the bulk of commercial bank loans to non-oil developing countries carried variable interest rates and since, as previously mentioned, the 1973–80 period was marked by high inflation rates, one might have thought that the terms of financing would have been particularly onerous for middle-income non-oil developing countries. As pointed out in the World Economic Outlook (1981) however, real interest rates paid by the non-oil developing countries were much lower in 1972–80 than in 1960–71. In other words, while nominal interest rates were much higher in the latter period, they were lower in relation to the changes in the non-oil developing country export prices.70 Indeed it was not until 1981 that (Eurodollar) interest rates exceeded the rate of increase of non-oil developing country export prices. In addition to the relatively favorable cost of financing, the average maturity of non-oil developing country debt also was longer in the 1973–80 period (see Colaço (1980)).
As with the low-income countries, the availability of external finance permitted those middle-income non-oil developing countries with relatively poor performance in the purchasing power of exports to maintain higher growth in the volume of imports than would otherwise have been possible. Specifically, other net oil importers were able to expand their volume of imports at an average rate of 3.2 per cent per year (1973–80) despite an average yearly increase of only 1.7 per cent in the purchasing power of exports.71 In contrast, the two non-oil developing country subgroups with the best export performance over the 1973–80 period, namely, major exporters of manufactures and net oil exporters, both had import volume increases over this period that were below the increases in the purchasing power of their exports.72
Although data for investment are considerably more sketchy than for imports, it appears that the availability of external financing also permitted some non-oil developing countries to maintain investment rates in the face of large current account deficits, thereby likewise reducing the impact of the adverse external environment on growth rates. In this connection, World Bank estimates suggest that foreign capital financed approximately 14 per cent of the total investment of all oil importing developing countries during 1975–78.73
Some evidence on the contribution that capital, both domestic and foreign, makes to growth of real income in developing countries is presented in Table 23. The estimated elasticities of real GDP growth with respect to the growth rate of capital range from 0.12 to 0.25. In those studies (e.g., Balassa (1978)) where the sample is composed only of relatively high-income developing countries with large manufacturing sectors, foreign capital emerges as more productive than domestic capital, whereas the opposite result holds in studies (e.g., Michalopoulos and Jay (1973)) that use a more heterogeneous sample of developing countries. It should be recognized that while the estimates in Table 23 confirm that the elasticity of real GNP in developing countries with respect to capital is considerably smaller than that with respect to labor, the evaluation of the contribution of each of these two factors to growth requires that these elasticities be multiplied by the respective average growth rates of the factors themselves. Here, the growth rate of capital in non-oil developing countries is typically about three times greater than the growth of the labor force.74 World Bank figures for the 1970–79 period indicate that the average annual growth rate of gross domestic investment in low-income non-oil developing countries was 6.4 per cent versus 1.9 per cent for the labor force. The corresponding figures for middle-income oil importers were 6.0 per cent and 2.2 per cent, respectively, while those for middle-income oil exporters were 10.3 per cent and 2.5 per cent, respectively.75 Seen in this light, the contribution of capital to growth becomes much closer to that of labor.
|Michalopoulos and Jay (1973)||0.24 (domestic)||0.60|
|Balassa (1978)||0.16 (domestic)||0.92|
Evidence that imports contribute to real GDP growth in non-oil developing countries is more scanty, in large part because imports are usually regarded as a function of real income, rather than the reverse. Nevertheless, Hicks (1980) in a recent cross-country study of growth of real GDP per capita in 83 developing countries during 1960–77, found that “.. .the growth rate of imports continues to be the dominant variable explaining variations in the growth rate of output” (p. 19). This study reports an elasticity of real GDP growth per capita with respect to the growth rate of imports of 0.24—and this elasticity is higher than those for the investment rate and alternative measures of human capital. Hicks’s (1980) results can, therefore, be interpreted as providing support for the view that the crucial channel by which foreign exchange shortages constrain non-oil developing country growth is via their adverse impact on those countries’ ability to import.
Structure of Production
One of the identifying characteristics of non-oil developing countries is that a relatively high share of GDP (and of employment) originates in agriculture. World Bank figures for 1979 reveal that, among the non-oil developing countries, agriculture accounted for 34 per cent of GDP in the low-income countries and 14 per cent in middle-income countries. This can be compared to agriculture’s share in GDP of 4 per cent in industrial countries and of only 2 per cent in capital-surplus oil exporters.76
The clear implication of these differences in the structure of production is that economic growth in low-income non-oil developing countries, as well as in quite a few middle-income non-oil developing countries, is much more sensitive to developments in agricultural production than elsewhere.77 Agricultural production has at least three features that are important for the subject of this paper. First, over the past two decades, the growth of production in agriculture has been less than one half as high as that in either industry or services in all country groups, developing or industrial. For example, agricultural production in low-income non-oil developing countries grew at an average annual rate of 2.0 per cent in 1970–79 versus 4.2 per cent for industry and 4.5 per cent for services. Similarly, in industrial countries the corresponding figures were 0.9 per cent for agriculture, 3.2 per cent for industry, and 3.4 per cent for services.78 Thus, the larger share of a low productivity sector (agriculture) in non-oil developing countries means that their overall growth rate is likely to be lower than elsewhere. Second, because agricultural production is sensitive to weather (especially in those low-income countries where irrigation facilities are limited), and because agricultural products typically comprise a large part of non-oil developing countries’ exports and imports, real GDP growth and the current accounts of many of these countries can change rather markedly from year to year for reasons that have relatively little to do with the state of economic activity in industrial countries.79 Third, agricutural production in non-oil developing countries is also responsive to a whole gamut of domestic policy incentives, ranging from producer prices and availability and cost of inputs to access to advice on farming methods.
Review of individual country experiences by the World Bank in its World Development Report 1981 testifies to the influence of agriculture on growth performance in the non-oil developing countries during the 1973–80 period, and to the sensitivity of agricultural production to relative profitability.80 To take a prominent example, India’s 4.3 per cent average GDP growth rate over 1974–79 is attributed in no small part to its average annual increase of 3.3 per cent in foodgrain production. The introduction of high-yielding seed varieties plus fertilizer and irrigation investment contributed to this latter development. Despite this overall agricultural success story, it is also worth noting that an extreme drought in 1979 led to a sharp cutback in agricultural production and, in turn, to a 4.5 per cent decline in India’s real GNP in that year. As an indication that the right agricultural incentives can spur agricultural production, the World Bank cites the favorable agricultural performance of Ivory Coast, Malawi, and in the 1960s, of Kenya. On the negative side of the ledger, poor agricultural policies are held as partly responsible for the disappointing growth and export performance of several sub-Saharan low-income non-oil developing countries over the 1974—78 period. Low agricultural prices that discouraged production, overvalued exchange rates that discriminated in favor of industry at the expense of agriculture, lack of research on appropriate farming methods for the region’s soil and climate, and inefficient distribution networks for delivering inputs to farmers and for marketing their output—all hurt performance, although drought, wars, and civil conflicts also adversely affected growth in that region.
Economic Policy Strategies
Thus far, we have concentrated on the external factors affecting economic growth in non-oil developing countries. Clearly, however, as in industrial countries and oil exporting countries, the rate of economic growth achieved by individual non-oil developing countries, or groups of them, was also a function of the economic policies pursued by these countries. Aside from the generalized high inflation rates and weak current account positions that placed rather severe constraints on the stance of financial policies in non-oil developing countries during 1973–80,81 it is necessary to focus attention on the types of domestic policies that seem to have been associated with superior development performance among that group of countries.
To this end, the authors have reviewed the policy lessons that emerge from several studies of individual country experiences that have recently been completed by World Bank and IMF staff (i.e., Balassa (1980), Havrylyshyn and Wolf (1980), Keesing (1979), Frank (1981), and IMF (1982)).82 An overview of the main messages for non-oil developing country policy strategies would seem to be the following.83
(i) At the broadest level, outward-looking development strategies have fared far better in terms of economic growth, employment, economic efficiency, and adjustment to external shocks than inward-looking strategies. The outward-oriented strategies are characterized by provision of similar incentives for domestic and export production; the tolerance of import competition for all but a narrow range of domestically produced goods; the maintenance of realistic real exchange rates, realistic public utility prices, and positive real interest rates; and the operation of an automatic and stable incentive system. In contrast, inward-oriented development strategies are typically characterized by protection against imports that extends well beyond the infant-industry stage; a bias in favor of import substitution and against export promotion; wide variation in degrees of import protection across industries; maintenance of overvalued real exchange rates, low public utility prices, and negative real interest rates; greater reliance on administered allocative mechanisms than on relative prices; and prevalence of sellers’ markets that provide little inducement for productivity advances or for meeting users’ (buyers’) needs.
(ii) Although a country’s export performance is partly a function of its economic structure and comparative advantage, there do seem to be a few general guidelines for success. The broad lesson is that good export performance requires that exporting be profitable relative to other activities in the economy; for if it is not, economic agents will instead spend their time, energy, and resources on these other activities. This means providing exporters with a unified and stable (permanent) system of incentives. Foremost among these is a realistic exchange rate, for no other policy instrument in non-oil developing countries will have as pervasive an influence on the relationship between the export price (in local currency) and the costs of producing exports. But the right exchange rate, together with supporting macroeconomic policies, is only one of many conditions necessary for export success. In the case of manufactured exports, access to duty-free or low-duty imported inputs by exporters, provision of infrastructure (e.g., roads, harbors, power), a legal code that does not unduly raise the exporters’ labor costs, provision of credit to exporters on terms that are competitive with that extended to other suppliers, export subsidies (where they can be justified on second best and infant industry grounds and which fall within the guidelines of the General Agreement on Tariffs and Trade), and commitment to quality control by exporters themselves, all have a contribution to make. For agricultural exports there must be a similar system of incentives, including market-related producer prices, access to necessary inputs (fertilizer, seeds, irrigation), provision of information and instruction on suitable farming techniques, and provision of related infrastructure.
(iii) Turning to import substitution, an analogous set of policy lessons can be deduced. First, while a policy of import substitution has played a useful role in most countries’ development of infant industries, its contribution declines rapidly as these industries mature. If import substitution is pushed beyond the infant industry stage, economies of scale are sacrificed and the protected industries typically do not find the competitive environment they need to develop efficiently. Second, if a strategy of import substitution is followed, it is apt to be most costly in those non-oil developing countries with small domestic markets. Similarly, import substitution is most costly when applied to industries (e.g., producer goods—such as steel and chemicals) where economies of scale are known to be important. If such industries are selected for special attention because they are thought to generate favorable externalities, they should be aided by measures (e.g., subsidies) that lower their prices and expand their scale of output. Third, import tariffs are preferable to quantitative restrictions because: (a) the costs of the former are easier to determine; (b) tariffs are less costly to administer and generate revenues for the government rather than scarcity rents for importers; and (c) tariffs are less subject to large sudden changes because they usually require legislative action for confirmation. Fourth, import protection is usually less harmful if applied to consumer goods industries than to producer goods because the latter have more significant input-output links in the economy and hence are needed as inputs by other industries. Fifth, a typical but disappointing outcome of protection against imports is that the terms of trade typically shift against agriculture, with obvious adverse effects on agricultural production. Such a development needs to be avoided, especially in those non-oil developing countries where comparative advantage lies in agriculture. Sixth and finally, those non-oil developing countries which already had allowed imports to compete freely with domestic production were better able to cope with sharp rises in import prices because they could then reduce imports without eliminating crucial imports that are inputs into production. In contrast, in cases where imports are already limited to only crucial imports any further fall in import volumes almost invariably reduces economic growth.
(iv) Guidelines or policy lessons for managing the transition from an inward-looking to an outward-looking development strategy are subject to somewhat more controversy. Nevertheless, Keesing (1979) has made a rather comprehensive set of suggestions for such a transition in the case of a developing country that is suffering from a payments crisis and faltering growth. Some of his recommendations can be compactly stated as follows. Because a highly distorted trade policy cannot be reformed in one or two steps and because the expectations and perceptions of economic agents are crucial for success, the sequence of policy measures should be planned several years ahead, and the general direction of these measures should be announced in advance. The initial measures should be directed almost exclusively at expanding exports.84 Successful export growth in turn will make it relatively easy to liberalize imports later as import capacity increases. A realistic exchange rate, a balanced government budget, and a reduction in the foreign exchange budget until import capacity expands are essential ingredients for a suecessful pro-trade policy, as are moderate wage increases and realistic prices for publicly supplied goods and services. As part of the transition program, savings need to be increased and investment cut back or switched to projects with low foreign exchange requirements. Available foreign exchange should be channeled into maintenance of current production rather than investment, and care should be taken to alleviate the squeeze on working capital in those industrial firms that are reasonably efficient. In the latter part of the transition, the key element is a well-designed, ambitious development program that puts to good use the import and output capacity generated by export growth.
The Reduced-Form Relationship Between Industrial Country Growth and Growth in Non-Oil Developing Countries
The preceding sections make it clear that economic growth in non-oil developing countries depends on a host of internal and external factors, only one of which is the rate of growth in industrial countries. Nevertheless, it may be illuminating to explore what the simple reduced-form relationship between the two growth rates has been, how this relationship differs among the analytical subgroups of non-oil developing countries, and whether it has been growing stronger or weaker over time.
To provide some simple answers to these questions, growth of real GDP in non-oil developing countries was regressed on industrial countries’ real GDP growth for the period 1965–80, and for the subperiods 1965–72 and 1973–80. To minimize the influence of erratic year-to-year changes, the variables in the regression were defined as three-year moving averages of the annual growth rates; however, the qualitative nature of the results is unaffected if the unsmoothed data are used. As with the earlier reduced-form tests, the regressions were estimated for all non-oil developing countries, and for each of the four analytical subgroups. The results appear in Table 24.
|Country Group||Time Period||Constant||Elasticity of Non-Oil Developing Country Growth with Respect to Industrial Country Growth||R2|
|All non-oil developing countries||1965–80||4.55||0.29||0.11|
|Major exporters of manufactures||1965–80||4.21||0.691||0.22|
|Net oil exporters||1965–80||6.42||0.02||0.01|
|Other net oil importers||1965–80||4.09||0.241||0.25|
Indicates statistical significance at the 5 per cent level.
Indicates statistical significance at the 5 per cent level.
The first noteworthy aspect of Table 24 is the low proportion of the variance in non-oil developing country growth rates that can be explained by (a constant and) industrial country growth rates.85 For all non-oil developing countries taken together, industrial country growth explained only 11 per cent of the variance of non-oil developing country real GDP growth over the 1965–80 period. Table 24 also reveals, however, that this finding is quite sensitive to both the groups of non-oil developing countries included in the sample and the time period. Specifically, among the non-oil developing countries, growth rates in the low-income group and in net oil exporters seem to bear no significant relationship to industrial country growth rates, whereas economic growth of major exporters of manufactures and of other net oil importers does seem to be related to industrial country growth. This is consistent with our earlier reduced-form results for the links between industrial country income growth and non-oil developing country export volume, and between non-oil developing country export volume and growth rates where these same two middle-income oil importing country groups showed the most export sensitivity to industrial country income growth and the closest relationship between export performance and economic growth. As for the size of the elasticity of economic growth in the non-oil developing countries with respect to economic growth in the industrial countries, our estimates suggest that each 1.0 per cent increase in industrial country real GDP is associated with about a 0.2 to 0.3 per cent increase in non-oil developing country real GDP, with the elasticity ranging from about 0.7 for major exporters of manufactures and 0.25 for net oil importers to practically zero for low-income countries and net oil exporters.86 In a similar exercise using (year-to-year) changes in real GDP for all non-oil developing countries over the 1960–77 period, Cline (1980) found an elasticity with respect to (lagged) industrial country real GDP growth of 0.21, which is quite close to the estimates in this paper (for all non-oil developing countries). Finally, while there are not enough observations to give much force to the subperiod results, there is a marked tendency for the elasticity of non-oil developing country growth with respect to industrial country growth to be higher in the 1973–80 period than for 1965—72 (where the results are simply not credible). In short, although admittedly crude, the results in Table 24 support the arguments presented earlier in this paper, namely that faster industrial country growth is associated with faster non-oil developing country growth; that this association is, however, relatively weak; that it is strongest for middle-income oil importing countries, namely, major exporters of manufactures and other net oil importers; and that it was probably stronger in the 1973–80 period than during 1963–72.