Exports and Economic Growth: Further Evidence

Transcription

1 Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized World Bank Reprint Series; Number Sixty-eight Bela Balassa Exports and Economic Growth: Further Evidence Reprinted from Joui rial of DeveloInnent Ecoinomics 5 (1978)

2 Journal of Development Economnics 5 (1978) O North-Holland Publishing Company EXPORTS AND ECONOMIC GROWTH Further evidence Bela BALASSA* Johns Hopkins University, Baltimore, MD 21218, U.S.A. The World,Bank, Washington, DC 20433, U.S.A. Received July 1977, revised version received September 1977 This paper investigates the relationship between exports and economic growth in a group of tleven developing countries that have already established an industrial base. Separate consideration is given to manufactured and to total exports; in the case of the latter, adjustment is made for domestic and foreign investment and for increases in the labor force. 1. Conceptual uia measurement issues In examining the effects of exports on economic growth in countries which have established an industrial base, we test the hypothesis that export-oriented policies lead to better growth performance than policies favoring import substitution. This result is said to obtain because export-oriented policies, which provide similar incentives to sales in domestic and in foreign markets, lead to resource allocation according to comparative advantage, allow for greater capacity utilization, permit the exploitation of economies of scale, generate technological improvements in response to competition abroad and, in labor-surplus countries, contribute to increased employment.' In turn, once the 'easy' stage of import substitution is over, substituting domestic production for imports entails rising costs due to the loss of economies of scale in small national markets and the relatively capital intensive nature of *The author is Professor of Political Economy at the Johns Hopkins University and Consultant at the World Bank. This paper was prepared in the framework of the Development Strategies in Semi-industrial Countries research project undertaken at the World Bank and directed by the author. He is indebted to Ian Little, Paul Streeten, Larry Westphal, and an anonymous referee for helpful comments. 'Some of these factors (e.g. increases in employment) represent a once-for-all gain while others (e.g. technological change) will have a continuing effect. Furthermore, to the extent that marginal saving ratios exceed average ratios, the rate of economic growth will be permanently higher under export orientation as compared to import substitution under protection. As noted below, this conclusion also follows if increases in factor supplies are accompanied by a shift of existing resources from export industr6s into import substituting industries under protection. Reprinted by permission from Journal of Development Economics 5 ( 1978, Ncrth- Holland Publishing Co.), pp

3 182 B. Balassa, Exports and economic growth the products involved. As a result, the domestic resource cost of saving foreign exchange through continued import substitution under protection will exceed the domestic resource cost of earning foreign exchange through exports and the difference will tend to increase over time. 2 Intercountry differences in trade policies may be expressed in a variety of ways. Under one alternative, intercountry differences in the growth of exports are taken to reflect the extent of export orientation; i.e. the choice between using resources for exports as against import substitution under protection. Thus, ceteris paribus, different rates of export growth will be associated with differences in trade policies. 3 It is of further interest to examine the relationship between export growth and the growth of GNP net of exports. In an intercountry context, the correlation between these variables may be taken to reflect the indirect effects of exports operating through changes in incomes and costs. 4 In turn, the correlation between export growth and GNP growth will provide an indication of the total (direct plus indirect) effects of exports on economic growth. In the case of countries that started from a low base, export growth rates are sensitive to the choice of the initial year. Correspondingly, further interest attaches to relating the absolute increment in exports to the absolute increment in GNP, which is less sensitive to the choice of the base year. This measure, the so-called incremental export-gnp ratio, too, can be transformed so as to relate the increment in exports to that of GNP net of exports. All these measures, as well as a variant of the measure proposed by Michaely, 5 have been used to investigate the relationship between export expansion and economic growth in an eleven-country sample. The sample includes countries which have established an industrial base; within the con- 2An extreme case of import substitution under protection is that considered by Harry Johnson (1967), where increases in the capital stock bring about a shift of labor into the protected capital-intensive industry that, in turn, gives rise to a decline in the economy's consumption possibilities. And, even if absolute impoverishment will not ensue, the shift of labor into the protected capital-intensive industry will adversely affect efficiency and growth. 3 This alternative has been objected to by Michaely on the grounds that 'since exports are themselves part of the national product, an autocorrelation is present; and a positive correlation of the two variables is almost inevitable, whatever their true relationship with each other' (1977, p. 50). However, import-replacing domestic production, too, is part of the national product, so that in an intercountry framework export growth rates reflect alternative uses of resources. 'There will be indirect effects emanating from exports to the extent that the size of the stimulus will be larger than under import substitution, leading to multiplier effects if resources are not fully utilized. Also, exporting permits cost reductions for goods sold domestically through the exploitation of economies of scale in industries producing for domestic as well as for foreign markets. «While Michaely correlates the average size of the annual changes in the ratio of exports to GNP with the average annual change in per capita GNP, it is more appropriate to use the rate of growth of GNP as the variable affected by export growth. This adjustment, however, changes the results but little.

4 B, Balassa, Exports and economic growth 183 straints imposed by datl availability, they have been chosen so as to provide representation of the -rincipal tendencies as far as trade policies in developing countries are concer-ied. Among the countries included in the sample, Korea, Singapore and Taiwan adopted export-oriented policies at an early stage; they provided essentially free-trade treatment to exports; and granted some additional subsidies which equalized, on the average, incentives to exports and to import substitution. Israel and Yugoslavia started early with export promotion but their efforts slackened somewhat afterwards. Argentina, Brazil, Colombia, and Mexico continued further with import substitution and provided export incentives only from the mid-sixties onwards; at the same time, these countries did not generally ensure exporters free access to imported inputs. Finally, Chile and India pursued inward-oriented policies throughout the period under consideration, and their relatively weak export promotion efforts were frustrated by farreaching controls on imports and investment. 6 The period of investigation chosen is Since in a number of the countries included in the sample policy changes occurred in the mid-sixties, calculations have also been made for the and subperiods. The estimates concern the relationship between total exports and GNP as well as that between manufactured exports and manufacturing output. 2. The empirical estimates Table 1 shows the empirical results obtained by the use of the Spearman rank correlation coefficient, which has also been employed by Michaely. It is apparent that the statistical significance of the estimates generally improves between the first and the second subperiods. These differences in the subperiod results may be explained, at least in part, by the relatively low level of manufactured exports in several countries at the beginning of the period. Not surprisingly, a weaker relationship is obtained in regard to the indirect - as compared to the total - effects of exports. This is the case in particular for estimates made on the national economy level, where calculating indirect effects involves deducting the value of exports from GNP that is a value added concept. Nevertheless. the results for the entire period, and for the second subperiod, point to the importance of the indirect effects of exports. With one exception, the observed correlations in regard to the total effects of exports are higher for GNP than for manufactured output. The opposite conclusion applies, however, to the indirect effects of exports, again reflecting the sensitivity of the estimates to deducting the value of exports from GNP. Finally, the results obtained by the use of the Michaely measure for the entire period, as well as for the two subperiods, show a higher rank correlation than 6 For a description of the policies applied, see Balassa (1977).

5 184 B. Balassa, Exports and economic growth Table I Spearnian rank correlation coefficients between exports and outpr"t growth in selected developing countries.' Manufacturing Total Export growth vs the growth of output (0.112) (0.001) (0.008) (0.001) (0.001) (0,001) 2. Export growth vs. the growth of output net of exports (0.301) (0.002) (0.005) (0.067) (0.004) (0.003) 3. Incremental exportoutput ratios vs. the growth of output (0.001) (0.011) (0.002) (0.008) (0.001) (0.002) 4. Incremental exportoutput ratios vs. the growth of output net , of exports (0.229) (0.008) (0.002) (0.458) (0.010) (0.031) 5. Increments in exportoutput ratios vs. the growth of output (0.178) (0.052) (0.067) (0.035) (0.002) (0.003) 6. The average ratio of exports to output vs the growth of output (0.001) (0.018) (0.005) (0.163) (0.003) (0.008) "Notes: (1) Manufacturing exports have been related to the output of the manufacturing sector and total exports to GNP. The relevant export-output ratios are averages for the periods concerned. (2) Levels of significance are shown in parentheses. (3) Regression equations estimated by the inclusion of a constant term are not shown because of the lack of statistical significance of the constant. that obtained in Michaely's sample of 23 countries, with GNP per head exceeding 300 dollars (0.523). The differences may be explained by reference to the heterogeneity of the sample used by Michaely that includes countries, such as Algeria, Guatemala, and Paraguay, which rely chiefly on primary exports as well as countries, such as Cyprus, Malta, and Panama, where service exports tend to dominate. 7 The greater homogeneity of our sample may also explain that we have obtained a relatively high correlation between export shares and the growth of GNP (0.703 for ) while Michaely found no such relationship t will be observed that the relationsbip between exports and economnic growth will not be invariant to the commodity composition of exports, as the possibilities of exploiting economies of scale and increased capacity use have relevance primarily for manufacturing industries. 8 Note further that the Michaely measure gives a somewhat lower correlation in cxport- GNP relationships than the measures proposed here, with larger differences shown for manufactured exports. These differences may be explained by the sensitivity of the Michaely measure to random factors resulting from the fact that it uses incremental export shares.

6 B. Balassa, Exports and economic growth Adjusting for factor inputs Attempting to explain GNP growth in terms of export growth has the disadvantage of omitting other relevant variables. Michalopoulos and Jay (1973) attempted to remedy this deficiency by using domestic and foreign investment and labor as explanatory variables together with exports in an intercountry regression designed to explain intercountry differences in GNP growth rates. The inclusion of exports in a production function-type relationship is warranted on the grounds that exports tend to raise total factor productivity for reasons adduced in the introduction. Using data for 39 developing countries in the period , Michalopou'los and Jay found that intercountry differences in domestic and in foreign investment and in labor growth explained 53 percent of the intercountry variation in GNP growth rates, while adding an export variable raised the coefficient of determination to These results are reproduced in eqs. (1) and (2) in table 2. We have applied the method utilized by Michalopoulos and Jay to the pooled data of ten out of the eleven countries under study for the periods and ; for lack of some of the relevant data, Singapore had to be excluded. The results shown in eqs. (3) and (4) indicate that adding the export variable in the regression equation raises the coefficient of determination from 0.58 to At the same time, all regression coefficients are significant at the 5 percent level in both equations. The regression coefficient of the export variable has the same value as in the Michalopoulos-Jay equation, indicating that a one percent increase in the rate of growth of exports is associated with a 0.04 of 1 percent increase in the rate of growth of GNP, In turn, the coefficients of the foreigrn capital and the labor variables are higher, and the coefficient of domestic capital is lower, than in the earlier study. The latter result may be explained by the fact that in most of the ten countries domestic investment was rising rapidly during the period under c'j.szu,ation and its effects were not yet fully absorbed in the national economy. As shown in equations (5) and (6), the results are not substantially affected if the current dollar value of exports is replaced by the purchasing power of exports or by the incremental export-gnp ratio. In the latter case, a 1 percent increase in the rate of growth of exports appears to be associated with a 0.05 of 1 percent increase in the rate of growth of GNP. The results obtained as regards the relationship between export and GNP growth may be compared to the estimate derived in a cross-section investigation of a partly-overlapping group of ten countries by Anne Krueger (1977, ch. XI) who found that an increase in the rate of growth of exports of 1 percent tends to raise the rate of growth of GNP by 0.06 of 1 percent. But while Krueger's estimates reflect an adjustment for a time trend, no adjustment has been made for changes in labor and capital. 9 At the same time, Krueger has found that, on 9 Regressing the GNP variable on the exporl variable alone, we have obtained elasticity values of 0.06 to 0.07.

7 186 B. Balassa, Exports and economic growth Table 2 Intercountry regression analysis of the growth of GNP.' Coefficients of independent variables Equation Dependent -- no. variable KD Kr L X PPX IXR A 2 1 Y (7.81) (3.35) (2.44) Y (9.62) (2.33) (2.81) (4.82) Y U (3.23) (2.42) (1.74) Y (3.33) (2.40) (1.99) (3.57) 0,77 5 Y (3.59) (2.44) (1.82) (3.34) Y (2.32) (2.32) (1.66) (1.86) 0.65 "Note: Eqs. (1) and (2) have been taken from Michalopoulos and Jay (1973); eqs. (3) to (6) have been estimated for the pooled data of ten countries for and The gross national product (Y), and labor (L) have been expressed as the ratio of the absolute change between the initial and the terminal year divided by in;tial year values. Thc same procedure has been followed for exports, which have alternatively been expressed in ternms of current dollar values (X) and in terms of their purchasing power (PPX), derived by deflating dollar values by the index of unit values of manufactured exports of the developed countries. IXR is the incremental export-gnp ratio as defined in the text. K, is the average current account balance during the period in question, expressed as a proportion of initial year GNP, and KD the average difference between gross fixed capital formation and current account balance, expressed as a proportion of initial year GNP. Data on labor refer to the labor force rather than employment. the average, countries with liberalized regimes had a GNP growth rate 0.7 percent higher than others even after differences in export performance are taken into account. 4. Implications of the results 'The next question concerns the implications of intercountry differences in export performance for the rate of economic growth. For this purpose, we have compared predicted values derived from eq. (4), calculated by using actual export growth rates for the individual countries, with hypothetical values derived from the same equation on the assumption that each country had the average export growth rate calculated for the group as a whole. The calculations have been made for the period According to the results thereby obtained, the increase in Korea's GNP would have been 37 percent smaller if its export growth rate equalled the average

8 B. Balassa, Export and economic growth 187 for the countries concerned. The corresponding proportion is 25 percent for Taiwan. At the other extrem-, in Chile, India, and Mexico, respectively, the increase in GNP would have been 14, 12 and 8 percent greater if these countries had average export growth rates. The corresponding estimates for the remaining countries of the group fall in the - 6 to +9 percent range (table 3). The differences are accentuated if the results are expressed in per capita terms. Thus, according to the calculations, increases in per capita incomes in Korea would have been 43 percent smaller and in Taiwan 33 percent smaller, if export growth rates in these countries had been identical to the average for the suanple as a whole. Conversely, increases in rer capita incomes in Chile, India, and Mexico, respectively, would have been 21, 22, and 17 percent higher in such an eventuality. As noted earlier, export growth rates are sensitive to the choice of the base year. This is the case, in particular, in Brazil where exports in 1966 were at a low level due to the adverse effects of import substitution policies. Coffespond- Table 3 Hypothetical gain or loss in the growth rates of GNP and per capita GNP assuming average export growth rates and incremental export-onp ratios.' Export growth Incremental export-onp ratio GNP Per cap. GNP GNP Per cap. GNP Argentina Brazil Chile Colombia India Israel Korea Mexico Taiwan Yugoslavia , 'Note: Hypothetical growth rates have been calculated from eqs. (4) and (6), under the assumption that the country in question had the average growth rate of exports or average incremental export- GNP ratio calculated for the sample group as a whole. The difference between these hypothetical growth rates and the predicted growth rates from the same equations for the country concerned has been designated as the gain (+) or loss (-) in the growth rate of GON and per capita GNP assuming average export growth and incremental export-onp ratios. F

9 188 B. Balassa, Exports and economic growth ingly, we have also made estimates using incremental export-gnp ratios.' 0 As shown in table 3, the results by-and-large correspond to the estimates derived by the use of export growth rates. These results point to the fact that trade orientation has been an important factor contributing to intercountry differences in the growth of incomes. It is further apparent that income increments have been achieved at a substantially lower cost in terms of investment in countries that have followed a consistent policy of export orientation. Thus, taking the period as a whole, incremental capital-output ratios were 1.76 in Singapore, 2.10 in Korea, and 2.44 in Taiwan. At the other extreme, these ratios were 5.49 in Chile and 5.72 in India [World Bank (1976)]. During the same period, incremental capital-output ratios were between 3 and 5 in the remaining countries of the sample, with improvements shown over time in countries with increased export orientation. While figures for subperiods are subject to considerable error, it appears that the greatest improvement was experienced in Brazil following its pronounced policy change. Brazil's incremental capital-output ratio declined from 3.84 in to 2.06 in , when the low figure for the second period presumably also reflects increased capacity utilization at higher export levels. In indicating that export growth favorably affects the rate of economic growth over and above the contributions of domestic and foreign capital and labor, the estimates presented in this paper provide evidence on the benefits of export-orientation as compared to policies oriented towards import substitution. At the same time, the empirical results tend to underestimate the effects of export growth on the growth of GNP, since the method applied does not take account of the implications of export growth for the other variables included in the equation. Yet, there is evidence that exports and domestic savings are positively correlated [Weisskopf (1972)]. Also, the improved balance-ofpayments situation attendant on the expansion of exports may increase the attractiveness of the country concerned for foreign capital. 'OIt may be suggested that these ratios are affected by the size of the country. This is not the case in our sample, however. Thus, the bottom group of countries includes Chile that kgs the smallest population Xfter Israel. In turn, the group that experienced the least deviations btween the two sets of calculated values includes countries with greatly different population size, such as Argentina, Brazil, Colombia, Israel and Yugoslavia. Finally, among countries in the first group, Korea had the third largest population in our country sample. References Balassa, Bela, 1977, Export incentives and export performance in developing countries: A comparative analysis, World Bank Staff Working Paper No. 248 (World Bank, Washington, DC). Krueger, Anne 0., 1977, Foreign trade regimes and economiic development: Liberalization attempts and consequences (forthcoming). Johnson, Harry G., 1967, The possibility of income losses from increased efficiency or factor accumulation in the presence of tariffs, Economic Journal 77,

10 B. Balassa, Exports and economic growth 189 Michaely, Michael, 1977, Exports and growth: An empirical investigation, Journal of Development Economics 4, Michalopoulos, Constantine and Keith Jay, 1973, Growth of exports and income in the developing world: A neoclaisical view, AID Discussion Paper No. 28 (Agency for International Development, Washington, DC). Weisskopf, Thomas E., 1972, The impact of foreign capital inflow on domestic savings in underdeveloped countries, Journal of International Economics 2, World Bank, 1976, World Tables 1976, Washington, DC.