EXPORTS AND ECONOMIC GROWTH A RE-EXAMINATION OF THE CAUSALITY RELATION IN SIX COUNTRIES, 1981-2005 NUSHIWAT, Munther * Abstract: The paper argues that, in most cases, causality runs from economic growth to export growth. To support this argument, we start by examining the evidence on causality in the empirical studies done on this subject. Then we empirically test the direction of causation between exports and economic growth using Granger s causality test on time-series data of six countries (Brazil, India, Indonesia, South Korea, Mexico and Thailand). The test reveals that; out of the six cases, the direction of causality can be inferred in two cases; one from growth to exports, the other from exports to growth. Finally we consider the early experiences of the developed countries, and the role of the internal supply factors that determine the supply of output including that of exports. Keywords: Exports, Economic Growth, Development, Causality. JEL classification: F19, O19. I. Introduction Import-biased policies prevailed in many developing countries during the 1950 s and early 1960 s. In the late 1960 s and the 1970 s, export-biased policies became increasingly common. These trends in trade polices were based on the changing convictions, among policy makers, that import substitution in the first case, and export promotion in the second case, are causes of economic growth. 1 If we test the relation, between exports and growth empirically, using econometric methods, it is certain that we will find a positive correlation, as all previous studies did. The estimation methods used were conclusive in determining the positive correlation relation, not in determining the causality relation. Since exports are a component of aggregate output, a quantitative assessment of relevant data will show that a positive correlation between them exists. We argue in this paper that, it is more likely that causality runs from economic growth to export growth. After examining the empirical evidence and testing the causality relation using Granger s test, we examine the developed countries early experiences, and we consider the role of the supply factors. 2. Examining the empirical evidence: Almost all of the research, on this subject, tested the existence of correlation, between exports and growth, not the direction of causality. We make reference here to the * Munther Nushiwat, Marymount Manhattan College, New York (NY, USA ), e-mail: nushiwat@yahoo.com 1 The theoretical basis of the argument that exports cause economic growth is that exports ease two constraints on economic growth, the foreign exchange constraint and the savings constraint. In addition, export industries are likely to be efficient, because of improved economies of scale, facing international competition, and are likely to have positive spell-over effects on other sectors in the economy.
most referred-to articles on this subject. For example 2, Emery, 1967; Michaely, 1977; Balassa, 1978; Tyler, 1981; Feder, 1982; Kavousi, 1984; and Doschos, 1989. The direction of causality was concluded, on theoretical basis, to be from exports to growth, giving support to export-biased policies. Four of the main studies that tested the direction of causality relation between exports and growth are the following : Jung and Marshall (1983), tested the direction of causality in 37 developing countries during the period 1950-1981. The exports-to-growth direction was found in four countries while the growth to exports direction was found in three countries. Some of the countries that are widely believed to owe their high rates of growth to export promotion policies, like Korea, Taiwan, and Brazil, provided no support for the export to growth direction of causality. Darrat (1986), tested the direction of causality, between exports and growth, in four countries that are widely believed to owe their high rates of growth to export-biased policies. The four countries are Korea, Taiwan, Hong Kong, and Sangapore, and the period covered is 1960-1982. The results of the study do not support the export-led growth hypothesis. For Korea, Hong Kong, and Sangapore, the tests revealed that exports and growth are causally independent. For Taiwan, causation ran from output growth to exports, which is contrary to the export-led growth hypothesis. Chow (1987), tested the direction of causality in eight countries. One indicated no causality, one had causality running from exports to growth, and six countries had bidirectional causality. From this estimate, the general case is bi-directional. The case of Mexico, the only case with the direction of causality being from exports to growth, pursued an economic policy that was biased against exports and foreign investment. Since the 1940 s and through the period of the estimate (1960-1980), Mexico pursued an import-substitution policy. Non-oil exports, although growing, were declining as a percentage of GDP, from 5.78% in 1960 to 2.92% in 1980, down from 14% in 1940 (Heath, 1998). Mexico appears to owe its steady growth, during the two decades of the estimate, to its import-substitution policy in addition to the increase in revenues from oil exports, not to manufactured exports. In Chow s estimate, exports included oil exports that were significantly growing as of 1973 as a result of the drastic increases in oil prices and the discovery of new oil resources in Mexico. Bahmani-Oskooee et al (1993), tested the direction of causality in 20 developing countries for varying periods between the 1960 s and the 1980 s. The study included three countries (Korea, Taiwan, and Thailand) that are widely believed to follow exportled policies. The study concluded that only Taiwan, out of these three countries, had the direction of causality from exports to growth. Korea and Thailand had a bi-directional causality. The overall conclusion of the study, which attempted to test the hypothesis of export-led growth, was inconclusive. More recent research used country studies and were generally inconclusive. A study of the role of exports in Irish growth for the period 1950-1997 concluded that a bidirectional causality existed between exports and growth of output (Doyle, 2001). Another example is the case of India. The empirical testing of the export-led growth thesis for the period 1950-1996, concluded that causality between exports and growth 2 There are many studies that tested the correlation between exports and growth. We refer here only to a few of them, as we will do in referring to the studies that tested the direction of causality. 6
Nushiwat, M. Exports and Economic Growth. Causality Relation in Six Countries, 1981-2005 runs in both directions and that the causality from output to exports is stronger than that from exports to growth (Chandra, 2003). 3 3. Testing the causality relation between exports and growth To test the direction of causation, using time-series data, between exports and growth we apply Granger s test (1969). Using this method, the direction of causality can be established between two variables, Y and X, as follows: 1. If Y can be predicted better by using past values of X than by past values of Y alone, making the coefficients of the lagged X variable statistically significant, then causality runs from X to Y. 2. If X can be predicted better by using past values of Y than by past values of X alone, making the coefficients of the lagged Y variable statistically significant, then causality runs from Y to X. The above can be explained using the following two equations: Y t = a 0 + a 1 Y t-1 + a 2 Y t-2 + + a i Y t-i + b 1 X t-1 + b 2 X t-2 + + b i X t-i + U t (1) X t = c 0 + c 1 X t-1 + c 2 X t-2 + + c i X t-i + d 1 Y t-1 + d 2 Y t-2 + + d i Y t-i +V t (2) Where t-1, t-2, and t-i are one-year, two-year, and ith-year lags; and U t and V t are taken to be serially uncorrelated white-noise series. The causality test between the two variables X and Y in equations (1) and (2) is as follows: 1. Causality runs from X to Y, if b i (for i=1,2,,n ), the coefficients of the lagged X variable in equation (1), are collectively statistically significant. 2. Causality runs from Y to X if d i (for i=1,2,,n ), the coefficients of the lagged Y variable in equation (2), are collectively statistically significant. 3. Bi-directional causality is concluded if b i and d i (for i=1,2,,n ), in equations (1) and (2) respectively, are collectively statistically significant. 4. The test is inconclusive if b i and d i (for i=1,2,,n ), in equations (1) and (2) respectively, are collectively statistically insignificant. Data: In applying this test of the direction of causality, between exports and growth, the output variable is the annual growth rate of real GDP; and the export variable is the annual growth rate of exports. Both are in real terms, deflated using the GDP deflator for each country. Using growth rates of output and exports, rather than the levels of output and exports, makes their stationarity less suspect, as proven by previous empirical studies. The symbols for the two growth rates, output growth and export growth, are y and x respectively; and the two equations used in the estimates are: y t = a 0 + a 1 y t-1 + a 2 y t-2 + a 3 x t-1 + a 4x t-2 + u t (3) x t = ß 0 + ß 1 x t-1 + ß 2 x t-2 + ß 3 y t-1 + ß 4 y t-2 + v t (4) The a 1 to a 4 and ß 1 to ß 4 are the coefficients of the explanatory variables, and a 0 and ß 0 are the constant terms in the two equations (3) and (4), respectively. The test of causality is applied to six countries. Those that are known to had pursued export promotion policies, and/or had high growth rates, with available data on both 3 An example that supports the exports to growth hypothesis is the case of Bangladesh. The results of an empirical study for the period 1976-2003 suggested that causality runs from exports to growth (Khawaja and Nath, 2005). 7
variables for the years of the test period, 1981-2005. The data for the two variables y and x were calculated after data from International Financial Statistics. The use of two-year lags for the two variables is adequate. Using three-year lags yields consistent results, as using two-year lags, without altering any of the inferences about the direction of causality. We expect that the estimates will indicate that, for most of the countries, causality runs from output to exports. For the rest of the countries, causality runs from exports to growth and in both directions. Table (1) Test of causality between export growth and GDP growth, at the 5% level of significance: Dependant Dependant Null hypothesis Variable: y Variable: x accept:yes Direction Country F-Statistic R 2 F-Statistic R 2 reject:no of causality Brazil.081.016 1.693.253 yes, yes inconclusive India 1.946.280 3.157.387 yes, no y to x Indonesia.578.104.986.165 yes, yes inconclusive Korea.713.125.266.051 yes, yes inconclusive Mexico.290.088.747.130 yes, yes inconclusive Thailand 4.608.480 1.152.187 no, yes x to y y = Growth rate of real GDP, x = Growth rate of exports, in real term. Period: 1981-2005. The results of the test of the direction of causality are presented in table (1). According to these results, in two cases out of the six cases tested above; we reject the null hypothesis and accept the alternative hypothesis, at the 5% level of significance, that the parameters are different from zero. These two cases are India and Thailand. In the case of India, causality runs from output growth to export growth. While in the case of Thailand, causality runs from export growth to output growth. 4 Considering previous studies, the results are not completely unexpected. Tests of causality of the relation between exports and growth in most previous studies were generally inconclusive. 4. The developed countries early experiences. Historically, exports did not have a leading role in the early development of the developed countries. It was economic growth that caused export growth. Evidence can be shown in the following three cases: 1. Western Europe, 2. The United States 3. Japan. 1. In Western Europe s early experiences of industrialization and development, export growth was not the cause of economic growth. Internal factors, that brought technologies based on inventions, and the release of labor from a breaking-down feudal system to the expanding industrial centers, are credited with the high rates of growth in a new system of mass production during and after the industrial revolution. Exports, when added to the increasing internal demand, contributed to economic growth in later stages. 2. The United States nineteenth century s experience showed few signs of exportled growth. As pointed out by Kravis (1970), exports remained a small proportion of, and 4 In the case of Brazil, only at the 25% level of significance the null hypothesis is rejected, inferring that causality runs from output growth to export growth. 8
Nushiwat, M. Exports and Economic Growth. Causality Relation in Six Countries, 1981-2005 lagged behind, total output and were concentrated in agriculture, which had a slower growth rate than the rest of the economy. The United States, a successful and documented case, owed its pattern and speed of development to internal factors. 3. In the case of Japan, in the 1930 s, pressure of military requirements caused steel, machinery, chemicals, and other heavy industries to dominate industrial expansion. After the war, the heavy industries were ready to resume a leading role in industrialization (Danison and Chung, 1976). Exports grew as a result and contributed to the growth of the growing economy. The study by Boltho (1996) investigated whether Japan s growth was caused by exports or domestic forces during the periods 1913-1937, 1952-1973, and 1973-1990. The results of five very different tests suggest that domestic forces, rather than exports, were behind Japan s exceptional performance during these periods, with overwhelming evidence for the period 1952-1973. However, there were examples of short-time periods during which cyclical upswings were influenced by foreign demand. 5 In all three cases, exports, although were growing, were not the cause of economic growth but the result of it. 5. Considering the supply factors For a developing country, the existence or increase in demand, whether external or internal, does not necessarily cause an increase in output and employment. This is because the sources of supply are not sufficiently developed to expand the supply of output, including that of exports. To support this argument, we will examine the following: 1. The role of government s policy. 2. The economic and institutional constraints on growth. 3. The allocation of resources according to comparative advantage. 4. The limitations on export growth. 1. There are differences between the developed countries experiences, in their early stages of development, and the developing countries recent experiences. As pointed out by Amsden (1992); for the first group, pioneering technology as the basis of competitiveness, was the principal reason for economic development in the first and second industrial revolutions. For the second group, denied the competitive technological advantage, they had to grow by borrowing technology and with more state intervention than was needed in the case of the developed countries. The major factor behind the newly industrialized countries high economic growth rates, in the 1970s and 1980s, were government s development policies with the objective of the overall development of the economy. Policies of export subsidies, incentives, and currency devaluations were implemented to increase exports, with the belief that export-led growth was the vehicle for development. These export-led policies were part of the overall development policies that included building the needed infrastructure, training and education, and expansionary fiscal and monetary policies. 2. An export-biased policy, under favorable foreign demand conditions, is expected to stimulate the supply of exports to a certain extent. But the economic and institutional constraints on supply growth, including the supply of exports, are more likely to be eased through overall development plans. It is true that the two major economic constraints on growth (foreign exchange and savings) are expected to be reduced by export growth. However, a certain level of development is needed for exports to be effective in influencing growth. 5 These periods are 1914-1981, 1931-1933, 1975-1977, and 1983-1985. 9
3. Allocation of resources, according to comparative advantage, that is caused by export growth, may not be effective in causing economic growth. If comparative advantage is the basis of trade, it does not necessarily lead to economic development. This is because of the costs of worsening terms of trade of light manufactured and primary goods, as they are traded against manufactured goods of the industrialized countries; in addition to the costs of export promotion policies like currency devaluations and export subsidies. The newly industrialized countries main exports, in the early stages of development, consisted mainly of light manufactured goods. Few of them moved to higher stages of technology in later stages. 4. There are limitations on the growth of exports that may be caused by three factors: (a) An increase in foreign competition, as more developing countries pursue export-led growth policies and produce similar competing products for export. (b) The problem of accessibility to foreign markets that may face an exporting developing country when its exports start to harm the importing developed countries' workers and producers. The developing exporting countries may have to seek alternatives to export promotion policies when the developed importers demand the removal of barriers facing their own exports to the developing countries. 6 (c) The appreciation of the domestic currency of the exporting country as exports and the current account surplus increase, what is referred to as the Dutch disease. Finally, positive effects of exports on growth can be found in Guisan(2006) and (2007). Guisan analyzed bilateral relationships between foreign trade and development, having into account demand and supply relationships and concludes that industrial development usually favors exports and the increase of exports usually implies an increase of imports, which have an average positive effect on production per capita. The positive effects of imports, as complementary factors of production for industrial and/or non industrial sectors is remarkable, and the total effect of foreign trade is positive on real production and income per inhabitant. 6. Concluding remarks: 1. The previous empirical evidence does not provide substantial support to the hypothesis that export growth causes economic growth. The studies that teste d the causation relation were generally inconclusive. 2. The test results, in this paper, were generally inconclusive. The direction of causality was inferred only in the cases of India and Thailand. Causality ran from output to growth in the first case, and from exports to growth in the second case. 3. The early experiences of economic growth of the industrialized countries were not export-led. Examining the cases of Western Europe, the United States, and Japan reveals that economic growth preceded and caused exports to grow. When exports grew, at later stages, they contributed to economic growth. 4. Demand for exports, in the other countries, is not the only determinant of the exports of a country. The domestic supply factors, of the exporting country, must be sufficiently developed to respond to the demand for exports. 5. Considering the case of China, the latest and largest economy to pursue export-biased policies; China was able to pursue these policies because it had attained a certain level of 6 As an example, the unresolved trade issues between the United States and China include the exchange rate between their currencies. 10
Nushiwat, M. Exports and Economic Growth. Causality Relation in Six Countries, 1981-2005 development. References Amsden, Alice (1992), "A Theory of Government Intervention in Late Industrialization", in "State and Market in Development: Rivalry or Synergy", edited by Putterman and Rueschemeyer, Lynne Rienner Publishers, Boulder, Co. Bahmani-Oskooee, M., Mohtadi, M., and Shabsigh, G. (1993), Exports, Growth, and Causality in LDCs, A Reexamination, Journal of Development Economics, 27, pp.535-542. Balassa, Bela (1978), "Exports and Economic Growth, Further Evidence", Journal of Development Economics, 5, pp.181-189. Bolto, A. (1996), Was Japanese Growth Export-led, Oxford Economic Papers, 48, PP. 415-432. Chandra, Ramesh (2003), Reinvestigating Export-led Growth in India Using a Multivariate Cointegration Framework, The Journal of Developing Areas, vol. 37, no 1. pp.73-86 Chow, Peter (1987), "Causality Between Export Growth and Industrial Development, Empirical Evidence from the NICs", Journal of Development Economics, 26, pp. 55-63. Darrat, Ali (1986), Trade and Development: The Asian Experience, Cato Journal, vol. 6, pp. 695-699. Denison, Edward, and Chung, William (1976), "Economic Growth and its Sources" in "Asia s New Giant", edited by Patrick, Hugh and Rosovsky, Henry, the Brookings Institution, Washington, D.C. Doschos, Demetrios (1989), Export Expantion, Growth, and the Level of Economic Development: An Empirical Analysis, Journal of Development Economics, 30, pp. 93-102. Doyle, Eleanor (2001), Export-Output Causality and the Role of Exports in Irish Growth: 1950-1997, International Economic Journal, vol. 15, no.3, pp.31-53. Emery, R. (1967), "The Relation of Exports and Economic Growth", Kyklos, vol. 20, pp. 470-486. Feder, Gershon (1982), "On Exports and Economic Growth", Journal of Development Economics, 12, pp. 59-73. Greene, W. (1997), Econometric Analysis, Third edition, Prentice Hall, Upper Saddle River, NJ. Guisan, M.C.(2006). Industry, Foreign Trade and Development: Econometric Models of Europe and North America, 1965-2003, International Journal of Applied Econometrics and Quantitative Studies, Vol. 3-1, pp.5-30 Guisan, M.C.(2007). Industry, Foreign Trade and Development: Econometric Models of Africa, Asia and Latin America 1965-2003, International Journal of Applied Econometrics and Quantitative Studies, Vol. 4-1, pp.5-20 Gujarati, Damodar (1995), Basic Econometrics, third edition, McGraw-Hill, New York. Granger, C (1969), Investigating Causal Relations by Econometric Models and Cross Spectral Methods, Econometrics, no 3, pp. 424-438. Heath, Jonathan (1998), "Orig inal Goals and Current Outcomes of Economic Reform in Mexico", in "Mexico s Private Sector", edited by Roett, Riordan, Lynne Rienner Publishers, Boulder, Co. IMF, International Financial Statistics, different issues. Jung, Woo, and Marshall, Peyton (1985), Exports, Growth, and Causality in Developing Countries, Journal of Development Economics, 18, pp. 1-12. 11
Kavousi, Rostam (1984), "Export Expansion and Economic growth, Further Empirical Evidence", Journal of Development Economics, 14, pp.241-250. Khawaja, A. and Nath, H. (2005), Export-led Growth in Bangladesh: A Time Series Analysis, Applied Economic Letters, vol. 12, no.6 Kravis, Irving (1970), "Trade as a Handmaiden of Growth: Similarities between the Nineteenth and Twentieth Centuries", Economic Journal, 80, pp. 850-873. Michaely, Michael (1977), "Exports and Economic Growth, an Empirical Investigation", Journal of development Economics, 4, pp.49-53. Tyler, William (1981), Growth and Export Expansion in Developing Countries: Some Empirical Evidence, Journal of Development Economics, 9, pp. 121-130. Annual growth rates in real GDP and real exports: y: Annual growth rate in real GDP (in percentages).x: Annual growth rate in real exports (in percentages). Br Br In In Id Id Ko Ko Mx Mx Th Th Year y x y x y x y x y x y x 1981-4.2 21.3 6.5 3 7.9-1.3 6.7 14.3 8.6 6.2 5.9 4.9 1982 0.7-9.3 3.8 5.3 2.2-11.2 7.3 1.1-0.6 13.7 5.4 1.2 1983-2.8 14.3 7.4 4.6 4.2 10.2 11.8 14.9-3.9 11.5 5.6-7.3 1984 5.4 22.1 3.7 13.3 7 7 9.4 9.6 3.6 10.5 5.8 15.2 1985 7.9 6.9 5.5-12.5 2.5-11.1 6.9 2.7 2.6-4.5 4.6 10.7 1986 7.5-17.7 4.9 4 5.9-6.9 11.6 23.2-3.8 5.6 5.5 15.8 1987 3.5 16.5 4.8 12.9 4.9 28.8 11.5 19 1.9 9.4 9.5 24.2 1988 0 18.5 9.9 18.3 5.8 7.9 11.3 6.4 1.2 5.8 13.3 29.1 1989 3.1 0 6.6 23.5 7.5 11.9 6.4-9.3 3.3 5.7 12.2 19.1 1990-4.4-8.3 5.7 5.7 7.2 12.1 9.5-0.4 4.4 5.3 11.6 8.6 1991 0.2 4 0.5 20.9 7 10.3 9.1 3.2 3.6 5.1 8.1 12.6 1992-0.8 0 4.6 9.8 6.5 14.3 5.1 7.8 2.8 5 7.6 10.4 1993 4.1 9 3.5 8.1 6.5-6.2 5.8 7.2 0.6 8.1 8 10.9 1994 5.7 6.4 7.9 7.5 7.3 3.9 8.4 10 4.4 17.4 8.7 11.7 1995 4.3 3.4 7.4 17.9 8.2 7.6 9.2 18.1-6.1 33 9.3 17.6 1996 2.6-1.2 7.7 3.2 7.9 5.7 7.1 3.4 5.1 5.5 5.9-0.6 1997 3.3 4.9 4.8 6.9 4.7 13.1 4.6 21.7 6.8 0.7-1.4 20.7 1998 0.2 10.6 6.5 9.6-13.1 65.3-6.9 32.7 5 6.2-10.5 9.7 1999 0.7-1.1 6.1 10.2 0.7-32.2 9.5-7.4 3.8 4.1 4.5 3.3 2000 4.4 7.5 4.4 18.3 4.9 21 8.5 13.4 6.6 7.4 4.7 20.1 2001 1.3 11 5.8 1.7 3.8-3.3 3.8-3.8 0-11.3 2.2 0.7 2002 2 9 3.7 18.1 4.4-12.6 7-0.3 0.7-1.9 5.3 2.7 2003 0.5 8.3 8.5 10.3 4.7-1.3 3.1 11 1.5 5.4 7.1 9.3 2004 5 19.8 7.6 40 5 11.8 4.7 21.5 2 10.4 6.2 14.3 2005 2.2-4.8 8 21.3 5.6 10.4 4 0.4 2.9 4.2 4.5 9 Source: Calculated after data from International Financial Statistics, different issues. Journal publis hed by the EAAEDS: http://www.usc.es/economet/eaa.htm 12