Journal of Development Economics26 (1987) 55--63.North-Holland
CAUSALITY B E T W E E N E X P O R T G R O W T H A N D I N D U S T R I A L DEVELOPMENT
Empirial Evidence from the NICs* Peter C.Y. C H O W Hoover Institution, Stanford, CA 94305, USA Southeastern Louisiana University, Hammond, LA 70401, USA
Received August 1985, final version received May 1986 This paper investigates the causal relationship between export growth and industrial development in eight Newly Industrializing Countries (NICs). Results of Sims' causality test show that for most of the NICs, there is a strong bidirectional causality between the growth of exports and industrial development. These findings support the export-led growth strategy in that expansion in exports not only promote the growth of national income but also lead to structural transformation of the developing countries.
1. Introduction Development economists have exhibited increased interest in recent years in the relationship between export expansion and economic growth. They have investigated the effects of export expansion upon the economies of the developing countries from various aspects. For example, contributions of export growth to developing countries have been measured by its impacts on the enhancements of (1) national income [Michaely (1977)], (2) production of non-export goods [HeUer and Porter (1978)], (3) capital efficiency and capability to manage external shocks [Balassa (1978,1981)], (4) the scale effects and externalities [Tyler (1981)], (5) resources reallocations [Feder (1982)], and (6) the total factor productivities [Kavoussi (1984)]. Most of the researchers have concluded that export growth has made substantial contributions to the economies in the developing countries, but only a few of them have ever thoroughly investigated the specific impacts of exports upon the
*This research was funded by a grant from the College of Business at Southeastern Louisiana University. The author wishes to express his gratitude to an anonymous referee of this journal for helpful comments on an earlier version of this paper. Constructive discussions on this project with T.W. Anderson, Jagjit Brar and Eden Yu are also gratefully acknowledged. However, the usual caveat applies. 0304-3878/87/$3.50 © 1987, Elsevier Science Publishers B.V. (North-Holland)
56
P.C.Y. Chow, Export growth and industrial development
development of manufacturing industries or the overall industrialization of the less developed countries (LDCs).I The purpose of this study is to test causality between the growth of exports of manufactured goods and development of manufacturing industries in selected developing countries during the 1960's and 1970's. 2 Our objective is to empirically validate the a priori assumption regarding the existence of causal relationship between expansion of exports of manufactured goods (X) and growth of manufacturing outputs (MFG) in the newly industrializing countries (NICs). Although the causal patterns established in this study may not be applicable to all LDCs, the findings shed considerable light upon the additional contribution of export growth to successes realized by exportoriented developing economies. Establishment of the causal pattern between export growth and industrial development in the NICs has important implications for development strategies. If there is a definite unidirectional causality from export expansion to the development of manufacturing industries ( X ~ M F G ) , then it will lend credence to the export-led growth strategy; exports will not only promote the growth of national income, but also lead to structural transformation in the developing economies. 3 If the causative process is of the opposite direction ( M F G ~ X ) , then it would imply that the development of manufacturing industries may be a prerequisite for developing countries to expand their exports. If the causative process is bidirectional (X~-MFG), then export growth and the development of manufacturing industries have a reciprocal causal relationship. However, if there is no definite causality between export growth and the development of manufacturing industries, then alternative strategies rather than export-promotion may be needed to structurally transform the developing countries. In light of the above, it is obvious that a rigorous test of causality between 1In these studies, some researchers have employed non-parametric tests, whereas others used regression analyses. For example, Balassa estimated the Spearman rank correlation coefficients between manufacturing exports and outputs of manufactured goods, and found that these coefficients were generally lower than those of total exports and GNP for a more homogeneous group of countries during the 1960-73 period [Balassa (1978)]. Tyler found that the expansion of manufacturing exports and GDP growth had a strong positive relationship for all middle income countries [Tyler (1981)]. Also, Kavoussi found that 'in the more advanced developing countries, the effect of export expansion on the growth of total factor productivity is very sensitive to the share of manufactured goods in total exports' [Kavoussi (1984)]. 2The causality test is very sensitive to the number of observations. Therefore, the period covered in each country is extended to the year for which the latest data are available. The data for years prior to 1960 could not be included because the countries analyzed in this study did not pursue export promotion strategies before that time [Balassa (1981)]. 3Generally speaking, as developing countries become more advanced, the structure of their economies gradually shifts from primary industries (i.e., agriculture) to secondary industries (i.e., manufacturing) and finally to tertiary industries (i.e., service). Therefore, growth of manufacturing industries in the LDCs can be regarded as a proxy to measure their first stage of industrial development. Balassa argued that the development of the manufacturing sector is a 'part and parcel of overall economic development'. See Balassa (1981, essay 1).
P.C.Y. Chow, Export growth and industrial development
57
expansion of exports and development of manufacturing industries is necessary to evaluate various patterns of industrial development in the NICs. Even though the finding may be unique to the NICs with successful export performances, it may provide other LDCs with important insights in mapping out their future development strategies. In this study, the causality test is conducted by using the Sims technique [Sims (1972)]. The annual data on exports and manufacturing production from eight NICs were utilized for the decades of 1960's and 1970's. 4 The empirical findings indicate that for most of the NICs studied in this paper, there is a strong causal relationship between export growth and industrial development. A majority of these countries exhibited bidirectional causalities between the growth of exports and development of manufacturing industries. Therefore, in these countries export growth and industrial development are mutually beneficial and reinforce each other. This paper is divided into four sections. The discussion of causal relationship between exports and development of manufacturing industries in section 1 is followed by a brief explanation of econometric technique of the causality test in section 2. The results from empirical studies are summarized in section 3. The final section is devoted to conslusions and policy implications. 2. Econometric model of causality Christopher A. Sims has developed a practical technique of testing causality in a bivariate model [Sims (1972)]. According to Sims, one can regress Y on past and future values of X, and 'if causality runs from X to Y only, future values of X in the regression should have coefficients insignificantly different from zero, as a group' [Sims (1972, p. 545)]. Therefore, for the purpose of this study, the following linear equations with distributed lags are postulated: M F G = F(X, 3 past lags and 3 future lags of X),
(1)
M F G = f ( X , 3 past lags of X),
(2)
X = f ( M F G , 3 past lags and 3 future lags of MFG),
(3)
X = f ( M F G , 3 past lags of M F G ) ,
(4)
4The concept of causality here refers to the statistical analysis of lead and lag relationship between economic times series. Therefore, the results need to be qualified as to the possible sensitivity to the choice of the time period. For the Asian NICs under study, most of their export-led growth strategies started with the 1960's while other NICs in Latin America did not start until the 1970's. Due to the data constraint, our period of observation is more consistent with the export promotion policies undertaken in the Asian NICs than in other countries.
58
P.C.Y. Chow, Export growth and industrial development
where M F G stands for output of manufacturing industries and X stands for exports of manufactured goods. In order to test the hypothesis that coefficients for future values of independent variable are jointly equal to zero, F-statistics were calculated in the following way: F = (RSS2 - RSS~)/(dJ2 - dJO RSS,/dfi
where R S S I , R S S 2 a r e the residual sum of squares (1) and (2) and dfl, df2 are the degrees of freedom in (1) and (2), respectively. Eqs. (3) and (4) were treated the same way, except that in these equations M F G is the independent variable, and X the dependent variable. Since regression analysis on time-series data is very likely to exhibit autocorrelation among residuals, a pre-filtered treatment of all variables was conducted. Sims suggested a filter of (1-0.75L) 2 where L, L 2 are lag operators, s Essentially, variable Yt is replaced by Y* = Yt- 1.5 Yt- 1+ 0.5625 Yt- 2, so is variable X,. Therefore, two degrees of freedom are lost to accommodate the requirement of white-noise residuals. Another three and six degrees of freedom are lost in eqs. (2) and (1) [(4) and (3)], respectively, due to the arrangement of distributed lags. To test the causality between export growth and industrialization, a group of eight most successful export-oriented NICs was chosen. They are Argentina, Brazil, Hong Kong, Israel, Korea, Mexico, Singapore and Taiwan. The selection of these countries is largely based upon the availability of the needed data. 6 These countries are very good representatives of the exportoriented economies because their share of manufacturing exports in the world market accounted for 80Vo of total manufacturing exports from developing market economies in 1980. If we exclude the Organization of Petroleum Exporting Countries (OPEC) from the developing market economies, then their share of manufacturing exports rises to 83.14~o of total exports from all non-OPEC LDCs. In these countries, the share of manufacturing output in total Gross Domestic Product (GDP) ranged, respec5While there is no a priori reasoning for using this filter, Sims argued that 'this filter (10.75L) 2 approximately flattens the spectral density of most economic time-series' [Sims (1972, p. 545)]. Also, both Geweke and Hsiao concluded that results of causality tests are 'similar' [Geweke (1979)] and qualitative conclusions are not changed by using different filters [Hsiao (1979)]. 6A similar process of selecting NICs was used by Balassa. Using the criteria of per capita income and share of manufacturing output in GDP in 1978, Balassa called a group of ten most advanced developing countries Newly Industrializing Countries (NICs). They are Argentina, Brazil, Chile, Hong Kong, Israel, Korea, Mexico, Singapore, Taiwan and Uruguay. The statistical data on exports in Chile and Uruguay were not disaggregated under Standard International Trade Classification (SITC) until the mid-1960's. Therefore, Chile and Uruguay are excluded from our sample.
59
P.C.Y. Chow, Export growth and industrial development
tively, f r o m 24Yo in Israel a n d M e x i c o , to 3 3 ~ o - 3 4 ~ in A r g e n t i n a a n d T a i w a n in 1980. E a c h of these c o u n t r i e s has, at least, a p e r c a p i t a i n c o m e of U n i t e d States $1,500 in 1980.
3. Empirical evidence A m o n g the t h i r t y - t w o r e g r e s s i o n e q u a t i o n s t h a t were e s t i m a t e d , m o s t of t h e m h a v e relatively high R 2. T h e F - s t a t i s t i c s for the e s t i m a t e d e q u a t i o n s are significant, except for e q u a t i o n s for A r g e n t i n a . F o r the sake of brevity, o n l y the F - s t a t i s t i c s w h i c h i n d i c a t e the results of c a u s a l i t y tests in e a c h c o u n t r y are r e p o r t e d in t a b l e 1. 7 T h e c a u s a l i n f e r e n c e b e t w e e n e x p o r t s a n d o u t p u t s of m a n u f a c t u r e d g o o d s for each c o u n t r y is s u m m a r i z e d in the last c o l u m n of t a b l e 1.
Table 1 Sims' test of causality between export growth and development of manufacturing industries,a
Country
Export growth (X) on manufactured outputs (MFG)
Manufactured outputs (MFG) on export growth (X)
F-ratio
Results
F-ratio
Results
fail to
0.69 (6,5) 10.28c (6,5) 4.88¢ (6,7) 35.41 b (6,5) 23.90b (6,5) 2.91 (6,5) 17.21b (6,5) 24.28b (6,9)
fail to reject Ho reject H o
X~-MFG
reject H o
X~-MFG
reject H o
X~MFG
reject H o
X~MFG
fail to reject Ho reject H o
X--,MFG
reject H o
X~-MFG
Argentina 2.32 1960-80 (6,5) Brazil 10.88b 1960-80 (6,5) Hong Kong 9.34b 1961-83 (6,7) Israel 16.35b 1960-80 (6,5) Korea 68.00b 1960-80 (6,5) Mexico 12.05b 1960-80 (6,5) Singapore 21.07b 1960-80 (6,5) Taiwan 16.54b 1960-84 (6,9)
reject Ho reject Ho reject H o reject H o reject Ho reject H o
reject Ho reject Ho
Causal inference no causality
X~MFG
aH o is the null hypothesis that X(MFG) does not cause MFG(X). Figures in
parentheses stand for the degrees of freedom. bSignificant at the lYoolevel. cSignificant at the 5Yo level.
7For interested readers, detailed results are available from the author upon request.
60
P.C.Y Chow, Export growth and industrial development
From the results in table 1, one can conclude that by applying Sims' criteria, the growth of exports and development of manufacturing industries had bidirectional causalities in Brazil, Hong Kong, Israel, Korea, Singapore and Taiwan. In Mexico, the causality runs from exports to the development of manufacturing industries, implying thatexpansion of exports causes the development of manufacturing industries but not the other way around. 8 Thus, these findings empirically support what has been a priori argument for export expansion as a development strategy, or more precisely for emphasizing exports of manufactured goods in the NICs. The causal process is far less significant in either direction in Argentina, implying that the country can promote industrialization without relying on export growth. One must be more cautious, however, about this conclusion because the data in Argentina are less reliable due to her hyperinflation during the period under study. 9 The results of this study suggest that for most of the small open economies like Hong Kong, Israel, Korea, Singapore and Taiwan, the development of manufacturing industries and export growth have reciprocal causal relationships. To put it more concisely, they are mutually interdependent in the development process. This finding is rationalized by the observation that export growth in these countries can expand their limited domestic markets, and contribute to the economies of scale necessary for industrial developments. Furthermore, export growth integrates domestic economy with regional and/or global economies thereby expanding the dimension of competition to international markets. Competition promotes resources reallocations in developing countries as they transform from less productive farming sector to relatively more productive manufacturing sector [Feder (1982)]. Therefore, factor productivities are improved through export growth [Kavoussi (1984)]. The development of manufacturing industries in these countries makes their industries more competitive. This in turn enhances their comparative advantage, which further promotes the growth of exports and thereby of national income of these countries. Therefore, the development of manufacturing industries and expansion of exports could accompany and reinforce each other in the development process. These results are consistent with the earlier studies by Michaely (1977), Heller and Porter (1978) as well as Tyler (1981). SSims' test requires the evaluation of the absolute size as well as the patterns of the estimated lag coefficients. In all the regression equations, the coefficients of past lags are much more significant than those of future lags. An overall evaluation of the coefficients of past lags also indicates that the 'net' effect on the dependent variable is positive. Hence, export growth is positively causing the development of manufacturing industries and vice versa in those bidirectional cases. 9The Consumer Price Index which is based on 1970= 100 increased from 16 in 1960 to 16,075 in 1980 whereas the Wholesale Price Index which is based on 1970= 100 increased from 14.3 to 58,331 during the same period. Data were derived from Statistical Yearbook, United Nations, various issues.
P.C.Y. Chow, Export growth and industrial development
61
4. Conclusion
Contributions of export growth to development of manufacturing industries were subjected to a rigorous econometric test of causality in this study. Although the quantitative technique employed in this paper is far from foolproof, it is adequate for our purpose, t° The results support the hypothesis of reciprocal causality. To our knowledge, this is the first study to empirically establish the causal relationship between export growth and industrial development; all other studies had simply assumed a priori that such a relationship does exist. The validation of this causal relationship has far-reaching policy implications for developing countries. Except for Argentina, export growth has significantly influenced the process of industrial development as indicated by unidirectional causality in Mexico, or by bidirectional causality in Hong Kong, Israel, Korea, Singapore and Taiwan. Even in the case of Brazil, which has a relatively large domestic market in contrast to the above five small open economies, export growth had a bidirectional causal linkage with the development of manufacturing industries. The case of Mexico indicates that export expansion has accelerated industrialization, yet industrialization did not contribute to export growth. It can be concluded that depending on the size of the domestic market, export growth can cause industrialization, either unidirectionally, or bidirectionally by influencing the development of manufacturing industries. The results further confirm the advantage of export-led growth strategy for small open economies. Countries such as the Asian NICs can expand their limited domestic markets by exporting their manufacturing outputs to international markets, as the growth of exports of manufactured goods and industrial development have proven to be complementary. Even though this study has validated the contributions of export growth in selected NICs, their export performances may not be completely duplicated in other lesser developed countries. A simultaneous adoption of export-led growth strategy by a large number of developing countries as a group may be self-defeating as it can generate excessive competition among themselves in the world market. Detailed discussion of the applicability of the results of this study to other developing countries is beyond the scope of this paper. However, it is important for other LDCs, especially for those who are chasing the Asian NICs, 11 to consider both the supply-side and the demandside effects of export promotion strategies.
~°For detailed discussion of the econometric techniques of causality test, see, for example, Jacobs et al. (1979), Hsiao (1979) and Wu (1983). 1tSome developing countries in Southeast Asia are granted the status of 'quasi-NICs', and are pursuing to reach the full rank of NICs in the coming decades. See Miyohei Shinohara, 'Trade and Industrial Adjustments in the Asia-Pacific Region and Japan', in M. Dutta (1984, p. 51).
J.D.E.
C
P.C.Y. Chow, Export growth and industrial development
62
Appendix: Sources of data The data on exports of manufactured goods for each country were derived by subtractng non-ferrous metals (SITC 68) from the sum of SITC 5 to 8. Whenever it was necessary, the values of exports in the country were converted into U.S. dollars based on current exchange rates. All data except for Taiwan and Hong Kong were obtained from the Yearbook of International Trade Statistics, United Nations, various issues. Except for Hong Kong and Taiwan, the figures of manufacturing outputs were obtained from the Yearbook of National Account Statistics, United Nations, various issues. Manufacturing outputs are classified under category No. 3 in the table of 'Gross Domestic Product by Kind of Economic Activity'. Whenever it is necessary, real outputs were derived by adjusting the current values of outputs with implicit price deflator, or consumer price index, whichever was available. Data for Taiwan were derived from the Taiwan Statistical Data Book, 1984, CEPD, Executive Yuan, 1984. They are labeled as industrial exports instead of exports of manufactured goods. The data on exports in Hong Kong and Singapore included re-exports because of the entrep6t status in these two city-states. Data of manufactured products for Hong Kong in the 1960-65 period were estimated by multiplying the percentage share of MFG in GDP with the estimates of GDP and updated to 1983 using the supplement of the Key Indicators of Developing Member Countries of Asian Development Bank, Vol. XVI, April 1985.
References Balassa, Bela, 1977, Export incentives and export performance in developing countries: A comparative analysis, World Bank Staff Working Papers, no. 248 (World Bank, Washington,
DC). Balassa, Bela, 1978, Exports and economic growth: Further evidence, Journal of Development Economics 5, 181-189. Balassa, Bela, 1981, The newly industrializing countries (Pergamon Press, New York). Dutta, M., ed., 1984, Studies in the United States - Asia economic relations (Acros Press, Durham, NC). Feder, Gershon, 1982, On exports and economic growth, Journal of Development Economics 12, 59-75. Geweke, John 1979, Testing the exogeneity specification in the complete dynamic simultaneous equation model, Journal of Econometrics 7, 163-185. Granger, C.W.J., 1969, Investigating causal relations by econometric models and cross spectral methods, Econometrica 37, July, 428-438. Heller, Peter S. and Richard D. Porter, 1978, Exports and growth: An empirical re-investigation, Journal of Development Economics 5, 191-193. Hsiao, Cheng, 1979, Causality tests in econometrics, Journal of Economic Dynamics and Control 4, 321-346. Jacobs, R.L, E.E. Leamer and M.P. Ward, 1979, Difficulties with testing for causality, Economic Inquiry 17, July, 401-413. Kavoussi, Rastam M., 1984, Export expansion and economic growth: Further empirical evidence, Journal of Development Economics 14, 241-250. Kessing, Donald B., 1967, Outward-looking policies and economic development, Economic Journal, June, 303-320.
P.C.Y. Chow, Export growth and industrial development
63
Michaely, Michael, 1977, Exports and growth: An empirical investigation, Journal of Development Economics 4, 49-53. Sims, Christopher A., 1972, Money, income and causality, American Economic Review LXII, no. 4, 540-552. Tyler, William, 1981, Growth and export expansion in developing countries: Some empirical evidence, Journal of Development Economics 9, 121-130. Wu, De-Min, 1983, Tests of causality, predeterminedness and exogeneity, International Economic Review 24, no. 3, 547-588.