This action might not be possible to undo. Are you sure you want to continue?

# JOURNAL OF ECONOMIC DEVELOPMENT Volume 22, Number 2, December 1997

**Dynamics of Real Exports and Real Economic Growths in 13 Selected Asian Countries
**

Matiur Rahman* and Muhammad Mustafa**

1

II

This paper is an exploration of the dynamics between real exports and real economic growths in 13 selected Asian countries. It applies the cointegration and error-correction procedures. The unit root test reveals that both the time series are nonstationary in each country and individually they depict I(1) behavior. The evidence of cointegration and short-run as well as long-run Granger causality including the directions thereof vary from one country to another.

I. Introduction Controversies surround the uneasy causal nexus between exports and economic growth in the context of both developed and developing countries. The hypothesis of export-driven economic growth lacks economic consensus. As a result, the empirical studies on this issue are voluminous and expansive. Dollar (1992) examined the experience of 95 less developed countries (LDCs) for the period 1976-85 and found that outward-oriented countries do grow faster than more inward-oriented countries. Bahmani-Oskooee et al. (1991) provide results in support of export promotion hypothesis. Dutt and Ghosh (1994) also find exports and economic growth are cointegrated for a majority of their sample countries, but fail to test the question of direction of causality. Dodaro (1993) demonstrates that export growth and GDP growth display weak bidirectional causality. Levine and Renelt (1992), and Levine and Zervos (1993), using sensitive and extreme bounds analyses respectively, demonstrate that the relationship between economic growth and some of its determinants is fragile. Recently, Amoateng and Amoako-Adu (1996) find a causal linkage between economic growth and exports for 35 African countries. Numerous other studies (e.g., Syran and Walsh (1968), Michalopoulos and Jay (1973), Michaely (1977), Salvatore and Hatcher (1991), Van den Berg and Schmidt (1994), Balassa (1978, 1985), Tyler (1981), Feder (1983) and Kavoussi (1984)) investigated the association between the growths of exports and aggregate national output. With some qualifications, most of these studies found that there exists a positive association between growth of exports and GNP growth. Jung and Marshall (1985) tested for Granger causality using time series data on thirty-seven developing countries and found limited qualified support for export-led growth hypothesis. Chow (1987) found bidirectional causality between growth of exports and growth of industrial output in eight

** Professor of Finance, Department of Accounting, Finance, & Economics, McNeese State University, Lake Charles, LA 70609. ** Professor of Economics, Department of Agribusiness and Economics, South Carolina State University, Orangeburg, SC 29117.

81

Singapore. but the effect is not robust. South Korea. The cointegration methodology is appropriate to the nonstationary time series. his results cast some doubts on the significance of the positive contribution of the manufactured export sector to the growth process of low-income African countries. the most appropriate procedure is the simple Granger causality test. this undertaking is timely and quite useful to some Asian countries because of their enormous trade and growth potentials. The application of simple ordinary least squares (OLS) to the nonstationary time series is likely to lead to erroneous conclusions due to spurious correlation. Finally. Henriques and Sadorsky (1996) and Abhayaratne (1996) also found evidence of bidirectional causality and feedbacks between economic growth and export growth which may be termed as “virtuous circle”. Indonesia. the most appropriate procedures are cointegration and error-correction models. China.S. Provided the time series data are found stationary. Japan. Afxentiou and Serletis (1991) drew a general conclusion that industrial countries’ export promotion policies are not instrumental in the stimulation of GNP growth. exports and factor inputs (capital and labor) in Germany. Section III reports the results. market reforms and trade liberalizations. privatization. offers conclusions and states the policy implications. it is necessary to examine the stationary/nonstationarity property of time series data to determine the most appropriate econometric technique in order to avoid incorrect conclusions. This conclusion is in agreement with a similar one reached by Jung and Marshall (1985) with respect to developing countries. Italy. Germany and Japan experienced export-led growth. and Helpman and Krugman (1985) suggested the existence of bidirectional causality between growths of national output and exports. Pakistan. II. The remainder of the paper is organized as follows. the cointegration regression is specified as follows: 82 . U. and U. To begin with this examination. In light of the above controversies and inconclusive evidence on the direction of causality between economic growth and export growth. nor are GNP growth policies necessarily effective in fostering export growth. Sharma et al. In particular. India. Section II briefly outlines the empirical methodology. Ukpolo (1994) also supports the hypothesis of a positive linkage between the non-fuel primary exports and economic growth. To carry out the investigation.K. and the U. The thirteen countries considered in this paper include Bangladesh. Nepal. Hansen (1994) finds that exports have a positive effect on economic growth in G-7 countries. Section IV summarizes the results. Alam (1991). In the case of nonstationarity in the time series data. These countries have been selected because of their eminence and rapidly growing importance in international business. However.S. Kunst and Martin (1989).K. In the case of the U. (1991) investigated causal relationships between economic growth. For Italy. they could not find any causal relation between exports and output. reverse causality was found between exports and GNP growth. increased emphasis on export promotion as a vital tool for enhancing economic growth. over the period 1960-87. Pomponio (1996). According to their findings.JOURNAL OF ECONOMIC DEVELOPMENT newly industrialized countries. deregulations. Empirical Methodology At first. Sri Lanka. Thailand. the issue merits further investigations. Japan. Philippines and Malaysia. this paper applies the relatively new cointegration and error-correction procedures.. Bhagwati (1988).

and et is the stochastic error term. r < n. Here. The variables xt and yt are integrated of order d (i. unit root tests are to be conducted for which the following equations are considered: (2) (3) Each time series has non-zero mean and non-zero drift. The cointegration procedure. the following ADF regression is considered: (4) The ADF test is applied on to infer about the null hypothesis of no-cointegration.e. and .Dynamics of Real Exports and Real Economic Growths (1) where xt = log of real GDP. i.. Again.. I(d)) if the time series data on xt and yt have to be differenced d times to restore stationarity. for d = 1. Equation (5) is subject to the condition that is less than full rank matrix. provided in Engle and Yoo (1987). This procedure applies the 83 . it allows explicit hypotheses tests of parameter estimates and rank restrictions using likelihood ratio tests. The empirical exposition of Johansen and Juselius methodology is as follows: (5) where Vt denotes a vector of log of real GDP and log of real exports. For d = 0. The Engle-Granger (1987) cointegration procedures are not without drawbacks since they do not consider explicitly the error structure of the data processes. is the speed of adjustment matrix and is the cointegration matrix. yt = log of real exports. That is why the estimation should include both a constant and a trend term in each specification. 1992). For nonstationarity in each variable. xt and yt are stationary in levels and no differencing is needed. first differencing is needed to restore stationarity.e. as developed in Johansen (1988) and Johansen and Juselius (1990. Most importantly. The relevant null hypothesis is that or against the corresponding alternative hypothesis that or . The null hypothesis is rejected if the calculated pseudo t-value associated with is greater than its critical value. A failure to reject the null hypothesis would imply that each variable is nonstationary. avoid the above drawback by allowing interactions in the determination of the relevant economic variables and being independent of the choice of the endogenous variable. Next.

there will exist an error-correction representation (Engle and Granger (1987)). dollar. The same sample period could not be used for each country because of the data availability constraints. (1984)). If xt and yt are found cointegrated by either ADF procedure or Johansen-Juselius procedure or both. It is likely that these two procedures will provide contradictory evidence in some instances. Equations (5) and (6) should not include the error-correction term for the detection of Granger causality between two variables (Bahmani and Payesteh (1993)). Only the annual data have been used in this study because GDP data are usually available on yearly basis in most of the developing countries. preclude the short-run dynamics and Granger causality. If . if and is not zero. Again. A lack of cointegration does not. In these two equations. the series xt and yt are cointegrated when at least one of the coefficients or and . If ’s are not all zero.. In the absence of a long-run relationship. all the data have been collected in nominal terms and in U.S. If two variables are cointegrated.JOURNAL OF ECONOMIC DEVELOPMENT maximum eigenvalue test ( max) and the trace test ( trace) for null hypotheses on r. Its importance lies in its ability to combine short-run dynamics and long-run relationship in a unified system. however. 84 . Again. All the data have been collected from the various monthly and annual issues of International Financial Statistics. and Engle and Granger (1987) on cointegration. 1988). Sometimes. published by the International Monetary Fund (IMF). Initially. Their real magnitudes have been obtained by deflating the nominal values with the respective consumer price indices/GDP deflators depending upon the data availability. max test is expected to offer a more reliable inference as compared to trace test (Johansen and Juselius (1990)). then xt will lead yt in the long run. the long-run Granger causality will stem at least from one direction. If ’s are not all zero. the Johansen and Juselius test procedure suffers from its supersensitivity to the selection of the lag structures. As a result. movements in yt will lead those in xt in the short run. Davidson et al. this study pursues both the ADF and Johansen-Juselius procedures for cointegration. they yt will lead xt in the long run. It has enjoyed a revival in popularity due to the recent work of Granger (1986. Annual data are employed in this study. Hendry et al. The error-correction model (ECM) was first introduced by Sargan (1964) and subsequently popularized by numerous papers (i. (1978). The error-correction model may take the following form: (6) (7) The reverse specification is considered due to plausible bidirectional causality. Of these two tests. it is desirable to exclude the insignificant lags to improve the efficiency of OLS estimates of parameters (Baghestani and Mott (1997)). movements in xt will lead movements in yt in the short run.e.

15000 (6) JAPAN (1965-1994) -2.00000 (2) THAILAND -2.79200 (2) -1.91120 (3) -0.20000 (2) -0.47500 (5) -0.95016 (4) -3.608 0(4) -2.39290 (4) -3.81800 (4) -1.07558 (3) -5.73430 (4) -14.24800 (4) -1.58900 (4) -2.21670 (4) -10.54570 (2) -2.65090 (3) -5.99420 (2) -3.18150 (4) 0.57820 (4) -5.43090 (4) -3.5041 (4) SINGAPORE (1972-1993) -1.06690 (4) -3.59077 (4) -3.54080 (4) -3.15000 (4) -1.30300 (4) -4.57620 (4) -6.35100 (4) SRI LANKA (1965-1991) -1.64580 (4) -3.78000 (2) INDIA (1965-1994) -1.71000 (4) 2.59300 (3) -3.37800 (2) -2.52100 (1) -1.28753 (0) -4.60600 (4) -5.55400 (1) -0.17156 (4) -3.16700 (4) -1.05500 (4) -2.17080 (4) -4.55536 (4) -4.86914 (4) -3.21000 (2) INDONESIA (1968-1992) -2.51000 (2) -1.51200 (4) -2.01210 (4) NEPAL (1975-1994) -3.93400 (4) -3.16140 (4) -3.22400 (3) -2.22400 (4) -2.64120 (4) 1.58000 (2) -3.26130 (4) -3.95490 (4) 1.50800 (4) -0.98780 (4) -4.93500 (4) 0.Dynamics of Real Exports and Real Economic Growths III.75211 (3) PHILIPPINES (1965-1994) -1.48370 (4) -4.92750 (4) -2.94300 (2) -0.71750 (4) -2.69230 (2) -4.47500 (4) -3.32100 (2) -1.57000 (4) -3.5 00 (5) -0.43580 (4) -3.45070 (4) -3.69640 (4) -4.33200 (4) SOUTH KOREA (1966-1994) -2.6520 (2) -0.19200 (4) -3.58000 (2) -2.87990 (4) 183.2101 (3) -2.60100 (2) CHINA (1978-1994) -2.06430 (4) -1.15480 (4) -3.01800 (4) -3.54609 (1) 0.90180 (4) -4.84140 (4) -0.03100 (2) -0.24200 (4) -3.36110 (4) -3. Empirical Results The unit root test results are reported as follows: Table 1 ADF Test of Unit Root: Equations (2) and (3) Variable RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX ADF (with Trend) ADF (without Trend) Level Difference Level Difference BANGLADESH (1971-1994) -0.92451 (4) -4.7312 (2) -1.11956 (4) -4.25000 (4) -1.19600 (4) 85 .0965 (4) -4.09730 (2) -0.47100 (5) PAKISTAN (1965-1994) -1.

277 0.274 (3) NEPAL -0.832 0. each time series on log of real GDP and log of real exports is nonstationary both with trend and without trend at 5 percent and higher levels of significance.JOURNAL OF ECONOMIC DEVELOPMENT Table 1 (Continued) Variable RGP REX ADF (with Trend) ADF (without Trend) Level Difference Level Difference MALAYSIA (1965-1994) -2. Table 1 also demonstrates clearly that each time series becomes nonstationary after it is differenced once only. $).938 1.732 (4) SINGAPORE -1.521 (1) INDONESIA -1.050 0.02610 (2) -0. The optimum lag-lengths. $).868 (1) JAPAN -1.390 0. are determined by the final prediction error (FPE) criterion (Hslao (1981)).270 0. At first.850 2.41270 (4) RGP = log of real GDP (U.910 1. the next logical step is to determine if both series are cointegrated.S.428 (2) SOUTH KOREA -1. as reported in parentheses.920 1.367 (1) INDIA -3.918 Adj R2 0.001 1.055 (5) PAKISTAN -3.220 1.735 1.170 (1) SRI LANKA -2. The cointegration results based on this procedure are reported as follows: Table 2 Cointegration Tests Based on ADF Procedure: Equation (4) Dependent (Xt) RGP RGP RGP RGP RGP RGP RGP RGP RGP RGP RGP Independent (Yt) REX REX REX REX REX REX REX REX REX REX REX ADF Statistics BANGLADESH -3. and REX = log of real exports (U.350 0.800 1. Critical ADF values at 5% level of significance are -3.390 0.8600 (without trend).230 0.197 (3) CHINA -5.91160 (4) -4.050 0.410 (with trend) and -2.689 (2) THAILAND -3.695 (5) DW 1.67560 (4) -3.87900 (4) 1. Since each time series is nonstationary.890 2.13510 (4) -3.02920 (4) -8.850 1. Evidently.250 86 .77200 (2) -1.250 0.S. the ADF procedure is applied.

2878 r 0 14.944 NEPAL r 1 13. log of real GDP and log of real exports are found cointegrated at 10 percent and higher levels of significance in Bangladesh.8841 (5) 18.222 (4) MALAYSIA -3. are determined by the FPE criterion for each country. -3. Likewise. REX) (RGP. are -4.3727 (9)* 13. The DW-values are reported to indicate white noise and the numerical values of s are reported to indicate the relative strength of long-run association between the variables.9587 (5) 87 .1230 r 0 24.480 0. India. Thailand. REX) Null Hypothesis Max BANGLADESH r 1 12. REX) (RGP. REX) (RGP. as reported in Engle and Yoo (1987). The existence of cointegration implies a long-run equilibrium relation between log of real GDP and log of real exports in the above Asian countries. Based upon a comparison of the calculated values of the ADF statistic with its above critical values.07.123 (5)* 27. REX) (RGP.4283 (9)** 55.0135 r 0 33. no evidence of cointegration is found for Japan.061 (5) DW 1.5737 INDIA r 1 13. as reported in parentheses.4283 r 0 36.340 The critical values of ADF statistics.6423 (5) 10. To overcome any confusions about the cointegrating relation. the optimum lag-lengths.835 Trace (2)** (2) (5) (5) (5) (5)* (9)* (9)* (5)* (5) 12.0135 (5) 33.86170(2) 13. The absence of cointegration implies that log of real GDP and log of real exports move being independent of each other in the long run without any tendency to converge. and -3.8707 SRI LANKA r 1 18.03 at the 1. The results are as follows: Table 3 Johansen’s Procedure of Cointegration Test: Equation (5) Data Vector (RGP. In contrast.860 Adj R2 0.904 1. South Korea and Sri Lanka. The optimum lag-lengths are reported within parentheses and they are determined by the final prediction error (FPE) criterion.37.Dynamics of Real Exports and Real Economic Growths Table 2 (Continued) Dependent (Xt) RGP RGP Independent (Yt) REX REX ADF Statistics PHILIPPINES -4. Pakistan.2878 (2)*** 16. the Johansen-Juselius Procedure is applied next.8955 PAKISTAN r 1 10. Indonesia. China.7468 (5) 11. Nepal.7468 r 0 17. 5 and 10 percent levels of significance respectively. Philippines and Malaysia.

5092 (5)* 34.3994 (7) SINGAPORE 10. based upon the max test in the cases of Pakistan.5092 (5)* 33. REX) (RGP.880 (6) MALAYSIA 10. Nepal and South Korea.2244 (2)* 25.9202 (4) 10.0334 (4) 32.5491 (3)* 12.896 (2)*** 10.4959 (5) SOUTH KOREA 19.0382 (4)* PHILIPPINES 10. China. Since neither approach is perfect and the evidence is mixed on cointegration between the two variables for several countries. they are determined by the FPE criterion.7159 (7) 10. Likewise. REX) (RGP. REX) Null Hypothesis JAPAN r 1 r 0 CHINA r 1 r 0 r 1 r 0 r 1 r 0 r 1 r 0 r 1 r 0 r 1 r 0 r 1 r 0 10. the null hypothesis of r 0 is rejected at the 95 percent and 90 percent confidence levels respectively. REX) (RGP. REX) (RGP. Indonesia.22440(2)* 25.59075(2) 10. no references have been made to trace test because max is more reliable. Philippines and Malaysia. Singapore.3741 (2) 16. Indonesia. the error-correction models (6) and (7) are estimated for each country. REX) (RGP.5490 (3)*** 12.5220 (2)* THAILAND 10. as stated earlier. South Korea. Nepal. REX) (RGP. The null hypothesis of r 0 (signifying no-cointegrating relation between variables) is rejected at 99% confidence level. These findings contradict those from the ADF procedure for Japan. Thailand. Sri Lanka.3165 (7)** * 32. In other cases. In Bangladesh and Japan.JOURNAL OF ECONOMIC DEVELOPMENT Table 3 (Continued) Data Vector (RGP.3165 (7)** * 41. The most noteworthy exception is India here which shows no-cointegrating relation between the variables based upon both max and trace tests.36632(2) INDONESIA 10. Sri Lanka. REX) (RGP.1982 (4) 11.2509 (3) Max Trace *** significant at the 99% confidence level *** significant at the 95% confidence level *** significant at the 90% confidence level The optimum lag-lengths are reported in parentheses.005 (5) 19.2780 (4)* 18.3708 (6) 10.278 (4)** 19.7018 (3) 11. The results are reported as follows: 88 .4898 (6)* 20.0334 (4) 32.4898 (6)* 19.0717 (4)* 10.3741 (2) 16.

17)=1.1874*** NEPAL m=4 (2.177 -0.751 t-statistics of the coefficient of et-1 -2.18)=4.172*** m=3 (2.9)=7. n=3.47 0.4125 0.005 -2.939 -3. n=3.020* JAPAN m=3 (2.3435 0.109** -2.6564*** SINGAPORE m=4 (5. n=5. n=4.818* -0.238** -0.60 0.8)=3. n=3.17)=1.586 -2.552** -0.19)=1.926 CHINA m=1 (2.437** -0.9048 0. n=2.722* PHILIPPINES m=5 (3. n=2.085* SOUTH KOREA m=4 (3.516** -3.122 0.716** m=5 (5.65** m=1 (4.12)=7.660 0.42 0.17)=6.17)=11. n=5.003** -1.736 BANGLADESH m=4 (4.440 0. n=5.17)=3.006 -2. n=5.17)=2.315 -0.477 -1. n=5.18)=17.23)=0.8672 m=4 (3.321 0. F-Statisticsb 0.355* INDONESIA m=3 (2. n=5. n=5.100* m=3 (5. n=1.750 0.4023* 89 .1614* m=4 (3.676*** -0.581 0.3185 0.44 0.18)=1.5178* THAILAND m=5 (4.064 -2.869*** -2.291** -3.101 0.101** -0.902 m=2 (3.11)=2. n=5. n=3. n=5.694** m=5 (3.808 0.632 0.4518* INDIA m=4 (2.492*** m=2 (3.15)=5.477* SRI LANKA m=1 (2.7912** PAKISTAN m=5 (3.20)=2. n=1. n=3.7475 0.90 m=3 (5.18)=4.63 0. n=3.18)=3.10)=3.643 -1.2358 m=5 (2. n=2.2512 0.Dynamics of Real Exports and Real Economic Growths Table 4 Granger-Causality Test: Equations (6) and (7) Dependent Variable RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX “Causal” Variable REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP Lag Ordersa n=5.74* m=5 (4.14)=6. n=2.198 -2.6)=6. n=4.18)=2.17)=2.58 0.331 -0.011* -4.507 0.7499 0.

Indonesia (at 99% confidence level). I(1) Ho: cannot be Pakistan rejected. South Korea. the relative strength of causality between the variables varies from one country to another as reflected through the numerical values of s. I(1) Ho: cannot be India rejected.76** n=3.487** MALAYSIA n=2. The F-statistics (with degrees of freedom in parentheses) tests the joint null hypothesis that all coefficients of the “causal variable” are simultaneously equal to zero. *** significant at 90% confidence level. After excluding the highly insignificant lagged differences. and Nepal (at 99% confidence level). I(1) 5 Summary of The Empirical Results Ho: No-Cointegration ADF Cointegration Cointegration Cointegration Johansen-Juselius Cointegration No-cointegration Cointegration Causality Long-Run REX RGP Short-Run REX RGP RGP REX RGP REX RGP REX RGP REX 90 . b. IV.593** Notes: a. Pakistan (at 99% confidence level). joint F-test is conducted throughout.365 0. Singapore and Malaysia. m = lag length of dependent variable.JOURNAL OF ECONOMIC DEVELOPMENT Table 4 (Continued) Dependent Variable RGP REX “Causal” Variable REX RGP Lag Ordersa F-Statisticsb 0. m=3 (3. *** significant at 95% confidence level. The error-correction models (6) and (7) are estimated by using OLS for each country.575* -2. There are evidence that real GDP growth is causally prior to real export growth in India (at 95% confidence level).20)=3. *** significant at 99% confidence level. short-run bidirectional causality is detected on the basis of joint F-test without any indication of long-run causality between the variables. For Granger causality. There are evidence of bidirectional Granger causality between real export growth and real GDP growth in Pakistan. The results reveal that real export growth is predominantly causally prior to real GDP growth in Bangladesh (at 90% confidence level) and Japan (at 90% confidence level). Summary. m=3 (3. the ECMs are then re-estimated. In the case of Singapore.20)=3. China. However. Conclusions and Policy Implications The empirical results from the preceding section are summarized as follows: Table Ho: Unit Root Country and Order of Integration Ho: cannot be Bangladesh rejected. The significance of a long-run relation is usually determined by the t-value associated with the respective numerical coefficient of the error-correction term.35 t-statistics of the coefficient of et-1 -2. Sri Lanka (at 90% confidence level). n = lag length of “causal variable”. Lag orders are selected based on the FPE criterion.

these countries are in an economic growth mode caused by export promotion. Furthermore. In other words. Only Japan has higher real economic growth due to higher real export growth in the short run as well as in the long run. I(1) Singapore Ho: cannot be rejected. Table 5 reveals that China. (↔) indicates bidirectional causality. I(1) Thailand Ho: cannot be rejected. I(1) Indonesia Ho: cannot be rejected. (→) indicates unidirectional causality. The general policy implications of the above empirical findings are as follows: (i) countries with short-run and long-run virtuous circles should emphasize both higher economic growth and export promotion policies. I(1) South Korea Ho: cannot be rejected. Bangladesh. Thailand and Philippines have evidence of bidirectional causality in the short-run and long-run unidirectional causality from real export growth to real GDP growth. Pakistan and Nepal enjoy bidirectional causality between the two variables in the short run while the long-run unidirectional causality flows from real GDP growth to real export growth. South Korea and Malaysia experience bidirectional causality between real GDP growth and real export growth in both short run and long run.Dynamics of Real Exports and Real Economic Growths Ho: Unit Root Country and Order of Integration Sri Lanka Ho: cannot be rejected. As a result. I(1) Ho: cannot be China rejected. I(1) Ho: cannot be Nepal rejected. I(1) Philippines Ho: cannot be rejected. Also. I(1) Ho: cannot be Japan rejected. Somewhat differently. India. I(1) Table 5 (Continued) Ho: No-Cointegration ADF No-cointegration No-cointegration No-cointegration Cointegration No-cointegration Cointegration No-cointegration No-cointegration Cointegration Cointegration Johansen-Juselius Cointegration Cointegration Cointegration Cointegration Cointegration Cointegration Cointegration Cointegration Cointegration Cointegration Causality Short-Run RGP REX RGP REX REX RGP RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX RGP REX Long-Run RGP REX RGP REX REX RGP RGP REX RGP REX REX RGP RGP REX No Long-Run Causality REX RGP RGP REX where. the virtuous circle hypothesis is satisfied in these countries and hence they enjoy spectacularly high real economic growths. (ii) countries with short-run virtuous circle and long-run unidirectional causality from real export growth to real economic growth should accord more importance to export promotion in conjunction with pro-growth policy. Sri Lanka and Indonesia experience unidirectional causality from real GDP growth to real export growth in both the short run and long run. Surprisingly. (iii) countries with short-run virtuous 91 . Singapore reveals short-run bidirection causality between the variables without any evidence of long-run causality at all. I(1) Malaysia Ho: cannot be rejected.

(iv) countries with short-run and long-run unidirectional causality from real GDP growth to real export growth should accord more emphasis to higher economic growth policy to spur exports.JOURNAL OF ECONOMIC DEVELOPMENT circle and long-run causality from real economic growth to real export growth should attach added emphasis to higher economic growth policy pari passu with the continuation of export promotion. every individual time series is. fairly long to yield meaningful cointegration results. thus. 92 . In our opinion. we have a sample period of at least 20 years for each country. In closing. and (v) countries experiencing unidirectional causality from real export growth to real economic growth should continue the ongoing export promotion policy. The above policy prescriptions can be briefly categorized as “leaning with the wind”.

1991. 181-89. “Cointegration and Error Correction: Representation. Policy Choices and Economic Growth in Developing Countries after the 1973 Oil Shock. “Outward-Oriented Developing Economies Really Do Grow More Rapidly: Evidence from 95 LDCs.S. “Exports and Growth: A Reconsideration of Causality. “Econometric Modelling of the Aggregate Time Series Relationship between Consumers Expenditure and Income in the United Kingdom. “An Empirical Investigation of the Export Growth-Economic Growth Relationship. and S. H. “Exports. Engle. Payesteh. M. J. F.F.” Econometrica 55. Estimation and Testing.” Journal of Development Economics 36. Protectionism. Amoateng. Export and External Debt Causality: The Case of African Countries. Dodaro. Bhagwati. 55-63. “Trade Orientation and Macroeconomic Performance in LDCs: An Empirical Study. Serletis. MIT Press. 143-59. Dutt.” Economic Development and Cultural Change 39.” Journal of Development Economics 26.J. Davidson.W. P. 227-44. Alam.C. 1985. 21-27. Granger. 1960-1992.W. 1987. 839-48. 661-77. Afxentiou. Bahmani.” Journal of Macroeconomics 15. 1988. “Developments in the Study of Cointegrated Economic Variables. August 1986.” Oxford Bulletin of Economics and Statistics 48. ______. Yeo. 23-35. Feder. C. December 1978. “Exports and GNP Causality in the Industrial Countries: 1950-1985. 213-27. and A. 1997. C. “Budget Deficits and the Value of the Dollar: An Application of Cointegration and Error-Correction Modeling. J. 1996. 251-76. “Some Recent Developments in a Concept of Causality. 199-211. S. O. S..D. 167-79.” Journal of Development Economics 12..Dynamics of Real Exports and Real Economic Growths References Abhayaratne. 1991. 93 . 661-92. 405-15. Chow. 1976-1985. H. R.. and B. “On Exports and Economic Growth.. 59-73. “Foreign Trade and Economic Growth: Evidence From Sri Lanka.E. Baghestani. “A Cointegration Analysis of the U.. Engle.” Journal of Econometrics 38.. D. R. Yoo... ______. 1978.J. “Causality Between Export Growth and Industrial Development. Balassa. “Economic Growth.” Kyklos 44. 1991.F. 44-48. 1992. and B.” Journal of Macroeconomics 19.S. Amoako-Adu. 1983. “Exports. and S. K. 1996. Ghosh. Cambridge.S.” Journal of Development Economics 5. Ghiath.” Journal of Econometrics 35. 523-44. Dollar. and D..” Journal of Developing Areas 27. 1994. D. Mott. Granger.H.” Applied Economics 28. and T. 269-83. Massachusetts... 1987. Bahmani-Oskooee..P.Y. Srba. and S. “Exports and Economic Growth: Further Evidence. G. “Forecasting and Testing in Cointegrated Systems.P. 567-70... and C. B. Money Supply Process. Mohtadi. Growth and Causality in LDCs: A Re-Examination.” Economic Development and Cultural Change 40. Hendry. M.” Applied Economics Letters 1. 1993. 1987. September/October 1988.” Economic Journal 88....” Applied Economics Letters 3. 1993.S.” Journal of Development Economics 18. A.

F.F. Kavoussi.M. Pagan.J..” American Economic Review 82.R. D. R.. 1-12. 168-76. Syran. Michalopoulos. Jay. and P. Intriligator. and D. 1993. and T. 7-25.. C.. Helpman. 1991. 49-53. London: Butterworths.M. D.. “Exports and Economic Growth in Industrialized Countries. 1996.” in Handbook of econometrics II. E. Norris.. US Agency of International Development AID Discussion Paper. 94 . “Autoregressive Modeling and Money-Income Causality Direction. Hsiao. and B. “A Causality Analysis of Growth and Export Performance.” Oxford Bulletin of Statistics. Levine. edited by Z.. and D.J. J. R. Jung.” Journal of Development Economics 4. “Statistical Analysis of Cointegration Vectors.” Journal of Monetary Economics. “Investment Data and the Empirical Relationship Between Exporters.” Review of Economics and Statistics 71. Hendry. A. Sharma. D.” Journal of Development Economics.. 1996.. and P.. “Exports and Growth: An Empirical Investigation. Martin. I. “Testing Structural Hypotheses in Multivariate Cointegration Analysis of the PPP and the VIP for U. 1985.D. 426-30. “Wages and Prices in the UK: A Study in Econometric Methodology.” Journal of Economic Dynamics and Control. and D. ______ and K. 231-54. Amsterdam: North-Holland.. 1968.. Michaely. 28. Henriques. 540-55. 541-45. Sadorsky. “What Have We Learned about Policy and Growth from Cross-Country Regressions?” American Economic Review 83. 14. ______. Growth and Causality in Developing Countries. “Export-Led Growth or Growth-Driven Exports? The Canadian Case.J.” Washington. 211-44. “Dynamic Specification. 942-63. P. 85-106. 1977. Sargan.. Salvatore. MIT Press.. 1973. 699-703. Renelt.” Canadian Journal of Economics XXIX. S. Johansen. May 1992. R.W. “Growth of Exports and Income in the Developing World: A Neoclassical View. 1991. May 1990.K. Krugman.” Applied Economic Letters 1. January 1981.C. S. 107-10. Levine. Walsh. Juselius. Government and Economic Growth. R. June-September 1988. Sargan. “The Relation of Exports and Economic Growth: A Note. 697-708. “On Export and Productivity: A Causal Analysis. 1992.C.” Atlantic Economic Journal 24. and K. R. Kunst.” Journal of Econometrics.” Journal of Development Studies 27. and J.” Journal of Development Economics 18. Cheung. M. X.” Applied Economics 23. Pomponio.” in Econometric analysis for national economic planning. “Inward Oriented and Outward Oriented Trade Strategies. Hatcher. and S..JOURNAL OF ECONOMIC DEVELOPMENT Hansen. Massachusetts.M. Zervos. “Maximum Likelihood Estimation and Inference on Cointegration with Applications to the Demand for Money. 1989. M. 241-50.. Hart et al. 1985. 1964... Cambridge. Griliches and M.Z.D. 1984.. 1984.. and P. 1994. edited by P.. Marshall. No. “Exports. “Export Expansion and Economic Growth. “A Sensitivity Analysis of Cross-Country Growth Regressions. Market Structure and Foreign Trade.” Kyklos 21. 169-210. C. S.

249-68.Dynamics of Real Exports and Real Economic Growths Tyler. 1981. “Foreign Trade and Economic Growth: Time Series Evidence from Latin America. 1994. 445-49. Schmidt. 121-30. 95 . Van den Berg. 1994.. Ukpolo.” Applied Economics 26. H.” Journal of Development Economics 9.R.” Journal of International Trade and Economic Development 3... and J. W. “Export Composition and Growth of Selected Low-Income African Countries: Evidence from Time Series Data. “Growth and Export Expansion in Developing Countries: Some Empirical Evidence. V.