A time series analysis of the role of imports in India's phenomenal economic growth.

Saunders, Peter J.

Abstract

This paper investigates the role of imports in India's impressive economic growth. Annual data ranging from 1970 to 2005 are used to investigate the impact of imports on economic growth that is measured by the real GDP. Johansen's (1988) cointegration tests indicate that these two variables are cointegrated. Therefore, a long run relationship between imports and the real GDP exists. A vector error correction (VEC) testing framework is used in a further data analysis. VEC tests indicate that imports have impacted positively India's economic growth in the short run. Therefore, imports may have played an important role in India's recent economic growth.

I. INTRODUCTION

India's economy has experienced phenomenal economic growth in recent years. It has become one of the world's fastest growing economies. (1) International trade has undoubtedly played a key role in India's economic growth. India's international trade liberalization policies that started in the early 1990s have led to an unprecedented surge in both exports and imports. Exports increased 320 percent from 1995 to 2005, while during the same time period imports experienced an astonishing 438 percent increase! Throughout the 1970-2005 period, large and ever expanding trade deficits were prevalent in India's economy. (2) Yet throughout the same time span, India's output continued to grow. The traditional view of the relative importance of exports and imports in economic growth cannot provide a satisfactory explanation for India's impressive economic growth. Traditional economic theory asserts that exports impact economic growth positively while imports affect it negatively. This conclusion is self-evident from the treatment of exports and imports in conventional GDP accounting practices. The conventional GDP accounting approach measures the monetary value of all goods and services produced in an economy during a specific period of time. Items that are purchased abroad, i.e., all imports, are subtracted from the overall monetary value of the GDP. All domestically produced goods, including those exported abroad, add to the overall GDP monetary value. Therefore, exports are considered a positive item in the GDP account, while imports reduce its numerical amount. Given this traditional treatment of exports and imports, trade deficits arising from a growing excess of imports over exports are assumed to impact economic growth adversely.

Given the above short discussion of the respective roles of exports and imports in the traditional view of economic growth, it is clear that this approach cannot explain satisfactorily India's recent impressive economic growth. In fact, it appears that the economic growth in India may well have been partly due to an unprecedented increase in India's imports, rather than to a substantially smaller increase in its exports. This assertion has a theoretical basis. Several hypotheses are readily available to explain how imports can impact economic growth positively. First, the basic concept of comparative advantage can explain imports' positive impact on economic growth. International trade that relies on importing those goods that can be produced at lower costs abroad and exporting those goods that are produced more efficiently at home creates wealth in both trading countries. This wealth creation enables more investment to occur in all trading countries. Investment is the key element of economic growth. Therefore, through this channel, imports can cause economic growth in the importing country. Second, import competition can improve the productivity and the efficiency of domestic industries and businesses that face foreign competition. It can also lead to innovation in the importing country. Domestic businesses that operate more efficiently due to the foreign competition produce higher output. In this way, higher domestic economic growth is achieved. Additionally, the financial consequences of trade deficits may have a positive impact on economic growth. Trade deficits can be financed by a sale of various financial instruments or real assets to foreigners. The proceeds of these maturing financial instruments or real assets will provide them with cash that will ultimately be used to buy domestically produced goods. These purchases can also lead to domestic economic growth.

The role of imports and trade in economic growth has been investigated extensively in economic literature. The initial research in the area of trade and economic growth was undertaken by Little, Scitovsky and Scott (1970) and followed by Balassa et al. (1971). Nelson and Winter's (1974) research supports the hypothesis of imports leading to economic growth. Economic growth occurs because imports often lead to innovation and the creation of competition. At the same time, import competition may affect negatively the profitable export sector. Lucas (1988) outlined yet another channel through which international trade can impact economic growth. According to his hypothesis, sectoral growth is due to the learning-by-doing that takes place in the export and the import sectors of a domestic economy. Since the export sectors typically enjoy comparative advantage, they are likely to grow faster than the importing sectors that have no such advantage over their foreign competitors. Hence, according to Lucas, imports are likely to harm economic growth. Grossman and Helpman (1992) reached a very different conclusion about the impact of imports on economic growth. They suggested that imports can lead to improved access to a larger number of intermediaries. The ability to utilize these intermediaries enhances domestic industries' productivity and, thereby, spurs economic growth. Similar conclusions about the effects of imports on economic growth were reached by Lawrence (1999). According to Lawrence, import competition led to total productivity growth in the U. S. Rodrik (1999) suggested that imports of inputs may promote economic growth. Imports can lead to innovation due to their promoting competition. Lawrence and Weinstein (1999) also noted this beneficial impact of imports on economic growth in their study of the Japanese economy during the 1964-1973 time period. According to the two authors, imports are important conduits for economic growth more because of their contribution to competition than to the production of intermediate goods. Awokuse (2007) examined the impact of trade on three transitional economies. Trade led to economic growth in these three economies.

Numerous cross-sectional studies have analyzed the impact of imports on economic growth. Among others, earlier research by Barro and Sala-i-Martin (1992), Dollar (1992), and Edwards (1992) lent some empirical support to the results reported by Lawrence (1999). These cross-sectional studies indicated that those countries that followed protectionist policies experienced slower economic growth. However, most of the latter studies analyzed the differences between the inward and the outward orientation of the countries under empirical investigation. Although these studies analyzed the impact of trade on economic growth, they give no indication of how imports alone affect economic growth.

Empirical evidence on the impact of imports on economic growth is mixed and inconclusive. Some of the above reviewed studies suggest that imports may impact economic growth positively, while others reach a very different conclusion on the role of imports in economic growth. Given this fact, a further empirical investigation of the relationship between imports and growth can provide valuable information on this yet unresolved, important economic issue. The objective of this paper is to provide such evidence by analyzing the role of imports in India's successful economic growth. The study of India's economy is particularly important and relevant because of the predominance of imports in that country's international trade. The focus of the present research is on investigating only the relationship between imports and economic growth and on analyzing the causal flows between these two variables. This objective can be accomplished most effectively within a reduced form modeling of India's economic data. (3) Therefore, a vector error correction (VEC) reduced form estimation is adopted to analyze the causal impact of imports on India's economic growth. The remainder of the present paper is organized in the following way. Its first section outlines the data used in the present research and the initial test results of these time-series data. These tests are comprised of unit root and cointegration testing of the data. VEC analyses of the data are undertaken thereafter. VEC test results are reported in the following section of this paper. Overall conclusions on the role imports in India's economic growth are reached thereafter.

II. UNIT ROOT AND COINTEGRATION TESTS

As stated above, the objective of this paper is to investigate the role of imports in India's economic growth. The best measure of India's economic growth is its real gross domestic product (RGDP), while total imports (IMP) are the most appropriate measure of all imports. Therefore, annual data on RGDP and IMP ranging from 1970 to 2005 are used to investigate the relationship between imports and growth in India. (4) Economic analysis of this time period can provide meaningful information on the impact of imports on India's economic growth as its growth accelerated during these years.

Several preliminary statistical steps must be undertaken in time-series analyses of any relationship under investigation, such as India's imports and real GDP. These initial steps include unit root and cointegration testing of the data. Their objective is to determine the subsequent test structure, such as a VAR or a VEC testing framework. Time-series data typically exhibit a trend over time. Such data are nonstationary as they contain unit roots. Time-series data that contain unit roots are not suitable for any subsequent econometric modeling. (5) Therefore, the first step in investigating the relationship between imports and India's real GDP must be to test for the presence of unit roots in each of the two time-series. This objective can be accomplished by deploying the augmented Dickey-Fuller [Fuller (1976); Dickey and Fuller (1979)] (ADF) test. The ADF test determines the degree of integration of each individual time-series under investigation and, thereby, the appropriate subsequent data estimation methods. The ADF test results on RGDP and IMP data are summarized in Table 1 below.

The ADF test specifications included the intercept and the deterministic time trend variable in all test cases. The SIC criterion was used in the lag selection of the two test variables. Test results reported in the above Table 1 indicate that beth test variables, RGDP and IMP, are nonstationary in their levels. As such, these variables contain unit roots. Further ADF analyses of the RGDP and IMP data indicate that the first differences of levels are stationary. Therefore, the two data series are integrated of order one, I(1). Given the fact that the RGDP and IMP data are I(1), it is possible to assume that there exists a long run relationship between real GDP and imports in India. Such a possibility can be investigated by deploying cointegration testing of the two time-series. Cointegration is an appropriate testing framework for all time-series data that are integrated of order one, I(1). (6) Cointegration means that even though individual time series are nonstationary, one or more linear combinations of these time-series is stationary. (7) In such a case, a long run relationship exists among such time-series. Consequently, cointegration tests can provide key information on the role of imports in India's long run economic growth.

Cointegration tests are numerous and well known in the economic literature. These tests include the Engle and Granger (1987) estimation, the Stock and Watson (1988) test, and the Johansen (1988) test, among others. A detailed outline of these cointegration tests would be redundant. However, it is worth noting that even though the common objective of these tests is to find the most stationary linear combination of the vector time-series under empirical examination, there are some important statistical differences among these tests. This fact is noted by Gonzalo (1994) as well as by Dickey, Jansen, and Thornton (1991). These authors find Johansen's test superior when compared with the other above mentioned cointegration tests. Consequently, Johansen's cointegration test is used in the present research to investigate the long run role of imports in India's economic growth. Johansen's cointegration test results are summarized in Table 2 below.

Johansen's test results of the RGDP and the IMP data indicate that these two variables are cointegrated. (8) Cointegration tests involve testing the [H.sub.0] of no cointegrating equation while the alternative [H.sub.1] assumes the existence of at least one cointegrating equation. Test results reported in Table 2 indicate the rejection of the [H.sub.0] at the five percent level of statistical significance, as indicated by the trace statistic of 21.782 and the maximum eigen value statistic of 15.887. They also indicate the existence of one cointegrating equation. These results provide key empirical evidence on the role of imports in the growth of India's economy. They indicate that imports and economic growth are related in the long run. Therefore, it would be fair to conclude that imports may have contributed to the long run economic growth in India, rather than impeding it. It could also mean that the real GDP growth had influenced imports in the long run. Therefore, finding the existence of a stable long run relationship between imports and real GDP, as indicated by the above cointegration tests, does not necessarily mean that imports caused economic growth in India. Cointegration tests give no indication about the causality in the imports and real GDP relationship. This crucial determination can be made by deploying the Granger (1969) causality testing concept within the confines of further VEC data analyses.

III. VEC ESTIMATION

The next obvious step in analyzing the time-series data that are individually integrated of order one I(1) and cointegrated, is to investigate the short run relationship among such variables within a VEC testing framework. (9) Such an investigation can yield meaningful information about the short run dynamics of the relationship between imports and India's economic growth. It can also provide some empirical evidence on the causal role of imports in India's economic growth. The Engle and Granger (1987) VEC estimation procedure is the most appropriate estimation technique to be used in the present research because there is a direct connection between this procedure and the previously conducted Johansen's (1988) cointegration tests. The residuals from cointegration tests are used in the subsequent VEC estimation. (10) The Engle and Granger VEC estimation requires several steps to be taken in the time-series data analyses. The data must be subjected initially to unit root and cointegration testing. VEC estimation is dependant upon the results of these tests. This methodology is followed in the present research. Individual time-series data must be I(1) and cointegrated. Thereafter, the residuals from the cointegrating equation can be used in the Engle and Granger VEC estimation. All these conditions were met in the present research. Therefore, the Engle and Granger VEC procedure is adopted to investigate the short run relationship between imports and India's economic growth. This procedure involves estimating the following two equations:

[DELTA][RGDP.sub.t] = [alpha] + [rho][z.sub.t-1] + [2.summation over (j=1)] [[beta].sub.j][DELTA][RGDP.sub.t-j] + [2.summation over (j-1)] [[lambda].sub.i][DELTA][IMP.sub.t-j] + [[epsilon].sub.t] (1)

[DELTA][IMP.sub.t] = [[alpha].sub.1] + [[rho].sub.1][z.sub.t] - 1 + [2.summation over (j=1)] [[beta].sub.1][DELTA][IMP.sub.t-1] + [2.summation over (j=1)] [[lambda].sub.1][DELTA][RGDP.sub.t-j] + [[epsilon].sub.1] (2)

The above equations are estimated in the first differences of levels of variables. Therefore, [DELTA] RGDP and [DELTA] IMP are changes in the real GDP and imports, respectively. Since RGDP and IMP are cointegrated, this result implies that at least one of the coefficients ([rho] or [[rho].sub.1]) is nonzero. The [z.sub.t] terms are the residuals from the previously estimated cointegration equation. It is assumed that these residuals are white noises. The analysis of these coefficients plays a crucial role in the interpretation of the VEC estimates. It provides information about the state of equilibrium in the model under investigation. In the present ease, this analysis can provide key information on how imports affect economic growth in India in the short run. In general, focusing on the lagged [z.sub.t] terms provides an explanation of the short run deviations from the long run equilibrium. These lagged coefficients are the speed of adjustment coefficients. Their estimation results determine whether equilibrium or disequilibrium exists in the system under investigation. (11) If disequilibrium exists, then a further analysis of these coefficients can indicate the direction of causality among the variables under the VEC estimation.

Table 3 above summarizes the VEC estimation results. These results provide key information on the role of imports in India's economic growth in the short run. The coefficient of the lagged [z.sub.t] term of 0.100 in equation (1) is statistically significant while the same coefficient of 0.023 in equation (2) is statistically insignificant. This implies that a state of a disequilibrium exists in the present model. Furthermore, these results indicate that imports may have played an important role in the short run economic growth in India. At the same time, it appears that India's economic growth may not have been a major factor in its substantial increase of imports. It is also clear that imports have had a positive impact on India's economic growth, as the [z.sub.t] coefficient in equation (1) has a positive sign. Therefore, international trade that caused large trade deficits in India seems to have also been an important factor in India's spectacular economic growth in recent years. This conclusion is contrary to the conventional view of the role of exports and imports in economic growth.

A further joint analysis of equations (1) and (2) can provide additional important information on the relationship between imports, and economic growth in India. When the test results of these two equations are analyzed together, they can give some indication of causal flows between imports and growth. The direction of causality, in the Granger (1969) sense, can be inferred from the signs and the statistical significance of the two coefficients in focus. One way to interpret the joint test results of the two equations' estimates is to conclude that imports have Granger-caused economic growth in India in the short run. This is crucially important new empirical evidence on the role of imports in India's economic growth. While cointegration analyses indicated that imports and growth are related in the long run, VEC tests suggest that imports may have actually caused economic growth in the short run. Therefore, the present research supports the earlier findings of Lawrence (1999), Roderick (1999), Lawrence and Weinstein (1999) as well as the conclusions about the role of imports in economic growth reached in earlier numerous cross sectional studies [Barro and Sala-i-Martin (1992), Dollar (1992), and Edwards (1992)]. However, it is important to note that the research methodologies followed in these studies are different from the present study's VEC reduced form modeling. Therefore, the present research provides a new, unique look at the role of imports in economic growth in India.

IV. OVERALL CONCLUSIONS

India's economy has experienced spectacular economic growth since it liberalized its international trade policies in the early 1990s. This economic growth has been accompanied by a surge of both experts and imports. However, India's imports have continued to surpass its exports. Standard economic theory maintains that imports have a detrimental impact on economic growth. Therefore, this theory cannot satisfactorily explain India's successful economic growth. Given this fact, it is certainly important to investigate the role of imports in India's economic growth success story. The present research undertakes such an empirical investigation. The novelty of the present research lies in its sole focus on the role of imports in India's economic growth as well as on addressing the issue of causality in the import and economic growth relationship. The entire investigation is undertaken within a reduced form, time-series testing framework.

Annual data on imports (IMP) and real GDP (RGDP) ranging from 1970 to 2005 are used to investigate the impact of imports on India's economic growth. The initial data analyses are comprised of unit root and cointegration tests. The ADF tests indicate that both IMP and RGDP time-series are nonstationary and integrated of order one, I(1). This result suggests the possibility of a long run relationship between imports and economic growth in India. This possibility is investigated by deploying Johansen's (1988) cointegration tests. Cointegration tests reveal the existence of a stable long run relationship between imports and economic growth (approximated by RGDP) in India. Therefore, it appears that imports may have played an important role in India's economic growth in the long run. However, the direction of a causal flow in the imports and economic growth relationship cannot be inferred from cointegration tests alone. VEC data analysis can provide this important information.

The Engle and Granger (1987) VEC testing framework is adopted in the further investigation of the relationship between imports and India's economic growth. VEC test results provide new key empirical evidence on the role of imports in India's economic growth in the short run. They indicate that imports have played an important causal role in this short run economic growth. At the same time, these results suggest that economic growth may not have been a major factor in India's substantial increase of its imports. Therefore, it appears that contrary to the conventional economic wisdom, imports may promote rather than harm economic growth in the importing country. That certainly seems to be the case in India, where an unprecedented surge in imports and economic growth have coincided.

References

Awokuse, T. O. (2007), "Causality between Exports, Imports, and Economic Growth." Economic Letters, 94(3), 389-95.

Balassa, B. and Associates (1971), The Structure of Protection in Developing Countries. Baltimore: The John Hopkins Press.

Barro, R. and X. Sala-i-Martin (1992), "Convergence." Journal of Political Economy, Vol. 100, 223-51.

Dickey, D. A., and W. A. Fuller (1979), "Distribution of the Estimators for Autoregressive Time Series with a Unit Root." Journal of American Statistical Association, 74(366), 427-31.

Dickey, D. A., D. W. Jansen., and D. L. Thornton (1991), "A Primer on Cointegration with an Application to Money and Income." Review, Federal Reserve Bank of St. Louis, 73(2), 58-78.

Dollar, D. (1992), "Outward-Oriented Developing Economies Really Do Grow More Rapidly: Evidence from 95 LDCs, 1976-1985." Economic Development and Cultural Change, 40, 523-44.

Edwards, S. (1992), "Trade Orientations, Distortions, and Growth in Developing Countries." Journal of Development Economies, 39, 31-57.

Enders, W. (1995), Applied Econometric Time Series. New York: John Wiley and Sons.

Engle, R. F. and C. W. J. Granger (1987), "Cointegration and Error Correction: Representation, Estimation, and Testing." Econometrica, 55(2), 251-76.

Fuller, W. A. (1976), Introduction to Statistical Time Series. New York: John Wiley and Sons.

Gonzalo, J. (1994), "Five Alternative Methods of Estimating Long-Run Equilibrium Relationships." Journal of Econometrics, 60, 203-33.

Granger, C. W. J. (1969), "Investigating Causal Relations by Econometric Models and Cross-Spectral Methods." Econometrica, 37, 424-38.

Grossman, G. and E. Helpman (1992), Innovation and Growth: Technological Competition in the Global Economy. Boston: The MIT Press.

Johansen, S. (1988), "Statistical Analysis of Cointegrating Vectors." Journal of Economic Dynamics and Control, 12, 231-54.

Lawrence, R. Z. (1999), "Does a Kick in the Pants Got You Going or Does it Just Hurt? The Impact of International Competition on Technological Change in the U.S. Manufacturing." in The Impact of International Trade on Wages, edited by Robert C. Feenstra. Chicago: The University of Chicago Press.

Lawrence, R. Z. and D. E. Weinstein (1999), "Trade and Growth: Import-Led or Export-Led?" National Bureau of Economic Research Working Paper No.7264, Cambridge, MA, 1999.

Little, I., T. Scitovsky, and M. Scott (1970), Industry and Trade in Some Developing Countries. London and New York: Oxford University Press for OECD.

Lucas, R. E. (1988), "On the Mechanics of Economic Development." Journal of Monetary Economics, 22, 3-42.

McCallum, B. T. (1993), "Unit Root in Macroeconomics Time Series: Some Critical Issues." Economic Quarterly, Federal Reserve Bank of Richmond, 79, 13-43.

Nelson, B. and S. Winter (1974), "Neoclassical vs. Evolutionary Theories of Economic Growth: Critique and Prospectus." The Economic Journal, 84, 886-905.

Reserve Bank of India, Handbook of Statistics on Indian Economy.

Rodrik, D. (1999), "The New Global Economy and Developing Countries: Making Openness Work." Washington, D.C.: Overseas Development Council.

Stock, J. H., and M. W. Watson (1988), "Variable Trends in Economic Time Series." Journal of Economic Perspectives, 2(3), 14-74.

PETER J. SAUNDERS

Central Washington University, East University Way, Ellensburg, WA

Notes

(1.) The average annual growth rate in the 1990s exceeded seven percent. India's economy continued to grow at an eight per cent average annual rate from 2000 to 2005.

(2.) During this time period, small trade surpluses only occurred in 1973 and 1977.

(3.) The main advantage of reduced form modeling of economic data, such as the present VEC estimation, is its ability to focus on the relationship among the few key variables under investigation, such as imports and growth. VEC data modeling accomplishes this objective. This type of estimation does not rule the possibility that other variables may play important role in the economic growth of India.

(4.) These data were obtained from the Reserve Bank of India, Handbook of Statistics on Indian Economy. Oil and non-oil imports were used as the measure of total imports while real GDP at factor costs was used as RGDP.

(5.) The classical regression model can only be used when all time-series data are stationary; i.e., when such data are unit root free.

(6.) The interested reader is referred to McCallum (1993), among others, for a detailed analysis of the relationship between unit root and cointegration testing.

(7.) See Dickey, Jansen, and Thornton (1991) for a more detailed discussion of cointegration testing issues.

(8.) The assumption of a linear deterministic trend in the data was used in the above cointegration tests.

(9.) Enders (1995) provides a more complete explanation of this point.

(10.) See Enders (1995), pages 373-81 for a detailed explanation of the connection between Johansen's (1988) cointegration test and the Engle and Granger (1987) VEC estimation.

(11.) If the [z.sub.t] coefficient is statistically insignificant, then the system under investigation is in equilibrium. Disequilibrium exists if the' [z.sub.t] coefficient is statistically significant. Table 1 Augmented Dickey-Fuller (ADF) Test Results for RDGP and IMP Variable Test Results RGDP (1) 3.376 RGDP (2) -3.813 * IMP (1) 1.355 IMP (2) 3.846 * (1) ADF test results for levels of variables. (2) ADF test results for the first differences of levels of variables. * Indicates statistical significance at the five-percent level. Table 2 Johansen Maximum Likelihood Test Results for RGDP and IMP (lags 1-2) Variables Trace Statistic Max-Eigen Statistic RGDP and IMP 21.782 * 15.887 * * Indicates statistical significance at the five-percent significance level. Table 3 VEC Estimates of Equations (1) and (2) Dependent Independent "t" Equation Variable Variable Coefficient Statistic (1) [DELTA] RGDP constant 80143.39 4.914 * z (-1) 0.100 4.007 * [DELTA] RGDP (-1) -0.398 -1.921 [DELTA] RGDP (-2) -0.210 -1.064 [DELTA] IMP (-1) 1.269 4.479 * [DELTA] IMP (-2) -0.045 -0.069 (2) [DELTA] IMP constant 13625.11 1.286 z (-1) 0.023 1.453 [DELTA] IMP (-1) 0.993 5.437 * [DELTA] IMP (-2) 0.647 1.547 [DELTA] RGDP (-1) -0.020 -0.151 [DELTA] RGDP (-2) 0.648 -2.142 * * Indicates statistical significance at the five-percent level. Numbers in parentheses indicate number of lags.