Trade and GDP growth: Causal relations in the United States and Canada.

• Author: Zestos, George K.
• Position: Statistical Data Included

1. Introduction

   Economic policies leading to economic growth and development have been studied by many economists for a long time. The literature in this area is rich; a number of candidate variables that may be related to economic growth have been considered and carefully examined. Some of these variables are investment, saving, inflation, inflation variability, governmental expenditures as a percentage of GDP, government deficit, and other mainly macroeconomic variables. Many economic models were constructed for the purpose of understanding economic growth and to shed light on this issue.(1) A group of economists has focused exclusively on the foreign sector, particularly on the relationship of exports, imports, and GDP growth. Emphasis on international trade dates back to the mercantilists more than two centuries ago. Mercantilists were firm believers that trade surpluses were the only favorable outcome for the domestic economy from international trade relations. Mercantilists supported export promotion and protection of domestic industries since they were preoccupied with the accumulation of gold reserves and not necessarily with the standards of living or the growth and economic development of the country per se. Many other authors, since the mercantilists, support the expansion of the foreign sector for a variety of reasons. One of these reasons is that expansion of the export sector allows countries to attain economies of scale by specializing in production. This is particularly important for smaller countries where the national markets are too small to allow specialization. Some economists in favor of expansion of the export sector stress the common belief that the export sector is the most efficient sector of the economy. It is the sector where workers enjoy the highest wages and firms earn the highest profits since only the most efficient firms can compete successfully in the global market. Other supporters of export promotion point out that development of the export sector permits countries to have access to higher le vels of technology and technologically rich capital. This access is crucial to developing countries. Such inflow of foreign capital and transfer of technology would not be possible without the export sector providing the means for payment since exports constitute the main source of foreign exchange. Export expansion allows countries to follow a speedier pace toward industrialization and economic growth.

   A variety of models have been suggested in the literature to study the effects of the foreign sector on the domestic economy and vice versa. A group of econometric models rely on Granger causality tests to explain relations between trade and the domestic economy. Three studies from the literature that employed similar methodology are reviewed herein. Ahmad and Harnhirun (1996) examined causality between exports and economic growth for five countries of the Association of Southeast Asian Nations (ASEAN). The countries were Indonesia, Malaysia, the Philippines, Singapore, and Thailand. Their model is a bivariate two-equation vector autoregression (VAR) covering the period 1966-1986. Ahmad and Harnhirum were able to test for cointegration in only four of the countries since exports and GDP for Thailand were not integrated in the same order. In the remaining four countries, they found that exports and GDP were not cointegrated; consequently, the error correction term could not be included in their model. Based on their results, Granger causality is supported from GDP to exports for each of the four countries. This finding runs against the common belief that Southeast Asian countries were exceptionally successful in achieving economic growth by following export promotion policies.

   Dutt and Ghosh (1996) studied causality between exports and economic growth for a relatively large sample of countries using the methodology of the error correction model (ECM). For the countries in which they found cointegration, the VEC model was estimated, and tests for Granger causality were performed. Canada and the United States were two of the countries in their sample, which covered the period 1953-1991. Dutt and Ghosh found no causality for Canada between exports and GDP in either direction, but they found causality from GDP to exports for the United States. In other countries the results were mixed. Some countries experienced export-led growth, others the opposite (growth-led exports), some showed bidirectional causality, and others demonstrated no causality. Their model differs from the present analysis since Dutt and Ghosh utilized a bivariate two-equation ECM model. An interesting feature of the empirical part of this paper is that the authors pointed out the source of causality for each country, that is, short- or long-run causality. This was based on the F- and t-tests, respectively.

   Restricted and unrestricted VAR models were employed by Ghartey (1993) to examine any causal relation between exports and economic growth for Taiwan, Japan, and the United States. Ghartey utilized Hsiao's version of Granger causality (Hsiao 1979). The three endogenous U.S. variables were GDP growth, export growth, and capital stock or the terms of trade as the third variable. For the United States, it was found that economic growth causes export growth, while the opposite is true for Taiwan. A feedback causal relationship or bidirectional causality between exports and economic growth was found for Japan. It is clear from these and other studies not reported here that there exists inconclusive evidence regarding the causal relations of trade and GDP growth. Such inconclusiveness should be attributed partly to different methodologies and periods covered by these studies as well as to genuine differences between these economies. Similar results with the present study for Canada and the United States were found i n another study by Tao and Zestos (1999) for Japan and Korea. For Korea, a country that has a more trade-dependent economy than Japan, Granger causality tests indicate stronger causal relations.

   This paper investigates causal relations between the growth of GDP, exports, and imports for Canada and the United States, using a trivariate VEC model. Evidence of cointegration for both countries allowed us to estimate the VEC model. Granger causality tests were performed on the basis of the estimated VEC model. The model distinguishes two types of causality: long-run causality and short-run causality. The Granger causality tests reveal that the existence, direction, and degree of Granger causality in the two countries differ substantially. These results can be explained by historical differences of the two countries' economies. Stronger causal relations were revealed in the growth of GDP, exports, and imports for Canada than for the United States. These differences are attributed mainly to differing degrees of openness of the two countries to the world economy.

   This paper is structured as follows. In section 2, we discuss international trade and development theories. Granger causality tests, cointegration, and the VEC model are also presented in this section since they are the main tools of our analysis. In section 3, we describe and present graphically the Canadian and the U.S. data and report the unit root tests for all variables. Since the variables, exports, imports, and GDP were integrated of the same order and cointegrated, the model was estimated. The estimated VEC model is presented in section 4 together with the results of the Granger causality tests. In section 5, a conclusion and a summary of the paper are given.

2. International Trade and Economic Development Theories

   Many authors have stressed the positive effects of the export sector to the rest of the economy, including Balassa (1978, 1985), Krueger (1980), Feder (1983), and Bhagwati and Shrinivasan (1978). It has also been suggested that growth of output causes growth of exports (Jung and Marshall 1985). Other groups of economists have opposed the export-led growth approach. Nurkse (1961) advocated the "balanced growth" theory, while Prebisch (1962) supported the import substitution approach; the latter is diametrically opposite to the export-led growth hypothesis. The import substitution approach dictates self-sufficiency of the country and thus absolute trade protection. These trade and development theories have had an unparalleled influence on long-run economic policies adopted by countries. The two polar cases, export-oriented growth and import substitution, have split the developing countries into two distinct groups. The first is represented mainly by the Southeast Asian countries and the second by the Latin Amer ican countries.

   Economists have constructed many models for the sole purpose of explaining how trade expansion contributes to economic growth. Estimation of a single equation and correlation analysis dominated the early contributions. A few authors, such as Kwak (1994), adopted the factor growth accounting method, while other economists, such as Pack and Page (1994) and Esfahani (1991), utilized the neoclassical growth model. Some of these models employ cross-section multi-country data (see Afxentiou and Serletis 1991), while other models utilize time-series data for one country or a selected group of countries.

   Granger Causality Models

   Several econometric studies focus exclusively on Granger causality (Granger 1969). Granger causality from a variable X to a variable Y is established when knowledge of past values of X improve the prediction of future values of Y, over and above the prediction that is based on knowledge of past values of Y alone. (2)

   The simplest standard causality test is the pairwise Granger causality test. This is a bidirectional test for Granger causality regarding only two variables. Granger causality from X to Y is established when the coefficients of the lagged differences of X are found to be jointly...