Export Growth and Economic Growth: An Application of Cointegration and Error-Correction Modeling

Abstract

Most recent studies that have used time-series data to investigate the causality between a country's export growth and its economic growth have failed to provide uniform support for the export-led growth hypothesis.' These studies have used the Granger or Sims procedure in order to investigate the possibility of dual causal relationship between export growth and economic growth (measured by output growth). For example, Chow, who adopted the Sims procedure and investigated the causal pattem between export growth and growth in manufactured output, found dual causality in the cases of Brazil, Hong Kong, Israel, Korea, Singapore, and Taiwan; one-way causality from export growth to output growth in the case ofMexico; and no causality in the results for Argentina.2 By contrast, Jung and Marshall, who used the Granger concept of causality, found statistical evidence supporting the export-led growth hypothesis in only 4 out of 37 countries.3 Jung and Marshall's findings received further support by Bahmani-Oskooee et al., who employed the Granger concept of causality combined with Akaike's Final Prediction Error (FPE) criterion and concluded that we find some support in favor of the export-led growth hypothesis, though the evidence is at most inconclusive in evaluating competing hypotheses.4 Darrats and Hsiao6 are other studies that have used time-series data and rejected the export-led growth hypothesis for most LDCs in their sample. There are three major shortcomings associated with all the time-series studies just mentioned. First, none ofthe studies have checked for the cointegrating properties of the time series involved. Standard Granger or Sims tests are only valid if the original time