A test for the export-led growth hypothesis in possibly integrated vector autoregressions

Abstract

This article re-examines the direction of causation between exports and output by testing for Granger noncausality under the framework of four-variable vector autoregression. Both the methodologies developed in Toda and Phillips (1993, 1994) and Toda and Yamamoto (1995) are applied. A country is recommended to enforce export promotion policies as its appropriate development strategy, if its manufacturing sector takes up a relatively large share.

Notes

¹ Although these papers among others did consider the issues of integration and of cointegration among variables, they appear to ignore the role of the error-correction term when conducting Granger noncausality tests.

² Since the first oil crisis occurring in 1974 might have a significant adverse impact on these countries, a dummy variable D74 is included in our VAR systems. It takes a value of one in 1974 and zeros otherwise. However, it is noteworthy that estimates in Table 1 do not take into account D74.