Export-led growth and the Japanese economy: evidence from VAR and directed acyclic graphs

Abstract

This paper explores the causal relationship between real exports and GDP growth in Japan using two recently developed causal modelling approaches. Using Japanese time series, the paper employed the augmented VAR methodology developed by Toda and Yamamoto to test for Granger non-causality. Then, a more recently developed technique of directed acyclic graphs (DAG) was also used in providing over-identifying restrictions on the innovations from a vector autoregression (VAR). In contrast to prior analyses, the application of DAG techniques allows for the examination of both contemporaneous and dynamic causal structure of the exports-productivity nexus. The empirical results reveal that the causal path between exports and GDP growth in Japan is bi-directional. Furthermore, other variables such as capital and foreign output are also significant determinants of productivity growth in Japan.