37 Pages Posted: 20 Dec 1998
Date Written: October 1998
This paper looks at the patterns of causation between income, export, import, and investment growth for 25 developing countries. Our approach differs from previous efforts in a number of ways. First, we examine each country individually in order to allow for complete heterogeneity and properly account for the stochastic trending properties of the data. Second, we apply model selection techniques which are based on in-sample goodness-of-fit criteria and ex-ante predictive ability criteria to identify the best model for each country, and then look for evidence of reverse causation. Finally, we propose a rather novel device based on simple contingency tables which allows us to assess whether our models are capable of accurately predicting turning points in GDP growth. We find that countries with high trade exposure fare poorly in this dimension and posit that the GDP growth in such countries is best modeled using an index of global business cycle conditions, in addition to the above variables. Overall, we find that in around two thirds of the countries examined, growth is best explained by exports and/or imports. Further, and in contrast to previous findings of bi-directional causality, around 70% of the countries exhibit uni-directional causality.
Notes: Readers need Scientific Workplace fonts to be installed for electronic document to be fully readable.
JEL Classification: F43, O47
Suggested Citation: Suggested Citation
Krishna, Kala and Swanson, Norman R. and Ozyildirim, Ataman, Trade, Investment and Growth: Nexus, Analysis and Prognosis (October 1998). Available at SSRN: https://ssrn.com/abstract=137248 or http://dx.doi.org/10.2139/ssrn.137248