Openness, Investment and Growth

Abstract

In the empirical growth literature both investment and openness to international trade have been identified as important determinants of growth. These relationships appear to be robust in a number of specifications. However, Sachs and Warner claim that openness is such an important determinant of investment that the coefficient on investment will be insignificant in growth regressions, which already account for openness. I re‐examine this result and can only support it if I use their model specification and estimation method. I suggest that their ordinary least squares estimation suffers from both endogeneity and omitted variable bias. Using panel data analysis, I show that ignoring unobserved country‐specific effects and endogeneity problems cause investment to be insignificant. Estimating the model with the Blundell and Bond system general method of moments estimator, which allows me to address the omitted variable as well as the endogeneity bias, I find that investment is significant in the Sachs-Warner model despite controlling for openness. Thus, my regressions confirm that openness has a significant, positive effect on income. However, this variable is not so important that it drives investment out of the model. My re-examination of the Sachs- Warner models shows that results from single cross-country growth regressions can be misleading when unobserved country-specific effects and endogeneity problems are ignored.