Abstract:
A number of recent papers study the impact of institutions, trade and geography known as “deep determinants” of economic development using cross-section data. This paper instead employs a panel data approach to examine the impact of these three determinants on per capita income. Our approach enables us to account for unobserved heterogeneity across countries, an issue that cannot be addressed in a cross-section framework. Moreover, employing the Hausman and Taylor (1981) approach allows us to obtain direct parameter estimates of the time invariant explanatory variables like geography or some institutional measures, making our results comparable to the existing cross-section iterature. Also, by using lagged explanatory variables whenever possible we can account for contemporaneous correlation between these variables and the idiosyncratic error term. We find that the quality of institutions and openness to trade both have positive and statistically significant coefficient estimates throughout most specifications, while geography, captured by malaria ecology measure, has negative estimates that are often, but not always statistically significant. In terms of their economic impact, institutional measures appear to have the strongest impact, followed by openness to trade measures. In comparison, geography measures have rather small elasticity estimates.