Estimating the determinants of foreign direct investment inflows: how important are sampling and omitted variable biases?

This paper investigates the importance of factor endowment vis-à-vis institutions in explaining the locational choice of foreign investors during the 1990s.Using dynamic panel estimation on data for transition economies, we find that low labour costs, bureaucratic efficiency ("institutions"), agglomeration economies and natural resource abundance are key factors explaining foreign investors' decisions.However, sampling proves fundamental as these overall determinants mask deep and, so far empirically unexplored, differences between groups of recipient countries.For example, for the former Soviet Union economies we estimate that labour costs are no longer crucial, but abundance of natural resources and (interestingly) lower levels of human capital are.For Eastern Europe, we find that external liberalisation (one aspect of economic reform) is crucial in foreign investor's decisions.The main message is that minimising sampling biases and accounting for previously omitted variables yields a different, much richer picture than previously available.