Abstract
Employing a panel dataset of bilateral inward and outward foreign direct investment (FDI) for 31 Asian countries and territories over the period 2001-2012, we estimate the knowledge-capital (KK) model to find the dominant type of FDI within Asia. We consider alternative estimation methods to deal with zero values, which form the majority of the bilateral observations. Based on a series of model selection and diagnostic tests, we conclude that Lognormal Hurdle and Poisson Pseudo-Maximum-Likelihood are the most appropriate. Controlling for host and source country characteristics, our findings suggest that vertical FDI is the dominant type between Asian countries. However, we find little empirical evidence in support of the KK model’s predictions for its key variables, such as total GDP and skill difference, when country fixed effects are included. Some factors (distance, trade costs to both source and host country, the GDP difference between source and host country, and a common spoken language) are found to have statistically significant impacts on the volume of FDI between Asian countries, regardless of whether or not fixed effects are included.