Determinants of German Foreign Direct Investment in Latin American and Asian Emerging Markets in the 1990s

Many empirical studies in the area of foreign direct investment (FDI) exclusively focus on flows between industrialized countries. This article makes a contribution to the still relatively sparse literature on FDI in emerging markets by estimating determinants of German FDI flows to Latin America and Asia during the past decade. Using data contained in a newly available Bundesbank microdatabase, an FDI flow variable, constructed from year-to-year differences in FDI stocks adjusted for certain otherwise distorting factors, is empirically tested with respect to several exogenous variables previously found to be significant in the literature. These include so-called non-traditional factors such as country risk and agglomeration effects which are widely regarded as influential for FDI in emerging market economies. This study therefore focuses on estimating the effects of various risk measures and finds that country risk, and partially political risk, is indeed detrimental to investments of German enterprises. Moreover, German FDI in Latin America are found to have been market-seeking while those in emerging Asia tended to exploit low factor costs. Methodically, this paper uses the SUR estimation technique which allows for the contemporaneous correlation of disturbances as well as first-order autocorrelation of the time series disturbances and cross-sectional heteroskedasticity. In arriving at a parsimonious regression for each region, an Extreme Bounds Analysis (Leamer, 1983 & 1985) is performed to select individual variables robust to the inclusion of other explanatory variables. Making empirical use of German firm-level data, additional estimations are performed for direct investment of the manufacturing sector and three of its sub-sectors. Regarding the latter, the hypothesis that capital-intensive industries react particularly strongly to the changes in the regulatory environment of the host country is confirmed by the data.