Location Determinants of Greenfield Foreign Investments in the Enlarged Europe: Evidence from a Spatial Autoregressive Negative Binomial Additive Model

This paper addresses two important methodological issues in the analysis of industrial location: spatial dependence and nonlinearities. To this end, we estimate a semi-parametric spatial autoregressive negative binomial model using data on the number of inward greenfield FDI occurred over the 2003-2007 period in 249 European regions. Results support the view that multinational firms' location choices are very spatially dependent, even controlling for a large number of regional characteristics. A spatial lag model with a non-parametric spatial filter allows us to purge the residuals from spatial dependence and yields sensible changes in the magnitude of some estimated coefficients. We also provide robust evidence of nonlinearities. In particular, we find that the effect of agglomeration economies fades down as the density of economic activities reaches some limit value.