Neural Network Approaches to Estimating FDI Flows: Evidence from Central and Eastern Europe

Abstract

Central and East European (CEE) countries are in an economic transition process that involves convergence of their economic performance with the European Union. One of the principal engines for the necessary transformation toward EU average economic performance is inward foreign direct investment (FDI). Quantitatively examining the causes of FDI in the CEE region is thus an important research area. Traditional linear regression approaches have had difficulty achieving conceptually and statistically reliable results. In this paper, we offer a novel approach to examining FDI in the CEE region. The key tasks addressed in this research are a neural network (NN)—based FDI forecasting model and a nonlinear evaluation of the determinants of FDI. The methodology is nontraditional for this kind of research (compared with multiple linear regression estimates) and is applied primarily for the FDI dynamics in the CEE region, with some worldwide comparisons. In terms of mean square error (MSE) and R 2 criteria, we find that NN approaches better explain FDI determinants' weights than do traditional regression methodologies. Our findings are preliminary, but offer important and novel implications for future research in this area, including more detailed comparisons across sectors, as well as countries over time.