Neoclassical theory versus New Economic Geography. Competing explanations of cross-regional variation in economic development

This paper tests whether the success of new economic geography (NEG) and its wage equation is replicated using data for NUTS-2 regions in Europe, under the challenge of a competing neoclassical growth model and the need to control for additional effects. Testing is accomplished by fitting both rival models and by setting up a composite spatial panel model within which rivals are nested, using a panel of 255 European regions over the time period 1995-2003. While both competing models individually provide good explanations of the data, only the NEG model retains explanatory power in the presence of region-specific fixed effects and it is dominant within the composite model. In addition, the NEG model parameter estimates correspond to theoretical expectation, whereas the parameter estimates from the Solow model are counterintuitive. However estimation of the NEG model is problematical because of endogeneity and measurement error.