Relative investment specialisation inside EU: an econometric analysis for EU-regions

Models of economic geography predict an increase in concentration when transport costs diminish. These models, however, focus solely on the analysis of the production factor labour. Neglecting capital, though, might lead to disturbed results - particularly in the European context where capital, compared to labour, traditionally is the more mobile factor of production. In addition, a profound analysis of regional, not national, patterns of concentration is still missing in recent research. We regard regional data of the Eurostat REGIO database for the time period 1985 to 1994 and calculate indices of relative concentration of gross fixed capital formation (GFCF) for each region and each year. One important feature that can be detected is the higher level of relative investment concentration when regarding the more disaggregated NUTS 2-regions in comparison to NUTS 1-regions. In a descriptive analysis higher concentrated regions have been found to perform worse in economic terms than lower concentrated regions with respect to unemployment rate, number of patents, total regional GDP and total regional GFCF. Peripheral regions are often highly concentrated and of poorer economic performance than core regions. Economic centres, especially the region of Bruxelles and the Île de France, are highly concentrated as well. However, they demonstrate a good potential of high economic performance. As no causal relationship can be derived from this purely descriptive analysis, econometric analyses have to be conducted to test for the significance of potential determinants of agglomeration. The theoretical basis for the empirical investigation of the determinants of capital agglomeration are the results of models of the new international trade theory, the new economic geography as well as of gravity models (recently extended to the analysis of direct investment flows). They focus on determinants such as market size, factor endowments, distance (proxy for transport costs), scale intensities, technological capacities or research intensities, labour costs and liberalisation or integration effects. Results from cross-sectional and pooled regression analyses - controlling for problems of potential enodgeneity - point at a high importance of market size and regional size (having a decreasing impact on the level of investment concentration), the unemployment rate, and the centrality or population density of a region (increasing impact). The empirical effects of transport costs - though an important theoretical determinant in the new economic geography - are rather mixed. But, the impacts of integration, i.e. economic openness and capital account liberalisation, seem to enforce concentration tendencies. Between estimates, explaining the variation between regions, lead to results very similar to those of the cross-sectional and pooled estimates. By means of fixed effects estimates with highly significant region-specific effects and estimates in first differences, there are only few significant determinants to be detected. Thus, the variation within a region and the change in the level of concentration over time seem to be the result of random disturbances and not to underlie systematic changes. The robustness of our results will be checked applying spatial econometric tools taking account of regional interdependencies which are likely to occur when analysing regional effects and especially agglomeration tendencies.