Agglomeration and human capital: an extended spatial Mankiw-Romer-Weil model for European regions

Abstract Over recent decades a handful of very rich European regions have increased the gap separating them from the European average in terms of labour productivity. In this paper we extend a spatial version of the Mankiw, Romer and Weil to highlight the differential impact of spillovers linked to human capital over regions with agglomeration economies. The theoretical model leads to a cross-sectional spatial Durbin model specification, which is estimated for 121 EU regions for 1995–2014 period. An interaction term of human capital with a dummy variable is used to capture the externalities of human capital in the most agglomerated regions. The dummy relies on composite indexes of agglomeration that consider the spatial dimension of the data. The indexes let us distinguish between regions intrinsically agglomerated, mainly the capital cities, and those that become agglomerated by the spatial relationship with their neighbours. Our main result shows that investment in human capital has stimulated labour productivity growth in those regions with the greatest potential to benefit from agglomeration economies and it is an important source of divergence across European regions.