Spatial and sectoral composition effects of agglomeration economies in the Netherlands.

In this article we test for dynamic inter- and intra-industry externalities on the urban level in the Netherlands. We argue that previous contributions might be sensitive to untested spatial and sectoral composition effects of urban data. We conclude that research results are better controlled when analysed on lower spatial scales, that results improve in robustness when spatial dependence in the form of spatially lagged versions of explained (growth) variables is introduced in the econometric models, and that results are more informative when hierarchical urban regimes are tested for. Introducing spatially lagged versions of explanatory agglomeration variables is informative but leads to less robust outcomes. In general our research results are more conclusive on inter-industry externalities circumstances when outcomes of city-industry as well as sectoral research designs are compared with the same dataset. [ABSTRACT FROM AUTHOR]