The spatial structure of French wages: Investigating the robustness of two-stage least squares estimations of spatial autoregressive models

Résumé 0

Two stage least squares are a popular method of estimation of spatial auto-regressive models, where the dependent variable in an area is a function of the value of the same variable in contiguous areas. Existing literature on this topic points out, however, that this creates problems of consistency. Nevertheless, studies such as Fingleton (2003) show that such an approach is being used to test the central hypothesis of New Economic Geography that increasing returns to agglomeration lead to the concentration of economic activity. It is therefore important to investigate the validity of the methodology in this case.The focus of this study is twofold: first to replicate the methodology of Fingleton (2003) on the French case and investigate the presence of increasing returns to agglomeration in the spatial structure of wages in France. Secondly, because of the econometric problems inherent to the specification pointed out in the literature, the study tests the validity and robustness of the results obtained.The first central finding is the significant presence of such returns to scale for France, similar to the ones found in the UK and in other studies of French spatial wage disparities. The second finding is that rigorous tests on the instrumentation strategy defined in Fingleton (2003) reveal that the instruments are typically strong and lead to consistent estimates. Finally, the methodology is shown to be robust to changes in the specification of the spatial weights matrix, and that taking into account a larger time-dimension through a simple pooled regression is valid and leads to an improvement of the significance of the parameters.