Local public goods and the geography of economic activity[Biens publics locaux et géographie de l'activité économique]

This thesis studies how heterogeneity in the quality of local public goods may influence the geography of economic activity, through the study of two mechanisms generating such heterogeneity. In the first two chapters, the local public good I study is technological knowledge. The analysis is rooted in a vast body of literature showing that knowledge flows exhibit a strong spatial bias. The first chapter studies the dynamics of link formation between innovators, and their link with the aggregate effect of distance on knowledge flows. The analysis shows that innovators learn about new knowledge gradually, using the contacts of their own contacts. Inserting this fact in a network formation model yields predictions about the size of innovators and a relation between size and the distance of citations, which are met in the data. The second chapter takes these local innovation networks as given, and investigates how they influence firms' location choices through their decisions to relocate R&D labs. I show that innovative firms are more mobile than the average firm, and that denser innovation networks attract them while a poor position in their local network makes them more likely to leave. I then study theoretically the problem faced by firms able to relocate their R&D labs with limited information about the other locations.The third chapter studies a different problem in which a local public good provided by the public sector is spatially heterogeneous, due to wages set centrally. It shows that public sector workers exert positive spillovers on private sector workers, implying that heterogeneous levels of public good provision distort the geography of private sector activity.