Household Sorting and Politics: Empirical Evidence for the Metropolitan Area of Porto (Portugal)

Sorting and heterogeneity of households at metropolitan level has been a subject of many empirical studies. Interest in the subject goes beyond the knowledge of household mobility because it also has importance for the debate on the Tiebout’s hypothesis that households “vote with the feet” by choosing the best bundle of local public goods and local taxes. A result of households “voting with the feet” is a more heterogeneous public provision of local public goods by municipalities and, at the same time, a smaller distance between household preferences and public provision of local public goods. Household movements are determined not only by the bundle of public goods-local taxes in each jurisdiction but also by restrictions to mobility such as real estate market and availability of jobs. Several authors argue that aggregate empirical studies are not sufficient to test empirically the Tiebout’s hypothesis because sorting and heterogeneity may be a consequence of those restrictions. So they defend instead the implementation of micro studies where we observe the motivations of household movements. In our study we analyze sorting and heterogeneity using an aggregated approach, but we provide empirical evidence for two different periods: a first period in which there is no role for local politics in the sorting and heterogeneity process; a second period where there is an increasing role of local politics on the decision of households to locate. The empirical evidence on sorting and heterogeneity in those two periods gives us a hint about the importance of local politics on household decisions to locate in a metropolitan area, and therefore provides empirical evidence on the importance of the Tiebout’s hypothesis compared with other determinants of household mobility. The empirical study is conducted for the Metropolitan Area of Porto for a long period in order to include non-elected local governments and elected local governments. The empirical evidence we gather covers a wide period that goes from 1920 till 2011 using census data.