DOES THE CHOICE OF GEOGRAPHIC UNITS MATTER FOR THE VALIDATION OF GIBRAT'S LAW?

As we hypothesize, the selection of observation units may become a critical consideration for an empirical investigation of Gibrat’s Law, according to which the growth of cities is expected to be independent of their sizes. In order to verify this hypothesis, we investigate whether the validation of Gibrat’s law may depend on the aerial units used for the analysis i.e., individual localities vs. integrated urban areas. The present study examines that possibility, using 1990-2000 population growth data for two levels of geographic resolution 4,667 local administrative units (i.e., municipalities) and 2189 contiguous urban areas in 40 European countries. According to our findings, the association between population size and growth tend to differ across subsets of localities, being positive across localities with fewer location advantages and negative elsewhere. As a result, the strength and direction of the ‘growth-size’ association may change, depending on the relative shares of location subsets used for aggregation.