The types of city size distributions and their evolution

The widely recognized rank-size law is being questioned as the single best rule for city size distribution by an increasing number of studies. However, these studies suffer from inconsistent findings and some deficiencies, including differing definitions of a city across countries, samples with few countries, less systematic examination of socioeconomic influencing factors and no attention being given to the dynamic changes in best-fit city size distribution types for a given country. With a larger sample of countries and a more comparable and consistent definition of cities, our empirical analysis supports two findings from the existing literature. First, there is no single, universal distribution but rather a plurality of national city sizes. Second, the main body of a country's city size distribution is generally characterized by a lognormal distribution, while the upper tail tends to follow a Pareto distribution. Furthermore, our exploratory analyses reveal that specific city size distributions are associated with nations' socioeconomic development and urbanization levels. Finally, this paper presents the changes in the best-fit city size distribution type of a country across different periods and proposes a hypothesis on the general evolutionary trend.