The urban structures between the Member States of the European Union is very different for historical, geographical, economic reasons. However, the population is spread across geographic areas in a way that, although continuously changing, is not possible to define as random. Indeed, countries have faced a strong tendency toward agglomeration, namely population gathers within proper areas like cities, and currently the agglomeration within cities "is an extremely complex amalgam of incentives and actions taken by millions of individuals, businesses, and organizations" Eeckhout (2004, p. 1429). Then, the creation of the European Union, the distortions caused by the introduction of a single currency in countries structurally so different and the expansion of mobility of people, capital and services due to the constitution of the so-called Schengen Area from the beginning of '90s might have had some impacts on the dynamics of city populations. This paper provides a study of the hierarchical structure of the cities within the EU Member States with particular attention on agglomeration forces by means of two very well-known empirical regularities: Zipf's law, as a proxy for agglomeration forces, and Gibrat's law as a test for stationarity. Indeed, the Zipf coefficient can be seen as a measure of urbanization: the larger the value of the coefficient, the more even the population of cities in the urban system. There are several potential explanations for variations in its value, one of these can be found in a model of economic geography a la Krugman (1991) and Fujita et al. (1999). These models can be viewed as models of unevenness in the distribution of economic activity and moreover, they state that for certain parameter values, economic activity is agglomerated, while for other parameter values, economic activity is dispersed (i.e. a city system will be more agglomerated the greater are scale economies, the lower are transport costs and the lower the share of international trade in the economy) By means of parametric and non-parametric analysis the main conclusions of this paper are the following: the hierarchical structures of Member States is more even than expected. Moreover, the European Union seems to be split in three distinct areas: an area characterized by the validity of Gibrat's law (temporary idiosyncratic shocks might have permanent impacts on the city structure); an area characterized by the presence of mean reversion (any exogenous shock is used up in certain amount of time); a small area where the effects of the shocks is magnified in the big cities. Finally, we find that only the constitution of the Schengen Area and the share of international trade seem to have a weak impact on the hierarchical structures of Member States.
