This paper analyses empirically the dynamics of regional inequalities in GDP per capita. Our starting hypothesis is that the evolution of regional inequalities should follow a bell-shaped curve depending on the level of national economic development.

 A number of authors going from Kuznets (1955) to Lucas (2000) have provided extensive theoretical arguments along this line suggesting that growth, because of its very nature, is unlikely to appear everywhere at the same time. Regional inequalities should then rise when countries start developing and then fall once a certain level of national economic development is reached as long as spillovers are strong enough to transmit growth and technological progress across regions.

We test empirically these predictions by using regional data for a panel of European countries and by making use of semi-parametric estimation techniques. Our results provide strong support for a bell-shaped curve in the relationship between the national GDP per capita level and the extent of regional inequalities independently of the time period and regional administrative units considered. The nature of this non-monotonic relationship is not altered by the inclusion of other possible determinants of regional inequalities. A number of policy implications are derived from our results.