The Point: One major characteristic of the process of economic development is the movement of people from rural to urban areas. As a result, the percentage of population living in urban areas (the rate of urbanization) increases, with economic development usually going hand-to-hand with urbanization. According to classical theories (i.e., Lewis 1954; Kuznets 1955), this process is related to economy-wide inequality in a non-linear way: inequality first increases, as countries urbanize, and then declines as urbanization proceeds. This non-linear relationship between income (and urbanization) and inequality is known as the Kuznets’ inverted-U. But economic development is also associated with a change (usually an increase) in the number, absolute size and distribution of urban areas (cities). According to the urban economics literature, different cities and different size of cities are expected to experience different levels of mean income and of income inequality. Consequently, there is no reason to expect that when the number, size and distribution of cities changes, inequality will remain unchanged. However, this is an issue that to date remains understudied. The overarching aim of this paper is to analyze the relationship between the size and distribution of cities and income inequality, using panel data for as many countries around the world as possible, looking at nation-wide inequality, controlling for several determinants of inequality, and considering non-linearities in the relationship.
Contributions: (1) To show that beyond Kuznets’ hypothesis, there is a U-shaped relationship between average city size and inequality; inequality is expected first to fall and then to increase with average city size. A result that is found to be robust to a long list of controls, and different estimation techniques and identification strategies. (2) To analyze the mechanisms to which this result may relate to.
Method/Data: The paper builds on theory and evidence in urban economics regarding the relationship between city size and inequality. Econometric results rely on several panel data estimation techniques, including First-Difference and Instrumental Variables estimations. Several data sources are used, including the SWIID dataset for inequality, WB-World Urbanization Prospects, for data on population in cities, and historical data from Mitchel (2013).
Citation: Castells-Quintana, D. "Beyond Kuznets: Inequality and the size and distribution of cities" Journal of Regional Science (forthcoming)