Can large cities explain the aggregate movements of economies? Testing the ‘granular hypothesis’ for US counties

Abstract

The concentration of economic activity across space is a stylized fact of the literature in regional and urban economics. In the case of the United States (US), only 35 counties at the top of the distribution, out of a total of 3,138, generate more than one third of the country’s total personal income. Furthermore, US urban structure follows Zipf’s law. Under these conditions of concentration, we should presumably find some kind of granular hypothesis as suggested by Gabaix (Econometrica 79(3):733–772, 2011). This hypothesis means that the idiosyncratic behavior of the large units should be capable of explaining a significant part of the aggregate shocks. Gabaix found evidence to corroborate this hypothesis for firms: the idiosyncratic movements of the 100 largest firms in the US appear to explain about one third of the variations in output growth. In this paper we analyze whether this behavior is present in the case of urban concentration, as can be expected taking into account the degree of spatial concentration in the economy. We use data from the Bureau of Economic Analysis on personal income, which is disaggregated at a local level, US counties, from 1969 to 2011. The granular residual of the largest cities is calculated and used to explain the US aggregate economic evolution. The overall results provide support for the granular hypothesis: the idiosyncratic shocks to the top counties can explain a significant fraction of the volatility of US aggregate data.