Local Growth Empirics

Abstract

Using a newly constructed data panel on U.S. locality attributes, this paper sketches four sets of empirical facts on economic growth across U.S. counties. A first set of facts focuses on the time series and cross-correlation properties of local economic growth as measured by net migration, per capita income growth, and housing price growth. A second and a third set of facts focus on the geographical correlates of local growth over the 20th century and the non-government correlates of local growth over the period 1970 to 1990. A fourth set of facts focuses on the government fiscal policy correlates of local growth. Local economic growth from 1970 to 1990 is strongly negatively correlated with financial measures of initial local government size. This negative correlation is extremely robust across alternative specifications; an extensive set of control variables eliminates any obvious omitted variable bias; there is no indication of reverse causality; and the result is not driven by the elderly. Controlling for local government size, local growth is positively correlated with expenditures on elementary and secondary school education; it is negatively correlated with the percent of local tax revenue derived from personal income and selective sales taxes. A neoclassical model of local growth provides a framework for interpreting these correlations.