José G Caraballo-Cueto


An accurate estimation of GDP for sub-state divisions is needed for studies on the overall size and growth of smaller economies, assessment of the industrial composition by subdivisions, and the impact of economic policies, among other reasons. In this paper, I review two popular methods to infer GDP in counties, both of which rely heavily on payroll, without controlling for labor productivity differences between subdivisions. By considering labor productivity heterogeneity across counties, I propose a new methodology that appears to provide a better map of the GDP of counties. For forecasting, my approach uses payroll statistics, but to a lesser extent. The new method is tested with data from Puerto Rico, a suitable jurisdiction given its significant heterogeneity between counties. I am able to map changes in economic geography associated with changes in the state’s economic structure. Like many other jurisdictions, Puerto Rico has undergone a deindustrialization process that ended up concentrating GDP in one geographical cluster.