This paper develops a statistical model for measuring spatial interactions when estimating macroeconomic regimes and regime shifts. The model is applied to study the contagion and propagation of recessions in small regional economies in the United States from 1990 to 2015. The empirical analysis identifies two geographical concentrations (or clusters) where small regional economies were affected by recessions in similar ways. These clusters are interpreted as groups of regions that are potentially at-risk to collective economic distress, which is useful for national and regional policy makers. The first identified cluster is characterized by regional economies with important roles in the financial sector, while the second cluster is characterized by the oil and gas extraction sector. The empirical findings uncover an important propagation dynamic that would be overlooked if one were to apply the model without the spatial extension developed in this paper. Specifically, the evidence shows significant spatial spillovers between small regional economies, meaning that shocks to regions are expected to be higher, when shocks to neighboring regions are high on average. The magnitude of this effect is amplified for the period spanning and following the Great Recession.
Bayesian statistics business cycles endogenous clustering regime-switching regional economic analysis spatial econometrics time series econometrics