Nonintegration of Regional Labor Markets: The Case of Large US Cities from February 1980 through December 1983

Two basic models of regional unemployment are tested by use of unemployment rates of US metropolitan areas over the period February 1980 through December 1983. In the first model, regional unemployment is suggested to be the region's share of disequilibrium in the national labor market. The functional integration of regions with the national economy is a fundamental requisite for symmetry in this model, and symmetry is required for the model to have any explanatory power. Unfortunately, given the preponderant lack of strong covariance found here between regional and national labor markets, such functional integration does not seem to hold in general. The suggestion in the second model tested here is that regional labor markets should be spatially integrated under neoclassical rules. Here spatial integration, as measured by the spatial autocorrelation of regional unemployment rates, is found to be periodic rather than consistent. It appears that the spatial integration of US regional labor markets is not robust in the face of volatile macroeconomic circumstances. Unfortunately, this lack of robustness results in an asymmetrical system of interregional labor markets. The major problems of the two models result from their ageographical origins.