In this paper we derive an alternative measure for structural unemployment using a stochastic frontier analysis. This measure, by empirical design, is always less than total unemployment and it is, thus, more consistent with the theoretical description of structural unemployment than its usual interpretation as a smoothed long-run average of total unemployment. We find that our measure does not always track the long-run trends in total unemployment in the U.S. and when compared to the existing measures can produce different insights about the evolution of structural unemployment. Demographic and regional evidence provides some validation for our approach and allows us to determine how demographic and regional factors are related to the variation in structural unemployment across time and regions.
