Relative cohort size and unemployment in the United States: Evidence from the regional level

Since the early 1970s it was argued in different studies that shifts from relative smaller to larger youth cohorts raise the unemployment rate (cohort crowding hypothesis). In contrast, Robert Shimer (2001) provoked a debate with his controversial result that the overall unemployment rate tends to be lower when many young people supply labor (a high proportion of young workers incentivize firms to create new jobs because younger workers undertake more search activities, which reduce the firms recruitment costs). In contrast to other studies, he uses state level instead of national level data. The present study contributes to the literature by providing a theoretical model and spatial econometric estimates. The theoretical framework considers differences in job finding and separation as well as spatial interactions. I argue that the age groups differ in their employment-related attributes (e.g., productivity, matching efficiency, and labor turnover), apart from cohort size. Considering the stylized fact that age cohort unemployment rates decline with increasing age in an unchanged order over time, independent of cohort sizes and business cycles, I argue that age cohorts matter in theory. I use two different regional data sets for the US, the original data of Shimer at the state level and a new data set at the county level. Unemployment rates are taken from the Current Population Survey (CPS) and population shares are taken from the U.S. Census Bureau. In addition, I consider different cutoffs for the division between younger and older workers. The analysis I offer, undertaken to identify the demographic effects on unemployment, provides theoretical implications and empirical findings for the US labor market. Using a spatial Durbin model, I find no significant effects at the state level. Using county level data, the estimates provide strong evidence, that (spatial) age cohort differences are important driver of the change in overall unemployment. Total short-run elasticities vary about unity while total long-run elasticities are elastic. The results provide strong evidence that current shifts in working age population age cohorts reduce overall unemployment.