Sectoral shift theories of unemployment: evidence from panel data

This paper examines the response of sectoral real wages and location probabilities to oil price shocks using U.S. micro-panel data (the National Longitudinal Survey of Young Men). The goal is to determine whether the observed response patterns are consistent with so-called sectoral shift theories of unemployment. These theories predict that shocks that change sectoral relative wages should increase unemployment in the short run and lead to labor reallocation in the long run. Consistent with these predictions, the oil price changes of the 1970s resulted in substantial movements in industry relative wages and significant reallocation of labor across industries, while both oil price increases and decreases resulted in short run increases in unemployment. However, equilibrium sectoral models imply that real shocks that change relative wages across sectors should induce flows of labor into those sectors where relative wages rise. In fact, real oil price shocks are found to have substantially reduced respondents location probability in the construction industry, which had a wage increase relative to all large industries. The industry with the greatest increase in employment share was services, which had among the greatest wage declines. These are clear contradictions of the predictions of equilibrium sectoral models. Nevertheless, a more general class of models where both relative wage movements and quantity constraints generate labor flows appears to be quite consistent with the data.