Unemployment Cycles: Aggregate Disturbances And Sectoral Shifts

This thesis consists of three essays, all of which examine the relative importance of aggregate disturbances and sectoral shifts in the explanation of unemployment cycles.;Chapter 1 of the thesis uses a search-theoretical framework to analyze a multi-sector economy in which labor productivity is subject to aggregate and sectoral shocks. After a critical review of the existing literature in this area and questioning the previous methods of shock decomposition into aggregate disturbance and sectoral shift components, this paper develops a new method for partitioning the effects of these two shocks on unemployment. It is argued that the mean and the dispersion of industrial stock prices, wages or labor productivity can be used to measure aggregate disturbances and sectoral shifts respectively.;Chapter 2 investigates empirically the theoretical results developed in Chapter 1. The mean and the dispersion of industrial stock prices, wages and labor productivity for both the U.S. and Canadian economies are computed to represent aggregate disturbances and sectoral shifts respectively. Industrial Solow residuals are used to approximate labor productivity changes. Using vector autoregression and analysis of variance techniques, this chapter demonstrates that aggregate disturbances consistently explain at least twice as much of the unemployment variations as sectoral shifts across all data sets. In the U.S., the aggregate disturbances contribute 15 to 30 percent of unemployment changes while the sectoral shifts only account for 5 percent. In Canada, the effect of sectoral shifts on unemployment is negligible while the aggregate disturbances have a significant 40 percent explanatory power.;Chapter 3 further enhances the importance of aggregate disturbances on unemployment by studying the empirical fact that aggregate unemployment is highly correlated with the dispersion of industrial unemployment rates in both the U.S. and Canada. It is found that this high correlation is almost entirely attributable to the effects of aggregate disturbances. Due to the fact that industries have different cyclical sensitivities and that these cyclical sensitivities are positively correlated with industries\u27 mean unemployment rates, the industrial unemployment dispersion increases during an economic contraction and decreases during an expansion, and hence the positive correlation between the dispersion and aggregate unemployment. Several tests are performed to confirm this conclusion