Sectoral shifts of labor demand can have significant effects on aggregate rate and duration of unemployment, and this is known as sectoral shifts hypothesis. To mea¬sure the sectoral shifts, past studies use David M. Lilien’s dispersion measure which represents the effect of the changes in the distribution of sectoral shocks on aggregate rates of layoffs. This dissertation proposes an improved measure of sectoral shifts and tests the sectoral shifts hypothesis. It shows that, when the distribution of sectoral shocks is asymmetric, dispersion alone is not sufficient to capture the effect of the changes in distribution and, the skewness of the distribution can have a significant role in the approximation of aggregate rates of layoffs. Empirical results show a sig¬nificant effect of the skewness measure on the aggregate rate of unemployment. The results also lend a strong support for the sectoral shifts hypothesis in Lilien type and Abraham-Katz type models, which contrasts with the rejection of the hypothesis in previous studies of the Abraham-Katz type models. One concern about these empirical results is that the classical measures of disper¬sion and skewness are very sensitive to the presence of outliers and consequently the test of the hypothesis can be distorted by this presence. Strong evidence exists for the presence of outliers in the distribution of estimated sectoral shocks. Various robust measures of dispersion and skewness are computed. The sectoral shifts hypothesis is still strongly supported when the robust measures are used. This reinforces the empirical findings under the classical measures. When the mobility of workers across sectors is limited because of frictions in the labor market, workers who become unemployed due to sectoral shifts of labor demand will experience a longer duration of unemployment because of the time associated with switching sectors. Therefore, for a given rate of unemployment, a higher proportion of these workers will increase the average duration of unemployment. Empirical results show that sectoral shifts have a statistically greater effect on the average duration of unemployment than cyclical fluctuations. Sectoral shifts help explain unusual upward trends in the duration of unemployment in the 1990s.

This thesis investigates the effect that variation in employment between industries has had on the depth of recession faced by Metropolitan Statistical Areas (MSAs) in the United States. This analysis is limited to the previous two national recessions. I use regression analysis to find that increases in variation in employment has a significant effect on the maximum increase in unemployment rate in MSAs after controlling for relevant MSA characteristics. In this framework I also find that increases in education could mitigate the negative effects of this variation. I include several other measures of depth of recession including the fall in economic conditions and length for real GDP to recover to its pre-recession levels. I find that the measure of variation is significant in explaining falls in the economic conditions, but not so in explaining the length it takes for each MSA to recover its real GDP.