Date of Award

5-2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Economics

Committee Member

Dr. Curtis Simon, Committee Chair

Committee Member

Dr. Scott Baier

Committee Member

Dr. Michal Jerzmanowski

Committee Member

Dr. Andrew Hanssen

Abstract

The first chapter of this dissertation examines the phenomenon of labor reallocation at the level of industry during periods of recession and recovery. Whether permanent shifts of industries’ labor demand curves contribute to cyclical unemployment remains a highly controversial issue. With a focus on the timing of recessions and recoveries, I evaluate the empirical support for two competing explanations of cyclical unemployment: the pure sectoral-shifts hypothesis and the pure aggregate disturbances hypothesis. Although recessions are considered times of low aggregate demand, they also coincide with remarkably large permanent changes to the sectoral distribution of labor demand. Strikingly, I show that, for declining sectors, the majority of jobs destroyed during a recession are lost permanently and do not reemerged during the subsequent economic recovery. This fact contradicts the idea that recessions are solely periods of weak aggregate demand. In addition, the observation that employment gains in expanding sectors tend to concentrate during economic recoveries casts doubt on a pure sectoral-shifts story. The findings suggest that, on their own, these two hypotheses provide incomplete explanations for the relatively high unemployment observed after business cycle downturns. The second chapter builds on the idea of the first by studying the impact of changes in local industry labor demand on unemployment transitions. Most research examining the relationship between local labor market conditions and unemployment summarize these conditions in the form of Bartik's (1991) index. Such studies overlook an important component of local demand conditions, namely, the fortunes in workers' prior industries. If labor is perfectly mobile between sectors, the performance of an unemployed person's prior industry should not affect their job finding prospects after controlling for local aggregate labor demand. In reality, however, jobseekers may face significant economic costs associated with switching industries. For example, a worker displaced from automobile manufacturing may incur a substantial wage reduction from switching to the retail sector due to the loss of industry-specific human capital. To the extent that a worker is tied to their previous line of production, it becomes useful to distinguish between labor demand in their prior sector and the level of aggregate labor demand within their locale. By combining several U.S. datasets spanning the years 2003-2015, I find that a 10-percentage point increase in labor demand within an individual’s prior local industry increases the probability of exiting unemployment by 2.7-percentage points after controlling for aggregate demand. Moreover, I document that the magnitude of this effect increases with a jobseeker’s age and level of educational attainment. These findings suggest that jobseekers may be more vulnerable to demand conditions in their prior industries than previously appreciated. The third chapter, written jointly with Mallika Pung, examines the effect of local labor demand shocks on the earnings losses of U.S. displaced workers. Specifically, we ask whether changes to local labor demand help explain the magnitude of displacement-related earnings losses after controlling for the national business cycle. We combine data from Displaced Workers Surveys (DWS) and the Bureau of Economic Analysis (BEA) to construct a novel dataset linking displaced workers to measures of local labor demand. Using a generalized difference-in-differences approach, our estimates suggest that a one standard deviation increase in local labor demand reduces the mean earnings loss associated with job displacement by 14 percentage points, after controlling for national business cycle fluctuations. Using a quantile regression (QR) approach, we document significant heterogeneity in the effect of predicted demand shocks across the earnings loss distribution. The effect of labor demand is strongest in the lower region of the distribution with little or no effect in the upper region. In contrast, we find that local labor demand is an important determinant of earnings losses across the entire distribution of income.

Share

COinS