DURATION OF UNEMPLOYMENT SPELLS ENDING IN WITHDRAWAL: THEORY AND EVIDENCE (SEARCH THEORY, DISCOURAGED WORKERS, SORTING HYPOTHESIS, GROSS FLOW DATA)

MARTI SUE BOGART, Purdue University

Abstract

Labor economists have spent considerable time studying discouraged workers. One issue which has not received emphasis in the literature is the length of time spent unemployed before dropping out of the labor market. This study focuses on the theoretical and empirical aspects of this issue. The theoretical basis for considering the factors affecting the length of the search horizon (i.e. the duration of unemployment ending in withdrawal) is provided by Pissarides' job search model. The model predicts that the search horizon will be shorter for women compared to men, for nonwhites compared to whites, for old people compared to younger people, and that the search horizon will fall as economic conditions deteriorate. Gross flow data, together with data on unemployed persons by duration, sex, race, and age published in Employment and Earnings, are used to construct estimates of a search horizon in order to test the hypotheses from the Pissarides' model. The first measure, T, focuses on the UN flows by duration of unemployment and is calculated as a weighted average of the time spent unemployed before leaving the labor market for those individuals who drop out between two months. This statistic is shown to be an inappropriate measure of the average search horizon. The second measure, EDQ, is the expected duration of unemployment ending in withdrawal based on the conditional probability of dropping out given that no job is found, q*. All of the hypotheses from the Pissarides' model with respect to how demographic variables affect the length of the search horizon are strongly supported by the empirical evidence on EDQ. The cyclical behavior of EDQ is not, however, as supportive of the model. Disaggregation by duration of unemployment of the data used to estimate the probability of finding a job, p, and the conditional probability of withdrawal, q*, is also considered. The results indicate that p declines as unemployment duration lengthens, but q* displays a u-shaped trend over duration.

Degree

Ph.D.

Subject Area

Labor economics

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS