Characteristics of Low-Wage Labor Markets: A Stopping Rule in Cluster Analysis Based on Maximum Intracluster Distance

摘要

The study tests the segmented labor market theory by providing an objective method of assigning jobs to sectors; as contrasted to methods where sectors were defined using the characteristics which were to be examined, or where occupations were subjectively assigned to sectors. Jobs are defined as a 2-digit industry-occupation group. Eight job traits are examined, and using regression analysis on the Panel Study of Income Dynamics, estimates are generated from the standardized marginal impact of each job on each trait. Cluster analysis is applied based on these traits to group the jobs into two or more groups. The jobs cluster into one large group and a small group containing such diverse occupations as physicians and secretaries in mining industries. However, there is no significant difference in the grouping and in one large group containing all jobs. There is no evidence in the data that jobs break down into two or more distinct sectors and, thus no evidence for the dual labor market theory. Data is examined from the second of the wave of the Employment Opportunities Pilot Project. The study tests whether traits of workers and the EOPP program to aid job search influence the worker's time on the job. A hazard model is applied to the analysis of job duration employing several diverse explanatory variables. Most of the variables have significant impacts on the probability of a job transition EOPP had some impact on job duration.