Estimating the returns to community college schooling for displaced workers (Jacobson et al. 2004)

The authors matched data from unemployment insurance wage records to community college records to create a database. Unemployment insurance data covered 1987 to 1996 and included earnings, hours worked, hourly wages, employers, and employees’ demographics. The community college data set covered 1989 and 1995 and included information on students’ demographics, courses enrolled and completed, and grades. The authors identified those workers in Washington State ages 22 to 60 with at least three years of paid work experience who were laid off from 1990 to 1994 and had a valid unemployment insurance claim; in total, 97,000 workers fit these criteria. The authors then constructed a treatment group consisting of about 16,000 of those workers who had earned at least one college credit at one of Washington’s 25 community colleges by 1996; the remaining people in the sample comprised the comparison group.

The authors estimated a series of regression models comparing the outcomes of displaced workers who earned community college credits with those who did not. Regression models included time, worker, and time-worker fixed effects.