This story’s
analysis of the effects of job-retraining in
Janesville, Wis., is the first in the United States that has examined this
question using data since the recent recession made jobs harder to find.  It is patterned after a few older
studies elsewhere that used similar methods to identify dislocated workers who
went back to school, examine how much they were working and earning afterwards,
and compare them with a group of dislocated workers who had not been retrained.
The analysis was performed in collaboration with two labor economists, Kevin
Hollenbeck, a senior economist at the Upjohn Institute for Employment Research
in Kalamazoo, Mich., and Laura Dresser, associate director of the Center on
Wisconsin Strategy at the University of Wisconsin-Madison.

We focused
on students who retrained at Blackhawk Technical College, a two-year,
state-supported institution that attracts the vast majority of laid-off workers
in south central Wisconsin who go back to school. Several kinds of raw data
went into the analysis. The Wisconsin Department of Workforce Development
provided two datasets. First, to identify people who were unemployed, we used the
department’s records of unemployment claims from the summer of 2008 to the fall
of 2011, for residents of Rock County, WI (Janesville is the county seat) and
neighboring Green County. Most of Blackhawk’s students come from those
counties, and we wanted to be able to compare the students to other jobless people
in the same places. The second dataset from the department contained Unemployment
Insurance wage records, a kind of data kept by every state of each employee’s
wages that all employers are required to report. The records show quarterly
earnings–in other words, what the employee is paid over three months. We used
wage records for the same two counties. We also used information from
Blackhawk, provided by Michael Gagner, the college’s director of institutional
effectiveness.  The Blackhawk records
consisted of all students who enrolled in credit programs between the summer of
2008, when large number of jobs began to disappear from Janesville, and the
summer of 2010. We stopped with 2010 so that the students in our analysis would
have had time to finish their schooling and look for a job. The student records
contained basic demographics, such as age, sex, race and ethnicity, as well as
academic information, including whether they needed remedial work, the academic
program they pursued, and whether they graduated.

None of the
records identified individuals by name. All three datasets – the
unemployment claims, wage records, and college records – originally contained
Social Security numbers. We used these to link the datasets. This
linking was performed by Matias Scaglione, a labor economist in the Department
of Workforce Development’s Office of Economic Advisers, who then removed Social
Security numbers. We identified Blackhawk’s dislocated workers primarily
by identifying students who had received unemployment benefits at some point
during the period we examined. We identified others from a questionnaire that
Blackhawk gives all new students that asks, among other things, their
employment status. Students who answered that they were “unemployed” or
“dislocated” were included in the analysis. By comparing Social Security
numbers, we made certain that no student was counted more than once.

Once we had
identified the dislocated workers who were retrained, we did several kinds of
analysis. We created a pre-recession (and pre-layoff) period of 2007 and
compared that with a “post” period of the final year for which we had
information.  In this way, we
compared how many had any wages—that is, were working for
pay—before and after they retrained. We could not identify whether people
had full-time or part-time jobs, so we divided them into “consistent workers,”
who had some earnings each quarter of the year; “intermittent workers,” who had
at least one quarter with earnings and one with no earnings; and people who had
no reported earnings. The data contained earnings only from Wisconsin, not from
any other state, but other information suggests that relatively few people in
the area have jobs elsewhere. We also compared their “pre” wages with those
afterwards. And we compared all these results with those from the group of
unemployed people in Rock and Green counties who had not gone to the college.
We also compared the academic performance of the college’s dislocated workers
with that of other students on campus at the same time. The analysis could not
fully control for potential differences in “quality” between the dislocated
workers who were retrained and the other group of unemployed people, in large
part because there was no way to gauge the education levels of the group that
did not undergo retraining.

Amy
Goldstein is a staff writer on leave from the Washington Post,
focusing on Janesville, Wis., as a microcosm of the effects of vanished
jobs on people and the places where they live. The Joyce Foundation,
Harvard University’s Radcliffe Institute for Advanced Study, the
Woodrow Wilson International Center for Scholars, the University of
Wisconsin-Madison Institute for Research on Poverty, and ProPublica
have provided support. She can be reached at [email protected].