Recalls from Unemployment

More than 40 percent of the workers who become unemployed end their unemployment spell by returning to work for their last employer.

Unemployed individuals are typically assumed to be searching for a job in an environment in which they have imperfect information about employers, and employers have imperfect information about potential employees. This lack of information, coupled with the heterogeneity of jobs and workers, slows the job-finding process. However, if a worker who separates from an employer and experiences a spell of unemployment eventually returns to work for her former employer, then much of the heterogeneity may be irrelevant. The employer and the employee already know what to expect from each other. In Recall and Unemployment (NBER Working Paper No. 19640), Shigeru Fujita and Giuseppe Moscarini show that recalls from former employers are surprisingly common. They study the implications of these recalls for individual labor market experiences as well as for the efficiency of the labor market matching process.

The authors study data from the Survey of Income and Program Participation (SIPP), and find that more than 40 percent of the workers who become unemployed end their unemployment spell by returning to work for their last employer. These recalls include both workers on temporary layoffs and permanently separated workers. More than 20 percent of recalls are permanently separated workers who had no expectation of recall.

Unemployment spells that end with recalls are shorter, by about a month on average, than other unemployment spells. In other words, the probability of finding a new job at a different employer is much lower than the average exit rate from unemployment. The individuals who are recalled have longer job tenure before separation than other unemployed individuals. Recalls are also associated with better wage changes and with much lower occupational mobility. The well-documented negative duration dependence of unemployment -- the tendency for the probability that an unemployed worker will become employed to decline with the length of the unemployment spell -- nearly disappears once recalls are excluded. These findings suggest that recalls circumvent the usual search frictions of the worker-to-firm matching process.

In a recession, the probability that an unemployed worker will be recalled by her previous firm declines, but this decline is much smaller than the drop in the probability of being hired by a new employer. Because recalls are relatively stable over the business cycle, the probability of finding a new job at a new employer is more pro-cyclical and more volatile than the probability of being recalled. After removing recalls from the data on job finding, it appears that labor market "mismatch" was considerably larger during the Great Recession, and smaller in its aftermath, than conventional estimates would suggest. Recalls were relatively steady during the Great Recession in part because new jobs were so scarce that workers remained available for recall for much longer. Recalls helped to sustain the overall hiring rate during the Great Recession, but they did not rise as much as new hiring when the economy recovered in 2010-12.

--Claire Brunel