ILR Review

The Impact of Unemployment Insurance on the Search Process

A SERIOUS coding error in the data used in our recent article published in the April 1979 Review has been pointed out by Joe Stone of the Bureau of Labor Statistics. Our paper proposed a test of Mortensen's hy-pothesis that both future expected unem- ployment insurance benefits and benefits re-ceived during a current unemployment spell affect an individual's search intensity.

Regression 1, which remains unchanged, still provides no support for Mortensen's proposed effect of unemployment insurance benefits to be received during the subse- quent unemployment spell on the current job-search intensity of the unemployed. Regression 3, which remains unchanged, still indicates a distortion in the search process-in particular in the methods of search chosen-for current recipients of unem- ployment insurance benefits and the un- employed who are eligible and have applied for these benefits.

The error affects the results of the estimnation of Equation 2. It occurred because unemployed individuals who were eligible and had applied for benefits were assigned zero weeks left to receive these benefits rather than the maximum allowable duration of benefits according to the individual's state of residence. As a result, the value of unem- ployment insurance benefits for these individuals was inadvertently set equal to zero. Yet these individuals, other things equal, were shown in our original study to have a measured job-search intensity 74 percent higher than individuals currently receiving unemployment insurance, a difference related to the time involved in the ap- plication process rather than to actual job- search efforts.

Reestimation of Equation 2 controlling for this effect and correcting for the measure- ment error in the value of unemployment benefits results in one important change. The coefficient on the value of unemployment insurance benefits, though still negative, is riot different from zero for standard significance levels. A serious consequence is that the traditional disincentive effect of unemployment insurance on search in- tensity is not supported by our test. One explanation for this finding may be that individuals with larger values of unemployment benefits have a greater incentive to overstate search intensity since such benefits are dependent on search activity.