Standard theories of labor market search
predict that workers should search less when the returns to search are low,
yielding the counterfactual prediction that labor market participation and
other measures of search intensity should be strongly procyclical and unemployment
should be acyclical or even procyclical. I argue that this is a consequence of
how search intensity is modelled. In a discrete time setting, I model search
intensity as a worker's choice of the number of simultaneous applications to
make. My main result is that when the cost of making an application is small, a
worker who has at least an eighty percent probability of getting a job responds
to an adverse shock by increasing his search intensity. Workers who are less likely
to get jobs become discouraged, reducing their search intensity.