Collaborative Proposal: Reference-Dependent Job Search

Project Outcomes Statement

Can behavioral economics help us better understand the job search process? In this collaborative award, Johannes Schmieder (BU) and Stefano DellaVigna (UC Berkeley) pursued this question in two ways.

First, we used data from an unemployment insurance reform in Hungary which reformed the UI system from a single level of UI for 9 months to a 2-step system, with decreasing generosity after 3 months. We document that the pattern of exit from unemployment in response to this reform is best understood with job-seekers being reference dependent: they compare the current benefit level to recent earnings levels, and the more they are on the "loss" side, the harder that motivates them to search for a job. This type of pattern produces especially high search intensity at the beginning, and around the time when benefits decrease, just as observed in the data. The model makes a set of predictions which we check in the data.

Second, we also collect data on the search intensity of job seekers in Germany using a survey. While the administrative data used in the first project has advantages, it does not provide information on the within-person evolution of search effort during an unemployment spell. To attain information on that, we use an SMS-based survey, contacting unemployed workers in Germany through a collaboration with the IAB, the relevant local agency. Twice a week, we ask the respondents how many hours they searched for a job the previous day, as well as other related questions. This survey ran over a period of 2 years (November 2017 to November 2019) to obtain a large enough sample size. In the end 7800 individuals participated over a 4 month period leading to more than 110,000 observations on individual job search effort.

The analysis of the survey data produced 3 key results: 1) search effort is flat early on in the UI spell, 2) search effort exhibits an increase up to UI exhaustion and a decrease thereafter, 3) UI recipients do not appear to time job start dates to coincide with the UI exhaustion point.