What is keeping unemployed workers from finding employment? The role of information, job characteristics, and behavioral biases

The major constraint for existing research on job search is one of data availability. First, while many existing data sets contain information on unemployed job seekers and the jobs they find, very few data sets contain information on which jobs they applied for and even fewer contain information about other search activities, or about the vacancies that job seekers actually found and considered in the application process. Second, while data on job characteristics such as wages or occupations are often readily available, many important non-wage job characteristics, such as family-friendliness or future career prospects are not observed in the data. Finally, it is hardly ever possible to link data on job search behavior and labor market outcomes to comprehensive information on job seekers’ personality traits, their behavioral biases, and their social ties to family, friends, or former co-workers, which all can play an important role for workers’ labor market prospects.

To overcome these data constraints, our project will collect, process, and link data using a variety of sources and methodological approaches. 

First, in this project we will process data from a variety of new data sources: 

• data from UI recipients’ registrations of applied-for jobs with the public employment agencies (mandatory for Danish UI recipients since 2015) will provide information on individuals’ actual job applications. 
• data on vacant jobs posted online (on the online platform of the public employment agency, Jobnet.dk) will provide information on which jobs individuals could have applied-for.
• data on UI recipients’ online job search behavior (actual mouse clicks on the Jobnet.dk platform) will be used to measure which vacancies job seekers have successfully found. 

Second, the project involves collecting new survey data and constructing new measures from existing data: 

• a new survey of UI recipients will provide measures of job seekers’ beliefs and personality traits and the job characteristics they emphasize when evaluating jobs. 
• state-of-the-art machine learning techniques  will be applied to existing administrative data to measure important job characteristics such as family-friendliness and prospects for career advancements.

After linking information from all these data sources, we will conduct detailed analyses to answer the project’s research questions: 

• state-of-the-art econometric analysis of these novel linked data will uncover which belief biases are prevalent for different job seekers, which job vacancies they manage to find, which of these they apply for, and what the specific job characteristics are for these jobs.

Finally, to further draw lessons for practical policy-making, we will conduct a randomized trial:

• In collaboration with Danish Agency for Labor Market and Recruitment (STAR), we will conduct a randomized trial with UI recipients that tests our conclusions and their implications for policy.