Using AI to Expand the Job Search of Displaced Workers in the Aftermath of the Covid-19 Crisis

Regents Of The University Of Michigan - Ann Arbor, Ann Arbor MI

Investigators

Abstract

This award will support research that uses artificial intelligence (AI) and machine learning to improve matching of low skilled workers to available jobs. While the US job market recovered well from the COVID 19 pandemic disruption, the unemployment rate remains relatively high for under-represented groups even as unfilled vacancies have risen and stayed high. This suggests a mismatch between the information available to jobseekers and employers. This research will use field experimental methods to investigate whether AI-assisted algorithmic matching of skills and psychometric skills to on-line vacancies can help job seekers get better job matches. This research project has the potential to improve the functioning of the labor market for workers at the lower end of the skill distribution, hence increase employment for this group of workers. Besides providing evidence on the mechanisms through which AI and computational algorithms can be used to improve labor market efficiency, the results of this research can provide guidance on policies to increase employment, labor productivity, and economic growth. Because the research focuses on low wage workers, the results can decrease poverty as well as decrease income inequality and help establish the US as a global leader in poverty reduction policies. This project leverages data on job seeker characteristics and the requirements of jobs posted by online job search engines and use a randomized control trial (RCT) design to investigate whether assigning an AI-assisted algorithmic matching to job postings improves job matching. The RCT design has three treatment arms: (i) job offers in the job-seekers field in no particular order; (ii) vacancies in no particular order but with predicted match quality; and (iii) vacancies sorted to best match backgrounds and skills of jobseekers with predicted match quality. The control group are jobs listed in no particular order and without a match quality attached. The RCT will have a sample of 2600 job seekers, with heterogeneity across gender and space, recruited via advertisements posted on Monster.com’s social media accounts. Comparison of the first group with the control group will answer the questions of whether treatment reduces search cost hence improves matching, while comparing outcomes for the second and third arms to those of the control group and the first arm will answer the questions as to whether treatment help to lower the cost of incomplete information and improves job match. Besides providing evidence on the mechanisms through which AI and computational algorithms can be used to improve the functioning of labor markets, the results of this research project can provide guidance on policies to increase employment, increase productivity, and economic growth. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.