In Living Color: Does In-Person Screening Affect Who Gets Hired?
79 Pages Posted: 9 Apr 2020
Date Written: March 31, 2020
When hiring new workers, employers often screen large numbers of written applications before selecting a subset for more costly, in-person interviews. A large literature suggests that information frictions lead to screening on imperfect quality signals - e.g., educational pedigree and network-based referrals - and that these practices can perpetuate labor-market inequities. In theory, a reduction in the cost of in-person screening could lead to improvements in both efficiency and equity by reducing the use of blunt signals that disadvantage certain groups. We test this hypothesis by studying the introduction of a labor-market intermediary, the Accounting Rookie Camp (ARC), that greatly facilitated in-person screening in the labor market for PhD accountants. Using a novel data set with information on new PhDs, recruiters and market outcomes over 11 years, we estimate difference-in-difference models that leverage variation in the timing of ARC adoption. We find that degree program rank and adviser connectedness are strong predictors of initial job placements, but that their impacts are significantly reduced by participation in ARC. The results suggest that the historical returns to program reputation and adviser networks were driven partly by their signaling values, which were reduced by new the information channels created by ARC. They also indicate that in some respects, ARC adoption helped foster greater diversity in hiring. At the same time, we find no evidence that ARC reduced existing disparities in placements by gender and only weak evidence that it benefited under-represented minorities. Finally, using names to predict nationality and native language, we find that ARC led to worse placements for candidates whose native language is very different from English.
Keywords: Job Matching, Screening, Signaling, Hiring, Networks, Imperfect Information, In-Person Interview, Academic Labor Market
JEL Classification: D83,J7,J23, J44, M510
Suggested Citation: Suggested Citation