When minimum wages increase, employers may respond to the regulatory burdens by substituting away from disadvantaged workers. We test this hypothesis using a correspondence study with 35,000 applications around ex-ante uncertain minimum wage increases in three U.S. states. Before the increases, applicants with distinctively Black names were 19 percent less likely to receive a callback than equivalent applicants with distinctively white names. Announcements of minimum wage hikes substantially reduce callbacks for all applicants but shrink the racial callback gap by 80 percent. Racial inequality decreases because firms disproportionately reduce callbacks to lower-quality white applicants who benefited from discrimination under lower minimum wages.
*This project has been supported by a grant from the W.E. Upjohn Institute Early Career Research Award
Concealing candidate identities during evaluations, or "blinding", is often proposed as a tool for combatting discrimination. This paper studies how blinding impacts candidate selection and quality, and the forms of discrimination that drive these effects. I conduct a natural field experiment at an academic conference, running each submitted paper through both blind and non-blind review. Two years after the experiment, I collect measures of paper quality—citations and journal-weighted publication status—for each paper and link it to the experimental data. I find that blinding significantly reduces scores for traditionally high-scoring groups: applicants who are more senior (non-students), from top ranked institutions, and male. Blinding consequently alters the composition of applicants who are accepted to the conference. Despite these compositional changes, blinding does not worsen the conference's ability to select high-quality papers, and this is likely not driven by direct effects of the conference. I develop a model of evaluator discrimination that allows me to rationalize these effects and decompose non-blind disparities into two distinct forms of discrimination: accurate statistical discrimination and bias.