Abstract
We study labor market discrimination in the context of the US tech industry. Using a simple theoretical model, we show that blind hiring during the resume screening phase can result in employers recruiting more productive employees, regardless of their awareness of the applicants' group identity during interviews. We suggest a new two-step experimental approach that enables researchers to distinguish observed discrimination into either taste-based discrimination or statistical discrimination resulting from inaccurate beliefs.
In this study, we aim to test the following hypotheses:
1. Employers' subjective beliefs about the productivity of workers vary by race and gender.
2. Employers' subjective beliefs regarding the distribution of productivity by race and gender differ from the actual distribution of productivity by race and gender collected from UM students' coding test.
3. Job applicants from the discriminated group, who are hired by discriminatory employers, tend to have, on average, higher productivity compared to job applicants from the non-discriminated group, who are hired by discriminatory employers.
4. Blinding information about race and gender leads to changes in the average productivity as inferred by employers.
5. Blinding information about race and gender results in changes in the composition of hired workers.