Discrimination Between Religious and Non-Religious Groups: Evidence from Marking High-Stakes Exams

While religions frequently preach preferential treatment of fellow believers, the magnitude and economic implications of religiosity-based discrimination are largely unknown. Religiosity is often confounded with ethnicity, but it varies even within ethnicities and religious denominations. It is also seldom observed in administrative data. This paper exploits a setting that avoids these limitations. We analyse grading decisions in national matriculation exams in Israel, exploiting unique features that reveal student religiosity to the graders, and grader religiosity to the researcher. We find evidence of in-group bias between religious and non-religious groups, but in our setting this effect is very small. There seem to be two main reasons. First, religious in-group bias is limited to male graders only. Moreover, patterns of bunching in the grade distribution suggest the bias is primarily due to the religious - rather than secular - men. This is a small fraction of the grader population. A second potential reason is that many graders live in integrated communities. Our evidence suggests that living and working in close proximity to people with different levels of religiosity attenuates discrimination.