A theory of perceived discrimination

We develop a model in which individuals compete for a fixed pool of prizes by investing effort in a contest. Individuals belong to two separate and identifiable groups. We say that the contest is discriminatory if a lower share of prizes is reserved for one group than for the other. We show that it can be difficult for an observer to detect the presence or absence of discrimination in the contest, as both regimes can be observationally equivalent. In particular, one group's belief that it is allocated a lower share of prizes than the other group can be consistent with observed data even if no such group quotas actually exist. Conversely, the belief that the contest does not discriminate can be consistent with data when, in fact, discrimination exists. Incorrect beliefs will therefore not be revised, as the contest generates no evidence to the contrary.