The pure logic of discrimination

The paper expounds a simple non-cooperative game, which can model discrimination in market and non-market transactions. The model has two equilibriums where individuals do not discriminate and several where they do. One non-discriminatory equilibrium dominates in the sense of Pareto all discriminatory equilibriums. The second non-discriminatory equilibrium is dominated by all other equilibriums. While there is no discrimination in the first-best outcome, the introduction of discrimination can thus be Pareto-improving in some circumstances. For instance, in the model, a move from an equilibrium where they face discrimination to another equilibrium where they discriminate too can sometimes make the members of a group better off, without making anyone else worse off. Another implication of the model is that interactions between officials of the state and ordinary citizens generate a greater number of potential patterns of discrimination than interactions between ordinary citizens.