This paper analyzes the conditions under which the level of hate speech (expressing hostility towards racial and other minorities) in society can influence whether individuals commit hate crimes against minorities. More generally, we explore the conditions under which speech can influence behavior by revealing social attitudes. We propose a model in which potential offenders care not only about the intrinsic benefits from the crime and the expected costs of punishment, but also about the esteem conferred by like-minded individuals. The number of such individuals is uncertain, but can (in certain circumstances) be inferred from the level of hate speech. We assume that individuals trade off their expressive utility from voicing their true opinions against the costs imposed by formal and/or informal sanctions on hate speech. We characterize the separating and pooling equilibria of this asymmetric information game, and show that the costs of engaging in speech affect what views are expressed in equilibrium. Then, we specify a set of conditions where individuals have common prior beliefs, engage in Bayesian inference, and are risk-neutral in esteem under which speech is neutral, i.e. has no effect on behavior. Then, we relax these assumptions, taking into account the relevant psychological evidence, and derive the impact of hate speech on hate crime using a variety of different formulations. We conclude that those assumptions that appear to have the strongest empirical support (the correspondence bias in inference and the concavity of utility in esteem) imply that raising the costs of engaging in hate speech will deter hate crime.
Dharmapala, Dhammika and McAdams, Richard H., "Words that Kill? Economic Perspectives on Hate Speech and Hate Crimes" (2003). Economics Working Papers. 200305.