Modeling Social Preferences: A Generalized Model of Inequity Aversion

Abstract

Taking note of the wide variety and growing list of models in the literature to explain patterns of behavior observed in laboratory experiments, this paper identifies two tests, the Variety Test (ability of a model to explain outcomes under variety or alternative scenarios) and the Psychological Test (ability of a model to conform to psychological intuition), that can be used to judge any model of other-regarding behavior. It is argued that for a mathematical model to qualify as a social welfare function, it must simultaneously pass the two tests. It is shown that none of the models proposed to date passes these two tests simultaneously. The paper proposes a generalized model of inequity aversion which parsimoniously explains interior solution in the dictator game and dynamics of outcomes in other games. The paper postulates that one’s idea of equitable distribution is state-dependent, where the state is determined by psychological and structural parameters. The state could be fair, superior or inferior. Individuals in a fair state have zero equity-bias and split the pie evenly. Those in a superior (inferior) state have positive (negative) equity-bias and value more (less) than fair distribution as equitable distribution. Given psychological tendencies of an individual, every experimental design/structure assigns one of the three states to players which lead to individual-specific valuation of equity. Prediction about outcomes across different experiments and designs can be made through predicting their impact on equity-bias. All aspects of an individual’s behavior, such as altruism, fairness, reciprocity, self-serving bias, kindness, intentions etc, manifest themselves in the equity-bias. The model therefore is all-encompassing.