Yi Shi
Yi Shi
On the Nature of Conditional Cooperation
Abstract: In this paper, we experimentally study conditional cooperation, an instance of reciprocity that frequently arises in social dilemmas. We aim to better understand the nature of conditional cooperation and in turn the nature of reciprocity, given that existing models of reciprocity fail to explain some of the empirical regularities. Using within-subject comparisons between asymmetric sequential prisoner's dilemmas and modified dictator games, we provide clean evidence that conditional cooperation exists as a context-dependent phenomenon, distinct from altruism and inequity aversion. We then systematically examine how distinct payoff components influence conditional cooperation. In particular, our design manipulates the first-mover payoffs and the second-mover payoffs separately, an aspect that has not been explored in previous studies. We find that both payoff dimensions shape the second-mover decisions in different ways. Notably, the sacrifice borne by the first-mover, captured by their forgone payoff from defection, markedly affects the second-mover's reciprocal response, an aspect entirely absent from existing reciprocity models. Our results provide directions for extending and improving current theoretical models of reciprocity.
Kindness Matters: A Theory of Reciprocity
Abstract: A large body of experimental evidence shows that people are willing to sacrifice their own material payoffs to reward kind behavior and to punish unkind behavior, a phenomenon known as reciprocity. However, existing reciprocity models assess kindness only in terms of what the decision maker could receive, without considering the cost the other player incurs, leaving a range of experimental findings unexplained. This paper develops a model of reciprocity that addresses these limitations through two main aspects. First, we introduce a new definition of efficient strategy that successfully resolves paradoxes found in existing behavioral models. Second, we propose a novel definition of intentional kindness in which a player evaluates others' intentions by considering not only what they could receive, but also the payoff cost the other player bears in making that choice, captured through a new reference standard. Applying our model to well-known experimental games, such as the ultimatum game and the sequential prisoner's dilemma, we show that our model generates predictions consistent with experimental findings, which neither standard theory nor existing reciprocity models can fully explain.
Work in Progress
Gender bias in AI-mediated economic decisions
Restricted communication, information aggregation, and efficiency