Three Essays in Behavioral Economics on Cheating

Abstract

Information asymmetry lies at the heart of many economic and business interactions. When one party possesses private information, the incentive to exploit this advantage can give rise to cheating behavior. Classical economic theory assumes that self-interested agents will seize such opportunities to maximize their own benefit. Yet over the past decade, a growing body of experimental evidence suggests otherwise: individuals often cheat only marginally or refrain entirely, even when it is in their financial interest. This dissertation integrates insights from behavioral economics, experimental economics, and game theory to investigate how psychological motivations shape cheating behavior in settings with private information. Across three chapters, I examine different environments in which individuals face opportunities to cheat others. By incorporating preferences such as cheating aversion, social image concerns, and guilt aversion, I shed light on when and why people resist the temptation to cheat, how institutional settings mediate these decisions, and what these findings imply for policy design. In the first chapter, I examine how psychological preferences – Cheating Aversion (CA) and Perceived Cheating Aversion (PCA) – shape outcomes in credence goods markets. These markets are characterized by an information asymmetry: experts possess more knowledge than consumers about the appropriate service or treatment, creating opportunities for fraudsuch as overtreatment or overcharging. I study these markets under two institutional environments: (i) verifiability, where consumers can verify whether the prescribed treatment was actually provided, and (ii) liability, where experts are held accountable for adequate treatment. To address the challenge of defining cheating, I develop a social-norm-based framework that provides a benchmark for assessing the extent of dishonest behavior. I find that when prices are fixed, incorporating CA and PCA reduces unethical behavior and enhances efficiency in both institutional settings. However, when experts have monopoly power over pricing, the outcomes diverge: low sensitivity to CA and PCA can be advantageous for consumers, but high sensitivity may harm them under the verifiability condition. In contrast, under the liability condition, psychological concerns have no effect on consumer utility, and high sensitivity benefits only the expert. This chapter contributes to the credence goods literature by investigating how psychological factors related to cheating influence expert behavior, consumer surplus, and market efficiency. Additionally, the formal social-norm-based framework for defining cheating can be applied to a wide range of contexts beyond credence goods markets. In the second chapter (joint work with Martin Dufwenberg), we experimentally investigate how externalities influence cheating behavior in reporting tasks. We build on the design of Fischbacher and F¨ollmi-Heusi (2013), in which participants roll a die privately and self-report the outcome. In our adaptation, we introduce a “regular” audience: each participant is paired with another whose payoff is affected by the report but who does not observe the actual die roll. We manipulate the externality structure across three treatments: negative, zero, and positive. Given the presence of a regular audience, an important empirical question is whether experimenter observability influences reporters’ behavior. If it does not, then individual-level cheating can be identified without altering participants’ decisions. To test this, we compare two conditions: an observed condition, in which the experimenter observes the die-roll outcomes, and an unobserved condition, in which the experimenter does not. Our experimental findings show that participants were more likely to cheat when theirdishonesty also benefited others, highlighting the importance of incorporating externalities into models of cheating. Furthermore, we find that when a regular audience is present, experimenter observability has no significant e?ect on participants’ behavior. These results allow us to use the observed data from the zero-externality treatment to test competing theories, such as those proposed by Dufwenberg and Dufwenberg (2018) and Gneezy et al. (2018). Our analysis does not reject either model. To further distinguish between them, future work may require eliciting participants’ strategies explicitly. The third chapter builds on the behavioral themes of the previous chapters by develop-ing a theoretical model of cheating behavior in environments where misreporting imposes negative externalities on others. I construct a belief-dependent utility model in which a reporter’s payo? depends not only on the monetary benefit from misreporting but also on two psychological costs: perceived cheating aversion (PCA), the discomfort from being seen as dishonest, and guilt aversion (GA), the distress from letting others down relative to their expectations. I characterize a class of sailing-to-ceiling equilibria in which individuals lie partially, avoid downward lies, and engage in uniform overreporting. Comparative statics reveal that the effectiveness of externality-based deterrents depends on the relative strength of PCA and GA: when PCA dominates, increased externality reduces cheating, but when GA dominates, this effect may reverse. This model offers theoretical foundations for designing mechanisms that discourage cheating in real-world contexts where others are affected. Together with the first two chapters, this essay contributes to a unified understandingof how psychological motivations, institutional structures, and informational environments jointly shape cheating behavior.