We explore how regret influences strategic interaction and risky choice. Regret is captured by the payoff gap between what a player actually gets and what he believes he would have gotten had he chosen differently. Ex-post beliefs are critical to that evaluation, and the modeling therefore draws on tools from psychological game theory. Our analysis uncovers several non-standard implications. From a technical viewpoint, predictions depend in novel ways on information structure across end-nodes, assumptions regarding the precise nature of chance moves and mixed strategies, and the order in which play proceeds. From an applied viewpoint, regret can have powerful impact in a variety of economic settings including education, delegation, gambling, provision of public goods, and market entry.
Submitted
Prediction markets are designed to aggregate dispersed private information into prices. However, price informativeness becomes less clear when traders have intrinsic state-dependent preferences over the event being forecast. We report a laboratory prediction-market experiment with state-contingent securities and private signals about the state. For a subset of traders, one state triggers a psychologically undesirable task, whether or not they own securities, inducing a preference against that state. We vary whether these traders form a minority or a majority in the market. We find that such preferences impair information aggregation when such traders constitute a majority. By contrast, the effect is limited when these traders are in the minority. The results identify a channel through which non-monetary stakes can undermine prediction-market accuracy.
Submitted
This study investigates the reaction of workers to employer-sponsored general training that provides skills useful not only in the incumbent employer but also in other firms in the industry. While previous research has focused primarily on workers' responses to wage renegotiation, our work extends this understanding by exploring an additional dimension---workers' discretionary effort beyond their job duties, which is not verifiable. We conduct a laboratory experiment to observe workers' responses in such an effort to different training intensities. We find that workers generally increase their discretionary effort in response to general training, regardless of whether it is employer-sponsored or mandated. Moreover, the employer's intention behind offering training influences both effort and workers' renegotiation responses. Additionally, in settings where workers have the opportunity to penalize employers, they exhibit such behavior, though employer-determined higher training intensities mitigate the extent of such responses.
This paper studies the advance-purchase game when a consumer has belief-based loss-averse preferences, introducing a novel perspective by incorporating reference updating. It demonstrates that loss aversion increases the consumer's willingness to pre-purchase. Moreover, the paper endogenizes the seller's price-commitment behavior in the advance-purchase problem. The analysis reveals that the seller will commit to his spot price even with no obligation to do so, a behavior previously assumed in the literature.
Social-information messages are widely used to influence norm perceptions and norm-relevant behavior, but their effects may depend on what they report and how they are disclosed. This paper develops a belief-based framework connecting personal values and perceived social norms through inference about a latent appropriateness standard. The framework shows how messages affect perceived social norms, with potential implications for subsequent behavior, through two mechanisms: First, personal-value and perceived-norm statistics encode previous participants' private information differently because perceived-norm reports also incorporate an inference about others' assessments. Second, relative to private disclosure, public disclosure also affects recipients' beliefs about others' assessments, as they expect other group members to update using the same statistic. These mechanisms imply that, for equally sized changes on the same scale, perceived-norm statistics induce larger shifts in current perceived norms than personal-value statistics, and that public disclosure induces larger marginal responses than private disclosure. Thus, the effect of a norm-relevant message can depend not only on the value it reports, but also on how the reported statistic is constructed and who is known to observe it.