Misperceived Social Norms: A Unified Framework (Job Market Paper)
Perceptions of others are central to behavior under social influence, yet they are often inaccurate. This paper develops a unified framework to study how various kinds of misperceptions about others impact collective outcomes when individuals are subject to distinct types of social influence. In particular, I distinguish between two types of social influence: conformity to others' actions and conformity to others' attitudes. Although often treated interchangeably, I find that misperceptions have sharply different implications across the two: under conformity only to actions, misperceptions about others' attitudes or individual actions cannot shift collective behavior, whereas under conformity to attitudes, such misperceptions can significantly impact collective outcomes. I show that these two types of social influence also yield distinct policy and welfare implications. I discuss diverse applications of my framework, including norm-based interventions, the gender gap in competitive entry and the "friendship paradox" in networks.
Presented at EWMES (2025, scheduled) Stony Brook International Conference on Game Theory (2025), CESC (2025), SAEe (2024), European Meeting of ESA (2024), HEC Economics PhD Conference (2024), Liverpool University (2024), NL in Economics Online Seminar (2024), Annual Meeting of the EPCS (2023), IMPRS-UPF BeSmart Topics Workshop (2023), NoBeC Conference (2022, poster)
Previous title: A Theory of Misperceived Social Preferences and Norms
Preferences for Information Bias with Anticipated Learning (joint with Paul O. Richter)
Individuals are frequently exposed to information they do not actively seek, such as news shared by others. This paper studies how such anticipated information shapes agents’ learning decisions. We show that agents’ preferences for biased sources depend systematically on their prior beliefs and the structure of the anticipated information. We identify two novel patterns, matching and mismatching, in which agents select sources biased towards the same or the opposite action compared to the information they expect to receive. In social environments, when agents share information with their peers, the model predicts that groups of relatively uncertain people tend to coordinate on consuming news with the same bias, especially if the degree of bias is large. In other cases, they might coordinate on choosing information with diverse biases.
Presented at Torino University (2025), HEC Paris (2025), Séminaire parisien de Théorie des Jeux (2025), EEA-ESEM (2024), EWMES (2023), SAEe (2023), University of Zurich (2023), Universitat Pompeu Fabra (2023)
Previous title: Forecasted learning
Emotional Feedback Design: Theory and Experiment (joint with Evangelia Spantidaki-Kyriazi)
Individuals often derive utility from maintaining favorable beliefs about their own ability. Everyday decisions, like effort and task choice, influence the informativeness of performance signals. Therefore, agents may distort these choices to preserve self-image, a phenomenon known as self-handicapping. In this paper, we examine whether performance feedback design can attenuate self-handicapping behavior while retaining its motivational value. We propose that introducing moderate noise into performance feedback —thereby weakening the direct link between the feedback and underlying ability— can reduce self-handicapping incentives without undermining overall motivation.
Awarded with the REACT Grant by Universitat Pompeu Fabra