2026 Theses Doctoral

Essays in Behavioral Economic Theory

Si, Shuhua

This dissertation presents essays in behavioral economic theory, examining how cognitive constraints and informational frictions shape behavior in individual decisions and strategic interactions. It develops theory, alongside econometric and experimental analysis, to study these forces. The first two chapters focus on complexity in individual decisions and games; the third examines how informational precision shapes strategic signaling between agents and institutions.

In the first chapter, I develop a revealed preference framework, collective rationalizability, to test for variation in comparison difficulty while incorporating heterogeneity. Standard stochastic choice models assume that the degree of randomness in choices reflects the size of utility differences, but choice inconsistencies may also reflect difficulty comparing alternatives. Recent studies estimate such comparison difficulty by fitting choice models on data collected from different subjects under a representative-agent assumption. However, pooling data while assuming homogeneous preferences can produce apparent violations of standard models driven purely by heterogeneity, even absent any variation in comparison difficulty. Collective rationalizability characterizes whether violations of standard models can be explained by comparison difficulty alone, heterogeneity alone, or require both. I then provide a uniformly asymptotically valid statistical test for collective rationalizability and apply it to two existing experiments. In both cases, heterogeneity alone explains observed failures of stochastic transitivity better than comparison difficulty, despite being the more restrictive characterization, demonstrating that the two can be conflated in aggregate data without a framework that distinguishes them.

The second chapter, coauthored with Alessandra Casella and Olivier Compte, examines how individuals with limited cognitive capacity approach games with large and high-dimensional strategy spaces. Rather than seeking an optimal strategy within the full strategy set, we argue that players consider only a restricted number of strategies. We propose a general method for restricting the strategy set: an algorithm for constructing representative subsets, or grids, of strategies, each spanning the strategy space approximately uniformly. We then model individuals as if each restricted their strategy set to a randomly chosen grid. We apply the method to a Blotto-type resource allocation game which we also bring to the lab. We find a strong mismatch between the experimental data and the unique Nash equilibrium. Predictions over sufficiently coarse grids, by contrast, closely match the behavioral regularities and dispersion present in the data.

The third chapter, coauthored with Yangfan Zhou, studies a setting where a principal decides whether to approve an agent based on a noisy signal generated by the agent. High-quality agents can produce high signals on average at lower cost, but realizations are subject to noise that depends on the screening technology's precision. We uncover a paradoxical pitfall of precision: when precision is already high, further improvements reduce screening accuracy and lower the principal's welfare. This occurs because greater precision incentivizes more aggressive signaling from low-quality agents near the acceptance threshold, outweighing the direct benefit of improved precision. We also examine how commitment power helps mitigate this pitfall.