-
“A Representation of Preference over Preferences and Choices”
Job Market Paper
Presented at: Fall 2026 Midwest Economic Theory Conference (scheduled) ·
2026 Durham Economic Theory Conference ·
2026 Workshop on Choice & Welfare (QMUL) ·
2025 BSE Summer Forum ·
2024 Asian School in Economic Theory ·
2024 PETCO poster session
I develop an axiomatic model of preferences over outcome-menu pairs, which I call choices. An evaluator values each choice made by a decision-maker (DM) for its outcome and for the quality of the choice itself, reflecting her assessment of how well it aligns with an ideal preference. My key axiom, Relativity, captures how this quality depends on the available alternatives. The representation uniquely identifies three components: the evaluator’s preference for outcomes, the ideal preference she wants the DM to follow, and her beliefs about whether a choice is costless to the DM or requires him to give up something he values. Differences in these beliefs explain why, even when two people agree on ideals, whether a choice itself is good or bad remains highly subjective. The representation also generates a tradeoff between achieving good outcomes and making good choices. It provides a foundation for analyzing rewards, punishment, regulation, and welfare when the quality of choices matters.
-
“Consistency and Strategic Disclosure in Model-based Persuasion”
Presented at: 2026 Stony Brook International Conference on Game Theory ·
2026 PETCO poster session
Persuasion often works by supplying a model through which observed data are interpreted. I study persuasion when the receiver, beyond requiring that the model fit the observed signals well, demands that it be consistent, treating the signals as independent and identical draws rather than reading meaning into patterns. I characterize the persuasion cost of consistency in terms of the entropy of the empirical distribution of the signals, so that the more the signals differ from one another, the weaker the persuasive power. I then consider a sender who privately observes the signals and discloses only part of them at the time of persuasion, before the receiver sees the rest. When the signals are not all alike, the sender can avoid the cost of consistency entirely and fully manipulate the receiver's beliefs. When they are all identical, disclosing more of them makes the signals easier to fit, but consistency limits how much this added evidence can work in the sender's favor. The optimal disclosure is either a single signal or all signals; a single signal is best only when the receiver initially reads the signal as sufficiently bad news.
-
“Preference over Indifference”
Can indifference itself be desirable? People may value another person’s indifference as open-mindedness or dislike it as a lack of commitment. I develop an axiomatic model of how people evaluate others’ choices when decision-makers can declare indifference among several options. I characterize a “no preference for indifference” condition under which declaring indifference does not by itself contribute to the value of choices. I discuss deviations from this condition.
