« Back to Results

Economic Theory

Paper Session

Sunday, Jan. 4, 2026 10:15 AM - 12:15 PM (EST)

Philadelphia Convention Center, 201-C
Hosted By: American Economic Association & Committee on the Status of Women in the Economics Profession
  • Chair: Aislinn Bohren, University of Pennsylvania

Statistics Speak, but Anecdotes Stick

Arjada Bardhi
,
New York University
Nina Bobkova
,
Rice University

Abstract

This paper develops a unified framework to compare the persuasive power of two forms of evidence—a population-wide statistic versus an illustrative individual anecdote—when a policymaker seeks to secure just enough votes in a referendum. In our model, the policymaker commits in advance to reveal either the exact average impact of a policy across all citizens or the individual impact on one highlighted individual. Impact across citizens is correlated according to a Gaussian-mixture similarity function. Citizens, who are fully Bayesian, then revise their own expected policy impact by weighting the chosen signal according to either how representative they are in this population in the case of the statistic or how similar they are to the highlighted individual in the case of the anecdote. Good news from a statistic makes the median citizen more enthusiastic and echoes more steadily across the citizenry, whereas good news from an anecdote tends to have a local effect only. However, the statistic is more predictable than the anecdote. We first show that with a single anecdote, it is always best to feature the median citizen. The optimal form of evidence is the one most tightly aligned with the pivotal voter's own inference. Our main result establishes that the anecdote is optimal when only a narrow band of support is required for the policy to be approved, whereas a statistic is optimal if broad consensus is required for approval. The more similar the citizenry is---that is, the lower the polarization around the policy---the more the policymaker is likely to use an anecdote for persuasion. The ability to randomize over highlighted citizens never benefits the policymaker, and as the number of anecdotes grows, their joint effect converges to that of the statistic.

Screening Knowledge with Verifiable Evidence

Sulagna Dasgupta
,
University of Bonn
Zizhe Xia
,
University of Chicago

Abstract

A principal seeks to screen an agent based on his demonstrable knowledge of a subject matter, modeled as a binary state. The agent learns about the state through two kinds of opposing verifiable signals, each kind providing evidence in favor of one of the states. A high quality agent is more likely to possess evidence which is greater in both quantity and accuracy, than a low quality agent. In a symmetric setting, we show that under the optimal test, regardless of whether the agent can predict the state correctly, he is passed if his total amount of evidence provided is sufficiently high and failed if it is sufficiently low. Conditional on providing intermediate levels of evidence, the agent is passed based on a simple True-False test -- i.e., if and only if he gives the correct answer. Consequently, for intermediate levels of quality sensitivity of the principal, the optimal test is the simple True-False, which makes no use of verifiable evidence, even though it is available.

Early Voting and Late-Election Information

Kayleigh McCrary
,
University of Richmond

Abstract

Convenience voting (any form of voting that does not take place on Election Day at one's precinct) offers voters a "low-cost" method of voting, but at a price: those choosing to cast a ballot early forfeit their ability to incorporate late-election information into their vote. Given this information can matter for election outcomes, normative questions arise: from an ex-ante point of view, would society benefit from a wider availability of early voting, or is early voting already too available? To answer these questions, I develop a model in which voters choose whether to vote early, late, or not at all, and where both information and the realized cost of Election Day voting affect whether a particular voter votes or not. I find that ideologically moderate voters are less likely to vote early than those with more extreme policy preferences, but that, given a sufficiently low cost of early voting, these swing voters can be incentivized to vote early, creating an externality when they do so. In particular, swing voters who vote early lose the opportunity to "switch" their vote to the beneficiary of late-election information on Election Day. This vote could be decisive, benefiting the electorate as a whole. However, in the case that Election Day voting costs are correlated with ideological preferences, I show that early voting is welfare-improving. In this case, early voting decreases the cost of voting for one group relative to the other, ensuring that turnout is representative of the electorate.

Dynamic Disclosure with(out) Timestamps

Beixi Zhou
,
University of Pittsburgh
Aaron Kolb
,
Indiana University

Abstract

We study the role of timestamps in a dynamic disclosure game with an evolving state. At a random date, an agent obtains one piece of hard evidence of a hidden, binary state that evolves via Markov switching. When evidence carries a timestamp, the agent discloses good evidence immediately and bad evidence after a deterministic, timestamp-dependent delay. Without timestamps, equilibrium features a stockpiling phase in which the agent delays disclosure, followed by a purging phase with stochastic purging of evidence and then a phase of immediate disclosure of any good evidence; under some conditions, bad evidence is never disclosed. Timestamps prevent the agent from pretending good evidence is fresh and allow the agent to prove that bad evidence is old, thereby accelerating good-evidence disclosure and facilitating bad-evidence disclosure.

Discussant(s)
Doron Ravid
,
University of Michigan
Frank Yang
,
Harvard University
Alice Gindin
,
Middlebury College
Elliot Lipnowski
,
Yale University
JEL Classifications
  • C7 - Game Theory and Bargaining Theory