Publications (and soon to be published):

Recognition Probability in Legislative Bargaining, with Natalie Lee (paper), (Published online 2023,, Journal of Experimental Political Science

In legislative bargaining, the proposer is often able to extract a greater proportion of the surplus. However, a higher likelihood of being selected as the proposer can backfire, as it may reduce the probability that the agent is included in a winning coalition. We experimentally test the theoretical prediction of potentially negative returns to recognition probability in two-period legislative bargaining noted in Baron & Ferejohn (1989). 

We find that higher recognition probability benefits subjects in all treatments, except one in which we automate the second period. It is because proposers often favor the member with the greater recognition probability as a coalition partner, and such tendency varies depending on the proposer’s recognition probability, counter to the theoretical prediction. In all treatments, a vast majority of subjects exhibit a strict preference for higher recognition probability

Group Bargaining: A Model of International Treaty Ratification, with WonSeok Yoo (paper, earlier version at SSRN), Accepted at Games and Economic Behavior

We consider non-cooperative bargaining over a fixed surplus between two groups that may differ in size and the supermajority threshold they employ for within-group ratification. We find that total allocation to a group does not depend on group size and increases with the supermajority threshold. We use the Gini coefficient to study within-group inequality as an outcome of interest and find that inequality increases with group size and decreases with the supermajority threshold. Finally, we study delegation by concentrating the ability to influence proposals within a subset of group members. Delegation to a subgroup decreases the group's total allocation because non-delegates accept lower allocations. Inequality is higher if delegation is employed, and it is decreasing in the size of the delegate committee.

Working papers:

Identity-Based Elections, with Helios Herrera (paper, earlier version at CEPR, VoxEU article)

We propose a model of elections in which partisans wish to convince themselves that their party is the better choice. They select information sources from a broad array of media outlets with different biases to achieve that goal, but they may not always succeed due to their rationality which acts as a constraint. We explore how asymmetries between the two political sides skew electoral outcomes despite rationality. Here we consider salient examples such as asymmetric exposure or asymmetric trust in media, as well as propaganda, but this notion of partisanship is easily applicable to a wide variety of electoral contexts. 

Endogenous Limits on Veto Power in Dynamic Bargaining, with Ewout Verriest (paper)

We consider an infinitely repeated legislative bargaining model with a dynamically evolving status quo. Three players, one of whom is permanently endowed with veto power, must split a fixed budget in each period. Despite her additional power, the veto player cannot always asymptotically extract the full surplus. The non-veto players endogenously prevent each other’s expropriation when they are patient and have high initial allocations in the unique stationary, symmetric, stage-undominated, coalition-proof Markov perfect equilibrium. Further, we show that veto power and higher recognition probability may be strategic substitutes rather than complements. We also provide an intuition behind selfish egalitarianism between non-veto players. Our technique of employing coalition-proofness and iteratively generating a new equilibrium with novel predictions sheds light on the divergence in recent literature on the value of veto power and may be useful in other environments.

Career Concerns with Cost Uncertainty (paper

I consider a finite-horizon continuous-time career concerns model. As in Holmstrom (1999), there is symmetric uncertainty about a worker's ability. Additionally, the worker has private information about her cost of effort. The sequence of observed outputs allows learning about both attributes of the worker. Similar to other career concerns models, incentives to exert effort are driven in part by a signal-jamming motive, i.e., the desire to manipulate the market's beliefs about the worker's ability. This motive is present throughout the worker's lifetime, but its impact on worker effort gradually decreases over time as the market learns the worker's ability. In contrast, the motive to signal effort cost is more nuanced, and changes sign as time progresses. I find that early in her career, effort cost uncertainty pushes the worker to work harder to signal that she is has a low effort cost (is cost-efficient). During her middle and late career, the worker has an incentive to signal that she has a high effort cost (is cost-inefficient). These incentives to signal effort cost types are driven by the implication the effort cost type has on the market's posteriors regarding the worker's ability type. In the second phase of her career, the worker lowers her effort to seem like the cost-inefficient type. During her late-career, surprisingly, the worker increases her effort to signal to the market that she is cost-inefficient. While the focus of this paper remains theoretical, this paper suggests empirical implications and applications. The key applications considered are regarding the estimation of wages as a function of years of experience, the distribution of outputs, and performance-based pay.

Optimal Differentiation through Step-wise Learning (paper

I consider the problem of an entrant to an industry with a present incumbent, both vying to sell to a continuum of consumers. The ideal product designs for each consumer in this environment are distributed over an n-dimensional space. The entrant must first learn about the locations of consumers and then design a product. While the entrant can freely design her product, the incumbent has a valence advantage that the entrant must take into account. The valence advantage for the incumbent drives horizontal product differentiation because if the entrant chooses a product design too close to the incumbent, she will not sell to any consumers. 

I find that the entrant differentiates her product only as much as is needed to counteract the valence advantage of the incumbent. I also find that the optimal differentiation follows a predictable pattern that depends upon the strength of the incumbent. Against a weak (strong) incumbent, the entrant differentiates along dimensions with less (greater) expected variance, and less (greater) salience. Further, I show that the returns to research regarding consumer preferences along any dimension are initially convex, but must eventually become concave. Therefore, step-wise learning and specialization is optimal in this environment.

In progress:

Optimal Policy Choice and Campaign Financing, with Amy Hongfei Sun

We build a model in which two office-motivated candidates strategically choose policies while two policy-motivated large donors strategically choose how much to give. Voter ideal points are distributed over the policy space and they either abstain or vote non-strategically for the closer candidate. Candidates use own and donated funds to drive turnout. Donors give to candidates in close races and with more extreme positions. Self-funded candidates need to raise less outside funding and may choose more moderate positions.

Partisan Credulity and Fake News, with Helios Herrera

Some Non-Monotonicities in Finite-Period Legislative Bargaining , with WonSeok Yoo