Working Papers
Working Papers
September 2024 (First version: November 2020)
We examine the effect of item arrangement on choices using a novel decision-making model based on the Markovian exploration of choice sets. This model is inspired by experimental evidence suggesting that the decision-making process involves sequential search through rapid stochastic pairwise comparisons. Our findings show that decision-makers following a reversible process are unaffected by item rearrangements, and further demonstrate that this property can be inferred from their choice behavior. Additionally, we provide a characterization of the class of Markovian models in which the agent makes all possible pairwise comparisons with positive probability. The intersection of reversible models and those allowing all pairwise comparisons is observationally equivalent to the well-known Luce model. Finally, we characterize the class of Markovian models for which the initial fixation does not impact the final choice and show that choice data reveals the existence and composition of consideration sets.
with Hans Gersbach, October 2024
This paper introduces Voting with Random Proposers (VRP) procedure to address the challenges of agenda manipulation in voting. In each round of VRP, a randomly selected proposer suggests an alternative that is voted on against the previous round’s winner. In a framework with single-peaked preferences, we show that the VRP procedure guarantees that the Condorcet winner is implemented in a few rounds with truthful voting, and in just two rounds under sufficiently symmetric preference distributions or if status quo positions are not extreme. The results have applications for committee decisions, legislative decision-making, and the organization of citizens’ assemblies and decentralized autonomous organizations.
2024, Games and Economic Behavior, 147, pp. 288-304.
Ordinal random utility models (RUMs) are based on the presumption that fluctuating preferences drive stochastic choices. We study a novel property of RUM subclasses called exclusiveness, satisfied whenever the supports of all RUM representations of stochastic choice data, rationalizable by a RUM over preferences within a specific domain, also belong to that domain. We demonstrate that well-known preference domains such as the single-peaked, single-dipped, triple-wise value-restricted and peak-monotone are RUM-exclusive, alongside a novel domain we term peak-pit on a line. Building on existing characterization results, we show how these preference domains can be directly revealed from stochastic choice data, without the need to compute all RUM representations.
with Pencho Yordanov (2024), In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 6899–6912, Miami, Florida, USA. Association for Computational Linguistics.
We investigate whether LLMs display a well-known human cognitive bias, the attraction effect, in hiring decisions. The attraction effect occurs when the presence of an inferior candidate makes a superior candidate more appealing, increasing the likelihood of the superior candidate being chosen over a non-dominated competitor. Our study finds consistent and significant evidence of the attraction effect in GPT-3.5 and GPT-4 when they assume the role of a recruiter. Irrelevant attributes of the decoy, such as its gender, further amplify the observed bias. GPT-4 exhibits greater bias variation than GPT-3.5. Our findings remain robust even when warnings against the decoy effect are included and the recruiter role definition is varied.
with Christian Ewerhart (2020). Games and Economic Behavior, 123, pp. 182-206.
This paper studies fictitious play in networks of noncooperative two-person games. We show that continuous-time fictitious play converges to the set of Nash equilibria if the overall n-person game is zero-sum. Moreover, the rate of convergence is 1/t, regardless of the size of the network. In contrast, arbitrary n-person zero-sum games with bilinear payoff functions do not possess the continuous-time fictitious-play property. As extensions, we consider networks in which each bilateral game is either strategically zero-sum, a weighted potential game, or a two-by-two game. In those cases, convergence requires a condition on bilateral payoffs or, alternatively, that the network is acyclic. Our results hold also for the discrete-time variant of fictitious play, which implies, in particular, a generalization of Robinson's theorem to arbitrary zero-sum networks. Applications include security games, conflict networks, and decentralized wireless channel selection.