About

Avidit Acharya is a professor of political science at Stanford University, with courtesy appointments in political economy at the Stanford Graduate School of Business and senior fellowship at the Hoover Institution. He earned his PhD in political economy from Princeton University in 2012 and his BA in economics and mathematics from Yale University in 2006. His research focuses on political economy and formal political theory. Acharya's first book, Deep Roots: How Slavery Still Shapes Southern Politics, explores the lasting impact of slavery as an institution on political attitudes in the American South. His second book, The Cartel System of States: An Economic Theory of International Politics, offers a new understanding of the development and current state of the territorial state system. His work has been published in both economics and political science journals and has received awards such as the Elinor Ostrom best paper award, the Gosnell Prize in political methodology, and the Joseph Bernd best paper award. Acharya is also an editor at the journal Social Choice and Welfare and an advisory editor at Games and Economic Behavior.

Research topics

• Sociology
• Geography
• Philosophy
• History
• Economics
• Business
• Epistemology
• Engineering

Selected publications

• Learning Preferences from Conjoint Data: A Structural Deep Learning Approach

  arXiv (Cornell University) · 2026-04-12

  preprintOpen access1st authorCorresponding

  Conjoint experiments randomize multidimensional profiles, offering a powerful design for recovering structural preference parameters -- including marginal rates of substitution, willingness to pay, and the distribution of preferences across a population. Yet the dominant approach in political science has focused on nonparametric causal estimands that do not leverage this potential. We propose a structural approach that embeds a deep neural network within a random utility logit model, allowing preference parameters to vary as a fully flexible function of respondent characteristics. The neural network addresses the concern that a parametric specification may not capture the true data generating process, while double/debiased machine learning provides valid inference on average preference parameters. We apply our method to three prominent conjoint studies and find rich preference heterogeneity masked by reduced-form averages: a near-zero gender effect coexists with 83% preferring female candidates, opposition to undemocratic behavior is near-universal but varies sharply in intensity, and progressive tax preferences cut across every partisan subgroup.

  PublisherDOI

• Learning Preferences from Conjoint Data: A Structural Deep Learning Approach

  arXiv (Cornell University) · 2026-04-12

  articleOpen access1st authorCorresponding

  Conjoint experiments randomize multidimensional profiles, offering a powerful design for recovering structural preference parameters -- including marginal rates of substitution, willingness to pay, and the distribution of preferences across a population. Yet the dominant approach in political science has focused on nonparametric causal estimands that do not leverage this potential. We propose a structural approach that embeds a deep neural network within a random utility logit model, allowing preference parameters to vary as a fully flexible function of respondent characteristics. The neural network addresses the concern that a parametric specification may not capture the true data generating process, while double/debiased machine learning provides valid inference on average preference parameters. We apply our method to three prominent conjoint studies and find rich preference heterogeneity masked by reduced-form averages: a near-zero gender effect coexists with 83% preferring female candidates, opposition to undemocratic behavior is near-universal but varies sharply in intensity, and progressive tax preferences cut across every partisan subgroup.

  PublisherOA PDF

• Use of communication channels in 2x2 strategic games

  2025-07-23

  preprintOpen access

  How do we use language to coordinate with each other? This question is of interest both from the cognitive science of language perspective as well as the game theory perspective. Here we attempt to bridge the gap between the empirical tradition of studying language use for coordination in reference games with the theoretical and analytic framework of game theory. We conducted a series of 5 real-time online experiments where pairs of participants were matched to play a sequence of 40 2x2 games together (total N=616). Across the experiments, we varied the exact structure of trials, but all experiments involved trials with reward structures modeled on Prisoner’s Dilemma (PD) and Bach or Stravinsky (BoS). We manipulated the presence or absence of a chat interface as a between-pairs manipulation. Despite low levels of use of the communication channel, we found slight differences in outcome between the chat and nochat groups, where pairs who could communicate were slightly more likely to reach coordinated beneficial outcomes in BoS games. In contrast to the predictions of rational actors, participants in both chat and no-chat conditions generally aligned on both cooperating rather than both defecting in PD. Overall, we were limited in our ability to discern the role of communication channels in game strategies due to low uptake in using the communication channels when available.

  PublisherOA PDFDOI

• Motivated Reasoning and Information Aggregation

  ArXiv.org · 2025-12-10

  preprintOpen access1st authorCorresponding

  If agents engage in motivated reasoning, how does that affect the aggregation of information in society? We study the effects of motivated reasoning in two canonical settings - the Condorcet jury theorem (CJT), and the sequential social learning model (SLM). We define a notion of motivated reasoning that applies to these and a broader class of other settings, and contrast it to other approaches in the literature. We show for the CJT that information aggregates in the large electorate limit even with motivated reasoning. When signal quality differs across states, increasing motivation improves welfare in the state with the more informative signal and worsens it in the other state. In the SLM, motivated reasoning improves information aggregation up to a point; but if agents place too little weight on truth-seeking, this can lead to worse aggregation relative to the fully Bayesian benchmark.

  PublisherOA PDFDOI

• Political accountability under moral hazard

  American Journal of Political Science · 2024-04-03 · 9 citations

  article1st authorCorresponding

  Abstract Viewing the relationship between politicians and voters as a principal–agent interaction afflicted by moral hazard, we examine how political careers are shaped by the incentives that voters provide incumbents to work in the public interest. When moral hazard binds, the optimal way for voters to hold politicians accountable is to provide re‐election incentives that evolve dynamically over their careers in office. Under these incentives, first‐term politicians are among the most electorally vulnerable and the hardest‐working; politician effort rises with electoral vulnerability; electoral security increases following good performance and decreases following bad performance; and both effort and electoral vulnerability tend to decline with tenure. In extensions, we study limited voter commitment, voluntary retirement from politics, and adverse selection.

  PublisherDOI

• Ranked Choice Voting, the Primaries System, and Political Extremism: Theory and Simulations

  SSRN Electronic Journal · 2024-01-01 · 1 citations

  articleOpen access1st authorCorresponding
  PublisherDOI

• Electoral Campaigns as Dynamic Contests

  Journal of the European Economic Association · 2024-01-31 · 8 citations

  articleOpen access1st authorCorresponding

  Abstract We develop a model of electoral campaigns in which two office-motivated candidates allocate their budgets over time to affect their odds of winning. We measure the candidates’ evolving odds of winning using a state variable that tends to decay over time, and we refer to it as the candidates’ “relative popularity.” In our baseline model, the equilibrium ratio of spending by each candidate equals the ratio of their initial budgets; spending is independent of past realizations of relative popularity; and there is a positive relationship between the strength of decay in the popularity process and the rate at which candidates increase their spending over time as election day approaches. We use this relationship to recover estimates of the perceived decay rate in popularity leads in U.S. subnational elections.

  PublisherOA PDFDOI

• Selective Exposure and Electoral Competition

  The Journal of Politics · 2024-09-17 · 1 citations

  article1st authorCorresponding

  We study how selective exposure to information by voters impacts electoral competition between two policy-motivated candidates. Each candidate has stochastic valence that is realized after the candidates choose platforms. In our model of selective exposure, voters receive information about the candidates’ valences that is slanted to reflect their ideological preferences. Existing work predicts that selective exposure intensifies platform polarization relative to settings in which voters obtain information from a neutral source. We show instead that it can reduce platform polarization.

  PublisherDOI

• The Political Economy of the Middle Income Trap *

  SSRN Electronic Journal · 2024-01-01 · 2 citations

  preprintOpen access1st authorCorresponding
  PublisherDOI

• Historical Persistence

  Oxford University Press eBooks · 2023 · 15 citations

  1st authorCorresponding
  • Sociology
  • Epistemology
  • History

  Abstract Historical persistence refers to the idea that historical causes can have effects that persist long into the future. This chapter provides an overview of the literature on historical persistence, focusing on key substantive themes. It lays out a discussion of the theories behind historical persistence. It also considers some methodological issues that emerge in establishing causality and interpreting the empirical findings of the literature. Finally, the chapter concludes with a discussion of how the persistence literature could now benefit from a reorientation that seeks to answer the main open question in the literature: under what conditions does historical persistence take place, and under what conditions does it not?

  PublisherDOI

Frequent coauthors

• Kirk Bansak

  University of California, Berkeley

  93 shared

• Jens Hainmueller

  Stanford University

  93 shared

• Maya Sen

  18 shared

• Matthew Blackwell

  18 shared

• Alexander Lee

  NHS England

  9 shared

• Edoardo Grillo

  University of Padua

  7 shared

• Juan Ortner

  7 shared

• John E. Roemer

  Yale University

  5 shared

Education

• Ph.D., Political Science

  Stanford University

  2009

• B.A., Government

  Harvard University

  2004

Awards & honors

• Elinor Ostrom best paper award
• Gosnell Prize in political methodology
• Joseph Bernd best paper award