Research

Publications

• Sorting for K-Street: Post-Employment Regulations and Wage Setting in Congress
  The Journal of Politics, 2025, 87(2):664-679
  
  Abstract

  While post-employment regulations are a common tool to slow the revolving door in government, little is known about their effectiveness and consequences. Using the 2007 Honest Leadership and Open Government Act (HLOGA), I argue that policymakers strategically adjust their behaviors to maintain lucrative career options in the lobbying industry. HLOGA prohibited staffers-turned-lobbyists who earn at least 75% of a Congress member’s salary from contacting their ex-employers in Congress for one year. Using data on the complete set of congressional staff (2001-2016), I show that staffers sort below the salary threshold post-HLOGA. Employing various panel data analyses, I also find that selecting out of the regulation increases a staffer’s probability to become a lobbyist and ensures a substantial premium in revenues at the beginning of their lobbying career. These results explain why reforms of the revolving door fail and provide insights on institutional determinants of career incentives for non-elected public officials.
  
  article
  appendix
  replication




• Multilanguage Word Embeddings for Social Scientists: Estimation, Inference and Validation Resources for 157 Languages
  with Pedro L. Rodriguez, Arthur Spirling, and Brandon M. Stewart
  Political Analysis, 2025, 33(2):156-163
  2022 PolMeth Best Poster Award
  
  Abstract

  Word embeddings are now a vital resource for social science research. Unfortunately, it can be difficult to obtain high quality embeddings for non-English languages, and it may be computational expensive to do so. In addition, social scientists typically want to make statistical comparisons and do hypothesis tests on embeddings, but this is non-trivial with current approaches. We provide three new data resources designed to ameliorate the union of these issues: (1) a new version of fastText model embeddings, fit to Wikipedia corpora; (2) a multi-language “a la carte” (ALC) embedding version of the fastText model fit to Wikipedia corpora; (3) a multi-language ALC embedding version of the well-known GloVe model fit to Wikipedia corpora. These materials are aimed at “low resource” users who lack access to large corpora in their language of interest, or who lack access to the computational resources required to produce high-quality vector representations. We make these resources available for 30 languages, along with a code pipeline for another 127 languages available from Wikipedia corpora. We provide extensive validation of the materials, via reconstruction tests and some translation proofs-of-concept. We also conduct and report on human crowdworker tests, for our embeddings for Arabic, French, (traditional, Mandarin) Chinese, Japanese, Korean, Russian and Spanish.
  
  article
  appendix
  resources
  replication




• Bureaucratic Resistance and Policy Inefficiency
  with Kun Heo
  American Political Science Review, 2026
  
  Abstract

  Poor public service provision creates an electoral vulnerability for incumbent politicians. Under what conditions can bureaucrats exploit this to avoid reforms they dislike? We develop a model of electoral politics in which a politician must decide whether to enact a reform of uncertain value, and a voter evaluates the incumbent’s reform based on post-reform government service quality, which anti-reform bureaucrats can undermine. Bureaucratic resistance for political leverage is most likely when voters are torn between the reform and the status quo. Resistance lowers the informational value of government service for voters and can lead to policy distortions and accountability loss. When reform is moderately popular, resistance prevents beneficial reforms due to electoral risks and induces ineffective reforms by providing bureaucrats as scapegoats. Our model identifies a distinct mechanism of bureaucratic power and its implications for policy and accountability.
  
  article
  appendix




• Political Power of Bureaucratic Agents: Evidence from Policing in New York City
  Journal of Politics, 2026
  
  Abstract

  To what extent can bureaucrats manipulate public service provision for explicitly political ends? A growing body of work highlights the immense ability of bureaucrats to influence governments through campaign contributions, endorsements, collective bargaining, and organized election turnout. I explore a more fundamental mechanism of bureaucratic influence: bureaucrats strategically shirking responsibilities. Politicians depend on bureaucrats to achieve policy goals. This gives the latter leverage over the former. If bureaucrats deviate in their preferences from politicians and are organized in cohesive unions with strong tenure protections, they can collectively reduce effort to exert political pressure. I use data on New York Police Department (NYPD) 911 response times together with council members’ preferences on the FY2021 $1 billion cut to the NYPD’s budget. Employing difference-in-differences and spatial difference-in-discontinuities designs, I find that police reduced effort in districts of non-aligned politicians by slowing response times. This study informs the theoretical debate on principal-agent relationships in government and highlights the importance of organized political interests to explain policing in US cities.
  
  article
  appendix
  replication

Working Papers

• The Limits of Merit: Career Pipelines and Representation in US Law Enforcement
  R&R at the American Political Science Review
  2024 Best Paper Award, APSA Urban and Local Politics Section
  
  Abstract

  Civil service systems promise equal access to government employment, yet the bureaucracies they govern often remain unrepresentative: US police departments are systematically Whiter and more Republican than their jurisdictions. Scholarship and police reforms have focused on written entry exams, but bureaucratic careers extend well beyond entry. Drawing on novel data linking exam, career, and voter records for 99,000 exam passers and 58,000 NYPD officers, this study decomposes these disparities across the career pipeline. The application and written exam produce limited sorting. Instead, even among applicants with comparable exam scores, post-exam hiring favors Republican and White candidates, who are more likely to be promoted, receive awards, and remain in the agency. Equalizing post-exam hiring alone would increase appointments of Democratic and Black candidates by 78% and 27%. These findings redirect attention from the entry exam to the discretionary procedures that follow, and reveal how formally meritocratic institutions sustain unequal representation.
  pdf




• The Political Consequences of Police Slowdowns
  with Arvind Krishnamurthy and Dvir Yogev
  2026 LPEC Best Poster Award
  
  Abstract

  Police increasingly resist reform by strategically reducing service quality to shape electoral outcomes, yet we lack systematic evidence on how this bureaucratic resistance moves reform attitudes among voters and elected officials. We argue that police wield two tools grounded in their day-to-day service provision: fearmongering, public warnings that tie a reform to future service decline, and resistance shirking, the quiet withdrawal of effort afterward. Because voters cannot easily distinguish policy failure from bureaucratic resistance, the mere prospect of pushback can deter voter and policymaker support for reform. We test this with two survey experiments, alongside observational evidence on the prevalence of police fearmongering and slowdowns. A 2×2 vignette experiment with 1,676 mass-public respondents varies whether a police union signals cooperation or fearmongering and whether post-reform service quality improves or worsens, and an experiment of more than 500 sitting local elected officials varies whether police threaten resistance to a qualified-immunity reform. In the mass experiment, fearmongering is independently costly to politicians regardless of service outcomes, and service quality strongly shapes support for both reforms and incumbents. In the elite experiment, a threatening police response substantially reduces officials’ support for reform even though no service decline has occurred. Together, these results show that police possess strategic tools operating across the entire policy pipeline, creating anticipatory political costs for both voters and policymakers.
  pdf




• Cultural Conflicts, Professional Identity, and Workforce Sorting: Evidence from U.S. Public School Employees
  with Roxanne Rahnama and Tyler Simko
  
  Abstract

  How do cultural conflicts reshape the composition of the institutions they target? We argue that culturally conflictual policies threaten bureaucrats’ public sector motivation and shape employment patterns through two channels: how a policy constrains an employee’s discretion, and whether the ideological direction of the policy aligns with the employee’s own values. We test our argument by collecting administrative roster data on 2.9 million public school employees across 12 states. We link each employee to a contentious policy adoption in their school district, and identify how each policy impacts their individual employment choices. Using a matched triple difference-in-differences design, we find that school district bans on the instruction of Critical Race Theory increase the probability that Democratic employees move to other districts, while Republican employees become more likely to remain in their current districts. Further, we find book ban adoptions drive profession exit among Democratic librarians. Interviews with current and former public school employees corroborate both channels, showing policy misalignment imposes costs even on educators whose own subject is not directly targeted. Our findings reveal how nationalized local political conflict can lead to durable partisan sorting in the bureaucracy.
  Paper available upon request