Publications


Polity Size and the Congested Budget: Evidence from Italian Municipalities

with Massimo Morelli
Accepted, Journal of Politics

Once in office, politicians propose policies aimed at maintaining the support of their constituencies. This form of political activism increases with polity size – i.e., the number of politicians in government – but it may clash with capacity constraints, leading to a congestion effect whereby politicians’ plans are not enacted in practice. With novel data on Italian municipalities, we estimate the causal effect of polity size on a battery of planned and actual budget outcomes. We leverage a reform that introduced a new temporary population threshold where polity size changed discontinuously and estimate local treatment effects with a difference-in-discontinuities design. We document a congestion effect. Municipalities with larger polities have a larger planned budget which does not translate into a larger actual budget. The congestion effect decreases when bureaucratic capacity is high, proving how administrative capacity can be a binding constraint for politicians’ behavior.



A Costly Commitment: Populism, Government Performance, and the Quality of Bureaucracy

with Massimo Morelli and Matia Vannoni
2023, American Journal of Political Science
Covered by VoxEU.org

We study the consequences of populism for economic performance and the quality of bureaucracy. When voters lose trust in representative democracy, populists strategically supply unconditional policy commitments that are easier to monitor for voters. When in power, populists try to implement their policy commitments regardless of financial constraints and expert assessment of the feasibility of their policies, worsening government economic performance and dismantling resistance from expert bureaucrats. With novel data on more than 8,000 Italian municipalities covering more than 20 years, we estimate the effect of electing a populist mayor with a close-election regression discontinuity design. We find that the election of a populist mayor leads to smaller repayments of debts, a larger share of procurement contracts with cost overruns, higher turnover among top bureaucrats – driven by forced rather than voluntary departures – and a sharp decrease in the percentage of graduate bureaucrats.



A Dynamic Measure of Bureaucratic Reputation: New Data for New Theory

2022, American Journal of Political Science

Bureaucratic reputation is one of the most important concepts used to understand the behaviour of administrative agencies and their interactions with multiple audiences. Despite a rich theoretical literature discussing reputation, we do not have a comparable measure across agencies, between countries, and over time. I present a new strategy to measure bureaucratic reputation from legislative speeches with word-embedding techniques. I introduce an original dataset on the reputation of 465 bureaucratic bodies over a period of forty years, and across two countries, the US and the UK. I perform several validation tests and present an application of this method to investigate whether partisanship and agency politicisation matter for reputation. I find that agencies enjoy a better reputation among the members of the party in government, with partisan differences less pronounced for independent bodies. I finally discuss how this measurement strategy can contribute to classical and new questions about political-administrative interactions.





Under Review


The Executive Unbound? Politicized Bureaucracy and Partisan Procurement under DOGE

with Kyuwon Lee

The establishment of the Department of Government Efficiency (DOGE) during Trump’s second term marks an expansion of presidential authority over federal agencies. This institutional development provides a rare opportunity to examine whether presidents can leverage politicized agencies for political and electoral goals. Drawing on detailed procurement data and DOGE’s cancellation records, we find that Republican donor firms were less likely to face cancellations, whereas firms donating to Democrats were more likely to lose contracts. Cancellations were less frequent in Republican-held districts, conservative agencies, and states favorable to the Republican Party. Leveraging the timing of the 2025 Wisconsin Supreme Court election, we use a difference-in-differences design to show that Wisconsin-based firms experienced a sharp increase in cancellations following the election, underscoring the strategic timing of DOGE’s operations. Our findings shed new light on the consequences of agency politicization and align with the Trump administration’s effort to consolidate its support base.



The Shift to Commitment Politics and Populism: Theory and Evidence

with Massimo Morelli, Antonio Nicolò and Paolo Roberti

We present a theory of populism centered on commitment politics – a type of agency relationship in which candidates promise specific and monitorable policies. The shift to commitment politics is driven by increased distrust toward government institutions, itself a consequence of cognitive complexity and disinformation typical of modern social media environments. Candidates who adopt a commitment platform rationally choose all the complementary strategies associated with populism, including anti-elite rhetoric, misinformation, aversion to judicial independence, and bureaucratic expertise. The paper presents observational and experimental evidence from the United States on the supply and demand of commitment consistent with the model’s key predictions.



Group Appeals and Political Mobilization: Evidence from U.S. House Races

with Alonso Roman Amarales and Catherine E. De Vries

Group membership plays a crucial role in political conflict, with candidates frequently using group appeals to mobilize voters. However, the factors influencing candidates’ decisions to employ such appeals remain insufficiently understood. We propose an argument rooted in the notion that candidates selectively deploy group appeals to maximize electoral returns and we introduce a novel method for detecting group appeals in text, which is applied to the universe of tweets posted by U.S. House candidates between 2012-2021. We present three findings. First, candidates’ group membership and their district’s demographic composition are strongly associated with the frequency of group appeals. Second, protest events associated with specific groups increase appeals to those groups. Third, except for young voters, there is limited evidence that these appeals enhance turnout intentions. These findings offer new insights into how politicians strategically engage underrepresented groups in the electorate, contributing to our understanding of identity-based political mobilization.





Working Papers


Bureaucratic Information in Congress

Bureaucratic agencies produce a wealth of information that can be used by politicians when making policies. However, little is known about the extent to which members of Congress rely on bureaucratic information and what factors they consider when they do so. I introduce a novel measure of politicians’ reliance on bureaucratic information which uses natural language processing to extract and analyze bureaucratic information used by members of Congress in congressional speeches, newsletters, and social media. I find that members make greater use of information coming from ideologically similar bureaucracies, especially in floor speeches and during legislative hearings. Leveraging the Supreme Court’s decision in Seila Law vs. Consumer Financial Protection Bureau – which curtailed the independence of the CFPB – as a shock to the independence to the bureau, I find that members’ reliance on the information produced by the CFPB drops as a result of the court’s decision to. Statutory features insulating agencies from political control sharply increase members’ reliance on bureaucratic information. These findings have implications for theories of separation of powers and for the use of evidence in policy-making.



Personnel is Policy: Delegation and Political Misalignment in the Rulemaking Process

with Massimo Morelli, Jörg Spenkuch, Edoardo Teso, Matia Vannoni, and Guo Xu

Combining data on the U.S. federal rulemaking process with personnel and voter registration records, we study the consequences of political misalignment between regulators and the President. We show four main results. First, there are only small partisan cycles in the assignment of rules to regulators, and subject-matter expertise matters much more than partisan alignment in the assignment process. Second, rules overseen by misaligned regulators take systematically longer to complete. Third, misaligned regulators produce rules that have lower readability, are more likely to attract public opposition and to be challenged in court. Fourth, political leaders face a trade-off between alignment and expertise: we estimate that assigning rules only to aligned regulators would have lost 36% of the stock of expertise in the U.S. rulemaking process between 1997 and 2023.





Work in Progress


Campaign Contributions and Self-Reports: How Accurate Are Survey Responses Regarding Donation Behavior?

with Mike Barber and Brandice Canes-Wrone

Measures of whether an individual is a political donor commonly depend on self-reports from surveys. In the United States, the discrepancy between the percentage of self-reported donors and validated ones with the Federal Election Commission (FEC) has often been over 20 percentage points. Because “small donors” who give no more than $200 do not need to be reported to the FEC, it is unclear whether this discrepancy is due to overreporting, such as is prevalent for voting turnout. To investigate this issue, we conducted an original survey for which we had information on whether an individual is a validated FEC donor. The survey–which includes samples of the general population, affluent, and donors–includes questions on whether the respondent has given above the legal reporting threshold and whether they have ever donated. Additionally, we conducted an original list experiment to assess whether self-reported donating rates decline once the respondent’s answer is not identified individually in the survey response. The results suggest that overreporting is high for survey items that ask about the activity without reference to the amount, but drops substantially once respondents are asked whether they gave above the the legal threshold. More specifically, the list experiment suggests that only 3 percent of respondents donate even though it is common for over 20 percent of respondents to state that they donated in individual survey items. At the same time, the data on validated donors show that once respondents are asked whether they gave above the legal threshold, the rate of overreporting drops to only 5 percentage points.



The anti-Democrat Penalty: The Political Consequences of Regulatory Enforcement in the United States

A large literature explores the political drivers of regulatory enforcement and firm responses, but little is known about how employees’ political behavior changes when their employer is sanctioned. This paper investigates how enforcement actions by the Environmental Protection Agency (EPA) affect the political mobilization and partisan leaning of employees. Using a novel dataset linking firm-level enforcement data to individual employment records and historical voter files, I track over 14 million individuals across 150,000 regulated firms. Difference-in-differences estimates show that employees at sanctioned firms are less likely to register to vote and register as Democrats. Analysis of 411 million online reviews reveals employees do not blame the firm after the EPA penalty, suggesting the absence of anti-business attitudes as a result. This project highlights how bureaucratic enforcement can shape political behavior, offering new insights into the electoral consequences of regulatory governance.



Bureaucratic Responsiveness to Multiple Principals: Evidence from High-Frequency Social Media Data

with Hye Young You

We examine federal agencies’ responsiveness to multiple principals, focusing on the President, House Appropriations Committee members, and oversight committees. Utilizing a novel dataset of 2.17 million tweets from 94 agencies posted between 2010 and 2022, matched with 2.5 million tweets from members of Congress and presidents, we employ stance detection methods to classify tweets mentioning agencies as supportive, opposing, or neutral. Our findings reveal that agencies are highly responsive to supportive mentions, particularly from the President, resulting in a significant increase in Twitter activity. Conversely, agencies are less responsive to criticism. Using an event-study design, we show that when criticized by the President, agencies shift their focus to policy issues, indicating a reputation-maximizing strategy. This study underscores the unique role of social media in modern bureaucratic politics, where real-time signals from political principals significantly influence agency behavior and communication strategies.



A New Language-Based Measure of Bureaucratic Ideology

Classical models of American political institutions–spanning delegation, oversight, and information transmission–hinge on the ideological distance between the bureaucracy, presidents, and members of Congress. While ideology is well measured for presidents and legislators, comparable, time-varying measures for the bureaucracy remain scarce. This paper develops a dynamic, language-based measure of bureaucratic ideology, inferred from how members of Congress (MCs) discuss federal agencies. Using 8.5 million congressional speeches (1979–2022) mentioning hundreds of agencies, I apply natural language inference models to classify whether MCs express support, opposition, or neutrality toward each agency. These stances, combined with MCs’ DW-NOMINATE scores, produce time-variant estimates of agency ideology across administrations. Validation against expert assessments and prior metrics shows strong alignment. Applying the measure to congressional oversight, I find that greater ideological distance between MCs and agencies predicts more intensive questioning of bureaucratic witnesses. This scalable, low-cost approach enables new analyses of political–bureaucratic relations over time.