From Networks to the Ballot Box: How Parties Tilt the Playing Field, and Why It Matters

Growing up in Paraguay, I wondered why corruption survived the return of democracy. After a dictatorship that ended in 1989, many hoped elections would bring cleaner government and prosperity. Yet two presidents have been convicted of corruption, and for years Transparency International ranked Paraguay among the most corrupt. I came to see that politics often runs on relationships and favors as much as ideas and governance. Cash and public jobs offered around elections quietly forged loyalties, and those loyalties surfaced at the polls. That puzzle drives my research: to study how democracy works in practice, not just on paper, and how these informal channels can tilt elections and, ultimately, state performance.

Clientelism is the exchange of private benefits (such as short-term aid and hiring favors) for political support. Brokers are local go-betweens who know the community and match favors to households, expecting loyalty and voting for their party in return. Understanding how this system works, and why it persists, is essential for fair elections and transparent public programs.

Clientelism endures because it adapts. My coauthors (Frederico Finan, Horacio Larreguy, and Laura Schechter) and I study how political brokers in rural Paraguay use everyday social ties to identify voters who will reciprocate favors at the polls. Our research, which was recently published in the Review of Economic Studies, maps village‐level networks from both voters and brokers and shows that brokers aren’t merely local insiders. They are strategically embedded in village networks to “hear” about non-copartisan voters who are likely to return favors. This predictive targeting helps explain why vote buying survives even under secret ballots, as monitoring is replaced by inference, not force. For recent reporting on Paraguay’s machine politics, see this 2023 New York Times article on the latest election.

What we found, and why it matters

When brokers can learn more about a voter through shared ties, they are significantly more likely to target them. Network “hearing” tracks who is likely to reciprocate among non-copartisans. These findings could reframe anti-clientelism policy: improving vote secrecy or civic education is insufficient if brokers’ local advantage remains intact. Reforms may instead want to disrupt the mechanisms making clientelism profitable and feasible. For example, by reducing the returns to brokers’ targeted distribution, strengthening formal program eligibility that reduces voters’ vulnerability as shown by Bobonis and coauthors, and seeking ways to reduce political brokers’ information access on voters.

Beyond brokers: election administration as a partisan advantage

Clientelism is one way parties may tilt the field, election administration can be another. In Getting a Seat at the Table: Partisan Poll Workers and Electoral Bias (with Andres Carrizosa), we leverage Paraguay’s quasi-random assignment of voters to booths (alphabetical ordering by their surnames within booths at polling stations) to isolate the effect of poll-worker partisanship on recorded vote shares. The main idea behind our empirical strategy is that voters’ names are uncorrelated with their political preferences in many contexts (building on research by Cantu and Casas and coauthors), and if voters are assigned to booths based on their names, the average political preference of voting booths within a polling station is similar. However, something that does vary across booths within a station is who’s counting votes.

We also explore a particular election oversight mechanism present in some countries, which is that of adversarial vote-counting systems. In this system, poll workers are party representatives and mutual oversight is meant to ensure integrity. Yet in countries with dominant parties, parties often have de facto unequal capacities to send representatives to all booths. Through extensive data work on identifying the poll workers at each booth merged with administrative data on the partisanship of every Paraguayan citizen, we uncover the unequal allocation of partisan poll workers at each voting booth. We find small but systematic advantages for parties that hold a poll-worker majority at a booth, independent of voter preferences. In our main specifications, we find that partisan poll workers decrease an opposing party’s vote share by up to 1.1 percentage points and increase theirs by up to 0.7 points. While the measured effects are small, in close races these differences can still be pivotal.

What motivates brokers, and the costs to the state

Another thread of my research agenda asks why individuals become brokers. Anecdotal evidence suggests that grassroots party work can be a pathway to public sector jobs. In my job market paper, Patronage Hierarchies and Tax Collection, I examine how patronage shapes bureaucratic hiring and performance in the customs authority of Paraguay, with implications for state capacity and fiscal efficiency. If broker work improves public employment prospects, then this motivates the persistence of this practice with efficiency costs for revenue collection and service delivery more broadly, as hiring criteria extend beyond qualifications.

The broader research program

Across these projects, my goal is to understand how patronage and clientelism structure both electoral competition and the state. While dense social networks can improve risk-sharing and information flows, they may also enable clientelistic targeting by brokers, partisan control of vote counts can bias outcomes even when rules appear neutral, and patronage logics can distort who staffs the bureaucracy and how well it performs. Together, these dynamics explain how elections can be formally competitive yet structurally skewed. My work aims to raise transparency and reduce the payoffs to clientelistic practices and procedural distortions, so elections are not just formally competitive but substantively fair, leaving less room for corruption under democracy.