Money Maps Reveal How Democrats and Republicans Fund Each Other’s Campaigns

Money moves through the United States Congress in ways that most voters never see. Beyond the well-known flood of donations from individuals, corporations, and outside groups, members of the U.S. House of Representatives routinely transfer campaign funds to one another, using their own campaign committees and leadership political action committees to bankroll colleagues facing difficult elections. These internal transfers are perfectly legal, publicly disclosed, and strategically deliberate, yet they have long escaped systematic quantitative scrutiny. A new study published in PLOS Complex Systems by Yidan Sun and Mayank Kejriwal of the University of Southern California’s Information Sciences Institute changes that, treating the flow of money between lawmakers as a complex network and modeling, with statistical rigor, the forces that shape it.

The researchers compiled campaign contribution records from the Federal Election Commission spanning seven two-year election cycles, from 2009 to 2022. For each cycle, they built a directed network in which every one of the 435 elected House members appears as a node, and an arrow connects one lawmaker to another if the first, through an affiliated committee, contributed to the second’s campaign. Because contributions across party lines are vanishingly rare, accounting for less than half a percent of all records, the team constructed separate networks for Democrats and Republicans. The resulting money flow networks capture who gives, who receives, and how those patterns evolve across a period that includes the rise of Super PACs following the Supreme Court’s Citizens United decision in 2010.

Mapping the networks was only the first step. To explain their formation, Sun and Kejriwal turned to exponential random graph models, a class of statistical models that estimate the probability of observing a network given a set of structural and attribute-based features. Unlike ordinary regression, these models can capture endogenous dependencies among ties, such as reciprocity, whether a contribution in one direction makes a return contribution more likely; triadic closure, whether members who share contribution partners tend to give to each other; and in-degree centralization, whether incoming contributions concentrate among a small elite of recipients. The models also incorporate exogenous covariates, including each member’s cumulative House terms served, formal leadership status, shared state representation, and the number of Super PACs making independent expenditures supporting or opposing each member.

The technical machinery was demanding. The models were estimated with Markov Chain Monte Carlo maximum likelihood methods using the ergm package in R, and the authors ran extensive goodness-of-fit diagnostics, comparing statistics from one hundred simulated networks against the observed data for each party and cycle. To summarize effects across seven cycles, they pooled estimates with random-effects meta-analysis, reporting odds ratios with 95 percent confidence intervals. Sensitivity analyses confirmed robustness: replacing leadership measured in the next Congress with leadership held concurrently, and replacing Super PAC counts with total expenditure amounts, produced results consistent in direction with the main specification.

The headline finding is a striking divergence between the two parties. Democratic contribution networks remained remarkably stable and centralized across the entire study period. Incoming ties stayed concentrated among a small subset of recipients, reciprocity remained low and steady, and seniority showed a clear signature: each additional term served raised the odds of contributing by about 5 percent while lowering the odds of receiving by a similar margin. This is the statistical portrait of a hierarchical system in which electorally secure, senior Democrats funnel money to vulnerable colleagues, a pattern consistent with long-standing accounts of coordinated party strategy.

Republican networks told a different story. Early in the study period, Republican contributions were heavily centralized, with in-degree odds ratios far below one, indicating that a few members absorbed most incoming funds. Over the following cycles, that centralization steadily weakened; by 2021 to 2022 the estimate was statistically indistinguishable from a random distribution of recipients. Reciprocity among Republicans also collapsed, falling from an odds ratio of 0.70 in 2009 to 2010 to just 0.03 by the final cycle, meaning return contributions became almost nonexistent. Meanwhile, triadic closure grew stronger, with a pooled odds ratio of 1.91 for Republicans compared with 1.36 for Democrats, showing that Republican giving increasingly occurred within locally interconnected clusters of members who shared contribution partners.

The authors interpret this shift cautiously, noting that the timing coincides with the entry of Tea Party-aligned candidates and documented increases in intra-party factionalization. A contextual reading, they suggest, is that reduced reliance on leadership-centered giving coincided with more coordination among ideologically aligned subsets of the Republican conference. They are careful to frame this as interpretive rather than causal, but the structural evidence of declining hierarchy and rising local clustering aligns with scholarship describing organizational change within the Republican coalition after 2010. Democratic networks, by contrast, show no comparable drift toward decentralization, despite ideological heterogeneity within the caucus.

Leadership status emerged as a powerful predictor, particularly among Republicans. Members who would hold formal leadership positions in the next Congress, including Speaker, party leader, whip, caucus chair, or committee chair, had roughly double the odds of contributing compared with non-leaders, with a pooled odds ratio of 2.11 for Republicans and 1.49 for Democrats. Leaders also had significantly lower odds of receiving contributions, especially among Republicans. Crucially, when the researchers re-estimated the models using leadership held during the same Congress, the estimates barely changed, suggesting that future leaders already give at elevated rates before assuming office. That pattern is consistent with leadership selection drawing on demonstrated fundraising generosity, rather than contributions rising only after appointment, a finding with direct relevance to debates about how influence is built in Congress.

The Super PAC results may be the most politically provocative. Members targeted by more Super PACs, whether the spending supported or opposed them, were consistently less likely to give and more likely to receive internal contributions. For Democrats, each additional supportive Super PAC was associated with a 47 percent decrease in the odds of contributing and a 25 percent increase in the odds of receiving; each additional opposing Super PAC raised receiving odds by 41 percent. Republicans showed the same direction of effects. Because Super PACs are legally barred from coordinating with campaigns, the finding is remarkable: outside spending, though formally independent, appears to function as a public signal of electoral vulnerability that co-partisans read and act upon. The authors stress that the association is descriptive, since electoral risk could drive both Super PAC targeting and internal giving, but the stability of the pattern across seven cycles suggests a durable informational channel operating despite coordination prohibitions.

The study’s implications extend beyond academic network science. By quantifying how seniority, leadership, and outside spending shape the internal economy of Congress, the work provides an empirical foundation for ongoing debates about term limits, which rest partly on assumptions about the advantages of long tenure, and about the role of unlimited independent expenditures in American elections. The authors acknowledge limitations: the models exclude ideological proximity, caucus membership, and committee overlap, and each cycle is analyzed as an independent cross-section rather than within a temporal framework. Still, the dataset and replication scripts are publicly available, and the picture they reveal is unambiguous. Campaign finance within Congress is not random. It is a structured system, organized differently by each party, that responds measurably to institutional hierarchy and to the visible tremors of outside money, offering network science a new lens on how political power is financed from within.

Subject of Research: Network modeling of campaign finance flows among U.S. House members

Article Title: Structural modeling of campaign finance decisions in the U.S. House of Representatives

Article References: Sun, Y., & Kejriwal, M. (2026). Structural modeling of campaign finance decisions in the U.S. House of Representatives. PLOS Complex Systems, 3(5), e0000104. https://doi.org/10.1371/journal.pcsy.0000104

Image Credits: AI Generated

DOI: 10.1371/journal.pcsy.0000104

Keywords: campaign finance, U.S. House of Representatives, network analysis, exponential random graph models, Super PACs, political parties, seniority, leadership PACs, Citizens United, reciprocity, complex systems, Federal Election Commission