Essays on Modern Campaign Financing in the United States

Abstract

Campaigns are more expensive than ever. A growing number of Americans are contributing. Most of those who give, now give online. The consequences of broad-based, small-dollar, and public-facing online fundraising for 1) descriptive representation in the money candidates raise, 2) candidate messaging, and 3) donor welfare, are hotly debated, but understudied. In this dissertation, I examine modern fundraising from each of these angles. In Chapter 2, using a dataset matching federal donors in 2020 to a nationwide voter file and a nationwide property database, I find some gains in descriptive donor representation among small, online donors compared to large ones, though little evidence of observable platform effects. I then evaluate a proposed program of small-donor matching for House campaigns designed in part to remedy these inequalities. I find that such a program would likely achieve only marginal improvements in descriptive representation. In Chapter 3, using a comprehensive dataset of federal campaign fundraising emails from 2018 and 2020, I show how the content of these appeals varies with the electoral constraints faced by candidates: in more competitive races, Democratic (but not Republican) candidates appeal to potential donors with significantly less issue-focused content and less overall variation in language across districts. The public nature of online fundraising yields partisan asymmetries in candidates' abilities to optimize their fundraising messaging. In Chapter 4, coauthored with Will Schulz, we examine recurring online contributions and their sources. We estimate that automated contributions accounted for at least 8.6% and 14.0% of ActBlue and WinRed transactions in 2020, respectively --- more dollars than congressional candidates raised from unaffiliated PACs over the same period. Consistent with reporting that older donors are disproportionately misled by recurring contributions, particularly in the presence of pre-checked boxes that initiate them by default, we find that Republican donors with recurring contributions are older than similar donors without. We implement a survey experiment on Facebook to study pre-checked boxes. Conditional on giving, the rate of misled donors is high in all box conditions, suggesting that most donors who initiate recurring contributions do so unintentionally; however, pre-checked boxes dramatically increase the number of misled donors.