DATA AND METHODS

We exploited the variation in state campaign finance laws across states and time to identify the effects of reforms on the competitiveness of incumbent reelection contests.

This natural experiment approach to understanding the effects of state campaign finance reforms was implemented in several recent studies (Flavin Reference Flavin2015; Klumpp, Mialon, and Williams Reference Klumpp, Mialon and Williams2016; Primo and Milyo Reference Primo and Milyo2006; Reference Primo and Milyo2020; Primo, Milyo, and Groseclose Reference Primo, Milyo, Groseclose and McDonald2006; Stratmann Reference Stratmann2010; Stratmann and Aparicio-Castillo Reference Stratmann and Aparicio-Castillo2006).

We used the most recent version of the state legislative election returns data compiled by Klarner (Reference Klarner2019) to identify all general election races involving incumbents running for reelection in single-member districts. We then eliminated contests with multiple incumbents running for the same seat or with no major-party incumbent, which includes all contests for the nonpartisan legislature in Nebraska. Finally, we restricted the period to 1986–2018 to match the time span covered by our campaign finance data. This selection process resulted in almost 66,000 state legislative races.

The upper panel of table 1 describes the dependent variables that we examined. Each represents a different way to characterize the competitiveness of incumbent reelection races. These include whether the incumbent faced a major-party challenger; whether the incumbent’s vote was less than 60% or 55%; and whether the incumbent was defeated. In the 33 years covered by our study, fewer than 60% of incumbents were challenged by a major-party candidate (i.e., Democrat or Republican) and less than 25% (i.e., 15%) of incumbents were held under 60% (i.e., 55%) of the votes cast in their race. However, the most telling statistic is that only 5% of state legislative incumbents running for reelection were defeated in the general election. Consequently, because the base rate is so low, even a small change in the probability of incumbent defeat may lead to a relatively large change in the frequency of incumbents defeated.

Table 1

Descriptive Statistics for State Legislative Elections, 1986–2018

Notes: N=65,979. The sample includes all general-election races in single-member districts with a major-party incumbent running for reelection (races with multiple incumbents or in Nebraska are omitted).

We estimated the effects of state campaign finance reforms on each measure of competitiveness via linear probability models (i.e., for ease of exposition and to facilitate the interpretation of fixed effects). To control for anti-incumbent and partisan tides in any given election year, all specifications include indicators for year, as well as the year interacted with whether the incumbent was a Democrat. All specifications also include indicators for state-legislative districts to control for unobserved district-level fixed effects.Footnote ¹ We also adjusted estimated standard errors for clustering by state (Primo, Jacobsmeier, and Milyo Reference Primo, Jacobsmeier and Milyo2007).

The key independent variables of interest in our analysis are indicators for state campaign finance regulations (see the middle panel of table 1). We followed a previous study (Primo and Milyo Reference Primo, Milyo, Groseclose and McDonald2006) and focused on state laws that limit contributions to candidates from corporations and individuals or that provide some type of public funds for state legislative candidates. All data on state campaign finance laws are from Primo and Milyo (Reference Primo and Milyo2020).

One advantage of this approach is that state campaign finance regulatory regimes then can be characterized in one of five mutually exclusive categories. We included five indicators for each regime: (1) limits on contributions to candidates from corporations; (2) limits on contributions to candidates from corporations and individuals; (3) states with limits on contributions and partial public funding for state legislators; (4) states with limits on contributions and full public funding (i.e., “clean-money” reforms); and (5) states without limits on contributions and no public funding. The fifth regime is the omitted category in our analysis, and the presence or absence of any of the other four regimes is described with binary indicator variables. Table 2 is a breakdown of regulatory regimes by state over time.

Table 2

Campaign Finance Regulatory Regimes, 1986–2018

Note: See table 1 for definitions of campaign-finance regulatory regimes.

Slightly less than 15% of races occur in states that do not limit contributions to candidates (i.e., the omitted regime category), whereas almost 20% of races occur in states that limit only contributions to candidates from corporations (i.e., regime 1). However, the most common regulatory regime imposes limits on contributions from both corporations and individuals (and with no public financing); that is, more than half of all incumbent reelection races occur under these regulations (i.e., regime 2).

Public funding occurs only in states that also limit both corporate and individual contributions (i.e., about another 10% of the contests in our sample), although public-funding schemes in the states come in two variations. Earlier adopters of public funding (e.g., Hawaii, Minnesota, and Wisconsin) provided limited matching grants to candidates conditional on adherence to a ceiling on candidate spending (i.e., regime 3). These “partial-funding” reforms have since given way more recently to so-called clean money reforms in Arizona, Connecticut, and Maine (i.e., regime 4). These states provide “full” public financing to candidates who can raise sufficient seed money from small donations and agree not to supplement their campaign spending with funds from private contributors. Of course, public funding of either type is voluntary in that candidates may opt out and fund their campaigns entirely through private donations.Footnote ²

Whereas the campaign finance laws described previously occur in five mutually exclusive regulatory regimes (captured with four indicator variables), the same is not true for prohibitions on corporate independent expenditures. Before the 2010 US Supreme Court decision in Citizens United, almost half of the states prohibited independent expenditures by corporations in support or opposition of state legislative candidates (Klump, Mialon, and Williams Reference Klumpp, Mialon and Williams2016). Consequently, we included a separate indicator for the presence of prohibitions on independent expenditures by corporations, which encompasses almost one third of the contests in our sample.

Finally, we included indicators to control for the presence of legislative term limits in a state. Data on state-legislative term limits are from the National Conference on State Legislatures. Descriptive statistics for this variable, as well as the share of races involving the upper chamber of the state legislature or with Democratic incumbents, are shown in the lower panel of table 1. However, these last two variables are subsumed by the inclusion of fixed effects for legislative districts and (party of the incumbent X year) in subsequent regression models.