Campaign Finance Regulations and Public Policy

Abstract Despite a century of efforts to constrain money in American elections, there is little consensus on whether campaign finance regulations make any appreciable difference. Here we take advantage of a change in the campaign finance regulations of half of the U.S. states mandated by the Supreme Court’s Citizens United decision. This exogenously imposed change in the regulation of independent expenditures provides an advance over the identification strategies used in most previous studies. Using a generalized synthetic control method, we find that after Citizens United, states that had previously banned independent corporate expenditures (and thus were “treated” by the decision) adopted more “corporate-friendly” policies on issues with broad effects on corporations’ welfare; we find no evidence of shifts on policies with little or no effect on corporate welfare. We conclude that even relatively narrow changes in campaign finance regulations can have a substantively meaningful influence on government policy making.

D espite a century of efforts to constrain money in American elections, there is little consensus on whether campaign finance regulations make any appreciable difference. Here we take advantage of a change in the campaign finance regulations of half of the U.S. states mandated by the Supreme Court's Citizens United decision. This exogenously imposed change in the regulation of independent expenditures provides an advance over the identification strategies used in most previous studies. Using a generalized synthetic control method, we find that after Citizens United, states that had previously banned independent corporate expenditures (and thus were "treated" by the decision) adopted more "corporate-friendly" policies on issues with broad effects on corporations' welfare; we find no evidence of shifts on policies with little or no effect on corporate welfare. We conclude that even relatively narrow changes in campaign finance regulations can have a substantively meaningful influence on government policy making.
S cholars and observers alike are increasingly concerned about the dominance of moneyed interests in American politics. Studies have shown that elected representatives' votes most closely track the preferences of their affluent constituents and campaign donors, federal policy primarily reflects the influence of well-off citizens and well-funded interest groups, and lobbying groups overwhelmingly represent business and the well-to-do, not ordinary citizens. 1 Adding to these long-standing concerns, the Supreme Court's 2010 Citizens United decision decreed that (1) corporations and unions have the same speech rights as individuals and (2) spending to influence elections does not give rise to corruption, or the appearance of corruption, as long as it is not coordinated with a political campaign. Based on the latter finding, the D.C. Circuit Court soon ruled in SpeechNow v. FEC that contributions to and spending by "Super PACs" and other groups that are independent of campaign organizations cannot be constitutionally limited.
These court decisions applied to state as well as federal election laws. But not all states were affected.
Prior to Citizens United, 23 states had banned "independent expenditures" (IEs) by unions and/or corporations-exactly the type of restrictions struck down by the high court. By forcing these states to drop their bans, the Supreme Court altered the campaign finance environment in these states, resulting in a significant increase in IEs in state elections (detailed below).
While many have denounced the Citizens United decision, scholars have long suspected that campaign finance laws are largely ineffective and, at best, succeed only in shifting the way in which money is spent-"political money, like water, has to go somewhere. It never really disappears into thin air" (Issacharoff and Karlan 1999).
Estimating the effect of campaign finance laws has been hampered by the difficulty of drawing causal inferences from observational data. Federal elections are governed by a single set of regulations that change infrequently, and when such changes occur they are often confounded with other shifts in political context. While the U.S. states provide greater variation, analysis of these policies is similarly hindered by many potential confounds-states with stricter campaign finance regulations are likely to differ from other states in ways that cannot be fully accounted for.
Given these challenges, it is not surprising that previous analyses disagree as to whether campaign finance regulations influence public policy. In this research note, we leverage an exogenous change in state campaign finance law mandated by the Supreme Court's Citizens United decision to estimate its effect on policy outcomes. This provides a stronger basis for causal inference and lends greater support to the conclusion that campaign finance laws can shape the policies governments adopt. Citizens United affected IEs by unions, for-profit corporations, and advocacy organizations, such as 501 (c)(4)s, that took money from for-profit corporations or unions; other 501(c)(4)s were already free to make IEs due to the Supreme Court's 1986 Massachusetts Citizens for Life decision. (Following common practice in the campaign finance literature, we use "corporations" as shorthand for "for-profit corporations.") We hypothesize that Citizens United will have the effect of moving state policy in a pro-corporate direction. This expectation rests on three factors. First, 23 states had preexisting bans on corporate IEs but only 15 of those also banned IEs by unions. Therefore, dropping all of those bans more broadly affected corporations than unions. Second, as detailed below, the increase in IEs in state elections subsequent to Citizens United was considerably greater for business corporations than for labor unions. Third, previous research found that Citizens United increased Republican representation in state legislatures (Abdul-Razzak, Prato, and Wolton 2020; Harvey and Mattia 2019; Klumpp, Mialon, and Williams 2016). Because the Republican Party is, broadly speaking, more aligned with corporate interests than the Democratic Party (Brunell 2005), this partisan shift may both reflect and enhance the power of corporations in shaping state policy.
Our findings largely confirm these expectations. We estimate that the change in campaign finance regulations after Citizens United resulted in about a 4% reduction in treated states' top corporate income tax rate and about an 8% reduction in those states that had previously banned only corporate independent expenditures. We also find significant reductions in plaintiff-friendly civil litigation standards-a change consistent with corporate interests. We find no effects for policies without clear corporate interests at stake, namely abortion, gun control, and eminent domain laws.

STATE CAMPAIGN FINANCE REGULATIONS
Studies that have relied on observational analysis of state campaign finance laws have produced conflicting results. La Raja and Schaffner (2014;, for example, found that corporate spending bans had little effect on the partisan balance of power or the success of incumbents. Werner and Coleman (2012) found that state bans on independent expenditures, and state campaign finance regulation more generally, were unrelated to the degree of liberalism among state elected officials, the state minimum wage, or the degree of pretransfer income inequality in a state.
In contrast, Hall (2016) found that corporate contribution bans increased the share of money going to Democratic candidates and the proportion of Democrats in states' upper chambers, whereas Werner and Coleman (2015) found that states with independent expenditure bans were less likely to have antitakeover statues (that help to entrench corporate management). Finally, Witko (2017) found that bans on direct campaign contributions from corporations or unions (but not caps on contribution amounts) were associated with shifts in the number of such organizations engaged in electoral activities.
More recently, a number of scholars have looked to Supreme Court decisions to assess the effect of state campaign finance laws. Using the high court's 1976 Buckley v. Valeo decision, Harvey (2019) found that removing state limits on campaign spending led to Republican electoral gains. Three studies have used Citizens United to assess the effect of removing bans on IEs. All three found increased Republican electoral success in state legislatures (Abdul-Razzak, Prato, and Wolton 2020; Harvey and Mattia 2019; Klumpp, Mialon, and Williams 2016), and Harvey and Mattia (2019) also found greater conservatism among Republican legislators.
Of the studies that based their identification strategies on Supreme Court decisions, only Abdul-Razzak, Prato, and Wolton (2020) examined policy effects and found no significant effects. But their study considered the influence of Citizens United on legislators' efforts to increase intergovernmental transfers from the state to their local governments (in support of services such as education, health care, highways, housing, etc.), where corporate interests are limited and, in many cases, countervailing. Most substantive areas of government spending, for example, benefit one industry or set of industries but require higher state taxes to pay for (e.g., health care, highways, sewage). Consequently, it is unclear that we would expect greater corporate influence over state policy to influence the sort of intergovernmental transfers that this study examined. In sum, studies that relied on observational analyses of state campaign finance laws have produced mixed results, whereas those that have based their identification strategies on Supreme Court decisions have consistently found partisan and ideological effects but no evidence of an influence on policy outcomes.

STATE POLICY, INDEPENDENT EXPENDITURES, AND CITIZENS UNITED
We focus on state policy rather than the partisan or ideological composition of state legislatures for a number of reasons. First, policy is the aspect of state government that most directly affects citizens' lives. Second, interest group behavior is directed at proximate political objectives like candidate or party success only insofar as those outcomes enhance the probability of preferred policy changes. Third, shifting the partisan or ideological composition of a legislature is only one mechanism through which interest organizations might influence state policy; potential attacks by organized interest groups at the next election might influence the behavior of incumbents of either party or ideological leaning. Finally, in line with Harvey and Mattia's (2019) findings noted above, new members elected with greater assistance from business interests may behave differently than others of their same party or general ideological orientation.

Campaign Finance Regulations and Public Policy
Prior to Citizens United, 23 states had bans on IEs by corporations or by corporations and unions (Appendix D). In states that eliminated their outside spending bans, IEs increased by 127% compared with 48% in states that had no prior bans. 2 Moreover, among states with bans on IEs prior to Citizens United, the balance of IEs between unions and corporations shifted toward corporations. In the two election cycles prior to Citizens United (2006 and2008), corporations accounted for an average of 44% of all corporate and union IEs. But in the three election cycles after Citizens United (2012, 2014, and 2016), corporations accounted for an average of 68% of the IEs from these two sources in the affected states-slightly higher than the 62% for states that had never banned IEs. 3 Thus, Citizens United produced a substantial change in the nature of IEs, bringing states that were forced to drop IE bans more in line with states that had never had such bans in the first place. Thus, we view the Citizens United decision as a natural experiment in which some states with prior bans on independent expenditures were "treated," bringing their campaign funding environments more in line with those states that had never had IE bans.
One might wonder whether the growth in IEs in the states that were affected by Citizens United was large enough to account for any subsequent policy changes. Although the 127% increase in IEs in these states was substantial, IEs by corporations and unions made up only about one fifth of the total funds raised by state legislative and gubernatorial candidates. 4 Nonetheless, what may matter the most in enhancing the political influence of corporations, unions, and other highly engaged and knowledgeable actors is the ability to increase their spending when their interests are threatened. Corporate and union actors need not spend enormous sums as long as candidates and officeholders understand that they are highly informed and could increase their spending if they felt their interests were threatened.

HYPOTHESES AND EMPIRICAL STRATEGY
We hypothesize that the effect of the ruling will be strongest for those policies that have the largest and most widespread influence on corporations' well-being and weakest for policies that are favored by Republicans and conservatives but have little or no effect on businesses. More specifically, we test the following hypotheses: H1: Citizens United will move state policy in a procorporate and pro-Republican/conservative direction (consistent with previous research showing both partisan and ideological effects).
H2: Policies that have broad and substantial effects on corporations' welfare will be most strongly affected by Citizens United, while other Republican and conservative policies will be less strongly affected.
Because unions typically provide a counter-balance to the policy advocacy of the business community: H3: These patterns will be clearer in states that had previously banned only corporate IEs than in states that had banned both corporate and union IEs.
To test these hypotheses, we rely on the generalized synthetic control (GSC) method developed by Xu (2017;, which builds upon the synthetic control approach described by Abadie, Diamond, and Hainmueller (2010). Unlike the related difference-indifference analysis, synthetic control methods do not assume pretreatment parallel trends between the treated and control units. Xu's GSC method also has the advantage of estimating a single model with an estimate for the average treatment effect on the treated (ATT) across all the treated units. Finally, Xu's GSC method reports inferential statistics such as standard errors and confidence levels.
The GSC approach estimates the effect of a treatment by constructing, for each treated unit (in our case, each state affected by Citizens United), a synthetic control unit consisting of a weighted average of the control units (that is, the states not affected by Citizens United). The synthetic units are constructed such that each synthetic control matches as closely as possible the variation over time in the treated unit during the pretreatment period. The quantity of interest, then, is the difference between the observed outcomes in the treated units during the treatment period (i.e., post-Citizens United) and the counter-factual predicted values of the outcomes for those units, based on their synthetic controls.

DATA
Our outcome measures consist of five policies that vary in the extent to which they affect corporate interests. First, we look at states' corporate income tax rate (or, for the few states with graduated corporate income taxes, the top marginal rate). 5 We expect that this policy will have the broadest and strongest association 2 These figures are based on the 15 states with consistent reporting requirements for IEs over this period (see Hunt et al. 2020;Malbin et al. 2018, Malbin et al. (2018) further coded the sources of these IEs into business, labor, and other categories (e.g., political parties, issue advocacy groups). We are grateful to Michael Malbin for making these data available. These figures reflect corporate IEs as a proportion of corporate plus union IEs, averaged across the 10 states that were affected by Citizens United that had consistent IEs reporting requirements from the 2006 through the 2016 election cycles (see note 2 above). 4 Malbin et al. 2018, Table 2. with the lifting of IE bans because all profit-making corporations benefit from lower corporate tax rates. 6 Two policy outcomes that we expect to influence corporate interests less strongly are states' civil litigation policies and eminent domain laws. Civil litigation (or tort) laws benefit corporations when they place caps on the amount of damages that courts can award to plaintiffs, when they impose higher burdens of proof for punitive damages, and when they preclude joint and several liability. Defendant-friendly tort laws benefit many corporate sectors, including health care, manufacturing, transportation, and construction. Law firms are a clear exception, but IEs by law firms represent a very small proportion of all corporate IEs. 7 Consequently, we expect corporations on average to prefer more defendant-friendly tort laws.
Eminent domain laws regulate a state or local government's ability to appropriate private property for public use. Some important corporate sectors (e.g., real estate and construction) can benefit when governments use eminent domain to facilitate private development. However, other corporate sectors have little stake in eminent domain laws or may even be harmed by government's use of eminent domain. Consequently, we expect corporations as a whole to have only weak interests in laws that restrict governments' use of eminent domain.
Finally, we examine two policy outcomes that do not involve corporate interests, and would only be affected by Citizens United through the strengthening of Republican representation and conservative ideology: abortion and gun control. 8 Our measures of tort law, eminent domain, abortion, and gun control are all multi-item indices from Sorens, Muedini, and Ruger (2008) and Sorens (2019). All of our policy outcomes are scored on a 0 to 100 scale with higher scores reflecting policies less congenial to corporations or conservatives (see Appendix A).
Our treatment variable is whether a state had bans on corporate or corporate and union independent expenditures prior to Citizens United. To maximize the fit between each of our treated states and our synthetic controls, we include as covariates Republican vote share in the last presidential election, party control of state government, state GDP, state budget deficits, total number of large firms in the state, union membership as a share of the population, and the state unemployment rate (see Appendix B). Note: Corporate tax rates rescored to range from 0 to 100, with higher scores reflecting higher tax rates. Based on the generalized synthetic control method using interactive fixed effects model. Covariates include Republican presidential vote share in the preceding election, party control of the state legislature, lagged state GDP, lagged state budget deficits, total number of large firms in the state, union membership as a share of the population, and the state unemployment rate. Table 1 and Figure 1 show the estimated effects by year for corporate tax rates for all states that had IE bans prior to Citizens United and separately for states that only banned IEs by corporations. The "effects" in the pre-Citizens United period (i.e., 2000 through 2009 in Table 1 and the years to the left of the vertical white lines at "year 0" in Figure 1) are the differences in tax rates between the treated states and their synthetic controls. If we are successful at constructing well-matched synthetic controls, these quantities should be consistently close to zero, with small standard errors, prior to treatment. As Table 1 and Figure 1 show, this is indeed the case. The estimated effect of Citizens United is revealed by the differences between treated states and their synthetic controls in the post-Citizens United period. For all of the affected states (the left side of Table 1 and left panel in Figure 1), we see a decline in corporate tax rates (relative to their synthetic controls), consistent with H1. The bottom row of Table 1 shows that for the post-Citizens United period as a whole-that is, for the years 2010 through 2016)-the estimated effect of Citizens United was −2.83 (p < 0.01), equivalent to about 4% of the average state corporate income tax during this period.

RESULTS
The right side of Table 1 and right panel of Figure 1 show results for those states that banned IEs only from corporations, not unions. Consistent with H3, the overall effect across the post-Citizens United period is about twice as large for the eight states that only banned corporate IEs: −5.55 ( p = 0.001) vs. −2.83 ( p = 0.01). For this smaller set of   states, we estimate that Citizens United resulted in about an 8% decline in corporate tax rates relative to what would have been expected based on these states' synthetic controls. Table 2 and Figure 2 show the analogous results for our four other policy outcomes (details in Appendix C). With regard to plaintiff-friendly tort laws, states that were affected by Citizens United shifted in a more corporate-friendly direction relative to their synthetic controls (−2.95, p < 0.001). Contrary to H3, the effect of Citizens United appears to have been similar in the smaller group of states that banned only IEs from corporations (−3.29, p < 0.01).
In contrast to corporate tax rates and tort laws, none of our other three policy outcomes showed significant effects of Citizens United. As Figure 2 shows, only abortion laws appear to have moved in the predicted (conservative) direction, but there is considerable uncertainty around that estimate.
These results hold under a series of robustness checks (Appendices D-H). The estimates are unaffected by alternative measures of state-level party control and hold when using an interactive fixed effects estimator in lieu of the matrix completion estimator. Neither are the results sensitive to the exclusion of potential outliers. None of these procedures produce results that differ significantly from the full set of treated states based on the ATT over the whole post-Citizens United period or in the most recent year of our data (2016).
Taken together, the results for the five policy outcomes we examined are consistent with H1 and H2. Policies in states that were affected by Citizens United did shift in a pro-corporate/conservative direction (H1), but the only statically significant effects were found for the policies with the strongest and broadest hypothesized influence on corporations' well-being (H2). Our expectation that the effects would be stronger for states that only banned corporate IEs (H3) was clearly born out for corporate tax rates but not for states' tort laws.

DISCUSSION
Regulations on IEs by corporations and unions are only one small facet of states' campaign finance regulations, which also include limits on contributions to candidates and campaigns, disclosure requirements, various forms of public financing, and so on. While the exogenous nature of Citizens United is advantageous in establishing the causal effect of IE bans, the influence of the court's decision is likely to be rather modest within the broader scope of state campaign finance laws. But if this modest change in the legal avenues through which corporate (and, in some cases, union) actors can work to shape politics nevertheless had a discernable effect, more dramatic differences in campaign finance regulations are likely to have greater effects. In this light, our estimates of the effect of Citizen United, although modest in size, nevertheless suggest that campaign finance regulations are important drivers of policy in the American states.
Our findings, and those from other studies using similar identification strategies, strongly suggest that campaign finance laws do matter. Moneyed interests may always have an advantage in shaping election outcomes and public policies, but the ease with which money can be transformed into political influence does appear to shift depending on the restrictions that our campaign finances regulations impose and the opportunities for influence that they allow.

SUPPLEMENTARY MATERIALS
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0003055421000149.

DATA AVAILABILITY STATEMENT
Replication files are available at the American Political Science Review Dataverse: https://doi.org/10.7910/ DVN/082VIU.