Introduction

Scholarship on the effects of campaign expenditures in congressional elections is a notoriously thorny area of study. Indeed, while the scholarship is mixed on the effects of campaign spending for the challenger and incumbent respectively, the only consensus seems to be that campaign expenditures matter (Jacobson 2013). As congressional campaigns increase in cost, questions about the effectiveness of campaign spending has coupled with questions surrounding outside group expenditures.

For Example, in 2014, North Carolina saw the most expensive Senate contest in the nation’s history, totaling over $111,000,000 (Wallack and Hudak 2014). While a portion of this is due to the rising costs of running a campaign in the modern era, many blame the rising cost on the 2010 Citizens United v. FEC Supreme Court decision, allowing for broader sources for funds for outside spending groups. These groups, armed with the ability to raise unlimited amounts of money from corporations and unions, among many other sources, are now seen as having a larger role in the election process. In North Carolina, the incumbent Democrat Kay Hagan and her Republican opponent Thom Tillis spent $24,849,812 and $10,513,963, respectively – meaning the lion’s share of the amount spent in North Carolina came from these independent expenditure groups (Federal Elections Commission). Indeed, North Carolina was not an isolated case as the Senate races in Colorado, and Iowa cost nearly $100,000,000 a piece, and the races in Kentucky, Georgia, Alaska, Arkansas, and New Hampshire all cost more than $50,000,000 (Federal Elections Commission). Even more impressive is the amount of money spent per voter in these states where a single vote cost $120.59 in Alaska, $50 in New Hampshire, and $39.11 in Iowa (Wallack and Hudak 2014).

Of course 2014 should not be surprising, as the cost of campaigns has increased steadily since 2010. For example, in 2010, the most expensive Senate campaign was Florida’s Senate seat, which cost $84,300,000. Compare that to 2012s most expensive race, Virginia, which cost $87,900,000 and the aforementioned case of North Carolina at over $111,000,000 in 2014. The rising expenditures by both independent groups as well as campaigns raise the question about the effectiveness of money in elections. Farrar-Myers and Skinner (2012) find that independent expenditures have a statistically significant, though marginal effect on election results in the House of Representatives. Yet given the increased value in a Senate seat and the ability for outside groups to coalesce around a handful of races, it remains to be seen whether outside groups can effectively change the nature of senate elections.

Much of the scholarship on outside group spending and their effects on the modern era of American elections is still emerging. Given the increase in spending in Senate campaigns, we seek to answer what effect if any, outside group expenditures have on Senate campaigns in the post-Citizens United era. Specifically, we examine the effect of outside group spending made on behalf of campaigns and their effects on the overall vote share for United States Senate campaigns. We find that in general, outside spending in support of the candidate has a positive effect on the vote share for both challengers and incumbents. We also find that outside group spending against the challenger is effective in decreasing the vote share for the challenger but outside group spending against the incumbent is ineffective.

This article proceeds as follows. We begin with a discussion of campaign finance law and the scholarship surrounding its impact on the electoral process. We then explain our data and methodology. In the following section we discuss our results and we end with conclusions and suggestions for future research.

Campaign Finance and Congressional Elections

Voters in congressional elections often begin the electoral puzzle with little to no knowledge of the candidates or even contests involved. Naturally, this lack of knowledge yields an opportunity for campaigns and independent, “big money” groups to have a large impact. The breath of literature on congressional campaigns reveals that campaign spending does indeed matter (See Squire 1995 and Jacobson 2013 for summaries). Still, there is some disagreement in the literature as to the effects of campaign spending for the incumbent and challenger in which Abramowitz (1988), Ansolabehere and Gerber (1994), and Jacobson (1985, 1990, 2013) all find that challenger spending matters and incumbent spending does not, while Green and Krasno (1988, 1990), Erikson and Palfrey (1998), Goldstein and Freedman (2000) all find that both challenger and incumbent spending have an effect. Regardless of which side benefits, and for that matter, by how much, there is clearly an incentive to spend in congressional elections. This incentive to spend has been exacerbated by recent Supreme Court decisions that have seemingly increased independent spending by wealthy individuals as well as non-party groups and committees. Campaign finance laws restrict an individual, corporation, or union’s ability to impact elections, yet the recent Supreme Court decisions have allowed for a new avenue for individuals and groups to spend unlimited amounts of money in favor or against candidates in the form of outside spending.

As held in the Supreme Court case Buckley v. Valeo, 424 US 1 (1976), independent expenditures by groups or individuals cannot be limited as long as they do not coordinate with a candidate, as it would violate a person’s right to express their free speech. However, such expenditures have not received much attention prior to 2010. Scholarship on independent expenditures during the 1980s to the 2000s is fairly limited. In their study examining the effects of independent expenditures on senate elections outcomes from 1984 to 1999, Engstrom and Kenny (2002) find independent expenditures have a significant effect on election outcomes, but that this effect is largely dependent on electoral context, the candidate, and the type of spending. Furthermore, Skinner (2005) found that while 527-group activity did look quite similar to that of political parties, they fell short of eschewing the two major parties in power and activity.

The court’s 2007 decision on Federal Election Commission v. Wisconsin Right to Life, allowing independent groups to run issue ads during election years even past the 60 day cutoff – provided they do not say the magic words, “vote for” or “vote against,” further provided an avenue to impact the electoral process for wealthy individuals who were maxed out with their campaign contributions (Medvic 2014).

While this certainly gives a strong description of the nature of independent expenditures, the Supreme Court case Citizens United v. FEC changed the proverbial game. In January of 2010, the Supreme Court issued a 5-4 decision ruling that corporations and unions cannot be prohibited from influencing money as long as the money was spent was independent expenditures, or not in concert with a candidate. Shortly thereafter, the Supreme Court’s decision on Speechnow.org v. Federal Elections Commission integrated the Citizens United decision to conclude that individuals could effectively bankroll outside spending groups with unlimited amounts of contributions. As a result, the following 2010 midterm elections and the 2012 general elections saw skyrocketing expenditures by these independent groups (Farrar-Myers, Gulati, and Skinner 2013). Indeed, our own analysis of independent expenditures demonstrates that this trend extends beyond 2012 as outside spending increased again in 2014, as demonstrated in Figure 1.

Figure 1:

Average Outside “SuperPAC” Expenditures from 2010–2014.

When compared to campaign spending, outside spending steadily grew from 2010 to 2014 as a larger share of the overall amount of money spent in Senate races. Figure 2 demonstrates the increase in the average campaign expenditures compared to the average outside expenditures.

Figure 2:

Average Campaign Spending Compared to Average Independent Expenditures.

In fact, amongst Republicans in 2014, outside expenditures were on average, greater than campaign expenditures. It is also interesting to note that there is little partisan advantage with regards to outside spending outside of 2010, in which Republicans had a slight advantage over Democrats. 2010 and 2014 on average, were very close to equal.

Not only did spending increase, but the way in which outside groups spent their money changed as well. Chand (2014) finds that the Citizens United decision incentivized groups to forgo traditional PAC spending patterns of contributing to candidates in favor of purchasing their own ads. Finally unleashed from their legal limits, wealthy individuals seeking to impact the electoral outcomes of these races contributed the lion’s share of donations to independent expenditure groups after the 2010 decisions (Farrar-Myers and Skinner 2012; Herrnson 2013). This new participation favored wealthy individuals as it not only allowed for unlimited contributions beyond the legal limits to candidates, parties, and political action committees (PACs), but it also provided a shroud of secrecy as there are limits to the disclosure laws surrounding these groups (Herrnson, Deering, and Wilcox 2013). Interestingly, while the judicial decisions did provide wealthy individuals, corporations, and unions an extra avenue for financial participation, only wealthy individuals and unions took advantage as Gulati (2012) notes that businesses held back such contributions in order to preserve their corporate, non-political identity.

Still, there are questions about the effects of this new avenue of participation. The key difference between campaign spending and outside spending is that outside spending is independent of the candidate. This added component comes with difficulties as outside spending groups are in the dark when it comes to candidate messaging and strategy. Offering a pessimistic view for proponents of these groups, Washington Post reporters Eggen and Farnam (2012) find that independent expenditure groups had little impact in the 2012 election. In her work on interest groups, McKay (2010) finds that ideologically extreme groups are more likely to spend on elections than lobbying which may contribute to the confusion in messaging. Indeed, Hansen et al. (2015), find that most businesses sat out in terms of giving to independent expenditure groups, suggesting that more ideologically extreme efforts may be in play.

Still, Franz (2011) finds that in the 2010 election, the first election after Citizens United, interest groups were much more limited in their power than anticipated. Using data on political ads run during the 2010 election, he finds that not only did the ads do little to impact the outcomes of House and Senate races, but they also were not nearly as negative as expected either. Legal scholars suggest that 2012 offers only a “taste” of the future as Super PACs will grow in their reach and influence (Kang 2013). More empirically rigorous work by Farrar-Myers, Gulati, and Skinner (2013) demonstrated a statistically significant though marginal effect for independent expenditure groups in the 2010 and 2012 House elections.

Following all of this, we offer two hypotheses:

H₁: Indpendent expenditure groups will have a small, though statistically significant effect on the candidate’s vote share.

Given the fact that strategy and candidate messaging do have some importance, (Jacobson 2006), independent expenditure groups are more likely to fall short in terms of effect as they may miss the mark in messaging, especially given the extremism of the groups that participate monetarily. That is, since campaigns and independent groups cannot strategize together, the amount that independent groups can help the campaign’s messaging and overall campaign is limited. As a result, we should expect to see the candidate with the largest amount of independent expenditures in her favor with an advantage over her opponent in terms of vote share.

We also hypothesize that independent expenditure groups will affect the incumbent and the challenger differently. We hypothesize:

H₂: Outside Spending will benefit the challenger more than the incumbent.

Jacobson (2006) makes a compelling argument for the expectation of greater marginal returns on campaign spending for the challenger, arguing that incumbents should begin the campaign with a large advantage in name recognition and familiarity with voters. Indeed, he argues that it us up to the challenger’s campaign to argue that the incumbent should be replaced, and replaced by her (Jacobson 2006). As a result, we should expect to see outside spending have a stronger effect on the challenger’s vote share over the incumbent’s vote share. In the following sections we test our hypothesis and discuss the implications of the findings.

Data

To examine the effects of outside spending on Senate races, we collected data on campaign and outside group expenditures from the Center for Responsive Politics. ^[1] One problem in studying the effects of outside spending is that the term “outside spending” seems to be used synonymously with “SuperPAC,” however, the term covers spending from independent expenditure only groups (SuperPACs), traditional political action committees (PACs), which are regulated by the FEC, 501(c) non profit groups, 527 groups, corporations and unions, and political party committees. In this paper however, we are concerned with outside spending that deals exclusively with groups that are not regulated by the FEC. As a result, we exclude political party committees and traditional PACs.

Still, it is difficult to verify the overall validity of these numbers. Indeed, even the collection of these data may include or leave out money that is vital to the study of these types of expenditures. That is to say, because these funds are “dark” and somewhat unaccounted for, it is difficult to say that the numbers reported and collected by Open Secrets are truly the total funds used in a particular campaign. As a check to verify that the use of these data does reflect outside spending (excluding party committee and traditional PAC spending), we gathered data from the Wesleyan Media Project (WMP) for 2010 and 2012. ^[2] The WMP gathers data on every political ad run in these election cycles. Of primary interest to this study, one of the variables coded in the dataset is who funded the ad – the candidate, the party, coordinated between the candidate and party, or an interest group or other group. When we correlated the amount of money spent by outside groups in favor of a candidate with the number of ads run in favor of a candidate by an interest group or other group, the two variables were correlated at 0.79. While this is not a perfect correlation, we suspect that ads capture nearly all of the spending reported by these groups, but not necessarily all aspects of spending (i.e. signs, consultants, etc.).

There are still however, many problems with studying the effects of expenditures here. Indeed, we are not necessarily interested in the overall effect of the amount of spending. That is, if one candidate had groups spending millions of dollars to support her, it is likely that the other candidate also had millions of dollars supporting her as well. Similarly, competitive races are likely to have greater amounts of spending than non-competitive ones. That is to say, while spending has increased across the board, using raw spending amounts would bias the model and the results as competitive races as well as big state races had, on average, much more spending than less competitive and small state races. As a result, we focus on the ratio of outside spending in support of a candidate (calculated as the amount spent in support of a candidate divided by the total spent in support of either candidate in that specific electoral contest) and the ratio of outside spending against a candidate (calculated as the amount spent against a candidate divided by the total spent against either candidate in that specific electoral contest) as our two independent variables of interest. By focusing on ratios, we can see the impact of the volume of spending, avoiding, to some extent, endogeneity issues that so often plague these types of models.

We use these data to examine their effects on the vote share of the candidate. To do this, we run multiple fixed effects regression models. First, we examine the effects of the spending on the incumbent’s vote share. Since we only have 50 incumbents in our dataset, as a robustness check, we also examine the effects of spending on the challenger’s vote share. Given the partisan advantages that the Republicans had in 2010 and 2014 and the Democratic advantage in 2012, we account for these issues using a fixed effects model account for the year of the election.

We also include a number of control variables. We control for campaign spending as much of the literature on congressional elections demonstrates a clear importance on this type of spending. This is calculated in the same way we calculated the outside group spending (the ratio calculated as campaign expenditures by the campaign divided by the total expenditures in that specific electoral contest). As much of the literature demonstrates, we expect to see campaign spending to be more effective for challengers than incumbents, even when outside spending is taken into account. We also control for the quality of the candidate using a dummy variable in which a “1” corresponds to a candidate that has held statewide office (i.e. governor, secretary of state, etc.) or is a US House of Representative, and a “0” corresponds to all other candidates for the model examining challenger vote share. We assume here that higher quality candidates, coded here as a “1,” are likelier to gain support than the lesser known or lower quality candidates. We control for the ideology of the state by including a variable for Barack Obama’s vote share in 2008. We assume here that Republicans from states in which Obama did not do well will receive a higher vote share while Democrats from states in which Obama did well will receive a higher vote share. Finally, we control for party by using a Republican “dummy” variable in which a “1” corresponds to Republican candidates and a “0” corresponds to Democrats. In the following section, we review the results of our analysis.

Results

A first, descriptive cut of the data of the total outside group expenditures spent in favor of a candidate, calculated as the total expenditures spent in support of a candidate added to the total expenditures spent against the opposition, plotted against the vote share for the candidate demonstrates that there is indeed a bivariate relationship between outside group spending and a candidate’s vote share. This is illustrated in Figure 3.

Figure 3:

Ratio of Outside Spending Against the Candidate vs. Candidate Vote Share.

Still, the relationship is relatively weak in that the correlation is 0.47. When we separate the two forms of spending as the ratio of outside spending against a candidate and the ratio of outside spending for a candidate we get more nuanced bivariate results. Indeed, Figure 4 demonstrates almost no bivariate relationship between the ratio of outside spending against a candidate and said candidate’s vote share.

Figure 4:

Ratio of Outside Spending in Support of the Candidate vs. Candidate Vote Share.

Indeed, the linear estimation shows a positive relationship between outside spending against a candidate and that candidate’s vote share. Naturally this makes little sense, as outside spending against a candidate should decrease the candidate’s vote share. This suggests a very weak relationship between outside spending against a candidate and the ability of that spending to decrease a candidate’s vote share, at least in terms of a bivariate relationship.

The relationship between the ratio of outside spending in favor of a candidate and the candidate’s vote share on the other hand, demonstrates a much stronger correlation at 0.64 and an r-squared value at 0.4. This is illustrated in Figure 5.

Figure 5:

Ratio of Total Expenditures Spent in Favor of a Candidate vs. Candidate Vote Share.

This suggests that the cache that outside spending groups have seems to be in support of a candidate rather than the conventional wisdom of attacking a candidate, at least for Senate races. Still, this is only bivariate analysis that requires further rigorous testing to verify the results.

The results of the multivariate regression analysis are presented in Table 1.

Consistent with much of the literature on the effects of campaign spending, incumbent campaign spending did not have a significant effect on the incumbent’s vote share. Also consistent with expectations given that the tides were in favor of the Republicans in 2010 and 2014, Republican candidates had a five-percentage point advantage over Democrats. Most interesting to the study at hand however, is that the ratio of outside spending in support of the incumbent is statistically significant and positive. That is, for a change in the minimum value to the maximum value for outside spending, the expected vote share for the candidate increases by approximately thirteen percentage points.

Still, this analysis includes races in which incumbents may have received the benefit of all of the outside spending. As a result, we ran a second model limiting the analysis to races in which the total outside spending ratio was less than 0.75. We again see similar results though the effect of the ratio of outside spending in support of the candidate increased. While this is a more conservative model, we did lose ten observations out of our initial fifty, making more rigorous analysis difficult. Nevertheless, we did run a third fixed effects regression model predicting the vote share for challengers that yielded a larger number of observations with 105. The results of this model are presented in Table 2.

Again, consistent with previous research on the effects of campaign spending for the challenger, the ratio of campaign spending has a positive and significant effect for the challenger’s vote share. Similarly, the ratio of outside support for the challenger has a significant and positive effect on the challenger’s vote share. Different in this model is that the ratio out outside spending in opposition of the challenger has a negative and statistically significant impact on the challenger’s vote share.

We should note however that the coefficient for campaign spending is nearly five times greater than the coefficient for outside spending against the challenger. Furthermore, it is also nearly two and a half times greater than the coefficient for outside spending in favor of the challenger. This suggests that while both outside group spending in support and opposition for challengers is significant, it is offset by campaign expenditures. In addition, it also suggests that campaign spending is more effective than outside support. Again, as mentioned earlier, this is likely due to the fact that federal campaign regulations prohibit outside groups from coordinating with the candidate. Since many of the contributors and organizers of outside groups are ideologically extreme, their expenditures may appeal to only a certain portion of the population.

As with the incumbent model, we ran the challenger model again, limiting our analysis this time to those races in which the total ratio of outside spending was less than 0.75. Again we see similar results, though with some meaningful differences. The ratio of campaign spending is again positive and significant, though the coefficient is greater than the previous model by approximately 2. The ratio of outside spending in support of the challenger is also still positive and significant, but the coefficient is now smaller by approximately 4. This again lends evidence to the idea that outside spending, while effective, is less effective than campaign spending. ^[3]

When we compare the results to the previous model predicting the vote share for incumbents, there are also noteworthy differences. First, the ratio of outside support for incumbents is greater for incumbents than it is for challengers. Second, the ratio of outside opposition spending against the incumbent is not statistically significant, while the ratio of outside opposition spending against the challenger is statistically significant. Third, consistent with the literature on campaign spending, challengers are benefited by campaign expenditures while incumbents are not.

These differences may exist for a number of reasons. First, as the literature on campaign expenditures argues, one of the reasons that campaign spending benefits challengers over incumbents is due to name recognition, or the lack thereof. That is to say, since incumbents are relatively well known as compared to their challengers, challengers have more to gain from campaign expenditures. Similarly, since voters are less familiar with challengers, outside group attacks may resonate better. Still, we see that outside group spending in favor of the incumbent has a greater effect than outside group spending in favor of the challenger, though the difference is relatively small. Again, since outside groups may be more ideologically extreme, they may appeal to voters than incumbents are not targeting, or base voters who are excited by ideological measures. Still, given the limited data that we have access to following the aforementioned Supreme Court decisions, further research with data from future elections is still needed.

Conclusion

As mentioned at the onset of this study, understanding the effects of money on electoral politics is a notoriously thorny endeavor. This is especially true for Senate races as the number of observations is restricted as compared to House races. Nevertheless, in this paper we have moved our understanding on the effects of outside group spending on Senate campaigns incrementally forward. Specifically, we found that outside groups did have an effect in the Senate campaigns in 2010, 2012, and 2014. Outside group spending in favor of the candidate had a statistically significant and positive effect for both incumbents and challengers while outside group spending against the candidate had a statistically significant and negative effect for the challenger. Counter to our expectations however, the outside group expenditures had a greater impact on the incumbent’s vote share than the challenger’s vote share. Again, we think this may be because the challenger has less to gain from outsider groups who may offer ideologically extreme information in favor of the candidate as compared to the incumbent.

We also find that campaign spending for the challenger had a greater impact than either the outsider group expenditures against the challenger or the outsider group expenditures in favor of the challenger. This is probably due to the fact that the outsider groups cannot coordinate with the campaign, making their messaging less effective or even off base when compared to the campaign’s messaging.

These results raise a number of questions. First, the normative literature on money in politics has long questioned the potential for corruption. Given that these results demonstrate that outsider group spending does have a positive, albeit, small effect on a candidate’s vote share, it raises questions as to the incentives for individuals to spend independently, outside of the campaign system. In a similar sense, it also raises questions as to the equality of participation opportunities in American politics. That is, if contributions to outside groups are effective, then it offers yet another opportunity to impact congressional elections for wealthy individuals.

Outside of the normative concerns, how individuals are evolving as a response to the Supreme Court decisions is of continued interests. Indeed, we must exercise caution in using this study as the final measure of the effects of these expenditures. As Kang (2013) argues, the results in this study may simply be a result of the fact that the phenomenon is in its infancy and only has three elections under its belt. Still, it is difficult to dismiss the results entirely as the shortcomings that outside groups experience are not insignificant and are unlikely to change in the immediacy, especially those prohibiting contact with the candidate and her campaign.