Analyzing Campaign Contributions in Context

Abstract

Political campaign contributions are not made in a vacuum. Rather, contributions from individuals are affected by the political and legal environment experienced by potential donors. I study itemized contributions to candidates, parties, and political action committees (PACs) aggregated by congressional district for 1994-2006. The timing and competitiveness of presidential, congressional, and even state elections affect the expected benefits and opportunity costs of contributing to all types of committees, and this is reflected in the amounts contributed through different channels. Political party committees raising hard money contributions relied more on districts with high per capita income, education attainment, and urbanization after the Bipartisan Campaign Reform Act (BCRA) than before, but were still less dependent on these districts than soft money committees before the BCRA. Overall, the BCRA led to less reliance on districts with high socioeconomic status as sources of itemized contributions to committees regulated by the Federal Election Commission.

Keywords

campaign finance political parties Bipartisan Campaign Reform Act political contributions political action committees

It is well known that monetary contributions by individuals to political campaigns are a function of potential donors’ personal characteristics and whether they are asked to contribute (e.g., Brady, Schlozman, &Verba, 1999; Grant & Rudolph, 2002; James, 2009). Recent research using aggregate data has established that spatial context also matters, as evidenced by the effects of spatial autocorrelation coefficients and population density (e.g., Cho, 2003; Cho & Gimpel, 2007; Gimpel, Lee, & Kaminski, 2006). Yet there are at least two other kinds of context that should matter: political and legal. The timing and competitiveness of specific elections directly affect the amounts contributed to candidates who are running in those races, but should also have indirect effects on the amounts contributed through other channels. The most obvious, but not the only reason for this is that money is scarce and a dollar contributed to one campaign committee cannot be contributed to another committee. Legal regulations allow different maximum contributions to different types of committees, but more importantly a change in regulations alters the amounts that can be contributed as well as the incentives for candidates, parties, and their supporters to raise money through different channels. This, in turn, should affect the total amounts contributed through different channels as well as the extent to which different kinds of committees rely on donors with high socioeconomic status (SES).

The research on political campaign contributions nonetheless gives little attention to the effects of political and legal context on money flowing through different kinds of fundraising channels. Most existing studies of campaign contributions are surveys that ask whether each respondent has contributed to a political committee—any political committee—within a specified time period (e.g., Brady et al., 1999; Grant & Rudolph, 2002; James, 2009. But see Jones & Miller, 1985), or that focus on a specific subset of donors such as those who give to congressional campaigns (Francia, Green, Herrnson, Powell, & Wilcox, 2003). Similarly, studies of contributions aggregated by geography either focus on total dollars contributed to all federal committees (Cho, 2003; Gimpel & Lee, 2006; Gimpel et al., 2006) or one specific channel such as contributions to U.S. House candidates running in a different district (Gimpel, Lee, & Pearson-Merkowitz, 2008).

In this article, I analyze and compare itemized contributions made to a wide variety of federal committees controlled by or affiliated with candidates and political parties aggregated by congressional district during the 1994-2006 election cycles. These include contributions to campaign committees that support specific congressional and presidential candidates, party committees raising contributions subject to contribution limits (so-called “hard” money), “soft” money committees through 2002, and political action committees (PACs) that are not sponsored by a business firm, association, or labor union. I examine the effects of the timing and competitiveness of congressional, state, and presidential elections on contributions made through these different channels, as well as the extent to which contributions tend to come from districts with high per capita income, education attainment, and urbanization. I also examine how the determinants of contributions changed after the virtual elimination of soft money and the increase in hard money contribution limits due to the 2002 Bipartisan Campaign Reform Act (also known as McCain-Feingold; hereafter referred to as BCRA).

I find that contributions to many types of federal committees are affected by competition for federal and even state offices. The competitiveness of congressional races is, of course, a major determinant of contributions to candidates from donors residing in contested districts but state, district, or local party committees may also raise more money from competitive congressional districts, whereas most other types of federal committees raise less. Some federal committees also raise less money from districts in states that hold concurrent state and federal elections, while all types of party committees and some types of PACs raise more money during presidential cycles. While there is some evidence of competition for scarce contribution dollars, the overall result is that more money is contributed to federal campaign committees in total when there are more federal races and more competitive races. This is consistent with the claim that the major political parties and their allies act in concert to raise more money when the stakes are greatest (Herrnson, 2009), while leaving unanswered the question of what other types of activities compete with federal elections for potential donors’ scarce resources.

All types of federal committees raise more money from districts with higher SES as measured by per capita income, education attainment, and urbanization, but the magnitude of this effect varies substantially. It is smallest for contributions to local congressional candidates and state, district, or local party committees, and largest for contributions to nonlocal congressional candidates and PACs. The effect on contributions to national and Hill party committees and presidential campaigns lies in between.

Following implementation of the BCRA, party hard money committees relied even more on high-SES districts (but still less than soft money committees did before the BCRA), whereas PACs were less reliant on these same districts. The last result may reflect increases in the number of PACs rather than a shift in fundraising strategies. The overall effect of the BCRA was to reduce reliance on high-SES districts as sources for itemized contributions to committees regulated by the Federal Election Commission (FEC). Finally, while the major independent effect of the change in laws was an increase in hard money contributions, there are two exceptions: state, district, and local party committees, and leadership PACs. The first exception provides indirect confirmation that the primary purpose of soft money was not to support local party building activities.

Taken as a whole, my results demonstrate that while contributions to various types of federal committees share many of the same determinants, the expected benefits and opportunity costs of contributing through different channels are affected in different ways by the political environment where the donor lives. Moreover, a significant change in campaign finance regulations can alter the strategies used by campaign committees and the extent to which they rely on districts with high socioeconomic characteristics for the bulk of their contributions. Would-be reformers should consider these distinctions when thinking about which types of fundraising channels should be encouraged and the full effect of proposed changes in regulations.

The next section describes the different channels for political contributions included in this study and how these contributions may be affected by differences in political and legal context. The following section presents the empirical analysis: First, regression models of itemized contributions made through different channels are estimated, and then the predicted contributions under different scenarios are calculated to determine the effects of the political environment and changes in legal regulation. The final section summarizes the results and discusses their implications.

Different Channels for Federal Contributions

Campaign contributions from individuals can flow through many channels subject to regulation by the U.S. FEC. These include individual candidates’ campaign committees, political party committees, and PACs. Each of these types of committees is subject to different rules governing maximum contributions and, in the case of certain PACs, allowable solicitation practices (U.S. Government Printing Office, 2014).

Each of these broad types of federal committees can be subdivided further. Candidate committees obviously break down into presidential and congressional committees, and contributions to congressional candidates may be further subdivided into those made to candidates running to represent the district or state where the donor lives—what I will call “local” congressional candidates—and those running to represent some other district or state—“nonlocal” candidates. The same laws apply to both types of contributions and the same candidate committees receive both types. However, these two channels correspond to different forms of representation (Mansbridge, 2003), and previous research has shown that contributions to nonlocal House candidates tend to originate from a relatively small number of districts characterized by high SES (Gimpel et al., 2008).

Party committees can be broken down into national committees, the so-called Hill committees,¹ and all other state, district, or local—referred to hereafter as “S/D/L”—committees. Although national party committees can support all federal candidates with direct contributions and coordinated spending, they typically concentrate on the presidential campaign during presidential cycles, and may also spend money independently and transfer funds to state and local parties (Corrado, 2006; Herrnson, 2009; La Raja, 2008). Hill committees raise money to support candidates for the House and the Senate (Dwyre & Kolodny, 2006), and Gimpel et al. (2008) argue that they play an important role in soliciting and directing contributions by individuals to nonlocal congressional candidates as well. S/D/L party committees support federal candidates running locally as well as activities such as get-out-the-vote drives (La Raja, 2006). Prior to the 2004 cycle, party committees could also receive so-called “soft” money in unlimited amounts for the purpose of party building activities, but lax interpretation of the regulations and repeated pushing of the envelope (such as mixing hard and soft money and purchasing “issue ads” that advocated for a candidate without mentioning them by name) by both major parties led to the 2002 BCRA, which essentially eliminated soft money² and increased individual contribution limits for hard money (Malbin, 2006; Mann, 2003).

PACs can be subdivided according to whether they are sponsored by a parent organization such as a corporation or other business firm, labor union, or trade/membership/health association. These sponsored PACs are excluded from this analysis, as they can only solicit contributions from individuals who have a preexisting relationship with the sponsoring organization (U.S. Government Printing Office, 2014). Nonconnected PACs may be subdivided into issue PACs that focus on specific areas such as environmental policy, partisan or ideological PACs, and leadership PACs created by individual politicians for the purpose of furthering their own careers (Center for Responsive Politics, 2013; Currinder, 2003). Herrnson (2009) refers to leadership PACs as “party-connected” committees and notes that other PACs often act as party allies. Lowry (2013) compares contributions to sponsored and nonconnected PACs. Donors to sponsored PACs rarely make itemized contributions to other types of federal committees and their contributions are essentially unaffected by competitive congressional races. The effects of presidential cycles and the BCRA on contributions to sponsored PACs are statistically significant but substantively much smaller than their effects on contributions to nonconnected PACs.

There are still other channels available to individual contributors that are not subject to contribution limits enforced by the FEC. These include so-called “527 organizations” that became popular following the elimination of soft money (Weissman & Hassan, 2006) and nonprofit social welfare (501c4) organizations that have become favored vehicles following the Supreme Court’s decision in Citizens United v. Federal Election Commission (2010). And of course individuals can also make unlimited independent expenditures on behalf of their preferred candidates (U.S. FEC, 2014a). Political organizations that do not contribute to federal candidates or parties and do not expressly advocate for or against federal candidates or make electioneering communications are not required to register with the FEC, but are required to file the names of itemized contributors with the Internal Revenue Service (IRS). However, IRS’s online database does not include political organizations that used paper forms rather than filed electronically prior to 2012 (U.S. IRS, 2014). Assembling a data set of itemized contributions to all such organizations would be a very arduous task. Social welfare organizations have recently become popular vehicles for fundraising precisely because they need not disclose the names of their donors (Maguire & Novak, 2013).

It is unlikely that any analysis of political contributions by individuals will ever be truly comprehensive, given that many fundraisers and donors are constantly seeking new ways to skirt or evade regulations and disclosure requirements (see Mann, 2003). The effect of these omissions on my analysis is greatest in the 2004 and 2006 cycles, as money raised by political and social organizations not regulated by the FEC increased dramatically following the elimination of soft money (Weissman & Hassan, 2006). It therefore seems likely that contributors to these organizations most closely resemble contributors to soft money committees prior to 2004. The effect of the expansion of these alternative channels on contributions to the committees in my data set is subsumed within the overall effect of the BCRA.

Trends for 1994-2006

Data on itemized contributions—those made by individuals giving at least $200 to a committee during an election cycle (U.S. Government Printing Office, 2014)—were obtained from the U.S. FEC’s (2014b) website. These contributions are aggregated by congressional districts³ to distinguish between contributions to local and nonlocal congressional candidates and test for the effects of the local political environment. Data for the 1994-2006 election cycles are used to explore the effects of presidential cycles and a major change in campaign finance laws.

One limitation of these data is that they do not include contributions from donors who give less than $200 during an election cycle to a given committee. Data collected by Malbin, Dusso, Fortelny, and Gavin (2011) show that these “small” donor contributions made up 27.8% of all dollars contributed by individuals to congressional candidates, 45.2% to “Hill” party committees, and 49.0% to national party committees during the 2000 and 2002 cycles. After individual contribution limits were raised by the BCRA, these percentages declined to 20.2, 35.8, and 41.3, respectively, during the 2004 and 2006 cycles. Lowry (2013) shows that in the 2004 election cycle, small donor contributions made up more than half of all dollars contributed by individuals to issue and partisan PACs, but only about one quarter of contributions to leadership PACs. While the findings below do not apply to small donor contributions, it is reasonable to conjecture that differences in SES have less effect on small donor contributions per congressional district than on itemized contributions.

Figure 1 shows the average real itemized contributions per district made through various channels during 1994-2006 (contributions from Washington, D.C., and outside the United States are excluded). The data show a clear trend toward the nationalization of campaign finance. Contributions to local congressional candidates were the largest single category in 1994, accounting for 33.1% of the total. They were exceeded by contributions to nonlocal congressional candidates beginning in the 2000 presidential cycle, and by 2006, contributions to local candidates accounted for just 25.4% of the total.

Figure 1.

Average itemized contributions per congressional district, 1994-2006.

The figure also shows that the elimination of soft money following the 2002 cycle had little effect on total itemized contributions from individuals. Most of the loss was offset by increased fundraising by party committees and to a lesser extent PACs and nonlocal congressional candidates. In addition, itemized contributions to presidential candidates in 2004 greatly exceeded those in 2000. Although the percentage increases during 1994-2006 for contributions to various kinds of PACs are impressive—381% for partisan PACs, 619% for leadership PACs, and 92.5% for issue PACs—total fundraising by nonconnected PACs remained a relatively small part of the overall picture.

The Effects of Donors’ Political Environment and Legal Context

It is well known that people are more likely to donate when they are asked to do so (Brady & Verba, 1999; Grant & Rudolph, 2002), and likely donors may be solicited by multiple committees of various types. The local political environment can affect a prospective donor’s decision whether to contribute and to whom both directly and indirectly. The direct effect is that the expected benefits of donating to a particular committee are greater if that contribution may affect the outcome of an election. Gimpel et al. (2008) note that individual donors may give for various reasons including partisan motives, a desire to obtain access to lawmakers, a desire to show support for candidates who express strong ideological views, and a desire to show support for candidates from particular groups defined by race, gender, or some other attribute. Francia et al. (2003) use a more general typology of material, purposive, and solidary motives.⁴ With the possible exception of solidary motives, the expected benefits of a contribution should always be greater if it might actually influence the outcome of an election. In addition, local candidate and party committees have incentives to work harder to solicit contributions if the outcome of the election is in doubt. Thus, potential donors are more likely to give to local congressional candidates if they live in a competitive district. Conversely, it may not matter whether a potential donor resides in a battleground state during a presidential election year: Potential donors nationwide have an interest in the outcome and know that national party committees and presidential campaigns can direct resources to where they are needed most.

The political environment can also affect contributions indirectly because money is scarce, and the opportunity cost of a contribution made to one committee includes the benefits foregone from contributing those dollars to other committees (Varian, 1987). Thus, donors living in a district with a competitive race for House or Senate face a higher opportunity cost of giving to anything but their local candidate than donors in a noncompetitive district. However, the effect of this difference depends on whether the potential donor faces a binding budget constraint for all political contributions. Gimpel et al. (2008) conclude that donors in districts with contested local races simply “dig deeper into their pockets to finance both the local race and . . . high-profile races around the country” (p. 389). This is possible because the set of activities competing for potential donors’ discretionary spending may be larger than federal political campaigns.

In addition, the joint effect of political environment and legal regulations means that contributions to some committees may be complements rather than substitutes. For example, a donor who contributes the maximum legal amount to a presidential candidate may then make a contribution to the national party committee to provide further support for the campaign. A similar dynamic could apply to contributions to congressional candidates and Hill party committees. Finally, an intense campaign for president or Congress might stimulate interest and contributions to other committees as well, although we know very little about whether making a campaign contribution leads to other acts of political participation (Wilcox, 2001).

It is likely that each of these scenarios applies to at least some individuals during any given electoral cycle, and my analysis of data aggregated by congressional districts cannot be regarded as a test of any particular hypothesis about micro-level behavior by individuals. Nonetheless, we can draw conclusions about how variation in the political environment affects aggregate contributions to different kinds of committees.

Turning to the legal environment, contributions to different kinds of committees are subject to different maximum limits, and we might expect that committees with higher contribution limits receive a higher share of their contributions from districts with high SES. Moreover, the implementation of the BCRA following the 2002 election cycle allows me to examine the effects of a significant change in federal regulations. Again, there could be both direct and indirect effects. The direct effects were to virtually eliminate soft money while increasing the limit for contributions to candidates and the overall limit for hard money contributions to all committees including PACs (Malbin, 2006).⁵ One indirect effect was to alter the fundraising strategies of hard money committees as they attempted to make up for the loss of soft money. Both effects imply a greater reliance on districts with high SES. In addition, we might expect committees that support the same kinds of activities formerly supported by soft money to see an increase in their contributions, whereas those that support other kinds of activities should see no affect. For example, if soft money was used in substantial part to fund party building activities, then we should see an increase in contributions to S/D/L party committees following implementation of the BCRA. If, however, the main purpose of soft money was to fund activities designed to influence specific elections, then the increase in fundraising should be concentrated in the national party committees and Hill committees, as well as other committees not subject to FEC regulations.

Finally, although the numbers of national party committees, Hill committees, and to lesser extent candidate committees and S/D/L party committees remain fixed throughout the time period studied, PACs can be created or discontinued each election cycle. My data include 107 partisan PACs, 43 leadership PACs, and 155 issue PACs that received at least one itemized contribution during the 1994 cycle. For the 2006 cycle, those numbers are 321, 270, and 189, respectively. The number of PACs soliciting contributions at any given time may be a function of factors such as presidential election cycles and changes in the legal environment. In addition to these variables, each of my models includes a variable that captures approximately linear trends in factors affecting contributions over time.

Empirical Specification

To better understand the factors that account for differences in contributions made through different channels across congressional districts and over time, I estimated a series of mixed-effects regression models. The unit of analysis is the congressional district/electoral cycle, and the dependent variable is the natural log of real (2006) itemized dollars contributed.⁶

Independent variables include various measures of the political environment, socioeconomic characteristics, and legal regulations. Congressional elections are measured by variables for U.S. Senate and House contests and district partisan balance. I use separate dummy variables for contested (incumbent/challenger) Senate and House contests and open seat races. Partisan balance is measured as 50 – | 50 – Pct. Dem. |, where Pct. Dem. is the Democratic share of major party votes in the last presidential election. This variable is constructed so that higher values mean the district was more closely balanced between the two major party candidates. Closely balanced congressional districts may also overlap with competitive districts for state and local elections.

Districts with more competitive congressional races should obviously generate more contributions to local congressional candidates, and may also generate more contributions to S/D/L party committees. Notwithstanding Gimpel et al.’s (2008) findings, they should generate fewer contributions to most other types of federal committees due to opportunity costs and the scarcity of resources. As noted above, it is possible that competitive congressional races could stimulate increased contributions to other committees, but this effect is most likely limited to S/D/L or Hill party committees.

Three variables capture state elections. Even-year state elections is a self-explanatory dummy variable. Five states (Kentucky, Louisiana, Mississippi, New Jersey, and Virginia) hold state elections on odd-numbered years. Ceteris paribus, districts in states with concurrent state and federal elections may generate fewer contributions to federal committees than those in states with staggered elections. Similar to congressional races, I use separate dummy variables for contested and open seat governor’s races during even-numbered years.

Two additional variables capture the effect of presidential elections. Presidential cycle is a dummy variable coded 1 for 1996, 2000, and 2004. It is well known that national party committees raise more money in presidential cycles (see Corrado, 2006), but the effect of presidential cycles on other federal committees is less obvious. On the one hand, presidential candidates and national party committees compete with other committees for scarce financial resources. On the other hand, interest in election outcomes of all kinds may be heightened during presidential cycles, and nonpresidential committees may increase their fundraising efforts to take advantage of this as well as to offset the competition from the presidential campaigns. I also include another dummy variable for presidential battleground states (during presidential cycles only) as identified by Shaw (2006) to test whether it matters that potential donors are exposed to particularly intense campaigns.

I also include a set of variables that measure aggregate socioeconomic characteristics for each district: real per capita personal income, the percentage of adults with college degrees, the percentage of adults aged 65 and above, and the percentage of the population living in urban areas.⁷ Very little research has been done directly comparing donors who give to different types of committees,⁸ but a wealth of survey research demonstrates that individuals with more income and education and those who are older (at least up to a point) are more likely to make campaign contributions (e.g., Brady et al., 1999; Francia et al., 2003; Grant & Rudolph, 2002; James, 2009). While there is some possibility of the ecological fallacy (see King, 1997) and we cannot interpret the coefficients from a regression using aggregate data to imply causation at the individual level, previous research on aggregate contributions finds that these results tend to hold (Gimpel & Lee, 2006; Gimpel et al., 2006; Gimpel et al., 2008).⁹ Percent urban population taps into the effects of networks and spatial proximity. Individuals are more likely to make contributions if they are asked by someone they know or if they attend a fundraising event and this happens more often in densely populated areas (Cho, 2003; Cho & Gimpel, 2007).

Party committees responded to the 2002 BCRA by increasing their hard money fundraising efforts (Corrado, 2006). Most of the increase should come from districts that have relatively high income, education, and urbanization. Although the BCRA did not affect PACs directly, PACs that function as allies of political parties or are controlled by politicians may also have altered their solicitation strategies, and new PACs were created. I therefore include a set of terms interacting a dummy variable for post-BCRA cycles (2004 and 2006) with each of my socioeconomic variables, as well as a post-BCRA dummy variable by itself. Comparing the effects of socioeconomic variables on contributions before and after the BCRA will allow us to see the effect of the change in regulations on the geographic distribution of itemized contributions.

Two other variables are included as controls. The nominal threshold for reporting itemized contributions remained constant at $200 throughout the period covered by my data, but the real value declined by about 30% due to inflation. This alone should lead to an increase in itemized contributions reported. I therefore include the reporting threshold measured as 200 divided by the consumer price index. This variable should also pick up other factors that affect contributions over time, such as increases in the number of PACs soliciting contributions. Finally, the natural log of voting age population controls for the number of potential donors in a district. This can vary tremendously at the end of a decade. Summary statistics for the independent variables are reported in the Appendix.

Results

I account for the multi-level nature of my data (states, districts, election cycles) by estimating mixed-effects regression models with random intercepts at both the state and district levels and robust standard errors. Several previous studies explicitly model spatial autocorrelation between nearby counties or ZIP codes (Cho, 2003; Gimpel & Lee, 2006; Gimpel et al., 2006), but congressional districts can be very oddly shaped and it is not obvious how to define “nearby” areas. Gimpel et al. (2008) test for spatial autocorrelation in their analysis of contributions from congressional districts to out-of-district House candidates, but do not find any significant effects.

Each of my models of contributions to PACs includes some cases with no itemized contributions. I coded these cases as the natural log of the reporting threshold. I also estimated a set of Tobit regression models for PACs.¹⁰ The pattern of results is quite similar to those reported in Figure 2 below, but Tobit models cannot accommodate random intercepts at multiple levels. Results are provided in a Supplementary Appendix that will be made available on request.

Figure 2.

Point estimates and 95% confidence intervals for ratios of predicted contributions under “high” versus “low” scenarios.

Although contributions during one cycle do not cause contributions during another cycle, many of the same people are solicited and contribute cycle after cycle (Francia et al., 2003). Nonetheless, differences in district-level competition, differences between presidential and midterm cycles, and the effect of the BCRA all should lead to significant variation over time in aggregate contributions from the same district. I estimated a version of my models assuming first-order serial correlation within districts. The results are qualitatively very similar to those shown here (again save for the absence of random intercepts at multiple levels) and none of the estimated correlation coefficients exceed 0.19 in absolute value.¹¹

Regression Results

Table 1 shows the z scores for my mixed-effects regressions (complete regression results are in the Supplementary Appendix). It is hardly surprising that contributions to local congressional candidates are higher in districts with competitive congressional races. Otherwise, the coefficients on contested and open seat House or Senate races tend to have low statistical significance and are somewhat inconsistent. District partisan balance, however, has a negative effect on contributions made through every other channel (although the effect is essentially zero for S/D/L party committees) and the coefficients are significant at the 95% level for all but S/D/L and Hill party committees and soft money committees. Thus, differences in expected benefits and opportunity costs matter. The coefficients on even-year state elections also are negative for 7 of 10 types of federal committees and significant at the 95% level for nonlocal congressional candidates and soft money. Similar to House and Senate races, the effects of an even-year gubernatorial race are mostly insignificant and may be spurious as it is difficult to see how this could lead to increased contributions to federal campaign committees.

Table 1.

Z Scores for Regression of Ln(Real Itemized Dollars), by Type of Federal Committee, 1994-2006.

	Local congressional candidates	Nonlocal congressional candidates	Presidential candidates	S/D/L party committees	“Hill” party committees	National party committees	Soft money	Partisan PACs	Leadership PACs	Issue PACs
House contested	6.99**	1.12	1.90	−2.31*	−1.57	2.07*	−0.46	−1.85	−1.05	−0.27
House open seat	15.50**	−0.30	3.05**	−0.67	−1.97*	1.89	−0.03	−1.50	−1.70	−0.43
Senate contested	10.39**	−0.03	−0.08	1.11	−0.17	0.36	−0.97	1.07	−0.54	1.15
Senate open seat	20.00**	3.17**	−0.18	2.56*	1.02	−0.03	−1.90	1.19	0.56	1.18
District balance	4.37**	−3.09**	−4.04**	−0.16	−1.76	−3.01**	−1.48	−2.86**	−2.48*	−3.25**
Even-year state elections	−0.98	−3.06**	−0.78	−0.49	−1.38	−0.93	−3.36**	0.54	−1.51	1.24
Governor contested	−1.09	−0.59	−3.70**	0.71	0.97	0.71	1.22	0.55	−1.44	−0.14
Governor open seat	−0.60	0.08	−0.87	−1.26	2.56**	−0.86	1.23	1.55	−0.23	0.39
Presidential cycle	−0.86	0.90	—	7.86**	8.57**	29.91**	11.21**	2.76**	−3.85**	6.02**
Battleground state	0.98	−0.31	0.71	1.57	0.74	−0.58	2.01*	1.60	−1.75	−0.87
Per capita income	1.21	4.08**	4.02**	2.06*	1.99*	4.03**	1.44	1.05	5.32**	1.27
Percent college	3.63**	5.33**	6.23**	5.20**	6.51**	6.73**	2.99**	5.76**	2.38**	7.30**
Percent age 65+	0.95	1.03	0.87	1.76	3.92**	3.72**	−0.08	2.79**	0.17	2.90**
Percent urban	0.69	9.29**	5.78**	3.17**	3.30**	3.60**	7.22**	2.48**	6.35**	5.33**
BCRA × Income	0.64	1.42	−1.97*	1.47	2.44*	1.86	—	−0.39	−1.72	0.08
BCRA × College	−1.25	−1.52	2.33*	−0.52	−1.62	−0.70	—	−1.55	1.51	−1.36
BCRA × Age 65+	0.54	1.46	1.43	−0.85	−2.03*	1.20	—	1.05	0.48	1.22
BCRA × Urban	0.54	1.56	−0.38	−0.12	4.82**	2.90**	—	1.66	−2.22*	−2.14*
BCRA	−0.05	−1.70	0.83	−0.38	1.80*	−1.79	—	3.53**	1.73	2.50*
Reporting threshold	−1.39	−4.53**	−8.58**	−1.72	−3.99**	−5.23**	−13.56**	−3.27**	−8.07**	−4.02**
Ln(voting age population)	0.07	0.02	2.46*	0.33	2.74**	2.06*	1.72	2.40*	3.76**	1.52
Constant	1.60	2.26*	1.44	1.06	0.18	1.13	0.66	−1.38	−1.63	−0.19
Cases	3,045	3,045	1,305	3,045	3,045	3,045	2,175	3,045	3,045	3,045
Log pseudolikelihood	−3,039.9	−1,523.4	−876.5	−2,686.2	−1,534.2	−2,276.9	−3,044.0	−4,485.8	−4,335.5	−3,478.1

Note. Data are for hierarchical log-linear models with random intercepts at both state and district levels. Contributions to presidential candidates are for presidential cycles only. Soft money contributions are for 1994-2002 only. Z scores are calculated from robust standard errors. S/D/L = state, district, or local; PAC = political action committees; BCRA = Bipartisan Campaign Reform Act.

p < .05. **p < .01.

Contributions to all types of party committees and partisan and issue PACs are higher during presidential cycles, whereas those to congressional candidates are not significantly affected and those to leadership PACs are lower. The coefficients on presidential battleground states are mostly insignificant, save for leadership PACs (negative at the 90% level) and soft money committees (positive at the 95% level). The coefficient on S/D/L party committees is positive and just misses significance at the 90% level.

The effects of aggregate socioeconomic characteristics are generally as expected. The pre-BCRA coefficients on per capita income are positive in all equations and significant at the 95% level in 7 of 10. Coefficients on percent college graduates are positive and significant at the 95% level in every equation.¹² Districts with more residents aged 65 and above generate more contributions to Hill and national party committees and partisan and issue PACs, but otherwise the coefficients are not significant. The coefficient on percent urban is positive and significant at the 99% level for every contribution channel except local congressional candidates. Thus, all channels that involve raising money from the most fruitful districts to spend it wherever it will do the most good tend to rely disproportionately on urban districts.

Coefficients on socioeconomic variables that interacted with the post-BCRA dummy variable have varying signs and are mostly insignificant, whereas coefficients on the post-BCRA dummy variable itself are positive and significant at the 95% level for Hill party committees and partisan and issue PACs. The joint effects of these variables will be examined below.

Finally, the coefficients on the reporting threshold are negative for all types of itemized contributions and significant at the 95% level for all but local congressional candidates and S/D/L party committees. Thus, itemized contributions increase as the reporting threshold drops in real terms. This variable should also pick up the effect of increases over time in the numbers of PACs soliciting contributions.

Predicted Effects of Different Scenarios

The z scores reported in Table 1 show the direction and statistical significance of associations between each of the independent variables and the natural log of aggregate contributions, but the interpretation of the magnitudes of the effects is not always intuitive given the nonlinear functional form. Moreover, multiple independent variables are sometimes used to measure broader concepts such as district-level political competition or SES. Figures 2A to 2F address these issues by showing the point estimates and 95% confidence intervals for the ratios of predicted contributions under different scenarios. The vertical lines represent the 95% confidence intervals for the ratios of the predicted values with selected groups of independent variables set at “high” or “low” levels and all other independent variables set at their mean values; the horizontal hash marks are the point estimates.¹³

For district competitiveness (see Figure 2A), the “high” scenario assumes an open seat House election, a contested Senate election, and district partisan balance 1 SD above the mean. The “low” scenario assumes a contested House election, no Senate election, and district partisan balance 1 SD below the mean.¹⁴ Predicted contributions to local congressional candidates quadruple moving from the low to the high scenario whereas predicted contributions to S/D/L party committees increase by 9%, although the latter effect is not quite significant at the 95% level. Predicted contributions to all other types of federal committees decrease and the negative effect is significant at the 95% level for nonlocal congressional candidates and leadership and issue PACs. The very large percentage increase for local congressional candidates combined with the fact that it is the largest single channel for an average case in my data set means that, on balance, increases outweigh decreases: Total itemized contributions made through all types of channels shown are predicted to be $2.21 million per district in 2006 dollars under the high scenario compared with $1.69 million under the low scenario, an increase of 31%. Thus although there is competition between federal committees for scarce contribution dollars, on the whole my results are consistent with Gimpel et al.’s (2008) conclusion that donors in districts with competitive local races dig deeper into their pockets.

Figure 2B compares predicted values for a district in a state with an even-year, open seat race for governor to a district in a state with odd-year state elections. The joint effect of concurrent state elections and competitive governors’ races is negative for every type of channel except for partisan and issue PACs, both of which are small categories. The decrease in predicted contributions is significant at the 95% level for nonlocal congressional candidates and soft money, and total predicted contributions through all channels are 26% lower. The implied competition between federal and state campaign committees for scarce resources is a subject that has received little if any attention in the literature.

Figure 2C compares a district in a battleground state during a presidential cycle with an average district during a midterm cycle. The predicted effects on contributions to congressional candidates are minor and not significant at the 95% level despite very tight confidence intervals, but predicted contributions to all types of party committees as well as partisan and issue PACs and soft money committees increase substantially, with the largest percentage increases being for national party committees and soft money committees. Conversely, predicted contributions to leadership PACs are only 60% as large in presidential battleground states as during midyear cycles.

Figure 2D shows the independent effect of the BCRA on hard money contributions, holding all other variables equal to their sample means. Recall that this effect results from both increases in contribution limits for individuals and shifts in fundraising strategies by committees. Predicted contributions for local congressional candidates, S/D/L party committees, and leadership PACs are not significantly different at the 95% level; those for other types of federal committees are mostly up by 10% to 72%, except that predicted contributions to partisan PACs more than triple. The predicted sum of all hard money contributions increases by 26% post-BCRA; the predicted sum of hard and soft money contributions combined increases by 3%. This increase no doubt would have been larger but for the post-BCRA expansion of alternative channels such as 527 groups.

Finally, Figures 2E and 2F allow us to examine how the BCRA altered the extent to which federal itemized contributions come from districts with different socioeconomic characteristics. Gimpel et al. (2008) showed that contributions to out-of-district House candidates originate primarily from districts with high income, a high percentage of professionals, and high population density. Although my variables are slightly different, I find similar results for contributions to all nonlocal candidates, and in fact these contributions are affected to a greater degree than are contributions to presidential candidates or party hard money committees. The highest ratios of predicted contributions under high versus low scenarios prior to the BCRA are for soft money contributions and nonconnected PACs.

After the BCRA was implemented, the ratios of predicted contributions under high versus low scenarios were relatively unaffected for congressional candidates and S/D/L parties. They increased somewhat for Hill party committees and presidential candidates and somewhat more for national party committees, while actually decreasing for all three types of nonconnected PACs. These differences are statistically significant at the 95% level for Hill committees, national party committees, and all kinds of PACs, but not congressional candidates, S/D/L committees, or presidential candidates.¹⁵ Overall, the ratio for all predicted hard money contributions post-BCRA was 4.58 compared with a ratio of 4.94 for the sum of predicted hard and soft money contributions prior to the BCRA.

As noted above, my analysis does not include “small” (nonitemized) contributions to organizations in my data set, or contributions to organizations that are not required to file with the FEC. One possible consequence of not having data on small contributions is that the effects of variation in per capita income especially but also education attainment and urbanization may be exaggerated. This exaggeration would be greatest for party committees and partisan and issue PACs, which rely on small contributors more than other channels (Lowry, 2013; Malbin et al., 2011). Despite the demise of soft money, the rise of various other channels not subject to contribution limits means that individuals with higher SES almost certainly account for a greater share of total political contributions through all channels than they did prior to 2004.

Discussion

Political campaign contributions do not occur in a vacuum; they are affected by the political and legal environment experienced by prospective donors. This analysis examines contributions made through many different channels, but perhaps these channels can be usefully consolidated into three types. First are those that involve a subnational geographic connection between the donor and the recipient committee—these are primarily contributions to local congressional candidates, and also contributions to S/D/L party committees. Second are those channels that are closely tied to the national campaign for president that occurs in alternating election cycles—these are contributions to the presidential candidates themselves, as well as many contributions to national party committees. The third group consists of contributions to nonlocal congressional candidates, Hill party committees, and nonconnected PACs. I do not have direct evidence that parties orchestrate contributions through other channels (Gimpel et al., 2008), but this would provide a further reason to group Hill committees and contributions to nonlocal congressional candidates and perhaps partisan PACs in the same subset of channels.

Itemized contributions to local congressional candidates are driven to a large degree by the local political environment, whereas contributions to S/D/L party committees are slightly higher in competitive districts and perhaps presidential battleground states. Moreover, socioeconomic variables tend to be less important for explaining the variation in aggregate contributions made through these channels than through many other channels. Contributions to presidential candidates and national party committees are heavily dependent on the timing of presidential campaigns and socioeconomic variables. Contributions through other channels also vary somewhat between midterm and presidential cycles, and except for Hill party committees, they tend to be the most affected by variation in socioeconomic variables.

One thing that all channels except for contributions to local congressional candidates have in common is a significant, positive effect of district urbanization. Moreover, contributions to local candidates are becoming a smaller share of all itemized contributions over time and this trend will no doubt continue now that the cap on aggregate direct contributions during an election cycle has been eliminated (McCutcheon et al. v. Federal Election Commission, 2014). To the extent that candidates, parties, and their affiliated PACs skew their agendas to attract financial support from itemized individual donors, this would seem to give an advantage to urban interests over rural interests.

My results also demonstrate that aggregate contributions made through different types of channels are indirectly affected by differences in the political environment. Competitive congressional districts are predicted to generate fewer contributions to all types of federal committees except for local congressional candidates and S/D/L party committees, and in several cases, the difference is statistically significant at the 95% level. Similarly, most types of federal committees receive fewer contributions from districts in states with concurrent state elections, and leadership PACs receive fewer contributions during presidential cycles. This implies that prospective donors with limited resources respond rationally to differences in the expected benefits and opportunity costs of contributing through different channels. The effect is dampened, however, because there does not appear to be a “hard” budget constraint for federal political contributions. Rather, total contributions through all channels are higher in districts with competitive congressional races and during presidential election cycles.

Another set of findings concerns the effects of the 2002 BCRA. First, my findings reinforce the conventional wisdom that the world of political campaign finance will always adjust to changes in legal regulations (Mann, 2003). Moreover, the primary adjustments to the elimination of soft money by the BCRA were made by the national and Hill party committees and partisan PACs. Predicted contributions to S/D/L party committees were actually slightly lower. This implies that soft money was not being used primarily for party building activities, but rather was being used to promote the election of specific federal candidates.

Finally, the BCRA did lead to significant changes in fundraising practices and patterns. While party hard money committees especially relied more on congressional districts with very high levels of per capita income, education attainment, and urbanization after the law changed, they still were less reliant on these districts than were soft money committees before the law changed. Thus, the various channels that make up the “multilayered party coalition” (Herrnson, 2009) did not experience a drop in contributions despite the loss of soft money and the expansion of various unregulated channels, but did broaden the geographic base of their combined fundraising from individuals making itemized contributions.

These results are based on data through the 2006 cycle and one avenue for future research would be to investigate the effect of changes in fundraising by presidential campaigns due to the rejection of public funding for the general election by the Obama campaign in 2008 and both the Obama and Romney campaigns in 2012 (see Murray & Bacon, 2008). It seems unlikely that any major party presidential candidate will opt for public funding in the future. This may well affect the magnitudes of some of the effects involving presidential cycles, but the general results regarding the effects of political and legal context should remain, and the specific results regarding the implementation of the BCRA immediately following the 2002 cycle are not affected.

These results also have implications for further research that could be conducted using micro-level survey data. Although it may not be practical to collect large samples of contributions made through every different kind of channel, a useful distinction might be made between donors who contribute to a federal candidate running to represent their local jurisdiction or S/D/L party committee, those who contribute to support a presidential candidate or national party committee during a presidential cycle, those who contribute to a sponsored PAC (see Lowry, 2013), those who contribute to other committees regulated by the FEC, and those who make contributions not regulated by the FEC. Moreover, the effect of the timing of state and local elections on fundraising for federal campaigns has been overlooked in the literature. Further examination of the choices made by donors who must decide how to allocate scarce resources between contributions to different kinds of political committees as well as other uses for discretionary income may provide further insights into why people donate.

Footnotes

Appendix

Descriptive Statistics for Independent Variables

Variable	M	SD	Minimum	Maximum
House contested	0.75	0.43	0	1
House open seat	0.11	0.32	0	1
Senate contested	0.49	0.50	0	1
Senate open seat	0.20	0.40	0	1
District balance	39.4	9.0	3.9	50.0
Even-year state elections	0.90	0.29	0	1
Governor contested	0.32	0.47	0	1
Governor open seat	0.18	0.38	0	1
Presidential cycle	0.43	0.49	0	1
Presidential battleground state	0.14	0.35	0	1
Per capita income	20,386	5,641	7,928	58,625
Percent college graduates	23.6	9.2	5.6	61.3
Percent age 65+	12.5	3.1	4.7	30.3
Percent urban	77.1	21.2	13.1	100
Post–Bipartisan Campaign Reform Act	0.29	0.45	0	1
Reporting threshold	235.6	23.3	200	272.1
Ln(voting age population)	13.07	0.09	12.73	13.58

Note. Number of cases = 3,045.

Acknowledgements

I thank Jay Barth, Tom Brunell, Patrick Hultman, Adam Levine, Keena Lipsitz, John McAdams, Gary Wekkin, and Carole Wilson for helpful comments; Banks Miller for advice on hierarchical models; and Patrick Hultman and James Frazier for excellent research assistance.

Author’s Note

Previous versions of this article were presented at the 2008 meeting of the Southern Political Science Association, the 2010 meeting of the Midwest Political Science Association, and the University of Texas at Dallas.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Financial support from the National Science Foundation (Grant SES-0418129) is gratefully acknowledged.

Notes

Author Biography

Robert C. Lowry is a professor of Political Science at the University of Texas at Dallas. His research interests include political economy and American politics and public policy.

References

Brady

H. E.

Schlozman

K. L.

Verba

(1999). Prospecting for participants: Rational expectations and the recruitment of political activists. American Political Science Review, 93, 153-168.

Center for Responsive Politics. (2013). Political action committees [Interactive data summaries]. Retrieved from http://www.opensecrets.org/pacs/

Cho

W. K. T.

(2003). Contagion effects and ethnic contribution networks. American Journal of Political Science, 47, 368-387.

Cho

W. K. T.

Gimpel

J. G.

(2007). Prospecting for (campaign) gold. American Journal of Political Science, 51, 255-268.

Citizens United v. Federal Election Commission, 558 U.S. 310 (2010).

Corrado

A. J.

(2006). Party finance in the wake of BCRA: An overview. In Malbin

M. J.

(Ed.), The election after reform: Money, politics and the Bipartisan Campaign Reform Act (pp. 19-37). Washington, DC: Campaign Finance Institute.

Currinder

M. L.

(2003). Leadership PAC contribution strategies and house member ambitions. Legislative Studies Quarterly, 28, 551-577.

Dwyre

Kolodny

(2006). The parties’ congressional campaign committees in 2004. In Malbin

M. J.

(Ed.), The election after reform: Money, politics and the Bipartisan Campaign Reform Act (pp. 38-56). Washington, DC: Campaign Finance Institute.

Francia

P. L.

Green

J. C.

Herrnson

P. S.

Powell

L. W.

Wilcox

(2003). The financiers of congressional elections: Investors, ideologues, and intimates. New York, NY: Columbia University Press.

10.

Gimpel

J. G.

Lee

F. E.

(2006). The geography of electioneering: Campaigning for votes and campaigning for money. In McDonald

M. P.

Samples

(Eds.), The marketplace of democracy: Electoral competition and American politics (pp. 125-148). Washington, DC: Brookings Institution Press.

11.

Gimpel

J. G.

Lee

F. E.

Kaminski

(2006). The political geography of campaign contributions in American politics. Journal of Politics, 68, 626-639.

12.

Gimpel

J. G.

Lee

F. E.

Pearson-Merkowitz

(2008). The check is in the mail: Interdistrict funding flows in congressional elections. American Journal of Political Science, 52, 373-394.

13.

Grant

J. T.

Rudolph

T. J.

(2002). To give or not to give: Modeling individuals’ contribution decisions. Political Behavior, 42, 31-54.

14.

Herrnson

(2009). The roles of party organizations, party-connected committees, and party allies in elections. Journal of Politics, 71, 1207-1224.

15.

James

R. N.

III . (2009). An econometric analysis of household political giving in the USA. Applied Economics Letters, 16, 539-543.

16.

Jones

R. S.

Miller

W. E.

(1985). Financing campaigns: Macro level innovation and micro level response. Western Political Quarterly, 42, 301-324.

17.

King

(1997). A solution to the ecological inference problem: Reconstructing individual behavior from aggregate data. Princeton, NJ: Princeton University Press.

18.

La Raja

R. J.

(2006). State and local political parties. In Malbin

M. J.

(Ed.), The election after reform: Money, politics and the Bipartisan Campaign Reform Act (pp. 57-78). Washington, DC: Campaign Finance Institute.

19.

La Raja

R. J.

(2008). Small change: Money, political parties, and campaign finance reform. Ann Arbor: University of Michigan Press.

20.

Lowry

R. C.

(2013). Mobilizing money: Political action committees and political participation. American Politics Research, 41, 839-862.

21.

Maguire

Novak

(2013). Shadow money magic. Washington, DC: Center for Responsive Politics. Retrieved from http://www.opensecrets.org/downloads/shadowmoneymagic.pdf

22.

Malbin

M. J.

(2006). Assessing the Bipartisan Campaign Reform Act. In Malbin

M. J.

(Ed.), The election after reform: Money, politics, and the Bipartisan Campaign Reform Act (pp. 1-16). Lanham, MD: Rowman & Littlefield.

23.

Malbin

M. J.

Dusso

Fortelny

Gavin

(2011). The need for an integrated vision of parties and candidates: National political party finances, 1999-2008. In Green

J. C.

Coffery

D. J.

(Eds.), The state of the parties: The changing role of contemporary American parties (6th ed., pp. 185-204). Lanham, MD: Rowman & Littlefield.

24.

Mann

T. E.

(2003). Linking knowledge and action: Political science and campaign finance reform. Perspectives on Politics, 1, 69-83.

25.

Mansbridge

(2003). Rethinking representation. American Political Science Review, 97, 515-528.

26.

McCutcheon et al. v. Federal Election Commission, Slip Opinion No. 12-536 (April 2, 2014).

27.

Murray

Bacon

Jr. (2008, June 20). Obama to reject public funds for election. The Washington Post, p. A1.

28.

Shaw

D. R.

(2006). Race to 270: The electoral college and the strategies of 2000 and 2004. Chicago, IL: University of Chicago Press.

29.

U.S. Census Bureau. (2014). Census 2000 gateway [Data files]. Retrieved from http://www.census.gov/main/www/cen2000.html

30.

U.S. Federal Election Commission. (2014a). Citizens’ guide. Retrieved from http://www.fec.gov/pages/brochures/citizens_guide_brochure.pdf

31.

U.S. Federal Election Commission. (2014b). Detailed files about candidates, parties and other committees [Data files]. Retrieved from http://www.fec.gov/finance/disclosure/ftpdet.shtml

32.

U.S. Government Printing Office. (2014, January 1). Title 11—Federal elections. Code of Federal Regulations. Retrieved from http://www.gpo.gov/fdsys/pkg/CFR-2014-title11-vol1/pdf/CFR-2014-title11-vol1.pdf

33.

U.S. Internal Revenue Service. (2014). Political organization filing and disclosure. Retrieved from http://www.irs.gov/Charities-&-Non-Profits/Political-Organizations/Political-Organization-Filing-and-Disclosure

34.

Varian

(1987). Intermediate microeconomics: A modern approach. New York, NY: W. W. Norton.

35.

Weissman

S. R.

Hassan

(2006). BCRA and the 527 Groups. In Malbin

M. J.

(Ed.), The election after reform: Money, politics and the Bipartisan Campaign Reform Act (pp. 79-111). Lanham, MD: Rowman & Littlefield.

36.

Wilcox

(2001). Contributing as political participation. In Lubenow

G. C.

(Ed.), A user’s guide to campaign finance reform (pp. 109-126). Lanham, MD: Rowman & Littlefield.