Money and Elections: The Flow of Funds through the Federal Campaign System

Transcription

1 Money and Elections: The Flow of Funds through the Federal Campaign System by Susanna Supalla Submitted in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy Supervised by Professor Lawrence Rothenberg Department of Political Science Arts, Sciences and Engineering School of Arts and Sciences University of Rochester Rochester, New York 2017

2 ii Dedicated to my family, Elissa Newport, Ted Supalla, Zach Supalla, Ashley Supalla, Anita Newport, James Smith, and Charmaine Smith, for all of their support.

3 iii Table of Contents Biographical Sketch v Acknowledgments vi Abstract vii Contributors and Funding Sources ix List of Tables xi List of Figures xii 1 The Effect of Congressional Redistricting on the Flow of Contributions to State Legislative Candidates Factors Affecting Campaign Fundraising Redistricting and its Effects on State Legislative Fundraising Research Design and Data Findings Conclusions The Market for Hard Money: Funding Flows to Political Parties Post-BCRA 27

4 TABLE OF CONTENTS iv 2.1 Data State Parties and Fund Transfers as Networks Covariates Methods Findings Discussion A Appendix to Chapter The Effect of Ballot Measures on Issue Salience Direct Democracy The Cooperative Campaign Analysis Project Issues on the Ballot in Models Findings Discussion A Appendix to Chapter Bibliography 102

5 v Biographical Sketch The author was born in Rochester, New York. She attended Wellesley College, and graduated with a Bachelor of Arts degree in Political Science. After her undergraduate studies she worked as a Finance Director and Campaign Manager for local, state, and federal political campaigns in North Carolina and Virginia. She began doctoral studies in Political Science at the University of Rochester in She received a Master of Arts degree from the University of Rochester in She pursued her research in Political Science under the direction of Professor Lawrence Rothenberg.

6 vi Acknowledgments Thank you to my committee, David Primo, Kevin Clarke, Celeste Kidd, Cliff Smith, and most especially my advisor Larry Rothenberg. I also appreciate suggestions received from Lynda Powell, Dick Niemi, Matt Blackwell, Maya Sen, HyeSung Kim, Svanhildur Thorvaldsdottir, Brenton Kenkel, Rob Carroll, and members of the University of Rochester s Women Graduate Students in Political Science Working Group. Thanks also to conference panel attendees who gave me additional feedback on pieces of this work. Those include participants at the Midwest Political Science Association Conference in April 2013, the Annual Political Networks Conference in June 2013, the Southern Political Science Association Conference in January 2014, and the University of Rochester s American Politics Working Group.

7 vii Abstract This dissertation consists of three related papers that examine how the interactions between federal and state-level parties, campaign committees, and laws affect campaign fundraising and the greater electoral context. This research is particularly significant in light of recent federal and state laws that seek to restrict influences of money on politics. Chapter 1 examines how a state-level institution, state legislative control of redistricting, affects incentives for contributors to give to state rather than federal campaigns. Using data on state legislative campaign receipts and a research design that incorporates overlapping state and federal constituencies, I find that contributors in states with legislative control of redistricting (rather than an independent commission that does this work) target more of their money to state campaigns that may shape redistricting outcomes for years to come. This analysis shows that contributors are capable of sophisticated giving strategies influenced by the nature of their state institutions. Chapter 2 uses social network analysis to examine how political parties take advantage of variations in state laws to circumvent federal campaign finance reforms. McCain-Feingold s regulation of contributions to political parties, along with the absence of federal regulation of intra-party transfers, has led state parties to create segregated funds one for hard money raised according to federal

8 ABSTRACT viii campaign finance laws and one for soft money raised according to state campaign finance laws, which are often more permissive. I find that in some election cycles state parties behave strategically by trading money between states with less and more permissive campaign finance laws. Chapter 3 investigates how well-funded ballot measure campaigns affect issue salience in the 2012 election. A multi-wave survey, and ballot measure fundraising are used to examine the effects of exposure to ballot measures on individuals views of issue importance. Using a regression analysis, I compare respondents who live in states with ballot measures on each issue with respondents who do not experience well-funded ballot measure campaigns on the subject. I find that, while in the aggregate ballot measures do not affect issue salience among respondents, ballot measure exposure does alter views of issue importance among certain demographic groups.

9 ix Contributors and Funding Sources This work was supported by a dissertation committee consisting of Professors Lawrence Rothenberg (advisor), Kevin Clarke, and David Primo of the Department of Political Science and Professor Celeste Kidd of the Department of Brain and Cognitive Sciences. The data analyzed for Chapter 1 was provided by the National Institute on Money in State Politics (NIMSP), updated by Professor Adam Bonica of Stanford University, by the National Conference of State Legislatures (NCSL), by the United States Census Bureau, and by Professor Carl Klarner formerly of Indiana State University. The data analyzed for Chapter 2 was provided by the Center for Responsive Politics, the NIMSP, NCSL, and Congressional Quarterly s Voting and Elections Collection. The data analyzed for Chapter 3 was provided by YouGov and the NIMSP. All other work conducted for the dissertation was completed by the student independently. Graduate study was supported by a University Fellowship from the University of Rochester.

10 x List of Tables 1.1 State Characteristics in Average Treatment Effect (ATE) of Running in a Competitive District on Total Fundraising Receipts, in Average Treatment Effect (ATE) of Running in a Competitive District on Total Fundraising Receipts Average Treatment Effect of Running in a Competitive District on Individual Fundraising Receipts for OLS Estimation of the Effect of Redistricting Institution on Logged Total Fundraising Receipts, for Campaign finance laws in the 50 states, Exponential Random Graph Model (ERGM) Estimates for State Parties Fund Transfers, 2008 Cycle ERGM Estimates for State Parties Fund Transfers, 2010 cycle ERGM Estimates for State Parties Fund Transfers, 2012 cycle ERGM Estimates for State Parties Fund Transfers, 2014 cycle Competitive Statewide Elections, Competitive Statewide Elections, Competitive Statewide Elections,

11 LIST OF TABLES xi 2.9 Competitive Statewide Elections, Summary of Responses on Education s Importance Summary of Responses on Health Care s Importance Summary of Responses on Taxes Importance Summary of Total Amount Raised by Ballot Measure Committees in Number of States, and Numbers of Respondents in those States, by Subject Logit Estimates of Ballot Measure Effects on Education s Importance Logit Estimates of Ballot Measure Effects on Health Care s Importance Logit Estimates of Ballot Measure Effects on Taxes Importance Logit Estimates of Ballot Measure Effects on Education s Decreasing Importance Logit Estimates of Ballot Measure Effects on Health Care s Decreasing Importance Logit Estimates of Ballot Measure Effects on Taxes Decreasing Importance

12 xii List of Figures 1.1 Fundraising Advantage of Competitive Race, by Cycle Fundraising Advantage of Competitive Race, by Contributor Type, for Democratic State Parties Fund Transfers, 2004 Cycle Democratic State Parties Fund Transfers, 2006 Cycle Democratic State Parties Fund Transfers, 2008 Cycle Democratic State Parties Fund Transfers, 2010 Cycle Democratic State Parties Fund Transfers, 2012 Cycle Democratic State Parties Fund Transfers, 2014 Cycle Republican State Parties Fund Transfers, 2004 Cycle Republican State Parties Fund Transfers, 2006 Cycle Republican State Parties Fund Transfers, 2008 Cycle Republican State Parties Fund Transfers, 2010 Cycle Republican State Parties Fund Transfers, 2012 Cycle Republican State Parties Fund Transfers, 2014 Cycle

13 1 Chapter 1 The Effect of Congressional Redistricting on the Flow of Contributions to State Legislative Candidates In the 2010 election cycle, state-level campaign and committee expenditures reached $3.5 billion, nearly $1 billion more than these organizations spent in 2008 or in 2012 (National Institute of Money in State Politics 2013). This election cycle immediately preceded redistricting, drawing national attention to state-level races that could impact state and federal elections and policy. In some states, legislators elected in would propose new congressional district lines, allowing the party in power great leeway to affect the makeup of the state s congressional delegation. The national Democratic and Republican party organizations created new organs to support targeted state legislative candidates, particularly in swing districts in chambers with close seat margins between parties. Residents of such swing districts were exposed to a deluge of phone calls, mailers, and television ads, some from their local candidates and others from national groups. In 2010, the Ohio 21st House district was one such targeted district, with expenditures totaling close to $1 million. The Republican candidate for the open seat, Michael Duffey, said,

14 CHAPTER 1. EFFECT OF REDISTRICTING 2 They re investing in one particular vote on redistricting. It s like Lord of the Rings. One vote to rule them all (qtd. in Harrison 2010). We have some evidence from political parties and political action committees formed explicitly to influence redistricting outcomes that political elites target their state-level contributions with their probable effect on state- and federal-level outcomes in mind. But we do not know the extent to which a state s redistricting institution impacted its legislative candidates overall fundraising receipts in 2010 or whether individual contributors targeted their funds as the national groups did. Individuals, confronted with many competing solicitations for contributions, often give to build relationships directly with candidates and thus may be less likely to adopt the sophisticated giving strategies employed by parties. In this paper I analyze the impact of state legislative control of federal redistricting on fundraising receipts for state legislative candidates. I find that in the 2010 election cycle, competitive campaigns in states where legislatures redraw congressional district lines raise more, all else equal, than competitive campaigns in states where legislatures do not have such power. These competitive campaigns in states where legislatures control redistricting also receive more money from individual contributors, though individuals do not appear to target their money to such candidates as much as other contributor types, on average. The rest of this paper proceeds as follows: First, I discuss the range of factors we know to influence campaign fundraising receipts. Second, I discuss states redistricting institutions as factors that may also influence a campaign s fundraising receipts, particularly in the election cycle prior to redistricting. Third, I present my data and research design, in which I employ matching to understand the impact of legislative control of redistricting and district competitiveness on campaign

15 CHAPTER 1. EFFECT OF REDISTRICTING 3 fundraising receipts. Last, I present my findings, which highlight the importance of considering how our federal system shapes incentives to give to campaigns. 1.1 Factors Affecting Campaign Fundraising The literature on money in politics is extensive, with no shortage of analyses of why contributors give money and what factors contribute to a campaign s ability to rake in the dough (see Stratmann 2005). Contributors give for a host of reasons, but most often they fall into three categories (Francia et al. 2003, Joe et al. 2008). Investors are individual contributors who want to invest in politicians who will promote policies in their interests. Purposive givers have ideological or partisan reasons for contributing and seek out candidates whose policy positions align with their own. Lastly, solidary contributors give for personal reasons because they know the candidate or like socializing with politicians and members of the political elite. Many factors affect the incentives these various contributors have for giving to campaigns, and scholars have studied attributes of campaigns and their contexts to learn why some campaigns raise more than others. These reasons range from candidate characteristics to campaign characteristics, state institutions, district characteristics, electoral circumstances, economic climate, etc. While some factors affect a candidate s fundraising ability differently according to the context for instance, a candidate s party affiliation may help when it is the same as a popular president or hurt when it is not other factors can be used to reliably predict a candidate s fundraising ability. For instance, incumbents very often fare better than challengers in fundraising (Jacobson 1985). Senior legislators raise more than junior legislators, as do

16 CHAPTER 1. EFFECT OF REDISTRICTING 4 legislators who hold more prestigious committee assignments and leadership posts (Romer and Snyder 1994, Box-Steffensmeier et al. 2005). Often a candidate s fundraising prowess increases with experience in the legislature, as the candidate develops more relationships with potential contributors and also becomes more appealing to those seeking access or influence over policy making. Other factors affect campaign fundraising in more complex ways. For instance, competitive races may draw certain types of contributors. Solidary contributors are more likely to contribute to a candidate in a tight race to help the candidate win and gain or keep a friend in a powerful place. Purposive givers also want to give to a candidate in a competitive race in order to affect the outcome and gain or keep a policymaker who is on their side. Investors, on the other hand, may only want to give to a candidate who has a good chance of winning so they can ensure they would have access to the policymaker after the election. Investors may also want to be seen as valuable to a candidate who won a tight race, though, so often they may wait until the end of the campaign before donating to the candidate who seems most likely to win. Investors, when faced with a choice between giving to a competitive race and an unopposed candidate, may give to the unopposed candidate (McCarty and Rothenberg 2000, Box-Steffensmeier et al. 2005). There are also district characteristics and electoral circumstances that may affect the amount a campaign needs to raise in order to be competitive. For instance, the cost of advertising causes great fluctuations in the amounts raised by state legislative campaigns within states. As more and more campaigns air television ads, the cost of a competitive state legislative campaign relies more and more on characteristics of the district s media market(s). Large urban areas, for instance, frequently are part of large media markets, and the cost of running ads

17 CHAPTER 1. EFFECT OF REDISTRICTING 5 is higher because of the large number of viewers included in the market (though many of those viewers may not even be in the desired state legislative district) (Stratmann 2009). Comparatively little research has been conducted on the effects of state institutions on campaign fundraising. A state s campaign finance laws and limits affect a campaign s ability to raise money, as well as the competitiveness of its campaigns (Stratmann 2010, Stratmann and Aparicio-Castillo 2005). Also, some state institutions, such as term limits, can affect candidate fundraising by altering incentives for contributors and candidates (Apollonio and La Raja 2006, Powell 2012). Legislative control of redistricting has hardly been studied as an institution that may affect incentives to give and hence the flow of money in politics. (For a notable exception, see Makse 2013.) I now turn to legislative control of redistricting and its effects on contributor incentives. 1.2 Redistricting and its Effects on State Legislative Fundraising Every ten years, the states redraw district lines for congressional as well as state legislative seats after the Census Bureau releases its results. States have varying methods of redistricting (McDonald 2004). Most do so through a legislative process in which state legislators debate and pass a redistricting plan for the state s U.S. House districts, as well as the state legislature itself. Others employ an independent commission, which often includes political appointees. Table 1.1 lists the methods each state employed in the redistricting cycle. When legislatures have the power to redraw district lines, they have the power to gerrymander. La Raja writes, Redistricting enables the party in power to in-

18 CHAPTER 1. EFFECT OF REDISTRICTING 6 crease its swing ratio, or the proportion of seats it wins relative to its proportion of the popular vote (2009, 210). Researchers have identified many redistricting plans that have been passed and implemented as partisan gerrymanders, which create more seats for the majority party, or incumbent-protection gerrymanders, which make seats safer for incumbents (Arrington 2010, La Raja 2009, McDonald 2004, Gelman and King 1994). Further, partisan bias may occur not simply because of the partisan breakdown of newly created districts, but the weakening of the incumbency advantage due to changing district lines and more high quality challengers who enter following redistricting (McKee 2008, Murphy and Yoshinaka 2009, Yoshinaka and Murphy 2009, Hetherington et al. 2003, Carson and Crespin 2004). Those in charge of redistricting thus wield a great deal of power over the composition of the state s congressional delegation. The national parties have recognized the importance of the redistricting cycle. In the lead-up to the 2010 election, the Republican State Legislative Committee (RSLC) formed REDMAP, the Redistricting Majority Project, whose mission was to [win] state legislative seats that will have a critical impact on congressional redistricting in 2011 (RSLC 2010). Meanwhile, Democrats invested in the project Redistricting Facts, which highlighted how redistricting is misused, how its misuse affects you, and what you can do to keep your districts safe (Democratic Legislative Campaign Committee (DLCC) 2010). 1 As part of their efforts, the RSLC and the DLCC contributed to state parties and legislative campaigns across the country. Recently Makse (2013) examined the contributions made by parties across the country throughout the redistricting cycle, and he found that state parties 1 Democrats were in the majority in 60 of the 99 state legislative chambers prior to the 2010 elections (DLCC 2010).

19 CHAPTER 1. EFFECT OF REDISTRICTING 7 adopt specific strategies in states with legislative control of redistricting. In those states, majority parties protect their incumbents, while minority parties support challengers at the expense of incumbents in hopes of attaining the majority. Such strategies are not always present in states with independent commissions and in years early in the redistricting cycle. Considering again the most common motivations for contributors to give: the redistricting cycle should impact individual and PAC contributors, as well as parties, thus affecting a campaign s overall ability to fundraise. In a state with legislative control of redistricting, a contributor with ties to a member of Congress or the state s overall delegation whether for solidary, purposive, or investment reasons has greater incentive to give to a state legislative campaign whose outcome could affect the redrawing of congressional district lines. Contributors with disdain for the congressional delegation may face similar increased incentives to give to candidates of the opposite party. These incentives would propel contributors to become more involved in highly contested campaigns in such states, rather than less competitive campaigns in which they would not perceive their contributions as being able to change the outcome. In states in which legislatures had control over redistricting, by changing or maintaining the balance of power in the state legislature, contributors could impact their own congressional delegation for many years. In states with independent commissions or single members of Congress, contributors may not feel such increased incentives to participate in competitive state-level campaigns. With these hypotheses for how contributors may react to the state s redistricting institution, I now turn to the research design to explore the extent of the institution s effects on state legislative campaign fundraising.

20 CHAPTER 1. EFFECT OF REDISTRICTING Research Design and Data I employ matching to analyze the impact of legislative control of redistricting on campaign fundraising receipts, incorporating data on state legislative lower house campaign receipts, a few basic candidate characteristics, election returns, and overlap of state legislative and congressional districts. By conducting exact matching on four critical candidate and campaign characteristics while allowing competitiveness of races to differ within each pair of campaigns, I am able estimate the average difference in the amount of money raised between competitive and noncompetitive campaigns (Ho et al. 2011, Rubin 2006). By then calculating the same average differences in campaign receipts between competitive and non-competitive races on subsamples of matched pairs in states with and without legislative control of redistricting, I estimate the extent to which legislative control of redistricting affects the targeting of campaign funds to competitive campaigns. Rather than considering a state s redistricting institution the treatment, I adopt the state legislative campaign s competitiveness as the treatment. This is because I am interested in how the redistricting institution affects contributors targeting of competitive versus non-competitive campaigns. Thus the first step is to identify the average difference in fundraising receipts between competitive and non-competitive campaigns. The second step is to compare these differences across states with legislative control of redistricting and states with non-legislative control, a category that includes states with independent commissions and states with only one congressional district. In order to accomplish the first step, I calculate the average effect of a campaign being competitive on its fundraising receipts. This is the Average Treatment Effect (ATE), with the campaign s competitiveness serving as treatment. My results are robust to varying definitions of competitive-

21 CHAPTER 1. EFFECT OF REDISTRICTING 9 ness. Controlling for the many other district characteristics and electoral circumstances that impact a candidate s fundraising ability is a difficult task. Rather than seeking a wealth of data that has yet to be assembled at the state legislative district level, I adopt a research design that eliminates the need to control for these other characteristics. Specifically, I employ exact matching, pairing state legislative campaigns that are contained in the same congressional district. Because I match on the containing congressional district of the state legislative district, each pair shares some level of geographic proximity and in many cases also share media market(s) and other district characteristics that affect the amount of money raised by candidates, such as district wealth. Matching by containing congressional district is an imperfect way to account for these similarities, as geographic and demographic variation across congressional districts is common, but it allows for a simple research design. Further, matching by the state legislative districts containing congressional district means that each of our campaign pairs has some degree of similarity in the electoral context in which the campaigns occur. Constituents within the state legislative district pairs share exposure to the same federal-level and state-wide campaigns and are bombarded with similar fundraising appeals from other nearby campaigns. Within each state legislative campaign pair, the campaigns face similar competition for fundraising resources, as well as similar economic conditions and other factors that could affect contributors willingness to give. I also match campaigns based on three additional pre-treatment variables the party of the candidate, the incumbency status of the candidate, and whether the candidate wins. These variables account for other sources of variation in

22 CHAPTER 1. EFFECT OF REDISTRICTING 10 fundraising receipts. I rely on campaign fundraising data made available by the National Institute on Money in State Politics (NIMSP), updated and cleaned by Adam Bonica (2013). To measure key candidate and campaign characteristics that affect a candidate s fundraising capacity including party, incumbency status, and whether the district is an open seat, I use data from Klarner et al. (2013). I restrict my analysis to the most recent redistricting year of 2011 and the election cycle immediately preceding it, For most states, the final elections before the 2011 redistricting process occurred in The exceptions include the four states that hold elections in odd years (Louisiana, Mississippi, New Jersey, and Virginia), as well as the two states that hold elections in even years but every four years (Alabama and Maryland). 3 I adopt a dichotomous variable indicating whether the legislature controls redistricting in the 2011 decennial congressional redistricting. Table 1.1 lists the congressional redistricting institutions for the 2011 redistricting in each state. In 38 states, legislatures presently have control over the redistricting process, either by proposing redistricting plans or voting on plans drawn up by other institutions. Of these states, two have only one member of Congress and no need for redistricting, so they are coded as non-legislative control of redistricting. The 11 other states in our sample have independent commissions that redraw congressional district lines. 4 While the degree of control that elected officials in lower 2 Data are drawn from the National Conference of State Legislatures (NCSL) website: 3 New Jersey and Virginia campaigns, which occurred in 2009 and thus not very distant in time from most other states, are included in the analysis. Louisiana, Mississippi, Alabama and Maryland campaigns are excluded. 4 Because of the extensive work required to combine these datasets, I include only the lower house of the legislature. Nebraska has a unicameral legislature, so that leaves us with data for 49 lower houses.

23 CHAPTER 1. EFFECT OF REDISTRICTING 11 houses exercise over redrawing district lines varies across states coded as having legislative control, these officials play a greater role in redistricting than in states with independent commissions where elected officials only indirectly affect redistricting outcomes (by selecting potential appointees to the commission), if they have any say at all. At the state legislative level, data on campaign competitiveness is difficult to come by. I therefore adopt three inexact measures of competitiveness based on the number of candidates running and on vote tallies (Klarner et al. 2013). For matching, I dichotomize campaigns as competitive or non-competitive in two ways. The first of these is whether the candidate has any opposition. The second is whether the candidate wins by less than 55% of the vote, and the third is whether the candidate wins by less than 65% of the vote arbitrary cutpoints, but similar to those used in other analyses as defining competitive or non-competitive races (Stratmann 2010, Jacobson 2006). Lastly, in an OLS regression, I use a continuous measure of competitiveness that is the difference between the percentage of the vote earned by the highest vote getter and the second highest vote getter. This margin of victory has been used in many other analyses as a measure of competitiveness (Stratmann 2010, Niemi et al. 2006). In order to code the containing congressional district of each state legislative district, I use GIS software and the 2010 Census TIGER/Line Shapefiles provided by the U.S. Census Bureau for all state legislative lower house districts and all congressional districts. 5 There are 4,720 state legislative districts in the 49 lower house chambers. Of these state legislative districts across the country, 3,040 lie entirely within a single congressional district, while the remainder are split among two or more congressional districts. I discarded the split districts since whole dis- 5 Data drawn from

24 CHAPTER 1. EFFECT OF REDISTRICTING 12 Table 1.1: State Characteristics in Redistricting Institution: Legislative Independent Democratic Total Marginal Change in State a Process Commission Seats Seats Chamber c Congressional Districts AL AK AZ AR CA CO CT DE FL GA HI ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NV NH NJ NM NY NC ND OH OK OR PA RI SC SD b TN TX UT VT b VA WA WV WI WY a NE is excluded from the analysis because it does not have a lower chamber. b SD and VT have only one member of Congress so are coded as non-legislative control of redistricting. c Legislative chambers are coded as marginal if the Democratic percentage of seats before the 2010 election was <65% and >35%. Sources: Levitt (2010); Levitt and McDonald (2007); McDonald (2004); NCSL (2010); U.S. Census Bureau (2010).

25 CHAPTER 1. EFFECT OF REDISTRICTING 13 tricts are still plentiful and simplify the matching process. Of the whole districts, 2,789 had elections in I joined data from the various sources NIMSP, Klarner et al. (2013), and the Census Bureau. The resulting combined dataset yielded 4,319 state legislative campaigns eligible for matching. 6 Among these 4,319 lower house campaigns, I conduct exact matching with replacement on four discrete characteristics: the candidate s party, the candidate s incumbency status, the candidate s winning status, and the containing congressional district of the legislative district. 1.4 Findings I first examine differences in the ATE between campaigns in states with legislative control of redistricting and campaigns in states without. The competitiveness definition used here is whether the winning candidate has earned 65% of the vote or less. 7 I therefore form pairs in which one campaign is competitive under this definition and the other campaign is not. The first row of Table 1.2 contains the ATEs for these campaigns, of which there are 2,332 across the country. Of these, 1,824 are in states with legislative control of redistricting and 508 are in states without legislative control of redistricting. The ATE of running in a campaign with opposition is $21,471 in additional total fundraising receipts. In states with 6 The NIMSP and Klarner files have a negligible amount of missingness, but there is difficulty in combining the data in some instances. New Hampshire in particular presents a problem, as its legislative districts do not share a common naming or numbering across data sets; New Hampshire is therefore excluded. I also eliminate campaigns that do not belong to one of the two major parties. I further eliminate candidates who did not stand for the general election. 7 I also run these analyses using two alternate definitions of competitiveness: whether the candidate had an opponent and whether the winning candidate won by 55% or less of the vote. The results are substantively similar, with changes in the magnitudes of the ATEs but the same relative sizes.

26 CHAPTER 1. EFFECT OF REDISTRICTING 14 Table 1.2: Average Treatment Effect (ATE) of Running in a Competitive District on Total Fundraising Receipts, in Control of Redistricting: All States Legislative Non-Legislative All Lower House Candidates ATE $21,471 $25,199 $8,114 SE $3,070 $3,430 $6,492 N treated 1, N control 1, Winning Candidates ATE $22,575 $26,111 $8,706 SE $4,440 $4,904 $9,712 N treated N control Open Seat Candidates ATE $17,265 $23,438 $7,004 SE $5,831 $7,035 $9,069 N treated N control Candidates with opposition ATE $19,292 $23,216 $6,640 SE $3,578 $4,052 $7,276 N treated 1, N control Note: Here competitive districts are considered those where the winning candidate won less than 65% of the vote. legislative control of redistricting the ATE is $25,199, while in states without legislative control of redistricting the ATE is $8,114 a difference of $17,085, which is significant and large given that the campaigns in our sample only raised $57,000 on average. 8 Campaigns running with opposition greatly outperform campaigns running unopposed in fundraising; this difference is vast and significant in states with legislative control of redistricting. However, we have not yet taken into account several factors that may affect 8 This difference is significant at the.05% level.

27 CHAPTER 1. EFFECT OF REDISTRICTING 15 our results. For instance, some of the candidates in our sample may be only nominally running registering as a candidate but doing little to campaign or raise money. If this occurs substantially more in one state or a set of states, this could affect the size of the ATEs we find. I thus calculate the ATEs again, restricting my sample to winning candidates to eliminate the nominally running losing candidates. Winning campaigns in states with legislative control of redistricting raise an additional $26,111 on average if they are competitive, while similar campaigns in non-legislative control states have only an $8,705 advantage on average. This difference between average winning candidates campaign receipts in a state with legislative control of redistricting and in a state without is $17,406, a difference that is statistically significant at the.1% level. Our finding that contributors in states with legislative control of redistricting target their contributions to competitive campaigns to a greater extent than contributors in non-legislative control states holds true for winning campaigns. We also should take into account the seniority or leadership roles that may allow incumbents, particularly in uncontested races, to raise more money. These candidates often redistribute money to other candidates across the state, and this may interfere with our understanding of how competitiveness affects a candidate s ability to raise funds (Powell 2012). Therefore, I calculate the ATEs for matched pairs in which both campaigns are for open seats. This reduces our sample of campaigns to only 330, but these are still spread across states with legislative control of redistricting and without to allow us some ability to make inferences. In these races, none of the candidates has leadership roles or seniority that would make them more attractive to contributors. The difference across states with legislative control and without remains; campaigns for open seats in states with

28 CHAPTER 1. EFFECT OF REDISTRICTING 16 legislative control of redistricting raise $16,434 more than similar campaigns in states without, on average. 9 As an additional robustness check, I calculate the ATEs for matches in which both campaigns are opposed. Similar to the logic for eliminating nominally losing candidates, these results exclude candidates without opposition who may put no effort into fundraising, which could affect our results depending on how these candidates are distributed across the country. Here the difference remains roughly the same size at $16,576, achieving statistical significance at the 0.05% level. 10 There may remain other ways that states with legislative control of redistricting and states without differ that could account for the result that competitive campaigns have a larger fundraising advantage in legislative control states. As a final robustness check, I calculate the ATEs in states with legislative control of redistricting and in states without for the entire redistricting cycle, If the difference across subsets of states is due to redistricting institution, then the difference should be greatest in the election cycles immediately preceding redistricting. Early in the redistricting cycle it is less likely that the redistricting institution factors into the contribution decision. The states with legislative control of redistricting and without do not change, however, so we are effectively controlling for other dissimilarities across states that covary with redistricting institution. Figure 1 shows that the difference between the average advantage of a competitive campaign in states with legislative control and in states without is $17,085 in 2010 (as previously mentioned). In 2008, this difference was greater, at 9 With a p-value of 0.15%, this difference falls just short of achieving statistical significance, but the sample size is much smaller in this case, which could be a contributing factor. 10 I also run the analyses in Table 1.2 while excluding states with only one member of Congress. In such states, matched state legislative campaigns may be located across the state from each other rather than sharing similar geographic and demographic characteristics. The results are substantively similar.

29 CHAPTER 1. EFFECT OF REDISTRICTING 17 $40, $30,000 Average Treatment Effect $20,000 $10,000 Redistricting Control Legislative Non Legislative $0 Figure 1.1: Fundraising Advantage of Competitive Race, by Cycle. Note: Points represent ATEs of running in a competitive race on total fundraising receipts. Vertical lines represent 95% confidence intervals. $20,169. The 2008 election cycle featured record-breaking fundraising for presidential campaigns, and the enthusiasm spread to down-ballot candidates, especially those in competitive races across the country. 11 This enthusiasm, coupled with the 11 The average competitive campaign in 2008 raised $26,440 more than a non-competitive

30 CHAPTER 1. EFFECT OF REDISTRICTING 18 approaching redistricting, could have drastically increased the targeting of competitive campaigns in some states over others. As we go further back in time to elections further from the end of the redistricting cycle, we see the differences between the ATEs in states with legislative control and without decrease. The 2002 difference was the lowest at $7,191, meaning that in 2002 the difference in competitive campaigns fundraising advantages across states with legislative control of redistricting and without was much smaller than in 2008 and A difference among states still exists, which may indicate that other factors affecting fundraising may differ across state types, but the pattern remains that the difference is greatest in the two election cycles prior to redistricting, when redistricting is more likely to be on the minds of contributors. This comports with our hypothesis that contributors will target their money to competitive campaigns to a greater extent in legislative control states. To contributors in states with legislative control who are aware of the redistricting cycle and trying to influence redistricting outcomes with their pocketbooks, the competitiveness of a campaign is not the only thing that should matter. They should also care about whether their contributions have a chance of influencing the partisan control of the state legislative chamber, as the partisan control may determine how congressional district lines are drawn. In addition, they should care about the extent to which the state legislature will be able to change district lines. Thus we should see targeting of contributions to competitive campaigns occurring more in states where the legislative chamber s seat margin between parties is slim. I term these marginal chambers. We should also see more targeting of contributions to competitive campaigns in states that will gain or lose congressional districts, as existing congressional districts will have to be drastically reshaped in campaign.

31 CHAPTER 1. EFFECT OF REDISTRICTING 19 those states. I therefore further subset the states with legislative control of redistricting into states with marginal chambers and states with safe chambers. I further subset the states with marginal chambers into states with a change in the number of congressional districts in 2012 and states with a constant number of congressional districts. 12 See Table 1.1 for breakdowns of which states fall into each category. Once again, the predicted patterns exist in the data, and the differences are much greater than in previous analyses. See Table 1.3 for results. Here I define a marginal chamber as a lower house in which the Democrats hold between 35% and 65% of the seats. In marginal chambers where states legislatures control redistricting, competitive campaigns have a $33,366 fundraising advantage over non-competitive campaigns, on average. Meanwhile, in safe chambers where state legislatures control redistricting, competitive campaigns only have an average fundraising advantage of $7,290. We see a clearer difference, however, when we separate these states further. Also as predicted, in states with legislative control of redistricting and marginal chambers, the most targeting occurs in states where the number of congressional districts was expected to change after See Table 1.3 for results. The ATE for states with a change in the number of districts is $42,962, while the ATE for states with a constant number of districts is $29,605. Once again we see that contributors are targeting their money to campaigns that will have the biggest impact on redistricting outcomes the competitive campaigns for seats in marginal chambers where the legislature redraws the lines and will have the opportunity to drastically change those lines, substantially affecting congressional district lines 12 For this analysis we actually are interested in the states anticipated to have a change in number of congressional districts. Popular predictions of which states would gain or lose a seat proved to be accurate, so these two lists happen to be the same (Salvanto and Gersh 2010).

32 CHAPTER 1. EFFECT OF REDISTRICTING 20 Table 1.3: Average Treatment Effect (ATE) of Running in a Competitive District on Total Fundraising Receipts. Legislative Control of Redistricting States ATE $25,199 SE $3,430 N treated 910 N control 914 and Marginal Chamber States and Safe Chamber States ATE $33,366 $7,291 SE $4,575 $6,775 N treated N control and Change in Number of Congressional Districts and Same Number of Congressional Districts ATE $42,692 $29,605 SE $9,657 $6,046 N treated N control Note: Competitive districts are considered those where the winning candidate won less than 65% of the vote. Marginal chambers are considered those in which Democrats hold between 35% and 65% of seats. for years to come. I now turn to the question of what is driving this effect. We know, as discussed above, that some party organizations and political action committees (PACs) organize themselves explicitly for the purpose of funding targeted races in redistricting years, and so these organizations activities may be solely responsible for the differences in competitive races fundraising capacities in states with and without legislative control of redistricting. To check if this is the case, I also calculate the ATEs for each of the matched samples on fundraising receipts from individuals only (excluding candidate self-financing), PACs, and parties. These results are graphed in Figure 2. The difference in ATEs for states with and without legisla-

33 CHAPTER 1. EFFECT OF REDISTRICTING 21 tive control of redistricting is only substantial and significant for political parties. We see that in the states without legislative control of redistricting, PACs show some preference for non-competitive campaigns (though this result is not significantly different from zero), as investor contributors often do. Finally, we see that individuals seem to be not targeting contributions to competitive campaigns in states with legislative control of redistricting, but targeting them in states without legislative control. According to these results for all of the states, individuals and PACs are not following the parties lead. However, we see a different story when we split the legislative control of redistricting states by whether the chamber is marginal and whether the number of congressional districts changes. See Table 1.4. Now we see that individuals seem to be exhibiting the same giving patterns as parties, though to a lesser extent. Perhaps responding to party solicitations, individuals are targeting money to competitive campaigns in states where such contributions may affect redistricting outcomes. PACs, on the other hand, behave as the investor contributor type and support candidates in non-competitive races. Thus parties are the primary drivers of targeted giving, while individuals do so to a lesser extent. Finally, I run an OLS regression using all 4,991 opposed campaigns in the election cycle for which we have data on election returns. My dependent variable is the logged total fundraising receipts of each campaign. My independent variables of interest are a dichotomous variable for whether the state has legislative control of redistricting and the difference in competitiveness between the campaigns. Here, I use a continuous measure of competitiveness: the margin of victory for the winning candidate in other words, the difference between the percentage of votes earned by the highest vote getter and the second highest vote

34 CHAPTER 1. EFFECT OF REDISTRICTING 22 getter. In order to form a measure that is 0 when a campaign faces no competition and 1 when a campaign is tied with its opponent, I use the following transformation: District Competition = 1 (Highest Percentage Lowest Percentage) 100. Because we expect campaigns to raise more based on competitiveness when the state has All Individual PAC Party $30,000 $20,000 Average Treatment Effect $10,000 Redistricting Control Legislative Non Legislative $0 Figure 1.2: Fundraising Advantage of Competitive Race, by Contributor Type, for Note: Points represent ATEs of running in a competitive race on total fundraising receipts. Vertical lines represent 95% confidence intervals.

35 CHAPTER 1. EFFECT OF REDISTRICTING 23 Table 1.4: Average Treatment Effect of Running in a Competitive District on Individual Fundraising Receipts for All States Change in Number Constant Number of Congressional Districts of Congressional Districts ATE $6,663 $7,889 $6,183 SE $1,582 $2,806 $1,757 N treated N control Note: Competitive districts are considered those where the winning candidate won less than 65% of the vote. Sample limited to campaigns in states with legislative control of redistricting and marginal chambers, which all have more than one congressional district. Marginal chamber is defined as a chamber in which Democrats have <65% and >35% of the seats. legislative control of redistricting than when it does not, I include an interaction term between these two variables. I include the other candidate characteristics (party of the candidates, incumbency status of the candidates, and winning status of the candidates) as controls, as well as spending on the legislature per capita (as a proxy for professionalization of the legislature) and logged population. 13 Results are in Table 1.5. This table confirms the relationships we have seen among variables in the matching analysis. If we set each control variable to its mean or median, we find that an average campaign in our dataset is a winning Democratic challenger from a state with a population of 66,040 and expenditures on the legislature totaling $70,174,206. If we set the difference in competitiveness between this campaign and its competitor at 5%, for a district competition score of 0.95 on the 0 1 scale, this campaign, according to the estimates in Table 1.5, would raise $89,856 if it were in a state without legislative control of redistricting. If an identical average but competitive campaign were instead in a state with legislative control of redistricting, 13 Data drawn from Klarner (2012), U.S. Census Bureau,