Elections and Public Financing

Elections and Public Financing Annie Gleason Daniel Ferris Justin Eppley Mucio Godoy Stephen Sumner Xavier Smith - 1 -

TABLE OF CONTENTS Executive Summary 3 Introduction 4 Part I. Literature Review & Past Findings 6 Part II. Methodology & Scope: Expanding on Past Findings 9 Part III. Findings 16 Experiences Implementing Public Financing & Candidate Participation 16 The Impact of Public Financing on Campaign Money 20 The Impact of Public Financing on Electoral Outcomes 22 The Impact of Public Financing on Women Candidates 25 With-in Public Financing State Sub-Analysis 25 Part IV. Limitations of Findings 27 Part V. Policy Implications 29 APPENDIX - 2 -

EXECUTIVE SUMMARY In the United States, fundraising to run for elected office continues to capture increasing amounts of a candidate s time, even at the state level. For example, in 2008 the Republican State Leadership Committee spent $20 million alone on state legislative races, and the average winning candidate in a House race in 2008 that was contested spent nearly $2 million. 1 Our research indicates that public financing systems can help to reduce the amount of time legislators must devote to fundraising and, thus, spend more time addressing policy issues. Our data also indicates that as these races become closer in terms of money, they also become closer in terms of competitiveness. We find that public financing has a moderate but significant effect on improving the number of races that are competitive in both the state House and the state Senate. We also find that public financing has a strong effect on decreasing the disparity between incumbents and challengers in campaign contributions. 1 Wall Street Journal. < http://online.wsj.com/article/sb10001424052748703862704575099670689398044.html > - 3 -

INTRODUCTION The debate over the role of money in modern politics and elections in the United States has attracted increased attention since a number of corporations gave large and illegal campaign contributions to President Richard Nixon s 1972 re-election campaign. The debate, however, is not limited exclusively to national elections as a wide array of issues such as self-financing, corporate expenditures, large individual contributions, a lack of disclosure, and low participation levels also arise at the local and state levels and are seen by many as threatening the very core of American democracy. One method of attempting to curb the role of money in elections, the public financing of elections, dates back to the 1970s when the federal government, as well as states such as Minnesota, Michigan, and Wisconsin, enacted various reforms including public funding to encourage more competitive and transparent elections in the wake of the Watergate scandal. In the following decades, other laws reforming how candidates fund their campaigns have passed, many having gone through voter referendums. Although public financing in Minnesota, Michigan, and Wisconsin had higher participation levels initially, the states failed to update the amount of money available over time despite ballooning campaign costs which is why possible long-term increases in competitiveness and sustained participation in public financing have not been observed. Nevertheless, a re-emergence of campaign finance laws through ballot initiatives in a number of states occurred in the 1990s. Highly publicized decisions by political candidates whether to accept public funding also have brought more attention to how elections are financed. Contributions at the federal level attract the most attention. For example, in 2008 the candidates for President raised and spent over $1 billion. Even though candidates for state legislatures and governorships don t usually spend that much money on their campaigns, money, or lack of money, is seen as a key deterrent to possible entry into a race. Limiting the size of contributions is one method that has been used to attempt to reduce the role of special interests in politics and encourage parity in campaign war chests, but smaller donations lead some candidates to spend more time fundraising and less time governing. Some states and municipalities have turned to the public funding of elections to try and increase election competitiveness, contestedness, and participation rates among groups that traditionally are not well represented in government. Many states have also turned to term limits in an attempt to increase participation in elections. Then Senator Barack Obama's decision not to accept public funding was the first time that a candidate for president had done so in both the primary and general election since 1976. Republican candidate Senator John McCain opted to accept public funding during both the primary and general elections but never successfully made an issue of Obama s refusal that contradicted an earlier campaign pledge. George W. Bush and John Kerry each contemplated not accepting funding in the general election during 2004, but both candidates ultimately opted into public financing in the general after not accepting funds during the primary a decision that hurt Kerry s ability to raise funds needed to compete nationwide against President Bush. If candidates do not opt into the programs that are intended to increase competitiveness and minimize the role of private money in elections, it will be difficult to sustain or even assess a public financing system at the state or federal level. Participation in state public financing programs is just one of many factors that we took into account for our analysis. - 4 -

The Supreme Court has also weighed in on the private and public financing of campaigns with rulings in cases such as Buckley v. Valeo (1976), Randall v. Sorrell (2006), and Citizens United vs. Federal Election Commission (2010). The Federal Election Campaign Act of 1971 was the first comprehensive effort by the federal government that introduced a public financing system to Presidential elections and regulated campaign contributions and spending. After the original law was amended to include more reforms post-watergate in 1974, the Supreme Court upheld the enacted federal limits on campaign contributions but also ruled that to some extent spending money to influence elections is a form of constitutionally protected free speech in Buckley v. Valeo (1976). In the decision in Randall v. Sorrell (2006), the Supreme Court interpreted state imposed contribution limits as unconstitutional in Vermont because the allowable amount was so low. The Bipartisan Campaign Reform Act of 2002 ("McCain-Feingold Act") prohibited the use of soft money - unlimited campaign contributions to political parties - and established limits on independent expenditures for issue advocacy advertisements amongst other things. The Supreme Court has since overturned key parts of the legislation, nullifying some of its mandates. Most recently, in 2010, the Court held that corporations funding independent political advertisements in elections cannot be limited (Citizens United v. Federal Election Commission). In 2009, Senators Dick Durbin (D-IL) and Arlen Specter (D-PA) introduced the Fair Elections Now Act on the floor of the U.S. Senate. A companion bill was introduced in the House by Congress members from states with public financing: Rep. Larson (D-CT), Rep. Pingree (D-ME), and Rep. Jones (R-NC). The respective bills would be the first public funding system for the U.S. House and Senate where candidates would be entitled to lump sums after receiving a certain number of qualifying contributions of $100 or less. Similar bills have been introduced in recent Congressional sessions but have never made it out of committee. After a year-long polarizing fight over health care and other pressing issues left for Congress to tackle (energy, financial reform, and immigration, for example), it is not likely this bill will be brought to a vote this year. When the formal debate over expanded public funding for federal elections does occur, we hope our analysis of the effects of public funding at the state level will contribute to the dialogue. While New Mexico and North Carolina have full public financing systems for candidates running for some elected offices, they were not desirable to include on our study because gubernatorial and legislative candidates are not consistently included. Other states have partial systems with different levels of matching funds or lump sum grants that other studies have looked at. Our study focuses on elections in Maine, Arizona, and Connecticut before and after they adopted full public financing systems and compares them to states without comprehensive public financing laws. - 5 -

Part I. Literature Review Advocates of public financing reform argue that they can increase electoral competition by decreasing the fundraising gap between incumbents and challengers. The assumption is that with a narrowing of fundraising gaps, vote margins will also tighten. However, there is currently no scholarly consensus on that assumption. Several notable studies have considered the impact different public financing programs have had on electoral competition and draw varying conclusions. Most studies have only focused on the impact of specific state programs because of the challenges in performing a comprehensive nationwide analysis. For instance, one major challenge is that states that have already implemented public financing programs usually have established laws unique to those states. Donnay and Ramsden evaluated Minnesota s elections and possible impact of public finance program for the years 1966 to 1990. Minnesota s public financing system allows for participating candidates to receive private contributions during the campaign although they are bound to spending limits. Using pre and post-implementation analysis they found that the program had decreased competition when comparing the vote shares of the winning and losing candidates. They then incorporated a more complex statistical model to control for variations in the amount of public money accepted by each candidate, how much each was received from other sources, and total expenditures and concluded that challenger spending has a strong negative influence on the mean incumbent vote share. As a result, they suggested Minnesota s program has the potential to increase electoral competition with some modifications. They also recommended that since participating in public financing helped both challengers and incumbents the system should be modified to provide more assistance to challengers. 2 Mayer and Wood analyzed Minnesota s neighbor Wisconsin to determine public financing s impact on competitiveness by considering State Assembly and Senate races from 1966 to 1990. They defined competitiveness by incumbent reelection rates, average vote percentage of incumbents, percentage of incumbents who run unopposed, and the percentage of incumbents who win with less than 60% of the vote. They found that reelection rates for incumbents in both chambers stayed relatively high even after the implementation of public financing in 1978. They also found that incumbent s margin of victory remained high and actually increased after reform was instituted. Lastly, Mayer and Wood found that the number of unopposed incumbents increased after reform was instituted but there was an overall narrowing of the expenditure gap between incumbents and challengers between 1964 and 1990. 3 2 Patrick D. Donnay & Graham P. Ramsden, Public Financing of Legislative Elections: Lessons from Minnesota, 20 Legis. Stud. Q. 351 (Aug. 1995). 3 Kenneth R. Mayer & John M. Wood, The Impact of Public Financing Electoral Competitiveness: Evidence from Wisconsin, 1964-1990, Legis. Sud. Q. 20 (Feb., 1995), 83. - 6 -

Malbin and Gais broadened the scope of previous studies by evaluation public financing s impact on competition in Minnesota and Wisconsin together; both of which combined public finance programs with spending limits for participants. Furthermore, they focused on the lower house of the state legislatures and compared them to states that did not implement a program. They aggregated races for Minnesota and Wisconsin in the 1994 election and compared those to states with no public funding programs: Florida, Idaho, Kansas, Missouri and Washington. They considered the percentage of incumbents who received 55% or less of the vote, percentage of incumbents receiving 60% of the vote and the percentage of incumbents in uncontested races. Malbin and Gais found that, there is no evidence to support the claim that programs combining public funding with spending limits have leveled the playing field, countered the effects of incumbency, and made elections more competitive. 4 They did not discount the potential of the programs outright, but stated, This kind of a program is evidently not a long enough lever to move such a mountain; too many other factors influence competition to expect so much from one legal change. 5 A limitation to their conclusion is that the analysis only consisted of data from 1994. In 2000 and 2002, Maine and Arizona instituted public finance programs that offered full state funding for participating candidates for legislative and some statewide offices. This has led to evaluations that have added to a growing and lively debate over public financing s impact on competitiveness in the states. The Government Accounting Office conducted a study of Maine and Arizona in 2003 and concluded that although the number of legislative candidates who participated in the program increased between 2000 and 2002 in both Maine and Arizona, evidence for increased competitiveness was lacking. The GAO measured competitiveness in three ways the percentage of contested races, incumbents reelection rates, and the margin of victory of incumbents. They found varying results in the percentages of contested legislative races; incumbent reelection rates remained about the same and no clear trends in changes in incumbent victory margins. 6 Mayer et al took issue with the GAO s conclusions and methodology and concluded that both Arizona and Maine saw dramatic increases in competitiveness after implementing their full public financing programs in State House elections in terms of both contestedness and incumbent margin of victory. Mayer contends that the GAO study underestimated 4 Michael J. Malbin and Thomas L. Gais. The Day After Reform: Sobering Campaign Finance Lessons from the American States (New York: The Rockefeller Institute Press, 1998), 137. 5 Ibid. 6 General Accounting Office, Campaign Finance Reform: Early Experiences of Two States That Offer Full Public Funding for Political Candidates, GAO-03-453, May 2003, 29. - 7 -

competitiveness by using a more conservative statistical approach. 7 Mayer et al measured the percentage of incumbents who faced a major party opponent, the percentage of incumbents who were in a race in which the winner received less than 60% of the two-party vote and the percentage of incumbents who were reelected to office. They considered elections for the years 1990 to 2002. By considering a longer time period than the GAO study, they were able to note trends that the GAO could not. Furthermore, Mayer et al defined incumbent victory margins as competitive if incumbents received less than 60% of the two-party vote while the GAO used a slightly more conservative measure of 15% point margin of victory. These differences in approach help explain the differences in findings between the two studies. The disagreement about the impact of public finance programs on electoral competition among the studies discussed highlights the need for expanding the analysis across states. The increasing number of election cycles after implementation of public finance programs may lead to greater consensus in the future. At the moment, general consensus is found only around the measures of competitiveness incorporated by the GAO study - the percentage of contested races, incumbent reelection rates and the margin of victory of incumbents. In terms of competitiveness, it is important to note that there has been relevant research conducted about the impact of contribution limits, which is a tactic that some state governments have initiated in order to raise competition. States intended to level the playing field for challengers with these mandatory limits and a 2009 study by the Brennan Center of Justice and Dr. Thomas Stratman found that contribution limits led to more competitive elections in a 26 state analysis. In fact, the authors concluded that the lower the limit, the more competitive the election. The lowest contribution limits on the books ($500 and below) increased the likelihood that challengers would beat incumbents. Further, lower contribution limits were found to reduce incumbents considerable financial fundraising advantage. 8 A secondary finding of this study was that state public financing programs, similar to low contribution limits, could increase the competitiveness of elections. The study concluded that states with some form of public financing of campaigns coupled with contribution limits (Minnesota 1980-2006 and Maine from 2000 onwards) saw a decline in the mean incumbent margin of victory, from 57 percentage points in states with contribution limits but without public financing, and 30 percentage points in Minnesota and Maine. 9 7 Kenneth R. Mayer, Timothy Werner & Amanda Williams, Do Public Funding Programs Enhance Electoral Competition, Presented at the 4 th Annual Conference on State Politics and Policy Laboratories of Democracy: Policy in the American States, Kent State University, April 2004, 13. 8 Ciara Torres-Spelliscy, Kahlil Williams and Dr. Thomas Stratmann, Competition Policy for Elections: Do Campaign Contribution Limits Matter?, May 2009, available via Brennan Center for Justice at http://www.brennancenter.org/content/resource/electoral_competition_and_low_contribution_limits/ 9 Ibid. - 8 -

Lastly, there has been little research done on public financing s impact on gender diversity on any electoral level. Ken Mayer and Timothy Werner begin to address this knowledge gap through their study conducted in 2005. They considered legislative elections in five states Arizona, Hawaii, Maine Minnesota, and Wisconsin. They analyzed the number of female candidates from 1990 to 2002 and the number of female members of the legislature for the years 1991 to 2003 and found no discernable trends that that could be linked with public financing laws. They noted that Hawaii saw a consistent increase in the number of female candidates and elected officials but did not attribute these trends to public financing laws. Too few candidates participated in the program and even fewer of those were women. 10 Part II. Methodology & Scope OUR SAMPLE Our research seeks to expand upon past findings by examining state house, senate and gubernatorial elections in the only three states in the United States with full public financing for legislative and gubernatorial offices: Arizona, Connecticut, and Maine. We chose to study state house, senate and gubernatorial races because each may be affected differently by public financing. Comparison States The inclusion of Connecticut to our Public Finance State Comp. State #1 Comp. State #2 study adds to past literature on Arizona New Mexico Nevada public financing. We decided not to Connecticut Delaware New York include partial public financing Maine New Hampshire Washington states, such as Wisconsin and Minnesota, in our study because these systems have been studied at length in separate analyses and differ greatly from full public financing. Partial states often provide a fixed subsidy or matching grants to candidates who agree to spending limits; candidates provided with a public grant are then able to continue raising private contributions up to the limit. Given the disparity in partial public financing systems, we believe states with full public financing are purer systems. In addition, we build upon past findings by including two comparison states for each of the public finance states. We matched our comparison states with the three public financing states based on a host of key demographic, economic and political variables. Our matching method is described in detail in Appendix Item A. Finally, we collected data for all nine states from 1996 to 2008 in order to provide at least two pre-election cycles and at least one post-election cycle for all states. While 10 Kenneth R. Mayer & Timothy Werner, The Impact of Public Election Funding on Women Candidates: Comparative Evidence from State Elections, April 2005, available via Wisconsin Campaign Finance Project at http://campfin.polisci.wisc.edu/gender%20and%20public%20funding.pdf - 9 -

we could have done a simple pre-post analysis looking only at the public financing states, this would not have accounted for time trends that could impact our findings over the time period we are studying. Therefore, we chose to make both temporal and spatial comparisons, where the only presumed change occurring over the time period is whether or not a state operated under a public financing law. OUR DATA We draw on several data sources for our analysis. Election results were collected from the Secretary of State offices for the nine states in our study. We obtained contribution data from Follow the Money s National Institute on Money in State Politics. The institute is a non-partisan organization that examines the influence of money on state elections. Expenditure data was obtained from the Brennan Center for Justice and the Wisconsin Campaign Finance Project (additionally, some of our election data was obtained through the Wisconsin project). We chose to use contributions rather than expenditures for our analysis primarily because the data was accessible through Follow the Money. We believe total contributions (the war chest amassed by each candidate) provide an adequate measure of the fundraising capabilities and monetary advantage of each candidate. However, we also have a sample of expenditure data for Arizona, Connecticut and Maine. We calculated mean expenditures and contributions over our study period and did not find dramatic differences between the two figures. We believe expenditures may provide a better measure, as they indicate how much a candidate actually spent on his or her campaign. Nevertheless, average and median contribution and expenditure figures were similar over our study period and tended to trend in the same direction. In the Arizona House, for example, the mean expenditure level in 2008 was $47,745; the mean contribution level was $47,053. Given time constraints, we used contributions for this analysis. However, had we gone further to perform an analysis of primary elections, expenditure data would have been necessary, as our contribution data didn t differentiate between primary and general phases. A full comparison of average and median contributions and expenditures is found in the appendix. Finally, we used Follow the Money and Internet searches to compile gender data on each candidate in the races. We were unable to assemble gender data for New Hampshire because the Secretary of State s Office did not list first names for candidates in its 2000, 2002, 2004 data files; first names were provided in the 2006 and 2008 files. Candidate data for 1996 and 1998 was obtained separately through New Hampshire s biannual Manual for General Court pamphlets. OUR OUTCOME VARIABLES Advocates of public financing argue two central benefits to such a system: First, it allows candidates to spend less time fundraising by reigning in campaign costs. Second, it narrows the fundraising gap between incumbents and challengers, allowing challengers to be more - 10 -

competitive. Candidates who are not able to raise as much money may be at a competitive disadvantage. For this study, we focus on the competitiveness argument. We will examine whether public funding narrows the disparity in candidate fundraising and then further examine the impact on overall competitiveness of a race. Other studies point to evidence of a relationship between competitiveness in fundraising and electoral competitiveness; in a rigorous study, Neil Malhotra, found that public financing programs had the greatest effect on improving competition in districts where challengers (as opposed to incumbents only) accepted public grants 11. This appears to point to a relationship between funding and a candidate s ability to launch a competitive campaign. Public funding may also influence a candidate s decision to enter a race, as the promise of a grant may improve one s outlook on his or her ability to be competitive. In fact, past studies have shown that candidates cite the onerous fundraising process as one of the biggest deterrents to launching a campaign 12. Therefore, we also look at whether public financing increases the likelihood that an incumbent will be contested. Finally, we examine whether public financing leads to a more diverse set of candidates. We measure outcomes in the following ways: The funding disparity between candidates: (1) whether the contribution ratio is competitive (defined as the contribution amount of the top recipient divided by the contribution amount of the second highest recipient). We defined a competitive ratio as 1.5. The competitiveness of elections: (1) whether the election is close (defined as an election where margin of victory is 10, 15, or 20 percent) The contestedness of elections: (1) whether a candidate had any challenger at all (2) whether a candidate was challenged by a third party candidate The diversity of elections: (1) whether or not a woman was present in a given race 11 Malhotra, Neil. The Impact of Public Financing on Electoral Competition. State Politics and Policy Quarterly. Vol. 8. Fall 2008. http://www.stanford.edu/~neilm/sppq.pdf 12 Hamm, Keith and Robert Hogan. Campaign Finance Laws and Candidacy Decisions in State Legislative Elections. Political Research Quarterly. March 20, 2008. http://prq.sagepub.com/cgi/rapidpdf/1065912908314646v1.pdf - 11 -

LIMITATIONS OF DATA FOR OUR VARIABLES Before we could conduct our analysis, we had to address two data issues regarding uncontested races and multi-member districts. Uncontested races presented a challenge in our calculations of competitiveness measured by the vote margin. All races with vote margins above 10, 15 and 20 percent were labeled as uncompetitive in our three definitions of competitive, respectively. However, our results would be distorted had we included uncontested races in our competitiveness measure; candidates in uncontested races had vote margins of 100 percent (much higher than our competitive threshold) but did not actually have opponents. We are already looking at rates of contestedness separately; for our competitiveness variable, we are interested in examining the closeness of a race between two or more candidates. Therefore, all uncontested races were excluded from our measures of competitiveness but included in our contestedness and diversity measures. Secondly, two of the states in our analysis Arizona and New Hampshire have multi-member districts, presenting a problem when attempting to make data comparisons with single-member district states. To address this problem, we used a method developed by Neimi, Jackman and Winsky and replicated by Kenneth Mayer to create pseudo-single member districts 13. Using this method, the top vote-getter is paired with the lowest vote-getter in the opposite party. The second highest vote-getter is paired with the second-lowest vote-getter in the opposite party. Ultimately, single-member districts with one Republican and one Democrat are created through pairings. However, candidate can also be labeled as running unopposed. For example, if three seats are available, two Democrats and three Republicans run, and the top vote-getter is a Republican, then the top Republican candidate would be placed in a district by him or herself and labeled as running unopposed. Further, minor-party candidates are placed with candidates in the middle districts. For example, in New Hampshire s 2006 Belknap County District 2 race, four candidates were vying for two seats. Gail Morrison (D), the highest vote-getter, was paired with Dennis Fields (R), the lowest Republican vote-getter. William Tobin (R), the second highest vote-getter, was paired with William Joscelyn (D), the lowest overall vote getter. OUR ANAYSIS METHOD In order to analyze the impact of public financing laws on electoral outcomes, we used state assembly and state senate districts as the units of analysis. Senate and house races are measured in separate analyses. We chose to analyze these chambers separately because we believed senate and house races may be impacted differently by the existence of public financing, as senate candidates are generally more well-known and have more established political networks. Also, state senate races are more prestigious, and therefore, may produce more competitive elections. 13 Richard G. Niemi, Simon Jackman and Laura R. Winsky. Candidacies and Competitiveness in Multimember Districts. Legislative Studies Quarterly. Vol. 1. February 1991; and Mayer, Kenneth. Do Public Funding Programs Enhance Electoral Competition? University of Wisconsin. March 2005. - 12 -

Additionally, we ran descriptive statistics using the state as the unit of analysis for gubernatorial races. We cannot determine whether the gubernatorial results are statistically significant or perform difference-in-difference estimates, because we do not have enough data points. Instead, we provide a table of descriptive statistics for each gubernatorial election year in the nine states which can be found in the Appendix. Difference in Difference Technique To analyze the data, we used the difference-in-difference technique. To illustrate, the general model for any of our three variables of interest in analyzing Maine is as follows: Impact estimate = (Mainepost Mainepre) (Comparisonpost Comparisonpre) The technique has been widely used in empirical studies. For example, in a well-regarded difference-in-difference study, David Card and Alan Krueger examined the impact of an increase in New Jersey s minimum wage law on fast-food employment using nearby Pennsylvania as a comparison 14. In another study, Patricia Anderson and Bruce Meyer assessed the effect of a change in Washington State s unemployment law, employing the difference-indifference method and using all other states as controls 15. Additionally, Jonathan Gruber in 1994 used the difference-in-difference technique to estimate the impact of mandated maternity benefit laws passed in the 1970s in Illinois, New Jersey and New York; Gruber calculated estimates using Ohio and Indiana as controls for Illinois and Connecticut, Massachusetts and North Carolina as controls for New York and New Jersey 16. The difference-in-difference technique is useful because it allows us to evaluate the impact of public financing on each of the above outcome variables, absent of any time trends that could bias our estimate upwards if not accounted for in the model. For example, if the souring economy over the past few years compelled more voters to turnout and express their dissatisfaction with incumbents, the data may show that state races became more competitive as challengers garnered more votes. If we did not account for this time trend, we would wrongly attribute this increase in competitiveness to public financing. However, by including data from comparison states, we are able to tease out the true impact of the clean elections reform, absent of overall time trends. 14 Card, David and Alan Krueger. Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania. The American Economic Review. Vol. 84, No. 4. September 1994. http://emlab.berkeley.edu/~card/papers/njmin-aer.pdf 15 Anderson, Patricia and Bruce Meyer. The Effects of the Unemployment Insurance Payroll Tax on Wages, Employment, Claims and Denials. Journal of Public Economics. Vol. 78. 2000. http://bbs.cenet.org.cn/uploadimages/200311261248469802.pdf 16 Gruber, Jonathan. The Incidence of Mandated Maternity Benefits. The American Economic Review. June 1994. - 13 -

REMEDIES FOR POTENTIAL LIMITATIONS OF OUR ANALYSIS METHOD Our six comparison states, although without public financing, have various campaign finance regulations. New Hampshire has both voluntary spending limits and contribution limits of $1,000 per election for candidates not agreeing to spending limits and $5,000 per election for candidates agreeing to abide by spending limits 17. Expenditure limits in New Hampshire were adopted in 1995 and are set at $20,000 for the state senate and $0.50 per voter for the state house. Nevada has a limit of $5,000 per contributor for any state office, a law passed through voter initiative in 1994 18. Washington has had limits on contributions for statewide and legislative offices since 1992 19. Limits in the states are set at $1,600 for statewide offices and $800 per election for legislative offices. Delaware enforces a limit of $1,200 in contributions to a statewide candidate and $600 to other candidates per election cycle. New York s regulations limit contributions to $9,500 per senate candidate in the general election and $3,800 per house candidate. Contributions to gubernatorial candidates are capped at $37,800 in the general election. New York further limits contributions by an individual to $150,000 in the aggregate 20. Nevertheless, New York s contributions are considered to be lax when compared with other states 21. New Mexico is the one comparison state in our study that has a public financing structure in place a minor program implemented in 2003. We selected New Mexico as a comparison for Arizona for two reasons. First, it offers public financing for just a few races, including state Supreme Court justice, court of appeals judge and public regulation commissioner. Second, Arizona lacked a second match in our cluster analysis. Given that New Mexico does not provide public grants to state legislative and gubernatorial candidates, we opted to include the state in our comparison group in order to have sufficient comparison data for Arizona. New Mexico, previously without limitations on contributions, enacted a new law during the 2009 legislative session, which limits contributions from individuals to $2,300 per candidate per legislative election and $5,000 per candidate per statewide election. Contribution limit regulations in a state have been shown to impact our outcomes variables of interest competitiveness, contestedness and diversity. For example, in a 2001 study, Thomas Stratmann et al. conclude that changing the law from having unlimited contributions from 17 States Campaign Spending Limits. Illinois General Assembly. http://www.ilga.gov/commission/lru/appendixc.pdf 18 National Conference of State Legislatures. State Limits on Contributions to Candidates. Jan. 2010. http://www.ncsl.org/portals/1/documents/legismgt/limits_candidates.pdf 19 Wickert, David. Governor Signs New Contribution Limits for Local Elections. The News Tribune. http://blog.thenewstribune.com/politics/2010/03/26/governor-signs-new-contribution-limits-for-local-elections/ 20 National Conference of State Legislatures. State Limits on Contributions to Candidates. Jan. 2010. http://www.ncsl.org/portals/1/documents/legismgt/limits_candidates.pdf 21 Fiscal Reform for New York State. January 2006. http://www.osc.state.ny.us/reports/budget/fiscalreform.pdf - 14 -

individuals to candidates to having limits on contributions leads to a reduction in incumbent vote share of between 3.3 and 6 percentage points 22. Further, in a 2006 study, Robert Hogan and Keith Hamm find a strong, significant correlation between the strength of a state s contribution limits and the rate of contestedness in house districts statewide 23. The literature, thus, appear to highlights a correlation between the existence of contribution limits and electoral outcomes. Even so, we contend that as long as campaign finance regulations in each state remain constant throughout our study period, we are able to achieve credible estimates despite the presence of contribution or spending limits in our comparison states. The difference-in-difference technique compares the changes in electoral outcomes between states and not the raw numbers; therefore, baseline differences between states in competitiveness, contestedness and diversity are accounted for using this technique. Nevertheless, if campaign finance regulations vary widely within a state cluster for example, Maine, New Hampshire and Washington then our results may be influenced to a degree by the fact that, with the difference-in-difference technique, we have to assume that elections in the three states are conducted the same way. New Hampshire and Washington are intended to serve as proxies for the way Maine would have functioned absent public financing reform. To address this concern, we performed a second analysis computing difference-in-difference estimates in electoral outcomes within a state. All three clean elections states have high levels of participation in the public financing system; however, in all states each year, at least of handful of districts involved races with privately-funded candidates only. The Arizona House, for instance, averaged 79 percent participation in the public finance system between 2000 and 2008. This allows us to compare the remaining 21 percent of privately-financed districts against the publicly-financed districts. Unlike our between-state comparison, the within-state comparison is not susceptible to bias resulting from observed and unobserved differences between states. However, this study is limited by a low number of data points and potential selection bias. Given the low number of privately-financed districts, this comparison is limited by a low number of data points. In fact, because some years had such low numbers of privately-financed races, we decided to weight our outcome results based on the proportion of privately-financed races for all the years. For instance, Maine in 2006 had only two districts with privately-funded candidates only; one race was competitive within a 10 percent vote margin and the other was not. The ultimate result is that 50 percent of the districts in Maine privately-funded races were competitive. However, this result is clearly misleading, given that only two districts factored into the analysis. Therefore, these results are weighted to reflect more accurate outcomes. We conducted the within-state 22 Stratmann, Thomas and Francisco J. Aparicio-Castillo. Competition Policy for Elections: Do Campaign Contribution Limits Matter? 2001. 23 Hogan, Robert and Keith Hamm. Campaign Finance Laws and Candidacy Decisions in State Legislative Elections. 2006. - 15 -

analysis for Maine and Arizona, as Connecticut did not produce an adequate number of data points. Selection bias is another potential problem with our within-state analysis. Because candidates voluntarily opt into the public financing system, it s possible that the districts where neither candidate opts into the system differ fundamentally from the publicly-funded districts. For example, candidates in privately-financed, urban districts may pass on public grants because they anticipate high-spending campaigns, given higher campaign and advertising costs in urban areas. Or, it s possible that candidates in more conservative districts are ideologically opposed to public grants, prompting neither to opt into the system. Regardless of the cause, if privately-financed districts are different in some fundamental way from publicly-financed districts, then our results could be biased. Despite these two limitations, we believe this analysis is worthwhile because it provides another set of results. If both the between-state and within-state difference-in-difference estimates reveal similar results, then we can be more confident in the soundness of our overall analysis. Part III. Findings EXPERIENCES IMPLEMENTING PUBLIC FINANCING & PARTCIPATION Although Arizona, Connecticut, and Maine clean elections systems share similarities, each state's program has varied in its implementation. In all three states, the public financing program is just one element of a broader umbrella of policies regulating campaign finance; other elements include contribution limits and disclosure requirements. In fact, in assessing public financing reform in the three states, we will actually be assessing the impact of the overall reform public grants and spending limits rather than the public grants alone. Maine Maine was the first state to pass a full public financing law. The Maine Clean Elections Act was passed as a ballot initiative in 1996. It took effect in 2000. Candidates who opt into the system must raise a minimum amount of qualifying funds, in order to show they are viable candidates. Once in the system, candidates agree to spending limits of $20,082 for the senate and $4,362 for the state house. However, they may receive matching funds to remain competitive if outspent by non-participating opponent. Arizona Arizona voters passed the Arizona Clean Elections Act in 1998, also by ballot initiative. In Arizona, the clean elections movement gained steam following a series of state legislative - 16 -

scandals in the early 1990s, often referred to as AZScam. The system provides full public funding for statewide and legislative candidates. As in Maine, participating candidates must raise qualifying contributions and agree to spending limits; likewise, they are eligible for matching funds when facing a high-spending, privately-funded opponent or high levels of independent expenditures. In Arizona, both senate and house candidates are subject to a spending limit of $19,382. Connecticut Connecticut passed its public financing law in 2005, two years after then-gov. John G. Rowland resigned from office due to improper gift-giving practices. Connecticut s law was passed by the state legislature, as opposed to a citizen s initiative, as in Maine and Arizona. Gov. Jodi Rell, lieutenant governor for Rowland, took over following Rowland s resignation. Rell promised reform to government ethics in the state and was a supporter of a state clean elections law. In addition to offering public grants to qualifying candidates, Connecticut s law limits how much individuals can donate to political parties. It does not, however, limit the amount parties can funnel into individual campaigns. Candidates agree to limit spending to the amount of the public grant up to $85,000 for the senate and up to $25,000 for the house. Table 2 State Year Qualifying Amount Spending Limits Max Grant Amount Funding Source Source AZ 2002 GOV SEN HSE $22,050 (4,410 $500 contributions $1,100 (220 contributions) Overview of Public Finance Laws $957,333 (general election) $19,382 $957,333 (general election) $19,382 Tax check-off: $5 voluntary donations; surcharge on civil and criminal penalties Citzens referendum ME 2000 GOV SEN HSE 3,250 voters contributions not exceeding $50,000 150 voters contributions not exceeding $1500 50 voters not exceeding $500 $400,000 (general election) $20,082 (general election) $4,632 (general election) $400,000 (general election) $19,078 (contested) $7,631 (uncontested) $4,144 (contested) $1,658 (uncontested) Tax check-off: $3 individual, $6 joint; general fund; unspent seed money and candidate qualifying contributions; voluntary donations Citizens referendum CT 2008 GOV SEN HSE $250,000 in contributions of $100 or less $15,000 in contributions of $5-$100 $5,000 in contributions of $5-$100 During general election period, expenditures are limited to grant amount and unspent qualifying contributions $3,000,000 $85,000 $25,000 General fund appropriation, revenue from abandoned property; voluntary contributions State legislature - 17 -

Clearly, although the three states are similar in that they have clean elections systems, the stringency of their qualifying thresholds (shown in the table above) and spending limits vary widely. The states also differ in the nature of their state legislatures. Maine s legislature, for example, is far less professional than the Arizona or Connecticut legislature 24. It s likely that public financing has differing impacts depending on the professionalism of the legislature. What impact professionalism has, however, is less clear. On the one hand, participation may be higher in a citizen legislative state like Maine, as the public funds give potential, grassroots candidates the means to pursue their political ambitions. On the other hand, participation may be higher in more professional states, because seats are valued more and public financing, in theory, gives viable challengers a better chance against entrenched incumbents. PARTICIPATION FINDINGS An important indicator of whether or not we can judge the success of a public financing system on electoral outcomes is whether candidates are actually participating in the program. States with low participation essentially function as privately-financed states, because few candidates opt into the public system. It is not informative to examine the effect of public financing on election outcomes if participation is low. Participation in State Public Financing System AZ Senate AZ House ME Senate ME House CT % House PUB CT Senate % PUB FUNDING % PUB FUNDING % PUB FUNDING % PUB FUNDING FUNDING Combined House (AZ and ME) Combined Senate (AZ and ME) % PUB FUNDING % PUB FUNDING % PUB FUNDING 1996 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1998 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2000 33.3 73.3 48.6 29.1 0.0 0.0 41.7 48.6 2002 40.0 71.7 91.4 75.5 0.0 0.0 74.4 91.4 2004 50.0 83.3 94.3 92.1 0.0 0.0 89.6 94.3 2006 73.3 76.7 100.0 98.0 0.0 0.0 91.9 100.0 2008 66.7 93.9 94.3 96.1 76.2 94.4 95.4 94.3 Average 52.7 79.8 85.7 78.2 76.2 94.4 78.6 85.72 Table 3 Participation levels in each system depend on various factors, including the threshold a candidate must meet to qualify for public funds, how closely the public grant resembles actual campaign spending levels and the availability of matching funds to compete with highspending, privately-funded candidates. Maine and Connecticut, in particular, have achieved high participation rates. Both states offer fairly generous grant amounts per district population, with Maine senate and house candidates receiving $0.74 and $0.68 to spend per person and Connecticut candidates receiving $1.38 and $1.74 per person 25. In contrast, Arizona candidates who opt into the system receive only $0.16 per district resident. Arizona senators receive an almost identical grant to senators in Maine, even though Maine senators represent only about a 24 The Squire Index ranks states on the professionalism of their legislatures, based on factors such as salary and staffing level. 25 Mayer, Kenneth, Electoral Transitions in Connecticut, 2007. - 18 -

fifth of the number of people represented by Arizona senators. Arizona has the lowest participation rate among the three states, with 53 percent participation in the senate and 80 percent in the house. Connecticut, in its clean elections first year, had the highest participation in the senate, with 94.4 percent of candidates opting into the system. Connecticut, however, is excluded from the combined numbers for all years, including 2008, because all there are no numbers to faithfully compare Connecticut to. It would be misrepresentative to include Connecticut in presenting a trend of numbers over time. Although we can not pinpoint the specific factors in this study that make one public financing program more successful than another, we can examine which state performed the best postreform and draw conclusions about factors that may have influenced our results. Further, if all three states produce similar results, we can conclude that the impact of public financing holds up under various institutional structures. States & Offices Average Number of District Size seats State Legislature Characteristics Annual base salary (2009) Legislative staff per member (2009) Part-time or full Public funding grant amount per person Term limits Senate 170,000 30 701 total; 7.8 $0.16 AZ House 86,000 60 $24,000 per legislator Part-time $0.16 Yes 8 yrs ME $13,526 (1 st Senate 37,750 31,33, or 35 session); $0.74 $9,874 (2 nd 211 total; 1.2 House 8,750 151 session) per legislator Part-time $0.67 Yes 8 yrs Senate 98,000 36 617 total; 3.3 $1.38 No CT House 23,000 151 $28,000 per legislator Part-time $1.74 Table 4-19 -

Results Our findings are presented separately for each of our outcomes of interest. Generally, our findings show that public financing has had a significant impact on narrowing disparity in contributions between candidates and increasing the All Public House Races competitiveness of elections. Our results were largely mixed when considering the impact of public Outcome % RACES CONTESTED Difference 5.8 % RACES COMPETITIVE (20) 3.8 funding on the proportion of women who choose to % RACES W/ WOMAN 6.1 run for office and the rate of contestedness among % CONTESTED BY 3RD PARTY 10.9 third-party candidates. The following tables COMPETITIVE CONTRIBUTIONS 42.9 summarize the impact of public financing in Arizona All Public Senate Races and Maine relative to their four comparison states Outcome Difference (Connecticut will be considered individually later % RACES CONTESTED 12.7 because of its later implementation). All data for % RACES COMPETITIVE (20) 15.0 Arizona and Maine was pooled together and % RACES W/ WOMAN 15.5 % CONTESTED BY 3RD PARTY -11.3 compared to the data pooled together for New COMPETITIVE CONTRIBUTIONS 30.5 Mexico, Nevada, New Hampshire, and Washington. The difference column indicates the percentage point difference between outcomes in the public finance states versus their non-public-finance comparison states. For example, the proportion of contested House races in the public finance states was 5.8 percentage points higher than in the non-publicly financed comparison states. This summary for our pooled results indicates that public financing had positive results for each category except for the percent of races contested by a third-party candidate. Individual results for Maine, Arizona and Connecticut, however, showed more variation. THE IMPACT OF PUBLIC FINANCING ON CAMPAIGN MONEY Our data also showed that public financing systems greatly reduce the disparity between candidate s fundraising amounts. All three public financing states experience large gains in the proportion of competitive races measured by the contribution ratio (in our case, a ratio of 1.5 or less). For example, in the Connecticut House, the average proportion of competitive races between 1996 and 2006 jumped from 27.9 percent competitive to 70.4 percent competitive in 2008. These results are promising but also expected. Candidates who opt into the public financing system are prohibited from raising any additional money through private contributions. Therefore, we would expect less disparity in contribution levels with high rates of participation in the public financing system, as candidates receive similar amounts through the public grant system. With the exception of races in the Arizona Senate, more than 75 percent of the races in our three states had at least one candidate accepting public financing. In the Arizona Senate, only 53 percent of the races had at least one candidate opting into the system. While we would expect contribution levels within a race to be close in our public financing states, we still see some variation in the number of districts with competitive contribution ratios in Arizona, Maine and Connecticut each year because of varying levels of participation in the - 20 -

system. If only one candidate participates in the system in a district, then contribution ratios may still be uncompetitive as the privately-funded candidate can raise as much money as he or she can. Impact on percentage of total House races that have a competitive contribution ratio (defined as 1.5) Public Financing State Difference Comparison State1 Difference Comparison State 2 Difference Public Difference Arizona 30.2 New Mexico -2.8 Nevada 5.7 28.6 Connecticut 42.5 New York -0.5 Delaware 3.5 41.0 Maine 28.9 New Hampshire INC Washington -37.1 66.0 26 Table 5 When Arizona and Maine House races are pooled together and compared against their four comparison states, again as expected, the disparity in contribution levels decreases significantly. In the Maine and Arizona house races, the proportion of competitive races measured by the contribution ratio increased by 42.9 percentage points more than in the comparison states. It should be noted that in states without public financing laws, the level of contribution competitiveness deteriorated in many cases post-reform. This means that there were increasing disparities between candidates contribution levels in non-public-financing states, as public financing states greatly leveled the playing field in terms of money (most likely due to a publicly funded candidates inability to raise additional private contributions). For example, Washington s proportion of races with competitive contribution ratios decreased, as seen in Table 5, by 37.1 percentage points over the 12-year time period. Similar results are found in the Senate (Table 6). For complete year-by-year, state-by-state data please see the Appendix. Impact on percentage of total Senate races have a competitive contribution ratio (defind as 1.5) Public Financing State Difference Comparison State1 Difference Comparison State 2 Difference Public Difference Arizona 25.3 New Mexico -1.0 Nevada -14.3 33.0 Connecticut 32.8 New York 5.5 Delaware 1.2 13.8 Maine 28.3 New Hampshire -3.2 Washington -9.2 34.5 Table 6 26 INC means the data was incomplete. - 21 -

THE IMPACT OF PUBLIC FINANCING ON ELECTORAL OUTCOMES Impacts on Election Competitiveness Our data show that public financing increases the competitiveness of elections. This finding is consistent no matter what level we place the margin of victory at: 10, 15, or 20 percent. The findings for the House can be viewed below. Table 7 uses our most conservative estimate, using a margin of victory of 10 percent, and Table 8 uses our most liberal estimate of a 20 percent margin of victory. Impact on percentage of total House races that are competitive (margin of victory of 10% or less) Public Financing State Difference Comparison State1 Difference Comparison State 2 Difference Public Difference Arizona 13.0 New Mexico 6.9 Nevada 6.0 6.5 Connecticut 1.2 New York -3.4 Delaware 15.2-5.0 Maine 8.7 New Hampshire 6.5 Washington -6.1 8.5 Table 7 Impact on percentage of total House races that are competitive (margin of victory of 20% or less) Public Financing State Difference Comparison State1 Difference Comparison State 2 Difference Public Difference Arizona 7.6 New Mexico -1.7 Nevada -5.6 11.3 Connecticut 4.1 New York -2.5 Delaware 9.9 0.4 Maine 4.1 New Hampshire 5.4 Washington -2.8 2.7 Table 8 Senate races appeared to be more affected by public financing laws than state house races. It should be noted that Arizona had very low rates of competitive races prior to public financing. In 1996, only 13.0 percent of house races in Arizona had a vote margin of 10 percent or less between the winner and second-highest vote-getter. In 1998, 5.3 percent of races were competitive. Meanwhile, in Nevada in 1996 and 1998, 21.4 percent and 17.1 percent of house races were competitive. Likewise, in New Mexico in 1996 and 1998, 25.0 and 12.5 percent of races were competitive. The data show that Arizona house races post-reform are consistently more on par with the two comparison states. The average proportion of competitive races in the Arizona House pre-reform was 9.2; the average post-reform was 22.1. Nevada and New Mexico had much smaller gains over this time period. Given that Arizona started from behind, our results show that the state was able to gain ground on its comparison states. The results are similar in Maine, which started more on-par with its comparison states but gained more. In this case, Washington actually experienced fewer competitive races over the time period. Results in Connecticut indicated that the state lost ground relative to its comparisons. However, we only have data for one election post-reform in Connecticut, making it difficult to draw substantial conclusions. Impact on percentage of total Senate races that are competitive (margin of victory of 10% or less) Public Financing State Difference Comparison State1 Difference Comparison State 2 Difference Public Difference Arizona 6.0 New Mexico 11.7 Nevada -15.7 7.9 Connecticut 6.4 New York 6.9 Delaware 14.6-4.3 Maine 12.8 New Hampshire 0.1 Washington -10.3 17.9-22 -

Table 9 Impact on percentage of total Senate races that are competitive (margin of victory of 20% or less) Public Financing State Difference Comparison State1 Difference Comparison State 2 Difference Public Difference Arizona 14.9 New Mexico -4.6 Nevada -10.0 22.2 Connecticut 13.5 New York 8.0 Delaware 31.3-6.2 Maine 5.3 New Hampshire 2.7 Washington -14.0 10.9 Impacts on Election Contestedness Table 10 Levels of contestedness were often high in public financing and comparison states prior to reform. Therefore, we would expect to see a less pronounced gain for this outcome variable. In Maine, gains were moderate. Maine, prior to reform, already had high rates of contestedness. In 1996, for example, 92.7 percent of Maine House races were contested. Nevertheless, Maine rates of contestedness still increased by more post-reform than comparison state rates. This is largely due to the fact that the proportion of contested races in comparison state house and senate races stayed fairly constant when comparing the pre-reform time period to the post-reform period. Ultimately, the proportion of contested races in the Maine House increased by 2.4 percentage points more than in the comparisons, and the proportion of Maine Senate races increased by 4.6 percentage points more, as seen in Table 11 and 12 below, respectively. Impact on percentage of total House races that are contested by any candidate Public Financing State Difference Comparison State1 Difference Comparison State 2 Difference Public Difference Arizona 16.8 New Mexico -14.9 Nevada 3.1 22.8 Connecticut -0.7 New York -9.9 Delaware 13.5-2.5 Maine 5.5 New Hampshire 7.1 Washington -0.9 2.4 Table 11 Impact on percentage of total Senate races that are contested by any candidate Public Financing State Difference Comparison State1 Difference Comparison State 2 Difference Public Difference Arizona 15.7 New Mexico -25.5 Nevada 5.7 25.5 Connecticut 5.0 New York 2.1 Delaware 0.0 4.0 Maine 6.0 New Hampshire 2.5 Washington 0.3 4.6 Table 12 Arizona experienced even higher gains relative to its comparison states. However, similar to our competitive analysis, Arizona had low rates of contestedness relative to its comparisons at baseline. In the Arizona House, for instance, only 47.5 percent of races were contested prereform. Post-reform, 64.3 percent of races were contested on average between 2000 and 2008. Our results again show Connecticut losing ground to its comparison states on contestedness. An examination of the numbers shows that Connecticut largely remained steady in the proportion of contested races pre- and post-reform, while its comparison states had varied outcomes. Third Party Contestedness - 23 -

Our pooled data discussed at the beginning of the findings section shows the percentage of house races contested by a third-party candidate (e.g., Libertarian, Green etc.) increased by 10.9 percentage points in public financing states relative to non-public financing states. However, in senate races, we found an opposite result; the percentage of third-party candidates in public financing states decreased by 11.3 percentage points relative to comparison states. The proportions of races contested by a third-party candidate varied greatly from year-to-year in many of the states; this is due, in a large part, to the small number of races with a third-party candidate. With a very small number of races, just one or two additional races can greatly impact the overall proportion of races. Impact on percentage of total House races that are contested by third party candidate Public Financing State Difference Comparison State1 Difference Comparison State 2 Difference Public Difference Arizona 1.8 New Mexico -3.4 Nevada 3.3 1.8 Connecticut 0.0 New York -18.4 Delaware -1.9 10.1 Maine 9.9 New Hampshire -3.1 Washington 5.7 8.6 Table 13 Impact on percentage of total Senate races that are contested by third party candidate Public Financing State Difference Comparison State1 Difference Comparison State 2 Difference Public Difference Arizona 1.3 New Mexico 2.4 Nevada 33.3-16.5 Connecticut -6.2 New York -26.3 Delaware -13.5 13.8 Maine -4.0 New Hampshire 5.2 Washington 12.0-12.6 Table 14 With more time, we would have studied primary elections in addition to general elections. It s highly likely that more third-party candidates show up in the primary elections. By examining the primary elections, we would have more data points to observe. Also, we would gain a better understanding about whether the promise of public financing induces lesser-known candidates (which includes third-party candidates) to enter a race. By only examining the general elections, we are only able to examine part of the overall picture. - 24 -

THE IMPACT OF PUBLIC FINANCING ON WOMEN CANDIDATES Our final electoral outcome measure is gender diversity, or the percentage of races that contained at least one female candidate. The results favor the public financing states on this measure. In our pooled comparison, the proportion of house races with a female candidate increased by 6.1 percentage points more in public financing states than in comparison states; the percentage of senate races increased by 15.5 percentage points more. Our pooled data comparison indicates that public finance states made small gains in the proportion of districts with a female candidate while comparison states remained steady. However, results from yearto-year tended to be erratic, making it difficult to draw solid conclusions about diversity impacts. The results disaggregated from the pooled results are shown in Tables 15 and 16. Impact on percentage of total House races that have a woman candidate present Public Financing State Difference Comparison State1 Difference Comparison State 2 Difference Public Difference Arizona -3.0 New Mexico 5.0 Nevada -8.4-1.3 Connecticut 0.3 New York -6.6 Delaware -1.9 4.5 Maine 6.1 New Hampshire -5.6 Washington -2.5 4.5 Table 15 Impact on percentage of total Senate races that have a woman candidate present Public Financing State Difference Comparison State1 Difference Comparison State 2 Difference Public Difference Arizona 21.0 New Mexico -11.8 Nevada 22.7 15.5 Connecticut -1.7 New York -8.8 Delaware -44.8 25.1 Maine -7.1 New Hampshire -2.6 Washington -5.5-3.1 IMPACT OF PUBLIC FINANCING WITHIN PUBLIC FINANCING STATES Table 16 In order to further underscore our findings, we conducted a within-state analysis for Maine and Arizona, comparing the districts where at least once candidate accepted public financing to districts where neither candidate opted into the system. It is our hypothesis that races with one or more publicly-financed candidates may be more competitive than races within the same state where neither of the candidates in the race receive public financing. As we showed earlier, public grants reduce disparities in contribution levels between candidates and, therefore, may increase competitiveness in races. However, one could also hypothesize that privately-financed races may be more competitive, as it s possible that both candidates opted out of the system in anticipation of a high-spending race. Our within-state analysis revealed similar results to our larger analysis. In Maine, the proportion of competitive races (measured through a 10 percent vote margin) in publiclyfinanced house districts was 9.4 percentage points higher than in privately-financed districts over the eight-year post-reform period 27. In Arizona, the mean proportion of competitive 27 A district is considered to be publicly-financed if just one candidate accepts public financing. Privately-financed districts are those in which none of the candidates opted into the public financing program. - 25 -

publicly-financed house districts was 26.5 percentage points higher than in privately-financed districts post-reform. Table 17 indicates the difference when comparing the two types of races. On our contestedness measure, the differences between privately- and publicly-financed house districts were even higher. In Maine, 95.3 percent of publicly-financed districts were contested; in contrast, 75.3 percent of privately-funded districts were contested. The results in Arizona are even starker. Over the post-reform period, 73.8 percent of publicly-financed districts were contested versus 28.9 percent of privately-financed races. Arizona, however, has fewer districts and thus fewer privately-financed districts to examine for this analysis. Only 57 districts were privately-financed between 2000 and 2008 (in 2008 alone, only four districts fell into this category); in comparison, Arizona had 221 publicly-financed districts between 2000 and 2008. We try to address this disparity by weighting our results; however, Arizona estimates may need to be interpreted with caution. Within State (Public vs. Private Financed Races) House Outcome Arizona Maine Overall % RACES CONTESTED 33.8 19.7 26.7 % RACES COMPETITIVE (20) 54.4-3.3 25.5 % RACES W/ WOMAN 48.7 15.2 31.9 % CONTESTED BY 3RD PARTY 0.4 7.2 3.8 COMPETITIVE CONTRIBUTIONS 29.8 49.5 39.7-26 - Table 17 Similar to our between-state analysis, we found small-to-no results for our third-party contestedness variable. Again, with a small number of minority party candidates to assess, we have few data points to analyze for this outcome. Also similar to our between-state results were our contribution ratio estimates. Publiclyfinanced house districts in both Maine and Arizona had fewer funding disparities between candidates than privately-financed races. Again, because candidates who opt into the public financing program are prohibited from raising additional money, we would expect publiclyfinanced candidates to be more similar in contribution levels than privately-financed candidates who are not limited in their fundraising abilities. One notable difference when comparing our two analyses is the gender outcome variable. In Maine, our within-state analysis shows that women were much more likely to appear in publicly-financed house districts. In the eight years after reform, 49 percent of publicly-financed house district races in the state involved at least one woman; in comparison, 36 percent of privately-financed districts had at least one woman over the same period. Results from the state senate in Arizona and Maine were more difficult to interpret given the very low number of privately-financed districts (often in the single digits). In fact, senate results were negative across-the-board in Maine and largely negative in Arizona. Privately-financed districts performed better in almost all outcomes of competitiveness, contestedness and

diversity. However, it s difficult to conclude whether these estimates - which contrast starkly with the house analysis - are the result of a fundamental difference in the effect of public financing on the senate versus the house, or simply the result of a very small sample size producing questionable estimates. IMPACT OF PUBLIC FINANCING ON GUBERNATORIAL ELECTIONS The gubernatorial elections involved just a small number of races, making it difficult to draw any conclusions about electoral outcomes. However, the data does indicate that at least a few gubernatorial candidates in Maine and Arizona took advantage of the public financing system. Post-reform in Arizona, half of the candidates in 2002 and two-thirds of the candidates in 2006 opted into the state s program. In Maine, just one candidate out of four opted into the system in 2002, but in 2006, three out of five participated in the system. It s unclear, based on just two election cycles, whether this participation trend will increase, decrease or stay about the same; however, our data show that at least some gubernatorial candidates believe public funding is a worthwhile option, despite its accompanying spending limit restrictions. Part IV. Limitations of Findings The major limitation of our study is that we were limited to calculating difference-in-difference estimates. With the difference-in-difference technique, we must work under the assumption that unmeasured trends that impact our outcomes variables operate in a similar manner over time in our public financing states and their two comparisons. If one of our comparison states had a systematic tendency over our time period to perform differently in a way that impacts competitiveness or diversity, then our results could be biased. For example, if the redistricting process in one of our comparison states has a tendency to pit two incumbents against each other more often than in the public financing state, then our competitiveness results could be biased (incumbents being pitted together impacts our ability to assess changes in competitiveness). We must assume this isn t the case. We know that Maine and Arizona both underwent redistricting during the time period in our study; Arizona drew new districts in 2002 and Maine in 2004. In his study of public financing, Neil Malhotra reasons that the political and economic changes produced in each district as a result of redistricting do not significantly affect the primary institutional change, the presence of clean elections 28. Provided incumbents aren t often pitted against each other as a result of redistricting, a shifting of district lines simply shifts candidates and shouldn t dramatically thwart our ability to study the impact of clean elections on our electoral outcomes of interest 28 Malhotra, Neil. The Impact of Public Financing on Electoral Competition: Evidence from Arizona and Maine. State Politics and Policy Quarterly. Vol. 8, No. 3. Fall 2008. http://www.stanford.edu/~neilm/sppq.pdf - 27 -

(such as whether challengers were able to build more competitive campaigns against incumbents). Another potential confounder is term limits. If any states enacted or repealed term limits over the period of our study, the resulting impacts on competitiveness and contestedness related to term limits could be wrongly attributed to public financing. In fact, Washington s term limits law was voided by the state supreme court in 1998; however, given that this is at the very beginning of our study period, the change doesn t affect the consistency of our data over our 12- year time period. Nevada implemented term limits in 1998, but the law will not actually impact legislators until the 2010 election cycle because the term limit is 12 years. Given the assumptions we must make using the difference-in-difference technique, our next step with this analysis would be to perform a more rigorous fixed effects analysis that would allow us to control for additional covariates that may influence competitiveness, diversity or contestedness in our reform or comparison states (i.e., other changes to a state s campaign finance laws over the time period, such as the implementation of term limits). A second option would be to collect data for additional comparison states to determine whether our results hold up when public funding states are matched with other similar comparison states. A second major limitation of this study was our inability to tease out the exact effect of public financing on our electoral outcomes. We are only able to estimate the impact of the overall reform, which includes both the public grants and spending limits. Furthermore, all of the states in our study have additional campaign finance regulations in the form of contribution limits and disclosure requirements. The structure of each state s full campaign finance system likely impacts the effectiveness of the public grants. Researchers have devised a method to quantify campaign finance regulations into a single index, allowing for easier assessment of complicated systems 29. We would be interested in exploring how the full campaign finance structure in each state impacts the effectiveness of public financing; for example, we could hypothesize that public financing reform would be less effective in states that already have strict contribution limits. Candidates in these states are more limited in their fundraising abilities, which may keep spending in check. In contrast, states with lax contribution limits may produce wide disparities in funding levels of individual candidates; if public financing were implemented in this type of state, it may produce a bigger impact. One final limitation to our study was our inability to examine the impact of clean elections laws on primary elections. Because of time constraints, we were not able to complete data collection for primary races. However, we hypothesize that public finance laws may produce greater impacts on contestedness outcomes in primary elections than in general elections. By the time a general election takes place, many challengers have already been weeded out of the process. 29 Christopher Witko developed a Stringency Index, which attempts to quantify public financing laws, contribution limits and disclosure requirements into as single index for each state. Measuring the Stringency of State Campaign Finance Regulation. State Politics and Policy Quarterly. Vol. 5, No. 3. Fall 2005. - 28 -

Furthermore, we didn t see a strong impact of public financing reform on gender diversity in our general election study. However, given that primary elections involve more candidates, it s possible that more women were able to use public funds to launch a campaign, but never made it to the general election stage. Without being able to examine what occurred during the primary elections, we are only able to see a portion of the picture. With more time, we would want to examine the impact of public financing on all of our election outcomes during the primary elections. Part V. Policy Implications A lack of competition in state legislative races is a direct result of challengers having inadequate access to finances necessary to establish and sustain effective campaigns. Therefore, the implementation of a public finance system is an appropriate measure to take in order to ameliorate systems that lack contestedness by constraining the costs of mounting campaigns. Furthermore, enacting a public campaign finance system eases the financial barriers that prohibit potential candidates and results in more competitive races. Another important way in which public finance systems matter is through its effect on the presence of more diversity amongst candidates. Public finance systems seem to successfully raise the participation of women in state legislative races. While public financing is not a panacea for the overall level of competition in state legislative races, it positively impacts diversity of the challengers who select into the system and ensures that candidates are more reflective of the electorate. However, public financing systems do not operate in a vacuum. The structure of the overall campaign finance system in each state must be viewed holistically, including the importance of term limits, contribution limits, and matching funds. State legislatures should consider these various levers when considering campaign finance laws. Nevertheless, our findings indicate that public financing is a valuable tool to consider whenever a state s level of competitive elections are abysmally low, as was the case in Arizona prior to public financing. Public financing is not necessarily a silver bullet but it is a potential weapon to combat entrenched interests. - 29 -

APPENDIX A. CLUSTER ANALYSIS: METHOD USED TO SELECT THE COMPARISON STATES We sought to match states based on geographic proximity but also on demographic and political makeup. We were limited in our choice of comparison states, because all states with partial public financing for legislative or gubernatorial elections were eliminated from our pool of potential comparison states. Washington, for example, clearly does not share regional proximity to Maine; however, with the exception of New Hampshire, all New England states have public funding systems and could not be considered as potential matches (Massachusetts repealed its underused and unfunded system in 2003, but given that a system was in place during our study period, we opted not to include it in our matching analysis). Washington proved to be an adequate match for Maine on characteristics described below. A sample of the kinds of characteristics we matched states on are seen in the following table. % Education Democrat/Republican gap* % Manufacturing % Age (55+) %White (Associates or higher) Maine 13 12.6 28 80.5 34.8 Washington 15 11.4 23.3 95.3 39.9 New Hampshire 8 11.5 24.9 94.9 41.7 Arizona 2 9.6 23.3 77.7 33.1 Nevada 7 9.8 22.5 74.9 28.6 New Mexico 13 5.6 23.8 70.1 32.1 Connecticut 22 10.3 24.9 79.9 42.2 Delaware 22 10.3 25.1 72.6 34.2 New York 23 10.1 24.4 66.7 39.8 *Positive means Democratic advantage; zero, for instance, would be equal Democratic and Republican advantage To determine demographic, economic and political similarity, we performed a cluster analysis using statistical software. For our analysis, we gathered data on almost all of the states with no form of public financing 30. Measures included in the analysis were: employment, income, education, age, race, political leanings, legislative professionalism and an index that attempts to capture state culture. The first several categories were collected through the U.S. Census Bureau s American Community Survey for 2006 through 2008. For employment, we collected data from non-public financing states on the percent of the population working manufacturing, farming, and the professional sector and the percent employed by the government. For education, we used data on 30 A few states with no public financing in the Southeast were left out of the cluster analysis because they were unlikely to be good matches for our public financing states in the Northeast and Southwest. - 30 -

the percent of the population with an associate s degree or higher. Age is broken into two categories the percent of the population 55 and older and those below. Race includes percent white, black and Asian, and percent Hispanic. The income categories are the percentage of families earning above $100,000, between $50,000 and $100,000 and below $50,000. The political bent of a state is captured through a 2009 Gallup Poll measure that quantifies the gap between leaned party identification in a state using poll data tracking 31. The analysis takes into account independents with partisan leanings, in addition to those who identify with the Republican or Democratic parties. State legislative professionalism is quantified through the widely-used Squire Index, which takes into account factors such as legislative salary, staffing level and part-time or fulltime status in order to rank state legislatures. Lastly, in an attempt to capture differences in state culture, we included the paternalism index, a measure developed by academics at the University of Buffalo that takes into account taxes and regulatory policies, such as gun control, to determine how free a state is from government intervention 32. Because the paternalism index has not been widely used, we also ran an analysis without the index; the various state clusters did not change dramatically. The statistical program uses data from each state to group states with similar characteristics. Through the analysis, the program attempts to limit variation within state groupings and increase variation between groups. To view a sample of data used in the analysis, please see the next page. The table below includes data only for the states selected into our study. In total, 24 states were included in the analysis. 31 Gallup Poll State of the States: Political Party Affiliation. January 2009. http://www.gallup.com/poll/114016/state-states-political-party-affiliation.aspx 32 Sorens, Jason and William Ruger. Freedom in the 50 States: An Index of Personal and Economic Freedom. University of Buffalo. 2009. - 31 -

APPENDIX B. EXPENDITURES AND CONTRIBUTIONS COMPARISON Expenditures and Contributions Comparison Expenditures Contributions Mean Median Mean Median AZ House 1998 26711 18094 21871 15314 2000 25521 25370 30316 27169 2002 27923 28145 29363 29005 2004 32073 31207 31785 31373 2006 31042 32035 35870 33187 2008 47745 41613 47053 41017 %Change 98-08 78.7 130.0 115.1 167.8 AZ Senate 1998 23133 14460 32010 19698 2000 30428 26803 37179 31133 2002 28110 24652 32519 28095 2004 31230 29705 39970 33425 2006 36014 31863 38409 32869 2008 42099 35714 53432 40370 %Change 98-08 82.0 147.0 66.9 104.9 ME House 1998 5111 4250 5739 5026 2000 3395 3076 3990 3763 2002 4038 4596 4402 4877 2004 4587 4784 4952 4985 2006 5123 4959 5723 5324 %Change 98-06 0.23 16.68-0.28 5.93 ME Senate 1998 18163 17853 19673 18689 2000 18052 14695 19461 14988 2002 17080 19428 18634 19769 2004 23052 19243 24320 19805 2006 22027 22616 23184 23262 %Change 98-06 21.27 26.68 17.85 24.47 CT House CT Senate 2008 25862 28427 21909 29428 2008 90657 97138 79914 99505 The above table is a pre-post analysis of candidate expenditures in Arizona, Connecticut, and Maine. This analysis shows that contributions are a relevant proxy for candidate expenditures. - 32 -

APPENDIX C. GUBERNATORIAL DESCRIPTIVE STATISTICS STATE YEAR %WOMAN AZ 1996 % MINORITY CANDIDATES % PARTICIPATE PUB FUNDING % MIN PARTY CANDIDATES WINNER VOTE MARGIN AVG CONTRIBUTION TOTAL MEDIAN CONTRIBUTION TOTAL AZ 1998 50.0% (2) 0.0% (0) 0.0% (0) 50.0% (2) 25.4 $891,898 $728,312 AZ 2000 AZ 2002 25.0% (1) 0.0% (0) 50.0% (2) 50.0% (2) 1 $1,540,000 $1,920,000 AZ 2004 AZ 2006 33.3% (1) 0.0% (0) 66.7% (2) 33.3% (1) 27.1 $994,119 $1,370,000 AZ 2008 Gubernatorial Descriptive Statistics ME 1996 ME 1998 20.0% (1) 0.0% (0) 0.0% (0) 60.0% (3) 39.7 $189,710 $71,677 ME 2000 ME 2002 0.0% (0) 0.0% (0) 25.0% (1) 50.0% (2) 5.7 $904,864 $964,981 ME 2004 ME 2006 40.0% (2) 0.0% (0) 75.0% (3) 60.0% (3) 16.6 $940,473 $1,130,000 ME 2008 CT 1996 CT 1998 40.0% (2) 0.0% (0) 0.0% (0) 60.0% (3) 27.5 $2,090,000 $1,150,000 CT 2000 CT 2002 0.0% (0) 0.0% (0) 0.0% (0) 0.0% (0) 12.2 $4,490,000 $4,490,000 CT 2004 CT 2006 25.0% (1) 25.0% (1) 0.0% (0) 50.0% (2) 27.8 $2,750,000 $4,050,000 CT 2008 NY 1996 NY 1998 20.0% (2) 20.0% (2) 80.0% (8) 21.2 $10,100,000 $7,100,000 NY 2000 NY 2002 0.0% (0) 12.5% (1) 75.0% (6) 15.9 $17,700,000 $6,360,000 NY 2004 NY 2006 16.7% (1) 50.0% (3) 66.7% (4) 40.8 $8,780,000 $1,930,000 NY 2008 DE 1996 DE 1998 DE 2000 33.3% (1) 0.0% (0) 33.3% (1) 19.5 $1,030,000 $1,530,000 DE 2002 DE 2004 33.3% (1) 0.0% (0) 33.3% (1) 5.1 $907,148 $906,388 DE 2006 DE 2008 0.0% (0) 0.0% (0) 0.0% (0) 35.5 $2,120,000 $2,120,000 NM 1996 NM 1998 0.0% (0) 0.0% (0) 0.0% (0) 9.1 $2,720,000 $2,730,000 NM 2000 NM 2002 0.0% (0) 66.7% (2) 33.3% (1) 16.4 $3,730,000 $2,920,000 NM 2004 NM 2006 0.0% (0) 50.0% (1) 0.0% (0) 37.6 $6,840,000 $6,840,000 NM 2008 NV 1996 NV 1998 0.0% (0) 0.0% (0) 0.0% (0) 10.2 $3,420,000 $3,420,000 NV 2000 NV 2002 0.0% (0) 20.0% (1) 66.7% (4) 48.5 $689,900 $4,984 NV 2004 NV 2006 25.0% (1) 0.0% (0) 50.0% (2) 4.2 $2,610,000 $1,920,000 NV 2008 WA 1996 50.0% (1) 50.0% (1) 0.0% (0) 15.9 $2,130,000 $2,130,000 WA 1998 WA 2000 0.0% (0) 33.3% (1) 0.0% (0) 18.7 $2,190,000 $2,840,000 WA 2002 WA 2004 66.7% (2) 0.0% (0) 33.3% (1) 6.5 $4,260,000 $6,330,000 WA 2006 WA 2008 50.0% (1) 0.0% (0) 0.0% (0) 39.7 $11,800,000 $11,800,000 NH 1996 n/a n/a n/a n/a n/a NH 1998 33.3% (1) 0.0% (0) 33.3% (1) 35.3 $1,580,000 $1,580,000 NH 2000 50.0% (2) 0.0% (0) 50.0% (2) 5 $1,430,000 $1,420,000 NH 2002 0.0% (0) 0.0% (0) 33.3% (1) 20.5 $4,080,000 $793,904 NH 2004 0.0% (0) 0.0% (0) 0.0% (0) 2.1 $3,550,000 $3,550,000 NH 2006 0.0% (0) 0.0% (0) 0.0% (0) 46.5 $1,380,000 $1,380,000 NH 2008 33.3% (1) 0.0% (0) 33.3% (1) 42.6 $702,500 $702,500-33 -

APPENDIX D. STATE-BY-STATE, YEAR-BY-YEAR RESULTS STATE YEAR %CONTESTED %COMPETITITIVE (10) % COMPETITIVE (15) %COMPETITIVE20 %WOMAN % CONTESTED BY 3RD PARTY COMPETITIVE CONTRIBUTIONS AZ 1996 50.0 13.0 16.7 54.2 73.3 30.0 33.0 AZ 1998 45.0 5.3 15.8 50.0 73.3 13.3 20.0 AZ PRE 47.5 9.2 16.3 52.1 73.3 21.7 26.5 AZ 2000 50.0 16.4 18.2 60.0 83.3 30.0 53.3 AZ 2002 55.0 22.8 31.6 73.3 68.3 11.7 66.7 AZ 2004 58.3 10.7 19.6 48.4 66.7 30.0 53.3 AZ 2006 58.3 14.0 18.0 52.0 63.3 26.7 60.0 AZ 2008 100.0 46.7 58.3 65.0 69.7 19.0 50.0 AZ Post 64.3 22.1 29.1 59.7 70.3 23.5 56.7 NM 1996 63.2 25.0 44.1 47.1 36.8 20.8 8.8 NM 1998 69.2 12.5 38.7 51.6 42.0 12.5 8.7 NM Pre 66.2 18.8 41.4 49.4 39.4 16.7 8.8 NM 2000 78.6 17.1 25.0 38.6 44.3 29.2 11.4 NM 2002 48.6 17.1 23.5 41.2 41.2 16.7 4.3 NM 2004 41.4 28.6 42.9 50.0 45.7 8.3 2.9 NM 2006 43.5 36.7 50.0 63.3 46.4 8.3 7.2 NM 2008 44.3 29.0 35.5 45.2 44.3 4.2 4.3 NM Post 51.3 25.7 35.4 47.7 44.4 13.3 6.0 NV 1996 88.1 21.4 26.2 37.8 50.0 16.7 11.9 NV 1998 82.9 17.1 31.7 64.7 53.7 7.1 14.6 NV Pre General Election House Races Pre & Post Public Financing Comparisons (AZ: NM, NV) 85.5 19.3 29.0 51.3 51.9 11.9 13.3 NV 2000 88.1 35.7 41.0 58.8 29.3 8.3 21.4 NV 2002 85.7 42.9 42.9 50.0 71.4 4.8 14.3 NV 2004 95.2 12.5 27.5 42.5 57.1 22.6 19.0 NV 2006 73.8 25.8 32.3 48.4 14.3 16.7 23.8 NV 2008 100.0 9.5 21.4 28.6 45.2 23.8 16.7 NV Post 88.6 25.3 33.0 45.7 43.5 15.2 19.0 Summary of Statistics PRE-PUBLIC Arizona 47.5 9.2 16.3 52.1 73.3 21.7 26.5 Comp. 75.9 19.0 35.2 50.3 45.6 14.3 11.0 POST-PUBLIC Arizona 64.3 22.1 29.1 59.7 70.3 23.5 56.7 Comp. 69.9 25.5 34.2 46.7 43.9 14.3 12.5 Arizona Difference *16.82 *12.97 *12.89 7.64-3.04 1.83 *30.16 Comparison Differ *-5.93 *6.49-0.98-3.64-1.71 *0.02 1.53 Difference-in-Diffe *22.75 6.48 13.87 11.28-1.33 1.82 *28.63 *significant at 5% level - 34 -

STATE YEAR %CONTESTED %COMPETITITIVE (10) % COMPETITIVE (15) %COMPETITIVE20 %WOMAN % CONTESTED BY 3RD PARTY COMPETITIVE CONTRIBUTIONS AZ 1996 60.0 5.6 5.6 22.2 40.0 20.0 0.0 AZ 1998 43.3 23.1 38.5 38.5 30.0 13.3 13.3 AZ PRE 51.7 14.4 22.1 30.4 35.0 16.7 6.7 AZ 2000 70.0 26.3 42.1 57.9 73.3 20.0 40.0 AZ 2002 60.0 16.7 16.7 22.2 50.0 23.3 23.3 AZ 2004 53.3 6.2 25.0 37.5 40.0 20.0 20.0 AZ 2006 80.0 25.0 37.5 54.2 50.0 13.3 43.3 AZ 2008 73.3 27.3 45.5 54.5 66.7 13.3 33.3 AZ Post 67.3 20.3 33.4 45.3 56.0 18.0 32.0 NM 1996 66.7 25.0 46.4 53.6 45.2 2.4 7.1 NM 1998 NM Pre 66.7 25.0 46.4 53.6 45.2 2.4 7.1 NM 2000 57.1 12.5 25.0 45.8 36.6 11.9 2.4 NM 2002 NM 2004 40.5 17.6 23.5 41.2 33.3 2.4 7.1 NM 2006 NM 2008 26.1 80.0 60.0 60.0 30.4 0.0 8.7 NM Post 41.2 36.7 36.2 49.0 33.4 4.8 6.1 NV 1996 90.0 22.2 33.3 33.3 30.0 0.0 20.0 NV 1998 80.0 37.5 50.0 50.0 10.0 0.0 30.0 NV Pre General Election Senate Races Pre & Post Public Financing Comparisons (AZ: NM, NV) 85.0 29.9 41.7 41.7 20.0 0.0 25.0 NV 2000 100.0 10.0 40.0 40.0 40.0 10.0 0.0 NV 2002 100.0 0.0 33.3 33.3 0.0 0.0 33.3 NV 2004 100.0 10.0 20.0 20.0 40.0 60.0 20.0 NV 2006 63.6 28.6 42.9 42.9 45.5 36.4 0.0 NV 2008 90.0 22.2 22.2 22.2 88.0 60.0 0.0 NV Post 90.7 14.2 31.7 31.7 42.7 33.3 10.7 Summary of Statistics PRE-PUBLIC Arizona 51.7 14.4 22.1 30.4 35.0 16.7 6.7 Comp. 75.9 27.4 44.0 47.6 32.6 1.2 16.1 POST-PUBLIC Arizona 67.3 20.3 33.4 45.3 56.0 18.0 32.0 Comp. 66.0 25.4 33.9 40.3 38.1 19.0 8.4 Arizona Difference *15.67 *5.95 *11.31 14.9 21.0 1.3 *25.33 Comparison Difference *-9.87 *-1.99-10.1-7.3 5.5 *17.82-7.7 Difference-in-Difference *25.54 7.9 21.4 22.2 15.5-16.5 *33.02 *significant at 5% level

STATE YEAR %CONTESTED %COMPETITITIVE (10) % COMPETITIVE (15) %COMPETITIVE20 %WOMAN % CONTESTED BY 3RD PARTY COMPETITIVE CONTRIBUTIONS CT 1998 75.5 7.9 22.8 29.8 47 10.6 25.5 CT 2000 66.2 13 23 29 40.4 16.6 23.9 CT 2002 76.4 19.1 26.4 37.3 45.1 23.6 29.6 CT 2004 78.1 13.4 21.8 28.6 45.7 37.1 25.8 CT 2006 74.8 15.9 21.2 24.8 49 24.5 34.6 CT Pre 74.2 13.86 23.04 29.9 45.44 22.48 27.88 CT 2008 73.5 15.1 28.3 34 45.7 22.5 70.4 CT Post 73.5 15.1 28.3 34 45.7 22.5 70.4 NY 1998 87.2 5.4 8.5 8.5 50 45.3 12.2 NY 2000 86.7 6.9 9.9 13.7 44.7 35.3 20.3 NY 2002 91.3 6.6 8.8 14.6 48 50 19.8 NY 2004 76 8.8 10.6 16.8 40.7 12.7 20.3 NY 2006 74.7 8.8 15 22.1 43.3 12 13.2 NY Pre 83.18 7.3 10.56 15.14 45.34 31.06 17.16 NY 2008 73.3 4.5 7.2 12.6 38.7 12.7 16.7 NY Post General Election House Races Pre & Post Public Financing Comparisons (CT: NY, DE) 73.3 4.5 7.2 12.6 38.7 12.7 16.7 DE 1998 45.7 13.3 20 26.7 31.4 11.4 DE 2000 80.5 12.5 16.7 29.2 46.3 7.3 21.7 DE 2002 87.8 18.2 23.8 27.3 56.1 29.3 60 DE 2004 62.5 14.3 14.3 42.9 47.5 12.5 20 DE 2006 66.7 26.1 47.8 56.5 46.2 12.8 22.7 DE Pre 68.64 16.88 24.52 36.52 45.5 14.66 31.1 DE 2008 82.1 32.1 39.3 46.4 43.6 12.8 34.6 DE Post 82.1 32.1 39.3 46.4 43.6 12.8 34.6 Summary of Statistics PRE-PUBLIC CT 74.2 13.86 23.04 29.9 45.44 22.48 27.88 Comp. 75.9 12.1 17.5 25.8 45.4 22.9 24.1 POST-PUBLIC CT 73.5 15.1 28.3 34.0 45.7 22.5 70.4 Comp. 77.7 18.3 23.3 29.5 41.2 12.8 25.7 Conneticut Difference -0.7 1.2 *5.26 4.1 0.3 0.0 42.5 Comparison Difference *1.79 6.2 5.7 3.7 *-4.27 *-10.11 1.5 Difference-in-Difference -2.5-5.0-0.4 0.4 4.5 *10.13 *41 *significant at 5% level

STATE YEAR %CONTESTED %COMPETITITIVE (10) % COMPETITIVE (15) %COMPETITIVE20 %WOMAN % CONTESTED BY 3RD PARTY COMPETITIVE CONTRIBUTIONS CT 1998 86.1 6.5 19.4 19.4 30.6 13.9 29.2 CT 2000 75 7.4 11.1 18.5 30.6 19.4 19 CT 2002 83.3 16.7 20 23.3 47.2 27.8 23.1 CT 2004 88.9 20.7 20.7 27.6 44.4 50 34.8 CT 2006 86.1 20 23.3 33.3 36.1 16.7 37.5 CT Pre 83.88 14.26 18.9 24.42 37.78 25.56 28.72 CT 2008 88.9 20.7 24.1 37.9 36.1 19.4 61.5 CT Post 88.9 20.7 24.1 37.9 36.1 19.4 61.5 NY 1998 86.4 2 5.9 11.8 37.3 49.2 NY 2000 86.9 9.4 15.1 17 44.3 62.3 18.5 NY 2002 81 2 2 4 38.1 47.6 11.1 NY 2004 71 4.5 11.4 20.5 30.6 21 12.5 NY 2006 75.8 6.4 14.9 34 38.7 16.1 2.9 NY Pre 80.22 4.86 9.86 17.46 37.8 39.24 11.25 NY 2008 82.3 11.8 15.7 25.5 29 12.9 16.7 NY Post 82.3 11.8 15.7 25.5 29 12.9 16.7 DE 1998 100 33.3 33.3 50 55.6 44.4 DE 2000 100 14.3 71.4 71.4 45.5 0 25 DE 2002 100 9.1 18.2 45.5 63.6 27.3 70 DE 2004 100 20 40 60 63.6 18.2 0 DE 2006 100 16.7 16.7 33.3 53.8 23.1 33.3 DE Pre General Election Senate Races Pre & Post Public Financing Comparisons (CT: NY, DE) 100 18.68 35.92 52.04 56.42 22.6 32.075 DE 2008 100 33.3 83.3 83.3 11.6 9.1 33.3 DE Post 100 33.3 83.3 83.3 11.6 9.1 33.3 Summary of Statistics PRE-PUBLIC CT 83.88 14.26 18.9 24.42 37.78 25.56 28.72 Comp. 90.1 11.8 22.9 34.8 47.1 30.9 21.7 NY-only Comp 80.2 4.9 9.9 17.5 37.8 39.2 11.3 POST-PUBLIC CT 88.9 20.7 24.1 37.9 36.1 19.4 61.5 Comp. 91.2 22.6 49.5 54.4 20.3 11.0 25.0 NY-only Comp 82.3 11.8 15.7 25.5 29.0 12.9 16.7 Conneticut Difference 5.0 6.4 5.2 *13.48-1.7 *-6.16 32.8 Comparison Difference 1.0 *10.78 *26.61 *19.65 *-26.81 *-19.92 3.3 Comparison Difference (NY only) 2.1 *6.94 *5.84 *8.04 *-8.8 *-26.34 *5.45 Difference-in-Difference 4.0-4.3-21.4-6.2 25.1 13.8 29.4 Difference-in-Difference (NY only) 2.9-0.5-0.6 5.4 7.1 20.2 27.3 Note: Delaware Senate involves very small sample sizes (~6 races each year); as a result, the difference-in-difference estimate is also calculated for NY only *significant at 5% level

STATE YEAR %CONTESTED %COMPETITITIVE (10) % COMPETITIVE (15) %COMPETITIVE20 %WOMAN % CONTESTED BY 3RD PARTY COMPETITIVE CONTRIBUTIONS ME 1996 92.7 30.1 44.9 58.1 39.7 1.3 32.5 ME 1998 77.5 19.7 34.2 43.6 40.4 0.7 23.8 ME Pre 85.1 24.9 39.6 50.9 40.1 1.0 28.2 ME 2000 78.8 33.3 41.7 41.7 38.4 14.6 35.8 ME 2002 88.7 34.1 43.9 58.3 43.0 11.3 45.0 ME 2004 98.7 36.5 51.4 62.8 45.7 15.2 72.2 ME 2006 97.4 36.7 48.3 59.9 51.7 5.3 71.5 ME 2008 89.5 27.4 42.2 51.9 52.0 7.9 60.5 ME Post 90.6 33.6 45.5 54.9 46.2 10.9 57.0 NH 1996 70.1 29.0 40.1 53.1 44.9 5.7 NH 1998 60.2 31.3 45.4 60.0 43.6 2.0 NH Pre 65.2 30.2 42.8 60.0 44.3 3.9 NH 2000 59.2 33.2 50.7 64.9 0.3 NH 2002 63.3 32.0 45.6 58.9 0.1 NH 2004 81.5 30.7 44.1 58.1 0.1 NH 2006 78.3 48.4 61.0 75.2 49.0 1.3 NH 2008 79.3 39.1 56.8 69.7 50.8 2.0 NH Post 72.3 36.7 51.6 65.4 49.9 0.8 WA 1996 87.5 29.6 37.0 46.9 49.0 8.3 39.6 WA 1998 74.5 25.0 43.1 50.0 50.0 13.3 68.4 WA Pre General Election House Races Pre & Post Public Financing Comparisons (ME: NH, WA) 81.0 27.3 40.1 48.5 49.5 10.8 54.0 WA 2000 89.8 19.3 31.8 35.2 58.2 37.8 39.8 WA 2002 73.7 27.8 43.1 54.2 50.0 14.1 12.1 WA 2004 83.8 24.4 37.8 50.0 40.4 25.3 16.2 WA 2006 70.4 17.4 27.5 44.9 42.4 1.0 7.1 WA 2008 82.7 17.3 27.2 44.4 43.9 4.1 9.2 WA Post 80.1 21.2 33.5 45.7 47.0 16.5 16.9 Summary of Statistics PRE-PUBLIC Maine 85.1 24.9 39.6 50.9 40.1 1.0 28.2 Comp. 73.1 28.7 41.4 54.2 46.9 7.3 54.0 POST-PUBLIC Maine 90.6 33.6 45.5 54.9 46.2 10.9 57.0 Comp. 76.2 29.0 42.6 55.6 48.4 8.6 16.9 Maine Difference *5.52 *8.70 *5.95 *4.07 *6.11 *9.86 *28.85 Comparison Difference *3.13 *0.24 1.2 1.3 1.6 *1.29 *-37.12 Difference-in-Difference 2.4 8.5 4.8 2.7 4.5 *8.6 *66.0 *significant at 5% level

STATE YEAR %CONTESTED %COMPETITITIVE (10) % COMPETITIVE (15) %COMPETITIVE20 %WOMAN % CONTESTED BY 3RD PARTY COMPETITIVE CONTRIBUTIONS ME 1996 94.3 30.3 36.4 63.6 37.1 11.4 45.5 ME 1998 85.7 13.8 27.6 44.8 45.7 11.4 37.9 ME Pre 90.0 22.1 32.0 54.2 41.4 11.4 41.7 ME 2000 97.1 29.4 35.3 47.1 42.9 8.6 61.8 ME 2002 88.6 36.7 56.7 63.3 40.0 11.4 51.7 ME 2004 94.3 42.4 63.6 75.8 31.4 5.7 78.8 ME 2006 100.0 28.6 51.4 62.9 34.3 2.9 97.0 ME 2008 100.0 37.1 48.6 48.6 22.9 8.6 60.6 ME Post NH 1996 96.0 34.8 51.1 59.5 34.3 7.4 70.0 NH 1998 95.8 34.8 47.8 65.2 33.3 0.0 36.4 NH Pre 95.8 34.8 47.8 65.2 33.3 0.0 36.4 NH 2000 100.0 43.5 52.2 73.9 30.4 21.7 40.9 NH 2002 100.0 34.8 43.5 60.9 17.4 0.0 50.0 NH 2004 95.8 43.5 43.5 60.9 20.8 0.0 30.0 NH 2006 95.7 31.9 40.9 77.3 39.1 0.0 30.0 NH 2008 100.0 20.8 41.7 66.7 45.8 4.2 15.0 NH Post 98.3 34.9 44.4 67.9 30.7 5.2 33.2 WA 1996 80.8 52.4 66.7 71.4 46.2 3.9 52.6 WA 1998 70.8 37.5 50.0 62.5 45.8 0.0 26.7 WA Pre General Election Senate Races Pre & Post Public Financing Comparisons (ME: NH, WA) 75.8 45.0 58.4 67.0 46.0 1.9 39.7 WA 2000 80.8 26.1 39.1 43.5 46.2 38.5 26.3 WA 2002 58.3 50.0 50.0 50.0 37.5 16.7 46.2 WA 2004 89.3 37.5 45.8 58.3 42.9 14.5 39.1 WA 2006 79.2 38.9 38.9 50.0 37.5 0.0 18.8 WA 2008 73.1 21.1 36.8 63.2 38.5 0.0 22.2 WA Post 76.1 34.7 42.1 53.0 40.5 13.9 30.5 Summary of Statistics PRE-PUBLIC Maine 90.0 22.1 32.0 54.2 41.4 11.4 41.7 Comp. 85.8 39.9 53.1 66.1 39.7 1.0 38.0 POST-PUBLIC Maine 96.0 34.8 51.1 59.5 34.3 7.4 70.0 Comp. 87.2 34.8 43.2 60.5 35.6 9.6 31.9 Maine Difference *6 *12.79 *19.12 *5.34 *-7.1 *-3.978 *28.28 Comparison Difference *1.42 *-5.065-9.8-5.6-4.0 *8.60 *-6.18 Difference-in-Difference 4.6 17.9 *29.0 10.9-3.1-12.6 *34.5 *significant at 5% level

General Election House Races Pre & Post Public Financing Comparisons (Public: Comparison) % COMPETITIVE % CONTESTED BY COMPETITIVE STATE YEAR %CONTESTED %COMPETITITIVE (10) (15) %COMPETITIVE20 %WOMAN 3RD PARTY CONTRIBUTIONS Public 1996 80.6 25.3 36.8 57.5 49.3 10.9 32.7 Public 1998 68.2 14.9 28.2 44.7 49.8 4.9 22.7 Public Pre 74.4 20.1 32.5 51.1 49.55 7.9 27.7 Public 2000 70.6 19.4 22.4 54.1 51.2 21.9 40.8 Public 2002 79.1 30.7 40.2 61.1 50.2 13.1 51.2 Public 2004 87.2 29.4 42.6 60.3 51.7 22.4 66.8 Public 2006 86.3 31 40.6 58.7 55 13.1 68.2 Public 2008 92.2 33.3 47.2 55.9 57.3 13.7 57.3 Public Post 83.08 28.76 38.6 58.02 53.08 16.84 56.86 Comparison 1996 73.3 27.8 38.5 50.1 45.1 19.5 17.9 Comparison 1998 65 25.9 43 59.4 47.6 11.7 29.9 Comparison Pre 69.15 26.85 40.75 54.75 46.35 15.6 23.9 Comparison 2000 68.5 27.7 42.4 54.4 47.8 20.2 20.4 Comparison 2002 63.5 29.9 42.9 56.1 42.7 8.6 5.8 Comparison 2004 78.2 28 41.5 54.9 46.4 17.9 9.5 Comparison 2006 72.7 41.1 53 67.7 37.8 8.6 8 Comparison 2008 77.2 32.1 47.1 60.1 44.3 12.8 6.9 Comparison Post 72.02 31.76 45.38 58.64 43.8 13.62 10.12 Summary of Statistics PRE-PUBLIC Public 74.4 20.1 32.5 51.1 49.55 7.9 27.7 Comp. 69.2 26.9 40.8 54.8 46.4 15.6 23.9 POST-PUBLIC Public 83.1 28.8 38.6 58.0 53.1 16.8 56.9 Comp. 72.0 31.8 45.4 58.6 43.8 13.6 10.1 Public Difference *8.68 *8.66 *6.1 6.9 *3.53 8.9 *29.16 Comparison Difference 2.9 *4.91 *4.63 3.9-2.6 *-1.98 *-13.78 Difference-in-Difference 5.8 3.8 1.5 3.0 6.1 *10.9 *42.9 Analysis includes Arizona and Maine and their comparisons *significant at 5% level

General Election Senate Races Pre & Post Public Financing Comparisons (Public: Comparison) STATE YEAR %CONTESTED %COMPETITITIVE (10) % COMPETITIVE (15) %COMPETITIVE20 %WOMAN % CONTESTED BY 3RD PARTY COMPETITIVE CONTRIBUTIONS Public 1996 78.5 21.6 25.5 49 38.5 15.4 23.8 Public 1998 66.2 16.7 31 42.9 38.5 12.3 25.4 Public Pre 72.35 19.15 28.25 45.95 38.5 13.85 24.6 Public 2000 84.6 28.3 37.7 50.9 63.1 13.8 51.6 Public 2002 75.4 29.2 41.7 47.9 50.8 16.9 37.3 Public 2004 75.4 30.6 51 63.3 44.6 12.3 50.8 Public 2006 90.8 27.1 45.8 59.3 53.8 7.7 71.4 Public 2008 87.7 33.3 47.4 50.9 60 10.8 47.6 Public Post 82.78 29.7 44.72 54.46 54.46 12.3 51.74 Comparison 1996 74.4 34.5 51.7 56.9 47.4 2.6 21.1 Comparison 1998 82.8 35.9 48.7 61.5 48.3 0 31.9 Comparison Pre 78.6 35.2 50.2 59.2 47.85 1.3 26.5 Comparison 2000 76.5 23.8 38.8 52.5 41.6 21.6 16.1 Comparison 2002 80 37.5 45 55 60 8 47.1 Comparison 2004 72.1 31.1 36.5 50 42.3 10.6 21.1 Comparison 2006 82.8 34 40.4 61.7 44.8 6.9 19.1 Comparison 2008 70.2 27.6 39.7 58.6 53 8.3 12.5 Comparison Post 76.32 30.8 40.08 55.56 48.34 11.08 23.18 Summary of Statistics PRE-PUBLIC Public 72.35 19.15 28.25 45.95 38.5 13.85 24.6 Comp. 78.6 35.2 50.2 59.2 47.9 1.3 26.5 POST-PUBLIC Public 82.8 29.7 44.7 54.5 54.5 12.3 51.7 Comp. 76.3 30.8 40.1 55.6 48.3 11.1 23.2 Public Difference *10.43 *10.55 *16.47 8.5 *15.96-1.6 *27.14 Comparison Difference -2.3 *-4.4 *-10.12-3.6 0.5 *9.78 *-3.32 Difference-in-Difference 12.7 15.0 *26.6 12.2 15.5-11.3 *30.5 Analysis includes Arizona and Maine and their comparisons *significant at 5% level

Thank you for evaluating AnyBizSoft PDF Merger! To remove this page, please register your program! Go to Purchase Now>> AnyBizSoft PDF Merger Merge multiple PDF files into one Select page range of PDF to merge Select specific page(s) to merge Extract page(s) from different PDF files and merge into one