Giving to Governors: State Contribution Limits and the Incumbency Advantage

Consistent with elections throughout the United States, the majority of contributions in gubernatorial campaigns come from individuals. Understanding the incentives for why individual donors give money in varying political environments is essential for understanding the influence of political donors. Using contribution data from 2009 through 2015, this paper considers whether state campaign finance reform has had unintended effects for how incumbent and non-incumbent candidates raise money. This study demonstrates that individual donations vary in theoretically driven ways. Few factors, however, affect donor generosity more than the relationship between a candidate’s in-cumbency status and the severity of state contribution laws. Consistent with Jacobson’s early concerns, restrictive contribution laws appear to strengthen the position of incumbents and place non-incumbent candidates at a disadvantage. Where states apply less restrictive limits, however, the advantage held by incumbents among donors largely disappears.


Introduction
Following the Watergate scandal of the 1970s, a movement to limit contributions started with the passage of the Federal Election Campaign Act Amendments of 1974 and continued later with the Bipartisan Campaign Reform Act of 2002.States immediately followed the post-Watergate reforms by placing widespread limitations on the ways that citizens, parties, and groups participate in elections.Today, a large majority of American states have adopted limits on contributions to political campaigns.With these campaign reforms, the most active contributors face significant restrictions on how much money they may donate to state campaigns.According to the National Conference of State Legislatures (NCSL), individual contributors in 2015 faced restrictions on donations to statewide campaigns in 38 states. 1 Limits during general elections were also applied to contributions from state political parties (29 states), political action committees (37 states), corporations (44 states), and labor unions (42 states).Of those groups, corporations and trade unions were particularly affected with prohibited contributions in 21 and 17 states, respectively.
The stated goal of policymakers when passing campaign finance reform is to limit corruption, address political inequality, and reduce overall spending in elections by placing limits on the size of donations (Eom & Gross, 2006; Ramsden, 2002).However, according to La Raja (2008), campaign finance reform leveled the playing field among partisan interests rather than removing political corruption.Following the passage of contribution limits at the national level, displeasure formed among some scholars and pundits due to concern that contribution limits helped incumbents at the expense of challengers (Cox & Munger, 1989; Jacobson, 1976; Meirowitz, 2008; Snyder, 1993).Forcefully, Jacobson  (1976) reasoned that campaign finance reform would create "severe electoral handicaps" (Jacobson 1976, p. 2) for non-incumbents due to an incumbent's record of constituency service, travel allotments, and access to the media. 2For many challengers, according to Jacobson, campaign finance reform limits access to money necessary for educating citizens about their candidacy, thereby minimizing the effectiveness of their campaign.
At the state level, the most salient aspect of campaign finance reform has also been the creation of contribution limits as demonstrated by the widespread adoption of such limits, as discussed above, and the growth in scholarly attention to this topic (Bardwell, 2003; Butcher & Milyo, 2020; Eom & Gross, 2006;  Flavin, 2015; Hamm & Hogan, 2008; Stratmann, 2010; Stratmann & Aparicio-Castillo, 2006).Where states apply contribution limits on donations by individuals and other groups, elections become both less competitive and less contested with increased margins of victory and fewer candidates seeking election (Lott, 2006).Moreover, where contribution limits are more severe, incumbents spend larger sums of money relative to their opposition (Gross, Goidel, &  Shields, 2002), increasing the likelihood of an incumbent candidate's victory (La  Raja & Schaffner, 2015; Meirowitz, 2008).Contributions, where constrained, are also not distributed equally among incumbents and non-incumbents (Box-Steffensmeier & Dow, 1992).While individual contributors prefer different ideological outcomes than political action committees (PACs) or political parties (Barber, 2016; La Raja & Schaffner, 2015), existing research suggests state campaign finance reforms have had a number of important effects on the outcomes of elections and the types of officeholders serving in government.
In this paper, we evaluate the specific benefits and disadvantages of incumbency where states apply more or less restrictive campaign finance laws.Ninety-three gubernatorial elections (52 with incumbents and 41 open seat contests) from 2009 to 2015 are examined, including each party-nominated Republican and Democratic candidate and a small group of viable third-party candidates. 3alysis of the total amount donated to the general election campaigns for governor demonstrates that contributions are related to a variety of factors, but none more essential than state contribution limits and the presence of incumbent candidates on ballots.Consistent with expectations regarding contributors, the results demonstrate that they are well-informed electoral participants who give money in largely predictable ways.

State Limits on Individual Donations
From 2009 through 2015, states applied varying approaches to contributions to statewide campaigns, including races for governor.Consistent with national and statewide efforts to harness the influence of money in government, a large majority of states applied restrictions on how much an individual could give.Elsewhere, several states applied few or no restrictions on individual donations.These campaign finance policies suggest distinct state views about how money influences campaigns and the connection between officeholders and donors. 4ing data from the NCSL, Figure 1 shows state variation within the contiguous forty-eight states relating to limits on individual donations to statewide general election campaigns. 5The NCSL groups state policies by two-year cycles.
Throughout the years of this study, thirty-eight states utilized limits on contributions and thirteen states did not.Moreover, one state (Illinois in 2011) altered its campaign finance laws from unlimited individual contributions to restricted amounts, while no state shifted in the opposite direction.
Figure 2 delves further into the characteristics of state campaign finance laws by presenting the maximum contribution where states impose limits on donations 3 While third-party candidates are active in most gubernatorial elections and occasionally successful (e.g., Lincoln Chafee's 2010 campaign for Rhode Island governor), most receive minimal attention from contributors.Third-party candidates are included in the analysis only where they earned 5 percent or more of the general election vote.Inclusion of third-party candidates has no meaningful effect on the regression analyses.Results for the models without third-party candidates are available upon request.

4
The National Conference of State Legislatures' report, "Campaign Finance Regulation: State Comparisons" provides a detailed overview of the different approaches states take to campaign finance policies and which types of regulations are imposed by state.The report can be found at https://www.ncsl.org/elections-and-campaigns/campaign-finance-regulation-state-comparisons.Beijing Law Review  by individuals. 6The average limit throughout the years of this study was $4865.The average limit, however, increased from $4575 in 2009 to $5355 in 2015.In Figure 2, states are assigned colors by the maximum donation with yellow assigned to the largest values and blue to the smallest.States without limits are gray.New York allowed the largest sum for each year of this study.In 2009 and 2010, New York allowed individuals to give $37,800 to statewide candidates during the general election and that value increased to $41,100 by 2015.California likewise allowed large maximum contributions during the general election season.Their contribution limits increased from $25,900 in 2009 to $28,200 in 6 States differ in how they limit contributions.Some, like New York and North Carolina, use separate restrictions for primary and general election campaigns.Others, like Michigan and Wisconsin, apply limits for the entire calendar year.Figure 2 reports the state limit where general elections are restricted.Alternatively, where states impose a yearly limit, that value is reported.2 demonstrates further variation among the states and the modest limits imposed by most states, including Massachusetts where individuals from 2009-2014 could give no more than $500.For individuals participating in gubernatorial elections as donors, state contribution limits provide an important starting point for understanding their decisions to give.

Existing Tests for Motivations of Contributors
According to the National Institute on Money in Politics (NIMP), individual contributors were the largest source of contributions to gubernatorial campaigns from 2009-2015 with 65 percent of total donations from individuals. 7While voting is the essential form of electoral participation by citizens due to the impact of voting on who wins and how elected representatives perform once in office (Engstrom, 2012; Rosenstone & Hansen, 1993), decisions by individuals to contribute represent a primary form of political participation.Fortunately, contributions to presidential, congressional, and state level campaigns have received important scholarly attention (Barber, 2016; Barber, Canes-Wrone, & Thrower,   2017; Ensley, 2009; Francia et al., 2003; Rosenstone & Hansen, 1993), with decisions to contribute shown to be influenced by strategic donor goals and the context of campaigns and elections.
Studies of contributors find individuals give to political campaigns for a variety of reasons, including the desire for material benefits, ideological goals, and personal connections to a candidate.While individual donors prefer giving to ideologically compatible candidates (La Raja & Schaffner, 2015), most decisions to contribute are multifaceted and connected to various motives (Francia et al.,   2003).One consistent finding is the variety of incentives, solidary, purposive, and material, that compel individuals to donate (Barber, 2016; Barber,   Canes-Wrone, & Thrower, 2017; Francia et al., 2003).
Among contributors, two groups are prominent, those motivated by material gains ("investors") and others motivated by purposive or ideological goals ("ideologues"). 8For investors seeking material benefits and narrow policy outcomes, donations are given in part based on a candidate's likelihood of victory and the possibility that candidates will provide material benefits once in office (Ensley, 2009; Francia et al., 2003).With decisions contingent on a candidate's probability of winning, this is one reason incumbents outraise non-incumbents (Krasno, Green, & Cowden, 1994).Ideologues who are motivated by purposive goals, however, evaluate the policy positions of candidates and give according to their ideological alignment with a candidate (Barber, 2016; Gimpel, Lee, &   7   Like gubernatorial races, Barber (2016) finds that individuals are among the two largest sources of contributions to legislative campaigns (with PACs) and Barber, Canes-Wrone, and Thrower (2019)  find a similar pattern for presidential elections.Pearson-Merkowitz, 2008; Morton & Cameron, 1992; Snyder, 1993).As such, donors who are motivated by ideological goals show a tendency to support extreme candidates who are to the right or left on the ideological spectrum (Ensley,  2009; Francia et al., 2003; La Raja & Schaffner, 2015).Existing literature (Aldrich, 1995) reasons that candidates are aware of the purposive goals of contributors and take extreme policy stances to attract contributors.
While categories of contributors are not separately evaluated in this paper, theories of individual contributions are tested using high profile gubernatorial elections to understand donation amounts by individuals.As contributors make decisions about how much money to invest, they are expected to pay attention to the incumbency status, experience, and gender of candidates, as well as the closeness of elections and importance of the office.

Theory: When Should Contribution Limits Matter?
While theories of individual donors' reasons for giving separate along the investor and ideologue distinction, there are well-documented connections between the attributes of candidates and elections and the choices of citizens in elections.In studies of fundraising and voter turnout, officeholders consistently outperform challengers in congressional (Krasno, Green, & Cowden, 1994), state legislative (Engstrom & Monroe, 2006), judicial (Bonneau, 2007), and gubernatorial elections (Ansolabehere & Snyder, 2002; Brown & Jacobson, 2008).Petrocik and  Desposato (2004) frame this advantage as one where the benefits of office and name recognition diminish short-term political tides.9With access to a wider network of contributors (Crespin & Deitz, 2010) and the benefits of name recognition and status as a proven candidate (Ensley, 2009; Jacobson, 2009), incumbents consistently outraise challengers (Krasno, Green, & Cowden, 1994; La  Raja & Schaffner, 2015). 10e anticipate that individual donors in gubernatorial elections will treat incumbent and non-incumbent candidates differently.While donor agreement with candidates likely affects decisions to contribute, we expect that individual donors will give larger donations to incumbent governors due to improved familiarity with an incumbent and the governor's success in prior elections.
Hypothesis 1 (H1): Individual contributors will give larger donations to incumbent governors; whereas, contributors will give smaller donations to non-incumbent candidates.
Perhaps most relevant for understanding political donations, most states have adopted laws that restrict individual contributions to campaigns for governor.The structure of these limits varies with contribution limits in some states applied separately to primary and general elections.Elsewhere, individual dona-tions are limited for the entire calendar year.Throughout the years of this analysis, contribution limits ranged from $500 to $41,100 (New York in 2014 and  2015).In contrast, thirteen states allowed unlimited donations by individuals.Summary evidence reported in this article (see Figure 5) suggests that individual contributors do indeed find themselves constrained by limits on individual giving.The number of contributors donating the maximum amount exceeded more than one-fifth of voters and in some states about half of donors were impacted by state limits.While several states modified their campaign finance laws between 2009 and 2015, differences in campaign finance laws reflect actual limitations on how individuals may use their money in campaigns for governor.Consistent with the intent of contribution limits, we expect that more restrictive limits on individual donations will reduce the size of individual donations by placing a cap on an individual's ability to give.
Hypothesis 2 (H2): Individual contributors will give smaller donations where states apply more restrictive limits on individual donations; whereas, contributors will give larger donations where states do not use or apply less restrictive limits.
Despite their popularity throughout the states, contribution limits remain an area of disagreement among scholars with lingering questions about whether campaign contributions create an advantage for incumbents (Aranson & Hinich,  1979; Box-Steffensmeier & Dow, 1992; Cox & Munger, 1989; Meirowitz, 2008;  Snyder, 1993).In their study of California legislative elections, Box-Steffensmeier  and Dow (1992) found the removal of contribution constraints allowed investors to target and spread their donations throughout the legislature and across several positions of authority (i.e., those in leadership roles or serving in prominent committees).
By comparing regulated and less regulated campaign finance law environments, we expect contribution limits will differently affect incumbent and non-incumbent candidates in gubernatorial elections.With increased visibility, a record of performance, and improved certainty about a contributor's investment, individual donors in regulated states are expected to give larger, limit-approaching sums to incumbent governors.Non-incumbents in states with contribution limits will receive smaller donations relative to those directed to incumbents.In less regulated states, where states allow contributors to give large or unlimited sums to their preferred candidate, donors will target their contributions to both incumbents and non-incumbents.The result will be individual contributions to incumbents that are smaller than more limited states and closer to the amount received by non-incumbent candidates.
Hypothesis 3a (H3a): Individual contributors will give larger donations to incumbent governors, rather than non-incumbents, where states apply restrictive limits on individual donations.
Hypothesis 3b (H3b): Individual contributors will give more equitable amounts to incumbent and non-incumbent candidates where states apply larger limits or no restrictions on individual donations.

Data and Methodology
To evaluate contributions as a form of political participation, this study systematically examines ninety-three gubernatorial elections from 2009 to 2015. 11Data on contributions was collected by NIMP's Follow the Money data archive (https://www.followthemoney.org/),which includes all contributions by individuals, businesses, interest groups, political parties, and candidates throughout the period of this analysis, though the focus here is on individual contributions.Only direct donations to general election campaigns for governor are included in this study.
Through the information collected by NIMP, studies of state level contributions now benefit from a vast store of data relating to state and gubernatorial campaigns.Fortunately, recent efforts have used this contribution data to explore varied and important aspects of state level campaigns.In particular, Bonica's (2014) effort to measure the ideology of candidates and contributors demonstrates how significant advancements using campaign finance data can be made.Using records of contributors who donate across levels of government and institutions, Bonica created a common set of ideological measures for state governors, as well as state legislators, state supreme court justices, members of Congress, and others.Additionally, state campaign finance data has been used to explore various features of state elections, including findings that demonstrate externally financed candidates perform better than self-financed candidates in gubernatorial elections (Brown, 2013) and the unique effects of campaign finance laws on sources of funding for state legislative candidates (La Raja &   Schaffner, 2015).This research follows by using campaign finance data to explore the incentives for individual donors to give larger or smaller amounts to gubernatorial campaigns.

Dependent Variable
The primary dependent variable, Contribution Amount, is explored using the natural log of the total amount donated by an individual to a gubernatorial campaign during a general election. 12The natural log is used to achieve a normal distribution, which is not present when using non-transformed values.Further, to make estimates comparable throughout the seven years of this study, contributions were adjusted into constant dollars (2015) using the Consumer Price Index. 13This study explores contributions at the contributor-campaign level of 11 The year 2009 represents the beginning of this investigation since it is the first year that contribution limits for all fifty states were available from the NCSL.There were ninety-six elections between 2009-2015, but three elections (Connecticut in 2014 and South Dakota in 2010 and 2014) are not included in the data due to a lack of contribution observations. 12Self-contributions by candidates are not included in the dataset.Incentives for candidate contributions are expected to be quite different for candidates than non-candidates.Descriptive comparisons between candidate and non-candidate contributions indicate that contributions from candidates (mean = $38,031) are much larger than non-candidates (mean = $493).

13
Converting contributions to constant dollars is standard in the literature where using more than one year of data (see Engstrom & Ewell, 2010).
analysis to better understand how the cumulative amount donated to a general election campaign for governor relates to the incumbency status of candidates and the severity of state campaign finance laws, as well as control variables relating to elections, state environments, and the characteristics of contributors.As a result of the level of analysis, specific contributors are found throughout the data where contributors are active in multiple years and donations are given to different candidates.Only one contributor observation, however, is matched with a gubernatorial campaign.
This study explores only the small group of citizens that contribute to gubernatorial campaigns. 14While the choice to contribute rather than not contribute is a decision of interest, that decision is beyond the scope of this paper.Rather, using the data available from NIMP, we hope to understand the degree of generosity by donors active in elections for governor.To explore contributions, donors that gave $1 or more to a gubernatorial campaign are explored.

Key Independent Variables
One of the essential independent variables is whether a candidate is an incumbent governor or a non-officeholder.To measure the effect of incumbency, candidates are coded 1 where an incumbent governor seeks reelection and 0 where a candidate is not an incumbent.Prior research has validated the important role of incumbency, especially as incumbency relates to support from voters and generosity by contributors.
A second primary independent variable is the degree to which states limit contributions by individuals.To understand the impact of state campaign finance laws and restrictions on giving, we focus on limits on individual contributions.Contribution limits are explored using a categorical variable (ordinal contribution limit) that ranges from 0 -4.The base category (0) represents states from 2009-2015 that allowed the smallest restricted contributions (those between the 1 st and 25 th percentiles), states are coded 1 where between the 26 th and 50 th percentiles, states are coded 2 where between the 51 st and 75 th percentiles, and states are coded 3 where allowed contributions fall between the 76 th percentile and the maximum restricted contribution.States without contribution limits on individual donations receive a value of 4. To investigate the interactive relationship between contribution limits and contribution amounts where incumbent governors seek reelection, a multiplicative term (Contribution Limit*Incumbent) is added to the empirical model of contribution amount.The in- teraction terms tests for whether incumbent governors outraise non-incumbents among individual contributors where limits are more severe and whether that advantage decreases in less regulated states where larger donations are allowed.

Control Variables
In the empirical models, we account for a number of additional factors identified 14 Only about 10 percent of the electorate in the United States makes contributions (Rosenstone &  Hansen, 1993).
in the literature to affect political donations.These factors connect to a variety of features including attributes of candidates, elections, and state environments.To better understand the decisions of donors, we control for several traits of individual contributors.Among candidate features, we include a measure that ranks each candidate's political experience using Squire's (1992) profile index.Squire's index was designed for gubernatorial elections and considers the percentage of a state's electorate covered by a candidate's prior office, as well as the significance of that office.Scores for Squire's index range from 0 (no political office in the prior two years) to 600 (ex-governors or ex-U.S. senators). 15We also include a measure of a candidate's gender with female candidates coded 1 and male candidates coded 0. Regarding the partisan identification of a candidate, Republican candidates are coded 1 and Democratic candidates are coded 0.
Of the many election-specific factors that influence contributions, we control for open seat races and competitive state elections.For open seat races, a dichotomous measure is utilized with open seat races coded 1 and incumbent-challenger races coded 0. While there is debate about the appropriate threshold for competitive state elections (see Weber, Tucker, & Brace, 1991), we use the 55 percent marker advocated by Ray and Havick (1981) with elections won by 55 percent or less coded 1 and those above that threshold coded 0. To consider general state environments, a control variable for divided government is included in the models where states under the condition of divided government are coded 1 and unified control of state government is coded 0. 16 Additionally, the formal powers of governors are included as an institutional incentive for individuals to donate.We use the governor's formal powers index score by Ferguson (2012) to capture the degree of executive power in state government.Ferguson's (2012) measure of governor's institutional powers identifies the following: whether a state's executive branch is divided or unified, the tenure potential of office, appointment power, budget power, veto power, and control over the governor's political party.
To better understand the conditions for individual donations, we consider the effects of a contributor's gender, state of residence, and local income. 17To control for the relationship between gender and the contribution amount, female contributors are coded 1 and male contributors 0. In-state contributors are coded 1 and contributors from outside of the state are coded 0. Income tax data from the Internal Revenue Service's Individual Income Tax Zip Code Data is used as a proxy for a contributor's wealth. 1815 To develop scores for the candidate profile measure, data was collected from newspaper articles, official campaign websites, and biographies posted on political office websites.With very few exceptions, this information was available.Data were then coded using Squire typology.We follow Squire's coding scheme for current governors and assign a value of 600. 16In alternative models, we tested for a relationship between state population and individual donations.State populations had no observable impact on contribution amounts.17 The impact of state income and state population on individual contributions were also considered.There was no meaningful effect for either variable on the size of donations.
18 That data was transformed using the natural log of the adjusted gross income by zip code and matched to the zip code of a contributor using NIMP's Follow the Money database.
A year trend variable is also included within each model.While there is no expectation for a year trend, we control for the possibility that contributors gave increasing or decreasing amounts from 2009 through 2015. 19Table 1 describes the variables used in the empirical models of direct individual contributions.We expect the factors described in Table 1 and the sections that follow will influence decisions by individuals to be more or less generous.

Modeling Approach
With the dependent variable (Contribution Amount) a continuous measure, an ordinary least squares (OLS) design is used to estimate Models 1 and 2 of Table 2. Due to the structure of the data, we use robust clustered standard errors (RCSE) grouped by elections, which are robust to expectations about within-group correlation. 20The data is clustered by election since it is likely that the actions of contributors are correlated within elections, but independent from alternative elections. 21n Model 3 of Table 2, the empirical model of contribution amount is again estimated, yet here the bootstrap method is utilized.Due to the number of observations (1,150,552), there is a distinct possibility that Type I error may affect the interpretation of the results since RCSE are biased downward where empirical models include a large number of observations (see Harden, 2011).Type I error occurs where the agreed upon null hypothesis is rejected although it is actually true.Following the recommendation of Harden (2011), bootstrap clustered standard errors (BCSE) are utilized in Model 3 since they produce estimates that are less likely to be biased downward.As another benefit of the bootstrapping procedure, the results are derived from characteristics of the sampling distribution of estimators from 100 replications to better recognize the robustness of the results reported in Model 2. Like Models 1 and 2, contributions are observed at the individual level and the observational units are clustered by election.

Descriptive Patterns
To understand patterns relating to contributions to gubernatorial campaigns, Figures 3-5 illustrate state-by-state tendencies for individual contributions to general election campaigns for governor from 2009-2015.While the regression analyses reported in Models 1-3 evaluate the total contribution by an individual contributor, the mean and median values are presented in Figure 3 and Figure 4. Non-adjusted and non-transformed values are reported for ease of interpretation in Figure 3 and Figure 4. Figure 5 presents the percentage of donors in each gubernatorial election that reached the maximum individual contribution.With scholars and the popular media noting the unique political environments of the states, it should come as no surprise the wide variation observed among contributions.Donors were highly active in many states, giving large sums to gubernatorial campaigns, yet in other states individuals were significantly less generous.
The mean individual contribution for all states was $493 and the median contribution was $100. Figure 3 shows that the average contribution to gubernatorial campaigns ranged from $172 in Montana to $3111 in Illinois.Moreover, mean donor contributions in thirty-nine states fell below the $1000 threshold.Eleven additional states had average contributions that exceeded $1000 and five states -Illinois, Nebraska, Mississippi, California, and Pennsylvania -saw contributors give more than $1500, on average.For the states with highest mean sums, several, including Illinois, California, and New York, applied either weak or no restrictions on individual giving.With the median contributions displayed in Figure 4, a similar story unfolds with the smallest values found in Louisiana and Wisconsin, where the most typical contributions were both $50.In Mississippi and Kentucky, however, the median contributions for that same period were $900 and $750, respectively.Interestingly, neighboring Louisiana and Mississippi are two of just three states that hold their gubernatorial elections in the    year prior to a presidential election, yet their patterns of giving differ greatly.While a variety of factors certainly affect incentives to give larger or smaller sums, observed state differences suggest that political environments play an important role in creating incentives and disincentives for donations.Figure 5 presents the percentage of donors in each gubernatorial election that reached the maximum contribution in regulated campaign finance environments.As with Figure 3 and Figure 4, wide variation is observed in terms of the behavior of individual donors.Throughout the gubernatorial elections from 2009-2015, 10.2 percent of donors gave their maximum contribution.Several gubernatorial elections, however, saw a large majority of contributors stay away from maximum limits, including Arizona in 2010, New Hampshire in 2012, New York in 2014, and Ohio in 2010, where less than 1 percent of individual donors gave the maximum amount.Other states experienced different donor behavior.Ten elections had more than 20 percent of their donors give the maximum donation and two states, West Virginia in 2012 and Kentucky in 2015, provide examples of very engaged donors with maximum contributions reaching 49 and 50 percent, respectively.With both West Virginia and Kentucky limiting individual contributions to $1000 for the general election (a value equal or larger than the contribution limits in the most active donor states), there is a willingness among donors to contribute up to the state limit.

Model of Contribution Amount
To explore the generosity of contributors in gubernatorial elections from 2009 through 2015, two models, one a baseline and another an interactive model, are used in Table 2 to explore the determinants of contribution amounts.First, Model 1 of Table 2 specifies the baseline model that regresses the amount of a contribution on each of the explanatory variables of interest.Model 2 of Table 2 then estimates the effects of the explanatory variables from the baseline model and the additional interaction variable (Contribution Limit*Incumbent).At first glance, Models 1 and 2 appear very similar in terms of the substantive impact of the explanatory variables, yet Model 2 provides a more thorough understanding of donor generosity due to the important connection between incumbency and campaign finance laws.
Turning first to the baseline model, it performs well with seven statistically significant independent variables.Before discussing the key variables of interest, it is important to note that the basic features of candidates, elections, and state context explain 17.6 percent of the variance in contributions.The results from Model 1 fit well with previous state-based studies of contributions, as well as the theoretical predictions of this study.Of the variables connected to candidates and elections, the size of individual donations increased where elections were open seat contests.Where exploring the impact of state environments, divided government had a positive impact on political donations by individuals.Contributors also gave larger donations when they were male, lived in the state of an election, and resided in localities with higher average incomes.Additionally, the trend variable is statistically significant with smaller individual donations as the study advanced from 2009 to 2015.
With the primary variables of interest for Model 1, the estimate for incumbency status is both substantively and positively related to contribution amount, with larger donations where incumbent candidates vied for reelection.All things being equal, contributors gave 60 percent larger donations to incumbents than to non-incumbents.State laws that limit individual contributions, however, had no observable effect on individual donations in the baseline model.
The findings reported in Model 2, however, add an important dimension to our overall understanding of contributions with the financial advantages of incumbents more easily observed where states apply restrictive contribution limits.
In terms of the performance of the model, nine explanatory variables are statistically significant and 18 percent of the variation in individual donation amounts is explained.As with Model 1, predictors of donor generosity from previous studies perform as expected, including those variables tied to the features of elections, states, and contributors.Candidates affiliated with the Republican Party garnered larger individual contribution amounts.Open elections produced a positive and discernable effect on contribution amounts, as did divided government, showing individuals donate larger sums in these contexts.Contributor attributes also played an important role in determining contribution amounts, with in-state contributors donating substantially more than their out of state counterparts.Male contributors donated more relative to female contributors and living in a locality with greater wealth led to an increase in the amount contributed.As with the baseline model, the size of individual contributions de-creased throughout the years of this study.
The conditional effect of campaign finance laws where candidates are an incumbent or non-incumbent emerges as a dominant factor for explaining contribution activity in gubernatorial races.While the coefficient for Contribution Limit was not statistically important in Model 1, the Contribution Limit*Incumbent variable is directionally negative and has a strong discernible effect on donations.With interactions between binary variables (Incumbent) and continuous variables (Contribution Limit) difficult to interpret solely by the coefficient, Figure 6 displays the linear prediction of the natural log of contribution size where contribution limits range from the smallest and most restricted contribution to states with unlimited contributions.The plot indicates that contributions to incumbent candidates decreased by more than half from 5.44 ($229.85 in real value) to 4.71 ($111.40)where campaign finance laws became less restrictive and open to larger individual donations.The impact of contribution limits is quite different for non-incumbents, with contributions increasing slightly from 4.52 ($91.91) to 4.54 ($93.71)where comparing states with very restrictive limits to states with unlimited individual donations.While the degree that donations decreased for incumbent candidates is surprising, the similar donation amounts for incumbent and non-incumbent candidates in less regulated states fits the expectations of this study.The estimate for Incumbent, a constituent variable for the multiplicative term, is also statistically significant with a strong positive effect on donation amounts.To fully understand the impact of constituent terms, however, the effect must be considered in tandem with the interaction term. 22mpaign finance laws that limit contributions benefit or punish candidates quite differently depending on whether they hold office.From these results, there appears to be a distinct advantage for incumbents in states with restrictive limits, thereby creating a competitive imbalance between officeholders and non-officeholders.Where campaign finance laws allow larger contributions, the difference between donations to incumbent and non-incumbents is less meaningful, allowing incumbents and non-incumbents to perform almost equally among individual contributors.
Model 3 of Table 2 utilizes the bootstrap method of statistical inference to test the robustness of the empirical results reported in Model 2. Due to the large number of observations (N = 1,150,552) in Models 1 and 2, it is necessary to consider whether the results suffer from Type I error.With the bootstrapping procedure, the results for the amount contributed to gubernatorial candidates are again clustered by election.The results from Model 3 indicate no meaningful difference between the RCSE and BCSE specifications among the substantively important explanatory variables.With the primary variable of interest, Contribution Limit*Incumbent, there remains a statistically discernible and substantively 22 Brambor, Clark, and Golder (2006) reason "[s]cholars should refrain from interpreting the constitutive elements of interaction terms as unconditional or average effects, they are not" (p.71).negative relationship with the amount donated.Given the performance of the results using the bootstrapping procedure, there is confidence that the estimates reported in Model 2 do not suffer from Type I error.

Model of Contributor Quantity
A lingering question from the findings presented in Table 2 is the non-relationship between individual donations and the control variable for race competitiveness (i.e., elections won by less than 55 percent).While close elections encourage electoral participation (Cox & Munger, 1989; Engstrom, 2012), no relationship between contribution amount and competitive races for governor is found in Models 1-3.Contributors show no preference to give larger sums where winning candidates received less than 55 percent of the vote, the threshold used here for marginal elections.23Moreover, while not statistically meaningful, the direction of the coefficient is negative, suggesting marginal elections deter donor generosity.
To better understand this finding, an analysis of the number of contributors that support a gubernatorial campaign is performed in Table 3.While donors may not be more generous where elections are close, it is possible that marginal races encourage a larger number of contributors to be engaged.As such, Table 3 presents an ordinary least squares estimation with the dependent variable (contributor quantity) the total number of contributors to a gubernatorial general election campaign. 24Standard errors are clustered in Table 3 by candidate since gubernatorial campaigns should differently and independently solicit contributors to participate.The number of contributors varies greatly throughout the states with major party and significant third-party candidates receiving support from an average of 6666 contributors and a median value of 2530 contributors. 25he model of contributor quantity uses the same state and election-level variables as those presented in Models 1 -3, but also includes a control variable for each state's population. 26State population is added to this model since it is possible that larger state populations will correlate with a more active population of individual donors.The relationship between competitive races and contributor engagement is expected to be positive, or, where elections are closely contested, a larger number of contributors are expected to be involved.Moreover, the number of donors is expected to increase where a candidate is an incumbent, has political experience, and is a Republican.Open seat races, states with divided government, and offices with pronounced institutional authority should encourage a larger quantity of contributors.Unlike the model of contribution amount, stricter limits on contributions are expected to have a positive relationship with the number of contributors.Campaigns are expected to offset restrictive limits on contributions by soliciting a wider pool of donors, including those that give smaller amounts.
Most evident from Table 3 is the failure of many explanatory variables to reach a conventional level of statistical significance.Notably, while incumbents received larger individual contribution amounts compared to non-incumbents, as demonstrated in Table 2, they did not receive a larger quantity of contributions.While the direction of the coefficient in Table 3 is positive, it is not statistically significant.Three variables however, Competitive Race, Divided Government, and Formal Powers of Governors, explain engagement by contributors in gubernatorial elections.Where elections are closely contested and the winner receives less than 55 percent of the vote, the number of contributors increases by 6652, all else being equal.Whether giving smaller or larger amounts, contributors, like voters, react to increased electoral competition.Somewhat counterintuitively, divided state government was correlated with 2469 fewer contributors than where one party controlled a state's legislature and governor.Lastly, governors with the most formal powers added about 5923 contributors compared to states with the fewest formal powers.Though the closeness of elections had no impact on the size of contributions, the number of contributors does increase where elections are marginally decided.

Conclusion
In this article, many of the important expectations about individual participation in elections have been applied to political contributions, both in terms of the size of donations and the number of individuals supporting gubernatorial campaigns.Consistent with what the scholarly literature describes for both voting and contributions, the findings indicate that donations by contributors are connected to campaign and election signals, as well as laws that structure how individual donors participate in elections.Moreover, while electoral competition had no discernible impact on the average contribution, marginal elections encouraged donors to participate, suggesting a nuanced effect for close elections.
The results indicate that several characteristics of candidates and elections encourage donors to give larger or smaller sums, yet confirmation that a candidate's status as an incumbent and the laws that constrain donors emerge as essential considerations for understanding donor decisions.Ostensibly, incumbents experience an advantage in many states due to their visibility and expanded media coverage, but the story is incomplete without also considering the impact of campaign finance laws.States with more restrictive limits intensify the support of donors, leading to larger donations for incumbents than non-incumbents.In less regulated state environments, however, donation amounts are approximately equal among incumbents and non-incumbents.In less regulated states, the unrestrained ability of donors to contribute their preferred amount diminishes the advantage of incumbents.But why should donations to incumbents decrease to such a degree?It is likely that the constraints that limit donors in more restricted states also increase the appearance of an incumbent's electoral strength.For both donors seeking access and those in tune with an incumbent's policy agenda, their support is more generous for incumbents, which is due in part to that perception.Oppositely, where donors may give larger sums, they may level the playing field.The result, it appears, leads to roughly equal donations by supporters of incumbents and non-incumbents.
The result for the conditional relationship between contribution limits and donations to candidates by whether a candidate is an incumbent or non-incumbent, falls in line with general expectations about campaign finance laws.Treatments of contribution limits going back to Jacobson's (1976) influential study reason that incumbents are assisted by campaign finance reform.This analysis of individual contributors demonstrates that in terms of incumbent support, much depends on how restrictive states are in terms of their contribution laws.With more restrictive limits, non-incumbents operate at a financial disadvantage.
The degree of sophistication among contributors should be surprising to few.As La Raja and Schaffner (2015), Francia et al. (2003), and others have described, contributors are citizens who are fully aware of their goals, whether those interests are ideological or material in nature.As such, it is interesting that among individual contributors, smaller contribution limits may increase the likelihood that an incumbent will win.The results suggest that by democratizing the donation process in favor of smaller contributions, campaign finance reform has had unanticipated effects.In a constrained contribution environment, donors are more likely to support the known commodity, thereby reducing the ability of non-incumbents to raise sufficient sums to inform the public about their campaign.With fewer constraints on fundraising, however, this study confirms the incumbency advantage in gubernatorial elections can be altered, allowing non-incumbents the opportunity to raise amounts similar to incumbent governors, which should encourage more equal performance among candidates in gubernatorial elections.

Figure 6 .
Figure 6.The conditional effect of campaign finance laws on contribution amount.
*p ≤ 0.05, **p ≤ 0.01, two-tailed test of significance; Note: Statistics reported in parentheses are robust clustered standard errors for Models 1 and 2 and bootstrap clustered standard errors for Model 3.