The Effect of Shootings in the United States on Contributions
from Gun-Related Political Action Committees and on
Companies connected with the National Rifle Association
University of Amsterdam
MSc Finance, track Corporate Finance Simone Korrel, 10636242
Supervisor Dr. T. Caskurlu 1st of July 2018
Statement of Originality
This document is written by Student Simone Korrel who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Acknowledgment
I would like to thank my supervisor, Dr. T. Caskurlu. His comments and insightful questions have been very helpful in writing this paper. Furthermore, I would like to thank the Federal Election Commission and the Center for Responsive Politics for providing me access to their data and for their assistance. Lastly, I would like to thank Mother Jones for their publicly available database.
Abstract
I investigate whether Political Action Committees (PACs) try to influence behavior of politicians after an event has taken place that is related to the interests of PACs, but not directly related to legislation. I investigate this by analyzing how shootings affect the number and amount of contributions from gun-related PACs. For this study, I use data on contributions from the Federal Election Commission and data on political committees and candidates from the Center for Responsive Politics. I use a difference-in-difference model to study the amount and number of contributions before and after shootings. To investigate the number of contributions, I use a Poisson regression model. To investigate the amount of contributions, I use an ordinary least square (OLS) regression model. I find that the number of contributions from gun-related PACs increases significantly in states where a shooting has taken place. However, the increase in the amount of contributions is not statistically significant. Additionally, I study the effect of shootings on companies that are connected with gun right PACs, by considering the stock return of companies that partner with the National Rifle Association (NRA) around a shooting. The results indicate that shootings have no effect on the valuation of companies that have connections with gun right PACs. Next, I study the valuation of connections with gun right PACs, by considering the stock return around the announcement date that companies terminate their partnership with the NRA. The results suggest that connections with gun right PACs do have an effect on the valuation of companies. However, additional analyses have to be done to confirm this effect.
Table of Content
Introduction... 6
I. Literature Review ... 10
a. Reasons to Contribute... 10
b. Equilibrium between Candidate and Legislator ... 10
c. Influence of Contributions on Politics ... 11
d. The Debate about Guns in the United States ... 13
e. Influence of Contributions from Companies on their Valuations ... 15
f. Identification of Companies ... 15 II. Data ... 16 a. Contributions ... 16 b. List of Shootings ... 19 c. Affected companies ... 22 III. Methodology ... 23
a. Models to Test the Effect of Shootings on Contributions ... 23
b. Models to Test the Effect of Shootings on Companies Connected with the NRA ... 27
IV. Empirical Results ... 29
a. Empirical Results of the Effect of Shootings on Contributions... 29
b. Empirical Results of the Effect of Shootings on Companies Connected with the NRA ... 38
V. Robustness Tests ... 41
a. Different Post Shooting Periods ... 41
b. Different Event Windows ... 43
c. Different Companies ... 45
VI. Conclusion ... 45
References ... 49
Introduction
Contributions play an important role in politics in the United States. The influence of contributions on legislation is discussed in several papers. By showing an increase in the number and amount of contributions during elections and in times of congressional debates, previous studies show that contributions are used to influence election outcomes and behavior of elected people, respectively (e.g. Baron, 1989; Brunell, 2005; Grossman & Helpman, 2001; Stratmann, 1998; Welch, 1982). However, changes in contributions around events that are not directly linked to legislation have not been studied before. In the first hypothesis of this paper, I investigate whether contributions from ideological Political Action Committees (PACs) are affected by events that are not directly linked to legislation, but that are related to the interests of the ideological PACs. Specifically, I consider the effect of shootings on the number and amount of contributions from gun-related PACs. An increase in the number and amount of contributions suggests that PACs try to influence behavior of politicians after an event occurs that is related to the interests of the PACs. In the second hypothesis, I investigate the effect of shootings on the valuation of companies that are connected with the major gun right PAC, the National Rifle Association (NRA). Next, in the third hypothesis, I study the valuation of connections with the NRA, by considering the stock return around the announcement date that companies terminate their partnership with the NRA. With these hypotheses, I investigate whether the thoughts that people have about the NRA influence the thoughts people have about companies that are connected with the NRA.
The results support the first hypothesis, which states that gun-related PACs increase the number of contributions in states where a shooting has taken place. Additionally, the results show an increase in the amount of contributions in states where a shooting has taken place. However, the increase in the amount of contributions is not statistically significant. The results reject the second hypothesis, which states that shootings have a negative effect on the valuation of companies that are connected with gun right PACs. The third hypothesis, which states that terminating connections with gun right PACs has a positive effect on the valuation of companies, cannot be accepted. The results do show a positive and statistically significant cumulative abnormal return around the announcement that companies terminate their NRA-partnership, however additional analysis has to be done to draw conclusions on the valuation of connections with gun right PACs.
To investigate the first hypothesis, concerning the effect of shootings on the number and amount of contributions, I retrieve data on contributions from the Federal Election Commission
(FEC, 2018a). Contributions from 2009 until 2017 are included in this study, these are the latest four terminated election cycles plus the first terminated year of the current election cycle. Data on committees and candidates is provided by the Center for Responsive Politics (CRP, 2018c). For this study, I investigate changes in the monthly number and amount of contributions in states before and after shootings. The shootings that are considered are documented by Mother Jones (Mother Jones, 2018a). In total there are 48 shootings included in this study. Control variables which indicate whether a month is close to an election are included to control for increases in the number and amount of contributions due to elections. To generate these control variables, I hand-collect dates of primary, general and special elections. I investigate changes in the number of monthly contributions with a Poisson regression model and changes in the amount of monthly contributions with an ordinary least squares (OLS) regression model. The post-shooting period used, to investigate whether there is an increase in contributions after a shooting, ranges from the shooting until the next general election within a state. This post-shooting period is used since I expect that contributions will be affected in this period. Other than the effect of shootings on the amount and number of contributions from gun-related PACs, I also investigate the effect on contributions from gun right PACs and gun control PACs separately. The results are in line with the results of gun-related PACs in general. However, the effect on the number of contributions is not statistically significant for gun right PACs, whereas the effect is statistically significant for gun control PACs. For gun control PACs, the effect on the amount of contributions is also statistically significant, whereas this effect is not statistically significant for gun right PACs and gun-related PACs in general. In an additional analysis, I study the dynamics of the effect of shootings on contributions. The results indicate that the effect is short term and is most pronounced in the month of the shooting. In robustness tests, I investigate the effect of shootings on different post-shooting periods. The results with these different post-shooting periods support the results of the main analysis, specifically, that a shooting has a significant effect on the number of contributions on the short term.
In the second hypothesis, I investigate whether shootings have a negative effect on the valuation of companies that are connected with gun right PACs. In the third hypothesis, I investigate the valuation of connections with gun right PACs. With these hypotheses, I study whether the thoughts that people have about the NRA influence the thoughts people have about companies that partner with the NRA. To investigate these hypotheses, I use news articles from CNN Money and Thinkprogress (CNN Money, 2018; Thinkprogess, 2018). With use of these articles, I identify which companies have terminated their partnership with the NRA after the Stoneman Douglas High School shooting in Florida on February 14th 2018. Implicitly, the
companies that ended their partnership with the NRA after the shooting, were still connected with the NRA during the shooting. Therefore, the same sample of companies can be used to study both hypotheses. I investigate the second hypothesis by considering the stock return of companies that partnered with the NRA at times of the shooting. The results indicate that shootings have no effect on the valuation of companies that are connected with gun right PACs. I investigate the third hypothesis by considering the stock return around the announcement date that companies terminate their partnership with the NRA. The results suggest that connections with gun right PACs have a negative effect on the valuation of companies, however additional analysis has to be done to confirm this effect. To test whether the results of the main analyses are robust, I consider the cumulative abnormal return for different event windows around the shooting and around the announcement that companies terminate their NRA-partnership. These results are in line with the results of the main analysis and support the implication that a shooting has no significant effect on the valuation of companies that are connected with the NRA. However, connections with the NRA might have an effect on the valuation of companies.
The relevance of this study is reflected by the continuous debates on guns and campaign financing in the United States. The debate about gun control versus gun rights in the United States is stirred up by recent mass shootings (CRP, 2016). On February 14th 2018, 17 students and faculty members of the Marjory Stoneman Douglas High School in Florida were killed by a shooting incident. On October 1st 2017, 58 people were killed in Las Vegas and on June 12th 2016, 49 people were killed in a nightclub in Orlando. These are only some of the mass shootings that have taken place in the United States over the last few years (CNN, 2018). In the second amendment, it is stated “A well-regulated militia, being necessary to the security of a free state, the right of the people to keep and bear arms, shall not be infringed.” However, there are dissimilarities in the interpretation of the second amendment and there are discussions going on whether the amendment should be adjusted and whether gun control measures should be implemented (The Guardian, 2017). The other ongoing debate, about campaign financing, is stirred up by a significant increase in the amount spent on politics in the United States. Whereas in year 2000 spending on presidential elections amounted to $1.41 billion, in year 2016 this amount increased to $2.39 billion (Bauer, Ginsberg & Persity, 2018). Additionally, the number of interest groups has increased and is still increasing (Grossman & Helpman, 2001). PACs are one type of interest group that tries to influence politics. PACs are sorted by the Center for Responsive Politics in several sectors. Gun rights and gun control are examples of industries within the ideological or single-issue sector (CRP, 2018b). Whereas gun right PACs try to influence politics to enhance gun rights, gun control PACs strive for more gun
control and regulation. The combination of these two present-day debates in the United States, about guns and about the role of money in politics, shows the importance of this research and shows the public interest in this topic.
This paper makes several contributions to existing literature. First, to the best of my knowledge, this paper is the first to study whether the number and amount of contributions from PACs change after an event takes place that relates to the interests of PACs, but that is not directly related to legislation. Previous studies have investigated contributions in times of elections and in times of voting about bills, however, no studies have investigated contributions in times of an event that is comparable with the one in this study. By showing an increase in the number of contributions in states where a shooting has taken place, I show that PACs try to influence the behavior of politicians in these states. More generally, I show that PACs try to influence behavior of politicians in states where their interest is criticized or discussed.
Second, to the best of my knowledge, this paper is the first to investigate whether the critique on the NRA influences the valuation of companies that are connected with the NRA. Many people criticize the NRA and other gun right PACs after a shooting. Moreover, some people criticize companies that work together with the NRA. However, no paper has previously investigated whether the valuation of companies that are connected with the NRA is affected by shootings. Also, the effect of terminating partnerships with the NRA and approaching the value of connections with the NRA via this way has not been studied before. By adjusting the setup of this study, the valuation of connections of companies with other ideological PACs can be studied as well. For example, the effect of a new published report about the noxiousness of tobacco on the valuation of companies connected with pro-tobacco PACs could be investigated.
Third, in this paper I control for primary and special elections in each state individually. Whereas other papers control for primary elections on a national level by the creation of a variable that equals one for the entire period in which primaries take place in the United States (e.g. Stratman, 1998), this paper controls for each state individually for primary and special elections. I only control for general elections on a national level since the timing of the general election is the same amongst all states. To the best of my knowledge, another unique database similar to the one I prepare, including all relevant election dates, does not exist.
The remainder of this paper is organized as follows. Section I provides an overview of relevant existing literature and introduces the hypotheses. Section II presents and describes the data sources. Section III explains the econometric methodology. Section IV provides the results. Section V shows several robustness tests. Section VI gives a summary and discusses implications, limitations and some suggestions for future research.
I.
Literature Review
a. Reasons to Contribute
Many economists and political scientist have studied campaign contributions in the United States. Previous studies show that contributing is done either to influence who will be elected (e.g. Baron, 1989; Brunell, 2005; Welch, 1982) or to influence behavior of elected people (e.g. Grossman & Helpman, 2001; Stratmann, 1998). There are several papers specifically studying the contributions from PACs in the United States. Brunell (2005) studies the motivation of contributions from PACs and concludes that money is given by pursuing both an electoral and ideological strategy. PACs contribute a predominant amount of money to candidates from their preferred party to influence election outcomes and contribute only a small amount of money to politicians from the other party, to gain access to these politicians. The amount contributed to the less-preferred party is small and strategically, to ensure that it will not influence the election outcome for the less-preferred party. Brunell (2005) shows that corporate PACs contribute a predominant amount of money to the Republican party, whereas labor PACs contribute the most money to the Democratic party. Stratmann (1998) studies the timing and thereby the motivation of contributions. He shows that in the weeks around important events, PACs make more contributions. Important events are here considered as elections or times when legislation is considered in congress. Stratmann (1998) studies the number and amount of contributions from agricultural PACs during elections and during the debate about the Farm Bill, which is a bill that concerns legislation about the farm sector. By showing an increased number and amount of contributions both during election times and during important legislative events, in this case the Farm Bill, he shows that contributions are used to influence who will be elected and to influence voting behavior of elected people, respectively. With voting behavior, it is meant whether elected people vote against or in favor of specific legislation.
b. Equilibrium between Candidate and Legislator
The relationship between candidates and PACs is studied in several papers. There is an electoral equilibrium in which candidates offer services in exchange for contributions from interest groups (Baron, 1989; Kroszner and Stratmann, 1998). According to Baron (1989), in a symmetric equilibrium, when candidates are in symmetric positions, it is expected that the contribution-service offer of candidates becomes more attractive the higher their valuation of being elected is and the lower their cost of providing the service is. However, generally the equilibrium in elections is not symmetric, there is an advantage for incumbents. Baron (1989)
provides four explanations for this incumbency advantage. The first explanation concerns the recognition advantage of incumbents. Since incumbents have a higher chance to win elections, they will get more contributions even when they do not offer more services. Second, the higher the valuation of re-election for incumbents is, compared to the valuation of election for challengers, the harder incumbents work to provide services to interest groups. Third, incumbents have lower marginal costs of providing services. Therefore, they can provide more services per contribution. Offering more services generates more contributions and increases the probability of winning an election. Fourth, an incumbent is more likely to be aligned with interest groups with high demand for services. Kroszner and Stratmann (1998) study the equilibrium after the election of candidates. They conclude that legislators form committees and that an equilibrium will be established when there is repeated interaction between these committees and interest groups. Interest groups give contributions to committees of the legislators and the legislators put legislative effort in the interests of the interest groups. This legislative effort includes amongst others, voting in favor of the interest group, challenging other legislators, drafting bills and doing interviews with the media in which the interests of the interest group are supported.
Furthermore, Kroszner and Stratmann (1998) look at the role of competition between PACs to determine to which candidates PACs contribute. When relationships are long-term, and uncertainty is low, competing PACs contribute large amounts of money to different candidates. When relationships are not long-term, and uncertainty is high, the lower contributions from the competing PACs match with each other, meaning PACs contribute to the same candidates. Kroszner and Stratmann (1998) study this by looking at contributions from three competing PACs with interests in financial services, specifically commercial banks, investment banks and insurance companies. They find that when PACs contribute to members of the House Banking Committee, high amounts are contributed to different members, whereas when contributions are given to legislators who are not members of the House Banking Committee, contributions are matched between PACs.
c. Influence of Contributions on Politics
The influence of money in politics is widely discussed. To limit the influence of money in politics, there have been several adjustments to legislation in the United States (Grossman & Helpman, 2001). In 1907, it became prohibited for corporations and trade organizations, and in 1943 for labor unions, to contribute directly to candidates. Around 1970, unions and organizations found a way to indirectly contribute to candidates by using PACs. The Federal
Election Campaign Act of 1974 states that this practice was allowed, which resulted in an increase in the number of PACs (Grossman & Helpman, 2001). However, to limit the influence of contributions from PACs in the legislation, the contributions to candidates are limited at $5,000 per candidate per election and the contributions to national political parties are limited at $15,000 per calendar year (Bauer et al., 2018). However, state laws can, and do, differ from federal law, also with respect to limits on contributions. Consequently, PACs have found methods to contribute more than was intended by the Federal Election Commission (FEC). Later, the FEC stated that national parties are allowed to use contributions above the limit of $5,000 per candidate per election, to cover administrative costs of the party. Most important is that the contributions above the limit are used for the party and not for a specific candidate. This resulted in ‘soft-money’ transactions. National parties can raise contributions without limits as long as the spending is in favor of the party and not, at least not recognizable, in favor of specific candidates. In 2002, The Bipartisan Campaign Reform Act (BCRA) prohibited the soft-money contributions to political parties from corporates and unions (Grossman & Helpman, 2001). However, corporate and union PACs could still contribute. Furthermore, corporates and unions could still contribute to Independent Expenditure Committees, named Super PACs. Super PACs are one type of outside group that are not allowed to formally coordinate, but can be aligned, with a political party or candidate. Those independent committees have no limit on their expenditures and since a legislative decision made in 2010, there is also no limit on the contributions from corporates and individuals (Bauer et al., 2018).
Even though the FEC uses and adjusts legislation to limit the influence of money in politics, there is no unambiguous evidence that money does influence politics. In previous studies, the results about the effect of contributions on voting behavior of politicians are mixed. Several papers (Baldwin and Magee, 2000; Langbein and Lotwis, 1990; Silberman and Durden, 1976) state that contributions influence voting behavior. Silberman and Durden (1976) show that contributions from the AFL-CIO, a labor PAC, do influence voting behavior on legislation with respect to minimum wages. Langbein and Lotwis (1990) find in their study that voting behavior on the McClure-Volkmer Bill, a bill about guns, was influenced by contributions from gun control and gun right PACs. Welch (1982) also finds a significant effect, measured by voting behavior on milk prices after receiving contributions from milk PACs. However, the paper of Welch (1982) shows that the influence of contributions on voting behavior is small, relative to other factors like ideology, party and consistency. Chappell (1982) states that the conclusion cannot be made that contributions influence voting behavior of legislators and that votes are seemingly based on ideology or preference of constituents. Differences in results about the
influence of money on politics might be explained by focusing on different interest groups in different studies. The influence of contributions on legislative decisions varies per issue (Kau, Keenan & Rubin, 1982). It seems that for some issues, behavior of politicians is influenced by contributions.
d. The Debate about Guns in the United States
Recent shootings stir up the debate about gun ownership in the United States. Despite the public debate, no gun control measures are passed by the House and Senate in the last years (CRP, 2018a). An outstanding example is the Manchin-Toomey amendment that obliges background checks for all commercial gun sales. In April 2013, after the Newton Shooting, this measure came to a vote. The measure got 54 votes in favor and 46 votes against, whereas 60 votes are needed in favor to pass a measure in the United States legislation (CRP, 2018a). The Center for Responsive Politics states that almost all of the politicians who voted against the measure had received sizeable amounts of money from gun right PACs. After another shooting in 2015, the San Bernardino killings, the measure was discussed again, this time only 48 politicians voted in favor of the measure. Most Republicans support gun rights. Of the 54 Republicans, only 4 voted in favor of the gun control measure in 2015 (CRP, 2018a). Democrats, on the other hand, are split between supporting gun rights and supporting gun control. According to the Center for Responsive Politics, the voting behavior of politicians is incompatible with the sentiment of the public. In 2015, the Pew Research Center found that 85 percent of the Americans is in favor of background checks, in 2017 this was 84 percent, still a large majority (Pew Research Center, 2017). Democratic Senator Dianne Feinstein stated that the “Congress also has a problem – a debilitating fear of upsetting the gun lobby” (CRP, 2018a).
There are more and larger gun right PACs than gun control PACs. In 2016, gun right PACs contributed $2,133,760 to federal candidates, whereas gun control PACs contributed $86,405. Gun right PACs contributed 97 percent to Republicans and 3 percent to Democrats. On the other hand, gun control PACs contributed 100 percent to Democrats and 0 percent to Republicans (CRP, 2018b). Of the $42 million that gun right interest groups have spent to political candidates, parties and interest groups since 1989, $23 million was spent by the NRA (CRP, 2018a). The NRA is the largest pro-gun PAC, which currently claims to have more than five million members (Thinkprogress, 2018). The annual fee to be a member of the NRA is $40. However, on the NRA website it is stated that “It pays to be a member”. Members of the NRA get “five-star benefits” and “five star savings” (NRA, 2018). Those savings are applicable
on multiple services and products, varying from discounts on wines and travelling to insurance for health and cancer care. The NRA uses these benefits and savings to promote memberships.
The first hypothesis of this paper investigates whether gun-related PACs try to influence politicians with contributions after a shooting incident has taken place. By considering shootings, this paper is, to the best of my knowledge, the first to study whether an incident that is not an election or a congressional debate, but nevertheless is an event that affects the related PACs, influences the contributions from those PACs. After the mass shooting in Florida on February 14th 2018, there has been an interview at the CNN town hall with the senators of Florida, Republican Marco Rubio and Democrat Bill Nelson, and congressman Ted Deutch. Children and teachers who survived the shooting and parents of killed children were present at the CNN town hall to ask questions to these politicians, to the spokesperson of the NRA, Dana Loesch, and to Sheriff Scott Israel. This study investigates whether the NRA and other gun right PACs increase contributions to politicians in states where shootings occur comparable to the shooting in Florida. On the short-term, the NRA could for example try to influence what politicians were going to say at the CNN town hall. On the long-term, the NRA might contribute more in the next general election in Florida to increase the probability that a candidate who is pro-guns will be elected. I investigate whether the number and amount of contributions to politicians increases in states where a shooting incident has taken place. Both the short-term and the long-term effect are studied. I investigate the effect of shootings on contributions from gun-related PACs per state, since I expect that shootings will have the largest impact on opinions and behavior related to guns in the states where the shootings have taken place. Furthermore, since PACs are limited in the total amount of contributions they can give, due to limited cash inflows, there is no increase expected in aggregated contributions on a national level. It could be that more people donate to gun-related PACs after a shooting has occurred, to support the PACs. However, the increase in contribution inflow of PACs, and therefore the number and amount of contributions on a national level, are not studied in this paper. Based on previous literature, I hypothesize the following:
Hypothesis 1: the number and amount of contributions from gun-related PACs increases in states where a shooting incident has taken place.
Hypothesis 1a: the number and amount of contributions from gun right PACs increases in states where a shooting incident has taken place.
Hypothesis 1b: the number and amount of contributions from gun control PACs does not increase in states where a shooting incident has taken place.
Even though I expect that the number and amount of contributions from gun control PACs retains stable or decreases in states where a shooting incident has taken place, I still expect an increase in the number and amount of contributions from gun-related PACs in general. I expect this, since contributions from gun right PACs have the major share in the total number and amount of contributions from gun-related PACs.
e. Influence of Contributions from Companies on their Valuations
Previous research investigates the relationship between political support and firm valuations. Jayachandran (2006) shows a significant effect of changes in political power on market valuations of firms, what he names the ‘Jeffords effect’. When Senator Jeffords left voluntarily the Republican party, the Democratic party became the majority, and subsequently political power shifted. The market valuation of firms that previously contributed to Republicans declined, whereas the market valuation of firms that previously contributed to Democrats raised. This can be interpreted as firms knowing which party behaves in their favor and that they contribute accordingly to this party. Another interpretation is that behavior of politicians can be influenced by contributions. Comparable studies to the study of Jayachandran (2006), are studies of Roberts (1990), Claessens, Feijen, and Laeven (2008), and Fisman (2001). Roberts (1990) shows a significant effect on the valuation of firms in response to the dead of Senator Jackson. Firms with constituent ties to Senator Jackson experienced a negative abnormal return after his dead, whereas firms with constituent ties to Senator Nunn, his successor, experienced a positive abnormal return. Constituent ties were determined by geographic and resource constituencies. Outside the United States, the relationship between political support and stock performance is also studied (e.g. Claessens et al., 2008; Fisman, 2001). Claessens et al. (2008) find a positive stock performance in election years for Brazilian firms that contribute money to politicians compared to Brazilian firms that do not contribute money to politicians. Fisman (2001) finds that political connectedness of firms, measured by the Suharto Dependency Index, does influence valuations of firms in Indonesia. It seems that the relationship between firms and political parties, based on whether and how much firms contribute to politicians, influences firm valuations.
f. Identification of Companies
Sugarmann (1992) has written a book about the NRA in which “NRA's purported use of fear, intimidation, and money to promote firearms sales and defeat any pro-gun control legislative measures” (p.111) is discussed. In the paper of Elsbach and Bhattacharya (2001), in
which they study identification versus disidentification with the NRA, one of the respondents states “I feel bad if I find out I accidentally used a product made by a corporation that supports the NRA. I feel like I’m being untrue to myself and I feel guilty” (p.399). Whereas some people want to identify themselves with groups like the NRA, others, contradictory, want to disidentificate themselves with the NRA. In light of the debates about guns and the critique on gun right PACs after shootings, especially on the NRA, it could be argued that this critique can also influence firms that support the NRA.
In the second hypothesis, I study the effect of connections with gun right PACs on the valuation of companies. I investigate whether shootings have an effect on the valuation of companies that are connected with the major gun right PAC, the NRA. Furthermore, in the third hypothesis, I investigate whether companies that terminate their partnership with the NRA experience an effect on their valuation. While most studies measure political support of companies by measuring direct contributions from companies to political candidates or committees, I use partnerships with the NRA to identify connections between companies and the major gun right PAC. I retrieve information about companies that terminated their NRA-partnership from news articles. This is more extensively discussed in the data section. Based on described existing literature, I hypothesize the following:
Hypothesis 2: shootings have a negative effect on the valuation of companies that are connected with gun right PACs.
Hypothesis 3: terminating connections with gun right PACs, shortly after a shooting has taken place, has a positive effect on the valuation of companies.
II.
Data
a. Contributions
To investigate the first hypothesis of this research, I retrieve data on contributions from the Federal Election Commission (FEC, 2018a). The FEC publishes files with contributions from committees to candidates for each two-year election cycle. I use data of the latest four terminated cycles, from 2009 up to and including 2016, and data of 2017. Most recent data of year 2018 is not used since this election year is not terminated yet. The amounts and dates of contributions and committee and candidate identification numbers are useful for this study. To get more detailed information about the contributing committees, I use the committee
information file published by the Center for Responsive Politics (CRP, 2018c). The CRP assigns all committees a Primcode, which is a code that identifies the industry or ideology of committees. By using this Primcode, the PACs that are related to gun rights and gun control can be identified. The Primcode for pro-gun parties is J6200 and the Primcode for anti-gun parties is J6100. Furthermore, I use a file with candidate information from the Center for Responsive Politics to be able to track to which party candidates belong (CRP, 2018c).
I merge the files with information about committees and candidates and the contribution data in one dataset. After excluding transactions that did not come from PACs, approximately 1.59 million transactions remain in the dataset. Of these transactions, around 1.1 million transactions are contributions in money and have a positive transaction amount. I exclude contributions with negative amounts from the dataset, since these are refunds of committees to donors (FEC, 2018b). The final sample contains 4,766 contributions from gun-related PACs, 64 from gun control PACs and 4,702 from gun rights PACs. Of these contributions from gun right PACs, 2,550 contributions are from the NRA. In table I, panel A and panel B show the number and average amount of contributions from gun right and gun control PACs, respectively. Whereas the number of contributions from gun right PACs is much higher than the number of contributions from gun control PACs, the average amount of contributions from gun right and gun control PACs differs to a lesser extent. The average amounts are $1,635 and $1,868, respectively.
For this study, I collapse the daily number and amount of contributions from each PAC to monthly numbers and amounts from all PACs accumulated. This is useful, since on many days zero contributions are made. Furthermore, I am interested in the total number of contributions from gun-related PACs, therefore it is not needed to have data for each separate PAC. Figure 1 shows the timing of contributions from gun-related PACs. There are four peaks visible around the general elections, which take place in November in all even years. The trends towards the peaks are roughly similar for all election cycles. The amount of contributions increases up to the general election and drops shortly after the general election.
Table I
Number and average amount of contributions from gun right and gun control PACs This table shows contributions from gun right and gun control PACs that are included in this study. Panel A shows the number of contributions and the average amount per contribution from gun right PACs. Panel B shows the number of contributions and the average amount per contribution from gun control PACs. These contributions are made between 2009 and 2018. Data on contributions is retrieved from the Federal Election Commission (FEC, 2018a) and identification of the gun-related PACs is done with use of the committee information files from the Center for Responsive Politics (CRP, 2018c).
Panel A: Gun right PACs
Number of contributions
Average amount per contribution ($)
Dallas Safari Club 88 715
Gun Owners of America 58 2,716
National Association for Gun Rights 96 3,384
National Rifle Association 2,550 1,606
National Shooting Sports Foundation 455 1,671
Ohio Gun Collectors Association 79 1,120
Safari Club International 1,376 1,598
Total 4,702 1,635
Panel B: Gun control PACs
Number of contributions
Average amount per contribution ($)
Brady Campaign to Prevent Gun Violence 15 540
Giffords PAC 39 2,550
Pride Fund to End Gun Violence 10 1,200
Figure 1. Amount of contributions from gun related PACs from 2009 until 2018. This figure shows the total amount of contributions from gun-related PACs in the United States for each month between January 2009 and December 2017.
b. List of Shootings
In the United States there are shootings daily. According to Every Town, on an average day 35 Americans die due to homicides with guns (Every Town, 2018). Consequently, not all shootings are included in this study, rather the number of shootings to include in this study has to be limited. Hence, I focus on mass shootings that are documented by Mother Jones (Mother Jones, 2018a). Mother Jones is a nonprofit investigative news organization. One of the topics Mother Jones reports about is mass shootings. Mother Jones holds a database of public mass shootings wherein the motive seems to be unselective assassination. Several criteria are used by Mother Jones to identify a shooting as a public mass shooting, specifically, the offender takes the lives of at least four people, the assassinations are performed by one single shooter and the shooting takes place in a public area. Also, some ‘spree killings’ are included in the database, these are shootings in multiple places shortly after each other. There are a few cases, well substantiated, included even though one of the criteria is not fulfilled (Mother Jones, 2018b). Available data for each mass shooting in the database include the location, date, the number of fatalities and injured people, the name of the case, a summary of the case, the venue
of the shooting, data about the offender (e.g. race, gender, mental health) and data about the type of weapon that is used and where the weapon was obtained. In this study, I include in total 48 cases that have taken place between 2009 and the end of 2017. There are several states that experienced multiple shootings during the sample period. California is the state that experienced the most shootings, specifically 9 shootings, followed by Washington with 6 shootings. There are 14 states in which only 1 shooting has taken place and 6 states in which the number of shootings differs between 2 and 5. In the remaining states, no shootings that fulfill the criteria of Mother Jones have taken place. Table AI in the Appendix shows all shootings and corresponding dates and states that this study refers to. Furthermore, table AI reports the number of fatalities and the number of injured people per shooting.
Table II shows the average number and amount of contributions from gun-related PACs before and after shooting. The averages given are per state. In these descriptive statistics, months in a specific state are included in the treated group from the month in which the shooting occurred until the next general election. In other months, these states belong to the control group. Furthermore, included in the control group are all observations of states where no shooting has taken place during the sample period. Panel A shows the monthly average number and amount of contributions per state when the dataset is balanced, which implies that the dataset includes months with zero contributions. Panel B shows the average monthly number and amount of contributions per state when for each state only months with positive contribution amounts are included. I make the dataset balanced on state and month level, since otherwise months in which zero contributions are made are excluded from the dataset which would upwardly bias the average monthly number and amount of contributions. The descriptive statistics suggest that the period after shootings is correlated with a higher number and amount of contributions per month. In this study, I investigate whether the differences between the number and amount of contributions before and after shootings are significant when controlled for several factors. Among these factors are the primary, general and special elections. Dates of these elections are retrieved from the site of the FEC and are manually added to the dataset.
Table II
Monthly amount and number of contributions from gun-related PACs per state This table presents the average number and amount of monthly contributions from gun-related PACs per state. Observations in a specific state are included in the treated group from the month in which a shooting has occurred until the next general election. In other months, these states belong to the control group. Moreover, included in the control group are all observations of states where no shooting has occurred during the sample period. For all states, data from January 2009 until December 2017 are used. Panel A shows the monthly average number and amount of contributions per state when the dataset is balanced, which implies that the dataset includes months with zero contributions. Panel B shows the monthly average number and amount of contributions per state when for each state only months with positive contribution amounts are included. Amounts are reported in dollars.
Panel A: Balanced Sample
Obs Mean Std D. Min Max
Control Group
Number of Contributions 5,091 0.577 .904 0 7
Amount of Contributions (in $) 5,091 1,305.20 2,819.097 0 39,000 Treated group
Number of Contributions 417 1.180 1.342 0 8
Amount of Contributions (in $) 417 2,788.10 4,394.228 0 32,600 Total sample
Number of Contributions 5,508 0.623 .957 0 8
Amount of Contributions (in $) 5,508 1,417.50 2,992.999 0 39,000 Panel B: Unbalanced Sample
Obs Mean Std D. Min Max
Control Group
Number of Contributions 1,952 1.505 .857 1 7
Amount of Contributions (in $) 1,952 3,404.10 3,685.805 100 39,000 Treated group
Number of Contributions 250 1.968 1.205 1 8
Amount of Contributions (in $) 250 4,650.60 4,854.215 72 32,600 Total sample
Number of Contributions 2,202 1.557 .915 1 8
c. Affected companies
To investigate the second and third hypotheses of this research, I make use of news articles about changes in the relationship between companies and the NRA, the most prevalent gun right PAC. Connections between companies and the NRA cannot be identified with contributions from companies to the NRA, because of the structure of the NRA as an entity. The NRA itself is a 501c3 non-profit organization that has two main political arms - the PAC (the NRA Political Victory Fund - NRA PVF) and the 501c4 (the NRA Institute for Legislative Action - NRA ILA). Both the c3 and c4 are exempt from having to disclose their donors (individual or otherwise) and the majority of the money raised by the NRA is done via these entities. The PAC is required to file with the FEC and to disclose their donors. However, the federal election law forbids corporations from contributing directly to PACs. This means that any corporation that wants to give money to the NRA has to do so through the NRA arms that do not disclose their donors. Hence, to identify companies with connections to the NRA, I do not use contributions to the NRA, but rather partnerships with the NRA. Specifically, I focus on companies that end their partnership with the NRA. I study the stock price around the announcement date that companies end their NRA-partnership. Since partnerships are more noticeable for the public than contributions, I expect that partnerships have at least as much influence as contributions, on how people value the company.
Furthermore, I study the stock return of companies that are connected with the NRA around a shooting, to see whether people link a shooting to companies that are connected with gun right PACs. CNN Money and Thinkprogress both published a list, with in total 19 companies that terminated their partnership with the NRA shortly after the shooting at the Marjory Stoneman Douglas High School in Florida on February 14th 2018 (CNN, 2018; Thinkprogess, 2018). 9 out of these 19 companies are public and have stock price data available. The companies and their corresponding announcement date of terminating their NRA-partnership can be found in table III. 2 out of the 9 companies of the sample, Delta Air Lines Inc. and United Continental Holding Inc., made their announcement on a Saturday, on the 24th of February 2018. For these companies I use the first trading day after the announcement, Monday 26th of February 2018, as the announcement date for studying the abnormal returns. Examples of other companies that ended their partnership with the NRA are the private companies Paramount RX Inc, SimpliSafe Inc, and Starkey Inc. However, these companies are not included in the study, due to unavailability of data.
Table III
Announcement dates of companies terminating their NRA-partnership
Company Announcement date
Symantec Corp. 23-02-18
Hertz Global Holdings Inc. 23-02-18
Metlife Inc. 23-02-18
Avis Budget Group Inc. 23-02-18
Truecar Inc. 23-02-18
Chubb Limited 23-02-18
Teladoc Inc. 23-02-18
Delta Air Lines Inc. 24-02-18
United Continental Holdings Inc. 24-02-18
III.
Methodology
a. Models to Test the Effect of Shootings on Contributions
The first hypothesis is investigated with a difference-in-differences test. The independent variable of interest is shooting, which is a dummy variable that equals one for a specific state from the month that a shooting takes place until the month of the next general election. In other months, the dummy variable equals zero and the state belongs to the control group. Moreover, included in the control group are observations of states where no shooting has occurred during the sample period. I use this post-shooting period, from the month in which the first shooting occurs in a state until the next general election in that state, since I expect that a shooting can influence the number and amount of contributions directly as well as later, until the next general election. I study the number and amount of contributions from gun-related PACs, and gun right and gun control PACs separately, before and after a shooting incident takes place. One of the assumptions for a difference-in-differences test is to have a parallel trend between the control group and the treated group. It is expected that without a shooting, the contributions in both types of states would have the same trend. This implies that changes in contributions would have been the same amongst both types of states when no shooting incident occurred. It is not possible to show a pre-shooting period parallel trend in this study, since states where a shooting has occurred are identified as control group until the first shooting. Shootings occur regularly, and the timing of shootings differs per state. However, figure 2 and figure 3 show non-parametrically that states with and without shootings have a roughly comparable trend. The figures show the average number of monthly contributions during the sample period. Figure 2 shows the average monthly number of contributions in states wherein no shooting occurred
during the sample period. Figure 3 shows the average monthly number of contributions for states that experienced at least one shooting during the sample period. As the figures show, in states that experienced at least one shooting, the average monthly number of contributions is slightly higher than in states that did not experience any shootings. This higher number of contributions could be the effect of shootings. This effect is studied in this paper.
In this study, the dependent variable is a measure of how many contributions are done and how much is contributed in state 𝑖 at time 𝑡. The number of contributions is the sum of all positive contributions, the amount of contributions is the sum of amounts of contributions with a positive amount. Comparable to the study of Stratmann (1998), to study the effect of shootings on the number of contributions, a Poisson regression model is used. The following Poisson distribution is used:
𝑝𝑟(𝑛𝑖𝑡) = 𝑒−𝜆𝑖𝑡𝜆𝑖𝑡𝑛𝑖𝑡
𝑛𝑖𝑡! (1)
whereby 𝑛𝑖𝑡 is the number of contributions from gun-related PACs in state 𝑖 in month 𝑡. The Poisson parameter varies across states and time as ln(𝜆𝑖𝑡) = 𝑥𝑖𝑡𝛽, whereby 𝑥𝑖𝑡 is a vector of regressors which describe the characteristics of the observation in a specific state at a specific time. 𝛽 is the coefficient of interest and is estimated with the model. To study the effect of a shooting on the amount of contributions, I use an ordinary least square (OLS) model of the following form:
ln(𝐶𝑜𝑛𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛𝑠it) = 𝛼 + 𝛽1∗ 𝑆ℎ𝑜𝑜𝑡𝑖𝑛𝑔𝑖𝑡+ 𝛽2∗ 𝐺𝑒𝑛𝑒𝑟𝑎𝑙𝑖𝑡+ 𝛽3∗ 𝑃𝑟𝑖𝑚𝑎𝑟𝑦𝑖𝑡 (2) + 𝛽4∗ 𝑆𝑝𝑒𝑐𝑖𝑎𝑙𝑖𝑡 + γt + γi + 𝜀𝑖𝑡
I use the natural logarithm of the amount of contributions to reduce the impact of outliers in the dataset. Furthermore, I add a value of one to the monthly amount of contributions before taking the natural logarithm, to avoid that the log of zero amounts drives the results. Coefficient of interest is 𝛽1, which measures the effect of shootings on the amount of contributions within a specific state. I include the variables general, primary and special as control variables in the OLS regression model and as regressors in the Poisson regression model. The first variable, the dummy variable general, controls for whether there is a general election in the next months. General elections take place on a national level in November in each second year of election cycles. The second variable, the dummy variable primary, controls for whether there is a primary election in the next months. Primary elections take place in the months before the general election and the timing of primary elections differs between states and is therefore
included for each state specific. The third variable, the dummy variable special, controls for whether there is a special election in the next months. Every year a few special elections take place in a few states. Special elections are also state specific, consequently timing of special elections also differs among states. The variables general, primary and special are dummies that equal one in the month and in the two months before the specified election and are equal to zero otherwise. I include these variables, because I expect that most contributions are given during election times. A positive coefficient for an election variable suggests that contributions are used to influence election outcomes. Variables included are also described in the appendix, in table AII. γt and γi are time and state fixed effects, which reduce omitted variable bias. These fixed effects are also included in the Poisson regression model. State fixed effects control for time-invariant differences in characteristics of states, amongst others, differences in legislation, culture, and demographics. Time fixed effects control for time trends between years. Time fixed effects control, amongst others, for whether there are changes in federal legislation. In both the Poisson and the OLS model, state clustered-robust standard errors are used. A positive relation between shootings and the number and amount of contributions from gun-related PACs suggests that gun-related PACs try to influence the behavior of politicians with respect to their position to guns after a shooting has occurred in the states of these politicians. Furthermore, to investigate how fast a shooting has an effect on contributions, I perform another regression. In this regression, I include a dummy variable for the month of the shooting and leads and lags for surrounding months.
The contribution limit for PACs, of $5,000 per candidate per election, could impede PACs to increase the number and amount of contributions after a shooting. If the limit is reached in an election before a shooting takes place, no additional contributions can be done in the specific state until the next election. Furthermore, in following elections the amount contributed in that specific state can at maximum be as high as in the previous election. If PACs want to contribute more, but are unable to contribute more, results about the increase in the number and amount of contributions in this study might be downward biased. To investigate whether this limit has a substantial impact in this study, I consider the total amount contributed to candidates by PACs. In this study, 97.55% of the PACs contributed less than $5,000 per candidate per election. 92.86% of the PACs contributed even less than $4,000 per candidate per election. These numbers suggest that in most cases, PACs could have contributed more. Therefore, it appears that the limit has no large impact on the ability of gun-related PACs to increase the number and amount of contributions after a shooting.
Figure 2. Number of monthly contributions from gun-related PACs in states where no shooting has taken place between 2009 and 2018. This figure shows the average number of contributions per state in states where no shooting has taken place between January 2009 and December 2017.
Figure 3. Number of monthly contributions from gun-related PACs in states where at least one shooting has taken place between 2009 and 2018. This figure shows the average number of contributions per state in states where at least one shooting has taken place between January 2009 and December 2017.
b. Models to Test the Effect of Shootings on Companies Connected with the NRA
To test the second and third hypotheses, I perform several event studies. Announcements in which companies declare to terminate their partnership with the NRA are used as source for connections between companies and the NRA. I consider stock returns around these announcements, to investigate whether the valuation of companies is influenced by connections with the NRA. The companies that made the announcement to terminate their NRA-partnership after the shooting, were connected with the NRA in times of the shooting. Therefore, I can use the stock return of these companies around the shooting date to investigate whether the valuation of companies that are connected with the NRA, is influenced by a shooting.
The shooting which I use for this study is the recent mass shooting at the Marjory Stoneman Douglas High School in Parkland, Florida on February 14th 2018. In this shooting, 17 people were killed, of which 14 students and 3 staffing people. A few days after the shooting, several news articles published announcements that companies terminated their partnership with the NRA (CNN Money, 2018). Since the NRA is the major gun right PAC and since there are no news articles about terminating partnerships with other gun-related PACs, I focus on connections with the NRA. By considering the abnormal return and the cumulative abnormal return around the date of the shooting and around the announcement date of terminating partnerships with the NRA, the identification of companies with guns and the valuation that is given to connections with the NRA are considered, respectively.
I consider the abnormal returns and the cumulative abnormal return of companies that partner with the NRA around the shooting incident with use of the three-factor model of Fama-French (1993). The model of Fama-Fama-French to predict returns is:
𝑅𝑗𝑡 = 𝛼 + 𝛽𝑗𝑅𝑚𝑡+ 𝑠𝑗𝑆𝑀𝐵𝑡+ ℎ𝑗𝐻𝑀𝐿𝑡+ ∈𝑗𝑡 (3)
whereby the predicted return of firm 𝑗 at time 𝑡 is dependent on the return of the market index at time 𝑡. Furthermore, 𝑆𝑀𝐵𝑡and 𝐻𝑀𝐿𝑡 are included which are the average return on small market capitalization portfolios minus the average return on large market capitalization portfolios and the average return on high book to market equity portfolios minus the average return on low book to market equity portfolios, respectively. Moreover, an error term is included, which is expected to have a value of zero. The sensitivity of the return of firm 𝑗 to the excess market return is measured with 𝛽𝑗. 𝑠𝑗 and ℎ𝑗 measure the sensitivity of the return of firm 𝑗 to the difference between small and large capitalization stock returns and to the
difference between value and growth stock returns, respectively. The estimation period that I use to predict the normal returns is from 395 until 30 days before the shooting, a one-year estimation period. The abnormal return is generated as follows:
𝐴𝑗𝑡 = 𝑅𝑗𝑡− (𝛼̂𝑗+ 𝛽̂𝑗𝑅𝑚𝑡+ 𝑠̂𝑗𝑆𝑀𝐵𝑡 + ℎ̂𝑗𝐻𝑀𝐿𝑡 ) (4)
wherein 𝛼̂𝑗, 𝛽̂𝑗, 𝑠̂𝑗 and ℎ̂𝑗 are OLS estimates of 𝛼, 𝛽𝑗, 𝑠𝑗 and ℎ𝑗 (Cowan, 2005). The average abnormal return is generated as follows:
𝐴𝐴𝑅𝑡 = ∑𝑁𝑗=1𝐴𝑗𝑡
𝑁 (5)
wherein 𝑡 is the number of trading days relative to the event date, which implies days relative to the shooting. The cumulative average abnormal return is determined with the following function: 𝐶𝐴𝐴𝑅𝑇1,𝑇2 = 1 𝑁∑ ∑ 𝐴𝑗𝑡 𝑇2 𝑡=𝑇1 𝑁 𝑗=1 (6)
wherein 𝑇1 and 𝑇2 are the first and the last day of the interval, respectively. I consider the mean abnormal return and mean cumulative abnormal return of companies around the announcement date of terminating their partnership with the NRA in the same way as around the date of the shooting. However, in that case the announcement of terminating the partnership with the NRA is taken as the event, whereas in the previous case the shooting is taken as the event. The announcement date differs between companies, table II shows the announcement dates of each company. Although I expect that the largest effects of the shooting and the announcements are visible the day after the events, I study abnormal returns of a week because media can bring more news out in the week following the event. Moreover, it can take a few days before the news is incorporated in the stock prices of the companies. I test whether abnormal returns are significant with two tests, the generalized sign test and the standardized cross-sectional test. The generalized sign test focuses on the ratio of positive versus negative abnormal returns. The standardized cross-sectional test works the same as the Patell test, which is the standardized abnormal return test. However, the standardized cross-sectional test adds a cross-sectional variance adjustment to the analytical variance of the total standardized prediction error (Boehmer, Masumeci & Poulsen, 1991).
IV. Empirical Results
a. Empirical Results of the Effect of Shootings on Contributions
With the first hypothesis, I investigate the effect of shootings on the number and amount of contributions from gun-related PACs. In this study, the dependent variables are the monthly number of contributions and the monthly amount of contributions per state. Table IV shows key incidence-rate ratios and coefficients for changes in the number and amount of contributions in states where a shooting has taken place. The total number of states is 51, including Washington D.C. The total number of observations is 5,508, composed of 51 states times 108 months (9 years of 12 months). The dependent variable in column (1)-(3) is the number of monthly contributions per state and the columns report incidence-rate ratios which are generated with a Poisson regression model. Incidence-rate ratios are exponentiated coefficients and are easier to interpret than standard coefficients in this analysis.
Column (1) shows regression results for the base model which includes the variable shooting and the election covariates. Whereas the coefficients of primary and general elections are statistically significant positive at a one percent level, the coefficient of special elections and the coefficient of shootings, are not statistically significant. Column (2) shows results of a regression in which state fixed effects are added to the base regression model. Consequently, the general and primary election coefficients retain statistically significant positive at a one percent level, the coefficient shooting becomes statistically significant positive at a five percent level and the special election coefficient becomes statistically significant positive at a one percent level. Economically, the shooting coefficient of 1.222 implies that the number of monthly contributions is 1.222 times higher in states where a shooting has taken place, compared to states where no shooting has taken place and compared to the months before the shooting in the specific state. Column (3) shows results of the regression that also includes time fixed effects. Consequently, the general, primary and special election coefficients retain positive and statistically significant at a one percent level, and the shooting coefficient is positive and statistically significant at a ten percent level. Economically, the shooting coefficient of 1.164 implies that the number of monthly contributions is 1.164 times higher in states where a shooting has occurred, compared to states where no shooting has occurred and compared to the months before the shooting in the specific state. The positive and statistically significant coefficients of general, primary and special indicate that the number of monthly contributions is 1.675, 1.483 and 1.245 times higher in the months before the general, primary and special election, respectively, compared to months which are not close to an election.
The dependent variable in column (4) is the natural logarithm of the amount of monthly contributions per state and the column reports coefficients which are generated with an OLS model. The coefficient of shooting of 0.385 economically implies that a shooting increases the amount of monthly contributions in that specific state with 46.96%1. However, this coefficient is statistically insignificant. The coefficients of the general, primary and special election are significant at a one percent level and suggest that the amount of monthly contributions is higher in the months before an election compared to months which are not close to an election.
The results in table IV indicate that shootings do not have a significant effect on the total amount of contributions. However, the results indicate a significant increase in the number of contributions in states where a shooting has taken place. The combined results, that gun-related PACs contribute statistically significantly more often to states where a shooting occurred, but that the amount of contributions does not increase statistically significantly after a shooting, suggest that the average amount per contribution might decrease after a shooting has taken place. I test whether this is true with an OLS model. Table AIII in the appendix shows the results. The table shows that indeed the amount per contribution decreases significantly at a ten percent level. Economically, the coefficient of -0.098 implies that the average amount per contribution is 10.30%2lower in states where a shooting has occurred. The combination of an increase in the number of monthly contributions and the decrease in the average amount per contribution, might indicate that PACs spread contributions among more politicians or that PACs contribute more often, but lesser amounts, to the same politicians.
1 (𝑒0.385− 1) ∗ 100% = 46.96%
Table IV
The effect of shootings on monthly contributions from gun-related PACs
This table shows key incidence-rate ratios and coefficients for changes in the number and amount of contributions from gun-related PACs after shootings. The dependent variable in column (1)-(3) is the number of monthly contributions per state and the columns report incidence-rate ratios which are generated with a Poisson regression model. The dependent variable in column (4) is the natural logarithm of the amount of monthly contributions per state and the column reports coefficients which are generated with an OLS regression model. The independent variable shooting is a dummy variable that equals one in states where a shooting has taken place, from the month of the shooting until the month of the next general election. The variables general, primary, and special are dummy variables that equal one in the month and the two months before the general, primary and special election in a specific state, respectively. Standard errors are clustered at state level and are reported in parentheses. ***, **, * denote significance at the 1%, 5% and 10% level, respectively.
(1) (2) (3) (4) Dependent Variable: Number of Contributions Number of Contributions Number of Contributions Amount of Contributions Shooting 1.234 1.222** 1.164* 0.385 (0.375) (0.107) (0.095) (0.236) Primary 1.730*** 1.734*** 1.483*** 1.007*** (0.129) (0.082) (0.085) (0.180) General 1.936*** 1.940*** 1.675*** 1.002*** (0.172) (0.125) (0.119) (0.181) Special 1.254 1.249*** 1.245*** 0.674*** (0.179) (0.098) (0.082) (0.243) Constant 0.495*** 2.075*** (0.046) (0.126) /lnalpha 0.635 (18.224) Observations 5,508 5,508 5,508 5,508 Number of states 51 51 51 51
State FE No Yes Yes Yes
