You are on page 1of 37

Traders, Guns, and Money

Anand Gopal
Robert H Smith School of Business
University of Maryland, College Park
College Park, MD 20742
agopal@rhsmith.umd.edu

Brad N. Greenwood
Fox School of Business
Temple University
Philadelphia, PA 19122
bgreenwood@temple.edu

Abstract
In this work, we investigate how mass shootings influence the stock price of
firearms manufacturers. While significant anecdotal evidence suggests there is an
immediate spike in firearms purchase after such events, the reaction of financial
markets is unclear. On one hand, if the increased short-term demand represents an
unanticipated financial windfall for firearms manufacturers, the stock prices of the
firms may rise. On the other hand, mass shootings may result in perceptions of the
social contract between the firm and society being systematically violated. In this
case, the increasing potential for regulation may render the firms business model
untenable in the long run, leading to stock prices decreasing. We empirically
resolve this tension using a market movement event study of 93 mass shootings
between 2009 and 2013. Findings suggest that stock prices of firearm
manufacturers significantly decline over a 2, 5, 10, and 30 day window in the
wake of mass shootings. Furthermore, these losses are exacerbated by the
presence of a handgun and the number of victims killed. Interestingly, results are
not influenced by the location of the crime, insofar as there is no difference across
red or blue states, or the loss of a child.
Key Words: mass shooting, event study, social contracts, firearms, financial markets

Introduction
In recent years, the American public has continued to be traumatized by mass shootings; events
where individuals suffering from mental health issues or psychological trauma and access to
firearms open fire on strangers in seemingly senseless acts of violence (Kellermann and Rivara
2013). Although scholars and policy makers have speculated at length regarding the impetus for
such events, one plausible explanation which continually has surfaced is the easy access to
firearms in the US (Acquisti and Tucker 2010). In 2013, for example the Washington Post
reported that among OECD countries, the US had the highest number of firearms per capita, as
well as the highest gun-related deaths per capita (Fisher 2012). However, although the issue of
gun control remains a controversial topic receiving significant political and social debate,
ranging from the constitutionality of the governments attempts to regulate and control gun sales
to pre-emptive treatment for the mentally unstable (Graziano and Pulcini 2013), one aspect of
mass shootings and gun violence is often ignored: the response of financial markets to the role
firearm manufacturers play in these shootings. More specifically, if firearm manufacturers are
held responsible by financial markets for manufacturing the very products used in these
incidents. This broad question: how financial markets respond to these events in terms of the
stock prices of firearm manufacturers; forms the core of the research reported in this paper.
While gun violence within the US occurs in many disparate forms, the most distressing
and visible form of such violence are mass shootings. Defined by the Federal Bureau of
Investigation (FBI) as any incident where at least four persons are killed with a firearm in a
random act with little or no premeditation (Mayors Against Illegal Guns 2013)1, these events can
occur in private homes, in the shooters current or prior workplace, and even in schools and

These crimes are distinct from acts like killing sprees (where the perpetrator moves between multiple locations with no time
break between the murders) or serial killings where the murders occur over a protracted period of time.

universities. Names such as Columbine and Sandy Hook Elementary have become cultural
markers in the US, capturing the essence of the randomness of such events. The specific form of
weapon used in these events also varies; Mayors Against Illegal Guns (2013) report that
approximate 15% of shootings between 2009 and 2013 were with automatic weapons, resulting
in significantly more deaths compared to crimes where handguns are used exclusively. Finally,
in 43% of such incidents, the perpetrator ended up committing suicide, further implicating the
role of mental health as a critical factor.
While the specific details of each event differ, all represent seemingly senseless acts of
violence where close family members, co-workers, and even relative strangers were adversely
affected through the use of firearms. As firearms are centrally implicated in these events, extant
theory would suggest two potential responses. On the one hand, considerable anecdotal data
shows a sharp spike in firearm purchasing immediately after such events; suggesting an
immediate arming of individuals who may then act in time to prevent further bloodshed2. The
Financial Times of London, for example, reported that gun sales jumped 52% in the year after
the mass shooting in Newtown, CT (Crooks and Foley 2013). Unless significant environmental
factors change, a rational response to such events is to expect further such events in the future
and thereby anticipate the resulting increase in gun sales from each shooting. Alternatively, mass
shootings increase the calls for additional gun control, regulation, mental health checks, and gun
registration (Birkland and Lawrence 2009). For instance, the recent mass shooting in Santa
Barbara, CA spurred calls for greater gun control. These were especially noteworthy because
they were issued by parents of victims, thereby making them highly salient3. The effects of such

http://www.gallup.com/poll/179213/six-americans-say-guns-homes-safer.aspx
http://www.washingtonpost.com/politics/father-of-victim-in-santa-barbara-shootings-to-politicians-i-dont-care-about-yoursympathy/2014/05/27/8a030d10-e5ad-11e3-a86b-362fd5443d19_story.html
3

regulations, if any, are likely to introduce significant controls and restrictions on the sales of
firearms in the US, thereby undermining firm profitability in the long term.
This dichotomy of responses to these shootings illuminates a central tension about how
the market may respond to manufacturers of firearms. On the one hand, the increase in salience
of gun violence from such events, and the associated increased sales associated with them, is
likely to systematically increase the value of investments in the firearms industry, absent some
structural changes in the way the firearm industry functions (Cross and Pruitt 2013). Therefore,
one might expect investors, and the market, to respond positively to such events by investing in
gun manufacturers. Alternatively, the potential for additional gun control, regulation, and
registration suggests that despite the immediate spike in sales from each event, the long term
viability of the manufacturers business model is questionable. The effects of regulation and gun
control are particularly relevant here given the recent history of nations like Australia where such
regulations were largely successful (Lee and Suardi 2010) and severely stunted firearm sales.
Beyond the immediate effect of mass shootings on the stock prices of firearms
manufacturers, there are further likely to be moderating effects to the long-term valuations of
firearms firms based on the specific circumstances around the event. Events like Newtown and
Columbine struck a chord with society as the victims included children. Similarly, the use of
automatic weapons and handguns has raised the profile of certain mass shootings (Lawrence and
Birkland 2004). Thus, as an extension of our base research question, we also consider how
specific circumstances about the event, such as the victims including children, the use of
handguns, and the location of the shooting, affect the markets response to the event.
We address these questions using a traditional market movement model event study
(Binder 1998, Boehmer et al. 1991), an accepted approach in finance to model abnormal returns

in the wake of exogenous events. As, almost by definition, mass shootings are random, there is
perfect identification in terms of the effects of the event on a focal firms stock price, notably in
the short-run. We use the dataset compiled by the Mayors Against Illegal Guns (2013) wherein
every mass shooting in the US, defined as events where at least four persons were killed, and its
associated circumstances have been identified. This dataset covers the period between January
2009 and September 2013 and includes 93 specific events. Most firearm manufacturers in the US
are private firms but two specific firms are public and form the focus of this study Smith and
Wesson (SWHC) and Sturm, Ruger, and Co. (RGR). We use firm-level stock data from the
CRSP dataset to identify stock price movements associated with mass shootings.
Results indicate a significant decrease in the stock price of firearm manufacturers over a
2, 5, 10, and 30 day window using a market movement event study methodology (Binder 1998).
Interestingly, we see no significant decrease in the stocks of related firms, i.e. those in the same
SIC or Text-based Network Industry Classification (Hoberg and Phillips 2010a), which also
occurs, thereby speaking to the robustness of the results and the exclusivity of the penalty to
firearm manufacturers. In an exhaustive set of subsequent analyses, findings further indicate that
the penalty firearm manufacturers receive is intensified by the number of people slain and the
use of a handgun. Interestingly however, we see no moderating effect if the victims are children
or if the crime occurs in a conservative or liberal state, thereby underscoring the fact that the
dominant mechanism by which this penalty occurs is the risk of increased regulation.
Economically, these results translate to a $32.1M drop in market capitalization for Smith &
Wesson over the 4-year period of our sample and a $42.8M loss in market cap for Ruger.
Our work makes several notable contributions to extant literature. First and foremost, a
large corpus of research surrounding mass shootings and gun control has focused, rightfully, on

the public policy, medical, legal, and sociological ramifications of gun violence in American
society (Acquisti and Tucker 2010, Kellermann and Rivara 2013). Inasmuch as the stock market
and financial investors represent part of the social structures in which guns are embedded, we
argue that it is also important to understand how the market responds to such events (Cross and
Pruitt 2013). Second, we contribute to the extant literature on stigma and corporate responsibility
associated with industries or firms (Ramchander et al. 2012, Zavyalova et al. 2012). Current
work in stigma addresses the extent to which media influences stigmatized industries such as the
armaments industry (Vergne 2012) or the extent to which wrong-doing by one firm stigmatizes
other similar firms (Jonsson et al. 2009). In comparison, we study a context where a certain
industry operates in a zone where it has wide acceptance and legitimacy, while also concurrently
experiencing widespread disapproval. We therefore respond to recent calls to examine the effect
of broader stakeholder support and social perception on firm valuation (Chatterji et al. 2014,
Henisz et al. 2014). We further contribute to this literature by investigating how an industry is
affected by a series of negative events that are, to a greater extent, out of its control, but are
intimately associated with the industries core products. We also test the extent to which these
events affect associated industries, thereby representing a form of stigma contagion. Finally, we
add to the literature on sin stocks. Prior work indicates that institutional influences and public
disapproval actually work towards providing higher yields on stock associated with the gaming,
alcohol, and tobacco industries (Hong and Kacperczyk 2009). While firearms firms are not
included in sin stocks, our study adds to the literature evaluating the effect of adverse events
on the market reaction to such firms.
Related Literature
Why would a random event such as a mass shooting affect the stock price of a firearm
manufacturer? It is easy to conceive of reasons why a rational investor may find firearm
5

manufacturers attractive as a result of such an event. Chief among them is the copious scholarly
and anecdotal literature suggesting that Americans tend to view gun ownership as a first step
towards protection from crime (Acquisti and Tucker 2010, Feldmann 2012). In more systematic
surveys, Barry et al. (2013) provide a more nuanced view on public opinion; namely that general
gun owners like National Rifle Association (NRA) members are more tolerant of guns in society
and express the sentiment that guns are useful as protective devices. Non-gun owners diverge
sharply, often expressing the need for greater gun control and laws, such as banning the use of
automatic weapons for recreational purposes. More broadly even, surveys by the Pew Research
Center4 and the Economist5 suggest that, in the aggregate, there appears to be little change in the
overall opinion across the country even after mass shootings. Therefore, for gun buyers and
members of shooting associations who believe in the value of firearms for self-defense or
protection, the enhanced salience provided by events such as mass shootings are likely to lead to
greater demand for firearms in the short-run.
Beyond protection, a second motivating force also plays a role in creating demand for
firearms; namely the enhanced probability of impending gun regulation that may emerge as a
result of mass shootings. Anecdotal evidence from one firearm retailer in Virginia states as
much: "Normally what happens - and I've been doing this for 30 years is whenever they start
talking about gun control on the news and they start pushing that, people have a tendency to
think they're going to take away their right to buy the gun, and that usually spurs sales6. Prior
work in psychology further corroborates this argument, suggesting that the effects of news media
messages about mass shootings can unduly influence opinions among viewers towards additional

http://www.people-press.org/2012/07/30/views-on-gun-laws-unchanged-after-aurora-shooting/
http://today.yougov.com/news/2012/07/25/after-aurora-little-change-opinions-about-gun-cont/
6
http://www.csmonitor.com/USA/DC-Decoder/2012/0725/Why-gun-sales-spike-after-mass-shootings-It-s-not-what-you-mightthink
5

gun control; as well as restrictions on the availability of guns for people with mental health
issues (McGinty et al. 2013). Thus, in the immediate aftermath of a mass shooting, the effect of
the concentrated media coverage can lead to the appearance of a groundswell movement towards
firearm regulation. The immediate effect of this sentiment within the market is an increase in
firearm sales through the combined perceived need for self-defense as well as a heightened sense
of impending regulation; thereby creating unexpected financial windfalls for the firm.
However, there is also reason to believe that markets may react negatively to such events.
As is well established, firearm advocates have consistently argued that guns dont kill people,
people do7, and that firearms should not be singly implicated when mass shootings occur.
Regardless of the validity of such arguments, as outlined above, there is an expectation of
enhanced firearm control emerging from events of gun violence, even though firearm
manufacturers offer their products legally in the marketplace and bear no direct culpability for
such events. We argue that this expectation can be explained by extant work on business ethics
outlining the presence of a social contract between industry and society at large, growing out of
an expanded view of the firms stakeholders within society, and may lead to a systematic
reduction in the long-term valuations of the firearms manufacturer.
Stakeholder theory argues that the stakeholders of the firm are not simply their stockholders, but all entities within society who are affected by the actions of the firm (Henisz et al.
2014, Parmar et al. 2010). While such an expanded view of stakeholders may not sit well with an
economic view of the firm, where the stakeholders represent only the owners of the firm (Jensen
and Meckling 1976), it is necessary for the firm and its managers to suitably construct value
propositions for the firm within its broader social ecosystem (Walsh 2005). Thus, part of the

http://www.psychologytoday.com/blog/logical-take/201302/guns-don-t-kill-people-people-do

managers job is to identify conflicts between diverse group of stakeholders and to resolve these
appropriately so that value creation is maintained (Harrison et al. 2010). Moreover, ignoring
certain stakeholders, and their legitimate claims (such as potential employees or potential
customers), can have adverse effects for the firm (Donaldson and Preston 1995, Henisz et al.
2014)8.
If firms are to address multiple stake-holders in their operations, then it is necessary for
the firm to devise certain ethical standards that will guide decision-making (Chatterji et al. 2014).
Extant research in social contracts, based on relevant ethical standards of society, are exactly the
guidelines that managers rely on to make such tradeoffs (Dunfee 1991); thereby making them the
informal and formal guidelines that organizations use in establishing and governing their
relationships with the community (Donaldson and Dunfee 2002). Encompassing more than
simply the firm and its specific set of economic partners, social contracts represent implicit
understandings or contracts that bind industries, companies, and economic systems into moral
communities (p.1854). Maneuvering through a complex network of social contracts that the
firm is embedded in requires that it fulfill the ethical as well as economic duties required within
each social contract.
When firms are viewed to be non-compliant with social contracts, such as when they
contribute directly or indirectly to violence, communities and social entities have the means to
punish the firm appropriately (Henisz et al. 2014). Examples of such punitive action are not
difficult to find. Within the business domain, firms that practice deception in negotiation may
find themselves deprived of business opportunities or partnerships (Dunfee 1991). Communities
can impose higher transaction costs on deviant firms, customers can boycott products or services,

A full treatment of stakeholder theory is beyond the scope of this paper; we refer the interested reader to Parmar et al. (2010) for
a comprehensive review.

potential and current employers can dissolve their associations with the firm. Finally, society can
choose to limit market transactions, impose taxation, create barriers to economic activity, or
sanction certain forms of economic interactions relating to the firm or industry (Donaldson and
Dunfee 2002). The current stigma attached to the financial services industry (Roulet 2014), for
example, or the accounting scandals from the early 2000s are instances of industries which were
socially and legally sanctioned (Durand and Vergne 2014), even though not every action taken
by these firms represented a legal breach.
Applying this lens to mass shootings, we argue that the firearms industry has always
existed in a twilight zone where the social contract between firearm manufacturers and society at
large in the US has been somewhat fuzzy (Brenkert 2000). Unethical marketing and sales
practices have long been noted within the firearms industry in the US, making access to firearms
easier than it should be (Green 2000). Moreover, easily available protective devices for firearms
(such as trigger locks), though available, have often been underutilized by manufacturers,
primarily for financial reasons. Green (2000) writes: Too often, instead of showing leadership
in improving their products or practices and positioning their industry for long-term viability,
business owners and managers have merely resisted change. Eventually, bad practices undermine
the conditions for continued growth or invite heavy-handed government regulation (p.204).
Therefore, even though the purchase and storage firearms is legal and protected under the
Second Amendment of the US Constitution, the notion of moral responsibility is increasingly
being attached to firearm manufacturers as a result of mass shootings (Brenkert 2000). Thus,
mass shootings represent significant breaches of the tenuous social contract that exists between
firearm manufacturers and society at large. As these breaches become more numerous and
increasingly burdensome to society, there is an increasing likelihood that sanctions, in the form

of regulations, social disapproval, and gun control; will be enacted and these sanctions almost
uniformly affect the industrys financial outcomes (Donaldson and Dunfee 2002). Inasmuch as
each shooting further rends the social contract between manufacturers and the public at large, it
would follow that each shooting carries with it greater odds of regulation and social sanction of
the industry, thereby reducing the long-term valuations of the firm.
It is important to note that the role of stigma or social disapproval influencing the stock
prices of certain industries is not new. In finance, prior work suggests that even within the
professional investor class, there exist societal norms against investing in industries that promote
vice, especially within investment managers of pension funds, endowments, religious
organizations, banks, and insurance companies (Hong and Kacperczyk 2009). This class of
stocks, typically associated with gaming, tobacco, and alcohol industries, are also viewed as
being more risky due to the threat of regulation and litigation (Kim and Venkatachalam 2011).
Paradoxically, research also suggests that investments in such stocks may provide higher returns
to investors; because the market for their trading is rendered inefficient due to social norms
(Hong and Kacperczyk 2009). Within our context, a further reason for a negative effect of mass
shootings on firearm stocks could come from systematic divestment of such stock by
institutional investors such as pension funds and endowments.
In summary, while we provide arguments for why mass shootings may result in abnormal
stock movements for firearms manufacturers in the US, whether the net response from the
market is positive or negative remains an empirical exercise. Therefore, in the absence of a
compelling a priori expectation, we explore this question empirically.
Empirical Analysis
Data and Estimation Procedure
To complete our empirical analysis we create a unique dataset drawing from multiple publicly

10

available sources of data. First, to capture the stock prices of existing firms we pull from the
Center for Research in Security Prices (CRSP) US Stock Database. CRSP is currently the largest
and most comprehensive historical stock market dataset in the world and has been used
extensively in prior work. This gives us the daily stock price of every publicly traded firm in the
US, thereby allowing us to calculate both the movement of firearm manufacturers as well as
other firms in the market. Data on mass shooting is drawn from Mayors Against Illegal Guns
(2013) press releases. Spearheaded by former New York City and Boston mayors Michael
Bloomberg and Thomas M Menino, this rich source of data grants us access not only to the date
of shootings, but the number of people affected, location, and whether or not a handgun was
involved. For the baseline analysis, we only use the date of the event to study the markets
response. Descriptive statistics of the 93 events can be found in Table 1.
The primary empirical technique we employ is a market movement event methodology
(Binder 1998, Boehmer et al. 1991). Using this methodology, we study the abnormal returns
which accrue to firearm manufacturers after a mass shooting over a one, two, five, ten, and thirty
day window for the 93 mass shootings which occurred in the United States between January
2009 and September 2013. Accordingly, the dependent variable for this investigation is the
percent change in the stock price (r) of the two publicly held firearm manufacturers: Smith &
Wesson (SWHC) and Sturm, Ruger, & Co. (RGR). While many other manufacturers do exist
(e.g. Browning Arms, Remington, and Colt), these two are distinct in that they are the only
public firearm manufacturers in the US9. Following the standard event study methodology, we
regress r on the percent change in return to the S&P 500 Market Index over the window of the

It should be noted that the small N does pose normality problems (Barclay and Litzenberger 1988) for the study despite the fact
that we are using inter-day, as opposed to intra-day, data. However, we make note of two features of the design which reduce the
weight of the concern. First, by replicating the empirical test over 93 events we are able to convert the effective sample size to
186 firm-shooting observations. Second, following McWilliams and Siegel (1997) we replicate our estimations using
bootstrapped standard errors. Results remain consistent.

11

regression (1, 2, 5, 10, and 30 days before and after treatment). The treatment, i.e. the mass
shooting, is dichotomous and applied on the first full day of trading after the event occurs to
prevent partial information dissemination (i.e. the shooting occurring after trading has closed or
traders not being fully informed of the event) from contaminating the empirical analysis.
Utilization of a market movement event study offers us several significant advantages.
First, to the extent that the dependent variable is the percent change in stock price of firm i, we
are able to mitigate concerns about serial correlation of the error terms because the change not
attributable to the treatment is assumed to be a random walk, following the efficient markets
hypothesis (Malkiel and Fama 1970). Second, as the shootings themselves are exogenous and
unpredictable events, i.e. random, and the effect is measured in the short term, we need not
include additional covariates to account for the propensity for the stock to over or under perform
against the market in the long term, e.g. Fama-French factors (Fama and French 1993). More
simply, because the treatment can be assumed to be exogenous, we need only determine how the
stock price of firm i generally trends with the market. To do this we include the appropriate %
change in S&P 500 market return during day j. Finally, it is important to note that an inherent
trade-off occurs when choosing the window of investigation for event studies. Insofar as too
short a window does not allow the model to sufficiently capture how the market and the return to
the stock price of firm i correlate, and too long a window allows for other events to contaminate
the experiment, this poses significant challenges. To remedy this concern, we estimate the effect
over many different windows. We therefore estimate the effect of mass shootings on firearm
manufacturer stock price using the following equation and an OLS estimator:
= 1 1 + 2 + +

where r represents return for firm i on day j, x is the dichotomous treatment indicator, and

12

Market is the change in S&P 500 market movement on day j. and represent the constant and
error terms, respectively, and {1, 2} represent the parameters to be estimated. Intuitively, the
estimated model represents a differences-in-differences approach wherein the stock price
variation of firearm manufacturers is compared to the average movement of the stock prices of
the S&P 500 Index before and after the event, within the stipulated time window. Since the
shootings are random by definition, comparing stock prices before and after the event allows a
clean identification of the effect of the shooting event on firearm manufacturers. Needless to say,
the wider the time window of the analysis, the more the possibility of confounding events.
However, within the shorter time windows, the effects of the mass shootings on firearm
manufacturer stock can be identified cleanly and without bias. Results are in Table 2.
Results
Results from Table 2 add interesting nuance to our understanding about how mass shootings
influence financial markets. First, as expected, we see that single day market return is strongly
correlated with the S&P 500 market index movement, indicating that the stock prices of these
firms do not abnormally track with the market. Furthermore, although we see no significant
effect in the first day, post shooting, we see a significant drop in the stock prices of firearm
manufacturers between 49.5 and 22.4 basis points, per day, over a 2, 5, 10, and 30 day window.
To determine the economic impact of the shooting, we calculate the effect on the market
capitalization for the two firms. With the market cap of Smith & Wesson (Ruger), currently
standing at $509.11mm ($677.65mm)10 this translates to a $5mm ($6.7mm) equity loss over two
days and an equity loss of $34.2mm ($45.5mm) equity loss over a 30 day window. During the
time period of our analysis, the volatility of the firearms stock was reasonably high. For instance,
the volatility of Smith and Wesson (SWHC) stock was roughly 41%, which devolves to a daily

10

Price retrieved on January 1, 2015

13

volatility of roughly 2.45%, representing variability that is predictable. The unpredictable effect
of a single mass shooting in our model on SWHC stock over a 5-day period is roughly 25 basis
points per day, which amounts to a total effect of 1.25% negative shock on the price. In terms of
magnitude, this is roughly half of the daily volatility, suggesting a significant economic impact.
A similar calculation for Ruger stock shows the same qualitative outcome. Clearly, mass
shootings contribute significantly to the volatility observed in firearms stock, except that these
movements are all uniformly negative and unpredictable.
Empirical Extensions
Results from Table 2 provide striking prima facie evidence that the mass shootings cause firearm
stocks to react adversely. The punitive effect of these events, in terms of the firearm firm market
capitalization, appears to be unequivocal. However, many of these events are associated with
other characteristics that can be probed further to better understand how exactly these events may
be affecting the social contract between gun manufacturers and society. Specifically, it is worth
considering these empirical extensions for two reasons. First, these extensive tests allow us to
test the robustness of the results from Table 2, in terms of both falsifiability and identifying
boundary conditions for these effects. Second, these tests allow us to establish which specific
aspects of mass shootings may be more or less damaging to firearm manufacturers, in terms of
the perceived breach of the social contract.
Network of Firms Around Firearm Manufacturers
Most firearm manufacturers in the US are embedded in wider networks of industries that form
the extended value chain. This extended network of industries may sometimes be closely
associated with firearms (such as retailers who sell firearms, for instance) or observed to be
peripheral at best (the steel industry, for instance). When a negative event, such as an industrial
accident, is reported, associations with the errant firm spread to other firms or industries that are

14

viewed as being similar (Jonsson et al. 2009, Zavyalova et al. 2012). Similarity here can involve
other firms in the same industry, customers or suppliers of the focal firm or industry, or in some
extreme cases, other firms with subjective impressions of likeness (Vergne 2012). In such
cases, the stigmatized firm may act to reduce long-term disapproval (e.g. name changes,
divestment of the stigmatized unit). Similarly, firms that are stigmatized by association can act to
reduce common elements with the source of stigma and to renounce all economic transactions
with the source of stigma. However, in the short-run, these options are limited. Therefore, they
are likely to bear the immediate market costs of such associations. Even in the longer-term,
renunciations of all economic transactions with the errant firm or industry can have negative
economic consequences that are likely to reduce the long-term valuations. We examine if this is
case with firms in industries that are associated with the two firearm firms after a mass shooting.
We define similarity with the two firearm manufacturers here in two ways. The first is
based on membership in the same industry subclass as the firearms manufacturers, based on the
3-digit SIC code. Thus, we replace the two firearm firms with the firms in their SIC-3 industry
and repeat the estimation of the market movement model. In this case, the treated firms are those
in the same SIC-3 code. Results are shown in Table 4 and show no economically significant
difference in the movement of these stocks relative to the S&P 500. There is a small positive
effect on day 1 and day 2 but these effects dissipate in the longer-term regressions. When viewed
cumulatively, it is possible that the members of the SIC may be receiving the benefit of the spike
in short term sales attributable to the mass shooting. However, the lack of a long-term effect
significantly reduces the robustness of this effect.
As SIC-3 classifications have increasingly come under fire in finance for being static and
non-representative of the heterogeneous firms that often are classified together within the same

15

industry code (Ahern 2012, Tetlock 2011), recent work has resorted to text-based analytical
techniques to form similarity metrics between public firms, based on the firm 10K filings and
descriptions of product markets and competition therein, as a preferred alternative to the SIC-3.
One such classification scheme is the Text-based Network Industry Classification (TNIC)
scheme proposed by Hoberg and Phillips (2010a, 2010b). This scheme dynamically matches all
public firms with other firms using text analysis of the firms 10k filings and creates similarity
indices between firms which can be established and modified year-on-year (see Hoberg and
Phillips (2010a) for a complete description). Hoberg and Phillips (2010a) thus offer a counterpart
to the SIC-3 that allows us to identify all firms in a year that are similar to the two firearm
manufacturers (more details on this methodology are provided in Appendix A). We use these
firms in the analysis next and re-estimate the market movement model. Results are provided in
Table 3 and again show no significant effects. The results from the TNIC and the SIC-3 also help
us establish that the effects observed for the two firearm manufacturers are specific to those
firms and are not based on membership in an industry subclass.
Empirically, there appears to be no contagion effects of the social contract breaches from
the firearm firms to other firms. This adds further credence to the argument that perceived
regulatory pressure are likely at play for the observed reduction in the stock prices of firearm
manufacturers. Any such regulatory effects will affect other public firms that are part of the
ecosystem but not as directly as they will affect firearm manufacturers. For instance, retailers
like Walmart may not be as directly affected by impending regulation, even though firearms
constitute a reasonable share of its sales11. To further test the working hypothesis that perceived
impending regulation is the dominant mechanism at play we study the effect of event-specific

11

http://business.time.com/2013/01/11/walmarts-on-again-off-again-relationship-with-guns/

16

factors that may enhance calls for regulation next. It is important to note that in the interest of
robustness, all subsequent empirical analyses were also conducted using the SIC and TNIC firms
as counterfactuals; the results showed no significant effects of mass shootings on these firms
stock price movements. Results of these analyses are available from the authors upon request.
Scope and Composition of Victims
The next test we execute is related to the composition of the victims affected by the shooting. To
the extent that the number of victims will increase the amount of rhetoric the shooting receives
from media outlets, and by extension the amount of pressure legislators may face during the next
legislative session, establishing the correlation between the number of victims and the degree of
punitive action in financial markets is important. By definition, mass shootings lead to at least
four fatalities, but to the extent that a greater number of individuals are affected, there is
unconditionally higher salience to the event. Furthermore, we also investigate if children, i.e. if
anyone under the age of 18, were involved. Prior work argues that the greater the assumed level
of innocence amongst the putative victims of corporate wrongdoing, the greater is the stigma and
disapproval attached to the perpetrating firm or industry (Vergne 2012). Note also that studying
the number of fatalities, in addition to whether a child was involved, allows separating the scope
of the shooting from the composition of the victim pool, a distinction which is likely important
when considering media attention, salience, and the pressure on regulatory bodies. To estimate
these effects we perform two additional sets of analyses. In the first, we replace the dichotomous
treatment variable with the count of the victims, the natural log of the number of people slain
(+1), and the number of victims squared (to capture a second-order effect). Next, we split the
sample into crimes involving children and those not involving children and estimate split-sample
analyses to evaluate the markets response to each context.
Results of these estimations (Tables 5 and 6) add striking nuance to our previous
17

estimations. First, we note in Table 5 that as the number of victims increase (either in the base or
log model), the size of the punitive effect increases (Columns 1, 2, 3, 6, and 7). Moreover, and
strikingly, the scope of the shooting does not have the same durational effect on stock price as
the simple dichotomous indicator, indicating that the occurrence of a mass shooting provides a
more sustained effect than the number of victims in that event. This could also be reflective of
the fact that media attention wanes significantly beyond the first few days of the event (Durand
and Vergne 2014). Finally, we note from the inclusion of the squared term a significant
curvilinear effect. While the non-polynomial term is positive and significant in three out of the
five regressions (Columns 13, 14, and 15) in each case it is dominated by the negative and
significant second order term. This suggests, all else equal, that mass shootings with fewer
victims have a smaller effect on the extent to which the stock price is negatively affected. As the
number of victims increase, the punitive effect increases non-linearly.
In a striking result, we further observe that the presence of children amongst the victims
appears to have no significant effect beyond the punitive action associated with the event itself
(Table 6). With the exception of a single coefficient (Column 7), which is marginally significant
(p<0.1), children being victims of the mass shooting does not seem to influence stock price in
any meaningful way beyond the total number of victims involved. Children are, of course,
involved in relatively few of the incidents in this dataset but carry tremendous salience in terms
of media coverage, both in the US and abroad. We note that even though the dataset includes the
recent Sandy Hook shooting in Newtown, CT, where 20 elementary school children were killed,
this incident remains an aberration even within the context of mass shootings in the US. In
summary, we observe a strong and non-linear negative effect of the number of people killed on
the stock prices of firearm manufacturers, adding further weight to the argument that the

18

expectations of regulation is the mechanism that is driving these results.


Handguns Involved
The second check of interest relates to the weapons used in the shooting itself. To the extent that
a wide variety of firearms have been employed, significant insight can be gleaned from how the
choice of firearm influences the markets reaction to the mass shooting. To this end we
investigate whether or not the involvement of a handgun in the execution of the crime
exacerbates or attenuates the penalty these firearm manufacturers receive. We use handguns for
the following reasons. While assault weapons are largely regulated by the Violent Crime Control
and Law Enforcement Act of 1994, handgun regulation has become the purview of state and
municipal governments12. Furthermore, there have been limited attempts to ban the use of
shotguns and hunting rifles due to both their use in recreational sport (viz. hunting) and their
importance in agricultural and ranching work. In contrast, handguns have seen multiple, and
often failed, attempts to increase regulation at the federal level, including the 1993 Firearms
Safety and Violence Prevention Act. Further, the Federal Consumer Product Safety Commission
(CPSC) does not regulate handguns as it does almost all other consumer products. Finally, the
debate on the right to carry concealed handguns (Black and Nagin 1998, Bronars and Lott 1998)
and its effect on gun violence (Duwe et al. 2002) continues, making this a particularly salient
factor to consider in terms of impending regulatory action.
To determine if the weapon used to perpetrate the mass shooting influences the degree of
the penalty we split our sample into those involving handguns, and those not involving handguns
(similar to the analysis with children). We then replicate our estimations of the market movement
model. Results are in Table 7. Strikingly, results indicate that the negative market reaction
disappears if the shooting does not involve a handgun (Columns 1-5). However, results further

12

http://www.handgunlaw.us/

19

indicate (Columns 6-10) that if a handgun is involved there is a significant penalty for firearm
manufacturers. The market response to a shooting appears to be primarily driven by the presence
of a handgun, further supporting the argument that anticipated regulation, a key factor in the sale
of handguns, is the key mechanism here.
Location of Crime
Our final test pertains to the location of the shooting event. As discussed previously, legislation
relating to the regulation of firearms is not only enacted at the federal level, but at the state and
municipal level as well. Therefore, the level of current state and municipal regulation concerning
firearms is likely reflected in the majority views of the voting population of the region, i.e. there
is likely stronger local regulation in liberal regions like New York, Massachusetts, and
Washington DC; but weaker regulation in conservative regions like Texas, Idaho, and
Oklahoma. Would a mass shooting in a liberal region provoke the same market reaction as one in
a conservative region? Arguably, because the odds of additional regulation of any sort in a
conservative-leaning region of the nation is small (since regulation of any sort is viewed
negatively), the markets reaction to a mass shooting in such a region may be less severe than if
such an event occurred in a liberal region where additional regulation may be viewed positively.
While this is speculative, we devise a test to check for differences in the markets response based
on the regions ideological leanings to ensure the changes in stock price are the result of calls for
regulation at the federal level and not because of heterogeneous local regulation.
To conduct this investigation we introduce a new variable, the Cook Partisan Voting
Index13, into the regression at the state level. Used extensively in prior research (Cook et al.
1994, Patterson and Caldeira 1988), this variable provides a continuous indicator of the degree of
partisanship (i.e. Democrat v Republican) in each state bounded between -40 (extreme

13

http://cookpolitical.com/

20

liberalism) and +40 (extreme conservativism). 0 indicates a politically neutral location. We then
interact this variable with the mass shooting treatment to determine if the reaction varies based
on the partisan voting level of the location. The results, shown in Table 8, indicate no significant
moderating effect across location, thereby suggesting that regulation at the federal level (which
legally supersedes state and municipal regulation and would therefore be applied unilaterally) is
likely the mechanism driving the observed decrease in the stock price of firearm manufacturers.
In addition, if the reductions in stock prices are associated with strategic divestment of firearm
stocks by socially minded investment managers, such as those observed in the context of sin
stocks (Hong and Kacperczyk 2009), the ideological bent of the location may be irrelevant. The
market in such a context is location-agnostic and applies corrective action regardless. While we
cannot rule out this eventuality, it is one alternative explanation worth considering.
Discussion and Conclusion
In this work we investigate how mass shootings in the US influences the stock price of publicly
traded firearms manufacturers. Drawing upon prior research from stakeholder theory (Donaldson
and Dunfee 2002) and significant anecdotal evidence showing that mass shootings lead to an
immediate spike in the sales of firearms, we argue that an a priori relationship between these two
events is unclear. On one hand, the increasing demand for guns, as a source of protection and a
fear of impending regulation, should lead to a positive impact on the valuation of the firm due to
the systematic spikes in revenue and the expectations of such events continuing, absent structural
changes in the marketplace. However, financial markets may respond negatively to such mass
shootings because the long-term viability of the firms business models appears questionable and
each event represents a renewed call for regulation. Results indicate that when mass shootings
occur, there is a systematic negative reaction in the stock price of firearm manufacturers over a 2,
5, 10, and 30 day observation window. Moreover, we observe that the context of the crime can
21

moderate this penalty. While the number of victims and use of a handgun significantly increases
the adverse reaction from the market, we see no effect on stock prices if the event involves
children or if it happens in a democratic or republican state.
Taken in sum, these results strongly suggest that when mass shootings occur, investors
appear to be reducing their valuations of these firms. We contend that these results are reflective
of the systematic violations of the social contract existing between firms and society in the US as
a result of gun violence. Even though gun manufacturers are not directly responsible for these
events, there is increasingly a sense of moral responsibility that is attributed to them (Brenkert
2000) based on impressions formed of inadequate self-regulation by the firearms industry on its
own practices and operations (Green 2000). Economically, our results indicate a penalty between
49.5 and 22.4 basis points, per day, when shootings occur, resulting in lost market capitalization
of $34.2mm ($45.5mm) for Smith & Wesson (Ruger) over a 30 day window. Interestingly, these
negative opinions are not extended to other firms or industries that are closely associated with
the two firearm manufacturers we study. In other words, while the market appears bearish on gun
manufacturers as a result of mass shootings, it does not penalize the suppliers of these firms or to
the retailers selling weapons.
From a theoretical perspective, the context that we study poses an interesting departure
from much of the literature that addresses the economic costs of wrong-doing by organizations.
On the one hand, received research includes several studies that address the negative impact on a
firm or on an industry when there is significant evidence of corporate wrong-doing or fraud
(Greve et al. 2010, Karpoff et al. 2005, Ramchander et al. 2012). Punitive action on the part of
regulatory agencies or consumers in such cases is understandable and often takes the form of
sanctions. Furthermore, the implications may even extend to relatively innocent but similar firms

22

(Jonsson et al. 2009, Zavyalova et al. 2012). Alternatively, firms often experience negative
attention due to industrial accidents, such as in the case of the BP Oil Spill in 2010 (Kling et al.
2012) or the toys recall in 2007 (Freedman et al. 2012). In such cases, depending on the firms
response to the crisis, there may even be a positive reputational effect accruing to the firm
(Zavyalova et al. 2012). In the toys context, firms that were proactive in recalling toys
significantly reduced spillovers from contaminated toy product lines to other product lines while
accumulating a reputation for quick action14. In each of these contexts, the focal firm had
considerable agency and clearly bore responsibility for deliberate or accidental wrong-doing.
In the case of firearm manufacturers, especially in the US, there are significant
differences in terms of corporate agency as well as responsibility. In many ways, the firearms
industry is closer to the sin stocks industries in that they offer a product that is legal and
regulated. However, the products marketed by tobacco and alcohol firms still require some level
of individual agency on the part of the actual consumer. Firms marketing these products are able
to game these consumers appropriately (Smith and Cooper-Martin 1997) though corporate social
responsibility initiatives within such firms have tempered these marketing activities considerably
(Luo and Bhattacharya 2006). In contrast, with mass shootings, a disproportionate cost is borne
by individuals who are neither customers nor directly related to the firearm industry. Moreover,
there is rarely ever admittance of corporate wrong-doing nor are mass shootings considered to be
industrial accidents. In spite of these events, firearms enjoy a level of prestige and acceptance
that are rarely ever accorded to tobacco and alcohol firms, and operate in a context that is
historically and economically unique to the US.

14

The decisions made by the executives at Johnson & Johnson to withdraw millions of bottles of contaminated Tylenol
medication, at great cost, actually provided the firm considerable positive reputational capital (Cheah et al. 2007) even though
several patients died as a result of the contamination.

23

While a historical analysis of firearms in the US is beyond the scope of our analysis, the
economic implications of how these products are used can and should be studied (Acquisti and
Tucker 2010) and we add further nuance the extensive scholarly debate regarding firearms in
American society. While much has been written about the health-related and criminological
aspects of gun violence, there is relatively less work addressing how the markets interact with
firearm manufacturers. Perhaps simply because firearm manufacturers have realized the social
costs that may emerge from misuse of their products, most American firearm manufacturers are
private and many are partly owned by private equity (e.g. the Cerberus Group15). If the market is
truly a channel through which social norms are reflected, and is also a medium by which social
contracts between firms and society are governed (i.e. firms are implicitly sanctioned or
rewarded), most firearm firms simply exist beyond these channels. They operate in a region
where on the one hand, breaches of the social contract are irrelevant to the firms per se, and, on
the other, society has limited means by which it can sanction these firms. This is reflected in
prior work that has addressed first, the limited attempts made by firearm firms to apply the right
safeguards to their products so as to reduce misuse (Green 2000), and second, the abilities of the
firearms firms to lobby and influence government and policy makers (Kellermann and Rivara
2013). Our work is the first to provide a simple but compelling cost estimate of the penalty
extended to firearm firms, even in a limited way, by markets for bearing moral responsibility
for shootings (Brenkert 2000).
Our results from contextual factors associated with mass shootings add further to extant
theory in social contracts by providing boundary conditions for the economic impacts of social
contract breaches. First, we see no evidence of contagion of stigma from firearm manufacturers

15

http://www.businessweek.com/articles/2014-03-12/cerberus-gets-a-1-billion-bid-for-americas-biggest-gun-company

24

to other firms. Other industries where firms have been guilty of intentional wrong-doing has led
to stigma being transmitted but given the unique conditions surrounding firearms, i.e. firearm
firms have never accepted direct responsibility, we see limited spillovers to others. This could
partly be a uniquely American phenomenon, given the important role given to firearms within
the US Constitution. Nevertheless, this is a boundary condition for when stigma or reputational
fallout does not convey to other firms. In a similar vein, we see clear evidence for the number of
people killed in an event (reflecting scale or magnitude of the event) but not for children
(reflecting composition of the victims). While striking at first, this may simply be an artifact of
the low frequency with which children are affected. However, it suggests that the market is
appropriately responsive to the scale of violence and not just the event itself.
In terms of practical implications, our work here helps inform policy makers about some
of the costs associated with mass shootings, and therefore gun violence, in the US. Much of the
gun violence in the US occurs outside such mass shootings and therefore, in real terms the
economic costs we identify are relatively small. However, it is worth considering what the total
cost of gun violence is likely to be if all such costs could be reflected through the market. With
just two firms, we observe losses in market capitalization of over $70M. If more firearm firms
were public or the true costs of gun violence in other contexts could be quantified, there would
likely be a cleaner case made for reasonable gun regulation and control that would satisfy all
parties. For firearm manufacturers, our results reflect an uncomfortable reality that the market
appears to be pessimistic about the viability of the business model in the long-term, even though
short-term sales may be high. To the extent that their own operations, public relations, and image
contributes to these market responses, these firms may benefit from re-engaging with society at
large on what is expected from them via their social contract rather than operating through

25

lobbying groups and organizations like the National Rifle Association (NRA). A recent Pew poll
showed that 39% of respondents felt that the NRA exerted too much influence on behalf of
firearm firms16. This proportion jumped to 52% among moderate liberals and 77% among
liberals. These statistics suggest a gradual but steady gain in support for regulation, which affects
firearm firms long-term prospects. Our work here suggests that a change in approach may be
useful for these firms in protecting and enhancing their long-term viability.
Our work here has certain limitations that need to be acknowledged. First, we are limited
by the fact that only two publicly traded firearms firms are available for analysis in the US.
Given this small set, it may be reasonable to consider our work here as a case study rather than a
generalizable large-sample event study. We do note that the analyses of stigma utilize a larger
sample of firms across multiple industry subsectors, thereby examining the effects of mass
shootings on a wider canvas. However, as we use public data sources for our analysis, all of our
models can be independently verified. Second, given the randomness of mass shootings and the
exogeneity it provides, we estimate relatively simple models of stock movements after the event.
More complicated econometric models may be devised but it is our contention that the basic
results will not change, given the sharply diverging coefficients observed for the S&P 500 index
and the firearms stocks. Finally, while we theorize about social contract breaches in our work,
we cannot rule out alternative explanations that may drive stock prices.
In summary, we provide a market movement model of how stock prices of firearm
manufacturers in the US are affected by mass shootings, defined as random shooting events
where at least four people were killed. We argue that such an analysis is necessary to identify the
economic costs of such events so as to better inform policy makers as well as investors and

16

http://www.pewresearch.org/fact-tank/2014/04/24/5-facts-about-the-nra-and-guns-in-america/

26

firearm manufacturers. Findings suggest that mass shootings result in a significantly negative
effects on the stock prices of two publicly traded gun manufacturers in the US. In addition, we
show that these effects are exacerbated when more people are killed and when a handgun is used.
We argue that these effects are likely caused by heightened expectation of gun regulation in the
future, which will significantly limit the economic viability of firearm manufacturers. Our work
is thus one of the few to consider the economic costs of mass shootings via the stock market.
References
Acquisti, A., Tucker, C. 2010. Guns, Privacy, and Crime.
Ahern, K.R. 2012. Bargaining Power and Industry Dependence in Mergers. Journal of Financial
Economics. 103(3) 530-550.
Barclay, M.J., Litzenberger, R.H. 1988. Announcement Effects of New Equity Issues and the
Use of Intraday Price Data. Journal of Financial Economics. 21(1) 71-99.
Barry, C.L., McGinty, E.E., Vernick, J.S., Webster, D.W. 2013. After NewtownPublic
Opinion on Gun Policy and Mental Illness. New England Journal of Medicine. 368(12)
1077-1081.
Binder, J. 1998. The Event Study Methodology since 1969. Review of quantitative Finance and
Accounting. 11(2) 111-137.
Birkland, T.A., Lawrence, R.G. 2009. Media Framing and Policy Change after Columbine.
American Behavioral Scientist.
Black, D.A., Nagin, D.S. 1998. Do Right-to-Carry Laws Deter Violent Crime? The Journal of
Legal Studies. 27(1) 209-219.
Boehmer, E., Masumeci, J., Poulsen, A.B. 1991. Event-Study Methodology under Conditions of
Event-Induced Variance. Journal of Financial Economics. 30(2) 253-272.
Brenkert, G.G. 2000. Social Products Liability: The Case of the Firearms Manufacturers.
Business Ethics Quarterly 21-32.
Bronars, S.G., Lott, J.R. 1998. Criminal Deterrence, Geographic Spillovers, and the Right to
Carry Concealed Handguns. American Economic Review 475-479.
Chatterji, A., Durand, R., Levine, D.I., Touboul, S. 2014. Do Ratings of Firms Converge?
Implications for Managers, Investors and Strategy Researchers. Forthcoming, Strategic
Management Journal.
Cook, E.A., Jelen, T.G., Wilcox, C. 1994. Issue Voting in Gubernatorial Elections: Abortion and
Post-Webster Politics. The Journal of Politics. 56(01) 187-199.
Crooks, E., Foley, S. 2013. Profits up for Rifle-Maker Freedom Group on Sandy Hook
Anniversary. Financial Times of Lonton.
Cross, B.W., Pruitt, S.W. 2013. Dark Knights Rising: The Aurora Theater and Newtown School
Massacres and Shareholder Wealth. Journal of Criminal Justice. 41(6) 452-457.
Donaldson, T., Dunfee, T.W. 2002. Ties That Bind in Business Ethics: Social Contracts and
Why They Matter. Journal of Banking & Finance. 26(9) 1853-1865.
Donaldson, T., Preston, L.E. 1995. The Stakeholder Theory of the Corporation: Concepts,
Evidence, and Implications. Academy of Management Review. 20(1) 65-91.
27

Dunfee, T.W. 1991. Business Ethics and Extant Social Contracts. Business Ethics Quarterly 2351.
Durand, R., Vergne, J.P. 2014. Asset Divestment as a Response to Media Attacks in Stigmatized
Industries. Strategic Management Journal.
Duwe, G., Kovandzic, T., Moody, C.E. 2002. The Impact of Right-to-Carry Concealed Firearm
Laws on Mass Public Shootings. Homicide Studies. 6(4) 271-296.
Fama, E.F., French, K.R. 1993. Common Risk Factors in the Returns on Stocks and Bonds.
Journal of Financial Economics. 33(1) 3-56.
Feldmann, L. 2012. Why Gun Sales Spike after Mass Shootings: It's Not What You Might Think.
Christian Science Monitor.
Fisher, M. 2012. Chart: The U.S. Has Far More Gun-Related Killings Than Any Other
Developed Country. Washington Post.
Freedman, S., Kearney, M., Lederman, M. 2012. Product Recalls, Imperfect Information, and
Spillover Effects: Lessons from the Consumer Response to the 2007 Toy Recalls. Review
of Economics and Statistics. 94(2) 499-516.
Graziano, M., Pulcini, J. 2013. Gun Violence and the Role of Health Care: A Confusing State of
Affairs. AJN The American Journal of Nursing. 113(9) 23-25.
Green, R.M. 2000. Legally Targeting Gun Makers: Lessons for Business Ethics. Business Ethics
Quarterly 203-210.
Greve, H.R., Palmer, D., Pozner, J.E. 2010. Organizations Gone Wild: The Causes, Processes,
and Consequences of Organizational Misconduct. The Academy of Management Annals.
4(1) 53-107.
Harrison, J.S., Bosse, D.A., Phillips, R.A. 2010. Managing for Stakeholders, Stakeholder Utility
Functions, and Competitive Advantage. Strategic Management Journal. 31(1) 58-74.
Henisz, W.J., Dorobantu, S., Nartey, L.J. 2014. Spinning Gold: The Financial Returns to
Stakeholder Engagement. Strategic Management Journal. 35(12) 1727-1748.
Hoberg, G., Phillips, G. 2010a. Product Market Synergies and Competition in Mergers and
Acquisitions: A Text-Based Analysis. Review of Financial Studies. 23(10) 3773-3811.
Hoberg, G., Phillips, G. 2010b. Real and Financial Industry Booms and Busts. The Journal of
Finance. 65(1) 45-86.
Hong, H., Kacperczyk, M. 2009. The Price of Sin: The Effects of Social Norms on Markets.
Journal of Financial Economics. 93(1) 15-36.
Jensen, M.C., Meckling, W.H. 1976. Theory of the Firm: Managerial Behavior, Agency Costs
and Ownership Structure. Journal of Financial Economics. 3(4) 305-360.
Jonsson, S., Greve, H.R., Fujiwara-Greve, T. 2009. Undeserved Loss: The Spread of Legitimacy
Loss to Innocent Organizations in Response to Reported Corporate Deviance.
Administrative Science Quarterly. 54(2) 195-228.
Karpoff, J.M., Lott Jr, J.R., Wehrly, E.W. 2005. The Reputational Penalties for Environmental
Violations: Empirical Evidence*. Journal of law and economics. 48(2) 653-675.
Kellermann, A.L., Rivara, F.P. 2013. Silencing the Science on Gun Research. JAMA. 309(6)
549-550.
Kim, I., Venkatachalam, M. 2011. Are Sin Stocks Paying the Price for Accounting Sins? Journal
of Accounting, Auditing & Finance. 26(2) 415-442.
Kling, C.L., Phaneuf, D.J., Zhao, J. 2012. From Exxon to Bp: Has Some Number Become Better
Than No Number? The Journal of Economic Perspectives 3-26.

28

Lawrence, R.G., Birkland, T.A. 2004. Guns, Hollywood, and School Safety: Defining the
SchoolShooting Problem across Public Arenas*. Social Science Quarterly. 85(5) 11931207.
Lee, W.-S., Suardi, S. 2010. The Australian Firearms Buyback and Its Effect on Gun Deaths.
Contemporary Economic Policy. 28(1) 65-79.
Luo, X., Bhattacharya, C.B. 2006. Corporate Social Responsibility, Customer Satisfaction, and
Market Value. Journal of Marketing. 70(4) 1-18.
Malkiel, B.G., Fama, E.F. 1970. Efficient Capital Markets: A Review of Theory and Empirical
Work. The Journal of Finance. 25(2) 383-417.
McGinty, E.E., Webster, D.W., Barry, C.L. 2013. Effects of News Media Messages About Mass
Shootings on Attitudes toward Persons with Serious Mental Illness and Public Support
for Gun Control Policies. American Journal of Psychiatry. 170(5) 494-501.
McWilliams, A., Siegel, D. 1997. Event Studies in Management Research: Theoretical and
Empirical Issues. Academy of Management Journal. 40(3) 626-657.
Parmar, B.L., Freeman, R.E., Harrison, J.S., Wicks, A.C., Purnell, L., De Colle, S. 2010.
Stakeholder Theory: The State of the Art. The Academy of Management Annals. 4(1)
403-445.
Patterson, S.C., Caldeira, G.A. 1988. Party Voting in the United States Congress. British Journal
of Political Science. 18(01) 111-131.
Ramchander, S., Schwebach, R.G., Staking, K.I.M. 2012. The Informational Relevance of
Corporate Social Responsibility: Evidence from Ds400 Index Reconstitutions. Strategic
Management Journal. 33(3) 303-314.
Roulet, T. 2014. What Good Is Wall Street? Institutional Contradiction and the Diffusion of
the Stigma over the Finance Industry. Journal of Business Ethics 1-14.
Smith, N.C., Cooper-Martin, E. 1997. Ethics and Target Marketing: The Role of Product Harm
and Consumer Vulnerability. The Journal of Marketing 1-20.
Staff. 2013. Mass Shootings since January 20, 2009. Mayors Against Illegal Guns,
http://www.washingtonpost.com/blogs/wonkblog/files/2013/02/mass_shootings_200913_-_jan_29_12pm1.pdf.
Tetlock, P.C. 2011. All the News That's Fit to Reprint: Do Investors React to Stale Information?
Review of Financial Studies. 24(5) 1481-1512.
Vergne, J.-P. 2012. Stigmatized Categories and Public Disapproval of Organizations: A Mixed
Methods Study of the Global Arms Industry (1996-2007). Academy of Management
Journal amj. 2010.0599.
Walsh, J.P. 2005. Book Review Essay: Taking Stock of Stakeholder Management. Academy of
Management Review. 30(2) 426-438.
Zavyalova, A., Pfarrer, M., Reger, R., Shapiro, D. 2012. Managing the Message: The Effects of
Firm Actions and Industry Spillovers on Media Coverage Following Wrongdoing.
Academy of Management Journal.

29

Figure 1: Snapshot of the Mass Shootings Dataset

1
2
3
4
5
6

r
Market
NumKilled
Children
Handgun
Cook PVI

Mean
0.0009199
0.0002489
5.26087
0.3913043
0.6847826
-1.206522

Table 1: Summary Statistics


N 93 Mass Shootings
Std. Dev.
(1)
(2)
0.0350313
0.017966
0.4811
3.040239
-0.062
0.1011
0.4893739 -0.0734
0.0112
0.4658701 -0.0003
-0.0508
10.76885
-0.0571
-0.0469

30

(3)

(4)

(5)

-0.069
0.1818
0.1876

-0.1751
-0.2127

-0.0239

Table 2: Effect of Mass Shootings on Firearm Manufacturer Market Returns


Dependent Variable: Percent Change in Firm Stock Price
Model: OLS Market Movement Event Study
Dependent Variable
Time Window
Treatment
Market
Constant
N
R-squared

(1)
(2)
(3)
(4)
(5)
r
r
r
r
r
1 Day
2 Day
5 Day
10 Day
30 Day
-0.00265 -0.00495** -0.00279* -0.00243** -0.00224***
(0.00292) (0.00219)
(0.00147) (0.00103) (0.000761)
1.047***
1.076***
1.051***
1.100***
1.038***
(0.0929)
(0.0743)
(0.0489)
(0.0356)
(0.0269)
0.00326* 0.00322** 0.00171* 0.00284*** 0.00339***
(0.00168) (0.00139) (0.000992) (0.000714) (0.000542)
552
920
2,024
3,864
7,356
0.188
0.188
0.187
0.199
0.169
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

Table 3: Effect of Mass Shootings on TNIC Market Returns


Dependent Variable: Percent Change in Firm Stock Price
Model: OLS Market Movement Event Study
Dependent Variable
Time Window
Treatment
Market
Constant
N
R-squared

(1)
(2)
(3)
r
r
r
1 Day
2 Day
5 Day
0.00139
-5.68e-05 -0.000433
(0.00279) (0.00197)
(0.00129)
1.189***
1.103***
1.096***
(0.0888)
(0.0669)
(0.0428)
0.00153
0.00149
0.00135
(0.00161) (0.00125) (0.000868)
1,104
1,840
4,048
0.142
0.129
0.140
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

(4)
(5)
r
r
10 Day
30 Day
0.000226
0.000381
(0.000912) (0.000654)
1.049***
1.018***
(0.0314)
(0.0231)
0.000904 0.000877*
(0.000629) (0.000465)
7,728
14,712
0.126
0.117

Table 4: Effect of Mass Shootings on SIC Market Returns


Dependent Variable: Percent Change in Firm Stock Price
Model: OLS Market Movement Event Study
Dependent Variable
Time Window
Treatment
Market
Constant
N
R-squared

(1)
(2)
(3)
(4)
r
r
r
r
1 Day
2 Day
5 Day
10 Day
0.0152**
0.0114**
0.00350
0.000937
(0.00705) (0.00470)
(0.00316)
(0.00244)
0.445**
0.481***
0.507***
0.579***
(0.226)
(0.161)
(0.106)
(0.0851)
0.00117
-0.000487
0.00336
0.00497***
(0.00405) (0.00297)
(0.00213)
(0.00168)
1,618
2,692
5,922
11,309
0.006
0.006
0.004
0.004
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

31

(5)
r
30 Day
0.000160
(0.00174)
0.647***
(0.0620)
0.00499***
(0.00124)
21,610
0.005

Table 5: Effect of Mass Shootings Market Returns by Number of Victims


Dependent Variable: Percent Change in Firm Stock Price
Model: OLS Market Movement Event Study
(1)
(2)
(3)
Dependent Variable
r
r
r
Time Window
1 Day
2 Day
5 Day
NumKilled
-0.00126*** -0.00136*** -0.000451*
(0.000451) (0.000353) (0.000242)
Ln(NumKilled)

(4)
r
10 Day
-0.000204
(0.000170)

Market

1.099***
1.036***
(0.0356)
(0.0269)
0.00275*** 0.00227***
(0.00103) (0.000769)
3,864
7,356
0.198
0.168

Constant
N
R-squared

Dependent Variable
Time Window
NumKilled
NumKilled ^ 2
Market
Constant
N
R-squared

1.059***
1.088***
(0.0923)
(0.0740)
0.00899*** 0.00836***
(0.00274) (0.00214)
552
920
0.198
0.197

1.051***
(0.0489)
0.00281*
(0.00147)
2,024
0.187

(5)
r
30 Day
-3.22e-06
(0.000127)

(6)
r
1 Day

(7)
r
2 Day

(8)
r
5 Day

(9)
r
10 Day

(10)
r
30 Day

-0.0108**
(0.00462)
1.061***
(0.0927)
0.0216***
(0.00831)
552
0.195

-0.0115***
(0.00361)
1.088***
(0.0743)
0.0216***
(0.00648)
920
0.193

-0.00256
(0.00246)
1.050***
(0.0489)
0.00499
(0.00443)
2,024
0.186

-0.000518
(0.00174)
1.098***
(0.0356)
0.00260
(0.00312)
3,864
0.198

0.00113
(0.00129)
1.035***
(0.0269)
0.000248
(0.00231)
7,356
0.168

(11)
(12)
(13)
(14)
(15)
r
r
r
r
r
1 Day
2 Day
5 Day
10 Day
30 Day
0.000389
0.000826
0.00166** 0.00132** 0.00104***
(0.00144) (0.00112) (0.000770) (0.000543) (0.000401)
-6.62e-05 -8.78e-05** -8.50e-05*** -6.15e-05*** -4.25e-05***
(5.51e-05) (4.30e-05) (2.94e-05) (2.08e-05) (1.55e-05)
1.050***
1.079***
1.045***
1.096***
1.035***
(0.0926)
(0.0740)
(0.0488)
(0.0356)
(0.0269)
0.00277
0.000125
-0.00516*
-0.00301
-0.00167
(0.00586) (0.00456)
(0.00312)
(0.00221)
(0.00163)
552
920
2,024
3,864
7,356
0.200
0.200
0.190
0.200
0.169
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

Table 6: Effect of Mass Shootings of Children on Market Returns


Sample Split Based on Whether or Not a Child is a Victim
Dependent Variable: Percent Change in Firm Stock Price
Model: OLS Market Movement Event Study
Dependent Variable
Time Window
Sample
Treatment
Market
Constant
N
R-squared

(1)
r
1 Day
-0.00233
(0.00369)
0.955***
(0.118)
0.00461**
(0.00212)
336
0.165

(2)
r
2 Day

(3)
r
5 Day
No Children
-0.00369
-0.00225
(0.00270) (0.00183)
1.032***
1.063***
(0.0967)
(0.0633)
0.00379** 0.00234*
(0.00171) (0.00124)
560
1,232
0.171
0.187

(4)
r
10 Day

(5)
r
30 Day

(6)
r
1 Day

-0.00313** -0.00271*** -0.00300


(0.00129) (0.000963) (0.00478)
1.082***
0.993***
1.181***
(0.0467)
(0.0349)
(0.151)
0.00344*** 0.00379*** 0.00119
(0.000894) (0.000688) (0.00276)
2,352
4,488
216
0.187
0.154
0.223
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

(7)
r
2 Day
-0.00692*
(0.00370)
1.133***
(0.117)
0.00232
(0.00234)
360
0.213

(8)
r
5 Day
Children
-0.00364
(0.00245)
1.033***
(0.0772)
0.000717
(0.00165)
792
0.186

(9)
r
10 Day

(10)
r
30 Day

-0.00139
(0.00172)
1.120***
(0.0553)
0.00194
(0.00118)
1,512
0.214

-0.00156
(0.00124)
1.099***
(0.0422)
0.00280***
(0.000879)
2,868
0.192

Table 7: Effect of Mass Shootings on Market Returns


Sample Split Based on Whether or not a Handgun is Involved
Dependent Variable: Percent Change in Firm Stock Price
Model: OLS Market Movement Event Study
Dependent Variable
Time Window
Sample
Treatment
Market
Constant
N
R-squared

(1)
r
1 Day

(2)
r
2 Day

-0.00208
(0.00557)
1.443***
(0.219)
0.00275
(0.00315)
174
0.206

0.000526
(0.00382)
1.051***
(0.143)
0.000398
(0.00244)
290
0.159

(3)
r
5 Day
No Handgun
-0.00105
(0.00263)
1.009***
(0.0933)
0.000982
(0.00177)
638
0.156

(4)
r
10 Day

(5)
r
30 Day

(8)
(9)
(10)
r
r
r
5 Day
10 Day
30 Day
Handgun
-0.00121
-0.000891
-0.00373 -0.00750*** -0.00357** -0.00298** -0.00284***
(0.00178) (0.00131) (0.00342)
(0.00267) (0.00178) (0.00127) (0.000933)
1.081***
0.991***
0.940***
1.090***
1.066***
1.107***
1.055***
(0.0643)
(0.0498)
(0.102)
(0.0873)
(0.0574)
(0.0428)
(0.0320)
0.00218* 0.00225** 0.00335* 0.00454*** 0.00204* 0.00313*** 0.00391***
(0.00123) (0.000931) (0.00198)
(0.00169) (0.00120) (0.000876) (0.000665)
1,218
2,306
378
630
1,386
2,646
5,050
0.189
0.147
0.187
0.204
0.201
0.203
0.178
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

33

(6)
r
1 Day

(7)
r
2 Day

Table 8: Effect of Mass Shootings on Market Returns Moderated By Cook Partisan Voting Index
Dependent Variable: Percent Change in Firm Stock Price
Model: OLS Market Movement Event Study
Dependent Variable
Time Window
Treatment
Cook PVI
Treatment * Cook PVI
Market
Constant
N
R-squared

(1)
(2)
(3)
r
r
r
1 Day
2 Day
5 Day
-0.00264
-0.00483**
-0.00262*
(0.00295)
(0.00220)
(0.00148)
-1.82e-05
-0.000217*
-5.66e-05
(0.000157) (0.000129)
(9.23e-05)
9.14e-06
0.000103
0.000147
(0.000272) (0.000204) (0.000137)
1.047***
1.075***
1.051***
(0.0932)
(0.0743)
(0.0489)
0.00324*
0.00296**
0.00164
(0.00170)
(0.00139)
(0.000998)
552
920
2,024
0.188
0.191
0.187
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

(4)
r
10 Day
-0.00233**
(0.00104)
-4.61e-05
(6.65e-05)
8.49e-05
(9.63e-05)
1.101***
(0.0356)
0.00278***
(0.000718)
3,864
0.199

(5)
r
30 Day
-0.00212***
(0.000766)
3.07e-06
(5.07e-05)
0.000101
(7.07e-05)
1.038***
(0.0269)
0.00339***
(0.000545)
7,356
0.170

Appendix A: Detailed Description of the TNIC Industry Classification


TNIC industry classifications are based on the foundation that firms in the same industry use
many of the same words to describe their products in the business description section of firm 10Ks. Using 10-K text offers many advantages including improved signal strength, the ability to
measure competition dynamically even when the product market changes rapidly, and the ability
to use a general intransitive network structure. These features have proven useful in other
settings including research on mergers, asset prices, payout policy, and firm organizational form.
They are particularly well suited for use in IT industry, which is known for rapid change.
In order to construct TNIC industries, Hoberg and Phillips (2010a, 2010b) web crawl all
available 10-Ks from the SECs Edgar website and link each 10-K to the Compustat database in
each year using the central index key (CIK) as the unique firm identifier. A link table from CIK
to Compustat gvkey is provided by the Wharton Research Data Service (WRDS) through the
SEC Analytics package. Once business description sections are parsed from each 10-K,
common words (those appearing in more than 25% of all 10-Ks in a given year) are discarded.
Additionally, any word that does not appear as a noun or proper noun is also discarded. The
typical firm uses 200 unique words, and any firm using fewer than 20 is omitted.
Words are then mapped to numerical vectors and firm pairwise cosine similarity scores
are computed for every pair of firms in each year (cosine similarities are a popular tool in
computational linguistics, see Sebastiani (2002) for example). In particular, each firm i's
vocabulary can be represented by a vector Pi, which has a length equal to the number of unique
words, with each element being populated by the number one if firm i uses the given word, and
zero if it does not. These vectors are then normalized to have unit length (which is done by
dividing all elements of the vector by the square root of the vectors dot product with itself). The
normalized vectors are denoted as Vi. Because all firms are thus represented by vectors of length
one in the same space, it follows that TNIC industries imply that firms have unique locations on
a high dimensional unit sphere, and that they move along the surface of the sphere as their
products are revised from year to year.
35

To identify peers in the TNIC network, the cosine similarity between pairs of firms i and
j are calculated as follows:
, = ( )

The full TNIC network is thus fully described by an NxN square matrix that is populated with
cosine similarities between all firms in each year, where N is the number of firms. Due to the
properties of cosine similarities, all entries are real numbers in the interval [0,1]. Because firms
update their 10-Ks annually, the network (the entire square matrix) is time-varying. In the
current article, we consider the TNIC-3 network, which is calibrated to be as granular as is the
three digit SIC code network (SIC-3). This is done by deeming the firm pairs having cosine
similarities that are among the highest 2.05% of similarities as being rivals, as SIC-3 implies the
same fraction of all firm pairs are rivals.
TNIC industries have many important features. First, tests in Hoberg and Phillips (2010b)
illustrate that they are more informative than other frequently used industry classifications such
as SIC or NAICS code. Second, the entire classification is dynamically updated every year
because firms file updated 10-Ks every year as required by regulation S-K. Third, TNIC
industries are part of a generalized intransitive network, providing much research flexibility to
consider customized sets of rivals for each firm. In contrast, classifications such as SIC codes are
less powerful, rarely updated, and are constrained to be transitive.
References
Hoberg, G., Phillips, G. 2010a. Product Market Synergies and Competition in Mergers and Acquisitions:
A Text-Based Analysis. Review of Financial Studies. 23(10) 3773-3811.
Hoberg, G., Phillips, G. 2010b. Real and Financial Industry Booms and Busts. The Journal of Finance.
65(1) 45-86.
Sebastiani, F. 2002. Machine learning in automated text categorization. ACM Computing Surveys. 34(1)
1-47.

36