15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
When Do Leaders Free-Ride? Business Experience
and Contributions to Collective Defense
Matthew Fuhrmann Texas A&M University

Abstract: The logic of free-riding expects that individuals will underinvest in public goods, but people often behave in ways
that are inconsistent with this prediction. Why do we observe variation in free-riding behavior? This study addresses this
question by examining contributions to an important international public good—collective defense in military alliances. It
develops a behavioral theory of free-riding in which the beliefs of world leaders are important for explaining investments
in public goods. The argument holds that leaders with business experience make smaller contributions to collective defense
because they are egoistic and more comfortable relying on a powerful ally for their defense. An analysis of defense expenditures
in 17 non-U.S. members of the North Atlantic Treaty Organization from 1952 to 2014 provides evidence consistent with
the theory. The findings suggest that leaders with business experience are more likely than other heads of government to act
as self-interested utility maximizers.

Verification Materials: The materials required to verify the procedures and analyses in this article are avail-
able on the American Journal of Political Science Dataverse within the Harvard Dataverse Network, at:
https://doi.org/10.7910/DVN/9ZZOTW.

T
he free-rider problem is one of the most widely alliances—particularly those backed by a nuclear-armed
discussed social dilemmas in economic theory.1 superpower, such as the North Atlantic Treaty Organi-
According to the free-riding hypothesis, goods zation (NATO)—encourage weaker alliance partners to
that serve common interests produce suboptimal outputs. free-ride (e.g., Olson and Zeckhauser 1966; Palmer 1990;
Everyone shares the benefits produced by public goods, Snyder 1997). Based on this line of thinking, deterrence is
even those who do not contribute, so the incentives to in- a public good: Countries benefit from the persistence of
vest in them are often weak. Free-riding frequently occurs peace regardless of how much they contribute. Alliance
in everyday life. For example, many individuals contribute members should therefore underinvest in collective de-
little to public broadcasting services since they can con- fense. Many world leaders appear to accept this view.
sume their preferred programs even if they pay nothing. Barack Obama, for instance, complained in 2016 about
This dilemma is pervasive in international politics as well, Britain’s relatively modest contribution to NATO: “Free
and how countries respond to it can lead “to momentous riders aggravate me. . . . You have to pay your fair share”
decisions affecting war and peace” (Ostrom 1998, 1). (quoted in Goldberg 2016). At the same time, scholars
This study examines free-riding in the context and policy makers continue to debate whether NATO al-
of military alliances. Many scholars have argued that lies free-ride on the United States (see Lanoszka 2015).

Matthew Fuhrmann is Professor of Political Science, Texas A&M University, 4348 TAMU, College Station, TX 77843
([email protected]).
I worked on this study as an Andrew Carnegie Fellow (2016–18) and a Visiting Associate Professor at Stanford University’s Center for
International Security and Cooperation. Previous versions of this article were presented in research workshops at Duke University, MIT,
Stanford University, the University of California at Merced, the University of California at Santa Barbara, and the University of Houston.
I thank all of the participants in these workshops as well as Robert Art, Kyle Beardsley, William Clark, Bridget Coggins, Courtenay Conrad,
Sarah Cormack-Patton, Mark Crescenzi, Allan Dafoe, Bryan Early, Peter Feaver, Benjamin Fordham, Erik Gartzke, Andrea Gilli, Stacey
Goddard, Hein Goemans, Richard Hermann, Michael Horowitz, Jacques Hymans, Nathan Jensen, Hyeran Jo, Sabrina Karim, Joshua
Kertzer, Brad LeVeck, Christine Lipsmeyer, Danielle Lupton, Nikolay Marinov, Eric Min, Sara Mitchell, Cliff Morgan, Neil Narang, Vipin
Narang, Roger Peterson, Pablo Pinto, Brian Rathbun, Emily Ritter, Richard Samuels, Gerald Schneider, Kenneth Schultz, Patrick Shea,
Brad Smith, Martin Steinwand, Robert Trager, Rachel Wellhausen, and Guy Whitten for helpful comments. Yewon Kwon helped me verify
some of the codings of business leaders.
1
Scholars sometimes refer to the free-riding dilemma as the public good problem (Olson 1965) or buck-passing (Christensen and Snyder
1990).
American Journal of Political Science, Vol. 64, No. 2, April 2020, Pp. 416–431

C 2020, Midwest Political Science Association DOI: 10.1111/ajps.12502

416
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WHEN DO LEADERS FREE-RIDE? 417

Instead of seeking to determine if free-riding occurs, tries. I exclude the United States because it is the alliance
I ask a different question: What explains variation in con- leader on whom other countries may free-ride.4 My re-
tributions to collective defense among NATO members?2 gression analysis controls for observable confounders and
Many governments seem to exploit their allies’ coopera- employs country and year fixed effects to minimize the
tion by contributing relatively little to deterrence. Yet oth- risk of omitted variable bias. I assess the viability of my
ers spend more on this public good than the free-riding argument relative to alternative explanations by design-
logic would expect. My research objective is to explain ing two placebo tests. These tests identify contexts in
this puzzling discrepancy. By framing the issue in this which the alternative explanations would expect to find
way, I conceive of consistency with predictions from eco- a relationship between business experience and spending
nomic theory as a variable, mirroring the approach taken but my theory would not: (1) nondefense spending in
by some other recent work in political science (Kertzer NATO countries and (2) defense spending in non-NATO
and Rathbun 2015; Rathbun, Kertzer, and Paradis states.
2017). In support of my theory, I find that former busi-
I develop a behavioral theory of free-riding.3 Most nesspersons spend less on defense than leaders without
research on social dilemmas in international rela- business experience. I also find that a background in eco-
tions emphasizes structural factors such as group size nomics or finance without executive-level business ex-
(Axelrod and Keohane 1985), national wealth (Olson and perience is not robustly associated with smaller defense
Zeckhauser 1966), or regime type (Digiuseppe and Poast expenditures, although there is some evidence that these
2018). My focus, by contrast, is on individual leaders. I leaders also free-ride to some degree. This disparity sug-
seek to explain contributions to collective defense within gests that feelings of power and self-efficacy—not just
NATO by focusing on one leader-specific factor: business displaying egoistic tendencies—are an important part of
experience. the story. The placebo tests show that business experience
Leaders with executive-level business experience is not reliably associated with changes in spending when
should make smaller defense expenditures for two main my theory would not expect it to be.
reasons. First, they are more likely than leaders from We have known for some time that everyday citi-
nonbusiness backgrounds to display egoistic tendencies. zens respond differently to social dilemmas. However,
However, high levels of egoism may not automatically lead very little research has examined leader-level variation
to free-riding. I deviate from the economic theory of al- in responses to international social dilemmas. This study
liances (Olson and Zeckhauser 1966), which assumes that advances knowledge about collective action problems in
deterrence is a pure public good, by suggesting that a pow- world politics by developing a leader-centric theory of
erful country such as the United States may not ultimately free-riding and applying it to the case of military alliances.
come to the aid of an ally who is attacked. This brings me In doing so, it speaks to a central question in political sci-
to the second reason: Businesspersons have high levels ence: To what extent do individual leaders matter? Jervis
of self-efficacy and feelings of power (see Bandura 1982; (2013, 154) captures the current view of leadership: “Most
Fast et al. 2012), giving them confidence that they can get IR scholars place great stress on the incentives and con-
what they want from their alliance partners. They there- straints posed by the environment . . . the study of leader-
fore worry less about being excluded from the deterrence ship has fallen out of favor in political science.” According
benefits of an alliance or facing penalties if they make rela- to most scholarship in the field, the election of Donald
tively small expenditures on defense. Former businessper- Trump as U.S. president instead of another contender is
sons therefore see both the motive and the opportunity to relatively inconsequential for world politics. Many schol-
free-ride on powerful allies. Other types of leaders—for ars are skeptical that individual leaders make a difference
example, those with advanced training in economics— in world politics because the institutional environment in
may have the will to free-ride but not necessarily the which a leader operates shapes his or her decision making
confidence in their ability to manipulate a foreign (see Waltz 1959).
counterpart. My analysis suggests that this view on leadership is too
To test this theory, I analyze defense spending in 17 skeptical. Structural factors are undoubtedly important.
non-U.S. NATO members from 1952 to 2014, utilizing Indeed, this study will provide evidence that the effect
new data on the backgrounds of leaders in these coun- of business experience on defense spending is strongest
when leaders face relatively few political constraints. Yet
2
I use the terms collective defense and deterrence interchangeably.
3 4
See Hafner-Burton et al. (2017) for an overview of the new “be- However, as shown in the supporting information (SI), the find-
havioral revolution” in international relations. ings are similar if I include the United States.
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
418 MATTHEW FUHRMANN

it is a mistake to assume that they are the only thing that Public Goods and the Economic
matters (see also Horowitz and Fuhrmann 2018). My Theory of Alliances
analysis suggests that incorporating leader-based factors
into our analyses can provide a richer understanding of
Public goods are central to the study of economics and
international relations. I am hardly the first scholar to
political science. Goods or services are said to be “pub-
make this claim. A recent wave of research highlights
lic” if they serve the common interest (see Olson 1965).
the importance of leadership turnover and culpability
A collective good has one or both of the following char-
(Croco 2011; Wolford 2007) as well as the means of
acteristics: nonexcludability and nonrival consumption
entry and exit from office (Chiozza and Goemans 2011)
(Olson and Zeckhauser 1966, 267). Free-riding emerges
for understanding world politics. In addition, there is
when people underinvest in public goods, realizing that
growing evidence that leaders’ professional backgrounds
they can rely on others to do the heavy lifting while still
(Colgan 2013; Fuhrmann and Horowitz 2015; Horowitz,
reaping the benefits.
Stam, and Ellis 2015; Nelson 2014), education (Gift
Military alliances are widely believed to generate in-
and Krcmaric 2017), and beliefs (Rathbun, Kertzer, and
ternational public goods. The so-called “economic theory
Paradis 2017; Saunders 2011; Yarhi-Milo, Kertzer, and
of alliances,” first articulated by Olson and Zeckhauser
Renshon 2018) affect their behavior. This article pushes
(1966), holds that alliances display both features of a
the research agenda on leaders forward by suggesting
collective good. NATO members, they argue, cannot be
that heads of government, like all people, vary in their
excluded from the benefits of sustained peace at a reason-
cooperative preferences—and that this variation helps
able cost. Indeed, the NATO charter explicitly addresses
explain contributions to public goods, particularly when
the issue of excludability, declaring that “an armed attack
the executive faces few domestic political constraints.
against one or more . . . [members] shall be considered
Political commentators and politicians often draw a
an attack against them all.” Moreover, adding countries
link between leadership and business experience. Mitt
to the alliance does not decrease the level of peace avail-
Romney, for instance, suggested in 2012 that experience
able to other NATO members. There is therefore “little
in business should be constitutionally required to serve
doubt,” according to Olson and Zeckhauser (1966, 272),
as president. Four years later, Donald Trump played up
“that above all alliances produce public goods.” As a re-
his business career on the campaign trail: “Our country
sult, states must share the benefits of deterrence while
needs a truly great leader,” he said when announcing
shouldering the full costs of additional investments in
his candidacy. “We need a leader that wrote The Art of
peace. Most countries, then, will cease contributing to
the Deal.”5 This study provides concrete evidence that
the common defense before the alliance reaches the ideal
business leaders are unique. They are more likely than
level of defense-related output.
leaders without business experience to behave in a way
Subsequent studies challenged the notion that mili-
that is consistent with the predictions of economic theory,
tary alliances produce purely public goods. According to
which typically emphasize the maximization of one’s own
the joint product model, alliances generate three main
material self-interest.
outputs: deterrence, damage limitation, and country-
This article proceeds by discussing the economic the-
specific objectives (Sandler 1993, 459). These goods vary
ory of alliances. I then explain why individual-level factors
in their level of publicness. For example, a country may
are important for understanding variation in contribu-
have increased defense spending in the aftermath of World
tions to public goods. The subsequent sections present my
War II to counter a rebellion in one of its colonies. As San-
argument that business experience can predict a leader’s
dler (1993, 458) notes, this may provide public benefits
free-riding behavior, describe the data set used for the
to the colonial power’s citizens, but these gains would not
empirical analysis, and address possible barriers to infer-
“spill over to the nation’s ally.” It is also conceivable that
ence. I then present the findings from the two-way fixed
a state would spend more on defense to increase its social
effects regression analysis and the two placebo tests. The
status or to appease domestic political actors such as the
final section discusses the role of institutional constraints
military. Expenditures meant to serve country-specific
and concludes.
aims such as these produce mostly private benefits.
Given that alliances produce at least some private
benefits, a state that is bent on free-riding on a powerful
ally may still make considerable investments in defense.
5
Trump cowrote this 1987 book on business with Tony Schwartz. However, as long as some of the alliance’s outputs are
The text of Trump’s address is available at http://time.com/ nonexcludable and nonrival, there will be opportunities
3923128/donald-trump-announcement-speech/.
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WHEN DO LEADERS FREE-RIDE? 419

for free-riding, and we may observe this in national de- do not—despite being presented with the same scenario
fense expenditures. Of course, the degree of free-riding (e.g., Kurzban and Houser 2001). This has led scholars to
is likely to increase as an alliance produces a greater per- advocate for more research on the role that individual-
centage of public goods. level considerations play in shaping cooperation. Rap-
Some have argued that even the deterrence benefits port and Suleiman (1993, 193) write, for example, that
of alliances, which are most likely to be public, are ex- researchers should focus on “identifying personality and
cludable in the NATO context. This is because the United attitude variables that allow the classification of subjects
States may have strong incentives to refrain from fight- into different ‘types’ whose decision behavior in social
ing, especially once the Soviet Union developed long- dilemmas may be described by alternative models.”
range nuclear-capable missiles that could reach American Prior research suggests that there are two ideal types
cities (Gates and Terasawa 2003; Schelling 1966). French of people: proselfs and prosocials (e.g., Bogaert, Boone,
leader Charles de Gaulle famously wondered, for exam- and Declerck 2008; Fehr and Fischbacher 2002).6 Proselfs
ple, whether U.S. leaders would sacrifice New York to are rational egoists who care only about maximizing their
save Paris. This suggests that U.S. intervention following own gains. These individuals, not surprisingly, often act
a Soviet attack on a NATO ally in Europe is hardly au- in accordance with the predictions of standard economic
tomatic, making deterrence at least partially excludable. models. Prosocials, by contrast, value the material well-
There are good reasons to question an ally’s reliability: being of others, in addition to their own payoffs (Fehr and
Studies show that alliance commitments are violated in Fischbacher 2002, C2). Reciprocity and equality are two
war between 25% and 50% of the time (Berkemeier and key social preferences. Individuals who value reciprocity
Fuhrmann 2018; Leeds, Long, and Mitchell 2000). To the increase their contributions to public goods when others
extent that a NATO ally doubts the dependability of the in the group do so, while seeking to reduce private gains
U.S. security guarantee, free-riding may be a risky strat- accrued by free riders. Inequality aversion causes people
egy (see Digiuseppe and Poast 2018; Goldstein 1995). By to value burden sharing when it comes to the provision
contrast, leaders who have confidence in American com- of public goods (Fehr and Schmidt 1999). Those who
mitments may be more comfortable free-riding. As we display one or both of these social preferences are some-
will see, this point is important for understanding why times called “conditional cooperators” (e.g., Fischbacher,
former businesspersons are particularly likely to make Gachter, and Fehr 2001).
defense cuts. Kertzer and Rathbun (2015) present evidence that
social preferences are important in world politics. Their
experimental analysis shows that prosocials make more
evenhanded offers in bargaining games because they are
Individual-Level Variation more committed to fairness.7 Some constructivist-based
in Free-Riding scholarship also appreciates the significance of social pref-
erences for collective action. Most notably, Wendt (1994)
Prior studies support the notion that the nonsuperpowers argues that states’ social identities influence whether they
in NATO free-ride on the United States to some degree, can overcome collective action problems. If countries pos-
while pointing out that there is variation in free-riding itively identify with others in the group, he contends, they
behavior (e.g., Gates and Terasawa 2003; Oneal 1990; will be less likely to free-ride. When these positive feel-
Palmer 1990; Russett 1970; Sandler and Forbes 1980). ings are absent, however, “interests will be defined with-
There is disagreement about why this variation exists, but out regard to the other—who will instead be viewed as
arguments in this vein have something in common: They an object to be manipulated for gratification of the self”
focus on structural forces, like geography or a country’s (Wendt 1994, 386).
size, and downplay individual-level considerations. These The degree to which leaders value unselfish mo-
arguments assume that leaders will behave similarly when tives is important in an alliance context. I am not the
they are faced with identical economic, strategic, techno- first scholar to raise this possibility. Oneal (1990, 386)
logical, and political conditions. However, leaders have acknowledges that, in theory, weaker NATO allies may
distinct preferences for cooperation. have contributed more to the common defense over time
Research in economics and psychology has found that
individual-level factors are important for understand- 6
For other recent work in political science that uses this framework,
ing responses to social dilemmas. In particular, exper- see Kertzer and Rathbun (2015) and Rathbun, Kertzer, and Paradis
imental evidence shows that some people free-ride in (2017).
7
a manner consistent with economic theory but others See also Rathbun, Kertzer, and Paradis (2017).
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
420 MATTHEW FUHRMANN

because of changes in how much they value their material between themselves and one other person, business
well-being. But he quickly dismisses this conjecture: “It students tend to make one-sided offers in their favor,
is safe to assume,” he argues, “that most countries most whereas others are more likely to divide the money
of the time ‘satisfice’ egoistic objectives at a high level of evenly.
aspiration” (386). However, in light of considerable ev- Why might businesspersons be more selfish than
idence that egoism varies across actors, this may not be nonbusiness types? There are two main mechanisms (see
such a safe assumption. Gandal et al. 2005, 1233), which may be mutually rein-
Some leaders view alliance contracts largely through forcing rather than competing.
the lens of materialism. Others, by contrast, place a higher First, businesspersons are socialized to value their
value on nonmaterial factors such as equality and reci- own financial self-interest. Maximizing profits drives
procity, in addition to material considerations. To be sure, their professional success. These individuals therefore
when speaking about contributions to NATO, govern- have incentives to think about their firm’s material self-
ment officials outside the United States frequently in- interest. Business leaders may, of course, donate money to
voke the concept of “fair share.” British Defense Secre- charity or engage in other altruistic behaviors, but their
tary Michael Fallon said in December 2016, for instance, ultimate goal is profit maximization. This attitude may
“NATO is the world’s most successful military alliance and seep into other aspects of a businessperson’s life, causing
it is as good for the U.S. as it is for Europe. . . . The U.S. him or her to make self-interested choices in other con-
taxpayer can’t be expected to shoulder a disproportionate texts. As Ang (2000, 49) puts it, “Individuals preoccupied
burden” (quoted in MacAskill 2016). It is unfair, based on with wealth tend to be selfish and pursue individual as op-
this line of thinking, for European states to free-ride on posed to community goals.” Businesspersons, therefore,
the United States given that they benefit from the alliance may underinvest in public goods because doing so max-
as much as Washington. imizes their own material well-being. Based on this line
Identifying the degree to which a leader is a prosocial of thinking, businesspersons learn to behave as egoists as
type can help us better understand contributions to in- a result of their professional experiences.
ternational public goods. However, this is only part of the A second mechanism is based on self-selection rather
story. Allies may be excluded from the benefits of deter- than socialization. In this view, certain people are predis-
rence, so a leader’s tendency to free-ride also depends on posed toward behaving as utility maximizers, and those
his or her confidence in the alliance leader. Just as levels people are more likely to seek careers in business. Individ-
of egoism can vary across leaders, so too can perceptions uals who place a premium on fairness and social equality,
of the level of publicness in deterrence. by contrast, may pursue nonbusiness careers that are more
closely linked with those values.
Either way, business leaders think about collective
defense differently than their nonbusiness counterparts.
Business Experience and Because they are generally more selfish and less concerned
Contributions to Deterrence about fairness, these leaders view military alliances pri-
marily through the lens of their country’s own mate-
Experimental evidence indicates that individuals with rial self-interest. Consider, for example, U.S. president
business backgrounds are more likely to display proself Donald Trump’s views on military alliances. He seems to
tendencies. People with “entrepreneurial talent” tend to view alliance treaties as business contracts: The United
behave more selfishly than others, maximizing private States provides protection to its allies, and it should ben-
returns even at the expense of reducing public returns efit monetarily from doing so. If allies do not pay, the
(Baumol 1990; Murphy, Shleifer, and Vishny 1991). contract should be null and void, according to Trump’s
Consistent with this argument, Weitzel et al. (2010) view. As he put it in 1987, “Our world protection is worth
and Urbig et al. (2012) find that people who believe hundreds of billions of dollars to these countries and their
they are better at business than others are less likely to stake in their protection is far greater than ours” (quoted
display benevolence. On the basis of this evidence, Urbig in Wright 2016).
et al. (2012, 391) conclude that “selfishness deserves There is an important assumption embedded in this
attention as one of entrepreneurs’ distinctive personality logic: That leaders who are proself types will act egoisticly
characteristics.” Along similar lines, Kahneman, Knetsch, on behalf of the governments that they represent, not just
and Thaler (1986) show that business majors play ul- themselves. In advancing this claim, I am departing from
timatum bargaining games differently than psychology most laboratory experiments, which force people to think
majors: When deciding how to divide a sum of money about their own material well-being—not the material
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WHEN DO LEADERS FREE-RIDE? 421

well-being of an organization they might represent. This their alliance relationships and get support from allies
is reasonable, in my view, because national leaders are when they need it.8
charged with advancing the interests of the state. They People with business experience have higher levels
think about their government’s bottom line, not just their of self-efficacy than their nonbusiness counterparts (e.g.,
own pocketbooks (see also Rathbun, Kertzer, and Paradis Markman, Balkin, and Baron 2002). Former business ex-
2017, s41). Proself leaders, then, should be more likely to ecutives also tend to display feelings of power. For exam-
free-ride on other countries when doing so frees up more ple, as he was renegotiating Britain’s relationship with the
resources for their own government. European Union in January 2014, prime minister David
The experience of Ludwig Erhard, the chancel- Cameron, who served as vice president of communica-
lor of West Germany from 1963 to 1966, illustrates tions at the media company Carlton Communications
the connection between business experience and free- prior to entering office, confidently declared, “I believe
riding. Erhard worked for his family’s business and later I will get what I want” (quoted in Mason 2014). Lead-
served on the board of directors for a consumer re- ers from business backgrounds, then, should generally
search company. According to one of his biographers, have greater confidence that they can control their own
Erhard learned that “firms existed to make money” destinies in world politics. They are prone to believe—
(Mierzejewski 2004, 7). His emphasis on materialism in rightly or wrongly—that they can persuade allies to come
the context of business may have shaped his overall sys- to their aid in the event of a crisis, making free-riding a
tem of beliefs and values. Erhard reportedly “thought in more attractive strategy.
economic terms” and condemned the concepts of “so- Free-riding can invite political blowback and even
cial” and “just” (Mierzejewski 2004, 194). His apparent lead to the termination of an alliance. As Robert Gates,
proself inclination led to the expected behavior in the area then serving as Obama’s secretary of defense, put it in
of defense spending: Erhard generally reduced Germany’s 2011, “The blunt reality is that there will be dwindling
contributions to collective defense while he was in office. appetite and patience in the U.S. Congress . . . to ex-
However, having a proself orientation is not suffi- pend increasingly precious funds on behalf of nations
cient to generate free-riding in military alliances. As dis- that are apparently unwilling to devote the necessary
cussed previously, the deterrence benefits provided by an resources . . . to be serious and capable partners in their
alliance may be excludable. Beliefs about excludability own defense” (quoted in Shanker 2011). Leaders with
can vary across leaders. De Gaulle doubted the credibil- higher levels of self-efficacy and feelings of power believe
ity of American security guarantees. Erhard, by contrast, that they are insulated from an ally’s threat to walk away
seemed perfectly satisfied when U.S. president Lyndon from a treaty if they underinvest in collective defense.
Johnson affirmed the American defense commitments to There is some evidence that former business executives
Europe. Indeed, the West Germany leader once told U.S. have escaped political blowback that might have emerged
secretary of state Dean Rusk that West Germany would as a result of free-riding. Canadian prime minister Brian
be “lost without the United States” (Mierzejewski 2004, Mulroney, who was an executive at the Iron Ore Com-
191). Those who believe that they can rely on powerful al- pany of Canada before entering office in 1984, developed
lies for their defense will be more comfortable free-riding. a close relationship with Ronald Reagan while simulta-
Leaders who fear that they will be excluded from the de- neously underinvesting in defense. At one point, a U.S.
terrence benefits of an alliance, on the other hand, will State Department official suggested to Reagan that he
be much more reluctant to outsource their security to must pressure Mulroney to spend more. Reagan report-
an ally. edly responded by saying, “I want you to understand one
A leaders’ beliefs about whether they can get their way thing. I don’t talk to Brian Mulroney this way” (quoted
in international relations shape the degree to which they in Martin 2017). On this occasion, Mulroney seemed to
worry about being abandoned by an ally. People with high get what he wanted from Reagan. The theory does not re-
self-efficacy believe that they have an ability “to produce quire success, however, only that business leaders believe
and to regulate events in their lives” (Bandura 1982, 122). that they can control their own fate.
Similarly, those who possess feelings of power—defined
by Fast et al. (2012, 250) as “asymmetric control over val-
ued outcomes”—are more optimistic, take greater risks,
and are insulated from concerns about the loss of status
(Anderson and Galinsky 2006; Renshon 2015). Leaders
with high self-efficacy and a sense of power, therefore, 8
I thank an anonymous reviewer for helpful suggestions on this
should have confidence that they can effectively manage part of the argument.
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
422 MATTHEW FUHRMANN

Observable Implications and nonbusiness leaders in terms of changes in defense

spending in the absence of a military alliance. Free-riding
The argument developed in the preceding sections leads is an especially risky strategy in this context since there is
to the following hypothesis: not a superpower that is bound by treaty to foot the bill
for collective defense. According to at least some of the
H1: Nonsuperpower alliance members headed by alternative explanations—particularly the logic of busi-
leaders with business experience expend fewer ness pacifism—business leaders would still spend less on
resources on defense than leaders without busi- defense than their nonbusiness counterparts without for-
ness experience. mal security assurances. This leads to the third and final
prediction:
Hypothesis 1 will allow me to determine whether for-
mer businesspersons are different from nonbusinessper- H3: In the absence of superpower security guaran-
sons in terms of spending on defense. It will not, however, tees, there is not a clear difference in defense
shed light on why this relationship emerges—if it does, in spending between leaders with business experi-
fact, exist. There may be other reasons we observe a nega- ence and their nonbusiness counterparts.
tive relationship between business experience and defense
expenditures. It could be the case, for instance, that busi-
nesspersons simply favor lower levels of public spending.
If true, defense cuts made by business leaders may reflect Data Set and Variables
a broader strategy to get spending under control, perhaps
in order to balance the budget. Drawing on the logic of The theory developed above could apply to any multilat-
business pacifism (Huntington 1957, 222–230), former eral alliance, particularly those that include at least one
businesspersons may also have a general aversion to the superpower. I focus on one alliance: NATO. I construct a
military, preferring instead to focus on economic growth. panel data set that includes information on 17 non-U.S.
To assess these counterarguments, I design two NATO members from 1952 to 2014. The unit of observa-
placebo tests. These tests examine contexts where the al- tion is the leader-year.9 Incentives to free-ride may vary
ternative explanations would expect to find a relationship across countries and over time. After analyzing the full
between business experience and spending but my theory sample, I will focus separately on the pre– and post–Cold
would not, due to the absence of free-riding incentives. War periods. I will also limit the analysis to the “big three”
Evidence of a nonrelationship in these contexts would en- European powers in NATO—France, Germany, and the
hance the plausibility of my argument while weakening United Kingdom—which are widely seen as key players
support for the alternative explanations. in the alliance.
My theory does not anticipate a link between busi-
ness experience and nondefense spending. Free-riding on
a superpower is generally not possible, as it is for defense, Dependent Variable: Defense Expenditures
when it comes to other types of spending. However, if
The theory expects that leaders with business experience
business leaders simply have preferences for smaller pub-
will contribute fewer resources to defense than nonbusi-
lic expenditures, they should also cut nondefense budgets
ness leaders. I measure contributions to collective defense
to some degree. A negative relationship between business
based on a state’s annual growth in defense spending. An
experience and nondefense spending, therefore, might
ally’s defense expenditures effectively capture its contri-
call the logic of my argument into question, while lending
butions to deterrence. As a 2002 report from the U.S.
credence to the view that business leaders simply dislike
Department of Defense stated, “Defense spending is the
public spending. This leads to a second prediction:
most important single indicator of allied responsibility
H2: Nonsuperpower alliance members headed by sharing efforts” (see also Oneal and Diehl 1994, 377). I
leaders with business experience spend the same focus on changes in defense spending instead of overall
as (or more than) leaders without business expe- levels to compare annual expenditures to a preexisting
rience on nondefense.
9
Defense spending in states that lack superpower pro- As a robustness test reported in the SI, I use the leader–term in
tection offers a second opportunity to distinguish my ar- office as the unit of observation. When more than one leader served
in the same calendar year, the data set includes the longest-serving
gument from alternative explanations. My theory would leader only. I identify leaders based on Horowitz, Stam, and Ellis
not expect to find a large difference between business (2015), with my own updating from 2005 to 2014.
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WHEN DO LEADERS FREE-RIDE? 423

baseline, thereby measuring a leader’s cooperative prefer- FIGURE 1 Mean Level of Changes in Defense
ences more directly.10 Spending by Leaders’ Business
The dependent variable (Change in Defense Expen- Experience
ditures) is the percentage change in total defense expen-
ditures from year t-1 to year t.11 The relevant data come

2.5
from the Stockholm International Peace Research Insti-

Percentage Change in Defense Expenditures

tute’s (2015) Military Expenditure Database.

2
Other studies use defense expenditures as a percent-
age of the gross domestic product (GDP) as the depen-

1.5
dent variable (e.g., Oneal and Diehl 1994; Whitten and
Williams 2011). This measure will change based on two
things: defense expenditures and GDP. But governments

1
do not fully control their country’s GDP, and it is not
obvious that a leader’s tendency to free-ride affects eco-

.5
nomic capacity. Because the GDP-based measure of de-
fense spending can move significantly due to things that

0
have little to do with my theory, total defense expenditures No Business Experience Business Experience

offer a more direct test of the argument.12

Silvio Berlusconi of Italy (founder of a holding company),

Independent Variable: Business Experience Paul Martin of Canada (corporate CEO), Tansu Ciller of
Turkey (bank president), and Paul Vanden Boeynants of
Business experience leads to a reduction in defense expen- Belgium (creator of a meat-processing firm). The SI pro-
ditures, according to my theory, because businesspersons vides a complete list of former businesspersons along with
have proself tendencies and feelings of power. In mea- brief descriptions of their jobs.
suring this concept, I focus on experiences that are most The variable Business Experience is coded 1 if the
likely to be associated with these attributes: Those at the head of government in year t-1 had executive-level busi-
executive level where the individual played a key role in ness experience prior to entering office and 0 if he or
running or managing a company. Nonexecutive business she did not. I lag this variable to account for the time
experience (e.g., working as a salesperson) is alone insuf- that passes between decisions on defense spending and
ficient to meet my threshold. I classify leaders as having actual disbursements; my use of a 1-year lag assumes that
business experience if they did any one of the following governments determine their budgetary allocations in the
prior to entering office: established or owned a company, previous calendar year.15
worked as a firm’s chief executive officer, served on a cor-
porate board of directors, or worked as a senior manager
or an executive. I exclude high-level involvement in com-
mercial activities that are carried out on a small scale, such Bivariate Comparison and Barriers
as owning a family farm or running a local bookstore.13 to Inference
To identify leaders with business experience, I re-
searched the backgrounds of 193 heads of government As a first-cut look at the data, Figure 1 shows the mean
in NATO countries during the period examined in this level of Change in Defense Expenditures based on Busi-
study. Thirty-six leaders (18.650% of those examined) ness Experience. As the figure shows, former businessper-
had pre-office business experience.14 Examples include sons make smaller annual increases in defense spending
(0.19%) than nonbusiness leaders (2.36%), on average.
10
For a similar argument, see Plumper and Neumayer (2015, 254). This evidence is consistent with Hypothesis 1, but these
11
Prior to calculating the annual growth rate, I measure all figures figures are not directly comparable since business expe-
in constant (2014) U.S. dollars. rience is not randomly assigned. I account for confound-
12
The SI reports the results with four alternate dependent variables. ing variables that may determine both a government’s
13
I do not classify lawyers who open their own practices as
businesspersons. 15
Other studies take this approach (e.g., Whitten and Williams
14
The SI reruns the empirical models with changes to how I code 2011). The SI reports results from a nonlagged measure of busi-
business leaders. ness experience.
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
424 MATTHEW FUHRMANN

changes in defense spending and whether a leader with (see the SI).19 Based on my coding scheme, it is possi-
business experience is in office. ble for leaders to have experience in economics or finance
First, wealthier countries may produce more busi- as well as business.
nesspersons, meaning that the likelihood of someone with The inclusion of these variables in a statistical model
business experience rising to power in those countries will help reduce the likelihood that an observed correla-
is higher. At the same time, according to the economic tion between business experience and defense expendi-
theory of alliances, larger countries should be less likely tures (as depicted in Figure 1) is an artifact of the process
to free-ride. I control for a country’s economic capac- by which former businesspersons come to power. How-
ity with a variable (Economic Capacity) that measures a ever, it is likely that unmeasured country-specific factors
state’s logged real GDP in year t-1.16 This variable is from shape both the probability of a businessperson coming to
version 9.0 of the Penn World Table (Feenstra, Inklaar, power and the degree of investment in defense. I therefore
and Timmer 2015). include country fixed effects in the statistical models esti-
Second, in periods of low economic growth, those re- mated below to account for state-specific, time-invariant
sponsible for choosing a head of government may prefer a confounders. I also include yearly dummy variables to
leader with business experience to get the economy “back account for the possibility that events in a given year in-
on track.” Moreover, growth increases defense spending fluence both changes in alliance investments and the pos-
by providing leaders with additional resources to devote sibility of having a leader with business experience. The
to security (e.g., Whitten and Williams 2011). I therefore use of two-way fixed effects reduces the risk of omitted
include the variable Economic Growth, which measures variable bias.
the annual percentage change in a country’s real GDP in Pooled time-series analysis creates some well-known
year t-1. methodological challenges. I use panel-corrected stan-
Third, many scholars have argued that conservative dard errors to address heteroskedasticity with a Prais-
governments favor the military to a greater extent and Winsten transformation to correct for panel-specific
spend more on defense than liberal governments (e.g., AR(1) autocorrelation.20 The models also include a lagged
Schultz 2001; Whitten and Williams 2011). One might dependent variable since annual changes in defense ex-
also expect that leaders with business experience are more penditures may depend in part on the spending trend in
likely to represent conservative governments. The variable the previous year.21 I conducted Im, Pesaran, and Shin
Ideology measures the political ideology of the govern- (2003) tests for panel unit roots and was able to reject the
ment in year t-1 based on a 5-point scale, with higher null hypothesis of nonstationarity.
values indicating greater left-wing dominance (Seki and
Williams 2014).
Fourth, states can be expected to invest more in de-
fense during times of war. At the same time, leaders with Findings: Multiple Regression
strong military credentials may be more likely to be se- Analysis
lected for office during wartime, meaning that conflict
could reduce the likelihood that a former businessperson Table 1 displays the results of the analysis of defense ex-
comes into power. The variable War indicates whether a penditures in 17 NATO countries.22 The evidence is con-
country experienced war in year t, based on data compiled sistent with Hypothesis 1: Business Experience is negative
by Reiter, Stam, and Horowitz (2016).17 and statistically significant in Model 1 (p = 0.008), which
In addition, as noted previously, I account for other
professional experiences related to business that could re- 19
Being a trade union leader does not count as experience in eco-
sult in fewer contributions to defense.18 Economics and nomics or finance without relevant expertise. Nor does serving as
Finance Experience indicates whether a leader is a for- a high level political appointee (e.g., minister of finance).
mer economist or worked in the financial sector. There 20
My core findings are broadly similar when I use panel-corrected
are 55 leaders with careers in economics or finance standard errors without the Prais-Winsten transformation (see the
SI).
21
Including a lagged dependent variable with fixed effects can be
16
I standardize GDP values in millions of 2014 U.S. dollars. problematic (Angriest and Pischke 2009, 245). The SI shows that
support for my central hypothesis remains when I exclude the
17
This variable is not lagged because I expect war to have immediate lagged dependent variable, as well as when I use a lagged dependent
effects on defense spending. variable model without fixed effects.
18 22
In the SI, I also examine whether pre-office careers in the military I replicated Models 1–3 with each country excluded from the
and law influence defense spending. estimation sample (see the SI).
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WHEN DO LEADERS FREE-RIDE? 425

TABLE 1 Regression Analysis of Defense Expenditures in NATO Countries

(1) (2) (3) (4)

Full Sample Cold War Post–Cold War “Big Three”
Change in Defense Spendingt−1 0.152∗∗∗ 0.195∗∗∗ −0.070 −0.234∗
(0.037) (0.029) (0.088) (0.100)
Business Experiencet−1 −1.241∗∗ −1.599∗ −1.876∗∗ −3.177∗
(0.470) (0.702) (0.720) (1.318)
Economics and Finance Experiencet−1 −0.651 −1.597∗∗∗ 1.004 −1.858
(0.433) (0.467) (0.646) (1.389)
Economic Capacity (ln)t−1 1.158 −0.784 4.582 −14.972∗∗
(2.212) (1.949) (5.365) (5.131)
Economic Growtht−1 0.197∗ 0.353∗∗∗ −0.111 0.483+
(0.092) (0.080) (0.136) (0.293)
War 1.610 2.318∗ 1.352 6.318∗∗
(1.237) (1.149) (1.504) (2.297)
Government Ideologyt−1 −0.200 −0.344 −0.166 0.092
(0.269) (0.313) (0.357) (0.589)
Constant 15.812 35.521 −60.359 219.933∗∗
(26.513) (23.672) (72.156) (67.001)
Country Fixed Effects Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes
Observations 853 448 405 186
R2 0.373 0.414 0.203 0.577
Note: Standard errors are in parentheses.
+
p < 0.10, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001

includes all relevant country-year observations. This find- and 3 indicates that business experience exerts a stronger
ing holds when I analyze the Cold War (Model 2) and effect on defense expenditures in the post–Cold War pe-
post–Cold War (Model 3) periods separately. Model 4 riod. The effect is largest in Model 4: In the three Euro-
similarly reveals a negative and statistically significant pean major powers, switching from a nonbusinessperson
(p < 0.016) result for Business Experience in the three to a leader with business experience reduces the expected
largest non-American NATO members. change in spending by 3.177 percentage points. This car-
How important is business experience in influencing ries practical significance; the effect of business experience
changes in defense spending? Based on Model 1, switching on defense expenditures is the largest in the three coun-
from a nonbusiness leader to one with business experience tries that arguably matter the most, other than the United
results in a 1.241 percentage point reduction in growth States, for NATO’s politico-military effectiveness.
in defense spending the next year, on average, holding all What about the role of experience in economics or
other factors constant. To get a more concrete sense of this finance? The effect of these professional experiences on
effect, consider a case from the data set: Georges Pompi- defense spending is similar to the one produced by busi-
dou of France. Pompidou, who previously worked as a ness experience during the Cold War period (Model 2).
director at an investment bank, rose to power in France In general, however, economics and finance experiences
in 1969. Model 1 predicts that he would increase defense are less strongly associated with defense cuts. The p-values
spending by 1.72% in 1970. However, in a hypotheti- for Economics and Finance Experience are below 0.1 in just
cal world where Pompidou lacked business experience one of the four models displayed in Table 1. On top of
and everything else remained the same, the model pre- this, the coefficient on Economics and Finance Experience
dicts that France’s defense spending would have increased is positive in the post-Cold War period (Model 3) and
by 2.74%—a full percentage point greater—in the same about half the size of Business Experience in the full sam-
year. ple (Model 1). These findings might seem surprising given
The coefficient on Business Experience becomes more the previously established link between a background in
negative in Models 2–4. A comparison between Models 2 economics and more selfish behavior in public goods
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
426 MATTHEW FUHRMANN

TABLE 2 Regression Analysis of Nondefense Expenditures in NATO Countries

(5) (6) (7) (8)

Full Sample Cold War Post–Cold War “Big Three”
Change in Nondefense Spendingt−1 0.221∗∗∗ 0.087 0.143∗ 0.308∗∗∗
(0.042) (0.059) (0.070) (0.089)
Business Experiencet−1 0.273 −0.272 0.0534 0.201
(0.210) (0.363) (0.343) (0.423)
Economics and Finance Experiencet−1 −0.390∗ −0.631∗ 0.140 −0.562∗
(0.163) (0.259) (0.286) (0.282)
Economic Capacity (ln)t−1 −0.337 1.284 −1.885 5.915∗
(0.779) (2.468) (1.463) (2.413)
Economic Growtht−1 0.105∗∗∗ 0.030 0.143∗∗∗ 0.077
(0.029) (0.059) (0.040) (0.070)
War −0.632∗ −0.811 −0.249 −0.166
(0.321) (0.797) (0.354) (0.624)
Government Ideologyt−1 0.036 0.167 −0.063 0.007
(0.104) (0.208) (0.139) (0.143)
Constant 7.295 −11.773 26.703 −76.256∗
(9.840) (30.818) (19.720) (31.767)
Country Fixed Effects Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes
Observations 703 298 405 149
R2 0.470 0.524 0.478 0.727
Note: Standard errors are in parentheses.
+
p < 0.10, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001

experiments. Recall, however, that free-riding depends Hypothesis 2.23 If former businesspersons have a
on a proself inclination plus confidence that a leader can general preference for less public spending, we would
get what she wants from her alliance partners. Business have expected to find a negative relationship between
leaders are likely to check both of these boxes, whereas Business Experience and Changes in Nondefense Spending.
people with experience in economics and finance who However, the coefficients on Business Experience are
do not become executives may be less confident in their positive in three of the four models, and the p-values
ability to control events in world politics. are above 0.1 in every case. It is therefore unlikely that
we observe a negative relationship between business
backgrounds and defense expenditures because former
businesspersons seek budgetary cuts across the board.
Placebo Tests Second, I analyze defense expenditures among all
non-NATO democracies for which I have complete data:
In the placebo tests, I look for evidence of a negative re- Austria, Cyprus, Finland, Ireland, Israel, Macedonia,
lationship between business experience and government Malta, Sweden, and Switzerland. These countries share
spending when incentives to free-ride do not exist or some similarities with NATO members, but they are not
are substantially reduced. If we observe the same pattern part of the alliance. Because it would be more difficult for
in these other contexts, my interpretation of the above these states to reliably depend on a superpower to pro-
results might be called into question. By contrast, the ab- vide for their security, free-riding is less of an option for
sence of a negative relationship would increase confidence
that former businesspersons reduce defense expenditures
because of free-riding rather than some other reason.
23
First, Table 2 shows that business experience I calculate the dependent variable for this analysis by subtracting
defense expenditures from World Bank (2019) data on all govern-
does not appear to generate reductions in nondefense ment spending. Before making this calculation, I standardize both
spending among NATO countries, consistent with values in millions of 2010 U.S. dollars.
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WHEN DO LEADERS FREE-RIDE? 427

TABLE 3 Regression Analysis of Defense Expenditures in Select Non-NATO Countries

(9) (10) (11) (12)

Full Sample Cold War Post–Cold War “Big Three”
Change in Defense Spendingt−1 −0.101 −0.284∗∗∗ −0.194 −0.091
(0.117) (0.067) (0.216) (0.110)
Business Experiencet−1 −2.697 −2.604 −3.547 −1.699
(3.631) (2.128) (4.630) (1.632)
Economics and Finance Experiencet−1 −1.371 −3.032 0.277 −1.660
(3.922) (2.757) (4.654) (1.932)
Economic Capacity (ln)t−1 −7.387∗ −4.600 3.800 6.456
(3.147) (11.78) (11.761) (6.001)
Economic Growtht−1 −0.313 −0.377 −1.120+ −0.046
(0.413) (0.261) (0.669) (0.234)
Military Disputes 5.781∗∗∗ −0.327 −1.567 0.728
(1.364) (1.531) (4.029) (1.442)
Government Ideologyt−1 −5.006+ 3.481 −2.981 1.798
(2.805) (2.269) (3.629) (1.138)
Constant 97.726∗∗ 42.153 −19.219 −80.590
(29.825) (114.576) (123.213) (68.583)
Country Fixed Effects Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes
Observations 314 120 167 149
R2 0.312 0.967 0.262 0.538
Note: Standard errors are in parentheses.
+
p < 0.10, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001

them.24 The theory expects, therefore, that there should evidence we would expect to see if there were strong in-
not be a strong negative association between business centives for former businesspersons to reduce defense
experience and defense expenditures among these non- expenditures that are unrelated to free-riding.
NATO countries (Hypothesis 3). The evidence is broadly
consistent with this expectation (see Table 3).25 There is
much more uncertainty about the relationship between
Business Experience and Change in Defense Expenditures Discussion and Conclusion
in this sample, compared to what we saw in NATO coun-
tries.26 The p-values are relatively large, ranging from This article addressed a general question about invest-
0.221 (Model 10) to 0.458 (Model 9). I cannot rule out ments in public goods: What accounts for variation in
the possibility that there is a large negative effect among contributions to collective defense? Traditional explana-
these non-NATO states. Overall, though, this is not the tions for alliance-based free-riding focus on structural
variables such as a country’s economic capacity. By con-
24
trast, I developed a behavioral theory of free-riding in
Some of these countries have had non-NATO defense pacts
with nuclear-armed countries at various points since 1945 (see
which individual leaders are also important for under-
Leeds et al. 2002). I exclude these country-year observations from standing variation in contributions to deterrence. I focus
the analysis. on one particular type of leader: former businesspersons.
25
The conflict variable here measures the number of states My theory predicts that leaders with business experience
with whom a country experiences any dispute in a given year should make smaller contributions to defense than heads
(Gibler, Miller, and Little 2017), since only Cyprus and Israel of government without backgrounds in business. This
fought in wars. The “big three” countries in Model 12 are Austria,
Switzerland, and Sweden. The temporal scope is more limited in prediction rests on two main claims. First, former busi-
this analysis due to missing data. nesspersons are more likely to display egoistic tendencies
26
I exclude Israel from Model 10 because otherwise there are no and think in terms of their own government’s material
common periods for all countries. self-interest. Second, business leaders have higher levels
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
428 MATTHEW FUHRMANN

of self-efficacy and feelings of power, which gives them other factors undoubtedly shape the degree to which
greater confidence that they can get what they want from a leader’s preferences translate to outcomes, and under
their alliance partners. some conditions leaders might not matter at all. One
To assess the theory, I analyzed defense expenditures conditioning factor may be the level of executive auton-
in 17 non-U.S. NATO countries from 1952 to 2014 using omy. Jones and Olken (2005, 835) show that leadership
new data on the business experiences of world leaders. To matters for economic growth but add that “the effects
reduce the likelihood that the empirical model’s estimate of individual leaders are strongest in autocratic settings
of the relationship between business experience and where there are fewer constraints on a leader’s power.”
defense spending is an artifact of the process through We might similarly expect that the relationship between
which leaders come to power, I accounted for observable business experience and changes in defense spending is
confounding factors and employed country and year strongest when leaders have few institutional checks on
fixed effects. I carried out two placebo tests to address their authority.
the possibility that multiple explanations can account for The above findings are particularly surprising given
the same observed pattern. that most NATO members were democratic during the
The findings indicated that leaders with business ex- time period I analyze, meaning that the executives in
perience have smaller growth rates in defense spending, these states were institutionally constrained to a relatively
on average, than their nonbusiness counterparts. The high degree. Nevertheless, it seems likely that the do-
placebo tests showed that business experience is not asso- mestic institutional environment matters in this context.
ciated with less nondefense spending in NATO countries, Findings reported in the SI show that the link between
nor can we conclude that former businesspersons spend business experience and changes in defense spending is
less on defense in select non-NATO states when it is more especially strong in the small number of nondemocratic
difficult to free-ride on a superpower patron. When we country-year observations.27 At the same time, business
would not expect to find a relationship between business experience remains statistically and substantively signifi-
experience and spending based on the theory’s logic, we cant even after excluding these country-year observations
do not. This suggests that leaders with business experi- from the analysis, focusing on democratic country-year
ence spend less on defense because of a greater tendency observations only.
to free-ride—not because of the previously discussed al- Even within democracies, variation in the degree of
ternative explanations. executive autonomy can shape the influence of leader-
Economic models usually assume that people are specific factors. As an initial investigation into this issue,
egoistic. Free-riding is an attractive option, according to I use data on political constraints from Henisz (2002)
the economic theory of alliances (Olson and Zeckhauser to see whether the influence of business experience on
1966), but only if we assume that leaders have proself ori- defense spending weakens in institutionally constrained
entations and confidence in powerful allies to provide for settings. The results, which are also reported in the SI,
their security. Yet both of these things are variables rather show that it does. For relatively unconstrained leaders—
than constants. In alliance politics, just as in everyday life, those who are in power when the political constraints
some people are more egoistic than others and not all variable is below the sample mean—the effect of business
leaders view deterrence as a public good. I have identified experience on changes in defense spending is about two-
a particular type of leader—former businesspersons— and-a-half times as large as the effect when executives are
that is likely to behave in a way that is consistent with the more constrained. Among politically constrained lead-
classic economic theory of alliances. Nonbusiness lead- ers, the p-value for Business Experience increases to 0.190
ers, however, show little propensity to free-ride in the (95% confidence interval:−1.922 to 0.382). These find-
aggregate. This study has therefore identified a concrete ings are consistent with the view that leadership matters in
way in which business experience matters for interna- world politics but domestic constraints can attenuate the
tional relations and foreign policy. Future research might significance of leader-level factors on political outcomes.
build on this study by evaluating other ways that former Doing more to identify the structural forces that moder-
businesspersons shape world politics. ate the influence of leader-specific factors represents an
This study does not imply that leaders always mat- important avenue for future research on leadership (see
ter. My analysis shows that the mean changes in defense also Horowitz and Fuhrmann 2018, 2081).
spending are statistically different based on whether a
leader has business experience. Establishing this relation-
ship is an important first step for advancing knowledge 27
These results should be interpreted with extreme caution given
about leaders in the context of alliance free-riding. Yet the small number of observations.
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WHEN DO LEADERS FREE-RIDE? 429

References Gandal, Neil, Sonia Roccas, Lilach Sagiv, and Amy

Wrzesniewski. 2005. “Personal Value Priorities of
Economists.” Human Relations 58(10):1127–1252.
Anderson, Cameron, and Adam D. Galinsky. 2006. “Power, Op-
Gates, William, and Katsuaki Terasawa. 2003. “Reconsidering
timism, and Risk Taking.” European Journal of Social Psychol-
Publicness in Alliance Defense Expenditures: NATO Expan-
ogy 36(4):511–36.
sion and Burden Sharing.” Defense and Peace Economics
Ang, Swee Hoon. 2000. “The Power of Money: A Cross-Cultural 14(5):369–83.
Analysis of Business-Related Beliefs.” Journal of World Busi-
Gibler, Douglas M., Steven V. Miller, and Erin K. Little.
ness 35(1):43–60.
2017. “An Analysis of the Militarized Interstate Dispute
Angriest, Joshua D., and Jorn-Steffen Pischke. 2009. Mostly (MID) Dataset, 1816–2001.” International Studies Quarterly
Harmless Econometrics. Princeton, NJ: Princeton University 60(4):719–30.
Press.
Gift, Thomas, and Daniel Krcmaric. 2017. “Who Democratizes?
Axelrod, Robert, and Robert O. Keohane. 1985. “Achieving Western-Educated Leaders and Regime Transitions.” Journal
Cooperation under Anarchy: Strategies and Institutions.” of Conflict Resolution 61(3):671–701.
World Politics 38(1):226–54.
Goldberg, Jeffrey. 2016. “The Obama Doctrine.” The Atlantic.
Bandura, Albert. 1982. “Self-Efficacy Mechanism in Human https://www.theatlantic.com/magazine/archive/2016/04/
Agency.” American Psychologist 37(2):122–47. the–obama–doctrine/471525/.
Baumol, William. 1990. “Entrepreneurship: Productive, Un- Goldstein, Avery. 1995. “Discounting the Free Ride: Alliances
productive, and Destructive.” Journal of Political Economy and Security in the Postwar World.” International Organiza-
98:893–921. tion 49(1):39–71.
Berkemeier, Molly, and Matthew Fuhrmann. 2018. “Reassessing Hafner-Burton Emilie, M., Stephan Haggard, David A. Lake,
the Fulfillment of Alliance Commitments in War.” Research and David G. Victor. 2017. “The Behavioral Revolution
and Politics April–June, 1–5. and International Relations.” International Organization
Bogaert, Sandy, Christophe Boone, and Carolyn Declerck. 2008. 71(S1):S1–S31.
“Social Values Orientation and Cooperation in Social Dilem- Henisz, Witold J. 2002. “The Institutional Environment for In-
mas: A Review and Conceptual Model.” British Journal of frastructure Investment.” Industrial and Corporate Change
Social Psychology 47(3):453–80. 11(2):355–89.
Chiozza, Giacomo, and Hein E. Goemans. 2011. Leaders and In- Horowitz, Michael C., and Matthew Fuhrmann. 2018. “Study-
ternational Conflict. New York: Cambridge University Press. ing Leaders and Military Conflict: Conceptual Frame-
Christensen, Thomas J., and Jack Snyder. 1990. “Chain Gangs work and Research Agenda.” Journal of Conflict Resolution
and Passed Bucks: Predicting Alliance Patterns in Multipo- 62(10):2072–86.
larity.” International Organization 44(2):137–68. Horowitz, Michael C., Allan C. Stam, and Cali M. Ellis. 2015.
Colgan, Jeff D. 2013. “Domestic Revolutionary Leaders and Why Leaders Fight. New York: Cambridge University Press.
International Conflict.” World Politics 65(4):656–90. Huntington, Samuel P. 1957. The Soldier and the State: The
Croco, Sarah. 2011. “The Decider’s Dilemma: Leader Culpa- Theory and Politics of Civil-Military Relations. Cambridge,
bility, Domestic Politics, and War Termination.” American MA: Harvard University Press.
Political Science Review 105(3):457–77. Im, Kyung So, M. Hashem Pesaran, and Yongcheol Shin. 2003.
Digiuseppe, Matthew, and Paul Poast. 2018. “Arms ver- “Testing for Unit Roots in Heterogeneous Panels.” Journal of
sus Democratic Allies.” British Journal of Political Science Econometrics 115(1):53–74.
48(4):981–1003. Jervis, Robert. 2013. “Do Leaders Matter and How Would We
Fast, Nathanael J., Niro Sivanathan, Nicole D. Mayer, and Adam Know.” Security Studies 22(2):153–79.
D. Galinsky. 2012. “Power and Overconfident Decision- Jones, Benjamin F., and Benjamin A. Olken. 2005. “Do Leaders
Making.” Organizational Behavior and Human Decision Pro- Matter? National Leadership and Growth since World War
cesses 117(2):249–60. II.” Quarterly Journal of Economics 120(3):835–64.
Feenstra, Robert C., Robert Inklaar, and Marcel P. Timmer. Kahneman, Daniel, Jack Knetsch, and Richard Thaler. 1986.
2015. “The Next Generation of the Penn World Table.” Amer- “Fairness and the Assumptions of Economics.” Journal of
ican Economic Review 105(10):3150–82. Business 59(2):S285–S300.
Fehr, Ernst, and Urs Fischbacher. 2002. “Why Social Preferences Kertzer, Joshua D., and Brian C. Rathbun. 2015. “Fair Is Fair:
Matter—The Impact of Non-Selfish Motives in Competi- Social Preferences and Reciprocity in International Politics.”
tion.” The Economic Journal 112(478):C1–C33. World Politics 67(4):613–55.
Fehr, Ernst, and Klaus M. Schmidt. 1999. “A Theory of Fair- Kurzban, Robert, and Daniel Houser. 2001. “Individual Differ-
ness, Competition, and Cooperation.” Quarterly Journal of ences in Cooperation in a Circular Public Goods Game.”
Economics 114(3):817–68. European Journal of Personality 15:S37–S52.
Fischbacher, Urs, Simon Gachter, and Ernest Fehr. 2001. “Are Lanoszka, Alexander. 2015. “Do Allies Really Free Ride?” Sur-
People Conditionally Cooperative? Evidence from a Public vival 57(3):133–52.
Goods Experiment.” Economics Letters 71(3):397–404.
Leeds, Brett Ashley, Andrew G. Long, and Sara McLaughlin
Fuhrmann, Matthew, and Michael C. Horowitz. 2015. “When Mitchell. 2000. “Reevaluating Alliance Reliability: Specific
Leaders Matter: Rebel Experience and Nuclear Prolifera- Threats, Specific Promises.” Journal of Conflict Resolution
tion.” Journal of Politics 77(1):72–87. 44(5):686–99.
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
430 MATTHEW FUHRMANN

Leeds, Brett Ashley, Jeffrey M. Ritter, Sara McLaughlin Mitchell, Reiter, Dan, Allan Stam, and Michael C. Horowitz. 2016. “A
and Andrew G. Long. 2002. “Alliance Treaty Obliga- Deeper Look at Interstate War Data: Interstate War Data
tions and Provisions, 1815–1944.” International Interactions Version 1.1.” Research and Politics 1–3.
28(3):237–60. Renshon, Jonathan. 2015. “Losing Face and Sinking Costs:
MacAskill, Ewen. 2016. “UK Defense Secretary Tells U.S. Experimental Evidence on the Judgment of Political and
Only NATO Can Deter Russian Threat.” The Guardian Military Leaders.” International Organization 69(3):659–
(December 3). https://www.theguardian.com/politics/2016/ 95.
dec/03/uk-defence-secretary-warns-us-that-russia-is- Russett, Bruce M. 1970. What Price Vigilance? The Bur-
threat-only-nato-can-deter dens of National Defense. New Haven, CT: Yale University
Markman, Gideon D., David B. Balkin, and Robert A. Baron. Press.
2002. “Inventors and New Venture Formation: The Ef- Sandler, Todd. 1993. “The Economic Theory of Alliances:
fects of General Self-Efficacy and Regretful Thinking.” En- A Survey.” Journal of Conflict Resolution 37(3):446–
trepreneurship Theory and Practice 27(2):149–65. 83.
Martin, Lawrence. 2017. “To Rein In Trump, Canada Sandler, Todd, and John F. Forbes. 1980. “Burden Sharing, Strat-
Needs Brian Mulroney.” The Globe and Mail (April egy and the Design of NATO.” Economic Inquiry 18(3):425–
7). https://www.theglobeandmail.com/opinion/to-rein-in- 44.
trump-canada-needs-brian-mulroney/article33315006/
Saunders, Elizabeth N. 2011. Leaders at War: How Presidents
Mason, Rowena. 2014. “Cameron to Push for Cap on European Shape Military Interventions. Ithaca, NY: Cornell University
Migrants in UK Negotiations with EU.” The Guardian (Jan- Press.
uary 5). https://www.theguardian.com/uk-news/2014/jan/
05/cameron-cap-european-migrants-uk-negotiations-eu Schelling, Thomas C. 1966. Arms and Influence. New Haven,
CT: Yale University Press.
Mierzejewski, Alfred C. 2004. Ludwig Erhard: A Biography.
Chapel Hill: University of North Carolina Press. Schultz, Kenneth A. 2001. Democracy and Coercive Diplomacy.
New York: Cambridge University Press.
Murphy, Kevin, Andrei Shleifer, and Robert W. Vishny. 1991.
“The Allocation of Talent: Implications for Growth.” Quar- Seki, Katsunori, and Laron K. Williams. 2014. “Updating the
terly Journal of Economics 106(2):503–30. Party Government Data Set.” Electoral Studies 34:270–79.
Nelson, Stephen C. 2014. “Playing Favorites: How Shared Beliefs Shanker, Thom. 2011. “Defense Secretary Warns NATO of
Shape the IMF’s Lending Decisions.” International Organi- ‘Dim’ Future.” New York Times (June 10). https://www.
zation 68(2):297–328. nytimes.com/2011/06/11/world/europe/11gates.html
Olson, Mancur. 1965. Logic of Collective Action: Public Goods and Snyder, Glenn H. 1997. Alliance Politics. Ithaca, NY: Cornell
the Theory of Groups. Cambridge, MA: Harvard University University Press.
Press. Stockholm International Peace Research Institute. 2015. Mili-
Olson, Mancur, and Richard Zeckhauser. 1966. “An Economic tary Expenditure Database. Stockholm: SIPRI. https://www.
Theory of Alliances.” Review of Economics and Statistics sipri.org/databases/milex
48(3):266–79. Urbig, Diemo, Ute Wentzel, Stephanie Rosenkranz, and Arjen
Oneal, John, and Paul F. Diehl. 1994. “The Theory of Collective van Witteloostuijn. 2012. “Exploiting Opportunities at All
Action and NATO Defense Burdens: New Empirical Tests.” Cost? Entrepreneurial Intent and Externalities.” Journal of
Political Research Quarterly 47(2):373–96. Economic Psychology 33:379–93.
Oneal, John R. 1990. “The Theory of Collective Action U.S. Department of Defense. 2002. Allied Contributions to the
and Burden Sharing in NATO.” International Organization Common Defense. Washington, DC: U.S. Department of De-
44(3):379–402. fense.
Ostrom, Elinor. 1998. “A Behavioral Approach to the Rational Waltz, Kenneth N. 1959. Man, the State, and War: A Theoretical
Choice Theory of Collective Action: Presidential Address, Analysis. New York: Columbia University Press.
American Political Science Association, 1997.” American Po- Weitzel, Utz, Diemo Urbig, Sameeksha Desai, Mark Saunders,
litical Science Review 92(1):1–22. and Zoltan Acs. 2010. “The Good, The Bad, and the Tal-
Palmer, Glenn. 1990. “Alliance Politics and Issue Areas: Deter- ented: Entrepreneurial Talent and Selfish Behavior.” Journal
minants of Defense Spending.” American Journal of Political of Economic Behavior and Organization 76:64–81.
Science 34(1):190–211. Wendt, Alexander. 1994. “Collective Identity Formation and
Plumper, Thomas, and Eric Neumayer. 2015. “Free-Riding in the International State.” American Political Science Review
Alliances: Testing an Old Theory with a New Method.” Con- 88(2):384–96.
flict Management and Peace Science 32(3):247–68. Whitten, Guy D., and Laron K. Williams. 2011. “Buttery Guns
Rapport, Amnon, and Ramzi Suleiman. 1993. “Incremental and Welfare Hawks: The Politics of Defense Spending in
Contribution in Step-Level Public Goods Games with Asym- Advanced Industrial Democracies.” American Journal of Po-
metric Players.” Organizational Behavior and Human Deci- litical Science 55(1):117–34.
sion Processes 55(2):171–94. Wolford, Scott. 2007. “The Turnover Trap: New Leaders, Rep-
Rathbun, Brian, Joshua Kertzer, and Mark Paradis. 2017. utation, and International Conflict.” American Journal of
“Homo Diplomaticus: Mixed-Method Evidence of Varia- Political Science 51(4):772–88.
tion in Strategic Rationality.” International Organization World Bank. 2019. World Development Indicators. Washington,
71(S1):S33–S60. DC: World Bank.
 15405907, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/ajps.12502 by University Of Manitoba, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WHEN DO LEADERS FREE-RIDE? 431

Wright, Thomas. 2016. “Trump’s 19th Century Foreign Policy.”

Politico (January 20). https://www.politico.com/magazine/
Supporting Information
story/2016/01/donald-trump-foreign-policy-213546
Yarhi-Milo, Keren, Joshua D. Kertzer, and Jonathan Renshon. Additional supporting information may be found online
2018. “Tying Hands, Sinking Costs, and Leader Attributes.” in the Supporting Information section at the end of the
Journal of Conflict Resolution 62(10):2150–79. article.
Appendix A: Additional Robustness Tests
Appendix B: Leaders with Business Experience
Appendix C: Leaders with Economics and Finance
Experience