Political leaders and military spending

Abstract

How are patterns of military spending shaped by political leaders who have substantive policy preferences and need the support of their constituents to remain in power? A formal model developed to address this question indicates leaders’ preferences, political vulnerability, and motivation and their domestic audiences’ preferences jointly influence equilibrium military spending. I find variation in patterns of military spending consistent with the model's implications between 1946 and 2010. My analyses demonstrate that leaders’ desire to remain in power and individual-level characteristics and domestic audiences’ preferences jointly shape policy outcomes and, accordingly, suggest studying the interactions among them can provide insights into a range of topics central to peace science.

Keywords

Domestic politics leaders military spending political institutions

Decisions regarding the allocation of scarce economic resources to military spending have consequences for national security (Sandler and Hartley, 1995), interstate conflict processes (Rider et al., 2011), alliance politics (Palmer, 1990b), economic growth (Oatley, 2015), social spending (Carter et al., 2021), and social welfare outcomes (Sexton et al., 2019). Accordingly, deciding how much of their state's resources should be dedicated to the military is one of the most important decisions a leader will make while in power. Leader-centric research explains patterns of military spending as a function of leaders’ desire to remain in power and how political institutions influence the ease with which they can be replaced (e.g. Bueno de Mesquita et al., 2003). Rather than being motivated exclusively by retaining office, though, politicians have substantive preferences over policy outcomes that influence their decisions (Strøm, 1990; Whitten and Williams, 2011, Greene and Licht, 2024; Jones, Mattiacci, and Nordstrom, 2024; Wittman, 1977). How, then, do leaders’ desire to remain in power, policy preferences, and the political institutions they operate under interact to influence patterns of military spending?

I develop a game-theoretic model to analyze the relationship between leaders and military spending when both a leader and their constituents hold preferences over how a government allocates resources. Equilibrium military spending is a function of a leader's and their key constituents’ policy preferences, a leader's motivation for implementing policy, and how sensitive a leader's survival is to their constituents’ desire to remove them from power. Leaders rationally allocate spending distributions other than those preferred by their constituents when they are not exclusively motivated by their political survival, but doing so increases the probability that they will be removed from power. The extent to which a leader motivated by both policy and survival concerns deviates from their key constituents’ preferred policy is a function of their preferences and political vulnerability. Accordingly, the patterns of government spending we observe in the vast majority of cases reflect a weighted balance of leaders’ concerns and not simply what will maximize their chances of remaining in power or their substantively preferred policy. Moreover, and running counter to many arguments about leaders and regime type, the model indicates that in the cases when a leader does implement the policy that best secures her political survival, her decision is unrelated to the ease with which her key constituents can remover her from power.

I use the formal model to motivate three sets of cross-sectional, time series analyses during the period between 1946 and 2010. The first focuses on variation in patterns of military spending as a function of the influence members of the military have on whether a leader remains in power. The second and third analyses consider how the effect of leaders’ preferences, proxied by whether they previously served in the military, on patterns of military expenditures is conditioned on their political vulnerability. My findings are consistent with expectations. All else equal, military spending is higher in regimes where the leader's key domestic audience is disproportionately made-up of members of the military than in those where a leader's key constituents primarily consist of civilians. Focusing on leaders’ political vulnerability and policy preferences, I find that military spending is more sensitive to whether a leader has a military background when it is relatively difficult for their domestic audience to remove them from power (non-democracies vs. democracies and personalist dictatorships vs. other non-democracies).

This paper makes multiple contributions to our understanding of how leaders influence policy outcomes central to contemporary peace science research. First, where leader-centric explanations of conflict processes typically focus on how leaders’ political survival or personal attributes influence their decisions in isolation from one another (Carter and Chiozza, 2018; Wolford, 2021), my analyses demonstrate that the interaction between leaders’ desire to remain in power and individual-level characteristics is key to understanding the decisions they make. While I focus on military spending, this interaction should also influence leaders’ decisions in other foreign policy domains and conflict processes more generally. Second, explicitly allowing a leader to vary in the extent to which she is motivated by policy or survival concerns provides a useful framework for considering how leaders’ decisions and foreign policies vary as a function of the dynamic constraints they face throughout their tenure (Clark and Nordstrom, 2005). Third, my analyses offer further evidence that political institutions, across and within regime type influence policy outcomes as a function of how easy they make it to remove a leader from power and who in society gets to determine whether a leader is removed from office (Croco and Weeks, 2016; Hyde and Saunders, 2020; Chiozza and Khalifa, 2024; Weeks, 2012). Fourth, the formal model is organized around a set of observations about the world that are not often jointly considered by scholars working in the peace science tradition. In doing so, it provides a blueprint for future research on foreign policy and conflict processes analyzing leaders’ decisions when they are motivated by a desire to remain in power, they and their key constituents have substantive policy preferences, and variation exists in how easily leaders’ constituents can remove them from office.

Leaders and government spending

Political leaders have two basic motivations for enacting policies. The first is to secure their political survival, which is done by maintaining the support of their key domestic audience (Bueno de Mesquita et al., 2003). Incumbents ensure the political support necessary to remain in power partially by implementing the policies preferred by their key constituents, a process known as policy responsiveness (Burstein, 2003). One of the central ways incumbents are responsive to their key constituents is by allocating scarce resources to their constituents’ preferred policies (Arena and Nicoletti, 2014; Smith and Bueno de Mesquita, 2011).

There is strong evidence that constituent preferences influence patterns of government spending. Soroka and Wlezien (2010, especially p. 131) find that changes in public preferences for government spending on defense, major social programs, welfare, health care, education, and the environment are associated with subsequent changes in government expenditures in the United States, UK, and Canada. Using cross-national data from over 40 developed democracies, Brooks and Manza (2007) demonstrate that variation in public preferences over social welfare influences variation in countries’ welfare state spending. Moving beyond aggregate public opinion, democracies tend to allocate resources in a manner consistent with the preferences of the parties in control of government. For example, democratic governments led by left-wing parties tend to allocate more money to social spending and foreign aid and less money to military spending than do democratic governments led by right-wing parties (Palmer, 1990a, Soroka and Wlezien, 2010; Tingley, 2010).

Much of the explicit theorizing and empirical work on policy responsiveness has focused on democracies. This is unsurprising as it reflects the intuition that democratic politicians are more responsive to their constituents’ preferences than are non-democratic politicians.¹ However, autocrats also secure their political survival through policy responsiveness. Key notes that even “the least democratic regime … needs the ungrudging support of substantial numbers of its people. If that support does not arise spontaneously, measures will be taken to stimulate by tactical concessions to public opinion” (1961, p. 3). Selectorate theory argues all leaders retain the political support necessary to remain in power by allocating the resources available to them between public goods and private benefits (Bueno de Mesquita et al., 1999, 2004). Gandhi and Przeworski (2006, 2007) argue that autocratic leaders maintain their hold on political power partially through policy concessions. Similarly, a dictator keeps his ruling coalition satisfied and retains office through power sharing agreements and policy concessions in Svolik's (2009) model of authoritarian politics. Weeks (2012, 2014) demonstrates that the initiation of interstate conflict by non-personalist dictators is influenced by their key constituents’ preferences regarding the use of military force.

Importantly, patterns of military spending across regime type suggest that democratic and non-democratic leaders alike allocate economic resources in a way that is consistent with their key constituents’ policy preferences. Contemporary democratic leaders rely on a larger portion of the general public to remain in power than non-democratic leaders, who largely retain office owing to the support of members of the military and civilian elite (Acemoglu and Robinson, 2006; Bueno de Mesquita et al., 2003; Dahl, 1971). As discussed in more detail below, members of the public generally prefer lower military spending and higher social spending than do members of the military and wealthy civilian elite (e.g. Fordham and Walker, 2005; Holsti, 1998; Page et al., 2013). This variation in the policy preferences of leaders’ key constituents across regime type is consistent with the finding that non-democracies generally spend more on their militaries than do democracies (Carter and Palmer, 2015; Fordham and Walker, 2005).

The preceding discussion makes clear that any leader interested in remaining in power has an incentive to enact her constituents’ preferred policies and patterns of military spending across regime type reflect this. What much of the research on this topic misses, though, is that leaders are not solely motivated by retaining office. Politicians also have substantive preferences that influence the policies they pursue (Fenno, 1978; Strøm, 1990; Whitten and Williams, 2011). Leaders’ second motivation for implementing policy, then, is to obtain substantive policy outcomes they prefer. Consistent with this, other papers in this special issue find leaders’ policy preferences influence patterns of foreign aid (Greene and Licht, 2024) and interstate conflict initiation (Jones, Mattiacci, and Nordstrom, 2024). Importantly, regardless of the content of their preferences, a leader will face a trade-off between enacting their preferred policy and the policy that best ensures that they remain in power if a leader and their key constituents have different policy preferences (Alesina and Cukierman, 1990).

Two factors significantly increase a leader's incentive to “shirk” and implement their preferred policies instead of policies that reflect their constituents’ preferences. The first is when an incumbent is not motivated by a desire to remain in power (Alesina and Cukierman, 1990). Consistent with this, shirking is greatest when an incumbent cannot be re-elected owing to term limits, lame-duck status, or retirement (Carey et al., 1998; Rothenberg and Sanders, 2000). Second, a leader is more likely to pursue her own substantive preferences when her continued survival is relatively insensitive to her constituents’ assessment of her performance. This is because a leader's key constituents must be increasingly unhappy with her performance in office to replace her with a political challenger as the difficulty of removal increases (Bueno de Mesquita et al., 2003; McGillivray and Smith, 2008). While scholars have traditionally focused on differences across regime type, recent research stresses that important variation in leader vulnerability also exists within democracies and non-democracies (Croco and Weeks, 2016; Geddes et al., 2018; Hyde and Saunders, 2020’ Chiozza and Khalifa, 2024; Weeks, 2012). This implies that leaders’ incentives to pursue their preferred policies or be responsive to their key constituents’ preferences vary both across and within regime type.

The next section develops a model of leaders and military spending motivated by the preceding discussion. The model primarily serves two purposes.² First, it is organizational in that it explicitly links a set of features and relationships relevant to patterns of military spending that are rarely jointly considered by peace science scholars. Second, the model is used to motivate a set of expectations about how leaders' preferences and political vulnerability and the preferences of leaders' key constituents influence military spending.

A model of leaders and military spending

I model the relationship between leaders and military spending as a game between a country's incumbent leader $(L_{i})$ and her key domestic audience ( $D_{i}$ ). $L_{i}$ 's strategy set is $x_{i} \in (0, 1)$ , where $x_{i}$ represents the proportion of the state's available economic resources dedicated to military expenditures.³ $L_{i}$ and $D_{i}$ hold single-peaked preference profiles over $x_{i}$ , with $x_{L}$ representing $L_{i}$ 's ideal $x_{i}$ and $x_{D}$ representing the ideal $x_{i}$ of the member of $D_{i}$ holding the median preference over $x_{i}$ . I assume leaders and their key constituents do not have the same ideal spending distributions ( $x_{L} \neq x_{D}$ ) in order to focus on substantively interesting cases.⁴ $L_{i}$ 's and $D_{i}$ 's respective valuations for a given $x_{i}$ are modeled with the quadratic loss functions $z (x_{i}) = - (x_{L} - x_{i})^{2}$ and $v (x_{i}) = - (x_{D} - x_{i})^{2}$ , implying $\underset{x}{a r g m a x} z (x) = x_{L}$ and $\underset{x}{a r g m a x} v (x) = x_{D}$ .

$D_{i}$ 's strategy set is $w_{i} \in {r, k}$ , where $r_{i}$ replaces $L_{i}$ and $k_{i}$ keeps $L_{i}$ . Replacing a leader brings a benefit $(b_{i})$ to a domestic audience. Following Debs and Goemans (2010), $b_{i}$ is modeled as an exogenous preference shock unknown at the beginning of the game that follows the uniform distribution $U [0, 1]$ . The cumulative distribution function of $b_{i}$ is defined as F. Substantively, $b_{i}$ can be thought of as $D_{i}$ 's relative preference for a political challenger over the incumbent leader.

The model assumes replacing an incumbent imposes a cost on the domestic audience. This cost is a function of two things. The first is $D$ 's valuation of the mix of spending the leader distributes ( $v (x_{i}$ )). Allocating an $x_{i}$ close to their domestic audience's ideal policy should make a leader more popular with their constituents than a distribution of government spending that substantially differs from what their domestic audience wants. Relatedly, it should make a domestic audience wary of replacing a leader who spends government resources in a manner they like with a challenger for whom there is more uncertainty about the policies they will pursue (Shepsle, 1972). The second factor is how vulnerable a leader's political survival is to their constituents’ desire to see them removed from power ( $s_{i} > 0$ , with increasing values representing greater vulnerability). Substantively, $s_{i}$ can be thought of as how relatively easy or difficult a given political regime makes it to oust the incumbent leader. The cost of removing a leader is defined as $c_{i} (v (x_{i}), s_{i}) = 1 + v (x_{i}) s_{i}$ . This parameterization captures the idea that the consequences of implementing policy for a leader's political survival are conditioned by the institutional barriers a domestic audience has to overcome to replace an incumbent with a challenger and accounts for $v (x_{i}) \in (- 1, 0)$ .⁵

I assume leaders receive a reward $R_{i} > 0$ if they are retained by $D_{i}$ and a payoff of $O_{i} = 0$ if they are ousted. R therefore can be thought of as the relative value of remaining in power over losing office in a given situation (Debs and Goemans, 2010). Importantly, holding the value of incumbency constant, leaders vary in the extent to which they care about their political survival and obtaining their preferred policy. This is incorporated into the model with the term $m_{i} \in [0, 1]$ , where zero represents a purely survival-motivated incumbent and one represents a purely policy-motivated incumbent. The probability that an incumbent remains in power is defined as p. The reduced-form utility functions for a leader and her key domestic audience are written as follows:
$U_{L_{i}} = z m_{i} + (1 - m_{i}) [(p) (R_{i}) + (1 - p) (O_{i})]$
(1)
$U_{D_{i}} = {\begin{matrix} v & if k_{i} \\ v + b_{i} - c_{i} & if r_{i} \end{matrix}$
(2)
The timing of the game is as follows. First, the incumbent leader $L_{i}$ distributes $x_{i}$ . Next, the uncertainty over $b_{i}$ is lifted. The game ends with $D_{i}$ deciding to replace ( $r_{i}$ ) or keep ( $k_{i}$ ) $L_{i}$ . The model thus captures a scenario in which a leader implements a policy that will influence her continued political survival without being able to perfectly anticipate the environment in which she will have to compete with a rival for office.

Equilibrium behavior

The game is solved for pure strategy subgame perfect equilibria using backwards induction. I informally describe equilibrium behavior here and provide formal statements in the Online Appendix.⁶ Beginning with the domestic audience's decision, D keeps the incumbent if and only if the cost of replacing her is greater than the benefit of installing a new leader:
$D plays k iff b < c and r otherwise .$
(3)
Equation (3) indicates that whether an incumbent is removed from office depends on the relative cost and benefit of replacement to her constituents. As F is the cumulative distribution function of b, $F (c)$ identifies the probability with which D plays k and L remains in office. While L maximizes the probability she remains in office by allocating her domestic audience's preferred distribution of military spending ( $x_{D}$ ), her personal valuation of x and her motivation for implementing policy also influence her utility. Substituting $F (c)$ for p, expanding terms, and simplifying, $L$ 's expected utility for a given x is
$U_{L (x)} = (- {(x_{L} - x)}^{2}) (m) + (1 - m) [(1 + ((- {(x_{D} - x)}^{2}) s)) R]$
(4)
Equation 5 identifies the x that maximizes Equation 4 and, consequently, equilibrium military spending ( $x^{}$ ) in the model.
$x^{} = {\begin{matrix} x_{L} & if m = 1 \\ \frac{m R s x_{D} - m x_{L} - R s x_{D}}{m R s - m - R s} & if 0 < m < 1 \\ x_{D} & if m = 0 \end{matrix}$
(5)
A leader's optimal allocation of military spending is straightforward when she is motivated exclusively by policy or survival concerns. In the former case ( $m = 1$ ), she implements her ideal distribution of spending ( $x^{} = x_{L}$ ). In the latter ( $m = 0$ ), she implements her key constituents’ ideal allocation of military spending ( $x^{} = x_{D}$ ). If an incumbent is motivated at least marginally by both substantive policy outcomes and her political survival ( $0 < m < 1$ ), a leader's optimal allocation of military spending is a function of both her and her domestic audiences’ preferences, her relative motivation for implementing policy, how easy it is for her constituents to remove her from power, and her reward for retaining office ( $x^{} = \frac{m R s x_{D} - m x_{L} - R s x_{D}}{m R s - m - R s}$ ).

Figure 1 helps demonstrate how equilibrium military spending is shaped by a leader's and her domestic audiences’ preferences, her political vulnerability, and her motivation for implementing policy when she is at least marginally concerned with her political survival and substantive policy outcomes. Figure 1 plots $x^{}$ for a leader who prefers 75% of economic resources be allocated to military spending ( $x_{L} = 0.75$ , dotted red line) and a leader who prefers prefers 25% of resources be dedicated to the military ( $x_{L} = 0.25$ , solid blue line) as a leader's continued political survival becomes more vulnerable ( $s$ goes from 0.01 to 20). The first panel assumes leaders are motivated primarily by survival concerns ( $m = 0.1$ ); the second assumes leaders are equally motivated by policy and survival concerns ( $m = 0.5$ ); and the third assumes leaders are motivated primarily by policy concerns ( $m = 0.9$ ). All calculations assume the median member of each leader's core constituents prefers half of the state's economic resources be dedicated to the military ( $x_{D} = 0.5$ ) and a leader's reward for remaining in power is set to 1 ( $R = 1$ ).

Figure 1.
$x^{}$ as a function of $m$ , $x_{L}$ , $x_{D}$ , $s$ , and $R$ . All calculations assume $x_{D} = 0.5$ and $R = 1$ .

Figure 1 helps illustrate three results concerning the relationships among leaders, their key constituents, and equilibrium military spending. I describe the results, their intuition and implications, and how they are reflected in Figure 1 here and provide formal proofs in the Online Appendix.

Result 1. If $m \neq 0$ , $x^{} \neq \underset{x}{a r g m a x}$ $F (c (v (x), s))$ .

Proof. See Online Appendix. □

Result 1 states that leaders not motivated exclusively by survival concerns implement patterns of military spending that do not maximize their chances of remaining in power. Leaders maximize their chances of remaining in office by implementing their key constituents’ preferred allocation of military spending ( $x_{D}$ ). Leaders who are not purely survival-motivated ( $m \neq 0$ ), though, will always distribute an $x^{} \neq x_{D}$ because they obtain utility from having allocations of military spending reflect their substantive policy preferences.

This characteristic of optimal military spending is apparent in Figure 1. The numerical examples in Figure 1 assume $x_{D} = 0.5$ , implying that any leader would maximize their chances of remaining in power by distributing $x = 0.5$ . Yet, $x^{}$ never equals 0.5 because neither leader's ideal distribution of military spending ( $x_{L}$ ) equals $0.5$ and their utility increases as x approaches their ideal spending mix. Indeed, this will always be the case, as equilibrium military spending will not maximize the probability a leader retains office if they are at least marginally motivated by a desire to see their substantive preferences reflected in government policy.

The claim that a leader's allocation of military spending in equilibrium maximizes their chances of remaining in power only in a special case is straightforward in the context of the model. However, it stands in contrast to most contemporary scholarship on political leaders and the influence of domestic politics on military spending. Explanations of leaders’ influence on military spending in the peace science tradition commonly argue that leaders enact policies that maximize their chances of retaining enough domestic political support to remain in office. Most notably, selectorate theory attributes differences in military expenditures across regime type to differences in the mix of public goods and private benefits that most efficiently secure the political survival of leaders in large-W and small-W systems (Bueno de Mesquita et al., 2003, 2004). Goldsmith (2007) argues that variation in political competition primarily explains why survival-minded democratic leaders and dictators differ in how much of their resources they dedicate to the military during times of war and peace. Focusing on a different feature of politics across regime type, Fordham and Walker (2005) and Carter and Palmer (2015) argue that differences in democratic and non-democratic military spending reflect survival-maximizing leaders needing to maintain the support of domestic political audiences with systematically different preferences over government spending. Result 1 therefore suggests that a key feature of many explanations about leaders’ influence on military expenditures only accurately characterizes what is driving their decisions in a narrow set of cases.

Result 2. As $m \to 1$ , $x^{} \to x_{L}$ .

Proof.* See Online Appendix. □

Result 2 claims that the extent to which optimal military spending deviates from a leader's key constituents’ preferred spending distribution and approaches their own is increasing in the degree to which a leader is motivated by their substantive policy preferences. That is, leaders who care less about the potential consequences of their decisions for their political survival implement policies that more closely reflect their personal preferences than leaders who care more about their continued political survival.

Returning to Figure 1, Result 2's implications for equilibrium military spending are most easily seen by considering $x^{}$ for either leader at any given value of s across the three panels. For example, fixing $s = 1$ , optimal allocation of military spending when $x_{L} = 0.75$ is 0.53 when $m = 0.1$ , 0.63 when $m = 0.5$ , and 0.73 when $m = 0.9$ . If $s = 2$ , the same leader's optimal allocation of military spending is 0.51 when $m = 0.1$ , 0.58 when $m = 0.5$ and 0.7 when $m = 0.9$ . Accordingly, holding their political vulnerability and key constituents’ preferences constant, equilibrium military spending increasingly reflects the leader's substantive preferences as they become less motivated by retaining office.

Result 2 and the intuition behind it imply that the extent to which a leader is responsive to their key constituents’ preferences is not just driven by the ease with which the leader can be removed from power, as is commonly assumed to be the case in IR research (Bueno de Mesquita et al., 2003; Hyde and Saunders, 2020). Rather, the decision to deviate from the key domestic audiences’ preferred policy is also a function of the leader's individual-level motivation for implementing policy. Thus, Result 2 highlights the importance of how leaders’ personal attributes and desire to remain in power interact to shape their decisions and the policies we observe (Carter, Forthcoming).

Result 3. $x^{} \to x_{D}$ as s increases when $0 < m < 1$ , but not when $m = 0$ or $m = 1$ .

Proof. See Online Appendix. □

Result 3 states that optimal military spending more closely reflects the domestic audience's policy preferences as a leader's survival becomes increasingly vulnerable when a leader is motivated by survival and policy concerns, but not otherwise. That is, with the exception of pure ideologues and purely survival-motivated leaders, the extent to which optimal military spending reflects a leader's preferences is increasing in how difficult it is for her constituents to remove her from office. For almost all leaders, then, Result 3 is consistent with an idea central to most leader-centric explanations of foreign policy and conflict processes: leaders will implement policies favored by their key constituents as it becomes easier for their constituents to remove them from power and policies closer to their substantive preferences as it becomes more difficult for their constituents to replace them with a rival.

The consequences of Result 3 for equilibrium patterns of military spending are easily seen in Figure 1. Recall that $x^{}$ is based on the assumption $x_{D} = 0.5$ . Holding a leader's ideal distribution of spending and motivation for implementing policy constant, $x^{}$ moves away from the leader's preferred spending distributions and approaches their constituents’ preferred allocation of military spending as s increases in each panel. Thus, Figure 1 illustrates that among leaders motivated at least partially by policy and survival concerns, optimal military spending increasingly reflects a leader's preferences as their continued survival becomes less vulnerable to their constituents and their constituents’ preferences as their survival becomes more sensitive to their constituents’ preferences.

Importantly, a leader must be motivated by both her continued survival and personal policy preferences for her political vulnerability to influence government spending. If a leader is either a pure ideologue or purely motivated by a desire to remain in power, the ease with which her constituents can remove her from power does not influence equilibrium spending. A pure ideologue does not care whether the mix of spending increases or decreases their chances of remaining in power, as their decisions are exclusively driven by seeing their preferred policies implemented. At the other end of the spectrum, a purely office-seeking leader has no incentive to implement a spending distribution that differs from their constituents’ ideal point, as that is what maximizes their chances of remaining in power. This is the case independent of how easy or difficult it is for her constituents to replace her with a political rival.

That purely survival-motivated leaders are not influenced by their political vulnerability is important given the field's understanding of leaders’ decision-making. The above discussion associated with Result 1 makes this clear for our theoretical explanations of military spending (e.g. Bueno de Mesquita et al., 2004; Goldsmith, 2007). However, scholars explaining patterns of interstate conflict initiation, escalation and outcomes commonly argue that the ease with which a leader can be removed from power influences the decisions they make and that leaders make decisions that maximize their odds of retaining office (Carter and Chiozza, 2018; Wolford, 2021). The model developed here implies that these two features of numerous theoretical arguments cannot simultaneous hold. Leaders will only implement the policies that maximize their political survival when they are exclusively motivated by a desire to remain in power (Result 1). When a leader is exclusively survival-motivated, though, the ease with which her key constituents can remove her from power does not influence equilibrium spending (Result 3). This has implications for our explanations of leaders’ decisions and, consequently, their states’ foreign policies and international relations, a point I return to in the conclusion.

The preceding results indicate that, excepting those leaders who are purely survival- or policy-motivated, leaders’ preferences and political vulnerability and their key constituents’ preferences jointly determine equilibrium military spending. The next section describes three empirical analyses of observed patterns of military spending motivated by this implication.

Empirical analyses

Ideal empirical assessments of the model's implications require, among other things, measures of domestic audiences’ preferences over military spending ( $x_{D}$ ), leaders’ preferences over military spending ( $x_{L}$ ), and the vulnerability of leaders’ survival to their domestic audiences’ preferences ( $s_{i}$ ). Unfortunately, direct measures of these concepts do not really exist and many potential proxies are not available for a large number of the same cases over time. With this constraint in mind, I conducted three sets of empirical analyses that focus on how observable characteristics of leaders and regimes that correspond to the formal model's key concepts influence military expenditures. None of the analyses are dispositive, but all offer evidence consistent with the formal model's account of how leaders’ preferences and political vulnerability and domestic audiences’ preferences influence equilibrium military spending when leaders are motivated by both their political survival and substantive policy preferences. I begin with an assessment of whether military spending systematically varies as a function of the influence members of the military have on a leader's political survival.

Military and civilian audience preferences

The model indicates that military spending will reflect the preference of an incumbent leader's key constituents as long as he or she is at least marginally motivated by a desire to remain in power. If pure ideologues are as rare as scholars assume, patterns of military spending should vary as a function of the collective preferences of the people whose support is required for leaders to remain in power. While there are countless ways to classify members of a society, whether an individual is a member of the military or a civilian is particularly salient in terms of their preferences over government spending.

Military training socializes individuals to value a stronger military and, consequently, favor higher military spending (Huntington, 1957). Moreover, military expenditures beyond the level required to provide the public good of national defense finance private benefits and club goods for members of the military (e.g. higher salaries, the ability to purchase goods and services at reduced prices at base exchanges, and free or subsidized housing) but can crowd out consumption spending popular among the public (Fordham and Walker, 2005). Consistent with these observations, public opinion research finds that, compared with civilians, members of the military and veterans prefer higher levels of military spending and are less likely to support decreasing military spending to fund higher social spending (Bachman et al., 1977; Holsti, 1998; Holsti, 2001; Szayna et al., 2007). Accordingly, existing research indicates that members of the military prefer proportionately more government resources be dedicated to military spending than civilians.

This variation in preferences over government spending is important because political institutions affect the relative influence of the military and civilians in determining a state's political leadership. The political institutions of contemporary democracies make it impossible for a democratic incumbent to retain office with only the support of her country's military. In particular, the relatively high levels of political participation and low costs of voting associated with democracy result in democratic leaders requiring the support of a large portion of the general, civilian public to remain in power (Acemoglu and Robinson, 2006; Boix, 2003; Dahl, 1971). In contrast, non-democratic leaders can typically remain in power without the support of most of the general public, but must require the support (or at least acquiescence) of enough elites to retain office (Bueno de Mesquita et al., 2003). An important dimension along which non-democracies differ is whether their leaders primarily rely on military elites or civilian elites to stay in power (Geddes, 2003; Geddes, Frantz, et al., 2014; Weeks, 2012). With this in mind, we can usefully conceptualize non-democracies as being either a military or a civilian dictatorship, with the distinguishing feature being whether the modal member of a leader's key domestic audience is a member of the military or a member of the civilian elite. In the context of the model, the variation in preferences over government spending associated with members of the military and civilians, whether elite or members of the public, implies $x_{D}^{m i l} > x_{D}^{\bar{m i l}}$ , where $m i l$ represents a military regime. Given that equilibrium military spending when a leader is at least marginally motivated by survival concerns is influenced by their domestic audience's preferences, all else equal, observed military spending should be higher in military regimes than in non-military regimes, whether those regimes are civilian autocracies or democracies.

To provide an initial assessment of this claim, Figure 2 presents the distributions of annual defense burdens for military regimes (green) and non-military regimes (grey) between 1946 and 2010. The black circle in each distribution represents mean annual military spending, while the thick, medium and thin black lines respectively identify the 50th, 80th, and 95th percentile intervals of each distribution. A state's regime type is coded based on Geddes, Wright, and Frantz (2014). A state's defense burden, the percentage of a country's available economic resources allocated to the military in year t, is the most common indicator of national military spending in the defense economics literature as it captures the idea that government spending mirrors the relative preferences and priorities of a state's leadership better than a measure of raw expenditures (e.g. Fordham and Walker, 2005; Sandler and Hartley, 1995). Defense burden data are taken from Anders, Fariss, and Markowitz (2020), who divide a state's constant military expenditures by its surplus domestic product (SDP) assuming its citizens require $3 per day to meet their subsistence needs. Anders, Fariss and Markowitz persuasively demonstrate that measuring a state's defense burden based on the surplus domestic product its government can potentially extract from its population after subsistence needs are taken care of is more appropriate than basing it on a state's raw gross domestic product (GDP).

Figure 2.
Defense burdens in military and non-military regimes, 1946–2010.

Figure 2 is consistent with the model's claim that military spending will reflect a domestic audience's preferences unless a leader is a pure ideologue. The mean annual defense burden for military regimes is 23.08% of their surplus domestic product. In contrast, the mean annual defense burdens for other regimes in the international system is 13.04% of SDP. A t-test indicates that this 10 percentage point difference is significant at greater than the 0.01 level.

I also assess how regime type influences patterns of military spending via linear regression models with random country effects. The full dataset includes 7847 state-year observations from 154 countries between 1946 and 2010. Time-series cross-sectional data present multiple methodological challenges concerning unit heterogeneity and temporal dynamics. Scholars commonly account for unit heterogeneity with fixed-effects or random-effects estimators (Cameron and Trivedi, 2005). As fixed-effects estimators rely exclusively on within-panel variation and, consequently, miss important between-panel variation (Bell and Jones, 2015), I account for systematic heterogeneity in military spending with country-level random effects. I account for temporal dependence in Defense Burden by including a one-year lag of the dependent variable.

In addition to a specification that only includes Military Regime, a lagged dependent variable, and country-level random effects, I estimate a model that controls for a set of factors that could confound the theorized relationship between military spending and military regimes or serve as scope conditions on the hypothesized relationship.⁷ External and internal security threats are associated with higher levels of observed military spending and support for greater military spending among citizens, which could reduce the difference between the preferences of civilians and members of the military over military spending. I therefore control for whether a state was involved in an interstate war (Sarkees and Wayman, 2010) or civil war (Dixon and Sarkees, 2015), its level of external territorial threats (Miller, 2022), and whether it had a strategic interstate rival (Dreyer and Thompson, 2012). I control for a state's (logged) GDP per capita (Gleditsch, 2002, version 6) given that the economic resources states allocate to military spending are decreasing in their economic development (Carter and Palmer, 2016) and higher levels of economic development are associated with non-democratic forms of governance (Przeworski et al., 2000). Finally, I control for the level of support a country receives from a major power. The more support a protégé receives from a patron, the less it needs to provide for its own security via military spending (DiGiuseppe and Shea, 2021). This could influence whether the theorized relationship between audience preferences and patterns of military spending holds. I control for the levels of support that states received from the United States, the Soviet Union, and China with the measures of major power support developed by McManus and Nieman (2019).

Table 1 presents the results of the statistical models. Models 1 and 2, respectively, report the results of the base specification including only Military Regime, the lagged dependent variable, the country-level random effects and the fuller specification that adds the control variables discussed above estimated on the full state-year dataset. As expected, I find that military regimes allocate significantly more of their available economic resources to military spending than do non-military regimes.

Table 1.
Military spending in military and non-military regimes, 1946–2010.

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6

Military Regime 0.60* 0.41* 0.52* 0.45⁺ 0.55* 0.74*

(0.18) (0.19) (0.23) (0.25) (0.23) (0.27)

Interstate War 2.27* 3.13* 2.60*

(0.29) (0.48) (0.51)

Civil War 1.16* 1.40* 1.45*

(0.20) (0.29) (0.30)

Territorial Threat 0.27 1.11* 0.62

(0.22) (0.50) (0.57)

Rivalry −0.22⁺ −0.22 −0.16

(0.12) (0.18) (0.20)

GDP per Capita −0.15* −0.36* −0.72*

(0.07) (0.11) (0.16)

US Support −0.20* −0.42* −0.46*

(0.08) (0.15) (0.18)

Russian Support −0.01 0.06 0.29*

(0.08) (0.11) (0.13)

Chinese Support −0.05 −0.15 −0.18

(0.11) (0.19) (0.20)

Defense Burden $_{t - 1}$ 0.98* 0.96* 0.97* 0.95* 0.97* 0.95*

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

(Intercept) 0.08 1.42* 0.22* 3.03* 0.18⁺ 5.78*

(0.06) (0.57) (0.10) (0.95) (0.11) (1.29)

SD (Intercept) 0.00 0.34 0.00 0.38 0.00 0.55

SD (Observations) 4.13 4.07 5.09 5.04 4.99 4.95

Observations 7351 6962 4193 3988 3599 3414

Sample Democracies and non-Democracies Only non-democracies Civilian or military dictatorships

Two-tailed significance: ⁺p < 0.1; p < 0.05.

Table 1 also includes the results of two sets of robustness checks that estimate the main models on different samples. First, it could be the case that the estimates from Models 1 and 2 are driven by a combination of the observation that non-democracies generally dedicate more resources to military spending than democracies (Carter and Palmer, 2015; Fordham and Walker, 2005) and the inclusion of both democracies and non-democracies in the estimation sample. I therefore estimated a pair of models (Models 3 and 4) on a sample limited to those countries identified as non-democratic by Geddes, Wright, and Frantz (2014). Second, the sample of non-democracies includes a set of regimes where leaders’ key constituents are not best thought of in terms of whether they are members of the military or civilians. This includes monarchies, “warlord” regimes commonly organized around more traditional forms of political authority, “provisional” regimes frequently associated with political transitions, and “occupied” regimes associated with intervening foreign powers. I therefore estimated a pair of models (Models 5 and 6) on a sample limited to military regimes and non-democracies commonly referred to as civilian dictatorships (i.e. party-based regimes or personalist regimes). Consistent with expectations, I find that military regimes generally dedicate more of their available economic resources to military spending than other non-democracies (Models 3 and 4) or civilian dictatorships (Models 5 and 6), although the relationship is statistically significant at only the 0.1 level in Model 4.

Leader preferences and political vulnerability

The formal model indicates that, except for cases involving leaders motivated purely by ideology or their survival, the influence of leaders’ substantive policy preferences on military spending should be larger when their continued political survival is relatively insensitive to their constituents’ preferences. I assess empirical support for this claim by analyzing whether the effect of leaders’ prior military service on patterns of military spending is conditioned by political institutions that differ in terms of how easily a leader's key constituents can remove them from power.

There is increasing evidence that individuals’ background experiences influence their preferences and, consequently, the decisions they make if they later come to power (Horowitz et al., 2015; Krcmaric et al., 2020; Wolford, 2021). Importantly for my purpose, prior military service influences individuals’ preferences and beliefs about the importance of having a strong military and military spending (e.g. Holsti, 1998; Szayna et al., 2007). This suggests that leaders with prior military service generally prefer to allocate more of their countries’ economic resources to military spending than leaders with a civilian background.

To offer some face validity to the idea that leaders’ prior military service could influence patterns of military spending, Figure 3 plots the distribution of annual defense burdens broken down by whether the leader served in the military (olive green) or had a civilian background (grey) prior to obtaining office between 1946 and 2004. A leader's prior military service is taken from Horowitz et al. (2015). Countries led by political executives who previously served in the military dedicate an average of 18% of their surplus domestic product to military spending. States with leaders with a civilian background have a mean defense burden of 13.7% of SDP. A t-test indicates that this 4.3 percentage point difference is significant at greater than the 0.01 level.

Figure 3.
Defense burdens as a function of leaders’ prior military experience, 1946–2004.

The model indicates that the influence of leaders’ substantive preferences on military spending will be conditional on how sensitive their political survival is to their domestic audience's preferences. This implies that the effect of leaders’ prior military service on military spending should be larger when political institutions make it relatively difficult for their key constituents to remove them from power. I first assess this claim by analyzing whether the effect of leaders’ prior military service on military spending is conditional on whether they lead a democracy or a non-democracy.

It is typically harder for dictators’ supporters to remove them from power than it is for democratic leaders’ key constituents (Debs and Goemans, 2010; Hyde and Saunders, 2020; McGillivray and Smith, 2008). Replacing a non-term limited democratic leader with a political challenger most commonly requires an incumbent be defeated in a re-election bid. Voting for the opposition in an election is a relatively costless act of political participation in a democracy. In contrast, non-democratic incumbents commonly are removed from power through irregular events like a coup or revolution (Chiozza and Goemans, 2011). Participation in such events is riskier and costlier than voting for a domestic challenger in a democracy (Lake and Baum, 2001; McGillivray and Smith, 2008). This results in democratic leaders’ tenure being more sensitive to the wishes of their key constituents than non-democratic leaders’ political survival. Variation in political vulnerability across regime type should result in leaders’ substantive preferences having a greater influence on military spending in non-democracies. This implies that leaders’ prior military service should have a larger effect on military spending in non-democracies than in democracies.

I assess this claim by estimating Defense Burden* as a function of an interaction between the variable Democracy and Military Service. A country is coded as a democracy if it had a value of +7 or greater on the 21-point polity2 index from the Polity IV project (Marshall and Jaggers, 2005). I again estimate two specifications. The first includes Democracy, Military Service, their interaction, a one-year lag of the dependent variable, and random country-level effects. The second specification controls for whether a country was involved in an interstate war or a civil war, its level of territorial threat, whether it had an interstate rival, its level of GDP per capita, and its level of support from the United States, Russia, or China. The leader-year dataset covers 6794 leader-year observations of 1251 leaders from 155 countries.

As standard results tables limit one's ability to interpret multiplicative interaction terms (Brambor et al., 2006), I focus my discussion here on the results of a set of post-estimation simulations based on 1000 draws from the coefficient and variance–covariance matrices of each statistical model.⁸ Figure 4 reports the distributions of the marginal effect of a leader's prior military service on their country's defense burden for democracies (blue) and non-democracies (red) yielded by the post-estimation simulations from each model.

Figure 4.
Distribution of the marginal effect of military service on defense burdens in democracies and non-democracies.

Figure 4 suggests that the marginal effect of leaders’ prior military service on national defense burdens in non-democracies is positive: 96% of the simulated marginal effects are positive given the base specification (mean of 0.23) while 94% are positive given the fuller specification (mean of 0.22). The mean marginal effect of a military background on defense burdens in democracies is negative in both specifications (respectively, −0.15 and −0.08), although there is substantially more uncertainty around these estimates (26% of effects are positive with the base specification and 39% are positive given the fuller specification). The difference in the marginal effects across regime type is positive, suggesting that prior military service has a larger positive effect among non-democracies, in 90% of the simulations from the base specification, while 83% are positive when additional control variables are added to the model.⁹

The results reported in Figure 4 are consistent with the idea that, compared with non-democratic institutions, democratic political institutions make it easier for domestic audiences to hold their leaders accountable and, consequently, constrain leaders’ political autonomy and influence on policy outcomes (Hyde and Saunders, 2020). However, there is important variation in the extent to which non-democratic leaders’ political survival is vulnerable to their constituents’ satisfaction with their policy choices. In particular, there is strong evidence that personalist dictators are less likely to be removed from power by their domestic audiences than are leaders of other types of non-democracies, which allows personalist dictators greater policy autonomy and influence than other non-democratic leaders (Chyzh, 2014; Croco and Weeks, 2016; Weeks, 2014). In the context of the formal model developed here, this implies that equilibrium military spending should be more sensitive to the policy preferences of personalist dictators than the leaders of other non-democracies. Given the relationship between military service and preferences over government spending, this should result in the effect of leaders’ prior military service on military spending being larger in personalist dictatorships.

I assess this claim by estimating states’ defense burdens as a function of an interaction between the variable Personalist Regime and Military Service among those states Geddes, Wright, and Frantz (2014) identify as non-democracies. Personalist Regime is coded based on Geddes, Wright, and Frantz (2014). I again estimate two specifications. The first includes Personalist Regime, Military Service, their interaction, a one-year lag of the dependent variable, and random country-level effects. The second specification controls for whether a country was involved in an interstate war or a civil war, its level of territorial threat, whether it had an interstate rival, its level of GDP per capita, and its level of support from the United States, Russia, or China. The dataset covers 4455 leader-year observations of 521 leaders from 121 countries.

I again focus my discussion on the results of a set of post-estimation simulations based on 1000 draws from the coefficient and variance–covariance matrices of each statistical model and provide a results table in the Online Appendix. Figure 5 reports the distributions of the marginal effect of a leader's prior military service on their country's defense burden for personalist (red) and non-personalist (grey) non-democracies.

Figure 5.
Distribution of the marginal effect of military service on defense burdens in personalist and non-personalist non-democracies.

Figure 5 suggests that the marginal effect of leaders’ prior military service on military spending is more consistently positive in personalist regimes than in non-personalist regimes. Among personalist dictatorships, 95% of the marginal effects are positive given both the base and fuller specifications (mean marginal effects are 0.55 and 0.59, respectively). The estimated marginal effects of prior military service are smaller among non-personalist regimes under both the base (0.09) and fuller (0.04) specifications and are less distinguishable from zero (respectively, 68% and 56% of simulated marginal effects are positive). The difference in the marginal effects is positive in 87% of the simulations from the base specification and 90% from the fuller specification, suggesting that leaders’ prior military service has a larger positive effect in personalist dictatorships than in non-personalist authoritarian regimes.¹⁰

Conclusion

The focus on political leaders’ influence on conflict processes arguably represents the most important development in peace science scholarship since the bargaining model of war. This article makes theoretical and empirical contributions to our understanding of how leaders influence patterns of military spending and has implications for conflict processes more generally. The formal model indicates that optimal military spending is a function of a leader's and her constituents’ policy preferences, the relative extent to which a leader is motivated by a desire to remain in power and enact her preferred mix of government spending, and how vulnerable a leader is to being removed from power by her constituents. I use the model's results to motivate three sets of empirical analyses of patterns of military spending between 1946 and 2010. My analyses are consistent with expectations. I find that military spending is higher in regimes where the leader's key domestic audience is disproportionately made up of members of the military. I also find that patterns of military spending are more sensitive to leaders’ prior military service in non-democracies than in democracies and in personalist dictatorships than in other authoritarian regimes. I close with a brief discussion of a few larger implications of my analyses and some suggestions for future research.

My results highlight the utility of explicitly analyzing how leaders’ desire to remain in power and personal attributes, in this case their substantive policy preferences, interact to shape their decision-making. Existing research commonly argues that leaders’ decisions are a function of their desire to remain in power or their personal preferences, beliefs, or psychological traits (Bueno de Mesquita et al., 2003; Carter and Smith, 2020; Horowitz et al., 2015; Hyde and Saunders, 2020), but rarely considers how these two motivations interact (Wolford, 2021). The theoretical model indicates that equilibrium military spending is driven by the relative importance that leaders place on retaining office and obtaining their preferred outcome and, empirically, patterns of military spending are more reflective of leaders’ personal preferences when their political survival is relatively insensitive to their constituents’ preferences. All models by definition omit some features of the real world in favor of highlighting others, but my findings suggest there are benefits to explicitly considering how leaders’ desire to remain in power and personal attributes jointly influence topics central to peace science.

Indeed, arguably the theoretical model's most important implication is that a common aspect of institutional arguments that explain leaders’ decisions as a function of their desire to remain in power requires leaders to care about more than their continued political survival in order to hold on. Arguments about the influence of political institutions on foreign policy and international relations commonly assert that leaders implement the policies that maximize their chances of remaining in power and that the ease with which a domestic audience can remove an incumbent from power influences leaders’ decisions (Carter and Chiozza, 2018; Hyde and Saunders, 2020). The model developed here indicates that only purely survival-motivated leaders ever implement the policies that maximize their chances of retaining office in equilibrium. Importantly, though, the structural costs associated with removing these leaders from power are irrelevant to the policies such leaders implement. The institutional costs of leader replacement do influence the decisions of incumbents motivated by both their political survival and their substantive policy preferences, but those leaders never implement the policy that maximizes their chances of retaining office in equilibrium. Accordingly, political institutions and regimes can affect leaders’ decisions by influencing how easily they can be held accountable by their constituents, as is frequently argued, but this feature of political institutions does not influence the behavior of the myopically survival-focused leaders that typically populate these same arguments. This suggests explicitly considering that what leaders want to accomplish beyond remaining in power could produce a better understanding of when and how political institutions systematically influence leaders’ decisions, foreign policy, and international relations.

The theoretical model has a number of empirical implications beyond those explored here. First, the model indicates that the mix of military and non-military spending that best secures a leader's political survival will vary as a function of the leader's constituents’ policy preferences. There is evidence consistent with this in the context of war-time military spending across regime type (Carter, 2017), but I am unaware of research that analyzes how government spending systematically influences variation in cross-national leader survival during times of peace or beyond the democracy/dictatorship divide. Second, there are a number of political cleavages not considered here that could induce variation in the costs of removing a leader and the preferences of a leader's key constituents and, consequently, patterns of government spending. For example, whether a prime minister presides over a majority or minority government, how popular a leader is with the public, and the relative power of political parties in non-democracies all influence how easy it is to remove leaders from power and, therefore, the relative influence of leaders’ preferences and their key constituents’ preferences on leaders’ decisions. Third, I suspect that the underlying relationships among leaders, their constituents, and government policy described here extend beyond patterns of military spending to other domains, including but not limited to foreign aid (Greene and Licht, 2024), economic sanctions (McLean and Whang, 2014), international cooperation (Wolford and Ritter, 2016), and conflict initiation (Jones, Mattiacci, and Nordstrom, 2024).

Finally, the theoretical model indicates that leaders’ responsiveness to their constituents is the result of an interaction between an individual-level attribute and how easily they can be removed from power. I have generally discussed a leader's political vulnerability to their key constituents in terms of relatively static political institutions. However, both across and within regime type, a leader's vulnerability to her key domestic audience is probably best thought of as a dynamic and malleable constraint on their decision-making (Clark and Nordstrom, 2005; Hyde and Saunders, 2020). For example, public opinion and economic performance shape how easy it is to replace any leader, the costs associated with replacing a democratic leader with a political rival vary considerably as a function of the electoral calendar (Williams, 2013), and the difficulty of removing a dictator from power depends on whether he has consolidated political control, his tenure and his expected health (e.g. Bueno de Mesquita and Smith, 2018; Svolik, 2009). All of these factors should shape the relative extent to which the choices leaders make are driven by their personal preferences and consideration of how a given policy would influence their ability to remain in power. More generally, simultaneously considering variation in leaders’ attributes and their relative political security should improve our understanding of their decisions and conflict processes.

Supplemental Material

sj-pdf-1-cmp-10.1177_07388942231199164 - Supplemental material for Political leaders and military spending

Supplemental material, sj-pdf-1-cmp-10.1177_07388942231199164 for Political leaders and military spending by Jeff Carter in Conflict Management and Peace Science

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Military Regime	0.60*	0.41*	0.52*	0.45⁺	0.55*	0.74*
	(0.18)	(0.19)	(0.23)	(0.25)	(0.23)	(0.27)
Interstate War		2.27*		3.13*		2.60*
		(0.29)		(0.48)		(0.51)
Civil War		1.16*		1.40*		1.45*
		(0.20)		(0.29)		(0.30)
Territorial Threat		0.27		1.11*		0.62
		(0.22)		(0.50)		(0.57)
Rivalry		−0.22⁺		−0.22		−0.16
		(0.12)		(0.18)		(0.20)
GDP per Capita		−0.15*		−0.36*		−0.72*
		(0.07)		(0.11)		(0.16)
US Support		−0.20*		−0.42*		−0.46*
		(0.08)		(0.15)		(0.18)
Russian Support		−0.01		0.06		0.29*
		(0.08)		(0.11)		(0.13)
Chinese Support		−0.05		−0.15		−0.18
		(0.11)		(0.19)		(0.20)
Defense Burden $_{t - 1}$	0.98*	0.96*	0.97*	0.95*	0.97*	0.95*
	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
(Intercept)	0.08	1.42*	0.22*	3.03*	0.18⁺	5.78*
	(0.06)	(0.57)	(0.10)	(0.95)	(0.11)	(1.29)
SD (Intercept)	0.00	0.34	0.00	0.38	0.00	0.55
SD (Observations)	4.13	4.07	5.09	5.04	4.99	4.95
Observations	7351	6962	4193	3988	3599	3414
Sample	Democracies and non-Democracies	Only non-democracies	Civilian or military dictatorships

Footnotes

Acknowledgments

Phil Arena, Scott Bennett, Michael Bernhard, David Carter, Stephen Chaudoin, Giacomo Chiozza, Matt DiGiuseppe, Ben Fordham, Ben Jones, Doug Lemke, Amanda Licht, Burt Monroe, Tim Nordstrom, Glenn Palmer, Jim Shortle, Laron Williams, Chris Zorn, Heather Ondercin, the editor, and the reviewers are thanked for comments on various versions of this project. A special thanks to Glenn Palmer for once asking me, “Why would a leader ever not do what their winning coalition wants?” All remaining errors are my own.

Declaration of conflicting interest

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Jeff Carter

Supplemental material

All data, replication materials, and instructions regarding analytical materials upon which published claims rely are available online through the SAGE CMPS website: .

Notes

References

Acemoglu

Robinson

(2006) Economic Origins of Dictatorship and Democracy. New York: Cambridge University Press.

Alesina

Cukierman

(1990) The politics of ambiguity. The Quarterly Journal of Economics 105(4): 829–850.

Anders

Fariss

Markowitz

(2020) Bread before guns or butter: Introducing surplus domestic product (SDP). International Studies Quarterly 64(2): 392–405.

Arena

Nicoletti

(2014) Selectorate theory, the democratic peace, and public goods provision. International Theory 6(03): 391–416.

Bachman

Blair

Segal

(1977) The All-Volunteer Force: A Study of Ideology in the Military. Ann Arbor, MI: University of Michigan Press.

Bell

Jones

(2015) Explaining fixed effects: Random effects modeling of time-series cross-sectional and panel data. Political Science Research and Methods 3(1): 133–153.

Boix

(2003) Democracy and Redistribution. New York: Cambridge University Press.

Brambor

Clark

Golder

(2006) Understanding interaction models: Improving empirical analyses. Political Analysis 14(1): 63–82.

Brooks

Manza

(2007) Why Welfare States Persist: The Importance of Public Opinion in Democracies. Chicago: The University of Chicago Press.

10.

Bueno de Mesquita

Smith

(2018) Political loyalty and leader health. Quarterly Journal of Political Science 13(4): 333–361.

11.

Bueno de Mesquita

Morrow

Siverson

, et al. (1999) An institutional explanation of the democratic peace. American Political Science Review 93(4): 791–807.

12.

Bueno de Mesquita

Smith

Siverson

, et al. (2003) The Logic of Political Survival. Cambridge: MIT Press.

13.

Bueno de Mesquita

Morrow

Siverson

, et al. (2004) Testing novel implications of the selectorate theory of war. World Politics 56(3): 363–388.

14.

Burstein

(2003) The impact of public opinion on public policy: A review and an agenda. Political Research Quarterly 56(1): 29–40.

15.

Cameron

Trivedi

(2005) Microeconometrics: Methods and Applications. New York: Cambridge University Press.

16.

Carey

Niemi

Powell

(1998) The effects of term limits on state legislatures. Legislative Studies Quarterly 23(2): 271–300.

17.

Carter

(2017) The political cost of war mobilization in democracies and dictatorships. Journal of Conflict Resolution 61(8): 1768–1794.

18.

Carter

Chiozza

(2018) State leaders and foreign policy. In: Thies

(eds) Oxford Encyclopedia of Foreign Policy Analysis. Oxford, UK: Oxford University Press, pp. 722–739.

19.

Carter

(Forthcoming) Leader hawkishness, political survival, and interstate crises. Journal of Politics. https://doi.org/10.1086/726963

20.

Carter

Palmer

(2015) Keeping the schools open while the troops are away: Regime type, interstate war, and government spending. International Studies Quarterly 59(1): 145–157.

21.

Carter

Palmer

(2016) Regime type and interstate war finance. Foreign Policy Analysis 12(4): 695–719.

22.

Carter

Smith

(2020) A framework for measuring leaders’ willingness to use force. American Political Science Review 114(4): 1352–1358.

23.

Carter

Ondercin

Palmer

(2021) Guns, butter, and growth: The consequences of military spending reconsidered. Political Research Quarterly 74(1): 148–165.

24.

Chiozza

Goemans

(2011) Leaders and International Conflict. New York: Cambridge University Press.

25.

Chiozza

Khalifa

(2024) The Harsh Face of the Empire by Invitation: Coups in the US World Order. Conflict Management and Peace Science 41(2): 110–131.

26.

Chyzh

(2014) Can you trust a dictator: A strategic model of authoritarian Regimes’ signing and compliance with international treaties. Conflict Management and Peace Science 31(1): 3–27.

27.

Clark

Nordstrom

(2005) Democratic variants and democratic variance: How domestic constraints shape interstate conflict. Journal of Politics 67(1): 250–270.

28.

Clarke

Primo

(2012) A Model Discipline: Political Science and the Logic of Representations. Oxford, UK: Oxford University Press.

29.

Croco

Weeks

(2016) War outcomes and leader tenure. World Politics 68(4): 577–607.

30.

Dahl

(1971) Polyarchy: Participation and Opposition. New Haven, CT: Yale University Press.

31.

Debs

Goemans

(2010) Regime type, the fate of leaders and war. American Political Science Review 104(3): 430–445.

32.

DiGiuseppe

Shea

(2021) Alliances, signals of support, and military effort. European Journal of International Relations 27(4): 1067–1089.

33.

Dixon

Sarkees

(2015) A Guide to Intra-State Wars: An Examination of Civil, Regional, and Intercommunal Wars, 1816–2014. Washington, DC: CQ Press.

34.

Dreyer

Thompson

(2012). Handbook of International Rivalries , Washington, DC: CQ Press, 1494–2010.

35.

Fenno

(1978) Homestyle: House Members in Their Districts. Boston, MA: Little, Brown.

36.

Fordham

Walker

(2005) Kantian liberalism, regime type and military resource allocation: Do democracies spend less? International Studies Quarterly 49(1): 141–157.

37.

Gandhi

Przeworski

(2006) Cooperation, cooptation, and rebellion under dictatorship. Economics and Politics 18(1): 1–26.

38.

Gandhi

Przeworski

(2007) Authoritarian institutions and the survival of autocrats. Comparative Political Studies 40(11): 1279–1301.

39.

Geddes

(2003) Paradigms and Sand Castles: Theory Building and Research Design in Comparative Politics. Ann Arbor, MI: University of Michigan Press.

40.

Geddes

Frantz

Wright

(2014) Military rule. Annual Review of Political Science 17: 147–162.

41.

Geddes

Wright

Frantz

(2014) Autocratic breakdown and regime transitions: A new data set. Perspectives on Politics 12(2): 313–331. doi:10.1017/S1537592714000851

42.

Geddes

Wright

Frantz

(2018) How Dictatorships Work: Power, Personalization, and Collapse. New York: Cambridge University Press.

43.

Gleditsch

(2002) Expanded trade and GDP data. Journal of Conflict Resolution 46(5): 712–724.

44.

Goldsmith

(2007) Defense effort and institutional theories of democratic peace and victory: Why try harder? Security Studies 16(2): 189–222.

45.

Greene

Licht

(2024) Donor Political Preferences and the Allocation of Aid: Patterns in recipient type. Conflict Management and Peace Science 41(2): 155–176.

46.

Holsti

(1998) A widening gap between the US military and civilian society?: Some evidence, 1976–96. International Security 23(3): 5–42.

47.

Holsti

(2001) Of chasms and convergences: Attitudes and beliefs of civilians and military elites at the start of a new millennium. In: Feaver

Kohn

(eds) Soldiers and Civilians: The Civil–Military Gap and American National Security. Cambridge, MA: MIT Press, 15–100.

48.

Horowitz

Stam

Ellis

(2015) Why Leaders Fight. New York: Cambridge University Press.

49.

Huntington

(1957) The Soldier and The State: The Theory and Politics of Civil–Military Relations. Cambridge, MA: Harvard University Press.

50.

Hyde

Saunders

(2020) Recapturing regime type in international relations: Leaders, institutions, and agency space. International Organization 74(2): 363–395.

51.

Jones

Mattiacci

Nordstrom

(2024) How Leader's Type Shapes the Effect of Nuclear Latency on Dispute Involvement. Conflict Management and Peace Science 41(2): 177–193.

52.

Key

(1961) Public Opinion and American Democracy. New York: Knopf.

53.

Krcmaric

Nelson

Roberts

(2020) Studying leaders and elites: The personal biography approach. Annual Review of Political Science 23: 133–151.

54.

Lake

Baum

(2001) The invisible hand of democracy: Political control and the provision of public services. Comparative Political Studies 34(6): 587–621.

55.

Marshall

Jaggers

(2005) Polity IV dataset users’ manual. www.cidcm.umd.edu//polity.

56.

McGillivray

Smith

(2008) Punishing the Prince. A Theory of Interstate Relations, Political Insitutions, and Leader Change. Princeton, NJ: Princeton University Press.

57.

McLean

Whang

(2014) Designing foreign policy: Voters, special interest groups, and economic sanctions. Journal of Peace Research 51(5): 589–602.

58.

McManus

Nieman

(2019) Identifying the level of Major power support signaled for protégés: A latent measure approach. Journal of Peace Research 56(3): 364–378.

59.

Miller

(2022) A random item response model of external territorial threat, 1816–2010. Journal of Global Security Studies 7(4): ogac012.

60.

Oatley

(2015) A Political Economy of American Hegemony. New York: Cambridge University Press.

61.

Page

Bartels

Seawright

(2013) Democracy and the policy preferences of wealthy Americans. Perspectives on Politics 11(1): 51–73.

62.

Palmer

(1990a) Alliance politics and issue-areas: Determinants of defense spending. American Journal of Political Science 34(1): 190–211.

63.

Palmer

(1990b) Corralling the free rider: Deterrence and the western alliance. International Studies Quarterly 34(2): 147–164.

64.

Przeworski

Alvarez

Cheibub

, et al. (2000) Democracy and Development: Political Institutions and Well-being in the World, 1950–1990. New York: Cambridge University Press.

65.

Rider

Findley

Diehl

(2011) Just part of the game? Arms races, rivalry, and war. Journal of Peace Research 48(1): 85–100.

66.

Rothenberg

Sanders

(2000) Severing the electoral connection: Shirking in the contemporary congress. American Journal of Political Science 44(2): 316–325.

67.

Sandler

Hartley

(1995) The Economics of Defense. New York: Cambridge University Press.

68.

Sarkees

Wayman

(2010) Resort to War: 1816–2007. Washington, DC: CQ Press.

69.

Sexton

Wellhausen

Findley

(2019) How government reactions to violence worsen social welfare: Evidence from Peru. American Journal of Political Science 63(2): 353–367.

70.

Shepsle

(1972) The strategy of ambiguity: Uncertainty and electoral competition. American Political Science Review 66(2): 555–568.

71.

Smith

Bueno de Mesquita

(2011) Contingent prize allocation and pivotal voting. British Journal of Political Science 42(2): 371–392.

72.

Soroka

Wlezien

(2010) Degrees of Democracy: Politics, Public opinion, and Policy. New York: Cambridge University Press.

73.

Strøm

(1990) A behavioral theory of competitive political parties. American Journal of Political Science 34(2): 565–598.

74.

Svolik

(2009) Power sharing and leadership dynamics in authoritarian regimes. American Journal of Political Science 53(2): 477–494.

75.

Szayna

McCarthy

Sollinger

, et al. (2007) The Civil-Military Gap in the United States: Does it Exist, Why, and Does it Matter?, Vol. 379. Santa Monica, CA: Rand Corporation.

76.

Tingley

(2010) Donors and domestic politics: Political influences on foreign aid effort. The Quarterly Review of Economics and Finance 50(1): 40–49.

77.

Weeks

(2012) Strongmen and straw men: Authoritarian regimes and the initiation of international conflict. American Political Science Review 106(2): 326–347.

78.

Weeks

(2014) Dictators at War and Peace. Ithaca, NY: Cornell University Press.

79.

Whitten

Williams

(2011) Buttery guns and welfare hawks: The politics of defense spending in advanced industrial democracies. American Journal of Political Science 55(1): 117–134.

80.

Williams

(2013) Flexible election timing and international conflict. International Studies Quarterly 57(3): 449–461.

81.

Wittman

(1977) Candidates with policy preferences: A dynamic model. Journal of Economic Theory 14(1): 180–189.

82.

Wolford

(2021) What do we know about leaders and war? In: Mitchell

Vasquez

(eds) What Do We Know about War?, 3rd ed. Lanham, MD: Rowman & Littlefield Publishers, 244–259.

83.

Wolford

Ritter

(2016) National leaders, political security, and the formation of military coalitions. International Studies Quarterly 60(3): 540–551.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.11 MB