The Industrial Peace: Schumpeter, Conflict, and theInvestment-War Tradeo�∗

J. Tyson Chatagnier†

Emanuele Castelli‡

July 29, 2014

Abstract

Drawing on the writings of Joseph Schumpeter, we develop a theory of international con-flict that challenges proponents of the liberal peace research programs. We outline a simplemechanism whereby industrialization fosters peace, suggesting that industrialized states aremore peaceful because they can gain more by investing at home than by pursuing foreignmilitary conquest. We borrow from Schumpeter to argue that our mechanism is distinct fromtraditional measures of liberalism. Empirically, we propose a measure of industrial devel-opment, based on its economic structure. Using World Bank sector-specific economic data,our analysis shows that a more industrialized economy significantly reduces the likelihoodthat a state will be involved in a fatal military conflict. We show that this result is robustacross a number of model specifications, holds at both the monadic and dyadic levels, and isindependent of both democracy and capitalism.

∗Authors’ names are listed in reverse alphabetical order, and do not indicate unequal contributions. A previousversion of this paper was presented at the 2013 SGRI conference in Trento, Italy. We thank participants, as well asFilippo Andreatta, Alessandro Colombo, Joe Grieco, Jakub Grygiel, Christopher Hill, Gary Hollibaugh, Ted Hopf, KerimCan Kavakli, Pat Morgan, Angelo Panebianco, and Dan Reiter for insightful comments.

†Corresponding author. Research Center on International Politics and Conflict Resolution, Bruno Kessler Founda-tion and Johns Hopkins SAIS, Bologna Center. Email: chatagnier@fbk.eu

‡Research Center on International Politics and Conflict Resolution, Bruno Kessler Foundation. Email: castelli@fbk.eu

“War is the business of barbarians.”— Napoleon Bonaparte

Introduction

In his 1919 essay, “The Sociology of Imperialism,” Joseph Schumpeter argued that the shift to

an industrialized, capitalist mode of production would gradually eliminate interstate war. For

Schumpeter, this was primarily the result of a change in cultural and psychological attitudes

brought about by a transition to modernity, which rendered the idea of war obsolete. Indeed, at

the time of his writing, the futility and slaughter of the First World War were still vivid in the

European mind. However, he also suggested that military conflicts would become irrational for a

more basic and material reason. Specifically, he argued that as nations industrialized “what was

once energy for war [would become] simply energy for labor of every kind” (Schumpeter 1955, 69).

Moreover, as he later wrote, as a nation becomes more fully modernized in terms of both attitudes

and economic structure, “the more pacifist—and the more prone to count the costs of war—we

observe it to be” (Schumpeter 1942, 128–129).

Unfortunately, as the historical record shows, war and international violence have not disap-

peared, despite the spread of industrialization. Economic development has had contrasting e�ects

on interstate war: while industrial systems make it cheaper to produce and purchase, rather

than obtain with force, they also increase a state’s ability to project power (Gartzke and Hewitt

2010). On the other hand, the fact that today’s military disputes are confined primarily to specific

areas of the world (mainly underdeveloped and resource-rich regions) while other areas, such

as recently-industrialized East Asia, are now enjoying “the most broadly peaceful era in [their]

history” (Wainwright 2009), largely vindicates Schumpeter. What accounts for this divide? Does

industrialization a�ect a state’s tendency to go to war?

Drawing from Schumpeter’s argument, as well as insights from work on state-level causes

of war and peace, we argue that—all else equal—the transition to an industrialized economy

pushes states to pursue peaceful foreign policies. Our theoretical story begins from the observa-

1

tion that, regardless of regime type, states fight less today than in earlier times (Gat 2005), and

continues within the economically rationalist framework put forward by Schumpeter, arguing that

industrialized states fight less frequently because of a shift away from the material profitability of

war.

To demonstrate the empirical applicability of our argument, we examine the relationship

between the shift to an industry-based mode of production and the likelihood of (violent) military

disputes. Accordingly, our key explanatory variable is the size of a nation’s industrial sector, which

measures the percentage of GDP from industrial endeavors during a given year. This allows us to

distinguish, for example, between a�uent primary commodity exporters and fully industrialized

countries that generate wealth through manufacturing.

The paper proceeds as follows. We begin by critically reviewing the scholarship on state-

level causes of peace, focusing on the democratic and capitalist peace literatures, which serve

as a background for our analysis. Next, we outline our theoretical model, distinguishing it from

previous research on domestic causes of peace, and propose a basic mechanism that can shed

light on the nexus between peace and industrialization. The next two sections are devoted to

empirical analysis, which is conducted at both the monadic and dyadic levels. Finally, we draw

conclusions and suggest some paths for future research.

One Peace, Many Causes

While Europe was a place of near-constant warfare in the Middle Ages and early modern period,

the instances of war began to ebb in the 19th century, and have largely ceased since the end of

World War II. The same phenomenon has been observed in many other parts of the world, and

some scholars have noted its coincidence with the rise of liberalism. This relationship between

liberalism and peace—initially based on the so-called “Kantian tripod” (see Doyle 1983a,b)—has

been at the core of the international relations research agenda for the last three decades. Since

the earliest theoretical contributions (Babst 1972; Rummel 1979), hundreds of studies have been

2

published on the topic, many supported by robust quantitative analyses (Hayes 2012). These works

have resulted in a consensus that democratic countries do not fight each other (e.g., Maoz and

Russett 1992; Bennett 2006). According to Levy (1988, 662), the lack of wars between democracies

is “as close as anything we have to an empirical law in international relations.” The theoretical

rationale for this phenomenon rests upon the institutional di�erences between democracies and

autocracies. Democracies are constrained at a structural level by both institutions and the public,

rendering them less likely to begin or escalate conflicts, due to political costs from the domestic

public (Morgan and Schwebach 1992; Gelpi and Griesdorf 2001) and institutional checks and bal-

ances within the government (Morgan and Campbell 1991). Additionally, democratic governments

tend to have broad winning coalitions that require the provision of public goods: a leader who

loses a war loses (some of) the ability to distribute the goods required to retain o�ce (Bueno de

Mesquita et al. 2003).

Explanatory power notwithstanding, these studies do not imply that democracy is a necessary

and su�cient condition for peace. On the contrary, researchers have noted that the process

of democratization—especially when it is particularly quick or problematic (Ward and Gleditsch

1998)—may actually inhibit peace (Mansfield and Snyder 2002), and that the likelihood of engaging

in conflict is greater for states undergoing democratization than for either settled democracies or

autocracies (Hegre et al. 2001). In addition, even the “democratic” nature of the democratic peace

is not su�ciently clear, as there are today a number of “illiberal democracies” (Zakaria 1997)

and electoral autocracies (Levitsky and Way 2002), which blur the line between democratic and

non-democratic states. In fact, the underlying variable that ties together the research on the

democratic peace is not simply the existence of a democratic government, or of free and fair

elections, but the degree to which the domestic public enjoys political liberty. That is, liberal

features (including the rule of law, separation of powers, and limited government) may play a

larger role than democratic components (elections) in fostering peace among states (Mandelbaum

2007). Moreover, an institutional theory cannot explain how a monadic characteristic (institutional

constraints) can generate dyadic e�ects (separate peace among democracies): if democracies are

3

less likely to wage war in order to avoid domestic audience costs, then they should be peaceful in

general, and not only toward other democracies. This has prompted a shift in recent years toward

a focus on regime-similarity more generally (e.g., Peceny, Beer and Sanchez-Terry 2002; Gartzke

and Weisiger 2013).

Political variables alone, then, cannot completely explain the inter-democratic peace. For this

reason, scholars have begun including economic variables in their analyses. In doing so, some

(e.g., Hegre 2000; Dorussen and Ward 2010) have returned to Rosecrance’s (1986) “trading states”

hypothesis, in order to demonstrate that conflict among democracies can be impeded by free

trade. Although this “commercial peace” (Souva and Prins 2006; McDonald 2007) has not been

accepted uncritically (see Beck, Katz and Tucker 1998; Barbieri 1996; Rasler and Thompson 2005),

it has provided significant results both in terms of explanations and dyadic e�ects: it o�ers an

intuitive explanation—that trading for a good can be more attractive than fighting for it—and the

hypothesis that peace can be brought about through free trade o�ers a dyadic cause to explain

dyadic e�ects.

Only recently, due in part to the renewal of interest in the link between democracy and

markets (Gleditsch 2008), have researchers shifted their focus away from political liberalism and

toward economics. These authors have stressed the point that markets and development can be

important forces for peace between countries. A key factor behind this perspective is the fact

that the growth of global prosperity has eclipsed the growth of democracy in recent decades

(see Gartzke and Weisiger N.d.). Additionally, some scholars have demonstrated that the link

between political liberalism and peace is conditional upon economic performance: democracy

only begets peace for su�ciently wealthy states (Mousseau, Hegre and Oneal 2003; Gat 2005).

This has led to the formulation of a “capitalist peace” theory, originally articulated by Weede

(1995), which has been developed in a number of ways. Mousseau (2009) makes a “market

civilization” argument, suggesting that countries with strong connectedness among market actors

have a normative tendency toward contractual exchange, giving them distinct preferences for

cooperation over conflict. Gartzke contends that financial openness and capital market integration

4

can inhibit conflict at the dyadic level (Gartzke 2007). Finally, McDonald draws on the bargaining

literature and selectorate theory to demonstrate that states with less free markets tend to pursue

more aggressive foreign policies. This explanation holds even at the monadic level, leading him to

contend that the “peace created by capitalism is much stronger than that of democracy” (McDonald

2010, 164).

Given the strength of the capitalist peace’s challenge to the Kantian research program, some

authors suggest that it might not only serve as a complement (Gartzke and Hewitt 2010), but

may even subsume the democratic peace (Mousseau 2010). However, although the hypotheses

put forth by capitalist peace scholars are consistent, they lack “precise microfoundations that link

markets and their main attributes to peace” (Schneider and Gleditsch 2010, 110). This may be

one factor that has made it di�cult for researchers to agree on an operationalization of their

key concept. Moreover, political and economic liberalism tend to go hand-in-hand (Brawley 1993),

and it is common for democracies to adopt capitalist economies (Choi 2011), making it di�cult to

distinguish the e�ects of democracy from those of capitalism.

Although the liberal peace research program has contributed greatly to our understanding of

peace in the modern era, we believe that the criticisms noted above are both valid and serious.

We argue that a new theoretical framework is necessary, which avoids the issues outlined above.

In the next section, we elaborate more fully upon our theory and demonstrate that it addresses

the key problems that plague the liberal peace literature.

An Industrial Peace

We propose a novel explanation for peace among nations, based on the writings of Schumpeter.

In e�ect, we advance an “industrial peace,” wherein states avoid violent conflict and embrace

peaceful interactions as they reach su�ciently high levels of industrialization. First, we ground our

analysis in a well-established theoretical framework provided by Schumpeter’s insights (much as

Kant provides a foundation for democratic peace scholars), which suggests monadic explanations

5

for monadic e�ects. Second, our theory addresses the microfoundation problem by arguing that

an investment-war tradeo� provides industrialized countries with a rational incentive to avoid

war. Finally, we o�er a well-defined independent variable, which focuses on the primary aspect

of the modern economy: its level of industrialization. We argue that, to a greater extent than

other economic factors, industrialization is a key variable in reducing a state’s likelihood of waging

war. Notably, an industrial peace means that ownership of the means of production, be it by

private entrepreneurs or the state itself, is irrelevant. Our explanation does not exclude either

non-democratic states or states with centrally planned economies, so long as the economy is

properly structured. This can explain important variations in conflict behavior across both liberal

and illiberal regimes.

Our theory is related to an observation by the philosopher Ronald Glossop, who argues that

one of the primary causes of interstate conflict is that in “less developed countries, technology may

not be able to produce what is wanted more cheaply than it can be obtained by force” (Glossop

2001, 70). Our argument e�ectively inverts Glossop’s: highly industrialized countries possess

less incentive to engage in violent conflict because they have technology that can produce what is

wanted more cheaply than it can be obtained by force. Thus, we contend that the move away from

profit-seeking warfare, especially in Europe and North America (and more recently in East Asia),

is due to the e�ects of industrialization—defined here as the shift away from an agrarian mode

of production to a concentration on manufacturing, grounded in technological innovation and

continuous investment—which simultaneously made the holding of raw materials less important

and broke the “Malthusian trap,” significantly increasing material standards of living (Gat 2005,

77). Together, these factors made the acquisition of additional territory unnecessary, changing the

calculus of war and rendering it virtually obsolete from a material perspective (Kaysen 1990).

When a state shifts to an industrialized mode of production citizens’ everyday lives undergo a

material change. In industrialized economies, individuals are engaged in production. Workers are

committed to their tasks, while owners of the means of production find it more attractive to invest

and produce than to take from outside. In other words, “[t]he competitive system absorbs the full

6

energies of most of the people at all economic levels” and “there is much less excess energy to be

vented in war and conquest than in any [preindustrial] society. What excess energy there is flows

largely into industry itself” (Schumpeter 1955 [1919], 69). Moreover, once industrialized, continuous

investment is necessary to sustain economic growth, through Schumpeter’s process of “creative

destruction,” by which the state is further removed from the previous economic order.

This means that there exists an investment-war tradeo� for industrialized countries. In order

to sustain growth (a basic requirement for every industrialized economy), governments and private

entrepreneurs must reinvest profits in innovation. This creates job opportunities for potential

workers, enhancing their well-being and satisfaction with the status quo. Political leaders are

also better o� in this situation, as they can extract more revenue from a richer society. In

an industrialized economy, the risk of war threatens this virtuous mechanism of investment,

innovation, job creation, profits, and taxes, rendering war materially irrational. In other words,

each dollar that is spent engaging in militarized conflict—regardless of the money devoted to

military spending—is one dollar less to spend on the necessary activities of innovation and

economic growth. Thus, industrialization has left fewer states with what Schumpeter (1955 [1919],

4) refers to as a “concrete interest” in waging war. Although war is profitable in agrarian societies,

where land and resources are necessary for economic growth, industrialized societies grow simply

by improving upon resources.

Thus, the industrialization of the modern economy has removed many of the materialist incen-

tives for war. However, we do not argue that the material shift toward industry ensures instanta-

neous peaceful interaction between states. Indeed, some facets of industrialization may potentially

increase the likelihood of war, as realist scholars (Mearsheimer 2001, 63), “conquest pays” authors

(Liberman 1998), and lateral pressure theorists (Choucri and North 1975) have argued. However, we

believe that this does not necessarily mean that highly-industrialized, militarily-capable states will

necessarily display aggressive behaviors. Contrary to the realist view, highly-capable status quo

powers may also signal their commitment to fight if challenged, deterring conflict. With respect

to the cumulativity argument, we note that the very nature of industrialization has changed as

7

the shift “from smokestack to knowledge-based, high-technology production [...] has reduced the

cumulativity of industrial base” (Van Evera 1999, 115). Finally, we observe empirically that no

developed country has seized another during the last sixty years, and we would argue that this

is because industrial domestic resources (such as heavy industrial assets, industrial ouputs, and

machinery) have become less lootable and re-usable.

Instead, we simply claim that industrialization introduces a tradeo� between investment and

war that eventually leads to a reduction in the likelihood of conflict initially, and continues to

pacify state behavior over time. This is because, ultimately, the introduction of this tradeo� alters

the interests of the state. While agrarian societies and those that are reliant on the export of natural

resources can gain significantly from war because it is a cheap and e�ective means of resource

acquisition, gains in industrialized countries come from continuous production and innovation (see

Gat 2005, 84). Indeed, North, Wallis and Weingast (2009, 23) have suggested, wealth creation in

natural states usually comes from rent (exploitation of land, labor, and natural resources). However,

since the industrial revolution, with the shift to open-access societies, traditional sources of rent

have gradually eroded1 and innovation itself (creative destruction) has become a source of rent.

The cornerstone of our argument, then, is that industrialization removes incentives for states

to opt for war over trade (or even concessions). As states become more heavily industrialized—all

else being equal—they should eschew adventurous foreign policy. Examples of this phenomenon

abound in international relations. Consider the case of Asia during the last five decades, as

industrialization spread throughout the region. India and China, for example, were involved in

several border disputes, beginning in the 1950s, including a war in 1962. Following the onset of

industrialization in the two countries in the early 1980s (India with its Seventh Five Year Plan,

launched in 1985, and China with Deng Xiaoping’s Reforms), relations became significantly less

militarized (the final dispute including the use of force arising in 1987). A similar pattern can also

be found between China and Vietnam (who have had no fatal disputes since 1987). By contrast,

non-industrialized states tend to fight over economic issues. Examples of such disputes are

1Of course, the timing and the impact of industrialization can be di�erent for di�erent countries, and naturalstates exist even in industrialized areas (contemporary Russia can be considered a natural state).

8

common in Latin America, including the 1969 Football war between El Salvador and Honduras and

the 1965 Laguna del Desierto incident between Argentina and Chile. Thus, we primarily advance

a monadic explanation for international peace: as a state develops its industrial capabilities,

it becomes less likely to involve itself in violent international conflict. Our first hypothesis is,

therefore, a monadic one.

Hypothesis 1. A state’s level of industrialization will be negatively related to the probability that it

participates in a violent international conflict.

However, the most powerful explanations for international peace (the democratic and capitalist

peace theories) operate only at the dyadic level. In order to ensure that our results are not

somehow driven by one of these mechanisms, it is necessary to analyze dyadic-level data as well.

In so doing, we can account for the e�ects of joint democracy and joint capitalism in determining

the role played by industrialization. It should be noted that, while dyadic theories employ a “weak

link” argument, concerning themselves primarily with the lower of the two scores, an explanation

that is grounded in monadic e�ects should focus on the greater of the two. In other words, dyadic

explanations are e�ectively necessary conditions—both states must be democratic for democratic

peace theory to hold—while monadic explanations are su�cient conditions—if either state has

industrialized, then an industrial peace should hold.

Hypothesis 2. The level of industrialization of the more-heavily-industrialized state within a dyad

will be negatively related to the probability that the dyad enters into violent conflict.

Research Design

We examine our hypotheses with two di�erent units of analysis. As our first hypothesis is

monadic, we look at all countries between 1816 and 2001. Our second hypothesis requires a dyadic

approach, and so we examine all politically relevant dyads between 1816 and 2001.2 After removing

2The rationale for restricting the sample only to politically relevant dyads has been discussed in depth within theliterature (see, e.g., Maoz and Russett 1993; Choi 2011).

9

incomplete observations, our datasets range from 1960 to 1999.3 While this is a fairly short period,

relative to similar work, there are advantages to using a more compact era. In particular, when

examining several centuries, an analyst must specify a single theoretical mechanism and set of

empirical proxies for the entire period. Given the changing nature of the world and of international

conflict, this can be di�cult. By confining the analysis to a shorter era, we increase the likelihood

that our theory and measures are applicable across the entire period (see Rosenbaum 1999). We

see this as an important advantage for our analysis.4

For both the monadic and dyadic components of our study, we are interested in whether there

is a relationship between industrialization and violent conflict. For this reason, our dependent

variable takes a value of 1 if, during the year in question, the state or dyad becomes involved

in a new dispute that results in at least one fatality, and 0 otherwise. We choose to restrict

our analysis to fatal MIDs because our theoretical mechanism says nothing about the origins of

disputes. Rather, it suggests that industrialization should reduce the likelihood of violent conflict.

This operationalization is consistent with recent work on liberalism and peace (e.g., Hegre 2000;

Mousseau, Hegre and Oneal 2003; Gartzke and Weisiger N.d.), and averts the problems associated

with the reporting of many non-fatal disputes (see Weeks and Cohen 2006). Our measure of

fatal MID onset comes from the Militarized Interstate Disputes dataset (Ghosn, Palmer and Bremer

2004).5

Our primary variable of interest is the structure of a state’s economy. This goes beyond simply

development. Rather, we are interested in the types of endeavors in which the state is involved. In

particular, we want to distinguish states who invest heavily in manufacturing and industry from

wealthy “resource-rich states,” who gain primarily from selling raw materials (and thus cannot

3The dyadic dataset is shorter, due to a lack of data for some of our dyadic control variables after 1993.4Of course, it is not unreasonable to wonder about the robustness of our results, especially if the 1960–1999

period is unique in some way. Furthermore, if missingness is not independent of key explanatory variables, there is alegitimate concern that it could be driving our results. Indeed, one of Dafoe’s (2011) key criticisms of Gartzke’s (2007)analysis is that the listwise deletion of missing data significantly altered his results. To ensure that our findings arenot simply an artifact of data missingness, we have analyzed imputed datasets, covering all countries and politically-relevant dyads over both the time spans considered here and the entire range of the MID dataset (1816–2001), usingAmelia II in R (Honaker, King and Blackwell 2008). The details and results of these analyses are available in theappendix, and with respect to industrialization, are consistent with those presented below.

5All MID data were generated using EUGene v.3.204 (Bennett and Stam 2000).

10

profitably trade for them). We expect the behavior of these types of states to be significantly

di�erent. We operationalize this di�erence by looking at the contribution to each state’s GDP by

its industrial sector in a given year (World Bank 2013). We also include a measure of the size of

the service sector, as it is possible that states with service-oriented economies will also behave

di�erently. In our monadic analysis, we use the appropriate measure for each country-year; in

the dyadic section, because we theorize that the e�ect operates at the country level, we use the

greater of the two states’ industrialization levels. With respect to the service sector, we use the

size of the service sector for the state with the higher level of industrialization (agriculture serves

as the omitted category).

At both the monadic and dyadic levels, we control for democracy, using Polity scores (Marshall

and Jaggers 2002), and capitalism. In the dyadic case, we include both the lower score, measuring

joint democracy (Dixon 1993), and the higher, which measures regime dissimilarity (Oneal and Ray

1997). There is much less of a consensus, however, about what constitutes capitalism. Scholars

have o�ered a number of possible measures. Gartzke (2007) primarily defines capitalism in terms

of market openness, as measured by the International Monetary Fund (IMF), which proxies for the

degree to which the government engages in laissez-faire policies. McDonald (2010) takes a more

direct approach to government involvement, examining the distribution of property ownership

within a state, using IMF data on levels of governmental nontax revenue. Finally, Mousseau (2012)

proxies for capitalist norms, using World Bank data on per capita life insurance contracts in force

in each state. These measures all have their own merits as proxies for capitalism, and so we

examine a number of di�erent models, controlling for each measure in turn. This allows us to

assess the robustness of our results, with respect to various operationalizations of a key concept.

Studies of the capitalist peace, like those of the democratic peace before it, suggest that the

theoretical mechanism is dyadic in nature. Therefore, we adopt the “weak link” approach here, as

with democracy, using the lower dyadic capitalism score for each measure.

In addition to economic structure, democracy, and capitalism, there are a number of other

variables that potentially a�ect the likelihood of conflict onset. At the monadic level, we control

11

for the state’s development (logged GDP per capita), material capabilities, major power status, total

trade (in logged millions of U.S. dollars) in a given year, and whether it was a nuclear power. At the

dyadic level, we use Gartzke’s model as a baseline, controlling for dyadic economic development,

using the smaller (logged) GDP per capita for the dyad; the log of the stronger state’s share of

dyadic CINC scores; whether either state was a major power; the lower level of dyadic economic

dependence (dyadic trade divided by GDP); whether the two states are contiguous (up to 25 miles

of water); the (logged) distance between the states; whether the two states were involved in an

alliance; the total trade (in logged millions of U.S. dollars) between the two states in a given year;

Gartzke’s measure of UN voting similarity, using Signorino and Ritter’s (1999) S-scores; and whether

either state possessed nuclear weapons. The majority of these data come from the MID dataset,

while trade measures come from the Correlates of War (COW) dyadic trade dataset (Barbieri, Keshk

and Pollins 2009), and GDP comes from the Maddison project (Bolt and van Zanden 2013). To

account for the possibility of reverse causality, we lag time-varying covariates by one year.

Finally, Gartzke (2007) argues that there may be unobserved heterogeneity at the region level.

For this reason, he includes region dummies. We do the same, but we use the more traditional

COW coding for regions, creating dummy variables for North America, South America, Asia,

Europe, Africa, and the Middle East. At the monadic level, we simply indicate the region to which

the state belongs (Middle East is the omitted category). At the dyadic level, following Gartzke, we

code whether both states are in a given region.

As the dependent variable is dichotomous in both cases, we employ a logit framework, con-

sistent with previous research. Given the structure of the data, we are faced with two additional

problems, both of which could bias standard errors. The first concerns the fact that we observe

particular observations repeatedly over time. This generates the possibility of correlation within

groups that is not explicitly modeled. To deal with this, we cluster standard errors on countries at

the monadic level and on dyads at the dyadic level.

The second issue is one of temporal dependence: as peacetime progresses, states may become

more or less predisposed toward peace, irrespective of the covariates included in the model.

12

One method for treating temporal dependence is to use natural cubic splines to create a smooth

function of time (Beck, Katz and Tucker 1998). This approach has two disadvantages. First, splines

are unintuitive and are di�cult to interpret. For this reason, the substantive e�ect of time is

often unclear to the reader. Second, an appropriate implementation of the natural cubic spline

requires the selection of an appropriate number and placement of knots, which is a challenging

task, and one for which there is little clear guidance. The improper use of time splines can have a

significant confounding e�ect upon statistical inference (see Dafoe 2011). For these reasons, we opt

instead to use Carter and Signorino’s (2010) cubic polynomial approach, which has been shown to

provide results that are similar to optimally-implemented splines, while at the same time requiring

fewer additional assumptions and being relatively easy to interpret.6 We use t (time since the last

fatal conflict), t 2, and t 3 in all of the models below.7

Analysis

We are interested in determining both the monadic- and dyadic-level e�ects of industrialization.

To this end, we examine a series of models at two di�erent levels of analysis. We begin by looking

at how economic structure a�ects the probability of fatal MID onset at the country-year level, and

then move to an analysis at the dyad-year level.

Table 1 provides the results from the state-level analysis. Notably, beyond the controls for time

and region, relatively few variables have significant explanatory power. Although signed correctly,

neither the polity score nor the various proxies for capitalism attain statistical significance at the

p < .10 threshold in any model.8 Additionally, log-likelihood tests on restricted samples show that

the baseline model is preferred to all three models that include capitalism. This is no surprise, as

the theoretical mechanisms underlying both the democratic and capitalist peaces operate at the

6Our results are robust to the use of splines.7The range of the t 3 variable, relative to the ranges of other variables can introduce problems. We handle this by

rescaling the polynomial.8One concern is that industrialization is tapping the same concept as democracy or capitalism. We provide

empirical evidence that this is unlikely in the appendix.

13

Table 1: Country-year determinants of fatal conflictModel 1 Model 2 Model 3 Model 4(Baseline) (McDonald) (Gartzke) (Mousseau)

Industrial Sector −0.03∗∗∗ −0.04∗∗∗ −0.03∗∗ −0.03∗∗∗

(0.01) (0.01) (0.01) (0.01)Service Sector −0.01 0.01 −0.01 −0.01

(0.01) (0.02) (0.01) (0.01)GDP 0.09 0.08 0.06 −0.01

(0.20) (0.31) (0.34) (0.25)Capabilities 21.20∗∗∗ 24.26∗∗ 25.51∗∗∗ 19.47∗∗

(7.38) (9.79) (7.42) (7.83)Major Power −1.23 −0.84 −1.34 −1.15

(0.79) (1.01) (1.03) (0.88)Polity −0.00 −0.02 −0.00 −0.01

(0.02) (0.02) (0.02) (0.02)Trade 0.09 0.10 0.21∗ 0.15∗

(0.07) (0.11) (0.11) (0.08)Nuclear Weapons 0.34 −0.22 −0.11 0.31

(0.54) (0.64) (0.58) (0.57)North America −0.89∗ −1.73∗∗∗ −0.52 −0.91∗

(0.48) (0.63) (0.60) (0.49)South America −1.07∗∗∗ −1.28∗∗∗ −1.20∗∗ −1.09∗∗∗

(0.39) (0.48) (0.55) (0.38)Asia −0.21 −0.21 0.36 −0.22

(0.40) (0.44) (0.48) (0.42)Europe −0.45 −1.03 −0.45 −0.65

(0.55) (0.77) (0.76) (0.57)Africa −0.43 −0.45 0.12 −0.44

(0.41) (0.45) (0.47) (0.45)Time −2.08∗∗∗ −2.12∗∗∗ −1.92∗∗∗ −2.01∗∗∗

(0.33) (0.39) (0.45) (0.33)Time2 0.41∗∗∗ 0.38∗∗∗ 0.38∗∗ 0.39∗∗∗

(0.10) (0.12) (0.15) (0.10)Time3 −0.02∗∗∗ −0.02∗∗ −0.02∗ −0.02∗∗∗

(0.01) (0.01) (0.01) (0.01)Public Property 0.00

(0.01)Capital Openness −0.07

(0.07)Contract Intensiveness −0.00

(0.11)Constant −0.93 −1.25 −1.68 −0.76

(1.32) (1.78) (1.91) (1.58)N 3,272 1,994 2,088 3,115Log-likelihood −646.25 −342.09 −344.76 −605.72Cluster robust standard errors in parentheses.∗p < .10; ∗∗p < .05; ∗∗∗p < .01. All tests are two-tailed tests.

14

dyadic level. That their estimated e�ects are indistinguishable from zero is neither unexpected nor

problematic. It is, however, surprising that few other variables have any e�ect on the likelihood

of fatal conflict. Capabilities have the expected strong positive impact on the probability of war.

Foreign trade is also estimated to have a positive e�ect on the probability of war, though the

estimates vary across models, and only attain significance at the p < .10 level in two of the

models. The region-level e�ects are generally negative, indicating a lower probability of conflict,

relative to the Middle East, but only significant for the Americas. The results for time are consistent

with an inverse-N pattern, for which the estimated e�ect of time is always negative, from the first

year onward, implying that states that have been at peace are more likely to remain so.9

Turning to the economic structure variable, in all four models, industrialization has a robust

and negative e�ect on the likelihood of conflict, indicating strong support for our hypothesis

that industrialized states have a reduced likelihood of being involved in fatal disputes, relative to

less industrialized states. Across all four models, an increase in the level of industrialization is

associated with a significant reduction in the likelihood of fatal conflict. Interestingly, while the

estimate for the size of the service sector is generally negative, it never attains significance. This

suggests that, unlike industrialization, the shift from an agricultural to a service-based economy

will not significantly alter the probability that a deadly conflict occurs. Thus, at the monadic level,

there appears to be something peculiar to industrialization itself that makes war less likely. We

will return to this possibility later.

In addition to the statistical results, we find it useful to examine substantive results. After all,

statistical inference is a function of precision as well as importance. Thus, it is appropriate to

look at the estimated e�ects of changes in industrialization, democracy, and the three capitalism

variables, across our four models. Table 2 displays the results of first di�erence calculations across

the variables of interest. As our model is non-linear, we hold all other variables at an appropriate

profile of values. We opt to set all non-dichotomous controls—with the exception of the industrial

(where relevant) and service variables, which we set to 30%—to their (approximate) median values,

9Plots of the e�ect of time on the linear predictor can be found in the supplementary appendix.

15

Model Variable Change Model Variable Change

North America EuropeBaseline Industrialization −0.026∗∗∗ Baseline Industrialization −0.032∗∗∗

Baseline Democracy 0.001 Baseline Democracy 0.001McDonald Industrialization −0.029∗∗∗ McDonald Industrialization −0.024∗∗∗

McDonald Democracy −0.000 McDonald Democracy −0.000McDonald Capitalism 0.005 McDonald Capitalism 0.004Gartzke Industrialization −0.045∗∗∗ Gartzke Industrialization −0.034∗∗

Gartzke Democracy −0.004 Gartzke Democracy −0.003Gartzke Capitalism −0.001 Gartzke Capitalism −0.000Mousseau Industrialization −0.031∗∗∗ Mousseau Industrialization −0.033∗∗∗

Mousseau Democracy 0.001 Mousseau Democracy 0.001Mousseau Capitalism −0.000 Mousseau Capitalism −0.001South America AfricaBaseline Industrialization −0.017∗∗∗ Baseline Industrialization −0.022∗∗∗

Baseline Democracy 0.001 Baseline Democracy 0.001McDonald Industrialization −0.021∗∗∗ McDonald Industrialization −0.041∗∗∗

McDonald Democracy −0.000 McDonald Democracy −0.001McDonald Capitalism 0.003 McDonald Capitalism 0.007Gartzke Industrialization −0.016∗∗ Gartzke Industrialization −0.037∗∗

Gartzke Democracy −0.002 Gartzke Democracy −0.004Gartzke Capitalism −0.000 Gartzke Capitalism −0.000Mousseau Industrialization −0.022∗∗∗ Mousseau Industrialization −0.025∗∗∗

Mousseau Democracy 0.000 Mousseau Democracy 0.000Mousseau Capitalism −0.000 Mousseau Capitalism −0.000Asia Middle EastBaseline Industrialization −0.032∗∗∗ Baseline Industrialization −0.047∗∗∗

Baseline Democracy 0.001 Baseline Democracy 0.002McDonald Industrialization −0.041∗∗∗ McDonald Industrialization −0.044∗∗∗

McDonald Democracy −0.001 McDonald Democracy −0.001McDonald Capitalism 0.007 McDonald Capitalism 0.008Gartzke Industrialization −0.049∗∗ Gartzke Industrialization −0.062∗∗

Gartzke Democracy −0.005 Gartzke Democracy −0.006Gartzke Capitalism −0.001 Gartzke Capitalism −0.001Mousseau Industrialization −0.038∗∗∗ Mousseau Industrialization −0.054∗∗∗

Mousseau Democracy 0.001 Mousseau Democracy 0.001Mousseau Capitalism −0.001 Mousseau Capitalism −0.001∗p < .10; ∗∗p < .05; ∗∗∗p < .01. All tests are two-tailed tests.

Table 2: First Di�erences by Region

16

and the time since previous conflict to one year. We set dichotomous variables to zero. We run

six separate sets of first di�erences, in order to examine the e�ects in each of the six regions. The

values represent the change in predicted probability of conflict onset for a shift of one standard

deviation about the mean.10 Summary statistics for each of the variables of interest can be found

in the appendix.

Across all regions and models, a shift from a half-standard-deviation below the mean of

industrialization to a half-standard-deviation above it results in a meaningful and statistically

significant change in the predicted probability of fatal conflict onset. Importantly, not only are

the e�ects of the democracy and capitalism variables still insignificant, but their e�ects are

substantially smaller than the e�ect of industrialization. In every case, the absolute e�ect of a

shift in industrialization is several times larger than the corresponding shift in either of the other

variables. Moreover, the e�ects or both democracy and capitalism are e�ectively zero (with a

maximum of 0.007 for public property in Africa) in all instances, indicating a lack of even an

imprecisely measured substantive e�ect.

Our analysis at the monadic level provides significant support for our first hypothesis, and

gives us confidence in the workings of our theory. It suggests that industrialization plays a key

role in international relations, and one that is substantially greater than either democracy or

economic liberalism. However, the democratic and capitalist peace theories are dyadic in nature;

it is unfair to compare them to our industrialization mechanism at a monadic level. Indeed, if

our mechanism is at play on the international stage, then our results should be robust to a dyadic

level of analysis, given that we operationalize the variable appropriately.

Table 3 displays the results from our dyadic analysis. While dyadic democracy is correctly

signed, it is never significant. The higher democracy score, however, attains significance in each

model, and is, as expected, positively related to the probability of conflict onset. With respect

to the capitalism variables, none of the three attains significance at conventional levels, and the

estimates for capital openness and contract intensiveness are actually signed incorrectly. As in

10We use a parametric bootstrap with 1,000 draws to simulate confidence intervals.

17

Table 3: Dyad-year determinants of fatal conflictModel 1 Model 2 Model 3 Model 4(Baseline) (McDonald) (Gartzke) (Mousseau)

Industrial SectorH −0.04∗∗ −0.08∗∗∗ −0.07∗ −0.05∗

(0.02) (0.03) (0.04) (0.02)Service Sector −0.03 −0.00 −0.03 −0.04

(0.03) (0.04) (0.03) (0.03)GDPL −0.86∗ −0.17 −0.58 −1.32∗

(0.52) (0.56) (0.55) (0.68)Capability Ratio (logged) −1.32 −2.18∗∗∗ −0.81 −0.42

(1.30) (0.81) (1.50) (1.45)Either Major Power −0.22 −17.56∗∗∗ −16.88∗∗∗ −0.01

(0.62) (2.61) (1.43) (0.71)DependenceL −1.94 −0.77 −1.90 −1.37

(1.53) (0.74) (1.39) (1.14)PolityL −0.04 −0.10 −0.03 −0.05

(0.04) (0.08) (0.05) (0.04)PolityH 0.11∗∗ 0.13∗ 0.11∗ 0.11∗∗

(0.05) (0.07) (0.06) (0.05)Contiguity 8.58 −9.01 −10.56 9.04

(6.19) (10.07) (7.60) (6.60)Distance (logged) 0.83 0.80 0.59 0.88

(0.77) (1.34) (0.95) (0.83)Allies 0.09 −0.12 −0.10 0.18

(0.37) (0.48) (0.31) (0.35)Trade 0.24∗∗ 0.06 0.21 0.16

(0.11) (0.24) (0.13) (0.10)S-score −0.55 −1.48 0.70 −0.66

(0.74) (1.89) (1.35) (0.78)Nuclear Weapons 0.37 1.21 1.28∗∗ 0.62

(0.39) (0.93) (0.63) (0.41)North America 3.41∗∗ 4.45∗∗ 3.94∗∗ 3.47∗∗

(1.47) (1.75) (1.55) (1.46)South America −15.00∗∗∗ −14.20∗∗∗ −15.01∗∗∗ −14.46∗∗∗

(1.29) (1.65) (1.32) (1.31)Asia 2.31∗∗ 3.87∗ 2.54∗∗ 2.51∗∗

(0.97) (2.03) (1.26) (1.12)Europe −11.59∗∗∗ −11.65∗∗∗ −12.53∗∗∗ −11.43∗∗∗

(1.38) (1.91) (1.50) (1.68)Africa 2.31∗∗ 4.00∗ 3.31∗∗ 2.63∗∗

(1.13) (2.37) (1.55) (1.22)Middle East 2.83∗∗ 2.30 4.20∗∗∗ 4.08∗∗

(1.22) (1.63) (1.49) (1.60)Time −15.34∗∗∗ −13.96∗∗∗ −15.33∗∗∗ −16.32∗∗∗

(4.57) (3.93) (4.41) (4.71)Time2 2.02∗∗ 1.78∗∗∗ 2.08∗∗∗ 2.11∗∗∗

(0.82) (0.68) (0.80) (0.82)Time3 −0.07∗ −0.06∗∗ −0.07∗ −0.07∗

(0.04) (0.03) (0.04) (0.04)Public Property 0.02

(0.02)Capital Openness 0.12

(0.13)Contract Intensiveness 0.52

(0.37)Constant −5.17 5.75 10.60 −2.46

(7.60) (11.25) (9.74) (8.32)N 10,527 5,320 8,983 9,923Log-likelihood −108.59 2.56 −62.71 −90.55Cluster robust standard errors in parentheses.∗p < .10; ∗∗p < .05; ∗∗∗p < .01. All tests are two-tailed tests.

18

the monadic case, log-likelihood tests on restricted samples indicate a preference for the baseline

model. Industrialization, on the other hand, is correctly signed and attains significance at the

p < .10 level or better in all four models. This supports Hypothesis 2, and provides us with

greater confidence in our theory, as it appears to operate at both the monadic and dyadic levels.

The service sector, once again, exerts a negative e�ect on the likelihood of conflict, but the

measure is not su�ciently precise to conclude that this e�ect is greater than zero. Once again,

while industrialization matters, the service sector seems to be irrelevant, suggesting that this is

more than simply an economic growth or transformation argument. Rather, industrialization has

qualities that render it e�ective at dampening the prospects of war. Finally, the e�ect of time

is again an inverse-N shape, which is always estimated to be negative. Thus, peace again fosters

peace.

Some of the estimates in Table 3 vary widely (e.g., major power status). Given these issues, it is

conceivable that the model is not well identified. This is potentially due to near multicollinearity

among covariates. Indeed, a generalized version of the variance inflation factor shows severe

multicollinearity (V I F > 200) in a number of coe�cients. Removing the o�ending variables

provides more stable and intuitive coe�cients for the control variables. The results for democracy,

industrialization, and capitalism remain mostly unchanged (public property attains significance at

the p < .10 level). While we present the full model here, using standard controls, the results of

the restricted models are included in the appendix.

Given that the e�ects that we find for the variables of primary interest do not seem to be a

result of the significant multicollinearity in the models in the table above, we can be relatively

confident in the results. The e�ect of democracy is not statistically distinguishable from zero, nor

are the e�ects of capitalism. Industrialization, on the other hand, does have a meaningful e�ect

on the likelihood of fatal conflict. It is important, however, to examine not only the direction and

precision of our estimates, but also their substantive magnitudes. We turn now to an analysis of

the substantive results of Table 3.

The dyad-level substantive results are depicted in Table 4. Again, we use a one standard

19

Model Variable Change Model Variable Change

North America EuropeBaseline Industrialization −0.101∗∗ Baseline Industrialization −0.000∗∗

Baseline Democracy −0.038 Baseline Democracy −0.000McDonald Industrialization −0.121∗∗∗ McDonald Industrialization −0.000∗∗∗

McDonald Democracy −0.118 McDonald Democracy −0.000McDonald Capitalism 0.046 McDonald Capitalism 0.000Gartzke Industrialization −0.135∗∗ Gartzke Industrialization −0.000∗

Gartzke Democracy −0.032 Gartzke Democracy −0.000Gartzke Capitalism 0.028 Gartzke Capitalism 0.000Mousseau Industrialization −0.108∗ Mousseau Industrialization −0.000∗

Mousseau Democracy −0.046 Mousseau Democracy −0.000Mousseau Capitalism 0.089 Mousseau Capitalism 0.000South America AfricaBaseline Industrialization −0.000∗∗ Baseline Industrialization −0.077∗∗

Baseline Democracy −0.000 Baseline Democracy −0.061McDonald Industrialization −0.000∗∗∗ McDonald Industrialization −0.090∗∗∗

McDonald Democracy −0.000 McDonald Democracy −0.145McDonald Capitalism 0.000 McDonald Capitalism 0.061Gartzke Industrialization −0.000∗ Gartzke Industrialization −0.097∗

Gartzke Democracy −0.000 Gartzke Democracy −0.045Gartzke Capitalism 0.000 Gartzke Capitalism 0.042Mousseau Industrialization −0.000∗∗ Mousseau Industrialization −0.091∗∗

Mousseau Democracy −0.000 Mousseau Democracy −0.071Mousseau Capitalism 0.000 Mousseau Capitalism 0.154Asia Middle EastBaseline Industrialization −0.078∗ Baseline Industrialization −0.094∗∗

Baseline Democracy −0.061 Baseline Democracy −0.052McDonald Industrialization −0.083∗∗∗ McDonald Industrialization −0.021∗∗∗

McDonald Democracy −0.151 McDonald Democracy −0.139McDonald Capitalism 0.066 McDonald Capitalism 0.087Gartzke Industrialization −0.055∗ Gartzke Industrialization −0.148∗∗

Gartzke Democracy −0.057 Gartzke Democracy −0.026Gartzke Capitalism 0.057 Gartzke Capitalism 0.023Mousseau Industrialization −0.087∗ Mousseau Industrialization −0.099∗∗

Mousseau Democracy −0.074 Mousseau Democracy −0.030Mousseau Capitalism 0.164 Mousseau Capitalism 0.054Di�erent RegionsBaseline Industrialization −0.013∗

Baseline Democracy −0.030McDonald Industrialization −0.002∗∗∗

McDonald Democracy −0.026McDonald Capitalism 0.021Gartzke Industrialization −0.005∗∗

Gartzke Democracy −0.019Gartzke Capitalism 0.023Mousseau Industrialization −0.013∗∗

Mousseau Democracy −0.037Mousseau Capitalism 0.141∗p < .10; ∗∗p < .05; ∗∗∗p < .01. All tests are two-tailed tests.

Table 4: First Di�erences by Region20

deviation shift about the mean for each of the variables of interest. We generally set continuous

and ordered variables to their approximate median values in the dataset. As before, we set the

size of the economic sectors to 30%. Our profile assumes contiguous states (thus, zero distance),

and we set other dummy variables to zero. Finally, as before, we assume that the dyad is in its

first year of peace.

The dyadic results are somewhat di�erent than their monadic counterparts. While the statis-

tical di�erences between industrialization and the indicators of political and economic liberalism

are substantial, the substantive di�erences are not. The e�ect of a one-unit shift in industrial-

ization about its mean is very similar to the same shift for democracy or capitalism. In fact,

in many cases, the e�ect of democracy is larger than that of industrialization. Additionally, the

largest estimated e�ect in the table is the contract intensiveness variable, which is expected to

increase (rather than decrease) the probability of fatal conflict onset by 0.164 in Africa. While

our estimates of the e�ects of the liberalism variables are not precise enough to be certain that

they are non-zero, the estimated e�ects are large enough to be worth noting. Given the size of

the e�ects, despite the lack of statistical significance, it is di�cult to discount a possible role for

liberalism.

Both the statistical and substantive results provide support for our hypotheses, demonstrating

the impact that industrialization can have on the likelihood of fatal conflict onset. While neither

democracy nor capitalism has any e�ect on the probability of war at the monadic level, and

both have strong but uncertain e�ects upon dyads, industrialization has a large, significant, and

robust dampening e�ect at both levels of analysis, even when we control for various measures of

liberalism. Taken together, this suggests that economic structure plays a critical role in the conflict

behavior of states.

21

Conclusion

This paper began by proposing a mechanism for international peace based on the writings of

Joseph Schumpeter. We theorized that industrialization can have a pacifying e�ect upon a state’s

foreign policy. Our analysis demonstrates that this argument is consistent with the historical

record. Over the period analyzed above, states with higher levels of industrialization were less

likely to become involved in fatal militarized disputes than were their less-heavily-industrialized

counterparts. These results suggest that the “conquest pays” argument (applied to industrialized

societies) does not fit the empirical evidence, at least within the period considered here. More

importantly, our hypotheses hold at both the monadic and dyadic levels, even when controlling

for political and financial liberalism, and accounting for missing data. This suggests that indus-

trialization has an e�ect that is independent of liberalism, and may explain recent shifts to more

peaceful foreign policy by authoritarian powers, such as China (see Kurlantzick 2007). Further-

more, unlike democratization, from which it is easy to backslide into authoritarianism (as has

occurred repeatedly throughout Pakistan’s history, for example), states that have industrialized do

not revert to pre-modern, agrarian societies (see Huntington 1971, 290).

The argument that industrialization engenders a shift to a more peaceful mode of interstate

conduct is concordant with Schumpeter’s general claims about economic progress and war, in

which he noted that “[i]mperialism is [. . . ] atavistic in character” (Schumpeter 1955, 65). Further-

more, as Jervis (2005, 22) points out, “[t]raditional liberal thought [. . . ] stressed that economic

activity was so potent [. . . ] because it reconstructed social values to downgrade status and glory

and elevate material well-being.” This suggests a possible linkage between industry, democracy,

and capitalism, such that democratic and capitalist norms can be generated by a shift to industri-

alization. Thus, over time, the material roads to peace fostered by industrialization may give way

to immaterial causes. Some preliminary evidence for such a shift is that, although both material

and immaterial factors remain important today, the most common motivations for fighting have

changed considerably since the Industrial Revolution: material issues (territorial and economic

reasons) have waned, while immaterial issues (ideological, human sympathy) have waxed (Holsti

22

1991; Freedman 2005). While testing this hypothesis is beyond the scope of this paper, we note

that our results are not inconsistent with the attitudinal framework put forward by these authors

or by liberal peace theorists (e.g., Maoz and Russett 1993; Mousseau 2013), as we find evidence

that peace begets peace, which suggests that the peace that is fostered by industrial development

should be a lasting one. We invite future researchers to examine this relationship more directly.

A related observation is that while industrialization has a robust and significant e�ect on

conflict, the size of the service sector never attains significance. This may be the result of a

post-materialist value shift (see, e.g., Inglehart 1997), following the transition to a less-materialist,

service-based economy. This could facilitate a greater emphasis on ideology, such that a state

simply becomes involved in political and humanitarian conflicts, rather than interest-based dis-

putes. Thus, post-materialist, service-oriented states may be no less likely to fight, in general, than

materialist, agricultural economies; rather, the issues over which conflicts occur would change.

Investigating this mechanism is also beyond the scope of this paper, but we encourage others to

investigate the possibility.

While our results are robust and substantively powerful, this analysis is a first investigation into

a new theory. Thus, we o�er a few important caveats. The first concerns the temporal e�ects of

industrialization on war. Schumpeter’s argument is not that development will lead to a complete

cessation of international conflict, but that the shift to industrialization reduces the material

incentive to wage war. His main argument draws upon the economic motivation for warfare:

since the Industrial Revolution, he argues, few wars have been waged for material gain. While it

is always true that “many people stand to gain economically in any war” (Schumpeter 1955 [1919],

75), the spread of industrialization—especially high-tech, knowledge-based industrialization—has

rendered war unprofitable in more general terms. Aggressive, expansionist policies, then, may

remain as a legacy of the past, and are thus primarily the domain of the pre-industrialized state.

To the extent that industrialized states engage in such behavior, they are driven by habit and

inertia, rather than real and rational interest.

Second, and relatedly, our claims about the pacifying e�ects of industrialization are less expan-

23

sive than those of democratic peace proponents. In particular, we do not claim that industrialized

states never fight one another. Indeed, “atavistic” ideologies can be powerful factors that fan the

flames of conflict, even for relatively modern societies (see Schumpeter 1955, 65). This means

that imperialist or expansionist ideologies may still emerge—and, in fact, these were the primary

motivations for the two world wars, fought by industrial powers in the first half of the twentieth

century—despite any changes brought on by the shift to industrialization, as in the case of Britain

during the nineteenth century or Germany and Japan before World War II. Nonetheless, we suggest

that industrialization removes a key incentive for war, sharply reducing the likelihood that conflict

erupts.

Third, our primary analysis covers a relatively limited period, from 1960 through the 1990s.

An advantage of this design is that it increases the likelihood that the same mechanisms are

operational throughout the entire period being studied. A disadvantage is that we cannot be

certain that out results will hold outside of this range, though our supplementary analysis with

multiple imputation suggests that it is plausible. As more data become available, future research

should extend the period of analysis, and determine whether the theory continues to receive

empirical support

Finally, with respect to our dependent variable, we stress that our results concern fatal disputes,

rather than disputes in general. Industrialization does not necessarily reduce the probability that

states will become involved in conflict; instead, it makes peaceful resolution of conflicts more

likely (i.e., it decreases the probability that states actually engage militarily).

Our analysis also o�ers several policy implications. First, because the relationship between in-

dustrial development and peace is independent of political regime, our results suggest that, when

confronting other industrialized countries, Western democracies should engage them diplomati-

cally and economically (rather than militarily), regardless of the latter’s regime type. This could

be particularly applicable with respect to the emerging confrontation between the U.S. and China.

Second, the results provide both theoretical and practical implications for reducing international

conflict and promoting stability, especially in war-prone regions. Theoretically, it suggests that

24

encouraging economic development in underdeveloped countries might be better than fostering

democracy. While established democracies may share peaceful norms, focusing on democracy

first may not be optimal. As European history has shown, freedom takes time to consolidate,

and exporting the Western political model to non-Western areas has caused unexpected compli-

cations. Wealth and quality of life, on the other hand, are more universal, and might be attractive

even in autocracies and dictatorships. Practically, the model shows that peace can be facilitated

through industrialization. While nations accrue wealth from several sources (agriculture, finance,

exploitation of natural resources, and even foreign aid), industrial development alone can strongly

encourage peace, as “energy for war becomes energy for labor” (Schumpeter 1955, 69). Moreover,

because industry inherently requires continuous innovation (creative destruction), industrialized

nations will necessarily be presented with an investment-war tradeo�, in which they can profit

most by choosing economic development.

25

References

Babst, Dean V. 1972. “A Force for Peace.” Industrial Research 14(4):55–58.

Barbieri, Katherine. 1996. “Economic Interdependence: A Path to Peace or a Source of InterstateConflict?” Journal of Peace Research 33:29–49.

Barbieri, Katherine, Omar M.G. Keshk and Brian M. Pollins. 2009. “Trading Data: Evaluating ourAssumptions and Coding Rules.” Conflict Management and Peace Science 26(5):471–491.

Beck, Nathaniel, Jonathan N. Katz and Richard Tucker. 1998. “Taking time seriously: Time-series-cross-section analysis with a binary dependent variable.” American Journal of Political Science42(4):1260–1288.

Bennett, D. Scott. 2006. “Toward a Continuous Specification of the Democracy–Autocracy Connec-tion.” International Studies Quarterly 50(2):313–338.

Bennett, D. Scott and Allan C. Stam. 2000. “EUGene: A Conceptual Manual.” International Interac-tions 26:179–204.

Bolt, Jutta and Jan Luiten van Zanden. 2013. “The First Update of the Maddison Project; Re-Estimating Growth Before 1820.” Maddison Project Working Paper 4.

Brawley, Mark R. 1993. “Regime Types, Markets, and War The Importance of Pervasive Rents inForeign Policy.” Comparative Political Studies 26(2):178–197.

Bueno de Mesquita, Bruce, Alastair Smith, Randolph Siverson and James Morrow. 2003. The Logicof Political Survival. Cambridge,MA: MIT Press.

Carter, David B. and Curtis S. Signorino. 2010. “Back to the future: Modeling time dependence inbinary data.” Political Analysis 18(3):271–292.

Choi, Seung-Whan. 2011. “Re-Evaluating Capitalist and Democratic Peace Models.” InternationalStudies Quarterly 55(3):759–769.

Choucri, Nazli and Robert Carver North. 1975. Nations in Conflict: National Growth and Interna-tional Violence. San Francisco: WH Freeman.

Dafoe, Allan. 2011. “Statistical Critiques of the Democratic Peace: Caveat Emptor.” American Journalof Political Science 55(2):247–262.

Dixon, William J. 1993. “Democracy and the Management of International Conflict.” Journal ofConflict Resolution 37(1):42–68.

Dorussen, Han and Hugh Ward. 2010. “Trade Networks and the Kantian Peace.” Journal of PeaceResearch 47(1):29–42.

Doyle, Michael W. 1983a. “Kant, Liberal Legacies, and Foreign A�airs.” Philosophy & Public A�airs12(3):205–235.

26

Doyle, Michael W. 1983b. “Kant, Liberal Legacies, and Foreign A�airs, Part 2.” Philosophy & PublicA�airs 12(4):323–353.

Freedman, Lawrence. 2005. “The Age of Liberal Wars.” Review of International Studies 31(1):93–107.

Gartzke, Erik. 2007. “The Capitalist Peace.” American Journal of Political Science 51(1):166–191.

Gartzke, Erik and Alex Weisiger. 2013. “Permanent Friends? Dynamic Di�erence and the Demo-cratic Peace.” International Studies Quarterly 57(1):171–185.

Gartzke, Erik and Alex Weisiger. N.d. “Under Construction: Development, Democracy, and Di�er-ence as Determinants of the Systemic Liberal Peace.” International Studies Quarterly. Forthcom-ing.

Gartzke, Erik and J. Joseph Hewitt. 2010. “International Crises and the Capitalist Peace.” Interna-tional Interactions 36(2):115–145.

Gat, Azar. 2005. “The Democratic Peace Theory Reframed: The Impact of Modernity.” WorldPolitics 58(1):73–100.

Gelpi, Christopher F. and Michael Griesdorf. 2001. “Winners or Losers? Democracies in Interna-tional Crisis, 1918-94.” American Political Science Review 95(3):633–648.

Ghosn, Faten, Glenn Palmer and Stuart Bremer. 2004. “The MID3 Data Set, 1993-2001: Procedures,Coding Rules, and Description.” Conflict Management and Peace Science 32(2):133–154.

Gleditsch, Nils Petter. 2008. “The Liberal Moment Fifteen Years On.” International Studies Quarterly52(4):691–712.

Glossop, Ronald J. 2001. Confronting war: An Examination of Humanity’s Most Pressing Problem.Je�erson, NC: McFarland & Company.

Hayes, Jarrod. 2012. “The democratic peace and the new evolution of an old idea.” EuropeanJournal of International Relations 18(4):767–791.

Hegre, Håvard. 2000. “Development and the Liberal Peace: What Does it Take to be a TradingState?” Journal of Peace Research 37:5–30.

Hegre, Håvard, Tanja Ellingsen, Scott Gates and Nils Petter Gleditsch. 2001. “Toward a DemocraticCivil Peace? Democracy, Political Change, and Civil War, 1816-1992.” American Political ScienceReview 95(1):33–48.

Holsti, Kalevi Jaakko. 1991. Peace and War: Armed Conflicts and International Order, 1648-1989.Vol. 14 Cambridge University Press.

Honaker, James, Gary King and Matthew Blackwell. 2008. “Amelia II: A Program for Missing Data.”Journal of Statistical Software 45(7):1–47.

Huntington, Samuel P. 1971. “The Change to Change: Modernization, Development, and Politics.”Comparative Politics 3(3):283–322.

27

Inglehart, Ronald. 1997. Modernization and Postmodernization: Cultural, Economic, and PoliticalChange in 43 Societies. Vol. 19 Cambridge: Cambridge Univ Press.

Jervis, Robert. 2005. American Foreign Policy in a New Era. New York, NY: Routledge.

Kaysen, Carl. 1990. “Is War Obsolete?: A Review Essay.” International Security 14(4):42–64.

Kurlantzick, Joshua. 2007. Charm O�ensive: How China’s Soft Power is Transforming the World. NewHaven, CT: Yale University Press.

Levitsky, Steven and Lucan Way. 2002. “The Rise of Competitive Authoritarianism.” Journal ofDemocracy 13(2):51–65.

Levy, Jack S. 1988. “Domestic Politics and War.” Journal of Interdisciplinary History 18:653–673.

Liberman, Peter. 1998. Does Conquest Pay?: The Exploitation of Occupied Industrial Societies. Prince-ton, NJ: Princeton University Press.

Mandelbaum, Michael. 2007. “Democracy Without America: The Spontaneous Spread of Freedom.”Foreign A�airs 86(5):119–130.

Mansfield, Edward D and Jack Snyder. 2002. “Incomplete Democratization and the Outbreak ofMilitary Disputes.” International Studies Quarterly 46(4):529–549.

Maoz, Zeev and Bruce Russett. 1992. “Alliance, Contiguity, Wealth, and Political Stability: Is theLack of Conflict among Democracies a Statistical Artifact?” International Interactions 17(3):245–267.

Maoz, Zeev and Bruce Russett. 1993. “Normative and Structural Causes of Democratic Peace,1946-1986.” American Political Science Review 87(3):624–638.

Marshall, Monty G. and Keith Jaggers. 2002. “Polity IV project: Politi-cal regime characteristics and transitions, 1800-2007.” Available on line at:http://www.systemicpeace.org/polity/polity4.htm .

McDonald, Patrick J. 2007. “The Purse Strings of Peace.” American Journal of Political Science51(3):569–582.

McDonald, Patrick J. 2010. “Capitalism, Commitment, and Peace.” International Interactions36(2):146–168.

Mearsheimer, John J. 2001. The Tragedy of Great Power Politics. New York: WW Norton & Company.

Morgan, T. Clifton and Sally Campbell. 1991. “Domestic Structure, Decisional Constraints, and War:So Why Kant Democracies Fight?” Journal of Conflict Resolution 35(2):187–211.

Morgan, T. Clifton and Valerie L. Schwebach. 1992. “Take Two Democracies and Call Me in theMorning: A Prescription for Peace?” International Interactions 17(4):305–320.

Mousseau, Michael. 2009. “The Social Market Roots of Democratic Peace.” International Security33(4):52–86.

28

Mousseau, Michael. 2010. “Coming to Terms with the Capitalist Peace.” International Interactions36(2):185–192.

Mousseau, Michael. 2012. “Capitalist Development and Civil War.” International Studies Quarterly56(3):470–483.

Mousseau, Michael. 2013. “The Democratic Peace Unraveled: It’s the Economy.” InternationalStudies Quarterly 57(1):186–197.

Mousseau, Michael, Håvard Hegre and John Oneal. 2003. “How the Wealth of Nations Conditionsthe Liberal Peace.” European Journal of International Relations 9(2):277–314.

North, Douglass C., John J. Wallis and Barry R. Weingast. 2009. Violence and Social Order. Cam-bridge, MA: Cambridge University Press.

Oneal, John and James Lee Ray. 1997. “New Tests of the Democratic Peace: Controlling forEconomic Interdependence, 1950-85.” Political Research Quarterly 50(4):751–775.

Peceny, Mark, Caroline Beer and Shannon Sanchez-Terry. 2002. “Dictatorial Peace?” AmericanPolitical Science Review 96:15–26.

Rasler, Karen A. and William R. Thompson. 2005. Puzzles of the Democratic Peace: Theory, Geopoli-tics, and the Transformation of World Politics. London: Palgrave Macmillan.

Rosecrance, Richard N. 1986. The Rise of the Trading State: Commerce and Conquest in the ModernWorld. New York: Basic Books.

Rosenbaum, Paul R. 1999. “Choice as an Alternative to Control in Observational Studies.” StatisticalScience 14(3):259–278.

Rummel, R.J. 1979. Understanding Conflict and War: War, Power, Peace. Beverly Hills, CA: Sage.

Schneider, Gerald and Nils Petter Gleditsch. 2010. “The Capitalist Peace: The Origins and Prospectsof a Liberal Idea.” International Interactions 36(2):107–114.

Schumpeter, Joseph. 1955. Imperialism & Social Classes. Vol. 4 Cleveland, OH: World Publishing Co.

Schumpeter, Joseph A. 1942. Capitalism, Socialism, and Democracy. New York, NY: Harper.

Schumpeter, Joseph A. 1955 [1919]. The Sociology of Imperialisms. In Imperialism & Social Classes.Cleveland, OH: World Publishing Co.

Signorino, Curtis S. and Je�rey M. Ritter. 1999. “Tau-b or Not Tau-b: Measuring the Similarity ofForeign Policy Positions.” International Studies Quarterly 43:115–144.

Souva, Mark and Brandon Prins. 2006. “The Liberal Peace Revisited: The Role of Democracy,Dependence, and Development in Militarized Interstate Dispute Initiation, 1950–1999.” Interna-tional Interactions 32(2):183–200.

Van Evera, Stephen. 1999. Causes of War: Power and the Roots of Conflict. Ithaca, NY: CornellUniversity Press.

29

Wainwright, Elsina. 2009. “Conflict Prevention in Southeast Asia and the SouthPacific.” Center on International Cooperation, New York University. Avail-able online at: http://www.humansecuritygateway.com/documents/CIC_ConflictPreventionInSoutheastAsiaAndTheSouthPacific.pdf .

Ward, Michael D. and Kristian S. Gleditsch. 1998. “Democratizing for Peace.” American PoliticalScience Review 92(1):51–61.

Weede, Erich. 1995. “Economic Policy and International Security: Rent-Seeking, Free Trade andDemocratic Peace.” European Journal of International Relations 1(4):519–537.

Weeks, Jessica and Dara Cohen. 2006. “Red Herrings? Fishing Disputes, Regime Type, and InterstateConflict.” Presented at the Annual Meeting of the International Studies Association, San Diego,CA, March 2006 .

World Bank. 2013. “World Development Indicators.” Available online at:http://data.worldbank.org/indicator/NV.IND.TOTL.ZS .

Zakaria, Fareed. 1997. “The Rise of Illiberal Democracy.” Foreign A�airs 76(6):22–43.

30