Sage Journals: Discover world-class research

Abstract

This study questions the widely supported claim that economic globalization benefits developing countries by enhancing their prospects for civil peace. Drawing on ideas from across the social sciences, this study questions the nature of causation between economic globalization and civil peace and how it should be conceptualized and modeled. Instead of assuming a straightforward and static relationship whereby economic globalization directly increases the likelihood of civil peace, a range of alternative hypotheses are considered, including those about reverse and reciprocal causation and about dynamic and path-dependent effects. Using dynamic panel regression and vector autoregression, the study empirically examines how trade openness and foreign direct investment affect civil violence in a sample of all countries classified by the United Nations as developing. The results find positive links between economic globalization and civil peace, as much of the literature predicts, but show that this relationship mainly arises from complex patterns of causation rather than from the pacifying effects of economic globalization. These results are robust to various model specifications and sampling strategies. The study ends by considering whether economic globalization may indirectly promote civil peace by fostering long-term economic development.

Keywords

civil conflict dynamic effects economic globalization liberal peace vector autoregression

Introduction

Presently, there is considerable support for the notion that developing countries can build and maintain civil peace by integrating their national economies into global markets for trade and investment. Broadly known as the “liberal peace,” this idea builds on a long-standing theoretical tradition, stemming from Kant (1970 [1795]), which depicts peace and political stability as positive externalities of international trade and commerce. Originally, advocates of the liberal peace emphasized that trade between states reduces the likelihood of war between them (Rosecrance, 1986), and that democratic societies almost never wage war against one another (Doyle, 1986). These ideas are supported by significant amounts of empirical evidence (e.g. Chen and Zhou, 2021; Gartzke, 2007; Gartzke and Li, 2003; Hegre et al., 2010; Oneal and Russett, 1997; cf. Keshk et al., 2010; Sadeh and Feldman, 2020). Furthermore, the pacifying effects of trade may extend beyond direct trading partners to include third-party countries within larger trading networks (Kinne, 2012, 2014). In this way, networks of trading states may generate positive externalities that promote international peace among economically integrated states. This study, however, does not question this proposition, but instead examines civil conflicts in developing countries and whether economic globalization generates pacifying effects in this more limited context.

With global market integration proceeding apace in most developing countries, the question of whether the liberal peace also holds for these countries is salient. On this topic, the dominant view is that global trade and investment not only generate economic benefits for developing countries but also lay the material and social foundations for an enduring civil peace (Mandelbaum, 2002; Weede, 2004). This theoretical idea is supported by many quantitative studies, which find negative associations between economic globalization and incidences of civil violence in the developing world (e.g. Barbieri and Reuveny, 2005; Blanton and Apodaca, 2007; Bussmann and Schneider, 2007; Flaten and de Soysa, 2012; Gissinger and Gleditsch, 1999; Krause and Suzuki, 2005). There are, however, some discordant findings (e.g. Kollmeyer, 2024; Mihalache-O’Keef, 2018; Pinto and Zhu, 2022).

This study re-examines this topic. The preponderance of evidence suggests that economic openness has important pacifying effects for developing countries, yet this may not be the case for several reasons. (1) Reverse or reciprocal causation is likely, because global business actors should seek out peaceful countries when looking for overseas business opportunities and retrench their operations when violence occurs. If this happens, statistical analyses will find positive links between global market integration and civil peace, even if the former is not causing the latter. To my knowledge, two studies examine this possibility and both find limited support for the liberal peace proposition (Magee and Massoud, 2011; Martin et al., 2008). (2) Dynamic and temporal effects are likely, because civil violence and peacebuilding occur within complex social environments where feedback loops and path dependency are common (De Coning, 2020; Millar, 2020). Since episodes of civil violence often last for years, last year’s violence certainly influences contemporaneous violence and how local and global actors react to their changing circumstances. This complexity, however, is not captured by the static regression models used in most prior research (Barbieri and Reuveny, 2005; Blanton and Apodaca, 2007; Bussmann and Schneider, 2007; Flaten and de Soysa, 2012; Gissinger and Gleditsch, 1999; Krause and Suzuki, 2005). (3) In some cases, foreign investment in developing countries results in rent creation, heightened inequality, and intensified competition for scarce jobs (Bornschier and Chase-Dunn, 1985; Dixon and Boswell, 1996; Millar, 2016: 574–576). These conditions, in turn, may foment civil violence (Boswell and Dixon, 1990; London and Robinson, 1989; Millar, 2016; Pinto and Zhu, 2022).

To assess my arguments, I collect data on all major countries classified as developing by the United Nations (2019). The resulting sample contains 103 countries, which are observed annually from 1970 to 2017. Two modeling techniques are used. First, I use standard panel regression techniques to replicate support for the liberal peace proposition. These are static models, as they do not incorporate information about prior levels of violence. Then, I re-run these models under dynamic specifications, allowing me to see whether economic globalization retains its pacifying effects under this more realistic scenario. Next, I use vector autoregression (VAR) and Granger causality tests to help untangle the causal order among civil violence, foreign direct investment, and trade openness. This modeling framework is appropriate when feedback loops are suspected or the direction of causation is unclear.

To preview my findings, the empirical analysis generates limited support for the liberal peace proposition as it applies to civil conflict in developing countries. The static panel models find positive links between economic globalization and civil peace (as predicted by much of the literature), but these links become tenuous under the more realistic scenarios depicted by the dynamic panel models. Likewise, the VAR models yield similar results, showing that civil conflict clearly reduces global trade and investment, but that global trade and investment has very limited pacifying effects. Overall, results from both modeling frameworks are consistent with the notion that civil conflicts are inherently complex and dynamic (De Coning, 2020; Millar, 2020), and that the pacifying effects of economic globalization are limited (Magee and Massoud, 2011; Martin et al., 2008; Mihalache-O’Keef, 2018; Pinto and Zhu, 2022). Importantly, these results hold for various measures of civil violence and for various sampling strategies.

Theory and Empirical Studies

Development and Liberal Peace

Predating scholarship on economic globalization and civil peace, a well-developed literature examines the determinants of civil conflict in developing countries. Over several studies, Collier and Hoeffler (e.g. 1998, 2002) find that GDP per capita and economic growth negatively correlate with incidences of civil violence. Drawing on economic theory, they argue that these conditions create persistent grievances with the status quo, lower the “opportunity costs” associated with conflict, and leave the state without the resources needed to maintain social order. Theoretically, this implies that economic deprivation heightens the incentives but lowers the costs of armed rebellion, thereby making it more likely to occur. Equivalent arguments were made earlier in political science (Gurr and Duvall, 1973; Hibbs, 1973) and in sociology (Muller, 1985; Muller and Weede, 1990).

A well-functioning market economy is also thought to promote civil peace (De Soysa and Fjelde, 2010; Gartzke, 2007; Mousseau, 2012). In societies with effective markets, economic power and income-making opportunities are widely dispersed. This makes despotism less likely and allows people to improve their material conditions through everyday market participation. Here, an important assumption is that markets are “positive-sum” institutions whereby the welfare gains made by some often enhance the welfare gains made by others. In turn, this promotes liberal values that orient individuals toward the peaceful pursuit of well-being through market participation. Indeed, evidence suggests civil violence rarely occurs in societies with strong markets (De Soysa and Fjelde, 2010; Mousseau, 2012).

Similarly, political institutions should structure the likelihood of civil violence, but not necessarily in a straightforward manner. Democracy performs well in this regard, because it provides an institutional means for redressing grievances and lessens the state’s capacity for repression (Hegre, 2014). By contrast, citizens denied political rights may eschew peaceful action in favor of armed rebellion, but most autocratic states possess the ability and willingness to curb dissent. This creates the real possibility that civil peace can emerge from repression. Indeed, evidence suggests that civil violence disproportionately occurs in hybrid regimes, which are neither fully autocratic nor fully democratic (Fein, 1995; Hegre, 2001; Muller, 1985; cf. Vreeland, 2008).

Economic Globalization and Civil Peace

A more recent literature examines whether global market integration heightens the possibilities for civil peace in developing countries. Most studies on this topic are sanguine about the pacifying effects of economic globalization (Barbieri and Reuveny, 2005; Blanton and Apodaca, 2007; Bussmann and Schneider, 2007; Flaten and de Soysa, 2012; Gissinger and Gleditsch, 1999; Krause and Suzuki, 2005), although there are some dissenting views (Magee and Massoud, 2011; Mihalache-O’Keef, 2018; Pinto and Zhu, 2022).

How does participation in global markets induce civil peace, especially for developing countries? Scholarship on this topic highlights several possible mechanisms. One idea is that economic globalization indirectly promotes civil peace by raising living standards and creating new employment opportunities. Consistent with basic economic theory, evidence suggests that trade openness stimulates growth in developing countries (Frankel et al., 1996), and that the newly generated income improves living standards for ordinary citizens (Dollar and Kraay, 2002). In turn, rising living standards and new employment opportunities may promote liberal values as suggested above (i.e. De Soysa and Fjelde, 2010; Gissinger and Gleditsch, 1999; Mousseau, 2012).

This general idea can be framed in terms of “opportunity costs.” As argued by several scholars (Collier and Hoeffler, 1998, 2002; Muller, 1985; Muller and Weede, 1990), poverty associated with low development reduces the opportunity costs associated with civil violence and hence makes it more likely to occur. Thus, if economic globalization creates new and better economic opportunities, the expectation is that state actors and potential rebel groups will avoid violent conflict in order to retain and expand their globalization-induced welfare gains. In this way, the embrace of global markets heightens the opportunity costs associated with civil conflict, making rebellion a less attractive option (Barbieri and Reuveny, 2005).

In addition, sustaining domestic peace becomes a means of “signaling” to the global business community that particular countries are worthy candidates for investment and trade opportunities (Blanton and Apodaca, 2007). Conflict and aggressive posturing, by contrast, send the opposite signal. Hence, rational domestic actors should seek to avoid conflict as a means of maintaining their country’s reputation as an attractive site for foreign investment. The strength of these incentives, in turn, should increase as global market integration deepens, and they may arise not only from bilateral trade but also from third-party trade within larger trading and investment networks (Kinne, 2012, 2014).

The notion that countries signal to potential foreign investors builds on several ideas from political science and international relations. In general, it extends the “costly signaling” concept (Gartzke and Li, 2003). This perspective suggests that economic globalization should heighten the costs associated with aggressive and threatening posturing between states (and by extension, aggressive posturing within states), because global investors capital may interpret such behavior as genuine threats to their economic interests and retrench their business activities accordingly. In this way, threats to domestic peace should become costly and hence avoided. It also extends ideas from bargaining theory (Walter, 2009). Here, civil conflicts are interpreted as bargaining failures, because rational actors should prefer negotiated bargains instead of armed conflict as a means of resolving grievances. Extending this logic suggests that economic globalization promotes civil peace by further heightening bargaining pressures, as foreign investors become another vested interest that favors peace over conflict and has the power and resources to push for such outcomes.

However, recent research on conflict between states questions whether economic globalization may actually weaken some of these pacifying effects (Peterson and Wen, 2021; Sadeh and Feldman, 2020). A particularly salient point is that global market integration creates malleable trading and investment networks and thereby broadens the prospects of states to create new economic ties. These possibilities, in turn, mean that belligerent states may find it easier to replace trading partners, evade economic sanctions, and retain access to global credit markets, all of which make war less costly. Importantly, these ideas may apply to the prospects of civil violence as well.

Alternative Hypotheses

Rents and Negative Externalities

Dependency and related theories have long depicted global trade and investment as disproportionately benefiting economic actors in developed countries (Prebisch, 1950; Frank, 1966; Galtung, 1971). A key point is that developing countries need global trade and investment, but they are susceptible to exploitation/rent seeking due to their structurally disadvantaged positions in world markets and their weak states. One particular concern is that the jobs generated by foreign investment typically offer poor pay and undesirable working conditions. This can happen, as Lewis (1954) famously argued, simply because the pool of potential workers in developing countries often vastly exceeds the available job opportunities. Under such conditions, the wages of unskilled workers remain just above subsistence levels, with the benefits of rising productivity largely accruing to employers. Consequently, wages in developing countries generally lag far behind comparable wages elsewhere, even after adjusting for differences in labor productivity (see, for example, Kollmeyer, 2009, Table 1). Although wages should eventually rise as employment expands and markets diversify, the prevalence of poor pay can persist for long durations. For example, despite its extraordinary growth over recent decades, China has just recently reached its “Lewis turning point” (Fang and Yang, 2011). This reality suggests that proponents of liberal peace may over-estimate the degree to which foreign trade and investment generate widespread prosperity in developing countries.

Importantly, the prevalence of low-wage employment can spark civil conflict in several ways. One issue is that low pay, when coupled with high rents accruing to capital, creates systematic tensions between foreign firms and their workforces (Millar, 2016: 574–575). In addition, despite being low by international standards, the wages paid by foreign firms usually exceed the domestic rate, thereby creating a wage differential between workers in the domestic and foreign sectors (Bornschier and Chase-Dunn, 1985). The resulting inequality, and the relative scarcity of jobs with foreign employers, can create community tensions that sometimes turn into violent revolts (London and Robinson, 1989). Similar tensions can arise between those without jobs, often a sizable group in poorer developing countries, and those with jobs in the foreign sector (Millar, 2016: 575–576).

Clearly, such conditions are not conducive to domestic peace. However, as Pinto and Zhu (2022) argue, developing countries with strong and effective states can redirect rents toward socially useful causes, but in developing countries with weak or corrupt states, dominant foreign firms can limit the state’s autonomy to act in the national interest (see also Evans, 1995). In fact, the combination of high rents and high market concentration can foment unrest by tempting potential rebel groups to challenge the state for control over prized assets. These problems are particularly acute in the primary sector of the economy (Mihalache-O’Keef, 2018), which historically predominates in most developing countries (see also Kollmeyer, 2024).

Combining this more critical view of economic globalization with the liberal peace proposition leads me to surmise that, for developing countries, global trade and investment may promote peace in some ways, but also create unintended consequences that mute these pacifying effects linked to heightened opportunity costs and heightened signaling pressures arising from economic globalization. This leads to my first hypothesis:

Hypothesis 1 (H1). The effect of economic globalization on civil peace in developing countries is negligible, because positive and negative externalities largely offset one another.

Reverse or Reciprocal Causation

Another plausible scenario is that the causal order may run in reverse. Quite reasonably, one could surmise that global economic actors purposively seek out peaceful countries when looking for investment and trade opportunities in the developing world. They might do this because peace and stability clearly benefit their business operations, as conflict can impose severe costs and risks. For example, capital investment can be destroyed, markets disrupted, and consumers and workers harmed. Consequently, peaceful countries should attract more foreign investment and enjoy more trading opportunities than conflict-prone countries, resulting in a strong association between the two phenomena. Similarly, outbreaks of civil violence should have the opposite effect, actively driving away trade and investment. Importantly, whether peaceful countries attract trade and investment or whether conflict-prone countries drive them away, the observed statistical association between the economic globalization and civil peace will be positive (as claimed by much of the literature). However, under this scenario, the positive relationship arises from reverse causation. For interstate wars, this phenomenon is well documented (see, for example, Anderton and Carter, 2001; Hegre et al., 2010), but for civil violence, it receives less attention. Regarding the latter, Marano et al. (2013) and Martin et al.(2008) consider the possibility of reverse causation and find that civil conflict reduces trade flows, net of their control variables.

Similarly, it may be that global economic actors are not necessarily creating peace but reacting to outbreaks of civil violence and retrenching their business activities accordingly. Under this scenario, global economic actors become aware of civil violence in their host country and incorporate this information into their future business and investment decisions. Hence, if civil violence started last year (t – 1), global economic actors may rationally reduce their business activities this year (t), especially if the risk of continued civil violence is high. This scenario would create a strong statistical association between the two phenomena, but the causal relationship would run from civil violence in year (t – 1) to lower global trade and investment in year (t). This is a very different causal relationship than the one posited by liberal peace theory.

Importantly, this latter scenario is consistent with prominent economic and sociological theories. In economics, it is understood that economic actors form “rational expectations” or “adaptive expectations” in which present decisions are guided by information about prior events (Lucas, 1986; Lucas and Sargent, 1981). Likewise, in sociology, the concept of “reflexivity” describes how social actors monitor their actions in relation to their social environments, creating feedback loops in which prevailing practices are revised in light of new understandings (Giddens, 1993 [1976]; 1984: especially pp. 5–6). Since reflexivity becomes generalized, it applies not just to individuals but to firms, organizations, states, and whole societies. Importantly, both theories suggest that economic and social actors understand their environments and that this awareness creates feedback loops in which past events affect present actions, ultimately making human behavior adaptive, dynamic, and reflexive. This leads to my second hypothesis:

Hypothesis 2 (H2). Economic globalization and civil peace are bound together by reverse causation in which global business actors purposively seek out peaceful countries and avoid/withdraw from conflict-affected countries.

This hypothesis is also consistent with more statist views of trade. In particular, the “trade-flows-the-flag” paradigm highlights the potential for states to shape trade patterns in their interests, with diplomatic relations between states often paving the way for stronger trade ties (Pollins, 1989). Indeed, studies on interstate conflict find evidence that countries with friendly diplomatic relations trade more with each other (Chen and Zhou, 2021), and that when wars commence, bilateral and third-party trade shifts from adversarial states to friendly states (Feldman and Sadeh, 2018). All of this suggests that peaceful relations precede trade rather than trade creating peaceful relations.

A slightly different scenario combines Hypothesis 2 with the liberal peace proposition. Possibly, economic globalization and civil peace are linked by reciprocal causation, in which both phenomena are cause and effect. Under this scenario, global economic actors not only seek out peaceful countries, but once operating in a host country, they further entrench peace through their participation in the domestic economy. In other words, peaceful countries should attract global trade and investment, but then the resulting business activity may consolidate domestic peace. This scenario amounts to a more nuanced version of the liberal peace proposition, because it acknowledges the clear possibility of endogeneity. This leads to my third hypothesis:

Hypothesis 3 (H3). Economic globalization and civil peace are bound together by reciprocal causation.

From the perspective of liberal peace theory, Hypotheses 2 and 3 imply the presence of endogeneity caused by reverse or reciprocal causation. The literature on militarized conflicts between countries carefully considers this idea (Bussmann, 2010; Hegre et al., 2010; Kim and Rousseau, 2005). For example, Hegre et al. (2010) find evidence of endogeneity in the form of trade promoting peace as well as conflict reducing trade. Similar causal patterns hold for foreign direct investment (Bussmann, 2010). Yet, the literature on civil conflict in developing countries gives this issue less attention. It is considered by two studies—Magee and Massoud (2011) and Martin et al. (2008)—and both conclude that civil conflict disrupts global trade and investment more than global trade and investment foster civil peace. Importantly, the methodological concern is that, if reverse or reciprocal causation is present but unmeasured, standard regression models will inflate the pacifying effects of economic globalization, potentially leading to overly optimistic support for liberal peace theory.

Temporal Dependence and Dynamic Effects

Another scenario highlights the path-dependent and self-reinforcing nature of armed civil conflicts (see review of path dependency by Mahoney (2000)). Some countries experience multiple episodes of civil violence over decades, while others remain peaceful over this same period. Under both circumstances, countries are following self-reinforcing trajectories in which prior outbreaks of violence (or periods of peace) create the conditions for further violence (or further peace). For post-conflict societies in particular, violence often reoccurs because the negative effects of previous conflicts still linger. For example, old grievances are unresolved or the economy and state are not fully recovered from past conflicts. These conditions, in turn, heighten the chances of renewed conflict, trapping some countries in cycles of recurrent violence and stalled development.

This can be conceptualized as path or temporal dependency. For many social phenomena, including episodes of armed conflict, the passage of time since the last event is a strong predictor of the event under consideration (Beck et al., 1998; Oneal and Russett, 1997). This happens because historical events are structured by prior events. In the context of this study, this implies that past violence should heighten the chance of present violence, net of other relevant factors. Following Beck et al. (1998), researchers routinely control for the duration of peace in their models of conflict, but some research supporting the notion of a liberal peace in developing countries often overlooks this crucial factor (see, for example, Blanton and Apodaca, 2007; Gissinger and Gleditsch, 1999; Krause and Suzuki, 2005).

Dynamic effects are also important. There is a growing appreciation that civil conflict and peacebuilding occur within complex and dynamic social environments (De Coning, 2020; Millar, 2020; see also Goodhand, 1999). Over recent decades, small- to medium-scale armed conflicts within states have proliferated, and these types of conflicts are rarely short-lived. Their long duration is partially driven by their dynamism, which can unleash social forces that propel conflicts into the future. For instance, civil violence often generates new grievances, creates counter mobilizations, and builds up martial skills, all of which can sustain civil violence once it begins (Crenshaw and Robison, 2010: 242). Hence, countries in the midst of conflict face high odds of experiencing violence in the following year. Prior sociological research on political rebellions routinely includes lagged dependent variables (LDVs) for this reason (see, for example, Muller, 1985; Muller and Weede, 1990; White, 1993). Yet this issue is neglected by research supporting the liberal peace in developing countries (see, for example, Barbieri and Reuveny, 2005; Flaten and de Soysa, 2012; Gissinger and Gleditsch, 1999). Overall, this leads to my fourth hypothesis.

Hypothesis 4 (H4). Civil violence in the present year is substantially driven by civil violence in prior years. Other factors, including economic globalization, are less consequential.

Data and Methods

To assess my arguments about economic globalization and civil conflict, I collect data on an extensive sample of developing countries observed annually from 1970 to 2017. The sample comprises all member states of the United Nations (UN) meeting three criteria. First, the UN classifies its members into four broad categories: “developed economies,” “economies in transition,” “developing economies,” and “small island developing states” (UN, 2019: see Table A, B, C and H, respectively). For my study, I define “developing” as states in the latter two groupings (UN, 2019: Table C and H). From this subset, I drop newly formed states as well as countries with populations below 0.5 million in 2018. Note that post-socialist countries arising from the break-up of the Soviet Union and Yugoslavia are deemed “economies in transition” and not included in my sample. Unfortunately, three countries meeting all criteria—Afghanistan, Cuba, and North Korea—drop from the sample due to missing data on the economic globalization variables. The 103 countries in my full sample, and their relevant descriptive statistics, are shown in the Appendix 1 (see Table A1). However, later in the study, I amend this sampling strategy for robustness checks.

Measuring Civil Violence

My study measures civil violence with two sources of data. The first source is the Major Episodes of Political Violence (MEPV) database developed by the Center for Systemic Peace (CSP) (Marshall, 2019). The CSP/MEPV tracks and records “major episodes of political violence,” which it defines as the “systematic and sustained use of lethal violence,” perpetrated by organized groups and resulting in at least 500 deaths over the duration of the conflict (including combatants and noncombatants). In particular, I use the CIVTOT variable, which measures the whole range of intra-societal conflicts but excludes conflicts between states. This variable is originally coded along an 11-point scale, ranging from 0 (no major incidence of civic violence) to 10 (systematic and indiscriminate destruction of society), but recode it into a dummy variable, with 1 denoting country-years experiencing civil violence.

The second source of data is the Armed Conflict Dataset developed by the Uppsala Conflict Data Program (UCDP) and the Peace Research Institute Oslo (PRIO) (Gleditsch et al., 2002). The UCDP/PRIO records incidences of civil violence occurring between the state and armed domestic adversaries, which result in at least 25 deaths in a calendar year. These data are recorded in dummy-variable format, with “1” denoting country-years experiencing civil violence. I re-run my regression models using the UCDP/PRIO data, in part because its lower fatalities threshold may shed additional light on my research questions.

My study examines occurrences and onsets of civil violence using data from the CSP/MEPV and the UCDP/PRIO. The onset of civil violence is measured by coding “0” for country-years with no civil violence, “1” for country-years when violence commences, and “missing” for country-years with ongoing civil violence. If there are fewer than two peaceful years between onsets, the observations are coded as one episode. Unfortunately, this coding scheme results in substantial data loss because long spells of continuous or intermittent violence get coded as missing even though considerable amounts of violence are occurring. By contrast, the occurrence of civil violence is measured by coding “0” for country-years with no civil violence and “1” for country-years experiencing civil violence, irrespective of whether it is commencing or continuing. Many studies examine the onset of civil violence (e.g. Bussmann and Schneider, 2007; Collier and Hoeffler, 2002; Flaten and de Soysa, 2012; Mihalache-O’Keef, 2018; Pinto and Zhu, 2022). Others examine the occurrence of civil violence (e.g. Blanton and Apodaca, 2007; Kollmeyer, 2024; Magee and Massoud, 2011; Martin et al., 2008). Still others examine both (e.g. Barbieri and Reuveny, 2005).

For several reasons, my theoretical interests and modeling strategies are better aligned with the occurrence measure of civil violence. First, the onset measure sidesteps the complex dynamics of the conflict itself by coding conflict years as missing observations. My analysis seeks to address these dynamic effects directly (see Hypotheses 3 and 4), but this cannot be done if the conflict years are coded as missing. Second, the VAR modeling framework, which I use to untangle the temporal relationship between global market integration and civil violence, cannot easily handle the onset measure. Here, the problem is that VAR models use lags of civil violence as key regressors, but for the onset measure, these lags will be perfectly collinear as both are always coded as 0.¹ This problem, however, does not affect my panel models. In sum, my empirical analysis focuses on the occurrence measure, with the onset measure being used to provide additional insight and to make my findings comparable with the full range of prior research.

Measuring Economic Globalization

Two variables measure economic globalization. The first variable is trade openness. It measures a country’s integration into global markets for commodities and manufactured goods, and it equals the annual value of imports plus exports benchmarked against GDP. Data come from the Penn World Tables (Feenstra et al., 2019). The second variable is FDI inflows. The World Bank defines foreign direct investment (FDI) as the foreign ownership of a business or subsidiary in a host country, where the foreign investor holds at least a 10% equity stake and actively participates in the enterprise’s management. FDI inflows comprise the annual value of initial capital investments, the reinvestment of earned income, and any follow-up capital investments. This sum is reduced by the value of repatriated income, creating the possibility that FDI inflows can be negative. I benchmark net inflows of FDI against GDP. Data come from the World Bank (2019). I consider alternative measures of FDI, such as those based on stock rather than annual flows, but data are sparse for the years and countries in my sample.

Control Variables

My full models contain several control variables. First, I control for political regime type using data from the Polity IV Project (Marshall et al., 2019). The Polity IV Project gauges the “authority characteristics” of the world’s sovereign states, giving each an annual score ranging from −10 (for complete autocratic rule) to 10 (for complete democratic rule). I recode this data into four mutually exclusive dummy variables: autocracy (−10 to −6), partial autocracy (−5 to −1), partial democracy (0 to 5), and democracy (6 to 10). This allows for nonlinear effects in which democracy may not be the only regime type associated with low levels of civil violence (see Fein, 1995; Hegre, 2001; Muller, 1985; cf. Vreeland, 2008).

Petrostates are identified by a dummy variable (data from Cheatham and Labrador, 2021). Major exporters of oil and gas often become rentier economies, controlled by corrupt and enriched elites (Ross, 2012). This can distort development, fuel grievances, and provoke rebellions. Yet some petrostates evolve into stable autocracies, with long histories of civil peace (Colgan, 2015). Hence, petrostates are generally more prone to civil violence than nonpetrostates, but there is considerable variation around this mean effect.

In addition, following Collier and Hoeffler (1998, 2002), I control for socioeconomic factors associated with civil conflicts in developing countries. The data for these variables come from the Penn World Table (Feenstra et al., 2019). GDP per capita controls for levels of socioeconomic development, with the figure being adjusted for inflation and purchasing power parities. Economic growth captures the annual growth rate of the economy. Both variables should be negatively associated with civil conflict. Population size (logged) accounts for the fact that populous countries are more prone to grievances occurring somewhere within their borders simply due to their large populations. They are also less reliant on global trade and investment due to their large internal markets. This latter point is important given this study’s focus on international trade and investment.

Years of peace measures the number of years since the last civil conflict. It captures the strong possibility that global economic actors purposefully seek out peaceful countries, and thus, it helps to account for the adaptive and reflexive nature of global business actors. For post-conflict societies in particular, it also accounts for the gradual reduction in old grievances and the eradication of weapons caches, both of which may cause the re-emergence of violence. The measure starts from 1965 and counts the number of years since the last episode of civil violence, with the measure being created from the CSP/MEPV and UCDP/PRIO data.

Finally, I include time-invariant measures of ethnic fractionalization (data from Fearon, 2003) and the Cold War (coded 1 for years before 1990). I also try primary commodity exports/GDP (as suggested by Collier and Hoeffler, 1998, 2002), but drop it due to significant amounts of missing data.²

Analytical Strategy

Panel Regression

The analysis begins with standard panel regression models, which treat civil violence as a dependent and exogenous variable. In particular, I use logistic regression with random effects to assess whether economic globalization reduces the occurrence of civil violence within my sample. Estimation by logistic regression is appropriate because my dependent variable (DV) is binary (Hosmer, Lemeshow, and Sturdivant, 2013), and the inclusion of random effects (RE) is appropriate because panel models face complications associated with unobserved heterogeneity (Clark and Linzer, 2015). An alternative approach, via fixed effects (FE), cannot be used because many countries in my sample experienced no incidences of civil violence over the sample period, and because some of my independent variables are time-invariant. Under both conditions, the FE approach will create perfect collinearity between the country FEs and the time-invariant observations in my data. This problem, however, does not beset the RE approach. I do, however, include a time-trend variable to account for possible nonstationarity (Box-Steffensmeier, 2014: 125–149).

The inclusion of years of peace may cause statistical bias because it is temporally related to the DV (Beck et al., 1998; Carter and Signorino, 2010). To remedy this complication, Beck et al. (1998) recommend including cubic splines as control variables. More recently, Carter and Signorino (2010) note that cubic polynomials achieve the same effect (i.e. including the time-dependent regressor, its squared term, and its cubed term). I experiment with both techniques and find the Carter-Signorino method simpler and more intuitive.

Similarly, some of my panel models have dynamic specifications achieved by including an LDV as a regressor (Box-Steffensmeier et al., 2014: Chapter 3). Although LDVs are often used to control serial correlation, my motivation is purely theoretical. Episodes of civil violence often last for years, and in such conflicts, last year’s violence certainly influences the present year’s violence and the ways local and global actors react to their changing circumstances. If a country is experiencing a multi-year conflict, but the prior year’s violence is unmeasured, the resulting parameter estimates will absorb these dynamic effects and potentially generate biased results. This model specification, however, is incompatible with the onset measure because its 1-year lag is always coded as “0” for all nonmissing observations, thus leaving it without the variation needed to estimate its parameter.

To show the importance of this dynamic factor, my panel models of the occurrence of civil violence are estimated under static specifications (LDV and years of peace not included) and dynamic specifications (LDV and years of peace included).³ However, the LDVs is only included in the last models, because most of the dynamic effects are captured when years of peace = 0, and because some researchers worry that LDVs absorb too much variance (possibly underestimating the effects of other regressors).

VAR

To assess the possibility of reverse or reciprocal causation between economic globalization and civil violence (as implied by Hypothesis 2 and 3), I estimate several VAR models. Originally proposed by Sims (1980) for use in macroeconomics, VAR models are useful for estimating relationships among time-series variables when dynamic feedback loops are suspected or the direction of causation is unclear. Given my theoretical argument that the causal relationship among economic globalization and civil conflict is more complex than commonly recognized by liberal peace theory, I use VAR models and Granger causality tests to help establish the causal order among civil violence, foreign direct investment, and trade openness.

In VAR models, each variable receives its own equation, where it is regressed on its lags and the lags of the other variables of interest (Box-Steffensmeier et al., 2014: 106–124; Enders, 2014: Chapters 5 and 6). Given the assumption of endogeneity, there is no need to distinguish between the dependent and independent variables and control variables are not required. Consequently, in its most common form, the equations in VAR models exhibit the same battery of right-hand-side regressors but different left-hand-side regressors. When this occurs, the individual equations can be estimated as standalone models using standard regression techniques (Enders, 2014: 290).

This modeling system is adept at detecting causality. VAR models build on Granger’s (1969) insight that “the cause” necessarily precedes “the effect,” and that lagged values in time-series data can illuminate this temporal sequence. For this study, I use this modeling framework to test whether economic globalization reduces civil conflict, whether civil conflict reduces economic globalization, whether these factors are linked by reciprocal causation, or whether the factors are unrelated. To make this assessment, I first estimate the parameters of a VAR model.⁴ Then, Granger causality is deemed to adhere if variable Y_t is predicted not only by its own lagged values, but by the lagged values of variable X_t (or other variables of interest). If this occurs, X is said to “Granger-cause” Y. Unfortunately, this feature of VAR modeling sits uneasily with the onset measure of civil violence, because its 1- and 2-year lags will be perfectly collinear when onset is used as the left-hand-side regressor (see Note 1). This means these lagged parameters cannot be estimated, even though they are a crucial component of the VAR framework.

The lack of control variables can seem odd to some readers, but this is typical of VAR models. The reason for this distinctive model specification is that VAR models can determine the causal order among the endogenous variables without control variables. In some cases, however, researchers include “exogenous” variables in their VAR models, but typically only when trying to assess how some external factor (i.e. an exogenous shock) alters the patterns of association among the endogenous variables. These exogenous variables, however, are not being used to “hold constant” the influence of these covariates. Since my study is only concerned with understanding the causal order among my variables of interest, my VAR models do not use control variables. Nonetheless, as a robustness test, I re-run my VAR models but this time with prominent control variables included.

Finally, although VAR modeling was developed for time-series analysis, it can be adapted for panel regression (Holtz-Eakin et al., 1988). When used with panel data, time trends are included to guard against spurious regression (or variables are first differenced if nonstationarity is suspected), and the RE approach is used to absorb unmeasured country-specific effects.

Findings

Panel Regression Models

Table 1 presents standard panel regression models of the occurrence of civil violence in developing countries. The models are estimated via binary logistic regression, with the continuous regressors converted to z-scores to facilitate comparisons of the substantive effects. Model 1 begins by isolating the effect of global market integration on the likelihood of experiencing civil violence. In particular, the parameter estimates for FDI inflows and trade openness are found to be negative and highly significant, suggesting that each factor exerts its own pacifying effect. Hence, Model 1 provides initial support for the notion that participation in global markets reduces civil conflict in developing countries.

Model 2 adds several control variables. Note that four countries—Guyana, Libya, Papua New Guinea, and Somalia—drop from the sample due to missing data. Despite introducing the control variables, the revised parameter estimates for FDI inflows and trade openness remain negative and highly significant and their magnitudes change in only limited ways. Furthermore, the parameter estimates for the control variables are consistent with theoretical expectations and suggest that several economic and political factors affect the likelihood of civil violence occurring in developing countries. As expected, autocracies and democracies are less prone to civil violence than partial autocracies and partial democracies (see Fein, 1995; Hegre, 2001; Muller, 1985). Also, as expected, population size heightens the chance of civil violence, while GDP per capita and economic growth reduce the chance of civil violence (see Collier and Hoeffler, 1998, 2002). Overall, the model performs well, but its explanatory power is limited, as the pseudo R² is only 0.167.⁵

Model 3 introduces years of peace. This variable captures the adaptive and reflective nature of global business actors whose decisions are influenced by a country’s historical experience of peace and conflict. This model generates several noteworthy findings. Foremost, the magnitude and statistical significance of FDI inflows and trade openness drop considerably, leaving both with limited effects. By contrast, the control variables change in more limited ways. Hence, the inclusion of years of peace primarily alters the economic globalization variables, rendering them much less consequential. In addition, the explanatory power of the model increases notably, with the pseudo R² moving from 0.167 in the static model to 0.577 the dynamic model (cf. Models 2 and 3).

Table 1.

Logistic regression estimates of the occurrence of civil violence: Based on CSP/MEPV data (500+ deaths over conflict).

	Occurrence of civil violence (coded 1 or 0)
	(1)	(2)	(3)	(4)
Dynamic effects
Civil violence (t – 1)				4.995*** (0.302)
Years of peace^a (t – 1)			-14.227*** (0.699)	-1.125^ϯ (0.604)
Global economic flows
FDI inflows (%GDP) (t – 1)	-0.359*** (0.072)	-0.445*** (0.083)	-0.110(0.077)	-0.100(0.083)
Trade openness (% GDP) (t – 1)	-0.663*** (0.102)	-0.668*** (0.113)	-0.184^ϯ (0.131)	-0.128(0.125)
Control variables
Political regime type
Autocracy		0.451* (0.205)	0.459^ϯ (0.242)	0.131(0.236)
Partial autocracy		1.123*** (0.194)	0.965*** (0.252)	0.591* (0.263)
Partial democracy		1.081*** (0.181)	0.848*** (0.226)	0.695** (0.255)
Democracy (ref cat.)		—	—	—
Population size (log) (t – 1)		1.139*** (0.271)	0.912*** (0.162)	0.710*** (0.118)
GDP per capita (log) (t – 1)		-1.106*** (0.158)	-0.263* (0.128)	-0.248** (0.094)
Economic growth (t – 1)		-0.149** (0.053)	-0.112^ϯ (0.062)	-0.117(0.073)
Cold war (yes = 1)		0.601** (0.184)	0.659** (0.251)	0.311(0.280)
Petrostate (yes = 1)		0.965(0.821)	0.186(0.407)	0.191(0.257)
Ethnic fractionalization		0.112(0.267)	0.018(.129)	0.014(0.082)
Constant	-2.57*** (0.35)	-4.10*** (0.61)	-2.91*** (0.73)	-3.22*** (0.74)
Observations	4573	4360	4347	4347
Number of countries	103	99	99	99
Pseudo R²	0.079	0.167	0.577	0.743

Notes. Logistic regression with random effects. Coefficients with robust standard errors clustered on countries in parentheses. Time trend included but not reported. Continuous variables converted to z-scores to facilitate comparison.

Squared and cubed terms included but not reported.

***

p < .001, **p < .01, *p < .05, ^ϯp < .10.

Model 4 introduces civil violence (t – 1). This variable is added separately because some researchers worry that LDVs absorb too much variance (possibly underestimating the effects of other regressors), and because years of peace already captures some of the dynamic effects (especially when years of peace = 0). Nonetheless, this variable should better capture the path-dependent and self-reinforcing qualities of ongoing civil violence. Compared to the preceding model, the magnitude and statistical significance of FDI inflows and trade openness drops further, again casting doubt on the degree to which global market integration creates pacifying effects for developing countries.

Overall, the findings from the panel regression models broadly support several of the hypotheses. In particular, they provide support for Hypothesis 1 (that economic globalization’s pacifying effects are negligible) and Hypothesis 4 (that past violence is the most important determinant of present violence). In addition, the findings are compatible with Hypothesis 2 (that the pacifying effects of economic globalization are spurious because they partially reflect global economic actors retrenching their business activities when civil violence occurs).

Table 2 re-estimates these panel regression models but this time using the UCDP/PRIO data. Recall that these data use a lower fatalities threshold for constituting an episode of civil violence (25 deaths per year). This lower threshold allows for the possibility that economic globalization may suppress low-level civil conflicts, which are not detected by the CSP/MEPV data. Beginning with the static models (Models 5 and 6), the results continue to find that trade openness and FDI inflows generate negative and statistically significant effects on civil violence. This pacifying effect, however, wanes when the adaptive and dynamic effects are included (see Models 7 and 8). This is especially true of trade openness, which loses its statistical significance and falls considerably in its magnitude. Overall, these results provide some support for the liberal peace proposition by highlighting that economic globalization generates some pacifying effects related to low-fatality conflicts. Yet the results provide some support for Hypothesis 1 (that economic globalization’s pacifying effects are negligible) and Hypothesis 4 (that past violence is the most important determinant of present violence). It is worth noting that Models 4 and 8 are essentially onset measure models because they explicitly control for ongoing violence, which is captured when the 1-year lag of civil violence equals 1 and the duration of peace equals 0.

Table 2.

Logistic regression estimates of the occurrence of civil violence: Based on the UCDP/PRIO data (25+ deaths per year).

	Occurrence of civil violence (coded 1 or 0)
	(5)	(6)	(7)	(8)
Dynamic effects
Civil violence (t – 1)				3.206*** (0.262)
Years of peace^a (t – 1)			-12.685*** (0.675)	-3.556*** (0.861)
Global economic flows
FDI inflows (%GDP) (t – 1)	-0.170** (0.060)	-0.206** (0.067)	-0.134* (0.068)	-0.143* (0.069)
Trade openness (% GDP) (t – 1)	-0.425*** (0.102)	-0.444*** (0.116)	-0.104(0.127)	-0.115(0.126)
Control variables
Political regime type
Autocracy		0.463* (0.209)	0.302(0.233)	0.197(0.238)
Partial autocracy		1.056*** (0.197)	0.659** (0.237)	0.586* (0.245)
Partial democracy		1.626*** (0.185)	0.798** (0.220)	0.784** (0.234)
Democracy (ref cat.)		—	—	—
Population size (log) (t – 1)		1.267*** (0.248)	0.819*** (0.155)	0.742*** (0.140)
GDP per capita (log) (t – 1)		-0.420** (0.147)	-0.195^ϯ (0.126)	-0.229* (0.113)
Economic growth (t – 1)		-0.148** (0.053)	-0.116^ϯ (0.063)	-0.054(0.068)
Cold war (yes = 1)		0.475* (0.197)	0.951*** (0.248)	0.639** (0.258)
Petrostate (yes = 1)		0.541(0.750)	0.251(0.417)	0.325(0.354)
Ethnic fractionalization		0.131(1.154)	0.191(0.553)	0.042(0.469)
Constant	-3.33*** (0.35)	-4.29*** (0.81)	-4.54*** (0.71)	-4.34*** (0.71)
Observations	4573	4360	4360	4360
Number of countries	103	99	99	99
Pseudo R²	0.059	0.164	0.452	0.635

Squared and cubed terms included but not reported.

***

p < .001, **p < .01, *p < .05, ^ϯp < .10.

Table 3 examines the onset of civil violence using the CSP/MEPV and the UCDP/PRIO data. The panel regression framework accommodates the onset measure, although the LDV cannot be used because it is always coded as 0 or missing. Except for this issue, these models are identical to those presented in Tables 1 and 2, but now assess the onset rather than the occurrence of civil violence. Models 9 and 11 isolate the effects of global market integration. Here, the results show that trade openness but not FDI inflows generate negative and statistically significant pacifying effects. Models 10 and 12 then introduce the relevant control variables. Consistent with expectations, the modest pacifying effects found in the earlier model disappear.

Table 3.

Logistic regression estimates of the onset of civil violence: Based on CSP/MEPV and UCDP/PRIO data.

	Onset of civil violence(coded 1, “missing,” or O)
	CSP/MEPV data		UCDP/PRIO data
	(9)	(10)	(11)	(12)
Dynamic effects
Years of peace^a (t – 1)		-0.101(0.089)		-0.646(0.441)
Global economic flows
FDI inflows (%GDP) (t – 1)	0.064(0.098)	0.068(0.097)	0.025(0.098)	0.033(0.094)
Trade openness (% GDP) (t – 1)	-0.371** (0.139)	-0.136(0.174)	-0.272* (0.126)	-0.047(0.147)
Control variables
Political regime type
Autocracy		0.876* (0.380)		0.085(0.298)
Partial autocracy		1.108** (0.385)		0.501(0.320)
Partial democracy		1.383** (0.398)		1.122*** (0.298)
Democracy (ref cat.)		—		—
Population size (log) (t – 1)		0.601*** (0.155)		0.415** (0.132)
GDP per capita (log) (t – 1)		-0.220^ϯ (0.116)		-0.283* (0.137)
Economic growth (t – 1)		-0.097(0.120)		-0.064(0.095)
Cold war (yes = 1)		0.719^ϯ (0.419)		0.021(0.378)
Petrostate (yes = 1)		0.480(0.394)		0.255(0.336)
Ethnic fractionalization		0.140(0.441)		0.006(0.426)
Constant	-2.74*** (0.38)	-2.22* (1.08)	-2.91*** (0.37)	-2.81** (0.91)
Observations	3524	3338	3600	3426
Number of countries	101	97	99	95
Pseudo R²	0.005	0.017	0.003	0.014

Notes: Logistic regression with random effects. Coefficients with robust standard errors clustered on countries in parentheses. Time trend included but not reported. Continuous variables converted to z-scores to facilitate comparison.

Squared and cubed terms included but not reported.

***

p < .001, **p < .01, *p < .05, ^ϯp < .10.

VAR Models

The analysis now uses VAR models and Granger causality tests to untangle the causal relationships among civil violence, trade openness, and FDI inflows. Importantly, this modeling technique allows me to assess whether economic globalization and civil violence are bound together by reverse or reciprocal causation (as suggested by Hypothesis 2 and 3). These models are estimated with two lags, and the variables are treated as endogenous. When occurrence of civil violence is the left-hand-side regressor, the equation is estimated using logistic regression with RE, but when FDI inflows and trade openness are the left-hand-side regressors, the equations are estimated using general linear regression with RE. In addition, I report pairwise Granger causality tests (Granger, 1969), which assess whether the left-hand-side regressor is explained by the lags of the other right-hand-side regressors. This is determined with an F-test, run against the null hypothesis that the lags of the right-hand-side regressor are jointly equal to zero. Finally, I experiment with including years of peace and its derivative terms in these VAR models, trying to approximate an onset measure, but this specification does not meaningfully change the resulting parameter estimates.⁶

Table 4 reports results from the VAR models using the CSP/MEPV data. Starting with occurrence of civil violence as the left-hand-side regressor, the results indicate that prior levels of trade openness, and to a lesser degree, prior levels of FDI inflows are statistically related to the occurrence of civil violence. This is evidenced by the parameter estimates of the second lag of trade openness being negative and statistically significant and by the Granger causality tests. Regarding the latter, the F-tests reject the null hypothesis of no causation running from trade openness to occurrences of civil violence, but fail to reject the null hypothesis of no causation running from FDI inflows to occurrences of civil violence. These findings provide limited support for liberal peace theory, as there is some evidence that trade openness but not FDI inflows promote civil peace.

Table 4.

VAR model with Granger causality tests: Occurrence of civil violence based on CSP/MEPV data (500+ deaths over conflict).

	Occurrence civil violence_t (13)	FDI inflow_t (14)	Trade openness_t (15)
Occurrence civil violence_{_t – 1}	4.854*** (0.229)	-0.538* (0.254)	-1.593* (0.630)
Occurrence civil violence_{_t – 2}	1.029***(0.207)	0.361(0.292)	0.646(0.601)
FDI inflows_{_t – 1}	-0.023(0.015)	0.471*** (0.049)	0.080** (0.031)
FDI inflows_{_t – 2}	0.026(0.019)	0.171** (0.054)	0.049(0.035)
Trade openness_{_t – 1}	0.010(0.007)	0.014^ϯ (0.008)	0.825*** (0.036)
Trade openness_{_t – 2}	-0.022*** (0.006)	-0.005(0.007)	0.1251*** (0.034)
Observations	4573	4459	4470
Countries	103	103	103
R² (pseudo R²)	0.773	0.403	0.921
Granger causality F-tests
Occurrence civil violence →	—	5.23^ϯ (0.073)	11.48** (.003)
FDI inflows →	3.36(0.201)	—	14.15** (0.001)
Trade openness →	18.26*** (0.000)	4.25(0.102)	—

Notes. Model 13 estimated with logistic regression with random effects. Models 14 and 15 estimated with generalized least squares regression with random effects. Robust standard errors clustered on countries shown in parentheses. Time trend and constant included but not reported in all equations.

***

p < .001, **p < .01, *p < .05, ^ϯp < .10.

Models 14 and 15 examine FDI inflows and trade openness as left-hand-side regressors, respectively. As anticipated, the main findings are that prior outbreaks of civil violence reduce FDI inflows and trade openness, with the effect being particularly strong for trade openness. These results are consistent with my argument that global economic actors react to outbreaks of violence in their host countries and retrench their business activities accordingly (i.e. Hypothesis 2, the causal relationship runs from conflict to declining global trade and investment). However, in conjunction with evidence showing FDI inflows generate some pacifying effects (see Model 9), the overall results suggest the presence of reciprocal causation (i.e. Hypothesis 3, the causal relationship runs in both directions).

Interestingly, the results also reveal that FDI inflows promote greater trade openness. This makes sense because foreign-owned businesses in developing countries often produce largely for export and buy considerable amounts of intermediate and raw materials from abroad. Both business practices should boost overall trade flows.

Table 5 reports results from the VAR models based on the UCDP/PRIO data, which has a lower fatalities threshold for constituting an episode of civil violence. Starting with Model 16, the results indicate that prior levels of trade openness reduce civil violence, with the parameter estimates of the second lag of trade openness being negative and statistically significant and the Granger causality test rejecting the null hypothesis of no causation. Conversely, the results for FDI inflows and its relationship to civil violence are not statistically significant, as evidenced by the parameter estimates and the Granger causality test. Overall, the VAR model produces very similar results despite the different measures of civil violence (cf. Models 13 and 16). Both results provide limited support for liberal peace proposition, as there is some evidence that trade openness but not FDI inflows reduce civil conflict.

Table 5.

VAR model with Granger causality tests: Occurrence of civil violence based on UCDP/PRIO data (25+ deaths per year).

	Occurrence civil violence_t (16)	FDI inflow_t (17)	Trade openness_t (18)
Occurrence civil violence_{_t – 1}	3.622*** (0.223)	-0.055(0.162)	-1.655** (0.484)
Occurrence civil violence_{_t – 2}	1.443*** (0.244)	-0.116(0.184)	0.337(0.477)
FDI inflows_{_t – 1}	-0.025(0.019)	0.471*** (0.050)	0.082** (0.030)
FDI inflows_{_t – 2}	0.015(0.008)	0.171** (0.054)	0.048(0.035)
Trade openness_{_t – 1}	0.008(0.007)	0.015^ϯ (0.008)	0.825*** (0.036)
Trade openness_{_t – 2}	-0.016* (0.007)	-0.005(0.007)	0.1251*** (0.034)
Observations	4468	4464	4470
Countries	103	103	103
R² (pseudo R²)	0.798	0.402	0.920
Granger causality F-tests
Occurrence civil violence →	—	1.82(0.403)	24.13*** (.000)
FDI inflows →	3.26(0.196)	—	14.99** (0.001)
Trade openness →	8.90* (0.012)	5.62(0.060)	—

Notes. Model 16 estimated with logistic regression with random effects. Models 17 and 18 estimated with generalized least squares regression with random effects. Robust standard errors clustered on countries shown in parentheses. Time trend and constant included but not reported.

***

p < .001, **p < .01, *p < .05, ^ϯp < .10.

Finally, the last two models examine FDI inflows and trade openness as left-hand-side regressors. This generates one notable difference from the earlier VAR models. Now, given the lower fatalities threshold, prior occurrences of civil violence no longer reduce FDI inflows, at least to a statistically significant degree, but the negative effect on trade flows persists. This finding suggests that the relationship between economic globalization and civil violence is sensitive to the fatalities threshold used to measure domestic conflicts. In substantive terms, it infers that low-intensity civil conflicts are unlikely to disrupt the FDI plans of global economic actors operating in these countries, but the downward pressure on global trade is still evident. Combined, the VAR models assessing the UCDP/PRIO data, as well as the CSP/MEPV data, find evidence of reciprocal causation (i.e. Hypothesis 3), although the stronger effect seems to run from civil violence to declining global trade and investment.

Table 6 shows results from VAR models of the onset of civil violence for the CSP/MEPV and the UCDP/PRIO data. These results should be interpreted with caution as the 1-year and 2-year lags of civil violence onset cannot be estimated for Models 19 and 22 due to perfect collinearity. This essentially leaves these two models without important control variables. Nonetheless, the results provide further evidence that global market integration generates very limited pacifying effects for developing countries. The exception is the second-year lag of trade openness, which for both the CSP/MEPV and the UCDP/PRIO data results in negative and statistically significant effects of civil violent onset. In terms of reverse causation, the findings show that the onset of civil violence reduces FDI inflows and trade openness across both data sources, although not all these effects are statistically significant. Combined, these findings point toward a reciprocal causation (i.e. Hypothesis 3), with the pacifying effects of economic globalization being smaller than the disruptive effects of civil violence.

Table 6.

VAR models of the onset of civil violence using CSP/MEPV and UCDP/PRIO data.

	CSP/MEPV data(500+ deaths over conflict)			UCDP/PRIO data(25+ deaths per year)
	Civil violenceonset(19)	FDI inflow(20)	Trade openness(21)	Civil violenceonset(22)	FDI inflow(23)	Trade openness(24)
Civil violence onset_{_t – 1}	—	-0.941* (0.034)	-1.455^ϯ (1.003)	—	-0.144(0.239)	-1.332* (0.706)
Civil violence onset_{_t – 2}	—	-0.188(0.333)	3.130(1.093)	—	-0.313(1.119)	0.622(1.803)
FDI inflows_{_t – 1}	-0.013(0.034)	0.545*** (0.100)	0.051(0.054)	-0.002(0.022)	0.477*** (0.057)	0.056(0.540)
FDI inflows_{_t – 2}	0.000	0.145	0.011	0.011	0.196**	0.045
	(0.026)	(0.102)	(0.052)	(0.014)	(0.061)	(0.045)
Trade openness_{_t – 1}	0.018(0.011)	0.008(0.009)	0.808*** (0.038)	0.009(0.010)	0.009(0.008)	0.812*** (0.038)
Trade openness_{_t – 2}	-0.034** (0.011)	-0.001(0.010)	0.145*** (0.369)	-0.021* (0.010)	-0.001(0.008)	0.141*** (0.036)
Observations	3256	3372	3372	3305	3458	3458
Countries	101	101	101	99	99	99
R² (pseudo)	0.008	0.47	0.91	0.004	0.41	0.95

Notes. Model 19 and 22 estimated with logistic regression with random effects. Models 20, 21, 23, and 24 estimated with generalized least squares regression with random effects. Robust standard errors clustered on countries in parentheses. Time trend and constant included but not reported in all equations.

***

p < .001, **p < .01, *p < .05, ^ϯp < .10.

Robustness Checks

My analysis ends by performing several robustness checks. For the first three checks, I re-run my VAR models using the same specification but now amend the sample in ways discussed below. The checks examine the occurrence of civil violence, first with the CSP/MEPV data and then with the UCDP/PRIO data. The presentation of the results, shown in Table A2 in the Appendix 1, is slightly abridged although the basic details are fully reported.

The first robustness check assesses whether my general findings are skewed by the Cold War and its effect on civil violence. During this period, the United States increased trade and investment with strategic allies in Latin America and elsewhere, hoping to support these regimes. Similarly, proxy conflicts between the Soviet Union and the United States spurred intra-societal violence in some developing countries. My panel regression models include a dummy variable to capture these period effects, but my VAR models do not. Thus, as a robustness check, I re-run my VAR models but constrain the sample period to 1990 onward, thereby bypassing the Cold War era. As shown in Table A2, Check 1, the results from this robustness check continue to support the notion that FDI inflows and trade openness have limited pacifying effects, but that outbreaks of civil conflict distinctly reduce FDI inflows and trade openness. This finding holds for the CSP/MEPV and the UCDP/PRIO measures of civil violence, and generally infers that economic globalization’s purported pacifying effects are mostly spurious, as the relationship seem largely driven by reverse causation (as suggested by Hypothesis 2).

A second robustness check considers the possibility that petrostates are distorting my findings. Petrostates are prone to civil violence because their levels of development and state capacities are typically lower than their GDP per capita would suggest and because their rentier economies often create enriched elites whose control over prized resources can be directly challenged through armed insurrections. Yet, they also have high levels of trade openness, due to their extensive exportation of high-value oil and gas. My panel models but not my VAR models include controls to capture these effects. Hence, as a robustness check, I re-run the VAR models but drop the petrostates from my sample. The sample period remains from 1990 onward. As shown in Table A2, Check 2, this robustness check produces similar results to the first robustness check, with Check 2 showing even less reduction in civil violence due to increases in global trade and investment.

A third robustness check considers whether my results hold under a broader sample of countries. Although my sampling strategy is straightforward and systematic, it excludes a handful of countries that transitioned from “developing” to “developed” status over the sample period. This happens because my original sample selects countries deemed developing in 2018, but the sample period covers several decades. Looking at earlier lists of developing countries from the UN reveals several countries excluded for this reason, including most notably Israel, South Korea, and Singapore. Furthermore, my sampling strategy excludes post-socialist countries arising from the break-up of the Soviet Union and the end of communism in Eastern and Southern Europe. Some observers may feel these countries belong in my study’s sample as well. Therefore, as a robustness check, I re-run my VAR models but extend the sample to include all UN member states with populations greater than 0.5 million in 2018, except for the affluent democracies of Western Europe, North America, Oceania, and Japan. This broadens my sample from 103 to 127 countries, but the sample period remains 1990 onward.

This robustness check finds some evidence that economic globalization modestly reduces civil violence within this broad sample of countries (see Table A2, Check 3). In particular, the effect of FDI inflows on civil violence is negative and marginally significant for the CSP/MEPV data but not for the UCDP/PRIO data, and the effect of trade openness on civil violence is negative and marginally significant for the UCDP/PRIO data but not for the CSP/MEPV data. These results show modest pacifying effects. However, the evidence of reverse causation remains strong, as incidences of civil violence continue to dampen FDI inflows and trade openness within this enlarged sample of countries.

Another robustness check returns to the original sample but now examines whether the lack of control variables in my VAR models alter my findings. In this regard, one obvious consideration is that large countries are not only prone to civil violence, but they are also less reliant of trade due to their large internal markets. This means population size is a confounding factor, which in normal regression models would be held constant through its inclusion as a control variable. Check 4, shown in Table A3 in the Appendix 1, includes population size as a control variable and re-runs the VAR models. Pertaining to my variables of interest, the new parameter estimates are nearly identical to the old ones, and as expected, population size is positively associated with civil violence but negatively associated with trade openness.

A final robustness check returns to my original panel models and considers whether the results hold when the petrostates are dropped from the sample. Check 5, shown in Table A4, essentially revisits Models 4, 8, 10, and 12. The revised results change in very limited ways. The one exception is that trade openness now shows a modest pacifying effect on the onset measure when using the CSP/MEPV data. Otherwise, the results are essentially the same, and overall continue to support my general conclusions that the pacifying effects of global market integration are negligible.

Conclusion

This study reassessed the claim that global market integration fosters civil peace in developing countries by testing the effects of FDI and trade openness on the occurrence of civil violence in a large sample of developing countries. Building on ideas from economics, political science, and sociology about the adaptive and reflective nature of human actors and social processes, as well as other recent scholarship recognizing the complexity of peace and conflict dynamics, this study questions the nature of causation between economic globalization and civil peace and how it should be modeled. A range of alternative hypotheses were developed and tested with methodological techniques more adept at capturing complexity than those typically used to address this topic. In particular, the use of VAR models and Granger causality tests allow me to untangle the causal relationships among civil violence, trade openness, and FDI inflows, making it possible to test hypotheses about reciprocal and reverse causation.

My empirical evidence generates limited support for the notion that economic globalization fosters civil peace in developing countries. On this question, I find some evidence that global trade and investment generate pacifying effects, although these effects seem small and occur mainly in a broad sample that includes post-socialists and newly affluent countries. Conversely, my evidence more clearly illustrates that civil conflict reduces economic globalization (reverse causation), and that civil conflict has a strong autoregressive quality, meaning that once it begins, it has a self-sustaining capacity that propels it into the future. Combined, these findings point toward a reciprocal causation in which civil conflict disrupts global trade and investment more than global trade and investment fosters civil peace. This conclusion is consistent with research by Magee and Massoud (2011) and by Martin et al. (2008) and it generally holds across different measures of civil violence and different sampling strategies.

However, I stop short of concluding that economic globalization cannot function as a force for civil peace in developing countries. One idea not examined in this study is whether the purported pacifying effects of global trade and investment arise after a substantial delayed, coming many years after the initial investment. Clearly, development is essential for the construction of civil peace, and development through autarky seems implausible. In this way, liberal institutions, such as markets and rule-based governance, seem foundational to civil peace in the present era of global capitalism, but more neoliberal notions of development, coming in the guise of large-scale structural adjustments and fully liberalized markets for trade, investment, and finance, may generate the types of negative externalities that cause social and political instability. Crucially, whether more open economies engender civil peace or civil violence may hinge on the state, and whether it can redistribute market rents in ways that bolster development rather than allowing them to be appropriated by foreign firms and local elites (Evans, 1995; Peterson and Wen, 2021; Pinto and Zhu, 2022). Hence, a strong and accountable state may condition how global trade and investment affect the prospects for civil peace in developing countries. Future research may wish to address this question more carefully.

Nevertheless, if economic openness prompts peace via a long-delayed effect, this is still problematic for proponents of the liberal peace theory, because this notion has not been considered. It is also problematic for overly simplistic narratives of liberal peace used to open national markets to global capitalism (see Millar, 2016). But I cannot rule out, on the basis of this study alone, that economic globalization does not, over the longer term, have greater positive effects than those found in this study. Such questions must be tackled in future studies.

Footnotes

Appendix 1

Table A4.

Robustness check on panel models without petrostates in sample.

	Civil Violence
	Occurrences		Onsets
	CSP/MEPV	UCDP/PRIO	CSP/MEPV	UCDP/PRIO
Check 5
FDI inflows (% GDP) (t – 1)	-0.077(0.088)	-0.126^ϯ (0.073)	0.117(0.096)	0.050(0.099)
Trade openness (% GDP) (t – 1)	-0.141(0.144)	-0.180(0.149)	-0.375^ϯ (0.208)	-0.123(0.170)
All control variables included?	Yes	Yes	Yes	Yes
Observations	3755	3764	2881	2975
Number of countries	86	86	84	82
Pseudo R²	0.740	0.628	0.018	0.012

***

p < .001, **p < .01, *p < .05, ^ϯp < .10.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the Leverhulme Trust (grant no. RF-2021-530).

ORCID iD

Christopher Kollmeyer

Notes

Author biography

Christopher Kollmeyer is a senior lecturer in the Department of Sociology, University of Aberdeen, United Kingdom. His research uses advanced quantitative methods to understand large-scale changes in the social structures and political economies of modern societies and how they affect a range of outcomes of interest to social scientists, including income inequality, trade union participation, and civil violence.

References

Anderton

Carter

(2001) The impact of war on trade: An interrupted times-series study. Journal of Peace Research 38(4): 445–457.

Barbieri

Reuveny

(2005) Economic globalization and civil war. Journal of Politics 67(4): 1228–1247.

Beck

Katz

Tucker

(1998) Taking time seriously: Time-series-cross-section analysis with a binary dependent variable. American Journal of Political Science 42(4): 1260–1288.

Blanton

Apodaca

(2007) Economic globalization and violent civil conflict: Is openness a pathway to peace? Social Science Journal 44(4): 599–619.

Bornschier

Chase-Dunn

(1985) Transnational Corporations and Underdevelopment. New York: Praeger.

Boswell

Dixon

(1990) Dependency and rebellion: A cross-national analysis. American Sociological Review 55(4): 540–559.

Box-Steffensmeier

Freeman

Hitt

, et al (2014) Time Series Analysis for the Social Sciences. Cambridge, MA: Cambridge University Press.

Bussmann

(2010) Foreign direct investment and militarized international conflict. Journal of Peace Research 47(2): 143–153.

Bussmann

Schneider

(2007) When globalization discontent turns violent: Foreign economic liberalization and internal war. International Studies Quarterly 51(1): 79–97.

10.

Carter

Signorino

(2010) Back to the future: Modeling time dependence in binary data. Political Analysis 18(3): 271–292.

11.

Cheatham

Labrador

(2021) Venezuela: The rise and fall of a petrostate. Council on Foreign Relations. Available at: https://www.cfr.org/backgrounder/venezuela-crisis (accessed 2 March 2022).

12.

Chen

Zhou

(2021) Whose trade follows the flag? Institutional constraints and economic responses to bilateral relations. Journal of Peace Research 58(6): 1207–1223.

13.

Clark

Linzer

(2015) Should I use fixed or random effects? Political Science Research and Methods 3(2): 399–408.

14.

Colgan

(2015) Oil, domestic conflict, and opportunities for democratization. Journal of Peace Research 52(1): 3–16.

15.

Collier

Hoeffler

(1998) On economic causes of civil war. Oxford Economic Papers 50(4): 563–573.

16.

Collier

Hoeffler

(2002) On the incidence of civil war in Africa. Journal of Conflict Resolution 46(1): 13–28.

17.

Crenshaw

Robison

(2010) Political violence as an object of study: The need for taxonomic clarity. In: Leicht

Jenkins

(eds) Handbook of Politics. Handbooks of Sociology and Social Research. New York: Springer, pp. 235–246.

18.

De Coning

(2020) Adaptive peace operations: Navigating the complexity of influencing societal change without causing harm. International Peacekeeping 27(5): 836–858.

19.

De Soysa

Fjelde

(2010) Is the hidden hand an iron fist? Capitalism and civil peace, 1970-2005. Journal of Peace Research 47(3): 287–298.

20.

Dixon

Boswell

(1996) Dependency, disarticulation, and denominator effects: Another look at foreign capital penetration. American Journal of Sociology 102(2): 543–562.

21.

Dollar

Kraay

(2002) Growth is good for the poor. Journal of Economic Growth 7(3): 195–225.

22.

Doyle

(1986) Liberalism and world politics. American Political Science Review 80(4): 1151–1169.

23.

Enders

(2014) Applied Econometric Time Series (4th edn). Hoboken, NJ: John Wiley & Sons.

24.

Evans

(1995) Embedded Autonomy: States and Industrial Transformation. Princeton, NJ: Princeton University Press.

25.

Fang

Yang

(2011) Wage increases, wage convergence, and the Lewis turning point in China. China Economic Review 22(4): 601–610.

26.

Fearon

(2003) Ethnic and cultural diversity by country. Journal of Economic Growth 8(2): 195–222.

27.

Feenstra

Inklaar

Timmer

(2019) Penn World Table, 9.1. University of Groningen: Groningen Growth and Development Centre. Available at: https://www.rug.nl/ggdc/productivity/pwt/ (accessed 7 August 2019).

28.

Fein

(1995) Murder in the middle: Life-integrity violations and democracy in the world, 1987. Human Rights Quarterly 17(1): 170–191.

29.

Feldman

Sadeh

(2018) War and third-party trade. Journal of Conflict Resolution 62(1): 119–142.

30.

Flaten

de Soysa

(2012) Globalization and Political Violence. International Interactions 38: 622–646.

31.

Frank

(1966) The development of underdevelopment. Monthly Review 18(4): 17–31.

32.

Frankel

Romer

Cyrus

(1996) Trade and growth in East Asian countries: Cause and effect? NBER Working Paper No. 5732, August. Cambridge, MA: National Bureau of Economic Research.

33.

Galtung

(1971) A structural theory of imperialism. Journal of Peace Research 8(2): 81–117.

34.

Gartzke

(2007) The capitalist peace. American Journal of Political Science 51(1): 166–191.

35.

Gartzke

(2003) War, peace, and the invisible hand: Positive political externalities of economic globalization. International Studies Quarterly 47(4): 561–586.

36.

Giddens

(1993 [1976]) New Rules of Sociological Method: A Positive Critique of Interpretative Sociologies. Stanford, CA: Stanford University Press.

37.

Giddens

(1984) The Constitution of Society: Outline of the Theory of Structuration. Berkeley, CA: University of California Press.

38.

Gissinger

Gleditsch

(1999) Globalization and conflict: Welfare, distribution, and political unrest. Journal of World-systems Research 5(2): 327–365.

39.

Gleditsch

Wallensteen

Eriksson

, et al. (2002) Armed conflict 1946–2001: A new dataset. Journal of Peace Research 39(5): 615–637.

40.

Goodhand

(1999) From wars to complex political emergencies: Understanding conflict and peace-building in the new world disorder. Third World Quarterly 20(1): 13–26.

41.

Granger

CWJ

(1969) Investigating causal relations by econometric models and cross-spectral methods. Econometrica 37(3): 424–438.

42.

Gurr

Duvall

(1973) Civil conflict in the 1960s: A reciprocal theoretical system with parameter estimates. Comparative Political Studies 6(2): 135–169.

43.

Hegre

(2001) Toward a democratic civil peace? Democracy, political change, and civil war, 1816–1992. American Political Science Review 95(1): 33–48.

44.

Hegre

(2014) Democracy and armed conflict. Journal of Peace Research 51(2): 159–172.

45.

Hegre

Oneal

Russett

(2010) Trade does promote peace: New simultaneous estimates of the reciprocal effects of trade and conflict. Journal of Peace Research 47(6): 763–774.

46.

Hibbs

(1973) Mass Political Violence: A Cross-National Causal Analysis. New York: John Wiley & Sons.

47.

Holtz- Eakin

Newey

Rosen

(1988) Estimating vector autoregressions with panel data. Econometrica 56(6): 1371–1395.

48.

Hosmer

Lemeshow

Sturdivant

(2013) Applied Logistic Regression (3rd edn). Hoboken, NJ: John Wiley & Sons.

49.

Kant

(1970 [1795]) Perpetual Peace: A Philosophical Sketch. Chapter 5 in Kant Political Writings, Reiss (Editor). Cambridge, MA: Cambridge University Press.

50.

Keshk

OMG

Reuveny

Pollins

(2010) Trade and conflict: Proximity, country size, and measures. Conflict Management and Peace Science 27(1): 3–27.

51.

Kim

Rousseau

(2005) The classical liberals were half right (or half wrong): New tests of the “Liberal Peace” 1960–88. Journal of Peace Research 42(5): 523–543.

52.

Kinne

(2012) Multilateral trade and militarized conflict: Centrality, openness, and asymmetry in the global trade network. The Journal of Politics 74(1): 308–322.

53.

Kinne

(2014) Does third-party trade reduce conflict? Credible signaling versus opportunity costs. Conflict Management and Peace Science 31(1): 28–48.

54.

Kollmeyer

(2009) Consequences of North–South trade for affluent countries: A new application of unequal exchange theory. Review of International Political Economy 16(5): 803–826.

55.

Kollmeyer

(2024) Structural position in the global economy and major episodes of civil violence, 1970 to 2018. Sociology of Development 10(1): 1–28.

56.

Krause

Suzuki

(2005) Causes of civil war in Asia and Sub-Saharan Africa: A comparison. Social Science Quarterly 86(1): 160–177.

57.

Lewis

(1954) Economic development with unlimited supplies of labour. The Manchester School 22(2): 139–191.

58.

London

Robinson

(1989) The effect of international dependence on income inequality and political violence. American Sociological Review 54(2): 305–307.

59.

Lucas

(1986) Adaptive behavior and economic theory. Journal of Business 59(4): S401–S426.

60.

Lucas

Sargent

(1981) Rational Expectations and Econometric Practice. Minneapolis, MN: University of Minnesota Press.

61.

Magee

CSP

Massoud

(2011) Openness and internal conflict. Journal of Peace Research 48(1): 59–72.

62.

Mahoney

(2000) Path dependence in historical sociology. Theory and Society 29(4): 507–548.

63.

Mandelbaum

(2002) The Ideas That Conquered the World: Peace, Democracy, and Free Markets in the 21st Century. New York: Public Affairs.

64.

Marano

Cuervo-Cazurra

Kwok

(2013) The impact of conflict types and location on trade. International Trade Journal 27(3): 197–224.

65.

Marshall

(2019) Major Episodes of Political Violence (MEPV), 1946–2018 [data set]. Fairfax, VA: Center for Systemic Peace, George Mason University. Available at: http://www.systemicpeace.org (accessed 14 July 2019).

66.

Marshall

Gurr

Jaggers

(2019) Polity IV Project, Political Regime Characteristics and Transitions, 1800-2018 [dataset]. Fairfax, VA: Center for Systemic Peace, George Mason University. Available at: http://www.systemicpeace.org (accessed 14 July 2019).

67.

Martin

Mayer

Thoenig

(2008) Civil wars and international trade. Journal of the European Economic Association 6(2–3): 541–550.

68.

Mihalache-O’Keef

(2018) Whose greed, whose grievance, and whose Opportunity? Effects of foreign direct investments on internal conflict. World Development 106: 187–206.

69.

Millar

(2016) Local experiences of liberal peace: Marketization and emergent conflict dynamics in Sierra Leone. Journal of Peace Research 53(4): 569–581.

70.

Millar

(2020) Toward a trans-scalar peace system: Challenging complex global conflict systems. Peacebuilding 8(3): 261–278.

71.

Mousseau

(2012) Capitalist development and civil war. International Studies Quarterly 56(3): 470–483.

72.

Muller

(1985) Income inequality, regime repressiveness, and political violence. American Sociological Review 50(1): 47–61.

73.

Muller

Weede

(1990) Cross-national variation in political violence: A rational action approach. Journal of Conflict Resolution 34(4): 624–651.

74.

Oneal

Russett

(1997) The classical liberals were right: Democracy, interdependence, and conflict, 1950-1985. International Studies Quarterly 41(2): 267–294.

75.

Peterson

Wen

(2021) International trade, cooperation, and conflict: The role of institutions and capabilities. Foreign Policy Analysis 17(4): orab027.

76.

Pinto

Zhu

(2022) Brewing violence: Foreign investment and civil conflict. Journal of Conflict Resolution 66(6): 1010–1036.

77.

Pollins

(1989) Does trade still follow the flag? American Political Science Review 83(2): 465–480.

78.

Prebisch

(1950) The Economic Development of Latin America and Its Principal Problems. United Nations Economic Commission for Latin America. New York: United Nations.

79.

Rosecrance

(1986) The Rise of the Trading State: Commerce and Conquest in the Modern World. New York: Basic Books.

80.

Ross

(2012) The Oil Curse: How Petroleum Wealth Shapes the Development of Nations. Princeton, NJ: Princeton University Press.

81.

Sadeh

Feldman

(2020) Globalization and wartime trade. Cooperation and Conflict 55(2): 235–260.

82.

Sims

(1980) Macroeconomics and reality. Econometrica 48(1): 1–48.

83.

United Nations (2019) World Economic Situation and Prospects: Statistical Annex. Department of Economic and Social Affairs. New York: United Nations.

84.

Vreeland

(2008) The effect of political regime on civil war: Unpacking anocracy. Journal of Conflict Resolution 52(3): 401–425.

85.

Walter

(2009) Bargaining failures and civil war. Annual Review of Political Science 12: 243–261.

86.

Weede

(2004) The diffusion of prosperity and peace by globalization. The Independent Review 9(2): 165–186.

87.

White

(1993) On measuring political violence: Northern Ireland, 1969 to 1980. American Sociological Review 58(4): 575–585.

88.

World Bank (2019) World Development Indicators [database]. Washington, DC: World Bank.

Does economic globalization promote civil peace in developing countries?

Abstract

Keywords

Introduction

Theory and Empirical Studies

Development and Liberal Peace

Economic Globalization and Civil Peace

Alternative Hypotheses

Rents and Negative Externalities

Reverse or Reciprocal Causation

Temporal Dependence and Dynamic Effects

Data and Methods

Measuring Civil Violence

Measuring Economic Globalization

Control Variables

Analytical Strategy

Panel Regression

VAR

Findings

Panel Regression Models

VAR Models

Robustness Checks

Conclusion

Footnotes

Appendix 1

Funding

ORCID iD

Notes

Author biography

References