Revisionist Conflict and State Repression

International conflict can be fought over a variety of issues. Despite major progress in recent decades, political science’s understanding of the way that issues influence the onset and escalation of international conflict remains limited (Diehl, 2013). For decades, issues were considered a secondary cause for international conflict compared to attributes of the states fighting, such as military capability (Diehl, 1992). The foundational literature in the issue approach to understanding conflict (Mansbach and Vasquez, 1981; Vasquez, 1983; Vasquez and Mansbach, 1984) focuses on the tangibility or intangibility of issues and argues the less tangible issues between disputing states are, the more likely conflict is to arise between them. International conflict researchers have typically employed three different issue dimensions: territory, policy, and regime. Territorial conflicts are ones in which a revisionist state seeks to take (or take back) some piece of land. Policy conflicts are ones in which one state attempts to alter the foreign policy practices of another states, whereas regime conflicts are those in which a state’s ruling regime is being targeted by another state for removal from power. Revisionist states are those that seek some change in the issue area over the course of the conflict, as thus any or all states involved may be revisionists (Jones et al., 1996).

Since the early work exploring the role of territorial conflict on the escalation to war (Vasquez, 1993), we know that territorial disputes are the most likely issue over which states escalate militarized disputes to war (Senese and Vasquez, 2008). Territory as an issue links to basic economic, psychological, and identity ties, and is an issue over which hardliners may be able to bid for domestic political gain, making hawkish foreign policy more likely (Vasquez, 1993; Goddard, 2006; Gibler, 2012).³ Research has also consistently shown that territory has dramatic domestic political consequences. Gibler (2012) explains that territorial threat can lead to the centralization of domestic political power in executives, thus promoting autocratic regimes, whereas the lack of territorial competition may allow for more open political systems to emerge. Owsiak (2012; 2013) also shows that having stable or settled borders not only promotes long-term peace but also democratization.⁴

Although there is a wealth of literature exploring interstate conflict over territory, or the role of borders and land in international conflict, there is much less that examines the specific characteristics of the policy and regime conflicts. Diehl (2013) is critical of the lack of development in research that theoretically and empirically explores these issues. Given the overwhelming importance that regime type plays in conflict processes, conflicts where a state’s regime is the primary focus should certainly receive more attention. Regarding policy conflicts and their categorization in data and (lack of) theoretical attention, he notes that policy conflicts are “a polyglot of cases that no one, to this writer’s knowledge, has sought to unpack” (Diehl, 2013, 269). That said, although the central focus of the work on the linkages between civil and international conflict is focused more on the movement of rebel groups across borders or how civil wars can lead to interstate disputes (Trumbore, 2003; Salehyan, 2009), these disputes would often be considered either focusing on policy or regime. Such disputes would clearly have domestic political importance. Being the targets of regime conflicts in particular may present an existential threat for governments and regimes and thus be of primary importance domestically. That said, of the issues in interstate conflict, regime disputes are the rarest, whereas policy disputes are far more common.⁵

Given the divergent goals that interstate conflicts have for revisionist states, it is unreasonable to think these conflicts affect domestic politics, specifically repression, in a uniform way. I contend, like other work (Davenport, 2007b; Gibler, 2012), that the best way to understand how conflict will affect repressiveness is to explore international conflict through the lens of domestic political institutions. We know from work that explores the democratic peace phenomenon (Huth and Allee, 2002; Bueno de Mesquita et al., 2003; Chiozza and Goemans, 2011), democracies and autocracies have different priorities and goals for interstate conflict involvement, with democracies being more careful in selecting conflicts abroad in which they can ensure victory (a public good), whereas autocratic states might be more willing to gamble about victory when pursuing interstate conflicts (Huth and Allee, 2002; Chiozza and Goemans, 2011).

Below I lay out my general expectations for democratic and autocratic states. Given that democracies and autocracies approach conflict and repression differently, it seems reasonable to tell two theoretical stories. Below I make an argument that among the different types of international conflict issues, territorial ones are the most likely to lead to increases in repression by democratic states, whereas interstate disputes fought over policy disagreements are most likely to lead to increased by repression.

Democratic and autocratic states have fundamentally different approaches when it comes to repression. Ideally and conceptually, we imagine democracies to be systems in which leaders are kept in check through regular elections and must provide the public with generally accessible services and security (Bueno de Mesquita et al., 2003). We know from prior work that democracies are also likely to be much less repressive, on the whole, than autocratic states (Davenport and Armstrong, 2004; Bueno de Mesquita et al., 2005; Davenport, 2007b). Davenport (2007b) provides the most comprehensive investigation of how democracy and conflict interact and affect repression. He argues there are two major forces working to keep repression in democracies low. One is voice, which is the public choice mechanisms of elections, and the other is veto, which is the ability of political institutions to check the executive. Democracies typically maintain lower repression because they not only have to rely on public support to stay in office (voice) and repressing elements of the public may hurt their chances electorally, but also because of institutional checks at the elite level (veto). Davenport (2007b) finds that conflict can weaken both of these mechanisms’ ability to check against increases in state repression. A major domestic factor motivating democratic policy making is the need to deliver public goods. Selectorate theory (Bueno de Mesquita et al., 2003) explains that democracies avoid certain kinds of conflicts (i.e. ones they think they would have a hard time winning) because winning conflicts can be considered delivering on public goods. One may also think of the issues at stake in conflicts as corresponding to goods (either public or private). A government’s ability to win a conflict in a certain issue area may be seen as providing a public good or some share of private goods (beyond the regular public good of national security). If a conflict is salient enough, democratic publics may view repression during conflict as part of delivering the good of security (domestically as well as internationally), and thus may actually support the increase of repression against certain groups of the population. The mechanism at work here is not the conflict or issue of the conflict itself per se, but rather that the salience of it may lead to a public sense of insecurity that leads to at least tacit approval of increased repression (likely against minority groups seen as fifth columns or those opposed to the conflict effort of the state). We may expect this mechanism to be present when democracies increase repression in general, not just during interstate conflict. Other instances that might lead to acute and generalized sense of insecurity, such as after natural disasters (Wood and Wright, 2016), may also lead to a similar dynamic. But conflict may be a likely scenario in which we observe this mechanism at work.

Of the three conflict issues described above, victory in territorial conflict seems most likely to correspond to a “public” good. Although Bueno de Mesquita et al. (2003) emphasize the private nature of territory—seeing it as resources that can be distributed to a leader’s supporters—there are numerous reasons to think territorial salience is more complex and will contain multiple aspects that are salient to various groups for different reasons. Contested territory can be tied to all manner of intangible qualities that make it salient in a very public way (Hensel and Mitchell, 2005). Especially if territory can be tied to national identity or has some symbolic value, it becomes something that is not easily divisible. This can engender in a public a feeling of urgency around winning a conflict that other issues may not (as easily) provide. Thus, a public may be willing to go along with repression in the name of victory and the provision of the contested territory-as-public good. If there are groups tied to the “other” in the conflict, they may be particularly vulnerable to repression, as might groups opposed to the conflict. Policy and regime conflicts seem less likely to strike at the heart of a sense of public welfare the same way as territory. Aside from being more escalatory than other conflict issues (Vasquez and Henehan, 2001), territorial conflicts may also lead to greater in-group/out-group identification among the public (Hutchison and Gibler, 2007; Gibler, Hutchison and Miller, 2012). This should lead to a rallying effect around a government pursuing territorial competition abroad. Furthermore, Miller (2013) finds the citizenry in states that initiate territorial disputes are happier overall. Wright (2014) argues the public value of contested territory can drive repression domestically for democracies, finding that as territorial revisionist years become more deadly for democracies, they become increasingly likely to increase repression.

Although we know from prior research that democracies repress less than autocracies, in general there is no reason to think democratic publics would never support the use of repression by their governments. Indeed, when security concerns are paramount, publics become less tolerant of so-called “others” and may be more willing to support increased human rights abuse in the name of national security. Wallace (2013) finds experimental evidence among a sample of Americans, that people are more willing to support the use of torture against enemy combatants. Although Wallace finds a mitigating impact of international law prompts, a noticeable percentage (nearly 40%) of his samples still support the use of torture against enemy combatants. Conrad, Hill, and Moore (2017) find that different democratic institutions may lead to differing incentives for leaders to engage in certain types of torture. Conrad, Hill, and Moore (2017) describe a model of electoral democracies in which the public becomes more likely to support the increase of the use of torture, in part because of the public pressure on governments, noting that states that allow dissent openly are also more likely to experience “challenges to state authority in the form of crime, protest and terror, producing greater demand among the voting public” (Conrad, Hill and Moore, 2017; 4) to increase punishment against such bad actors, including torture.⁶ This is in contrast to autocratic states, where anti-state activity is more routinely cracked down against, and thus unlikely to see more public outcry for punishment.

Democracies are more likely to increase repression during territorial conflict when they are the revisionists in conflict. Although it seems plausible that states might increase repression when they are the ones defending the status quo of territory, I believe the revisionist states likely find the issue dimension more salient because they are the ones seeking change. Furthermore, it is plausible that revisionist states likely would need to secure an outright victory in disputes to achieve their territorial goals, whereas status quo states may only need to secure a stalemate to preserve the status quo. Thus, the issue of the contested territory may not be as resonant with the status quo state as the revisionist state. Thus, a government will be more likely to do what they can to win, including increasing repression against groups seen to be a hindrance to that effort, whether fifth columns or opposition. If, as Miller (2013) argues and finds, territorial initiators have happier citizens, the public may also be more supportive of leadership in general and thus more supportive of policies aimed at securing the contested territory. In other words, if repression occurs as the result of pursuing territory abroad, the public is likely to be supportive of it. Because of the public good nature of territory and the way competition over it translates rallying to support a government, I argue that democratic revisionists in territorial conflicts will become more likely to increase repression. I note there is research that emphasizes democracies should be less likely to contest territory (Gibler, 2012). My argument does not contest this notion, only noting that despite territorial revisionism by democracies being rarer than by autocracies, it is more likely to lead to increased repression by democracies when it does occur.

Autocratic regimes rely on the distribution of private goods to an elite group of supporters to maintain leadership. Repression, which keeps those outside of the winning coalition from mobilizing, is a normal practice of governance (Bueno de Mesquita et al., 2003). Along this line of thought, as long as the government can provide a consistent (or increasing) distribution of private goods to their support coalition, they should be able to stave off threats to their authority. International conflict can act as an exogenous shock to their distributive capacity, however. Like economic sanctions, which squeeze the resources available for autocrats to distribute (Wood, 2008), pursuing international conflict costs military resources and money to pursue. This could leave them with fewer resources available for distribution and make leaders vulnerable to internal challenges. As a result, one might expect, in general, autocratic states would be likely to increase repression when involved in conflict abroad.

That said, such a likelihood increase seems unlikely to be the same across all conflicts. Autocratic leaders often have a great amount of choice in the conflicts they pursue (Chiozza and Goemans, 2011; Huth and Allee, 2002). Although not all autocracies have uniform levels of policy-making freedom and face varying levels of domestic constraint (Weeks, 2014), they do have similar concerns that distinguish them from democracies. First, public opinion is not such a major constraining force for autocratic leaders, and they are more reliant on the distribution of private goods to a smaller group than the public (Bueno de Mesquita et al., 2003). Another distinguishing factor, emphasized especially by Chiozza and Goemans (2011), is that they are more likely to be removed in an irregular fashion, via coup, assassination, or arrest. This threat of irregular removal and what happens to leaders after such removals can be a major motivating force for involvement in conflict abroad. Repression is a standard policy through which autocratic leaders maintain control and power (Davenport, 2007c). Thus it is likely such states would take the opportunity in certain international conflicts to solidify control via increased repression. Certain conflicts will be more likely than others to provide such opportunities. I contend that more than the other issues, policy disputes can provide such opportunities and may be more likely to be domestically motivated than conflicts over other issues.

Whereas policy conflicts actually cover a number of different particular policies, as discussed by Diehl (2013), these are disputes fought to stop another state from pursuing certain foreign policy practices (Jones et al., 1996). An example of this is when, in 2001, China downed a United States (US) spy plane to response to US reconnaissance practices (Militarized Interstate Dispute 4280). Another example, from the same year, is China displaying shows of force in the Taiwan Strait in response to perceived Tawainese moves toward independence. This dispute resulted in shows of force by both China and the US by conducting naval exercises in the area. These are two seemingly very different reasons for engaging in disputes, despite two of same disputants. Both are coded as policy disputes. Because of the variance in potential reasons for disputes being coded as “policy” disputes, it can be daunting to come up with a precise expectation of how autocratic states might generally respond to repression. Such conflicts may be prime examples that fit a “gambling” foreign policy strategy, as described by Chiozza and Goemans (2011).

These are conflicts that may be less likely to escalate to war and do not concern (in general) existential issues (such as territorial integrity or the survival of the regime itself) (Vasquez and Henehan, 2001). They may therefore not face the squeeze in military resources (which can be used for repression) that potentially escalatory conflicts such as those over territory may lead to. In which case, conflict involvement may have diversionary incentives (Pickering and Kisangani, 2010). Thus, these conflicts may be an opportunity to shore up support from the elite and leaders may also take the opportunity to increase repression. From this argument, there also seems to be likely to be no difference in the motivation for repression for whether an autocratic state is revisionist in a policy dispute. Thus, policy conflict involvement seems the most likely issue in which autocratic states would increase repression.

Not all policy disputes are of the kind described above, however. Unlike disputes over spy planes or naval exercises, some policy disputes are much more domestically focused. Salehyan (2009) argues that ongoing civil conflict processes influence the likelihood of international conflict via the presence of external rebel bases. The hosting of a rebel base can lead to interstate conflict. Examining the case narratives from 1993–1999 (the years for which Salehyan (2009, 110–113) provides narrative analysis and for which the Correlates of War (COW) provides dispute narratives), I found that of the disputes that had a revision type and were tied to the presence of rebels in one country or another, all but three were coded as having a policy revision at stake.⁷ I believe such disputes will have direct consequences for the level of domestic repression. Such a possibility is accounted for by Chiozza and Goemans (2011) as well. They note that such international conflicts are direct opportunities to harm violent opposition directly to strengthen the state’s leadership politically. Trumbore further finds that states experiencing rebellion are more likely to use force abroad (Trumbore, 2003). Increased domestic repression would logically increase as a result of such pursuits of rebels across borders and the states that host them.

For this described dynamic of using international conflict as an opportunity to increase domestic support via repression, territorial conflict seems a less likely issue for this to occur than policy disputes. Territorial conflict may be equally salient for autocratic states as democratic states, but potentially for a different reason. Territory, in addition to its symbolic or intangible value, may also present an opportunity for resource grabs by states. Because of the potential private goods gain from taking territory, autocratic revisionists may find such conflicts very salient. But conflicts over territory are also the most likely to escalate to war. So although they may present an opportunity for an autocrat to strengthen their leadership position by taking more resources that can be used for private good distribution, they may also have to pour more military resources into the conflict to acquire those goods. Indeed, Gibler (2012) notes that the pursuit of territorial competition abroad may require building up of military capacity and the use of land armies. Whereas in the long term, this may provide larger militaries that can be repressive during peace time (or after the conflict is over), during periods of territorial conflict itself these military resources may have to be spent abroad, leaving a lower repressive capacity at home. As a result, autocratic states may not become more repressive during territorial conflict. When examining the effect of territorial revisionism on repression, Wright (2014) finds that autocratic states actually become less likely to increase repression as they experience more deadly territorial-revisonist conflicts, lending credence to the idea that as more resources are used in high-stakes conflicts, repressive capacity may diminish, at least in the short term. The above discussion about how autocracies approach territorial and policy conflicts differently yields the following key hypothesis:

The mechanisms underlying the increase in repression by democracies and autocracies in different issue conflicts are not the issues themselves, per se, but rather what the issues’ salience potentially represents to the leadership and key audiences of the different regime types. For democracies, increases in repression likely occur with at least tacit public backing, such as when the country as a whole feels threatened, because democratic leadership requires public backing to maintain itself. Territorial conflicts, although more rarely participated in by democracies, are more likely to be salient at a national-identity level than more narrowly focused policy conflicts. In contrast, although autocrats may have public backing for their repressive output, I would argue that for them to increase repression during conflict above already (likely) high levels of repression, it would need to be a salient threat to the regime’s ability to maintain its support with elites. Many policy conflicts may involve transnational rebel groups, which may specifically target autocratic regimes and thus present a more direct and salient threat to an autocratic leadership. Because the issues themselves are not direct analogues to the mechanisms themselves (public backing for repression or threats to the regime), I am only positing a probabilistic relationship between the issue type of international conflict and repression dynamics.

In addition to the above analyses, I have conducted several robustness checks, which are displayed in the online appendix to this article. Firstly, my above analyses split the sample according to regime type. I did this because theoretically I am arguing that democracies and autocracies have different processes for making decisions to repress and different reasons for engaging in conflict abroad. I also did this for ease of interpretation. Another route would have been to pool the full sample of observations and interact conflict issue with regime type. I conducted analyses along these lines and the results are substantively similar, with one exception. In the full sample analysis, the interaction between democracies and territorial revision is not significant in predicting imprisonment. Beyond this analysis, I also conducted a full sample interaction model including civil conflict at a lower threshold (25 or more deaths) and found substantively similar results.

My theoretical conception of regime type is dichotomous. In my abstract set up, the leadership of states is elected and accountable to the general public or subject to the whims of a small elite (Bueno de Mesquita et al., 2003). As such, I employed a commonly used dichotomous measure of democracy, a cutoff of six or higher on the Polity scale (Marshall and Jaggers, 2002). To explore the veracity of the above findings, I conducted additional analyses employing the Varieties of Democracy’s liberal democracy and electoral democracy (polyarchy) measures (Coppedge et al., 2016). I analyzed the full sample, interacting either the liberal democracy or polyarchy with conflict revision type, and found largely similar results to those above. The results employing the electoral democracy (polyarchy) variable are closest to those present into the main results, which should not be too surprising given the rather minimalist conception of democracy employed. The results employing the liberal democracy variable found significant results between the interaction of territorial revision and liberal democracy on the overall repression scale as well as for killing, which is surprising. The results of these robustness checks underscore the need for future research exploring the linkage between conflict behavior, repression, and different aspects of democratic states’ institutions.

In addition to the above concerns, one of the major potential questions that these above results raise is the degree to which the domestic repression and international conflict have reverse causation. In other words, does the increase (or as yet unobserved need for an increase) in repression lead to the onset of international conflict in the first place? Are the states that are more repressive the ones engaging in the conflicts that correlate to a subsequent increase in repression?

Although these questions are related to my study, they are somewhat beyond the scope of my analysis, because it is not taking on the question of conflict selection. That said, one way to try and get at this is to see if more repressive states are driving the findings that I observe above. To do that I ran additional models based on my primary analyses above (Table 1, Model 1; Table 2, Model 1). In these analyses, I interact the value for prior-year repression, included as a control above, with territorial revisionism for the democratic sample and interact prior repression with policy revisionism in the autocratic sample. I display these results in Table 3 below.

The results in Table 3 are suggestive that the more repressive states might drive the above findings only in the autocratic sample. In Model 1, for the democratic sample, we see the lower order term on territorial revisionism as statistically significant and positive, which means the least repressive democratic states are still likely to increase repression when revisionists in territorial MIDs. This suggests territorial MIDs are one of the rare “shocks” (when compared with autocracies) that lead to democratic states to increase repression. For the autocratic sample, however, neither the lower order term on policy revision nor the interaction term are statistically significant. The interaction term is positive and approaches one-tailed significance with a p value of 0.129. This suggests that more repressive autocracies might be increasing repression during policy MIDs.

The finding that more repressive autocracies are more repressive during conflict is perhaps not too surprising, given the research on the linkages between repression, civil conflict, and international conflict onset. Trumbore (2003) finds that states experiencing rebellion are more likely to engage in conflicts abroad, whereas Caprioli and Trumbore (2006) find that more repressive states are more likely to be involved in international conflicts. Beyond this, if policy MIDs really do contain more rebel chases, as the narratives of Salehyan’s (2009) cases suggest, then the findings for both my analysis in Table 2 and Table 3 are being driven by these cases. In other words, autocratic states experiencing rebellion (a direct threat to the elites and regime) are more likely to be repressive. When these rebellions lead to conflict abroad, these international conflicts also coincide with these increases in repression. These connected findings are all consistent with well understood mechanisms leading to autocratic repression.

To more fully explore the above question, one would need to explore jointly the processes of domestic repression and international conflict selection. This would need to likely incorporate an analysis of interstate conflict opponent in addition to modeling repressive targets. This is beyond the scope of the current project, but there is suggestive evidence that such a pursuit would be worthwhile.