Authors panelKristin M. Bakke, Andrew M. Linke, John O’Loughlin, Gerard Toal
Abstract
Post-war state-building is fraught with challenges as “war-makers” pivot to become “state-makers.” Citizen assessments of public good provision and physical security provide a measure of how state-building is perceived internally. State-building may also necessitate external dependence (Russia, for example, provides significant financial and military assistance to the post-Soviet de facto states), yet new state authorities want to be seen as more than puppets. We study the relationship between internal and external state-building dynamics in fostering citizen confidence in the post-war state. We use original population surveys to analyze public opinion and geographically disaggregated data on local violence from four post-Soviet de facto states—Abkhazia, Nagorno-Karabakh, South Ossetia, and Transdniestria—born of war. We examine the scalar relationships—from the individual embedded in the local context, to the regional (de facto territory) and supra-national (patron state and legitimacy in the international community)—that characterize them. We find that distrust of the patron state reduces trust in the de facto state president and translates into a lack of confidence in the prevailing order. As fears of conflict recurrence increase and disappointments about the economy worsen, these relationships are maintained across the pathways defined by the scale of patron trust-distrust.
Keywords
De facto states
Post-war societies
State-building
Local violence
Trust
External dependence
Public opinion
1. Introduction and argument
For armed groups in intrastate conflicts, winning the war against the state is just the beginning.1 A foundational challenge is for former “war-makers” to demonstrate to their own citizens that they now can make good on their (implicit or explicit) commitment to be “state-makers.” They have to convince their citizens that they are credible rulers, “not only able to kill and to destroy but to build and invest as well” (Schlichte, 2009, p. 96). In the absence of so doing, the post-war era is likely to be plagued by distrust and, possibly, cycles of violence, as no one can afford to let their guard down. The question, then, is, what explains why citizens do (or do not) have confidence in the post-war state’s authorities and prevailing order?
Coming out of a civil war, a central task on the road to make credible their commitment to rule is for the new authorities to be good state-builders (Flores & Nooruddin, 2011). Indeed, citizens’ confidence in post-war states likely hinges on their assessment of the provision of public goods such as welfare and, importantly, security in their local area. State-building is rarely, however, an exclusively internal endeavor. In the de facto states in the former Soviet Union, for example, Russia has played a central role as external patron. Russia provides significant financial and military assistance to Abkhazia, Transdniestria, and South Ossetia—and, through support of Armenia—indirectly to Nagorno-Karabakh. Yet for the authorities of post-war states, reliance on such external subventions and security measures complicates their claim to be credible rulers in their own right, as it creates a relationship of dependence. In particular, we argue that if citizens distrust the external patron, the domestic authorities’ efforts to foster citizen confidence in their rule might be jeopardized. Conversely, if the domestic population trusts the patron state, the domestic authorities’ credibility might be boosted by being associated with a benevolent helper.
To understand the dynamic relationship between the internal and external influences on people’s confidence in the post-war order, we utilize local-level data on violence and original survey data collected in 2010–2011 in the four post-Soviet de facto states cited above. Local-level data on violence allows us to investigate the importance of a context of post-war security (or lack thereof), whereas survey data enables us to examine social processes at the individual level. Given data collection challenges, there are few studies systematically examining individual attitudes across conflict-ridden states. One of our contributions to the emerging body of work employing surveys in post-war states (e.g. Blattman, 2009, Cassar et al., 2013, Samii, 2013, Bakke et al., 2014, Blair, 2016, De Juan and Pierskalla, 2016) is a geographically comparative perspective. Our analysis is territorially disaggregated and conceptually rooted in hierarchical scales of analysis, ranging from the individual embedded in the local context, to the regional (de facto territory), and the supra-national (patron state).
We begin by defining what we mean by de facto states and explain the “credible commitment problem” their authorities face vis-à-vis their residents. We develop our argument, which considers both the internal and external aspects of state-building that may shape people’s confidence in their rulers and prevailing order. The subsequent section describes the research design and data from the four post-Soviet de facto states, all born from violent struggles: Abkhazia (claimed by Georgia), Nagorno-Karabakh (claimed by Azerbaijan), South Ossetia (claimed by Georgia), and Transdniestria (claimed by Moldova). We then discuss our empirical findings and conclude.
1.1. De facto states born from violent struggles
In most separatist conflicts, in which non-state groups fight for independence or greater autonomy within a state’s border, the outcome entails no major change on the world political map. Relatively few struggles result in new states, and the separatists are more often either defeated or appeased with institutional solutions short of independence. In some cases, conflict results in the birth of so-called de facto states. A state is considered a de facto state if it possesses internal sovereignty, in the sense that it has administrative control over most or all of the territory it claims, but lacks international recognition as a state (external sovereignty) by the existing community of states (cf. Caspersen and Stansfield, 2010, Pegg, 1999).
The post-Cold War era has seen a proliferation of de facto states, most of them born from violent struggles with their parent states. The collapse of Yugoslavia saw the emergence of Republika Srpska Krajina (in Croatia) and Republika Srpska (in Bosnia-Herzegovina), as well as Kosovo after war and NATO intervention in 1999. The collapse of the Soviet Union led to the emergence of several breakaway “statelets,” including Chechnya, Nagorno-Karabakh, Abkhazia, South Ossetia, and Transdniestria. De facto states exist elsewhere across the world, too (for example, Somaliland and Iraqi Kurdistan), some also emerging during the Cold War (e.g. the Turkish Republic of Northern Cyprus and Taiwan). At least 21 de facto states have emerged since World War II (Caspersen & Stansfield, 2010, p. 4).2 More than half have either been recaptured by their parent state (Chechnya, for example) or, in a few cases, eventually recognized as states (Eritrea and East Timor have received full international recognition, whereas Kosovo is now recognized by some 111 out of 193 UN member states). Many de facto states endure despite their lack of international recognition.
All the post-Soviet de facto states were the result of violent separatist (or irredentist) struggles, each part of a dynamic of nationalist mobilization and counter-mobilization with their so-called parent state in the late 1980s-early 1990s. In Abkhazia and South Ossetia, the “titular” population after which the regions were named (not necessarily the majority of the population) sought autonomy and subsequently independence from Georgia; in Nagorno-Karabakh, the Armenian majority sought to exit Azerbaijan and join Armenia; in Transdniestria, a multiethnic coalition of Sovietized managers and workers mobilized against a pro-Romanian nationalist wave in Moldova. Table 1 provides an overview of the still-existing (as of 2017) post-Soviet de facto states—and, as of writing, we are witnessing a process of de facto state formation in eastern Ukraine.
Table 1. Still existing post-Soviet de facto states born of war.
Main separatist group(s) | Parent state | Violent conflict begins | Ceasefire/birth of de facto state | De facto state |
---|---|---|---|---|
Armenians/ Karabakhis | USSR/ Azerbaijan | 1988 | 1994 | Nagorno-Karabakh |
Dniester Slavs | Moldova | 1990 | 1992 | Transdniestria |
Ossetians | Georgia | 1991 | 1992 | South Ossetia |
Abkhaz | Georgia | 1992 | 1993 | Abkhazia |
1.2. Making credible promises to inhabitants of post-war states
One factor that may shape the relative endurance of de facto states—and shape stability in post-war states in general—is whether their inhabitants have faith in their authorities and prevailing order. By winning the war, the separatists have made a commitment, either implicit or explicit, to rule, and they now need to make that commitment credible.
A problem facing many war-torn and post-war states is precisely a credible commitment problem: the absence of an authority whom residents trust to protect their security (e.g. Posen, 1993, Walter, 2002). This absence of a credible central authority fosters distrust and makes way for spirals of insecurity and violence as groups within the state (think they) have to fend for themselves—particularly if accompanied by fear-laden myths of a zero-sum struggle (Brubaker, 2004, Kaufman, 2001). This security dilemma logic, first developed to explain behavior among states in the international system (Jervis, 1978), helps explain the dynamics between (former) warring groups in multi-ethnic states, in which the minority distrusts that the majority will have their interests at heart (Fearon, 1998). Yet the credible commitment problem is not just about fostering trust among former warring parties that the other “side” will stick to a peace agreement. It is just as much about fostering trust among the citizens that the post-war authorities will make good on their (implicit or explicit) promise to rule and provide public goods (Coyne and Boettke, 2009, Flores and Nooruddin, 2011, Keefer, 2008). This aspect of the credible commitment problem exists also after wars that end through military victories (and expulsion of the war-time enemy).
Although making credible their commitment to rule is a challenge for any post-war state government, it may be particularly pertinent in de facto states forged by violent territorial struggles. Given their lack of recognition as states, they have a double burden of earning their own inhabitants’ trust (demonstrating that they are up to the task), a challenge they share with any post-war state, as well as convincing the international community that they deserve to be states (Berg and Mölder, 2012, Caspersen, 2008, Caspersen, 2010). Indeed, even dysfunctional recognized states, such as Somalia, while often recipients of international criticism, do not have to convince the international community that they deserve to be states (cf. Krasner, 2009; Coggins, 2014).
Thus, an important theoretical (and policy) outcome in post-war states, including de facto states, is the perceived credibility of the ruler among the ruled. In this article, we think of credibility in two related ways. First, we want to know if citizens trust the executive power in the state. The very starting point in the security dilemma is the lack of trust in the ruler, the central authority—and there is by now a growing body of work emphasizing the importance of (and examining) political trust in post-war states (e.g. De Juan and Pierskalla, 2016, Wong, 2016). Second, we want to capture a broader sense of confidence in the prevailing order. We draw on the conceptualization of political legitimacy as citizens’ “actual quality consent” to rule. The philosopher Amanda Greene (2016) argues that the more widespread citizens’ quality consent, the more legitimate the rule. Consistent with our state-building argument, she believes that actual quality consent is shaped by citizens’ assessment of governance. The expression of quality consent, she argues, can be rather general, such as people affirming that things are going well. In the context of a post-war society, political legitimacy based on such a sense of confidence in the prevailing order is, like political trust, central for individuals overcoming the negative spiral of the security dilemma.
1.3. Internal and external influences on people’s confidence in their rulers
There are a number of reasons to believe (and evidence to support) the contention that confidence in a (de facto) state’s ruler and prevailing order is influenced by how successful the state is at providing its citizens with public goods (Bakke et al., 2014, Espinal et al., 2006, Flores and Nooruddin, 2016, Gilley, 2006, King, 2001, Norris, 1999, Organization for Economic Cooperation and Development (OECD), 2010). We draw on the insights of scholars who suggest that legitimate authority rests with an implicit (and often mythologized) social contract between ruler and ruled (Moore, 1978, Scott, 1972). The ruler provides benefits, most importantly social order, to the ruled, and the ruled, in turn, accept the ruler’s right to rule. Given the importance of security in the social contract between ruler and ruled (Lake, 2010, Wickham-Crowley, 1987), an ability to ensure “national security” both from external enemies and domestic instability is central to the credibility of both recognized and de facto states (Berg & Mölder, 2012).
As far as security goes, in all the post-Soviet de facto states, the wars concluded without formal settlements. Separatist forces, with variable and debated levels of Russian support, gained the upper hand in the armed struggle, leading to ceasefires. Whereas Transdniestria’s post-war record has been relatively free from political violence, it is a polity dominated by well-connected oligarchic networks (Bobick, 2011). Nagorno-Karabakh has been the scene of some elite-level struggles among former allies (De Waal, 2004), and there have been numerous clashes and sniper attacks on the boundary line with Azerbaijan. April 2016 saw a return to large-scale armed struggle, resulting in casualty levels not seen since the ceasefire brokered in 1994. Indeed, in 2011 and 2013, our fieldwork in Nagorno-Karabakh’s “border” regions revealed both long-standing and new tank emplacements and trenches—signs that the worry of war recurrence is a daily reality. Both Abkhazia and South Ossetia saw serious but brief outbreaks of fighting before the August War of 2008 (e.g. Welt, 2009). In August 2008, low-intensity skirmishes and invasion fear induced a Georgian military assault on South Ossetia, which triggered a short but deadly interstate conflict as Russian forces invaded to save their South Ossetian allies (Toal, 2017). Abkhaz forces, with the help of Russian air power, used the opportunity to establish control over the upper Kodor(i) Valley, which a Georgian warlord controlled until ousted by the Georgian government in 2006 (Marten, 2012). Both Abkhazia and South Ossetia have also seen internal political struggles that have occasionally spilled into violence, as well as criminal violence (International Crisis Group, 2007, International Crisis Group, 2010b).
Such violent incidents are likely to foster concerns about security, which, along with concerns about public goods provision more generally, raise the question of whether de facto states can do what states are “supposed to do.” Indeed, central to understanding the local population’s perceptions of de facto states (and post-war states more generally) is that the threat of violence is part of people’s consciousness (Linke & O’Loughlin, 2015). Some locales are more dangerous and conflictual than others (for example, because they are close to ceasefire lines or because of perceived hostility, as in the Gal(i) region of Abkhazia). If violence perpetrated by militants and military/police responses occur with perennial consistency in a region, residents in that locality are more likely to lack confidence in the central authority and prevailing order than where such events do not occur with regularity. Though there are nuances to this general expectation (we explore them below), the principle is straightforward. Reading about a violent attack distant from your home, workplace, and immediate social circle is different than personally fearing a deadly explosion in your daily activity space or hearing by word of mouth that a neighbor has been killed. We would expect events that are tangible to be more likely to affect attitudes than remotely occurring conflicts.
Yet state-building is rarely, if ever, a result of purely domestic dynamics. Post-war states need the help of external actors in fulfilling public goods provision—be that economic aid, democratization assistance, or even direct help in ruling and providing security (Brzoska, 2006, Fearon and Laitin, 2004, Krasner and Risse, 2014). Indeed, one way to overcome the security dilemma created by the absence of a credible central authority is for outside actors to serve as security guarantors (Fearon, 1998, Walter, 2002) or provide foreign aid (Flores & Nooruddin, 2011). In de facto states, non-recognition may increase the likelihood of external dependence on select or singular foreign patrons (King, 2001, Kolstø and Blakkisrud, 2008). As Nina Caspersen (2010, 82) notes, de facto states’ “lack of recognition and precarious position make them highly unattractive to foreign investors and they are, in most cases, also blocked from receiving international assistance and loans. As a consequence, most of these entities have to rely almost entirely on a patron state.”
Once external actors become involved in state-building processes—be it through economic aid, military intervention, security assistance, or as security guarantors—questions of legitimacy, dependency, and local ownership arise (cf. Caspersen, 2015, Paris and Sisk, 2007), as do questions about the credibility of the third party (Walter, 2002). Drawing on these insights, we posit that, to the degree that the external actor who backs the domestic authorities (in our case, the de facto state’s authorities) is considered untrustworthy, the authorities’ credibility will suffer. Conversely, if the domestic population thinks highly of their patron state’s leadership, the domestic authorities’ credibility will be boosted. Directly, this may go simply by way of association, based on a guilt-by-association mechanism (cf. Johnson, 2011). In particular, if an external actor exerts significant influence on the domestic authorities—in the form of, for example, providing critical military or economic support—people’s perception of the domestic authorities is unlikely to be independent of their view of the external actor exerting that influence. Indirectly, it is difficult for external actors to provide effective and efficient public good assistance if they are considered untrustworthy (Krasner and Risse, 2014, Suhrke, 2007), and, as a result, their involvement can tarnish the credibility of the local authorities. That is, while third parties are sometimes viewed as critical for successful conflict resolution and post-war stability and development, they can also serve the opposite role, undermining the credibility of the (de facto) state leadership in the eyes of large segments of their populations. Indeed, stated commitments to peace might be even less likely to be trusted with an untrustworthy powerful patron looming over the post-war society.
In Abkhazia, neither the entity’s state-building efforts nor people’s perceptions of the de facto state can be seen in isolation from substantial financial and military back-up from Russia (International Crisis Group, 2010a, Kolossov and O’Loughlin, 2011). For years, the authorities have been calibrating a fine balance between being grateful for Russian support, yet wanting to avoid being seen as entirely dependent on their powerful neighbor or subject to its dictates. In the words of the deputy foreign minister:
There is no direct attempt of the Russian Federation to influence our decisions; there is no dictating from Russia. I can’t imagine that happening. (…) Of course, the Russian presence here is felt. Abkhazia needs huge assistance: money and expertise. The only country offering that is Russia … 3
Indeed, since Russia’s recognition of Abkhazia in August 2008, in the wake of the Georgia-Russia war, a central debate in Abkhaz politics concerns the extent, depth and nature of Russian influence in the polity (Gerrits & Bader, 2016).
South Ossetia and Transdniestria are also recipients of Moscow’s aid, and Armenia serves as Nagorno-Karabkah’s patron (Kolstø, 2006). The former minister of industry in Transdniestria characterized Russia’s support as “symbolic aid.” This, he explained, took the form of favorable gas prices and support to the republic’s budget for pensions and education. The “symbolic aid,” he estimated, makes up about 20 percent of the republic’s budget.4 About Nagorno-Karabakh, Thomas De Waal (2004, 246) observes: “On an everyday level, Karabakh had become a province of Armenia. Karabakh Armenians were entitled to carry Armenian passports. Its currency was the Armenian dram. The budget was supported by free credits from the Armenian Finance Ministry” (for further examples, see O’Loughlin, Kolossov, & Toal, 2014).
Given this dependence on the external patron state, we are particularly interested in examining whether de facto state inhabitants perceive the leadership of the external patron to be trustworthy, expecting that high distrust of an external patron will have negative impact on people’s confidence in the de facto state’s authorities and prevailing order. As the external patron serves both as a security guarantor in these entities and backs them up financially, we also expect that there may be a conditional relationship at work. People’s perceptions of the danger of renewed warfare and experiences of violence will have a particularly damaging effect on the credibility of the de facto state if they do not trust the patron’s leadership. Similarly, because the patron is a source of financial support, people’s concerns about unemployment or lack of economic development will have a particularly damaging effect on local credibility if they do not trust the patron’s leadership. Conversely, if people are very worried about their physical and material security, trust in the patron state’s leadership might mitigate these worries. These conditional relationships, presented in Table 2, are at the heart of our analysis.
Table 2. Conditional expectations for the relationship proposed between trustworthiness of the external patron’s leadership, common state-building concerns among de facto state residents (physical and material security), and confidence in local rulers and prevailing order.
Empty Cell | Trust in patron state leadership | Distrust of patron state leadership |
---|---|---|
High worries about new war or lack of economic development | Medium confidence in de facto state’s ruler and prevailing order (patron trust can mitigate negative effects of high worries) | Low confidence in de facto state’s ruler and prevailing order |
Low worries about new war or lack of economic development | High confidence in de facto state’s ruler and prevailing order | Medium confidence in de facto state’s ruler and prevailing order (patron distrust may jeopardize positive effect of low worries) |
In sum, in assessing whether individuals have confidence in the de facto state authorites and prevailing order, we expect to see differences among individuals related to their experiences of violence locally, as well as the their personal perceptions of both governance within their de facto state and the patron state.
We indicate a note of caution on causation: We argue that there is a causal relationship between people’s views on state-building, conditional on their trust in the patron state’s leadership, and confidence in the ruler and prevailing order. There is a longstanding tradition of research on the determinants of political trust that relies on attitudinal variables (e.g. Mishler & Rose, 2001), and the state-building part of our argument is consistent with this body of research in post-war settings (e.g. Askvik et al., 2011, Bakke et al., 2014). That said, there are limitations to causal inference when using cross-sectional attitudinal variables. We try to overcome some of these concerns by using also non-perceptional measures of violence that capture people’s experiences of violence prior to the conduct of the surveys. Yet one objection to our argument is that it might well be that people’s trust in their own authorities shapes their trust in the patron state and not, as we argue, vice versa, via a similar association mechanism. There is, however, a strong theoretical reason why causality goes in the direction we propose. Both in developing countries and post-war societies, and certainly in post-war de facto states (Caspersen, 2010, King, 2001, Kolstø and Blakkisrud, 2008), external support to state-building often elicits intense debate about local ownership and local legitimacy because there is a relationship of dependence, in which the patron is the more powerful party. The recipient state needs the patron’s support to rule (but not vice versa), and the population in the recipient state is well aware that is the case. As such, it is of greater consequence for the recipient state authorities’ credibility than for the patron state authorities’ credibility whether the recipient state’s population trusts them.
2. Research design
2.1. Survey data
The public opinion surveys that allow us to explore state-building dynamics in Abkhazia (N = 1000), Nagorno-Karabakh (N = 800), South Ossetia (N = 460), and Transdniestria (N = 976) were designed by O’Loughlin et al. (2014). In each research site, shown in Map 1, the surveys were carried out by reputable private firms or organizations, employing local interviewers. Preliminary visits by the investigators and meetings with the presidential administrations and other local agencies ensured no interference with the surveys. For information about the conduct of the surveys and specifics of each locality, see O’Loughlin et al., 2011, O’Loughlin et al., 2013 and Toal and O’Loughlin, 2013, Toal and O’Loughlin, 2013b.

A number of steps were taken to ensure that we can engage in meaningful analyses across the four cases: the random selection of sampling points; the proportional distribution across the various local nationalities; the wide-ranging and lengthy number and nature of the questions (about 75 percent of the questions on any survey overlap with those in the three other sites); the use of local languages and trusted interviewers (including in Gal(i) in Abkahzia, the use of local Georgian/Mingrelian teachers as intermediaries and interview assistants); and the close timing of the surveys (March 2010 in Abkhazia, June 2010 in Transdniestria, October 2010 in South Ossetia, and November 2011 in Nagorno-Karabakh). We have elsewhere highlighted the special political and post-war conditions operating in the individual republics, which we here account for by analyzing the cases both individually and comparatively. This article focuses on the comparable struggle of leaders to gain their citizens’ confidence in an international environment in which their entities’ status as separate political units is challenged and dismissed as artificial by their parent states.
One problem we encountered in the Abkhazia survey, is that for many questions, the Georgian respondents opt for the “don’t know” option (see Bakke et al., 2014). Given the Georgian population’s precarious situation in Abkhazia (e.g. Human Rights Watch, 2011), this strategy is likely a way out of responding to politically sensitive and difficult questions. To avoid that these respondents are systematically excluded from the analysis by treating them as missing observations, we use copula methods to impute the answers (Nelson, 2010; for further description, see Bakke et al., 2014). This tendency of avoiding sensitive questions is unlikely to be driving “don’t know” responses in the other surveys. In Nagorno-Karabakh, all the survey respondents are Armenian, reflecting the ethnic make-up of the entity. In South Ossetia, the survey data analyzed here includes only South Ossetian respondents. In Transdniestria, the post-war state-building project has been built around creating a multi-ethnic entity (Kolstø & Malgin, 1998), hence there is less reason to expect personal insecurity to be driving “don’t know” answers among the responses of any ethnic group in particular. Relations between the ethnic groups there are significantly better than elsewhere, and the state constitution and government agencies guarantee equal rights to all groups, though in practice, ethnic Moldovans suffer discrimination in education, and in cultural and political entitlements.
2.2. Operationalization and description of variables
2.2.1. Dependent variables
The theoretical outcome of interest in this study is the credibility of the ruler among post-war states’ own citizens—in our case, de facto states’ inhabitants. As noted above, we think of credibility in two ways. First, given that doubts about the central authority is the theoretical starting point of the commitment problem that often characterize conflict-ridden or post-war states, we want to know if people have confidence in the executive power of the state. As such, we rely on a survey question that asks the respondents whether they trust the president (who is unnamed), based on a dichotomous yes/no question.5 The surveys show a strong variation from high trust in Nagorno-Karabakh (84 percent) and Abkhazia (82 percent) to lower rates in South Ossetia (68 percent) and Transdniestria (38 percent). Local contextual reasons for this variation are discussed in O’Loughlin et al. (2014). Second, as a broader and alternative measure, but with the same expectations, we want to capture credibility in the sense of confidence in the prevailing order. To do so, we analyze a question that asks our respondents whether things in the de facto state are moving in the right or wrong direction (a dichotomous question).6 In Nagorno-Karabakh, 62 percent indicate that things are “generally moving in the right direction”; in Abkhazia, 67 percent; in South Ossetia, 70 percent; and in Transdniestria, 28 percent. This is a common survey question, also used in the Eurobarometer surveys, which indicates the level of general (dis)satisfaction with the current state of affairs.
2.2.2. Independent variables
Because, we argue, confidence in the ruler and prevailing order are influenced by how well de facto governments perform the tasks that states do, we rely on variables that capture assessments of the provision of both material and security-related public goods. We use a survey question that asks people about economic public goods provision: “how big of a problem is lack of economic development or unemployment.” As for external security, we include a question that asks, “how big of a problem is the threat of another war?”7 The responses range from 1 (“no problem at all”) to 2 (“not a big problem”), 3 (“a rather big problem”), and 4 (“a very big problem”).
To assess perceptions of how trustworthy the external patron is, we rely on a question that asks respondents whether they trust the patron state’s leadership (again, there was no individual name prompt).8 We expect patron distrust to be negatively associated with respondents’ confidence in the president and prevailing order. In Nagorno-Karabakh, trust in the Armenian leadership among our respondents is 82 percent, while in Transdniestria, trust in the Russian leadership is at 70 percent. In South Ossetia, 87 percent of respondents trust the Russian leadership. In Abkhazia, trust in the Russian leadership is lower, at 74 percent, due to the lower trust among the Georgian/Mingrelian respondents (but for the titular ethnic group only, the ethnic Abkhaz, trust in the patron state is 87 percent).
To capture more than perceptions of safety and security, we include a non-perceptional, or objective, measure for local-level violence preceding the time of the surveys in each territory, expecting that people who live in geographic areas that have experienced violence in close proximity in the recent past are less inclined to have confidence in the ruler and prevailing order. As we argue above, local violence is integral in shaping people’s perceptions in post-war states, so in the second step of our analysis, we interact this measure for violence with our key independent variables of interest (more below). By treating the context of violence specifically in this manner, we move beyond viewing it as background noise or a context-level problem that should be controlled for in the analysis (Agnew, 1987).
We configure the violence measurement using several Geographic Information Systems (GIS) steps. Our violent events, coded per the protocol of the Armed Conflict Location and Event Data Project (Raleigh & Dowd, 2017), include the exact location and timing of a violent incident, as well as other qualities of the event, such as the actors involved and the type of event (whether it was carried out by military or non-military actors, and whether violence was directed against civilians or was a “battle” event between two armed actors). Here, we examine all types of events. The geographic coordinates of each event allow us to map them across the study areas. For Nagorno-Karabakh and Abkhazia, we have the exact sampling points used in the surveys (see Map 1); for the other two cases, we use the centroids of the small rayoni (counties) that constitute the primary sampling units. To aggregate the violence data to survey respondent locations in a GIS platform that merges both formats (survey and events), we measure the distance from each respondent’s location to the violent events in their de facto state.9 We define the temporal dimension so that all violence is included if it took place after the end of the de facto states’ respective major wars and before the date of survey enumeration. One benefit of our approach is that it does not rely exclusively on political boundaries of any administrative units to define which violent events affected survey respondents; violence proximate to a survey respondent may have an effect on his or her views even though the incident occurred just across a political boundary line. In our analysis, we capture a violent history by considering the raw count of all violent events within a 25 km threshold of each respondent since the end of the war in her/his de facto state of residence. We assume that it is the longer-term history of violence that shapes people’s confidence in the post-war order, and a 25 km threshold captures violence that can reasonably be assumed to be experienced by each respondent. There is some difficulty identifying distances that apply perfectly across different social settings, but this is a reasonable compromise. In a place like Transdniestria, 100 km is so big that it has little meaning and would encompass large parts of Moldova proper, but there are also conceptual lower limits. At one kilometer, for example, violent incidents are so rare that it would be almost meaningless for examining our expectations. In the appendix, we include a replication of our main findings at a five kilometer threshold, and the findings are consistent with what we find at a 25 km threshold.
To capture the conditional relationships hypothesized to be at work—how trust in the patron state is likely to condition the effects of people’s concerns about 1) new war and 2) lack of economic development—we introduce two interaction terms. We present the impact of the interactive hypotheses on conflict graphically, as predicted values with accompanying confidence intervals (Braumoeller, 2004, Clark et al., 2006, Franzese et al., 2001). For tables that contain the full set of variable estimates, see the online appendix. We first introduce the two-way interaction terms. Then, consistent with our argument about carefully examining individuals’ context of violence, we introduce figures that show how different degrees of background violence shape these relationships.
2.2.3. Control variables
We control for a range of alternative explanations. To assess people’s level of dependence on economic public goods provision, we include a survey question that asks the respondents to rate their family’s material situation. As for violence, although our expectation is that individuals’ perceptions of post-war violence will shape their confidence in the ruler and prevailing order, we control for perceptions of wartime violence as well. Those who suffered most during the war may be more fearful about its aftermath and possible recurrence and have a harder time believing in and trusting anyone, including those in power. To assess this possibility, we use a question that asks whether the respondents or their close relatives witnessed violence resulting in injury or death during the war.
Our main theoretical focus is on the state-building endeavors likely to be influenced by the patron states (economic development and prospects for war recurrence), though it may be that people’s concerns about state-building go beyond economic and physical safety. People’s worries about democratic development, as well as concerns for crime and corruption are likely to shape confidence in the rulers and prevailing order (Anderson and Guillory, 1997, Caspersen, 2008, Mishler and Rose, 2001, Seligson, 2002). Our analysis sample is reduced for every additional variable that we include (due to dropping “don’t know” responses). Therefore, we include analyses that control for these additional state-building concerns (concerns about crime and corruption, as well as perceptions of level of democracy) in the appendix.
Whereas our argument focuses on material aspects of state-building, we take into account that people’s identities may shape their confidence in the ruler and prevailing order. Indeed, both states and de facto states often aim to foster a collective identity. In some post-war and de facto states, authorities have highlighted the population’s shared war experiences and common enemy to create collective solidarity (Blakkisrud and Kolstø, 2011, Kolstø, 2006, Lynch, 2004, Organization for Economic Cooperation and Development (OECD), 2010). To examine whether such solidarity is associated with confidence in the authorities and prevailing order, we rely on a survey question that asks people how they “think now” of the war-time out-group (in the case of Abkhazia and South Ossetia, the Georgians; in the case of Transdniestria, the Moldovans; and in the case of Nagorno-Karabakh, the Azerbaijanis). If people feel negatively about the wartime out-group, they may be more likely to have confidence in the post-war order based on a clear common enemy image.
Due to the ethnocratic nature of the state-building efforts in most de facto states, we control for whether the respondents are a minority group in the de facto entity, i.e. not a member of the “titular” or dominant ethnic group. In the case of Nagorno-Karabakh and South Ossetia, this is not an issue, as all the respondents are of the titular group, but in Abkhazia and Transdniestria, members of the minority communities may have a different position in society than the titular groups. Thus, in Abkhazia, we consider whether respondents are non-members of the titular Abkhaz group, and in Transdniestria, we consider whether respondents are non-Russian. We do not interpret this variable in substantial terms; it is understood to be a control in the usual sense. Our main model also includes site fixed effects for each de facto state, as there is reason to believe that unobserved differences between the regions exist (including specific conditions at the time of the survey). We also show the results from running the models separately on the sample from each de facto state, which allows us to see if the main trends we observe are “carried” by any of the four entities.
We further control for general trust, gender, and year of birth. To the degree that political trust is associated with a predisposition to be trusting of people in general (cf. Eckstein, Fleron, Hoffman, & Reisinger, 1998), we include a question that asks respondents, “In general, would you say that most people can be trusted or you have to careful.” Experience of serving in the military is likely to shape male versus female responses, as can gender-based roles in these traditional societies. We anticipate that younger respondents, who have come of age and been socialized since the entities gained their de facto status around 1992, are more likely to have confidence in the ruler and prevailing order than older respondents, who were socialized in the Soviet context.
3. Empirical findings
We present our findings in figures that show predicted probabilities for the perceived credibility of the ruler and prevailing order, as assessed by trust in the president and confidence about the trajectory of the current state of affairs. Our discussion focuses on the conditional relationships between the internal and external aspects of state-building, but the full regression outputs (and robustness checks) are in the appendix. We have estimated the coefficients by using a generalized linear model (GLM) of binary outcomes with an interaction term for moderating variables and fixed effects for the four main de facto regions. Interaction terms in regression results are most effectively interpreted graphically, because varying effects can be observed across the range of combined independent variables’ values. In line with standard practice, we present predicted probabilities of observing the outcome across the range of constitutive terms (cf. Braumoeller, 2004, Clark et al., 2006, Franzese et al., 2001). First, in Fig. 1, Fig. 2, Fig. 3, we show how an external variable (patron distrust) modifies the link between internal variables (concerns about state-building) and confidence in the ruler and prevailing order. We first show the cross-case findings, then the findings for each de facto state. Second, we introduce figures demonstrating that key to understanding these dynamics is consideration for the geographical context of violence surrounding the people living in the region (how severe conflict has been at the local level). When interpreting our figures, note that the perceptual variables are coded so that a higher score (from 1 to 4) indicates that the respondent finds the question under consideration to be a “bigger problem.” The variable capturing “patron distrust” is dichotomous, with 1 indicating distrust in the patron’s leadership and 0 indicating that he or she trusts the patron.



Fig. 1 shows how respondents’ worries about the prospects of a new war with their parent state (Fig. 1A) and worries about economic development (Fig. 1B) shape their confidence in the prevailing order, conditional on whether they trust (solid line) or distrust (dotted line) the patron state. The shaded areas indicate confidence intervals (95 percent). Fig. 2 shows how the same concerns, about new war (Fig. 2A) and economic development (Fig. 2B), shape respondents’ confidence in the president, conditional on trusting the patron state.
As we would expect, the declining slopes of the lines indicate that growing worries among residents about the de facto state’s ability to provide for their security and material well-being is associated with less confidence in the ruler and prevailing order. The figures also demonstrate that patron state distrust is associated with lower confidence both in the president and prevailing order, compared with respondents who do trust the patron. Indeed, particularly when it comes to trust in the president (Fig. 2), the gap between the solid and dotted lines towards the right end of the x-axis indicates that patron trust might actually mitigate the negative impact of concerns about a new war or concerns about lack of economic development. Trust in the leadership of that “benevolent” patron—which serves as a security guarantor and provides financial assistance—helps boost credibility of the local ruler. Fig. 1, Fig. 2 also underscore that increased concerns about key state-building functions—providing for citizens’ physical and material security—diminishes the positive effect that patron trust, on its own, has on credibility of the ruler and prevailing order (note the negative slope of the solid lines). The findings confirm a complex interplay of individuals’ internal and external considerations in these settings of social and institutional formation. Worries about state-building diminish the credibility of the ruler and prevailing order, but particularly so when the external patron who backs the ruler is perceived as untrustworthy.
Next, we present the site-specific analyses, which suggest that our findings may be driven by the cases of Nagorno-Karabakh (Fig. 3A) and Abkhazia (Fig. 3B); overlapping confidence intervals makes it harder to distinguish between the patron effects in South Ossetia (Fig. 3C) and Transdniestria (Fig. 3D). Such variations among sites is not a surprise, and do not invalidate our general conclusions. First, the differences between sites are captured in our site-level fixed effects in the main models, which means that the estimates for the general relationships are not sullied by the differences. Second, we embrace the regional variation between de facto states as a meaningful representation of different institutional and historical conditions that can be the focus of future scholarly investigations.10
Building on the assumption that a context of violence is integral to people’s perceptions in post-war states, we now add an additional variable to our main analysis, the comparative findings presented in Fig. 1, Fig. 2. Our objective measure for violence at sub-national scales represents a third layer of complexity beyond the individual-level perceptions reported thus far, allowing us to examine how contexts of insecurity shape our findings.
We present the findings from our three-way interaction terms graphically. As in the previous figures, the solid and dotted lines are, respectively, trust and distrust in the patron, and the y-axis shows the probability of respondents trusting the president or having confidence in the prevailing order. In this extended analysis, we have changed the horizontal axis, which now shows the distribution of violent events that took place within the respondents’ locality (25 km) since the end of major fighting. To retain the dynamic effect of how varying concerns about fears of renewed war with the parent state and concerns about economic development shape people’s perceptions of the ruler and prevailing order, each figure contains four images. Each image illustrates the effects of violent events (horizontal axis) on the outcome (vertical axis) by level of patron (dis)trust (lines in each panel), and the row of images show how these associations vary across degrees of concern for new war (values of 1, 2, 3, and 4) or economic development (values of 1, 2, 3, and 4). The image on the far left shows what happens when respondents have a very low concern, i.e. a score of 1, for either economic development or new war, whereas the image on the far right shows what happens when they are highly concerned about either economic development or new war, i.e. a score of 4.
Echoing the conclusions from Fig. 1, Fig. 2, Fig. 4, Fig. 5 show that patron distrust is, in general, associated with lower levels of confidence (the dotted line of predicted probabilities is consistently below the solid line). The role of external forces is substantial and influential. Beyond that, our extended analysis shows that the severity of violence taking place at the local level is important for people’s confidence in the ruler and prevailing order in a nuanced and insightful manner. Whereas our findings above are based on estimates that control for the possibility that violence in an area could bias the estimated relationship, they tell us little about how and why violent settings affect responses. In the study of violent conflict, insights from political geography tells us that such dynamics should be modeled and interpreted explicitly, rather than just operating as controls (Linke & O’Loughlin, 2015).


We elaborate first on how confidence in the prevailing order is shaped by state-building concerns and patron (dis)trust across varying levels of local violence. Fig. 4A, in which the images from left to right show growing concerns about renewed war with the parent state, illustrates the probabilities of reporting high confidence in the prevailing order. It reveals something veiled in the results reported in Fig. 1A. For respondents who distrust the patron, their confidence is particularly low when they live in localities that have seen substantial amounts of violence during the post-war period. The steeply declining dotted lines suggest that people are less confident in places with more violent incidences. By contrast, this impact of violent contexts is mitigated by patron trust, as indicated by the solid lines. That is, the patron state’s role as a security guarantor mitigates also how internal violence shapes people’s confidence about the prevailing order. An unexpected result emerges in the upward slope of the line for patron trust when violence becomes higher in an area (compared to the downward slope of respondents who do not trust the patron). This demonstrates that patron trust can mitigate the effect of internal security concerns, but it is worthy of further investigation.
As expected (based on previous studies and our findings presented in Fig. 1A), respondents who trust the patron are also affected by violence occurring in their area at variable levels depending on their fear of renewed war with the parent state. Comparing the probabilities of the patron trust line in the far right hand panel, where fear of war is the highest (value of 4), to the far left hand panel showing the trend for low fear (value of 1), we see that the absolute values of the probabilities are smaller when concern for war is higher.
We see similar dynamics when looking at growing concerns about economic development (Fig. 4B). Consistent with the analysis so far, patron distrust is associated with lower confidence in the prevailing order. We also observe that as concerns for economic development grow, people’s confidence in the prevailing order is lower. Respondents who are the most likely to feel confident about the prevailing order are those who trust the patron and do not think that lack of economic development is a major problem (the far left panel in Fig. 4B). That is consistent with conclusions drawn from Fig. 1B. What these figures add to our understanding of how the state’s credibility is built and maintained is the importance of the relative level of violence in the local context.
Unsurprisingly, if people do not trust the patron, an increasingly violent context diminishes their confidence in the prevailing order. In the far left panel in Fig. 4B, we see that for people who are not particularly concerned about lack of economic development and who do trust the patron, a violent context is associated with a higher chance that they feel confident about the prevailing order. This finding seems counter-intuitive but, in this case, lack of concerns about the economy and trust in the patron trump the impact of the context of violence. In the far right panel of Fig. 4B, for people who are greatly concerned about lack of economic development, regardless of whether they trust the patron (but particularly when they distrust the patron), a more violent context is associated with less confidence in the prevailing order. Here, neither concerns related to internal public goods provision nor trust in the external patron can overcome the negative influence of living in a violent context.
In Fig. 5, we turn to how levels of local violence shape the effect that state-building concerns—about renewed war (Fig. 5A) and economic development (Fig. 5B)—and perceptions of the patron have on people’s trust in the president.
The findings in Fig. 5A are in line with our results in Fig. 2A, as the height of the estimated curves (their position on the vertical axis) for respondents who trust and also distrust the patron are lowest in the far right panel, where concerns about war are the greatest (value of 4). The figure reveals an unexpected dynamic, although it leads to a conclusion that adds precision. Fig. 5A shows that at lower levels of concern for a new war, an increasingly violent context is associated with increased trust in the president. In fact, only when respondents are greatly concerned about a new war, as indicated by the far right image, and distrust the patron do growing levels of violence diminish trust in the president. That is, when worries about external security, a violent local context, and patron distrust all combine, the prospects for trust in the president are very low.
Fig. 5B demonstrates how concerns for economic development are shaped by patron trust and severity of local violence and tells a largely similar story to our findings so far. Where the concern for economic development is highest, as indicated by the far right image, the overall level of trust in the president is lower (in the far right image, the trend for those both with and without trust in the patron are lower than in the far left image, where concerns about economic development are low). In a relationship similar to what we saw in Fig. 5A, there is evidence of a compound effect of patron distrust, high levels of violence, and, in this case, concern for economic development. Combined, these forces lead to the lowest levels of predicted trust in the president. From the solid line in the far right image, we also conclude that even when people are greatly concerned about lack of economic development, trust in the patron can mitigate the negative effects of living in an especially violent regional context.
4. Conclusion
The study demonstrates the nuances of complex and theoretically-informed expectations of how confidence in a ruler and ruling order—credibility—are generated and maintained, focusing empirically on de facto territorial states born from violent separatist conflicts. Although our empirical focus is on a particular type of post-war society and institutional entity, de facto states, we believe the findings relate more generally to challenges facing post-war scenarios. In former warzones that continue to experience periodic violence, the threats of geopolitical instability or a new conflict endure. There are good reasons to believe that inhabitants’ perceptions of both internal and external factors will shape their confidence of the (de facto) state’s governing authorities and the prevailing order. Our findings demonstrate that citizens’ worries about state-building and governing activities of the new territorial entity reduce a ruler’s credibility, but this is particularly the case when citizens do not have much trust in the external patron backing the ruler. Distrust of the external patron undermines the credibility of the local authorities in providing security and conditions of economic growth. That is, while post-war states may need the security and financial assistance offered by their patron states—and research has suggested that third parties can help overcome the credible commitment problem facing post-war states—our analysis shows that patron states do not have a uniform positive effect upon the local population’s perceptions of governance.
Rather than regarding the context of violence as simply a background curtain to the stage of political dynamics, we explicitly account for the contextual influences of continued violence in the second step of our analysis. By modeling the relationship between patron trust and confidence in the territorial ruling order as it varies by levels of local violence severity, we can estimate the relative importance of a specific and important contextual factor that shapes the rulers’ credibility. Thus, we can conclude that the people living in a poor economic environment express more confidence in their ruler and prevailing order even in a violent context because the ruler is tied closely to a patron that can provide security and economic aid. Such a conclusion indicates that citizens’ social setting has strong impacts on governance support at sub-national scales—and, indeed, parallels research emphasizing the role of contextual effects on, for example, voting (Agnew, 1987).
To the degree that confidence in the ruler and prevailing order is shaped by both internal and external dynamics, this has policy implications for the local authorities and their patrons. If post-war governments can provide public goods to their citizens and have a patron perceived as trustworthy, they have a recipe for fostering credible rule and stability within their (de facto) state. The role of the patron state, however, is a potential vulnerability. Geopolitical patrons like Russia will have their own interests and broader goals, and the de facto states in the former Soviet space are likely to continue to be places of contestation in a bigger geopolitical frame between Russia and “the West.” As we have shown in this study, the patron cannot take the views of the de facto states’ inhabitants for granted, but having their trust is in the interest of the domestic credibility of the de facto state’s government.
Conflict of interest
There is no conflict of interest.
Acknowledgements
This research is supported by a grant from the Human and Social Dynamics initiative, US National Science Foundation (grant number 0827016, John O’Loughlin, PI, Gerard Toal, Co-PI) and the UK Economic and Research Council (grant number ES/I035935/1, Kristin M. Bakke, PI). We thank Vladimir Kolossov, Russian Academy of Sciences in Moscow for his assistance and partnership in this project. We have received valuable feedback from several conference presentations, including the annual meetings of the European Political Science Association in 2014, the FLASCO-ISA Joint International Conference in 2014, and the British Conflict Research Society in 2014, as well as the OxPeace Conference in 2016. We also thank participants at seminar presentations at University College London and the University of Essex, as well as participants at the stakeholder conference on “Perceptions of Security and Governance in de facto and Partially Recognized States” at University College London in 2014 and at the workshop on “Russia and the de facto States in the Former Soviet Space” at the Norwegian University Center in St. Petersburg in 2017.