The Political Legacies of Rebel Rule: Evidence from a Natural Experiment in Côte d’Ivoire

Motivation and Theory

Does exposure to rebel rule affect civilians’ political attitudes—specifically, their attitudes concerning democratic citizenship—after formerly rebel-occupied communities are reintegrated into a unified postwar polity? If so, will the legacies of rebel rule foster increased citizen engagement in the formal political arena, or have the opposite effect? In line with previous work, we define rebel rule as a situation where a non-state armed organization exercises de facto control over the conduct of civilians and the provision of basic goods in a territory that is under the de jure authority of a recognized sovereign state (Arjona et al., 2015, 1–3).

A rich literature has examined the practices of social and political control that insurgents employ to govern civilians in rebel-ruled areas, and the varied motives behind these governance practices. While some rebel groups provide essentially no goods or security to civilians (Reno, 1999), most armed groups engage in governance in at least some domains of civilian life in an effort to create an alternative order and garner local support (Arjona, 2016; Huang, 2016). In cases where rebels take steps to preserve social order, they can engage civilians in different ways. Armed groups vary, for instance, in terms of whether they integrate non-combatants into wartime governance through local power-sharing arrangements, and whether they provide services to civilians, such as education, health care, dispute resolution, and policing (Huang, 2016; Mampilly, 2011; Mampilly & Stewart, 2021; Stewart, 2018). Variation in the scope and quality of rebel governance may be shaped by, among other factors, strategic dynamics (Kalyvas, 2006; Kasfir, 2005), ideology (Stewart, 2018), the strength of community organizations (Arjona, 2016; Péclard & Mechoulan, 2015; van Baalen, 2021), and rebel access to material resources (Huang, 2016; Weinstein, 2006).

Though this literature is impressive, evidence about the post-conflict legacies of rebel rule among civilians remains thin. Most research on the individual-level legacies of conflict focuses narrowly on the effects of victimization and exposure to violence, rather than exposure to political control by non-state armed groups. For instance, important research has studied how violence influences social capital and trust between citizens (Bauer et al., 2016; Collier et al., 2003), citizen attitudes toward the state (Bakke et al., 2014; De Juan & Pierskalla, 2016) and political participation after violent conflict (Blattman, 2009; Freitag et al., 2017). Perhaps counter-intuitively, much of this literature suggests that exposure to conflict can prompt cooperation and citizen engagement in political life after civil war (Bauer et al., 2016; Blattman, 2009; Freitag et al., 2017; Voors et al., 2012). By contrast, other studies find that exposure to wartime traumas tends to weaken trust and associational membership (Cassar et al., 2013; De Juan & Pierskalla, 2016; De Luca & Verpoorten, 2015). Yet the effects of nonviolent conflict processes related to rebel rule have received less attention.

Moreover, the literature that does exist on the political legacies of rebel rule has yielded mixed findings. Some studies suggest that interactions with non-state armed groups can boost postwar political engagement among civilians, a process we term the mobilization mechanism. Wood (2003) and Huang (2016), for example, argue that rebel-civilian interactions can increase the likelihood of active citizenship and postwar democratization. Whether instigated by rebel rulers—via rallies, persuasion, or other tactics—or by civilians who autonomously organize for self-protection, mobilization of civilians in rebel-ruled areas may create heightened political awareness and habits of political participation that persist into the post-conflict period and lead citizens to proactively engage with post-conflict state institutions. Viewed in this light, exposure to rebel rule may resemble other types of conflict-related experiences, such as witnessing violence or experiencing crime, that can cause individuals to become politically “activated” (Blattman, 2009).

Other work, however, suggests that exposure to rebel rule may have the opposite consequences for democratic citizenship. One such perspective arises from the literature on norms of citizen compliance. In this view, perceptions of the legitimacy and appropriateness of governing institutions—and corresponding duties to those institutions—emerge in part from a process of “mechanical ascription,” wherein long-term territorial control creates “robust informational monopolies that socialize populations accordingly” (Kalyvas, 2006, p. 125). Many citizens may cooperate with state authorities as a “natural” course of action, reinforcing local norms that state authorities deserve respect and that formal civic duties ought to be followed (Dreier & Lake, 2019). Following this understanding of citizen-state relations, exposure to rebel rule and its attendant rupture in state sovereignty may serve as a kind of exogenous shock for civilians that breaks long-held habits of compliance with the state and citizenship duties associated with formal government structures. This disrupted norms of compliance mechanism may be most likely in areas where insurgents govern arbitrarily or fail to recreate predictable state-like institutions.

Rebel rule can also create space for the construction of enduring forms of social solidarity and non-state governance that may later rival the post-conflict state, potentially further weakening democratic citizenship. For example, Kubota (2017) argues that governance by the Liberation Tigers of Tamil Eelan (LTTE) in Sri Lanka supplanted and co-opted state institutions, creating impediments to citizen trust and identification with the post-conflict government. Martin et al. (2021a) and Martin (2021) similarly found that many rebel-occupied communities in Côte d’Ivoire continued to rely on ex-rebel actors for informal services and protection, crowding out the role of redeployed police forces. Rebel rule can also bolster community-based institutions and patterns of local collective action in response to service shortfalls (Barter, 2016; Mampilly, 2011; Wood, 2008). In contrast to the mobilization mechanism, however, such practices of community solidarity may ironically reinforce opposition to and distrust of external governance authorities (cf Arjona, 2016) after war’s end if citizens evaluate postwar redeployed state institutions negatively relative to local self-help alternatives. If redeployed state authorities in former rebel-ruled areas are unable to convincingly (re)claim status as the primary guarantors of social order and essential goods, they may struggle to build popular legitimacy and restore the citizen-state social contract (Bakke et al., 2014; Bakke et al., 2018; Schmelzle & Stollenwerk, 2018, p. 464). We call this the rival governance mechanism (Table 1).

OPEN IN VIEWER

Table 1.

Rebel Rule and Postwar Democratic Citizenship: Theoretical Mechanisms.

Mechanism	Expected Direction of Effect
Mobilization	+
Disrupted norms of compliance	-
Rival governance	-

Overall, the existing literature remains ambiguous about which of these mechanisms is likely to be most salient, and therefore whether exposure to rebel rule is likely to strengthen or weaken post-conflict democratic citizenship in the aggregate. In part, this ambiguity can be traced to issues of methodology. Previous studies have rarely employed research designs that permit confident causal inferences about the postwar effects of rebel rule. While there are now a handful of studies that draw on fine-grained survey evidence (Bakke et al., 2014, 2018; Kubota, 2017, 2018), this work has not compared outcomes in areas ruled by armed movements to plausible counterfactuals (for example, otherwise similar areas that remain under sovereign government control). Moreover, because rebel rule is not randomly assigned across territory, it is difficult to distinguish between the effects of rebel rule and other confounding factors that may influence the choice of rebels to conquer and rule particular areas. For instance, if latent receptivity to rebels among the civilian population is also correlated with pro-civic attitudes—for whatever reason—then the estimated impact of rebel rule on postwar civic attitudes may appear positive even if the average causal effect is actually negative. Our study attempts to address this issue head-on. Crucially, and in contrast to past observational studies, we leverage a unique natural experiment that allows us to isolate the effects of rebel rule from other confounding factors.

A second possible reason for the ambiguity of the existing literature is that there is a good deal of variation in the nature of governance arrangements introduced by armed movements, as well as political dynamics across post-conflict environments. For instance, local exposure to rule by rebel groups that engage in extensive institution-building and ideological mass mobilization (as with some Marxist-Leninist insurgencies) may yield different patterns of citizen political engagement than exposure to ethno-nationalist rebellions or groups without clear political aims (Huang, 2016, pp. 90–102). ¹ Moreover, exposure to rebel rule in civil wars ending in decisive rebel victories may generate stronger attitudes of democratic citizenship than those ending in negotiated settlements or rebel defeat, if habits of participation in rebel governance structures are transferred to the post-conflict state. ²

These distinctions lead us to add important qualifiers to our study. First, as described below, our findings from Côte d’Ivoire likely pertain most closely to cases of “status quo” rebel rule (Mampilly & Stewart, 2021), where rebels provide core public goods but do not attempt to revolutionize existing social or political relations in society. Second, we study the legacies of rebel rule in a post-conflict context where the challenger armed group succeeded in deposing the incumbent regime and integrating into the state. Thus, if exposure to rebel rule in fact enhances democratic citizenship via the mobilization mechanism, we would expect to be able to detect such positive effects in our research setting.

Rebel Rule in the Ivorian Context

Rebel rule emerged in northern Côte d’Ivoire following the onset of civil war on September 19, 2002 when a group of Ivorian military officers, of mainly northern descent, attempted to overthrow the southern-based government of Laurent Gbagbo (Akindès, 2004). Although the mutineers failed to capture Abidjan, they quickly seized control in northern towns and cities under the aegis of the Patriotic Movement of Côte d’Ivoire (MPCI). Two additional rebel groups, the Ivorian Popular Movement for the Great West (MPIGO) and the Movement for Justice and Peace (MJP) emerged in the west of the country in November 2002. By January 2003, these rebel factions formally unified to become the New Forces (Forces Nouvelles (FN)). Côte d’Ivoire remained territorially partitioned until 2011, when FN military forces removed Gbagbo from power and installed the FN-supported political candidate, Alassane Ouattara. Insurgent forces integrated into the new national army and the FN administration in the north was dismantled as Ouattara redeployed state institutions throughout the country (Fofana, 2011). The Ivorian postwar government is widely seen as sympathetic to the former constituencies of the FN, and Ouattara’s Rally of Houphouëtists for Democracy and Peace (RHDP) party ³ enjoys strong electoral support in former rebel-controlled areas (Piccolino, 2018).

How did FN rebels rule civilians during the Ivorian conflict? Between 2002 and 2011, the FN occupation deprived millions of civilians from access to public services and caused significant social and economic disruption. While large-scale battles were rare, and the presence of international peacekeepers meant that populations in the rebel zone did not experience extensive state violence, local political order initially collapsed. State security personnel either joined the FN or fled, precipitating widespread looting and property damage. The Gbagbo government ordered civil servants to leave the north, and threatened to sack officials who did not comply (Chelpi-Den Hamer, 2007). Most government bureaucrats, teachers and medical professionals soon departed the rebel zone.

Initial efforts at governing FN-controlled territory involved the creation of 10 administrative zones. Each zone was managed by a military commander—popularly referred to as com’zones. These zone commanders used their wartime positions to gain wealth by taxing economic activity, controlling illicit resource extraction, and operating a ubiquitous network of rebel-operated checkpoints that created frustration among civilians as the conflict wore on (Bassett, 2011). Aware that complaints about living conditions in the north threatened the FN’s legitimacy, rebel leaders took steps beginning in 2003 to develop additional institutional structures to restore a sense of normalcy and order (Förster, 2015; Heitz, 2009; Popineau, 2017). The FN leadership created a civilian-staffed General Secretariat, headed by Guillaume Soro, which was responsible for social and political affairs. The Secretariat was represented by délégations générales in each of the 10 administrative zones (Fofana, 2012). To centralize control over rebel finances, the FN created a body called La Centrale Economique. These institutions helped the FN to project an image—however selective in presentation—of running a state-like administration complete with official stamps, transportation permits, and birth certificates.

Crucially, the FN often relied heavily on collaborations with existing community leaders and organizations in rebel-ruled territory, both to shore up their own legitimacy and ensure civilian access to basic goods and services (Fofana, 2012; Heitz, 2009). Health care, for example, was mainly provided by international humanitarian organizations, sometimes in cooperation with local NGOs or medical personnel who remained in the area. To meet demands for schooling and education, voluntary teachers’ initiatives emerged throughout the rebel-occupied zone (Chelpi-Den Hamer, 2007; Popineau, 2017). In the security domain, the FN usually retained the most important role, but commanders occasionally brokered arrangements with local dozo hunting associations, who acted as informal deputized police forces and enjoyed significant autonomy from the rebel command structure (Förster, 2015, pp. 213–216). Toward the terminal stages of the conflict, the FN also collaborated with government prefects who began to return to their offices in the north in 2009, though these officials typically lacked any real authority relative to the FN com’zones.

In many respects, then, the FN represents a typical case of contemporary rebel rule. As with many territorial non-state armed groups, the FN provided security and collected taxes, but did not intervene in all aspects of civilian life. This pattern is common: Huang (2016, p. 72), for instance, shows that rebels engaged in extensive institution-building in only 13% of armed conflicts occurring between 1950 and 2006. And similar to other rebellions in countries with a history of weak state penetration into society, where the prewar state bureaucracy coexisted with alternative governance providers, FN rebels had to negotiate their rule with a multiplicity of existing local actors, such as village chiefs, traditional hunters and religious leaders (Mampilly, 2011). Overall, while local variations in the quality of political order, service access, and civilian victimization existed (which we explore further below), rebel governance of civilians in FN territory from 2002 to 2011 is best conceptualized as what Mampilly and Stewart (2021, pp. 16–20) refer to as the “status quo” mode of rebel political control. Like other rebels in this category (such as the Eritrea Liberation Front or the Gerakan Aceh Merdekafree), the FN shared power with some local elites and co-opted existing institutions in order to administer civilians more efficiently, while refraining from extensive governance innovation or radical efforts to make political life more inclusive for civilians.

Research Design

To estimate the effects of local exposure to rebel rule on Ivorians’ political attitudes, we leverage the arbitrary placement of a ceasefire boundary that separated rebel- and government-controlled zones in the center-west and western regions of Côte d’Ivoire, running from Bouaké to Zouan-Hounien near the Liberian border. The northern side of this boundary was ruled by the FN, while the southern side remained under the control of the Gbagbo administration (Figure 1). The ceasefire line constituted a “solid” boundary on the rebel zone, beyond which FN forces did not operate. Our empirical strategy hinges on the fact that several regions in this zone—Gbêkê, Marahoué, Haut-Sassandra, Guémon, and Tonkpi—were partitioned into rebel- and government-controlled zones between 2002 and 2011 in a way that was arbitrary. The as-if random nature of the partition line location allows us to escape the selection bias problem caused by rebel groups choosing territory to occupy based on the attributes of local populations, and builds on the established practice of using arbitrary jurisdictional boundaries as the basis for causal inference (Dunning, 2008; McCauley & Posner, 2015).OPEN IN VIEWER

The key determinant of the ceasefire boundary location in our area of focus was the timing of French and rebel military mobilization in the early stages of the 2002 rebellion. Initially, the French government was reluctant to intervene in its former colony, but the mutineers’ rapid capture of northern territory shocked French leaders into action. Realizing that the rebels might overwhelm Gbagbo’s government and destabilize the country, France mobilized military forces stationed in Abidjan and Gabon at the end of September to prevent escalating violence (Charbonneau, 2012). The French interposition force—known as Operation Licorne—halted the southward advance of MPCI/FN units, resulting in clashes between French and rebel forces (Chelpi-Den Hamer, 2007). A small number of troops from the Economic Community of West African States (ECOWAS) also deployed at the end of 2002 to reinforce the peacekeeping presence along the frontline.

We focus on the western portion of the ceasefire line, and not the east, because the MPCI/FN’s military offensives in the fall of 2002 were concentrated on capturing this strategically important area (due to the production and shipping of cocoa), yet were only partially successful. East of Bouaké, rebels made few attempts to expand beyond their northern strongholds that had offered limited initial resistance, resulting in a territorial partition that was decidedly non-random. ⁴ In the center- and far west, by contrast, despite the conclusion of an initial ceasefire agreement between the rebels and the Ivorian government in October, the MPCI/FN made renewed offensives that prolonged fighting through the fall of 2002. The FN sought to move southward as far as San Pédro on the Atlantic coast in order to deny the Gbagbo government control over the cocoa-growing southwest, even though rebel leaders anticipated that southwestern communities were likely to offer resistance (Smith, 2002; International Crisis Group, 2003, pp. 20–21; phone interview with French military officer, 1 June 2010; interview with French military officer, 18 May 2010). In pushing further south, the rebels took advantage of the disorganization of the Ivorian national army, which was in chaos in the early phases of the war. What ultimately stopped the MPCI/FN offensive was the French intervention, which had reached a force of 3800 strong by December (Charbonneau, 2012, p. 513; Sciolino, 2002). After a second ceasefire in May 2003, the ligne de font stabilized permanently, cleaving through existing administrative boundaries and effectively partitioning in half the entire districts of Montagnes and Sassandra-Marahoué. Further details on the spatial-temporal expansion of the rebel frontline appear in Supplemental Material Appendix A.

In the end, the exact placement of the ceasefire line that spanned central and western Côte d’Ivoire was not a predetermined consequence of local demography, geography, or existing administrative boundaries—any of which could bias our study. The FN ended up in control of many ethnically inter-mixed localities where civilians expressed strong initial antipathy toward the rebel presence (Heitz, 2009). ⁵ If the timing or scale of the French military deployment had been different, or if the rebel offensive had progressed slower or faster—even by several days—the FN might have been stopped another 100 km to the north (and thereby failed to seize the cities of Man, Vavoua or Sakassou), or they might have advanced another 100 km south (and seized the cities of Daloa and Duekoué, for example). For many communities located close to the ceasefire line, therefore, “assignment” to exposure to the FN occupation essentially amounted to a coin-flip.

But how can we be confident that communities that were slightly farther north (and thus fell under rebel control) are not different in some way? To address this concern, we conduct a large number of placebo tests. Table 2 shows the estimates from ordinary least squares (OLS) regressions with standard errors clustered by enumeration area of several pre-treatment and fixed covariates on a binary indicator of placement in former FN-controlled territory. We find that for the placebo outcomes—including department-level education and poverty levels in 2002, individual ethnicity (Mande, Akan, or Krou), Christian, female, and years lived in the sub-prefecture—the effects are not statistically distinguishable from zero. In Supplemental Material Table A2, we also show pre-treatment balance across additional geographic and political variables including soil quality, cocoa production, annual precipitation, vote share for Ouattara’s RDR party in 2001 municipal elections, and pre-ceasefire violent events.

OPEN IN VIEWER

Table 2.

Placebo Tests.

	Dependent Variable
	Education	Poverty	Mande	Akan	Krou	Christian	Female	Years
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Treatment	−0.080	−0.086	0.122	−0.078	0.136	0.069	0.000	2.321
	(0.148)	(0.079)	(0.080)	(0.060)	(0.093)	(0.055)	(0.037)	(1.663)
Constant	0.864***	0.548***	0.264***	0.214***	0.200***	0.471***	0.500***	21.929***
	(0.128)	(0.066)	(0.032)	(0.041)	(0.043)	(0.037)	(0.025)	(0.878)
N	500	500	500	500	500	500	500	500

Importantly, the fact that there is no significant difference in the number of years that respondents lived in their enumeration sites in treated and control areas allays concerns about bias resulting from population displacement or migration across the ceasefire line. Population displacements occurred in both rebel- and government-controlled zones, with significant internally displaced people (IDP) settlement occurring in Guiglo, Yamoussoukro, and Abidjan. ⁶ The majority of IDPs, however, stayed with extended family and were likely to remain within their communities of origin, rather than relocate to other localities (Internal Displacement Monitoring Centre, 2012, p. 4). Moreover, we do not find clear evidence of “sorting effects” in our sample around the ceasefire line that would introduce bias. In Supplemental Material Table A3, we show that within the Krou, Mande, and Akan subsets in our sample, treatment status has no significant correlation with past migration, suggesting that cross-border migration did not “balance out” across groups. ⁷

We also anticipate the concern that FN rule in these “near-border” regions may have differed from the rest of the FN zone. Evidence from our respondent survey (Supplemental Material Table A5) indicates that Ivorians in near-border rebel areas were more likely to report being victims of violence, and were more likely to report that international peacekeeping forces visited their community between 2002 and 2011. This is unsurprising, given that the western zone—particularly Tonkpi/Montagnes—was a locus of partisan violence during the Ivorian conflict (Allouche & Bley, 2017; Balcells, 2017). However, we find that the prevalence of permanent FN camps and perceptions of the quality of wartime education and health services were quite similar in ceasefire line-adjacent communities compared to the rest of the FN zone. We also cross-check data from a separate survey of community elites in former FN localities (Supplemental Material Table A6). Our near-border areas of focus are not anomalous in terms of whether FN rebels recruited locally, whether FN commanders held public meetings in the community, whether community elites actively supported the FN, the degree of FN involvement in service provision, or the prevalence of rebel taxation.

Survey

In order to assess the political impacts of FN rule, we partnered with the United States Agency for International Development (USAID) to implement a survey in conflict-affected areas of western, northern, and central Côte d’Ivoire in August–September 2018. ⁸ The sampling frame consisted of all sub-prefectures that fell on the northern side of the ceasefire line (i.e., under FN control from 2002 to 2011), as well as sub-prefectures from an additional five regions near the ceasefire border that remained fully or partially under government control. The survey reached a total of 1200 respondents, with 300 coming from the government-controlled zone. Consistent with the standard Afrobarometer sampling methodology in Côte d’Ivoire, respondents were polled proportionally to the population size of the sub-prefecture or commune. All interviews were conducted face-to-face with tablets, with specific households selected via a grid-based random-walk method. For the main analysis below, we focus on the subset of 500 respondents in treated and control enumeration areas located close to the ceasefire border (less than 100 km).

Outcome Measures

To tap into the concept of democratic citizenship, we assess three sets of indicators. The first set of outcomes measure respondents’ confidence in the local institutions of the Ivorian state. We asked respondents whether they believed that the following institutions play a positive role in their community (1 = yes, 0 = no): the prefectoral corps, the police and gendarmerie, and the local municipal government. Second, we measured respondents’ attitudes about civic obligations and duties vis-a-vis the national political community. We asked respondents if they agree that paying taxes is something that a good citizen should always do, and if they agree that citizens have a duty to vote in elections (1 = yes, 0 = no). We also asked if respondents agree that citizens should organize with other community members to demand improved government service, contact the media to call attention to poor services, or contact a government official to complain. Finally, we measured attitudes about several illegal citizen protest actions, including whether respondents agree that it is acceptable for citizens to refuse to pay taxes, establish a road block, or occupy a public building. All of our questions build on standard polling instruments in the region, such as the Afrobarometer, and are adapted to the Ivorian case. We analyze indicators individually and construct a number of summary indices.

Treatment

To measure exposure to rebel rule, we use a binary indicator of whether the enumeration area was located on the northern side of the ceasefire line. Since the ligne de front was fixed and internationally monitored, populations south of this boundary should not have been exposed to FN control. To geolocate the ceasefire boundary, we relied on maps provided by the National Institute of Statistics in Abidjan.

Modeling Choices

To estimate the effects of the FN occupation, we use ordinary least squares (OLS) models that regress each outcome on our binary treatment indicator. ⁹ Below, we show a simple comparison between respondents in treatment and control areas. In the Online Appendix, we show models controlling for many additional covariates. We cluster standard errors by enumeration area.

Robustness

We conduct a series of tests to address lingering concerns about unobserved differences between areas that fell on the FN side of the ceasefire line versus areas that remained under government control, which could introduce bias. First, we look for potential prewar differences in local social capital, which could affect both the quality of wartime governance as well as current trust in institutions. To construct proxy measures of prewar social capital, we use fine-grained data from the 2002 household quality-of-life (ENV) survey, aggregated at the department level. Specifically, we measure the proportion of households that reported receiving financial support from a community organization in the last 12 months, the proportion of households that received a loan from a neighbor or community group to cope with financial distress, and the proportion of households that were willing to accept an adopted non-familial child in crisis. None of these measures of prewar social capital differ significantly across treatment and control areas (see Supplemental Material Table A8). Second, we constrain the sample to enumeration areas located within varied distances of the ceasefire line. If there are unobserved differences in our sample correlated with geographic location, the estimated effects of the treatment should vary between respondents in areas that are immediately adjacent to the ceasefire line and those at a near distance. Supplemental Material Tables A9–A11 show that the estimated effects of rebel rule are quite consistent when constraining the sample in 20 km increments. Third, we ran our analysis using “placebo borders” at varied intervals moving north from the actual ceasefire line location. Supplemental Material Table A12 shows that rebel-government borders at different locations do not return significant results. Fourth, Supplemental Material Table A13 shows that respondents’ political attitudes in both the rebel and government zones are largely unrelated to the distance to the ceasefire line. Finally, we estimate models with region-level fixed effects (Supplemental Material Table A14) to demonstrate that our results are not driven by specific survey areas.

Another potential concern is that our study faces a “cluster randomization” problem (McCauley & Posner, 2015) in the sense that Ivorians in central and western Côte d’Ivoire were not exposed individually to rebel rule, but rather they were exposed as a cluster along with their neighbors. If people who live near each other react in similar ways to the rebel occupation, but differently from people who live farther away, then potential outcomes will be correlated across clusters and the effective sample size of our study is reduced. To address this concern, we re-estimate our main regression models using enumeration area-level averages. Supplemental Material Table A15 shows that even using this more conservative approach (which reduces the sample size to 50), we find very similar estimates.

Given the time lag between the end of the conflict and the implementation of the survey (2011–2018), it is possible that intervening variation in the quality of postwar governance or state-provided patronage among the treated and control groups account for our results. However, in our research context, this type of bias is likely to cut against the direction of our findings. Given that the Ivorian conflict resulted in a victory for the FN-supported candidate, former rebel territories tended to be the primary beneficiaries of development resources and state patronage channeled by the Ouattara government (Piccolino, 2018, pp. 493–497). Analyses of data from the 2014 ENV survey confirms that households in the former Cenre-Nord-Ouest (CNO) zone reported greater access to loans, financial support, and aid from both government and international agencies. ¹⁰ Thus, it is unlikely that our findings are driven solely by lower-quality post-conflict governance in former rebel areas.

Finally, we consider whether our results can be explained in terms of differential exposure to violence or exposure to third-party (e.g., non-rebel) governance actors. We find that exposure to FN rule does significantly increase the risk of victimization—in government-controlled areas, 35% of respondents reported that they were victims of physical violence or property theft during the conflict; among respondents in rebel-ruled communities, the rate is 55% (p < 0.001). However, when we repeat the main analysis among victim and nonvictim subsets of respondents (see Appendix Supplemental Material Figure A3), we find broadly similar effect estimates among both groups. Moreover, when we add controls for whether respondents reported experiencing visits by UN peacekeeping forces in their community, or received assistance from a national or international NGO during the war, the estimated effects of rebel rule remain unchanged (Supplemental Material Table A16). Overall, these tests suggest that something about the experience of rebel rule itself is behind citizens’ changed political attitudes.

Exploring Causal Mechanisms

The analyses above suggest that rebel rule in Côte d’Ivoire had primarily negative effects for democratic citizenship. To explain this finding, we draw on both survey and qualitative evidence, including interviews conducted during two research trips in 2017 and 2018. Interviewees included a range of Ivorian community members including village leaders, civil society activists, public servants, elected officials, and ordinary residents in over 30 former rebel-ruled localities. ¹¹ We conducted this research prior to the implementation of the survey, but revisited our interview notes to help make sense of the survey findings.

One seemingly obvious explanation is that our findings capture a partisan galvanization effect, wherein exposure to “northern” rebel rule intensified mistrust in the state and skepticism about formal civic duties among Ivorians already inclined to view the postwar Ouattara administration unfavorably. Specifically, in the context of Côte d’Ivoire’s ethnically divided society (Akindès, 2004), we might expect that coethnics of FN rebels and the Ouattara government would respond differently to exposure to FN rule compared to non-coethnics, with increased antipathy to the postwar state concentrated among non-coethnics. We evaluate this idea by repeating our analyses among subsets of respondents who self-identified as belonging to Krou, Akan, and Mande ethnic groups. The Krou are generally seen as aligned with Gbagbo’s Front populaire ivoirien (FPI) party, while the Mande, and to a lesser extent the Akan, were seen as more supportive of the FN and President Ouattara. If FN rule were galvanizing anti-civic attitudes among those already inclined to distrust the post-conflict Ouattara regime, we would expect to see different outcomes across these sub-groups. Figure 3 shows that there are no such differences: across each set of outcomes, we see generally similar estimates among all three ethnic subsets.OPEN IN VIEWER

We point to three alternative mechanisms. Two of these mechanisms—disrupted norms of compliance and rival governance—are derived from claims in the existing literature. We also propose a third mechanism of unmet expectations, based on the authors’ inductive observations. While we cannot provide definitive tests of these mechanisms, we offer suggestive evidence for each on the basis of both qualitative and quantitative findings.

First, the experience of rebel rule and its attendant rupture in state sovereignty appears to have weakened local norms of compliance with state-like institutions. In interviews with redeployed state officials in former rebel areas, officials repeatedly complained that local populations were not subjected to the authority of the state for many years, and that this affected their ability to work with the community after reunification. One officer in the gendarmerie, who was redeployed in 2011 to the region of Gbêkê, recalled that when he came back to the region people were “unfamiliar” with the gendarmerie, ignored its authority, and were reluctant to follow official laws (Author interview, 17 July 2017). A sous-préfet in the same region drew a similar picture and insisted that the FN occupation was responsible for what he viewed as a lack of respect for the rule of law, including recent violent protests against the cost of electricity that degenerated into the pillage of a bank and the sous-préfecture building (Author interview, 19 July 2017). Interviews with local residents and civil society activists echoed similar themes. Many expressed the view that their communities had been “abandoned” by state authorities during the crisis, which created a reticence to place trust in formal institutions again.

Do the findings from our survey offer quantitative support for this mechanism? To probe this, we repeat our analyses among subsets of respondents who lived in areas where the FN established permanent military camps, versus respondents in areas where rebels only visited periodically. As Förster (2015) documents, areas with permanent FN camps tended to experience more continuity in terms of the presence of state-like authority—where local FN commanders often adopted the titles, protocols, and even uniforms of Ivorian state officials—compared to areas without permanent camps, where rebel rule was more informal and sporadic. Thus, comparing areas with and without permanent rebel camps allows us to indirectly assess the disrupted norms of compliance mechanism. Figure 4 shows that while the effects of rebel rule on favorable attitudes toward local state institutions and anti-state actions are not appreciably different across these areas, areas without permanent camps did see greater decreases in attitudes about civic obligations (a two sample t-test confirms the difference in estimates is significant at the p < 0.05 level).OPEN IN VIEWER

The rival governance mechanism was also supported by our interviews. In particular, rebel rule in Côte d’Ivoire often triggered community-level collective action, leading some Ivorians to view the post-conflict state as a largely irrelevant governance actor in their own lives. We heard numerous stories of ordinary citizens stepping up to fill service gaps created by the withdrawal of the state. During interviews in the city of Man, for example, interviewees described how university students, retired teachers and other educated individuals joined the ranks of volunteer teaching associations that allowed schools to re-open in many rebel-ruled towns. A member of the volunteer education association École pour Tous in Bouaké similarly recalled: “the initiative was really a community initiative involving every neighbourhood…” (Author interview, Bouaké, July 2017). In the FN-occupied town of Ouangolodougo (popularly referred to as “Ouangolo”), traders organized the Confédération des Fédérations des Filières Bétail Viandes des Pays Membres de l’Union Economique et Monétaire Ouest Africaine (COFENABVI-PAM-UEMOA) to facilitate the exporting of cattle amid the administrative division of the country (Speight, 2015, pp. 193–195). In other cases, civilians self-organized in order to mediate with FN rebels because existing local authorities had fled. In the town of Sakassou, for example, one member of a local women’s association explained that she felt obliged to take on new roles during the war, including negotiating with the local FN commander to resolve disputes and obtain essential resources for the community (Author interview, 23 July 2017, Sakassou). And in rare cases, local collective action was spurred by direct confrontations between Ivorian civilians and FN rebels. In the former rebel capital of Bouaké, residents organized public opposition to rebel taxation (Konan, 2007). Further north in Bounkani, tense relations between the ethnic Lobi community and FN rebels led local civilians to organize armed self-defense groups (Speight, 2015, pp. 114–115).

Many of these community-level governance mechanisms persisted after national re-unification in 2011, and provided a ready alternative to the redeployed state for addressing local issues. The grassroots education initiative École pour Tous, for example, continued to be active in northern Côte d’Ivoire, organizing efforts to improve education access (Popineau, 2017). Traders associated with COFENABVI-PAM-UEMOA resisted efforts of the municipal government in Ouangolo to reassert control over local taxation, and attempted to retain their wartime fiscal authority (Speight, 2015, pp. 200–203). In the security domain, although the state police and gendarmerie forces were redeployed across the north, many local dozo groups retained their informal policing roles in rebel-ruled territory, particularly concerning issues of cattle theft (Heitz-Tokpa, 2019, p. 156).

To probe the rival governance mechanism quantitatively, we analyze behavioral indices of respondents’ participation in both wartime and post-conflict local collective action. To construct the indices, we asked whether, during the rebel occupation and within the last 6 months, respondents had done the following actions: volunteered in a project to help their community, participated in a meeting to discuss problems in the community, joined a civil society organization, served in the governance system of their village or neighborhood, joined a local party branch, or attended a meeting or demonstration organized by a local party branch. We find that exposure to rebel rule is associated with increases in both wartime and current local collective action behaviors (p < 0.05). Moreover, the index of wartime collective action is positively correlated with both recent collective action and negatively correlated with attitudes about civic obligations (full results in Table A17). While not definitive, these results are consistent with the notion that rebel-ruled civilians developed habits of self-help and community solidarity that both outlasted the war and alienated citizens from a broader sense of civic duty toward formal political institutions.

As another piece of suggestive evidence for the disrupted norms and rival governance mechanisms, we momentarily leave behind our near-border identification strategy and zoom out to all FN-ruled localities covered by our survey (900 respondents across 90 enumeration areas). We ask: conditional on FN control of a locality, does the quality of governance under rebel rule covary with postwar citizen-state relations? Full results are reported in Supplemental Material Table A18, which show OLS models that regress our main attitudinal outcome measures on a large number of department, locality, and individual covariates. We find that, within the larger FN zone, respondents who judged that the availability and quality of basic services (specifically policing, health, and education) under FN rule were either “good” or “very good,” postwar attitudes about state institutions and civic obligations were also significantly more positive (p < 0.001) (though we see no relation between the quality of wartime services and postwar attitudes about anti-state actions). While we cannot interpret these patterns causally given that wartime governance may reflect unobserved traits of these communities (though we control for ethnicity, prewar education, poverty, and social capital), the pattern is nonetheless consistent with both mechanisms noted above: where rebel rule involved a more significant deterioration in the quality and reliability of state-like governance, local populations experienced a greater downward shift in the perceived efficacy of formal state institutions, and attendant obligations toward the broader civic community.

A final mechanism that emerged from our interviews revolves around a dynamic of unmet expectations. By this logic, citizen attitudes toward the postwar Ivorian state in rebel-ruled areas are shaped by resentments caused by the discrepancy between high expectations of a postwar “payoff” and the reality of slow postwar recovery and reconstruction. In many interviews, residents expressed the view that their postwar fates stood in marked contrast with the fortunes of the rebel leaders who once dominated their towns and cities. In the Worodougou area, for example, a poor region of the country where the population strongly supported the political goals of the rebellion and many people joined the FN, a sense of abandonment and deception was prevalent in 2017. Former combatants, residents, and officials alike complained that the leaders of the rebellion exploited the mineral wealth of the area during the division of the country, and then neglected it after taking power in Abidjan (see also Van Bockstael, 2019, p. 911; Grajales, 2020). Examining respondent perceptions about local economic conditions in the near-border sample, only 31% of respondents reported that they felt that local economic conditions had improved in their community since 2011—a notably low rate, given the country’s widely touted post-conflict economic recovery—and only 35% felt that government rehabilitation projects had benefited their locality. While these views are similar in former rebel and government-controlled areas (p > 0.1), perceptions of economic stagnation may have translated into greater frustration in the former rebel zone due to a sense of entitlement to a reward for having sustained the burden of the occupation.