What are the effects of international intervention on the rule of law after civil war? What role can UN peacekeeping in particular play in promoting rule of law in the world's weakest and most war-torn states? As peacekeeping operations have grown in scope and complexity, establishing the rule of law has become increasingly central to their mandates, and is now widely considered a prerequisite for sustained peace, good governance, and economic growth.Footnote 1
Yet with few exceptions,Footnote 2 most peacekeeping research continues to focus on a narrow range of outcomes, most of them related to violence: violence against civilians,Footnote 3 battles between governments and rebel groups,Footnote 4 and, of course, the recurrence of civil war.Footnote 5 This is especially true of cross-national research. These studies are important, but capture little of the variation in what peacekeepers actually do on the ground, or in the “multidimensional” goals they are mandated to achieve. In peacekeeping, as in war making, “a good deal is played out in daily contacts, often at a highly individual level.”Footnote 6 These dynamics are crucial to understanding the effects of international intervention on citizens and host states.
My analysis complements a small but growing literature on the impact of peacekeeping at the micro level.Footnote 7 Rule of law is a multidimensional problem, requiring not only that state authorities abide by legal limits on their power, but also that civilians rely on state institutions to adjudicate disputes and provide security when crimes are committed or violence occurs.Footnote 8 In postconflict settings, however, citizens often perceive formal (state) institutions as corrupt, inept, and inaccessible, and so opt to rely on informal (nonstate) alternatives instead. These alternatives can be quick and effective, but also suffer prejudices and inefficiencies of their own.Footnote 9 Some also rely on extrajudicial mechanisms of dispute resolution that undermine the rule of law.
I propose a theory to explain how international interveners promote the rule of law at the micro level, then test implications of the theory using an original survey and list experiment in Liberia. I focus on the “sociological” dimension of rule of law, namely, “to whom people turn for solutions to problems that would normally be considered legal.”Footnote 10 I show that exposure to the UN Mission in Liberia (UNMIL) increased Liberians’ reliance on state over nonstate authorities to resolve the most serious incidents of crime and violence, and increased nonstate authorities’ reliance on legal over illegal mechanisms of dispute resolution. I use multiple identification strategies to support a causal interpretation of these results, including an instrumental variables strategy that leverages plausibly exogenous variation in the distribution of UNMIL personnel induced by the killing of seven peacekeepers in neighboring Côte d'Ivoire.
I also find that UNMIL did not mitigate and may in fact have exacerbated citizens’ perceptions of state corruption and bias. However, while UNMIL's beneficial effects are still detectable two years later, even in communities that reported no further exposure to UNMIL personnel, its adverse effects on perceptions of state corruption dissipate over time. To my knowledge, this is the first study to explore decay in peacekeeping's micro-level effects. While I cannot say whether these effects will persist even after UNMIL's withdrawal, taken together, my results suggest the possibility of a durable and overall positive change in citizens’ attitudes toward state authority.
My study contributes to multiple literatures in international relations and comparative politics. Most directly, it contributes to research on international intervention and the consolidation and legitimation of state authority after civil war. Numerous scholars have argued that successful international intervention requires repairing the broken social contract between citizens and war-wracked statesFootnote 11—a task that some believe is impossible.Footnote 12 Yet surprisingly few studies include actual empirical data on citizens’ relationships with postconflict states, and fewer still address the thorny counterfactual question of how those relationships might have developed in the absence of third-party intervention.
Even studies that address these issues typically focus on state building in the capital city, “leaving more complex questions of extending state authority to the hinterlands behind.”Footnote 13 But demand for state authority does not simply “trickle down” from renewed supply: there is no a priori reason to believe citizens will automatically accept and comply with newly reconstituted state institutions, especially where nonstate alternatives exist.Footnote 14 Tensions between state and nonstate authorities often spark violence in postconflict countries,Footnote 15 and have stymied reconstruction in a variety of settings, including Iraq, Afghanistan, East Timor, Mozambique, Côte d'Ivoire, Sierra Leone, Guatemala, and beyond.Footnote 16 My study provides some of the first empirical evidence on the impact of peacekeeping on these vital but complex relationships.
My study also extends research on the difficulties of projecting state power into areas with “incomplete domestic sovereignty,”Footnote 17 especially where state and nonstate authorities compete for citizens’ loyalties;Footnote 18 on the possibility of legitimizing states through third-party provision of security and other public goods, for example, in the context of counterinsurgencyFootnote 19 or foreign aid;Footnote 20 and on the diffusion of international legal and human rights norms,Footnote 21 especially at the local level, and especially where international norms collide with local rules and customs.Footnote 22 Indeed, “sometimes the primary roadblock to norm change is the people within the state rather than the state itself.”Footnote 23 I show that international interveners can help overcome this roadblock, extending state authority and instilling respect for internationally recognized rule of law principles, even in one of the world's weakest and most war-ravaged states.
Theory
Rule of Law after Civil War
Establishing the rule of law after civil war is a multidimensional problem. On the one hand, the human and infrastructural capacity of the state must be restored: judges and police officers must be recruited and trained; laws must be amended; constitutions must be revised and ratified. But as important as these processes are, they do not address the “heart of the problem,” which is “to whom people turn for solutions to problems that would normally be considered legal.”Footnote 24 Rule of law is more than just a set of institutions; it is also a “normative system that resides in the minds of … citizens,”Footnote 25 and that cannot be manufactured by the “simple expedient of creating formal structures and rewriting constitutions and statutes.”Footnote 26 As Chesterman asks, “what would a woman do if her property were stolen—go to the police? Or what would a man do if his brother were murdered?”Footnote 27 Rule of law hinges crucially on the answers to these questions.
In many developing countries, security and justice are provided not by a unitary Weberian state but by a multitude of formal (state) and informal (nonstate) actors whose jurisdictions shift and evolve over time.Footnote 28 Examples of these informal actors abound: chiefs in Zambia,Footnote 29 warlords in Afghanistan,Footnote 30 and dozos in Côte d'Ivoire.Footnote 31 These are neither states nor citizens nor civil society organizations per se. Rather, they occupy the “pivotal ‘meso-level’ analytical space between the local and the national.”Footnote 32
The relationships that develop in this space are complex and context-dependent—sometimes cooperative, sometimes conflictive, often overlapping.Footnote 33 Formal and informal authorities may be equally essential for the provision of security and other public goods,Footnote 34 and organizations like the UN increasingly attempt to strengthen both simultaneously.Footnote 35 Yet “legal hybridity”Footnote 36 of this sort also poses practical and conceptual problems for international intervention. How to guarantee the UN's principle of “supremacy of law” where jurisdictional boundaries are blurred? How to ensure “avoidance of arbitrariness” and “procedural and legal transparency” across amorphous and ever-changing constellations of security and justice providers?Footnote 37 How to determine whether rule of law has been established when the state is just one of many potential venues for resolving disputes?
In posing an answer to these questions, I focus on three problems in particular that tend to afflict hybrid legal systems. First, most hybrid legal systems stipulate a constitutional division of labor between state and nonstate security and justice providers. In many countries, however, citizens routinely rely on informal institutions to adjudicate crimes over which the state claims both original and ultimate jurisdiction—murder or rape, for example. Whatever the merits of informal dispute resolution, blurred jurisdictional boundaries undermine basic tenets of the rule of law, and foment confusion among disputants who “cannot be sure in advance which legal regime will be applied to their situation,” sometimes with “serious social and political ramifications.”Footnote 38
Second, even when informal authorities operate within their legally circumscribed jurisdictions, some nonetheless rely on mechanisms of adjudication that violate key rule-of-law principles. Most hybrid legal systems authorize informal authorities to use adjudicative procedures not typically practiced in the courts, such as mediation. But most also prohibit some procedures as unacceptable—lynchings, for example, or trials by ordeal. No matter how effective these procedures may be, they preclude due process and flout basic human rights protections.Footnote 39 Rule of law requires that they be supplanted by legally sanctioned alternatives.
Finally and relatedly, citizens of many postconflict countries perceive state security and justice institutions as corrupt and biased, and fear that wartime state predation will persist into peacetime. This helps explain why many opt to rely on informal institutions instead. While some may come to rely on state institutions they perceive as unreliable,Footnote 40 distrust of the state often proves an impediment to citizen cooperation,Footnote 41 and thus to the rule of law.
The Role of International Intervention
How might international interveners like the UN help mitigate these obstacles? My core theoretical claim is that UN missions promote the rule of law by acting as surrogates for weak and war-torn states—proxies for centralized power in the absence of a powerful center. Especially in failed or collapsed states, exposure to the UN often constitutes civilians’ first re-encounter with centralized authority in the wake of conflict. UN missions assume responsibility for functions typically associated with state institutions, then perform those responsibilities at the local level. They literally “act out” the state.
In so doing, UN personnel help demonstrate, by example, the relative merits of formal over informal dispute resolution; sensitize civilians to increased third-party presence in and around their communities; create opportunities for state officials to “claim credit” for the benefits UN missions provide; and promote norms that legitimize the role of national over local institutions as purveyors of security and justice. Cross-nationally, UN missions transmit “internationally-approved norms of domestic governance” to host countries—a “globalization of the very idea of what a state should look like and how it should act.”Footnote 42 I argue that this same process can occur not just between countries, but within them as well. By engaging with civilians on a face-to-face, day-to-day basis, international interveners can instill loyalty to the “idea of the state,”Footnote 43 even where the reality remains absent and dysfunctional.
While peacekeepersFootnote 44 have developed a repertoire of strategies for engaging with civilians at the micro level, most of their interactions occur in the context of one of three activities: patrols, public works projects, and interventions to resolve impending or ongoing disputes. These activities mimic functions of the state as a provider of security and other public goods. While “seemingly routine,” they can nonetheless have a “profound effect” on civilians’ perceptions of the mission, the government, and the peace process.Footnote 45 In remote hinterland communities in particular, where few NGOs operate, the arrival of uniformed peacekeeping personnel is often an important event in and of itself—a fact that is often overlooked in studies focused on state building in the capital city.
Patrols help peacekeepers establish a physical presence throughout the host country, even where roads are rough and infrastructure dilapidated. But beyond merely providing security, patrols are a mechanism for consolidating and legitimating host state authority. This is explicit in the UN's own policies. Patrols “[create] an environment for the host state to begin reasserting its sovereign authority and re-establishing ties to local communities,” and “[extend] the geographical reach of host state police services.”Footnote 46 “Joint” patrols also create opportunities for citizens to interact with state security forces in a controlled, monitored environment, with the short-term aim of increasing citizens’ access, and the longer-term goal of depolarizing state/society relations.
Beyond patrolling, UN missions routinely intervene to prevent the escalation of small-scale conflicts into large-scale crises, and to ensure that criminal offenses are adjudicated in accordance with state laws.Footnote 47 Domestic police forces often participate in these operations as well, creating additional opportunities for contact with civilians. Interventions help enforce a legally prescribed division of labor between formal and informal institutions, but they also and more profoundly help peacekeepers demonstrate the virtues of formal over informal dispute resolution. Again, this is explicit in UN policies, especially around UN policing. Through their “independence, impartiality, commitment to UN values, and compliance with international human rights,” UN police are believed to “create strong positive expectations of host state police, foster popular confidence in the police, and engender legitimacy in the eyes of local populations.”Footnote 48 In this way, international interveners aim to legitimize their domestic counterparts through their own actions at the micro level.
Finally, many UN missions contribute to the provision of public goods—paving roads, building bridges, and repairing courts and police stations. These are low-cost, small-scale projects designed to demonstrate “early peace dividends,” open channels of communication with civilians, and—crucially for my theory—build the “legitimacy and capacity of local authorities.”Footnote 49 Public works also create opportunities for host states to claim credit for the UN's accomplishments. Citizens struggle to appropriately attribute credit and blame for public goods provision even in information-rich environments; in information-poor environments (like most postconflict settings), credit claiming is likely to have even starker effects.Footnote 50 Claiming credit may be especially likely in the context of UN missions, which, unlike NGOs, are expected—indeed, mandated—to improve citizens’ relationships with host governments. As a UN police officer explained to me, “all of our contributions are to be totally attributed to the [domestic police force]. Everything we do for them, they are supposed to get the credit.”Footnote 51
My theory resonates with classic constructivist accounts of the power of international organizations,Footnote 52 and with recent research on police/community relations, counterinsurgency, and foreign aid. Criminologists have found that mutually respectful contact can “depolarize” relations between civilians and police officers, and that positive interactions with individual officers can mitigate distrust of the police force as a whole.Footnote 53 Through patrols and other initiatives, I argue that UN personnel create opportunities for this sort of contact to occur, while also generating support for centralized authority. Studies of counterinsurgency further suggest that third-party public goods provision can improve perceptions not just of the third party, but also of the host government.Footnote 54 UN patrols, interventions, and public works can produce just such an effect. And scholars of foreign aid have shown that citizens often credit government officials for the benefits foreign donors provide, even when they know those benefits were paid for with foreign funds.Footnote 55 UN-provided public goods are especially ripe for credit claiming of this sort.
Importantly, my theory predicts that UN missions should have more durable effects than can be achieved through direct deterrence alone. I argue that by patrolling, building public works, and intervening to resolve disputes, UN missions change citizens’ beliefs about the role that national (rather than local) institutions can and should play in maintaining social order. But these same actions may also prevent crimes from being informally or illegally adjudicated, without changing underlying attitudes in any lasting way. In other words, in addition to deterring crimes, the UN may deter crimes that do occur from being adjudicated outside the bounds of the law—for example, by referring cases to the police.
While these two mechanisms are not mutually exclusive, a theory focused on direct deterrence alone would generate starkly different empirical predictions. My theory involves a change in citizens’ behaviors and beliefs, and so predicts that peacekeeping's effects should persist over time, even after peacekeepers depart. A theory premised on direct deterrence involves no such change, and so predicts that the effects of peacekeeping should dissipate quickly, especially after peacekeepers are no longer physically present. While I cannot definitively adjudicate between these two mechanisms, my results suggest a more lasting change than we would expect from direct deterrence alone.
Rival Hypotheses
While my theory is optimistic about the role international intervention can play in overcoming obstacles to the rule of law, skeptics might find four sets of countervailing concerns. First, while my theory assumes that UN missions are generally benevolent, peacekeepers are no strangers to scandal, and there have been many highly publicized cases of UN personnel engaging in unprofessional or even illegal behavior.Footnote 56 If peacekeepers are viewed as predatory state surrogates, then they may have the perverse effect of alienating citizens from the state and reinforcing reliance on nonstate alternatives.
Second and conversely, peacekeepers may prove so effective that they legitimize their own presence at the state's expense. Effective, legitimate international intervention may induce dependence: “to the extent that the public attaches legitimacy to the international trustee, the state may have more difficulty in earning the support and loyalty of those same individuals.”Footnote 57 In this case, exposure to peacekeeping may decrease reliance on formal institutions while increasing reliance on the UN.
Third, in many cases UN personnel depend on informal authorities to provide information and facilitate access to rural communities.Footnote 58 In the process, they may elevate the role these authorities play as arbiters of disputes,Footnote 59 and inadvertently embolden them to circumvent state laws.Footnote 60 In this case, peacekeeping may increase reliance on informal over formal institutions, and illegal over legal mechanisms of adjudication.
Finally and relatedly, some critics worry that because UN missions are constrained by language barriers and pressure from UN member states to avoid long, costly deployments, they are unlikely to affect state/society relations at all.Footnote 61 In other words, while there are multiple mechanisms by which international actors like the UN might reshape state/society relations, it is by no means a foregone conclusion that they will, or that their effects will be benign. However, the empirical record on this question is largely anecdotal, and even systematic studies tend to focus on “least-likely” cases where international intervention has most obviously foundered (the Congo, for example).Footnote 62 This is not to diminish the problems of negligence and abuse that have plagued some missions. It is simply to suggest that the local effects of international intervention should be treated as an empirical question—one to which we have few empirical answers.
Setting
Liberia is a small West African nation still recovering from fourteen years of civil war. Historically and today, rule of law in Liberia has involved a contest between elites on the coast and informal authorities in the interior whose resistance has long posed “the most important key” to state penetration.Footnote 63 A series of civil wars beginning in 1989 reduced the state to a “shell in which Liberians [continued] to live.”Footnote 64 In response, many took refuge in the same informal institutions that long governed social order in the hinterlands, and which survived the conflict without being “fundamentally altered.”Footnote 65
Liberia's informal sector comprises a hierarchy of paramount, clan, town, and quarter chiefs who operate in parallel to a network of sodalities known as “secret societies.” The 1986 constitution recognizes both customary and statutory law, but specifies different jurisdictions for each. Chiefs are legally authorized to resolve petty crimes and nonviolent domestic disputes, but are forbidden from adjudicating more serious crimes. Secret societies are not recognized under Liberian law, though they are important arbiters of disputes in rural areas. Customary institutions helped maintain local stability amid the upheaval of the civil war, and chiefs in particular remain indispensable for providing constitutionally sanctioned justice in areas beyond the state's reach, especially given corruption and severe resource constraints among the police and courts.
But legal hybridity also poses serious problems for the rule of law in Liberia. Informal authorities routinely ignore jurisdictional boundaries, creating endless opportunities to “forum shop.” They also tend to disadvantage youths, women, and ethnic minorities, many of whom believe the informal sector is “inherently biased against them.”Footnote 66 This helps explain why domestic violence, land disputes, and other interpersonal conflicts remain so pervasive in rural Liberia, and why they are so prone to escalation.Footnote 67 Informal authorities also sometimes demand that ongoing court cases be referred back to them, and some rely on illegal mechanisms of adjudication, especially trial by ordeal. This can include ingestion of poison, application of a heated cutlass to the suspect's skin, or immersion of the suspect's hand in a pot of boiling water. (In each case, it is believed that the innocent will be protected from harm, while the guilty will not.) Liberians are aware of the limitations of both formal and informal institutions, and many express frustration at the “lack of clear division of responsibilities” between the two.Footnote 68
The UN Mission in Liberia (UNMIL) deployed shortly after the signing of a Comprehensive Peace Agreement in 2003. UNMIL was in some respects a model force—large, adequately resourced, welcomed, and respected by most Liberians—but nonetheless it shared many similarities with other missions in the region,Footnote 69 and had a mandate that was “quite standard” for multidimensional operations of its kind.Footnote 70 From its inception, UNMIL prioritized establishing the rule of law, focusing on three goals in particular at the micro level: building confidence in the police and courts, enforcing the constitutionally mandated division of labor between formal and informal institutions, and eradicating trial by ordeal, which the UN views as a violation of human rights.Footnote 71
In pursuing these goals, UNMIL adopted a combination of the strategies described earlier.Footnote 72 While the mission lacked an executive policing mandate and therefore could not make arrests, it nonetheless assumed primary responsibility for responding to victims and investigating crimes, preventing felony offenses from being informally adjudicated, and protecting detainees from extrajudicial punishment. It also conducted public awareness campaigns to disseminate information about the newly reformed prisons, courts, and Liberian National Police (LNP); organized town hall meetings to encourage cooperation with the police; and faciltiated frequent contact between civilians and LNP officers through joint patrols, which began as early as 2003.
UNMIL personnel explicitly conceived of their role as one of state surrogacy. As one Head of Field Office explained to me, “the government was so feeble that UNMIL appeared as a proxy.”Footnote 73 But UNMIL was also acutely aware of the “Samaritan's dilemma,” and frequently allowed the government to claim credit for its achievements. In the security and justice sectors in particular, UNMIL consistently emphasized collaboration between mission personnel and their Liberian counterparts. LNP officers often cited the mentoring they received from UNMIL as evidence of improved performance—arguing, as one officer did during a town hall meeting, that while past “unethical behaviors” undermined confidence in the LNP, with the “training and discipline that UNMIL instilled in them, they are reformed and have changed.”Footnote 74
While UNMIL was large relative to the size of the country, it was tasked with rebuilding what was, at the time, the most “glaring world example of a ‘failed state.’”Footnote 75 In this sense, Liberia is a hard test for my theory to pass. The Liberian police and courts served as instruments of oppression during the civil war, and remained corrupt and dysfunctional after the fighting stopped. Citizens (rightly) feared state predation, and typically relied on nonstate security and justice providers instead. The boundaries separating the formal and informal sectors were (and remain) ambiguous, and jurisdictional conflicts were rife. At the same time, while Liberia is not “representative” of African countries, the problems afflicting it are common throughout the continent, and indeed much of the developing world.Footnote 76 UNMIL also shared enough similarities with other missions to ensure some degree of external validity—a point I return to in the conclusion.
Data and Measurement
Data
My analysis relies on an original survey covering 242 rural towns and villages in three Liberian counties—Lofa, Nimba, and Grand Gedeh. These counties have been priorities for both peacekeeping and state consolidation since the end of the civil war, and they are the site of recurring jurisdictional disputes between formal and informal authorities. They also have different ethnic compositions and suffered disparate forms and levels of violence during the civil war, thus capturing some of Liberia's diversity while still being similar and physically proximate enough to facilitate causal inference. While the 242 communities in my sample are not representative of Liberia, as I discuss in Appendix A.1.2, a comparison to a nationally representative survey conducted at the same time suggests they are quite similar to the average Liberian town or village. A more detailed justification for my case selection and a map of the 242 communities are provided in Appendix A.1.1.
Data were collected in three waves: between March and April 2009, November 2010 and January 2011, and February and April 2013. The first and second waves were the baseline and endline, respectively, for an unrelated field experiment on alternative dispute resolution.Footnote 77 The survey included two instruments: one administered to a random sample of twenty residents per community, and a second administered to four nonrandomly selected local leaders—typically a town chief, female leader, youth leader, and minority group leader. In each wave, data were collected from the same four local leaders and a new random sample of twenty residents, selected using the random walk technique described in Appendix A.1.3.
I complement the survey with semistructured interviews with over three dozen UN and government officials conducted during fifteen months of fieldwork in Liberia. While my analysis is primarily quantitative, I use qualitative data to motivate my measurement and identification strategies and inform my interpretation of the quantitative results. Further details on these interviews are provided in Appendix A.3.
Measurement
Operationalizing rule of law
Earlier I describe three common micro-level obstacles to the rule of law in postconflict settings: (1) citizens’ reliance on nonstate institutions to adjudicate crimes that fall unambiguously under state jurisdiction; (2) nonstate authorities’ reliance on illegal mechanisms of dispute resolution; and (3) citizens’ perceptions of state institutions as corrupt and biased. I use the survey to operationalize these three obstacles in Liberia.
To operationalize the first obstacle, respondents were read three hypothetical scenarios of crime and violence, then asked to choose the authority they would “most prefer” to respond. Options included three authority types: formal (police, military, and other “government people”), informal (chiefs and other “traditional leaders”) and UNMIL. All three hypothetical scenarios described incidents that fall unambiguously under state jurisdiction, and that pose a serious risk of further escalation: a murder, an outbreak of mob violence, and an ethnic riot resulting in fatalities.Footnote 78 I code each respondent's choice in each scenario, as well as their modal choice across the three scenarios.
To operationalize the third obstacle, respondents were also asked whether they view the state as corrupt or biased toward particular ethnic or religious groups. For comparison, they were asked the same questions about informal authorities and UNMIL as well.
Operationalizing the second obstacle requires measuring behaviors that are, by definition, illegal. Direct questions would likely overestimate compliance with state laws. To mitigate this problem, the third wave of the local leaders survey included a list experiment designed to measure the prevalence of trial by ordeal.Footnote 79 List experiments aim to improve the credibility of survey self-reports by allowing respondents to answer sensitive questions indirectly. In my case, respondents were first read a vignette describing a hypothetical burglary in their community, then read a list of options for “handling” the situation, including three control items and one sensitive one (trial by ordeal). Burglaries are criminal offenses, and by law should be reported to the police; as I learned in pretesting, however, informal authorities sometimes refuse to call the police when the perpetrator is unknown, instead using trial by ordeal to divine the burglar's identity. Trial by ordeal is less common for more serious crimes (like murder), or in cases where the perpetrator is known (like assault). Burglaries are thus an especially illuminating case for estimating reliance on the police and reliance on trial by ordeal simultaneously.
Treatment group respondents were read the following vignette (in Liberian English):
Let's say someone busts into your house and took something, and nobody knows who did the thing. I am going to tell you some different ways that people can handle that matter. I want you to tell me how many ways can happen in this community. Don't tell me which ones; just tell me how many.
1. Call the police.
2. Just leave it alone and do nothing about it.
3. Take it to the UNMIL people.
4. Call for sassywood or sand cutter or hot cutlass [trial by ordeal].
Control group respondents were read the same scenario and the three control items. A semi-sensitive option (“just leave it alone and do nothing”) was included among the control items to minimize the risk of floor and ceiling effects.Footnote 80 As I show in Appendix A.2.1 and A.2.2, the distribution of responses suggests this strategy worked, and a diagnostic test fails to reject the null of no design effects.
Following Corstange, control group respondents were asked each control question directly, rather than in the form of a list.Footnote 81 This approach has two advantages. First, it increases statistical precision by reducing the number of unknown parameters to be estimated. Second and more important, it allows me to test the impact of UNMIL's activities not just on the sensitive item, but on the control items as well. In most list experiments, control items are useful only insofar as they facilitate measurement of the sensitive one(s). But this discards potentially valuable data. In my case, the control items are theoretically relevant in and of themselves, since I expect exposure to UNMIL to increase reliance on the police while decreasing reliance on trial by ordeal. With Corstange's modified design, both of these effects can be identified.Footnote 82
Operationalizing exposure to peacekeeping
Most interactions between peacekeepers and civilians occur during one of three activities: patrols, public works, and interventions to resolve disputes. Because these activities are both frequent and universal, their effects in Lofa, Nimba, and Grand Gedeh are more likely to generalize to other parts of Liberia, and potentially to other countries as well. Unfortunately, while UNMIL maintains records on some of these activities, the records are not comprehensive, and in some years they are missing altogether. I therefore rely on reports from local leaders instead. Because local leaders often act as liaisons for UN personnel,Footnote 83 they should be able to provide accurate reports of peacekeepers’ activities in their communities.
I use these reports to construct three sets of independent variables:
1. Dummies that take a 1 if UNMIL patrolled weekly, monthly, occasionally, or never in the year preceding data collection.
2. A dummy that takes a 1 if UNMIL built or repaired infrastructure in the community in the preceding year.
3. A dummy that takes a 1 if UNMIL intervened to resolve disputes in the community in the preceding year.
To minimize measurement error, I code the modal response across the four local leaders for each indicator.Footnote 84 As an individual-level complement to these community-level proxies, I also code an indicator for whether residents interacted (or “spent time”) with UNMIL personnel. I show in Appendix A.5.1 that local leaders’ reports of UNMIL patrols and interventions are highly positively correlated with residents’ reports of interacting with UNMIL personnel, despite being measured in different ways using different samples. This should further mitigate measurement error concerns. Descriptive statistics are provided in Appendix A.1.5.
Empirical Strategy
UNMIL's activities are not randomly assigned. I use multiple identification strategies to isolate their effects.
Modeling UNMIL's Selection Process
First, I combine insights from UN Secretary-General progress reports with my own qualitative interviews to model the selection process underlying UNMIL patrols, interventions, and public works. While there is undoubtedly some slippage between the selection process described by UN personnel and the one implemented on the ground, the mission's selection criteria are, for the most part, transparent and quantifiable, facilitating identification.
My controls were measured in the first wave of data collection in 2008, and include population; prevalence of crime and violence; cell phone coverage; distance to the nearest usable road; average household wealth; availability of social services; proximity to natural resources; police and NGO presence; and exposure to wartime violence. I detail my rationale for each control in Appendix A.1.4. I further support my community-level analysis with an individual-level analogue that uses residents’ reports of interactions with UNMIL personnel as the independent variable.
Exploiting Variation Generated by the Deaths of Seven Peacekeepers in Côte d'Ivoire
Second, I leverage a tragedy in neighboring Côte d'Ivoire to generate plausibly exogenous random variation in the intensity of exposure to UNMIL during my third wave of data collection. In 2010, a contested presidential election in Côte d'Ivoire sparked a civil war between loyalists of incumbent Laurent Gbagbo and challenger Alassane Ouattara. With assistance from France and the UN Operation in Côte d'Ivoire (UNOCI), Ouattara's forces seized control of most of the country, culminating in Gbagbo's arrest in April 2011. Ouattara was installed as president shortly thereafter.
While the fighting subsided after Gbagbo's capture, militias continued to operate in the country's volatile western departments, several of which share a border with Liberia. On 8 June 2012, seven UNOCI peacekeepers were killed in an ambush near the Ivorian border town of Para. The killing shocked the UN on both sides of the border: while UNMIL and UNOCI had intensified joint operations soon after the start of the conflict, the ambush reoriented the missions’ attention much more specifically to Para and the surrounding area. As I show in Appendix A.4.2, this focus is evident in the data: respondents who lived near Para in 2012 were much more likely to report exposure to UNMIL than those who lived further away, even within a subsample of border communities.
I use proximity to Para as an instrument for individual-level exposure to UNMIL in 2012.Footnote 85 This identification strategy relies on two assumptions. First, I assume there is nothing especially strategically important about the area surrounding Para, and therefore that the ambush (and UNMIL's subsequent response) could have just as easily occurred anywhere else along the border. Second, I assume that, conditional on covariates, the only mechanism through which proximity to Para might have affected Liberians’ attitudes was the higher “dose” of peacekeeping they subsequently received.
Given that violence both during and after the Ivorian civil war occurred throughout the country's southwestern districts, and that the district in which the ambush occurred did not suffer any other attacks in the first half of 2012 (and only two more in the second half),Footnote 86 there is little reason to suspect that Para was more strategically important than any other border town. Still, communities near Para are, by definition, near the Ivorian border as well, and border communities may differ from others in many ways. To minimize the risk of confounding, for this analysis I restrict my sample to communities located within twenty kilometers of the border. (I show in Appendix A.4.3 that my results are robust to fifteen-kilometer and twenty-five-kilometer bandwidths as well.) These communities all share similar demographic, economic, and political profiles, lending additional credence to my identifying assumptions.
In Appendix A.5.4, A.5.5, and A.5.6 I also empirically explore three possible exclusion restriction violations: (1) that the ambush precipitated an influx of refugees into Liberia, (2) that it exacerbated violence in proximate Liberian communities, and (3) that it elicited an increase in LNP as well as UNMIL presence. Fortunately for my analysis (and probably for Liberians as well), the Liberian government closed the border with Côte d'Ivoire on 12 June, just four days after the attack, and did not reopen it until November of that year. This should have mitigated any consequences for communities on the Liberian side. Moreover, while Liberia's security forces did increase their activities along the border in response to the ambush, their capacity and presence remained extremely limited. After restricting my sample to border communities, I show that respondents living near Para were no more likely to be migrants than those living further away, and that communities located near Para were no more likely to report serious incidents of collective or interpersonal violence, and no more likely to report frequent patrols by the LNP. While the exclusion restriction is ultimately untestable, these nulls provide some evidence against the most plausible violations.
Exploiting Variation Generated by UNMIL's Initial Deployment Decisions
Finally, drawing on work by Mvukiyehe and Samii,Footnote 87 in Appendix A.4.1 I use the locations of UNMIL bases as a source of plausibly exogenous variation in the intensity of exposure to UNMIL personnel. As Mvukiyehe and Samii note, before peacekeepers deploy, they must make costly siting decisions based on highly incomplete information about potential host communities. Information quality is likely to be especially poor in rural areas, where data are scarce and obstacles to communication and transportation limit intelligence gathering. Initial siting decisions also depend on the preferences of the Department of Peacekeeping Operations in New York, which often reflect external considerations as much as they reflect factors internal to the host country itself.Footnote 88
I calculate the proximity of each community in my sample to the nearest UNMIL base that had already been established by the end of the mission's first full year of deployment (2004), after which new and better information should have become available. I then use that as an instrument for individual-level exposure to UNMIL. Following Mvukiyehe and Samii, to minimize any residual selection effects, I include control variables drawn from the same data sources that UNMIL consulted at the time of deployment, including an early 2004 rapid needs assessment by the UN Office for the Coordination of Humanitarian Affairs (OCHA). To minimize potential exclusion restriction violations arising from UNMIL's preference for establishing bases in or near cities, I restrict my sample to communities located within five kilometers of either a base or a city that did not host a base in 2004, using UNMIL's own deployment maps to identify locations that might have been considered as viable alternatives. Further details and results from this analysis are provided in Appendix A.4.1, and are consistent with those reported here.
Triangulation
Of course, none of these identification strategies is flawless. My goal is to triangulate between them. Triangulation is especially promising in my case, since each strategy is susceptible to biases of a very different sort. For example, communities located near UNMIL bases tend to be larger, more urban, and more accessible than average, while those located near Para tend to be smaller, more rural, and less accessible. Similarly, while unobserved selection processes may explain UNMIL's decision to engage with particular communities, those processes are likely very different from the ones that explain individual Liberians’ decisions to engage with UNMIL personnel. There is little reason to expect these biases to have the same magnitude or direction, and consistency across approaches should lend credence to a causal interpretation of my results.
Additional Empirical Challenges
Beyond selection bias, my analysis must overcome at least three additional empirical challenges: spillover, social desirability bias, and reverse causality. I address each in turn.
Spillover
In principle, the effects of UNMIL's activities might spill over from one community or individual to the next, causing bias. However, Liberia's dilapidated transportation and communication infrastructure mitigates the risk that peacekeeper presence in one location might affect citizens’ attitudes in another. More important, any spillover would likely bias my estimates toward the null. If, for example, UNMIL persuades citizens in one community to abide by jurisdictional boundaries, and those citizens convince residents of another community to do the same, then comparing the two will underestimate UNMIL's impact on the rule of law.
Social desirability bias
If UNMIL succeeds only in teaching respondents to say what they think survey enumerators want to hear—that state institutions are fair and transparent, for example, or that trial by ordeal is never used—then I will overestimate the mission's impact. This problem afflicts any study using survey data to measure outcomes, and mine is no exception. To mitigate this risk, all surveys were conducted in private by trained Liberian enumerators affiliated with a local research NGO (rather than the government or the UN), and respondents were repeatedly reassured of their anonymity. There is no taboo against openly expressing one's attitudes toward state laws and institutions in Liberia; indeed, as I show in Appendix A.1.5, many respondents were highly critical of the Liberian government, suggesting they were not shy about articulating their views. I also show that despite UNMIL's efforts to improve citizens’ perceptions of the state, its presence appears only to have exacerbated complaints of state corruption, at least in the short term. Social desirability bias cannot explain this result. My use of a list experiment to measure trial by ordeal should further ameliorate social desirability concerns.
Reverse causality
Individuals who prefer relying on the state may also be more likely to interact with UNMIL, raising the prospect of reverse causality. However, my results are inconsistent with this interpretation. Intuitively, if reverse causality were a threat, we would expect to observe a positive correlation between exposure to UNMIL and preferences for UNMIL as well. (It would be strange if citizens who preferred relying on the state were more likely to interact with UNMIL, while those who preferred relying on UNMIL were not.) I find little evidence of this positive correlation; indeed, in some specifications the correlation is negative. This should mitigate reverse causality concerns.
Results
Exposure to UNMIL Increased Citizens’ Reliance on Formal over Informal Institutions
Figure 1 reports the effects of UNMIL patrols, interventions, and public works on civilians’ preferences over potential security providers. The figure plots marginal effects and 90 and 95 percent confidence intervals from a multinomial logit regression given by 
$y_i = \alpha + \sum\nolimits_{j = 1}^J {\beta _j T_{c,j}} + \sum\nolimits_{k = 1}^K {\zeta _k X_{i,k} + \sum\nolimits_{l = 1}^L {\zeta _l W_{c,l}} + \sum\nolimits_{m = 1}^M {\lambda _m D_m} + \varepsilon _i} $, where y i denotes respondent i's modal preference across three hypothetical scenarios (murder, mob violence, and ethnic riots),Footnote 89 T i,j indexes five forms of exposure to UNMIL (public works, interventions, and weekly, monthly, or occasional patrols), X i,k and W c,l denote k individual-level controls and l community-level controls, respectively, and D m denotes district fixed effects. Observations are weighted by the inverse probability of sampling and standard errors are clustered by community.
Figure 1. Effects on citizens' preferences using local leaders' reports of exposure to UNMIL
I find strong evidence that UNMIL patrols increased citizens’ reliance on formal over informal institutions to resolve the most serious incidents of crime and violence. Residents of communities that reported weekly patrols were fifteen percentage points more likely to prefer relying on formal institutions, and sixteen percentage points less likely to prefer relying on informal ones. These countervailing effects are highly statistically significant both in absolute terms and relative to one another. The effects are smaller for monthly and occasional patrols, though not statistically significantly so.
Evidence for the impact of interventions and public works is more ambiguous. While both appear to have increased reliance on formal over informal institutions, the difference between these effects is only weakly significant. A possible explanation for the discrepancy with patrols lies in the distinct opportunities for interaction that each of these activities affords. Citizens were more likely to report “spending time” with peacekeepers in communities where UNMIL patrolled than in communities where it built public works or intervened to resolve disputes (see Appendix A.5.1). If peacekeepers affect civilians’ attitudes in part through interpersonal interactions, then it is perhaps unsurprising that patrols had more robust effects than interventions or public works.
Equally important, I find little evidence to suggest that exposure to UNMIL induced dependence on the mission itself. While interventions do appear to have strengthened preferences for UNMIL over informal authorities, the effects on preferences for the mission and preferences for the state are statistically indistinguishable. Patrols did not affect civilians’ reliance on UNMIL one way or the other. If anything, public works actually weakened preferences for UNMIL, both in absolute terms and relative to the state. While somewhat surprising, this is consistent with Pouligny's observation that in many cases, “far from the mission being thanked for its efforts,” public works are interpreted as evidence that peacekeepers “could have done more.”Footnote 90
Figure 2 replicates this analysis using citizens’ (rather than local leaders’) reports of interactions with UNMIL personnel. The figure again displays marginal effects from a multinomial logit regression with district fixed effects and individual and community-level controls.Footnote 91 Observations are again weighted by the inverse probability of sampling, and standard errors are clustered by community.
Figure 2. Effects on citizens’ preferences using citizens’ reports of exposure to UNMIL
Consistent with Figure 1, I find that residents who interacted with UNMIL were six percentage points less likely to prefer relying on informal authorities, and two percentage points more likely to prefer relying on formal ones. While the latter effect is not statistically different from 0, it is statistically different from the former. Residents who interacted with UNMIL were also four percentage points more likely to prefer relying on UNMIL. This effect is statistically significant on its own, and is statistically different from the effect on preferences for informal authorities. It is not, however, statistically different from the effect on preferences for the state. Together these results suggest that even if peacekeeping induces reliance on the UN, it does so at the expense of informal authorities, not formal ones. I obtain similar results when using proximity to the nearest UNMIL base as an instrument for individual-level exposure to UNMIL personnel (Appendix A.4.1).
Table 1 reports results using proximity to Para as an instrument instead. To minimize excludability violations, I restrict this analysis to communities located within twenty kilometers of the Ivorian border. Citizens who reported interacting with UNMIL as a function of proximity to Para were forty-two percentage points more likely to prefer relying on formal authorities. They were also eleven percentage points more likely to prefer relying on informal authorities, though this effect is not statistically significant and is statistically smaller than the effect on preferences for formal authorities. As I show in Appendix A.4.3, these point estimates are virtually identical at wider and narrower bandwidths. Corresponding first-stage results are reported in Appendix A.4.2.
Table 1. Effects on citizens' preferences using proximity to Para as instrument
Notes: Coefficients from two-stage least squares regressions using proximity to Para as an instrument for exposure to UNMIL in 2012. Estimates are weighted by the inverse probability of sampling. Standard errors, clustered by community, are in brackets. *p < .1; **p < .05; ***p < .01.
It is interesting that citizens who interacted with UNMIL were also fifty-three percentage points less likely to prefer relying on the mission itself—a substantively large and highly statistically significant effect. After the ambush near Para, citizens may have interpreted UNMIL's increased presence as a signal of its prior inability or unwillingness to adequately protect civilians in the volatile border region. Alternatively, UNMIL's highly militarized response to the incident may have alienated citizens in ways that more typical peacekeeping activities do not. This echoes concerns that Pouligny and others have raised about over-reliance on military (rather than civilian) personnel as the “face” of peacekeeping operations.Footnote 92 Whatever the explanation, the net effect on citizens’ reliance on the state is clear, and is consistent with the results of my selection on observables strategy.
Exposure to UNMIL Increased Informal Authorities’ Reliance on Legal over Illegal Mechanisms of Dispute Resolution
Figure 3 reports UNMIL's impact on local leaders’ responses to my list experiment. Given the need to estimate multiple conditional probabilities for relatively rare events, multivariate list experiment analyses can generally accommodate only a small number of regressors. For tractability I include community-level controls only. Point estimates are derived from the approximate maximum likelihood estimator proposed by Corstange.Footnote 93 Marginal effects and corresponding confidence intervals are derived from Monte Carlo simulations, following Blair and Imai.Footnote 94
Figure 3. Effects on local leaders’ responses to list experiment
I find that exposure to UNMIL increased the likelihood that the police would be called to respond to a (hypothetical) burglary, though the magnitude of the effect varies by type of exposure. In communities that reported weekly patrols, local leaders were approximately twenty-five percentage points more likely to believe the police would be called, and twenty-five percentage points less likely to believe trial by ordeal would be used. These effects are each statistically different from 0, and are statistically different from one another as well. Interventions had a similar effect on the likelihood of calling the police, and public works had smaller but still positive and statistically significant effects as well. While neither interventions nor public works affected trial by ordeal, the positive effect on calling the police is strongly statistically different from the null on trial by ordeal in communities that reported interventions, and weakly different in communities that reported public works.
Evidence is more mixed for less frequent patrols. Occasional patrols had no effect one way or the other. Monthly patrols, in contrast, appear to have had perverse effects. Local leaders were more likely to believe trial by ordeal would be used in communities that reported monthly patrols, and less likely to believe the police would be called, though the latter effect is not statistically significant, and the former only marginally so. These two effects are also only weakly statistically different from each other.
Consistent with Figure 2, I find that exposure to UNMIL increased the likelihood that UNMIL personnel would be called, though again, the magnitude of the effect varies. Local leaders were more likely to believe UNMIL would be called in communities that reported weekly patrols, interventions, or public works. They were less likely to believe UNMIL would be called in communities that reported monthly patrols, and no more or less likely to believe UNMIL would be called in communities that reported occasional patrols. Importantly, even in communities that reported weekly patrols, interventions, or public works, the effect on calling UNMIL is statistically identical to the effect on calling the police, suggesting again that insofar as the mission increased reliance on its own personnel, it did so at the expense of informal institutions rather than formal ones.
Exposure to UNMIL Increased Citizens’ Perceptions of State Corruption
Tables 2 and 3 report the effects of exposure to UNMIL on perceptions of state corruption and bias. For comparison, I include effects on perceptions of informal authorities and UNMIL as well. Table 2 tests the effects of community-level patrols, interventions, and public works; Table 3 tests the effects of individual-level interactions with UNMIL personnel. Both tables report coefficients from linear probability models with the same controls, weights, and fixed effects as before. Standard errors are again clustered by community.
Table 2. Effects on citizens' perceptions using local leaders' reports of exposure to UNMIL
Notes: Coefficients from OLS regressions. Estimates are weighted by the inverse probability of sampling. Standard errors, clustered by community, are in brackets. *p < .1; **p < .05; ***p < .01.
Table 3. Effects on citizens' perceptions using citizens' reports of exposure to UNMIL
Notes: Coefficients from OLS regressions. Estimates are weighted by the inverse probability of sampling. Standard errors, clustered by community, are in brackets. *p < .1; **p < .05; ***p < .01.
Counterintuitively, I find that, if anything, exposure to UNMIL exacerbated perceptions of state corruption. Residents of communities that reported weekly, monthly, or occasional patrols were, respectively, ten, eleven, and eight percentage points more likely to describe state institutions as corrupt. In communities that reported interventions, residents were four percentage points more likely to describe state institutions as corrupt, though this effect is only marginally significant. Similarly, citizens who reported interacting with UNMIL personnel were five percentage points more likely to describe the state as corrupt. These apparently contradictory effects may be a function of the somewhat mixed messages that UNMIL delivered, urging citizens to obey state laws and trust state institutions while also encouraging them to recognize and report acts of corruption and abuse.
Residents who reported interacting with UNMIL were also four percentage points less likely to describe UNMIL as corrupt, and four percentage points less likely to describe UNMIL as biased. Exposure to UNMIL appears to have had no effect on perceptions of informal authorities one way or the other. Residents of communities that reported interventions were six percentage points more likely to describe informal authorities as corrupt, but otherwise these results are null across the board.
UNMIL's Beneficial Effects Persisted, While Adverse Effects Decayed Over Time
How durable are these effects likely to be? If citizens become increasingly reliant on state institutions that they perceive as corrupt and biased, they may revert to nonstate alternatives over time. Moreover, if the impact of peacekeeping on rule of law depends on peacekeepers’ physical presence, then changes in citizens’ attitudes may prove fleeting. In Figure 4 I test whether exposure to UNMIL continued to affect citizens’ preferences over potential security and justice providers two years later. This specification is identical to the earlier one, except that the dependent variable and individual-level controls are measured in 2012, rather than 2010.
Figure 4. Persistence of effects on citizens’ preferences using local leaders’ reports of exposure to UNMIL
The results are broadly consistent: in communities that reported weekly, monthly, or occasional patrols in 2010, citizens continued to express a pronounced preference for formal over informal institutions after two years. Unsurprisingly, these effects are weaker in the medium term than in the short term, and the effects of public works, which were only weakly statistically significant to begin with, are no longer distinguishable from 0. But otherwise, the results are consistent with those before.
Table 4 extends this analysis by testing the effects of exposure to UNMIL in 2010 on citizens’ perceptions of state corruption and bias in 2012. Two years later, the apparently adverse effects that I observe in the short term decay to nulls. I do find some suggestive evidence that interventions mitigated perceptions of UNMIL as corrupt and biased, while public works exacerbated perceptions of UNMIL as biased. But the effects on perceptions of the state are substantively and statistically insignificant. Taken together, Figure 4 and Table 4 suggest that the adverse effects of UNMIL's presence dissipated over time, while the beneficial effects persisted (even if their magnitude decayed).
Table 4. Persistence of effects on citizens' perceptions using local leaders' reports of exposure to UNMIL
Notes: Coefficients from OLS regressions. Estimates are weighted by the inverse probability of sampling. Exposure to UNMIL is measured in 2010; citizens’ perceptions are measured in 2012. Standard errors, clustered by community, are in brackets. *p < .1; **p < .05; ***p < .01.
How likely are these effects to endure even after UNMIL withdraws? While I cannot answer this question directly,Footnote 95 in Appendix A.5.7 I show that the beneficial effects of UNMIL's presence persisted even in communities that reported no further exposure to UNMIL personnel. The adverse effects, in contrast, did not. These analyses should be interpreted with some caution because UNMIL may have discontinued patrols, interventions, and public works in particular communities for reasons that are endogenous to citizens’ perceptions. Still, my results raise the possibility that UNMIL induced a lasting change in modes of dispute resolution at the local level, and that the mission's impact did not depend on the continued physical presence of mission personnel.
Conclusion
Establishing the rule of law is central to the mandates of UN peacekeeping operations around the world. I use a survey and list experiment in Liberia to show that exposure to peacekeeping increased citizens’ reliance on state over nonstate authorities to adjudicate crimes that fall unambiguously under state jurisdiction, and increased reliance on legal over illegal mechanisms of dispute resolution. I also show that peacekeeping did not mitigate perceptions of state bias and, if anything, exacerbated perceptions of state corruption—but only in the short term. Over time, the beneficial effects of UN presence persisted, while the adverse effects decayed.
How generalizable are my results likely to be to other parts of Liberia, other countries, or other dimensions of the rule of law? I argue for generalizability to other regions of Liberia earlier and in Appendix A.1.2. But there are reasons to believe my results may generalize to other African countries as well, where the vast majority of peacekeepers are currently deployed.Footnote 96 UNMIL's mandate is similar to those of other multidimensional missions on the continent, and the tensions between formal and informal institutions that characterize the rule of law in Liberia are pervasive throughout Africa and beyond.Footnote 97 The three activities whose impact I assess (patrols, public works, and interventions) are also common to almost all peacekeeping operations around the world. These features of my case selection and research design lend support to the generalizability of my results.
Moreover, some of the potential scope conditions for my findings are not as restrictive as they may seem. While UNMIL has long been perceived as competent and legitimate, the mission was also embroiled in multiple scandals over the course of its mandate. Indeed, one of the few systematic studies of sexual exploitation and abuse by peacekeepers was conducted in Monrovia, and found astronomical rates of transactional sex between UNMIL personnel and Liberian women and girls.Footnote 98 UNMIL has been accused of negligence and bias as well.Footnote 99 Consistently virtuous behavior on the part of peacekeepers is therefore unlikely to be a scope condition. At the time of my study, Liberia was one of the weakest and most aid-dependent states in the world, and the Liberian police and courts remained corrupt and discombobulated. Consistently virtuous behavior on the part of state institutions is therefore unlikely to be a scope condition either.
I believe the most likely scope condition for my results is some prior degree of peace and stability. Previous analyses have shown that UN missions are more effective at consolidating peace than fighting an ongoing civil war.Footnote 100 While Liberia has suffered incidents of instability in the postconflict period, and while some have escalated into regional or national crises,Footnote 101 most have been localized and manageable. During periods of conflict, the short-term exigencies of waging war may crowd out the longer-term demands of establishing the rule of law. These periods also by definition necessitate a more militarized approach to peacekeeping, which previous studies have suggested are especially likely to alienate civiliansFootnote 102—a finding that resonates with my own results.
Whether my results will generalize to other dimensions of the rule of law is less certain, and is largely beyond the scope of this study. Still, my theory and findings provide some grounds for optimism. In hybrid legal systems, the actions of citizens and nonstate authorities can easily undermine other important requirements for the rule of law. For example, there is no way to ensure consistency between the law as enacted and the law as enforced,Footnote 103 or respect for the review powers of the courts,Footnote 104 or similar treatment for similar cases,Footnote 105 unless both citizens and informal authorities abide by jurisdictional boundaries and recognize state law as supreme. Related outcomes similarly rely on citizens’ willingness to comply with state laws and institutions: taxation, for example, or collective action in the provision and maintenance of public goods. The impact of international intervention on citizens’ relationships with state security and justice institutions may spill over into their relationships with other branches of government as well, promoting rule of law along multiple dimensions simultaneously. I leave this possibility for future research to explore.
Taken together, my results suggest that UN missions have complex but ultimately beneficial effects on the rule of law at the micro level, and that these effects are transmitted at least in part through localized, relatively mundane encounters with host populations. The salience of these day-to-day, face-to-face interactions has long been recognized in the literature on police/community relations and state/society relations more generally,Footnote 106 but has only recently begun to emerge in the study of peacekeeping. Micro-level peacekeeping is increasingly seen as key to the UN's success,Footnote 107 and micro-level analyses will become more urgent as international interveners pursue ever more ambitious mandates aimed at ever more transformative changes among host populations. Indeed, to the extent that UN missions are expected to promote stable liberal democracies at the macro level, the foundations of that process are presumed to emerge at the micro level, “in the political attitudes and behaviors of ordinary citizens.”Footnote 108
While such analyses are still relatively rare, micro-level studies are generating a mounting body of evidence that UN missions can reshape the social and political fabric of host countries in ways that belie more pessimistic perspectives. My results resonate with several recent studies—in particular, Mvukiyehe and Samii's finding that UN security committees increased the vote share of national over “parochial” candidates in Liberia;Footnote 109 Mvukiyehe's finding that exposure to UN personnel increased national but not local political participation in Liberia;Footnote 110 Nomikos's finding that the UN's presence in Mali improved relations between rival ethnic groups;Footnote 111 and Gordon and Young's finding that UN provision of security and humanitarian relief in Haiti improved civilians’ perceptions of the mission and increased their willingness to provide information to UN personnel.Footnote 112 While more analyses in more settings are needed, these studies suggest that peacekeeping operations can have more robust, positive effects on citizens than skeptics typically believe is possible.
Future studies should test the effects of international intervention on additional dimensions of social and political life. They should also further disaggregate peacekeeping to capture more variation in what UN personnel actually do in the field. Much existing research focuses on the role peacekeepers play in resolving commitment problems and mitigating information asymmetries. But peacekeepers do much more than this, from monitoring elections to training police officers to promoting gender equality and the rights of children. Research has not kept up with practice, and there are many questions that simply cannot be answered without more disaggregated data. Do peacekeepers’ attempts to control the circulation of small arms improve subjective perceptions of personal safety? Does their support to political parties and electoral commissions yield democratic dividends—increasing turnout, reducing electoral violence, and instilling confidence in the democratic process? Answering these questions requires disaggregated data, ideally at the micro level. Generating and analyzing this data is, in my view, one of the literature's most promising frontiers.
Supplementary Material
Supplementary material for this article is available at <https://doi.org/10.1017/S0020818319000031>.
