Introduction
Who governs policing determines for whom and how security is provided (Johnston and Shearing 2003). This is because security means different things to different people. To simplify, security governance is subject to tension between a top-down and a bottom-up view of security (Luckham and Kirk 2013). From the point of view of the state and elites, security implies the creation and preservation of a particular social order, even if this requires the use of violence. From the point of view of individuals, security refers to a basic entitlement to physical safety and integrity. Neither view is monolithic, of course. Security responsibilities tend to be distributed across tiers of government, with perspectives and priorities varying between lower and higher tiers (Devroe 2013; Hataley and Leuprecht 2014; Virta 2013). Comparing cities in Italy and France over ten years, Germain (2012) finds that if local police chiefs receive instructions from the mayor rather than the governor/prefect, they will focus on public tranquillity rather than social order. Mayors act on public disorders not just to respond to citizen demands for action, but also because acting on disorders — and framing these in terms of security — is a source of legitimacy for them and a successful move in electoral politics. On the other hand, governors in these two countries are not subject to electoral pressures. The governance of security ends up resulting from a complex power play between different individual and group interests and perceptions.
If who governs policing determines for whom and how security is provided, then in theory security can be made more inclusive—responding to bottom-up concerns, with policing agents viewing citizens as partners and striving to minimise harm—by re-distributing responsibilities and decision-making power among tiers of governance.
However, the effect of a change in formal institutional design is difficult to predict. When Colombia passed a decree in 2011 providing extended powers to governors and mayors in the territorial management of security, some mayors abused their position, having police commanders fired for personal reasons or preventing police commanders from performing their duties (Garcia Hernández et al 2016). Mayors were not always aware that they were the first police authority; in any case, they had significant leeway in interpreting legislative frameworks. On the contrary in Nigeria, constitutionally the federal government has exclusive domain over the police. Nevertheless, the state of Lagos was able to implement reforms to policing after it created a Lagos security trust fund in 2007 to supplement the funding of the federal police in the capital state (Agboga 2021). This arrangement enabled the state government to leverage its unique economic advantage to push forward significant improvements to the performance of the police in the capital state.
It is an important and challenging question, both from a theoretical and policy perspective, to understand the forces that constrain or enable the translation in practice of formal rules on governance in order explain such outcomes (Benz and Broschek 2021). The above-mentioned empirical studies offer important clues as to the factors that explain such divergences in the case of security governance. Causality is, however, notoriously tricky to establish. This article traces the effect of a constitutional change on the governance of security in Nepal in different tiers of governance, using process tracing to identify factors that prevented or enabled change to take hold, or diverted its effects. We first present our case study, mapping the strategies we employed to collect data. We then situate our investigation in the current literature on micro-level institutional change. Next, we operate a cross-section of institutional change in the governance of security in Nepal, tracing the forces that guided change across tiers of governance. We conclude with some theoretical and practical insights.
Nepal as a case study
Nepal constitutes a rare opportunity to observe and document institutional change to the governance of security as it unfolds. After emerging in 2007 from a period of popular upheaval and civil war, the Himalayan state adopted a new constitution in 2015, which marked the transition from a centralised system of government with some political decentralisation to the district and local (village and municipal) levels, to a federal system with autonomous provinces and political decentralisation to the local (urban and rural municipalities) level. The constitution reinstated locally-elected governments at the ward and municipal levels that had been suspended in 2002 at the height of the conflict. Importantly, the 2015 constitution transferred exclusive powers for peace and order to new provincial governments. The adoption of the 2015 constitution was followed in 2017 by a wave of elections at the ward, municipal, provincial and federal levels.
To document changes in the governance system in all its subtleties, we centre our study on a single municipality, Janakpur. The city is located in the plains along the Indian border known as the Tarai. The region’s traditional inhabitants, the Madhesis, have long felt discriminated against by Pahadis, the country’s hill-origin elites. Nepal introduced federalism in large part to assuage calls for greater autonomy emanating from the Tarai in a 2007 popular uprising, the Madhes andolan. In Janakpur, too, people had taken to the streets, and regular popular protests continued in the city until 2015, as Madhesi leaders feared their rights would not be adequately protected in the new constitution. When provincial governments were established in March 2018, the sub-metropolitan city of Janakpur served both as district headquarters and provincial capital, making it possible to observe in a single city all but the federal tier of government. This was also the only one of Nepal’s seven provinces not ruled by the Communist Party. Once established, the provincial government took the lead in holding the federal cabinet to account for the full implementation of the constitution, particularly the devolution of policing powers to the new provincial governments. In the words of a commentator, ‘Province 2 has been ground zero in the battle for federalism’ (Nepali Times 2018).1 If the new constitutional framework on security governance were to have an impact anywhere in Nepal, Janakpur would be the place.
Causality can be investigated in a single case study by using process tracing (Beach and Pedersen 2013; Bennett and Checkel 2015; Mahoney 2015). Key to good process tracing is the development of a detailed set of hypotheses likely to explain the outcome of interest — in this case, an observed change in the distribution of decision-making power over policing among tiers of governance. The method relies on a wide range of evidence, including academic literature, expert interviews, media accounts, and interviews with individuals directly involved with the process of interest (Bennett and Checkel 2015: 18). The diversity of sources of evidence is a hallmark of good process tracing.
Both authors gathered data over seven months of fieldwork in Janakpur, between October and December 2018, and February and July 2019. During this time, we conducted 61 interviews (17 of which were unstructured scoping interviews) and four focus group discussions with ward representatives, and conducted ethnographic observations. We were capturing the reality of security provision and governance as told by actors participating in or observing this system from insider viewpoints, rather than relying on the aspirational view described in legal and administrative texts. A review of newspaper articles published in the English-speaking Nepali press and available online added some historical and political context to the events we were witnessing and recording.2 From January 2019, the first author kept a running tab on important national-level debates and developments regarding security governance. We can therefore trace changes we observed in the governance of security right across the wards, the municipality, the district of Dhanusa, the Province 2 government, up to the federal institutions.
We recognise, of course, that the data we collected is not flawless. Attempting to reconstruct a multi-tiered governance network was an ambitious endeavour. Where we collected detailed ethnographic observation and interview data at the ward level, our data on the debates happening in higher tiers of governance is coarser and incidental. As our gaze lifts further away from Janakpur’s streets, our vision also becomes blurrier, increasingly relying on soundbites harvested from newspapers. We multiplied our methods because we could not be everywhere at once. But multiplying methods also enabled us to fact-check and triangulate our data, and compensate for and even out the inevitable biases of single research methods. In the end, we assembled a model to trace influence all the way to ministerial circles, providing a more complete picture of the systemic dynamics animating the Janakpur urban security network.
Explaining divergence
We use Crawford and Ostrom’s (2005: 138) concepts of rules-on-paper and rules-in-use to distinguish formally codified rules from shared rules, norms and strategies, whether written or not, which create expectations about other actors’ behaviour. Ostrom (2010: 647) defines rules-in-use as a ‘common understanding of those involved related to who must, must not, or may take which actions affecting others subject to sanctions’. This definition indicates that rules-in-use not only govern the actions of ‘those involved’ but also determine who are relevant actors in the first place. These rules-in-use may or may not correspond to formally written rules, referred to as rules-on-paper. Rules-on-paper and rules-in-use overlap when rules-on-paper are known and implemented, becoming rules-in-use. But some rules-on-paper are either not known or actively resisted, and hence do not translate into rules-in-use. Conversely, some rules-in-use have no formal basis in rules-on-paper; we reserve the term informal rules to designate rules-in-use that diverge from rules-on-paper.
Ultimately, policy effects are determined by rules-in-use, not rules-on-paper. To design institutions for better outcomes, it is necessary to explore when and how changes in rules-on-paper translate into changes in rules-in-use.
To provide a stronger focus to our causal investigation, we centre our analysis on two specific features of security governance we observed in Janakpur in 2019. First, some ward chairs were able to exercise strong leverage over local police stations despite having no constitutional prerogative to do so. Second, the provincial government had not gained significant influence on security provision even though the constitution provided it with both exclusive and concurrent powers over peace and order. Instead, as for the past decades, the actor most influential over security governance in Janakpur remained the Chief District Officer, in close collaboration with the Superintendent of the Nepal Police. The Chief District Officer chaired the District Security Coordination Committee, which brought together the Nepal Police, Armed Police Force, Nepal Army, and National Investigation Department (Nepal’s intelligence service). As such, he provided instructions to the Nepal Police and Armed Police Force. Beyond formal policing actors, the Chief District Officer also exerted (arguably weak) influence over the deployment of youth clubs as security auxiliaries during major festivals. Added to his direct reporting line to the Home Ministry, this means that the Chief District Officer was able to bypass the provincial government to act as a central node in the security governance system.
The rules-in-use had changed following the adoption of the 2015 constitution, but not in the ways the rules-on-paper would have suggested. These two distortions, one showcasing a change in rules-in-use not premised on the rules-on-paper, the other a change in rules-on-paper that did not affect rules-in-use, are the outcome of interest in our causal investigation. What hypotheses can we formulate that would simultaneously explain both outcomes?
Ostrom (2010: 647) considers two pathways for change for rules-in-use: following a change in rules-on-paper, or through emergent interactional effects. These two processes need not be viewed as distinct, however. If informal rules, norms and strategies can counteract or divert the effect of formal rules, they can do the same when formal rules are changed. Streeck and Thelen (2005: 10) argue that when analysing institutional change, formal, legal-political institutions must be distinguished from purely informal ones, such as customs, as both do not change in the same way. We disagree, finding value in considering both together as part of a complex process of institutional change.
Next, we turn to the literature on institutional change to identify a broad spectrum of hypothetical explanations that would explain the change in (and resilience of) rules-in-use observed in Janakpur, also accounting for the change in rules-on-paper. In this vast body of literature, we focus on explanations of micro-level institutional change (Kaufman 2017) affecting the way an institution functions—rather than the policy or formal rules that underpin it—where this institution is a multi-tiered security governance system. We group these in two broad categories: first, explanations emphasising change in the distribution of resources underpinning agency, such as financial resources, decision-making authority, knowledge, or connections, and second, explanations emphasising changes in ideational factors, such as beliefs, values and norms.
Resources and agency
The first family of explanations attributes a change in rules-in-use to a shift in the distribution of resources that underpin agency. Power explanations are often applied in policy studies to explain policy change, particularly at critical junctures (Mahoney 2015): actors leverage their resources, individually or within coalitions, to push through or resist a change in rules-on-paper. Applied to a multi-tiered governance system, such analytical explanations of change draw attention not only to the horizontal connections between actors bargaining around a policy change, but also to the vertical connections between tiers of governance.
New rules-on-paper are here significant in their own right: they may redistribute resources such as authority, fiscal revenues, and connections. However, new rules-on-paper do not wholly determine how rules-in-use will change. First, the same process of conflict and bargaining extends to how formal rules are applied once adopted. Second, resources are not equivalent: authority and financial resources may, for example, be in the hands of different tiers of governance, with control over funding trumping authority (Falleti 2005; Jones and Lister 2019). Together, these elements shape the new rules-in-use, determining whether old institutions will be wholly displaced, more subtly converted, or whether new institutions are merely layered over old ones (Mahoney and Thelen 2010).
Ideational factors
A second family of explanations attributes a change in rules-in-use to a change in underlying beliefs and values, which leads to the emergence of new norms. Here, norms are understood as shared expectations about how, for whom, and by whom security should be provided and governed. Norms are not necessarily uniform across a multi-tiered governance system, which can lead to tensions and conflict across tiers of governance (Agboga 2021; Madeira and Rodrigues 2015). A shift in norms will reorient actors’ behaviour, habits, and thus the rules-in-use.
Explanations differ as to what triggers the initial change in values and beliefs. Some studies of changes in security governance emphasise exogenous factors, such as urbanisation leading to increased awareness of security concerns in Belgian and Dutch cities (Devroe 2013), or the 9/11 attack in the United States, which motivated a decentralisation of security responsibilities to cities in Canada (Hataley and Leuprecht 2014). However, a normative shift can also stem from endogenous processes, where values and beliefs are reoriented out of repeated interactions, learning, deliberate advocacy, or the gradual replacement of political elites (Culpepper 2005; Pierson 2003).
Where does a change in rules-on-paper fit with this explanation? From an ideational perspective, it would be a manifestation of a deeper trend, where political actors close the gap between rules-in-use and rules-on-paper. Taken to the extreme, this argument would consider the new rules-on-paper entirely incidental to the underlying normative change, with no distinct causal effect. However, a change in rules-on-paper can influence ideational factors by endowing particular actors and actions with increased legitimacy.
Where we consider norms to be general rules and expectations about security practices, ideational factors also feed into the legitimacy attached to particular security actors. In a seminal article, Suchman (1995: 574) defines legitimacy as ‘a generalized perception or assumption that the actions of an entity are desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs, and definitions.’ Levi et al (2009) consider that the legitimacy of government structures can be elicited by the representativity and accountability of government actors and institutions, and their (perceived) effectiveness. While legitimacy is generally invoked as a self-reinforcing mechanism explaining the continuity of institutions (Mahoney 2000), legitimacy can also be shaken up by a change in representativity, accountability or effectiveness of actors and institutions. Furthermore, as with norms, the legitimacy of specific actors may not be uniform across a multi-tiered governance system; actors can be legitimate in the eyes of certain social constituencies but not others (Connell et al 2022).
Ideational and resource-based explanations are not mutually exclusive: part of the complexity of the translation of rules-on-paper into rules-in-use results from their combination (Capoccia 2015). For example, legitimacy is a type of belief attached to actors which they can also draw on as a resource; in multi-tiered governance systems, this may give a comparative advantage to lower tiers of government if they can draw on added democratic legitimacy (Germain 2012; Jacot-Descombes and Niklaus 2016).
In the next section, we probe causal mechanisms explaining the change in rules-in-use governing security that we observed in Nepal. Where other studies focus on a single tier of governance (Harguindéguy and Cole 2017; Toubeau 2017), we disentangle the factors that were at play across multiple tiers of governance, restricting, however, our exploration to a single policy domain: the governance of security.
The complex journey of institutional change in Nepal
New ideas are brewing
The 2015 constitutional change in Nepal rode on two broad waves of normative change: one nationwide pushing for a republican form of government with accountable security forces, and one more specifically in the Madhes demanding regional autonomy.
Until 2006, the monarchy in Nepal was deeply associated with a police force (and army) at the service of the king more than the people (Quinney 2011). In the districts, this authority was delegated to an appointed magistrate — first the bada hakim, then, after 1965, the Chief District Officer — who represented the central government in his area (Levi 1952; Scholz 1977).
However, governance norms came under increasing challenge in the mid-1990s as a radical fringe of the Communist opposition advocated for armed revolution to overthrow the monarchy. The support that the Maoist movement gradually garnered is evidence of mounting unhappiness with (and hence de-legitimation of) a mode of governance based on a centralised monarchy. The confrontation also undermined the stance of state security forces. The task of quelling the insurgency was left to the police, who became a prime target of insurgents and used increasingly ruthless tactics (Khanal 1998; Dhungana 2007). Nevertheless, security sector reform was never discussed in the public arena until the signature of the Comprehensive Peace Agreement in November 2006; the few who dared suggest the need to reform security institutions took personal security risks, as such suggestions could be construed as criticism of the Royal Nepal Army in a time of war (Pandey 2009; Upreti and Vanhoutte 2009).
Things changed in 2006 when this groundswell of discontent found its way into the discourse and agenda of the mainstream political parties. In April 2006, people across Nepal heeded the call of political parties to take to the streets and demand an end to the monarchy in what became known as the second Jana andolan (Hachhethu 2008; Hachhethu and Gellner 2010). This popular uprising constituted a turning point in the political awareness of the masses (Hachhethu 2008: 28). Previously held values, beliefs and norms were rapidly eroded through awareness and mobilisation. The main target of this change was the palace: the king relented by reinstating parliament, and in May 2006, the House of Representatives removed the title ‘royal’ from the armed forces and all other government bodies.
The new governance aspirations also extended to security governance. Evidence of a shift in public opinion is found in several large-scale surveys of political attitudes administered by non-government organisations between 2007 and 2010 (Hachhethu 2008; Hiscock and Benton 2008; Hiscock and Sharma 2009; Cochran-Budhathoki 2011). The Nepalese population agreed that the Nepal Police should be the primary law enforcement agency (Cochran-Budhathoki 2011: 5). Nevertheless, they were hungry for changes in policing practices, questioning how policing was hitherto controlled, directed and influenced. The population called for more civilian oversight of security providers and questioned the influence of political parties — and the government — on the Nepal Police. The politicisation of the police was widely felt to be the primary impediment to police effectiveness. Within police ranks, there was an appetite for more significant ties to local governments, at least for information exchange (Cochran-Budhathoki 2011: 71). This is not to say, however, that a new norm had already taken shape. Rather, the April 2006 uprising represents a turning point that opened up possibilities of change, questioning and searching for new, better arrangements. It was a necessary condition for institutional change, but it did not, in itself, determine the direction of change.
In the Tarai the premise of a nationwide debate on security sector reform combined with an additional concern: the politicisation of the Madhesi identity and the attendant demand for a federal system of government. A degree of regional autonomy has been a demand in the Tarai since the 1950s and certainly since the return of multipartyism in Nepal in 1990 (Karna 2018; Sijapati 2013). These demands found no echo in the 1991 constitution, but gradually, and perhaps opportunistically, demands for autonomy started permeating the agenda of mainstream political formations (Gyawali 2018).
The inhabitants of the Tarai were too diverse culturally, religiously and linguistically to easily fit under a common agenda, and the term ‘Madhesi’ itself was contested (Hachhethu 2007). This, however, started to change. Between August 2004 and March 2007, the proportion of Madhesis (excluding Muslims) who identified with a regional/ethnic identity rose from 28 to 47 per cent, while those identifying with a Nepalese national identity fell from 53 to 20 per cent (Hachhethu 2008: 77). The same trend was also observed for other ethnic identities, albeit to a lesser extent.
A federalist norm was gaining ground in the Tarai. Minorities saw federalism as the answer to a persistent problem of discrimination, of indigenous people and ‘backward’ castes, but also crucially, here, of the Madhesi people. After the Interim Constitution was adopted in January 2007, a second popular movement immediately arose from the Tarai, calling on political parties to respect earlier commitments to federalism. The Madhes andolan was described as ‘the strongest, most violent, and most effective set of street protests Nepal has seen’ (Hachhethu and Gellner 2010: 143). What was at stake was not just the democratic nature of the political system, however imperfect it had been, but also its centralisation.
Madhesis were successful in articulating their preference as a political agenda. The Madhes andolan forced an amendment to the transitional constitution in April 2007, with a new article 138 conceding that ‘[t]here shall be made progressive restructuring of the State with inclusive, democratic federal system of governance, by doing away with the centralised and unitary structure of the State so as to end discrimination based on class, caste, language, gender, culture, religion and region’ (Karna 2018: 9). As a territorially concentrated minority, Madhesis wanted to be governed by a regional government in their image; this government would be autonomous and enjoy full authority. Were their demands sketching a new norm on how the Tarai should be governed? The movement certainly succeeded in articulating a political agenda that helped shape the new constitutional order, but this norm, if a norm it already was, was far from shared uniformly across Nepal, as would be demonstrated by the Constituent assemblies.
Contested rules
The new constitution was negotiated by two successive Constituent assemblies (2008–2012 and 2013–2015) (Lecours 2014; Edrisinha 2015; Karna 2018). One of the most contentious issues, once federalism had been agreed to, was whether or not provinces would be allowed to mobilise security forces (The Kathmandu Post 2012). This reflected a deep-seated fear among Nepalese intellectuals and political elites that the Tarai might wish to secede (Gellner 1997: 28). The text that was eventually endorsed introduced significant changes on paper. It created new provincial governments and entrusted them with managing the provincial police as a distinct policing body alongside the Nepal Police. The constitution also stipulated provinces’ shared responsibility for law and order with the federal government. Nevertheless, these rules-on-paper were neither as fixed nor as prescriptive as would appear.
Indeed, the constitution’s wording could mean vastly different things depending on the rank up to which the provincial police could ascend and how responsibilities for law and order were shared. Even more confusing, the constitution allocated exclusive rights over peace and order to the provinces and exclusive rights over peace and security to the federal government. The ambiguous wording of the constitution allowed for both interpretations to take hold. In addition, by making peace and order a concurrent power of federal and provincial governments, the 2015 constitution gave the federal government a lever to override the provinces. Rules-on-paper were not simply an articulation of uncontested norms, as a hypothetical ideational explanation would have it, but reflected power struggles and compromises.
In effect, rules-on-paper continued to be contested and negotiated after the constitution was promulgated in September 2015. The Nepal Police headquarters had always considered that the task of supervising the provincial police should be devolved to the Nepal Police (Himalayan Times 2015; 2017). From 2016 to 2019, the Prime Minister and Home Minister used their executive powers to impose such a conservative interpretation of the new constitutional rules, one that would anchor the role of the Chief District Officer and the Nepal Police to the detriment of the provincial governments (My Republica 2016b; The Kathmandu Post 2016a; 2016b; Sapkota 2018). Finally, in January 2019, the Home Ministry tabled the long-awaited draft Federal Police Bill, based on one prepared by the Nepal Police headquarters. This proposed that, while provincial governments would be granted an oversight function, the provincial police would be deployed and controlled by the Chief District Officers appointed by the Home Ministry (The Kathmandu Post 2019; Kamat 2019b). A separate federal bill on internal security management tabled at the same time sought to entrench the role of the Chief District Officer and the Home Ministry. The 2019 draft provided Chief District Officers with direct supervision of the Superintendents of the Nepal Police and Armed Police Force and oversight over the provincial police (My Republica 2019a; Kamat 2019a).
The central role that Chief District Officers continued to play was incongruous. The new constitutional rules only referred to district-level authorities in two respects: the judicial system with its district courts, and district assemblies. The latter were seen as a type of ‘local unit’ whose role was limited to the election of the District Coordination Committee — itself tasked with coordinating municipalities — and to monitoring development projects and road construction. The resilience of the Chief District Officer can partly be explained by the control that the Home Ministry was able to exercise until 2018 over legislative developments. As the fate of Chief District Officers had not been explicitly addressed in the constitution, institutional change could proceed by layering new actors and channels of authority over an existing chain of command, rather than displacing old institutions with new ones.
However, by 2019, the playing field had changed. Provincial and federal elections had taken place as a direct outcome of the new constitution, and the federal parliament became an important arena for negotiating security governance. Pieces of legislation on the Nepal Police and provincial police, police adjustment, and peace and security were the objects of dozens of amendments by parliamentarians supporting meaningful federalism (Himalayan Times 2019b; Pradhan 2019; My Republica 2019c).3 Rules-on-paper were bitterly negotiated, during parliamentary debates or indirectly via media statements and coalition building, and with politicians activating, or threatening to activate, the Supreme Court. For the first time, the role of the Chief District Officers came under intense public scrutiny (Yadav 2019; Shrestha 2019; Dahal 2019).
From words to action
From then on, the reactive sequence also unfolded on a parallel track to the policy and legislative arena. To implement its constitutional prerogatives and turn the rules-on-paper into rules-in-use, the provincial government needed the federal government to devolve human and financial resources. This provided centralising forces with another avenue to resist change.
The provincial government in Province 2 was able to fill some key positions to exert its security prerogatives by, for example, recruiting a senior security advisor (My Republica 2019d). However, it did not have control over the civil servants deployed by the federal government, who could be (and were) transferred away from the province without consulting the provincial government (My Republica 2018). The adjustment of existing civil servants to the federal, provincial and local levels was completed only in late March 2019, 15 months after the provincial governments had been constituted (My Republica 2019b).
The situation of the police officers that would constitute the new provincial police was just as complex. The adjustment of Nepal Police officers to the provincial police was initially delayed by a lack of legal basis (Himalayan Times 2019a). But by December 2022, almost two years after the law on police adjustment had been passed, police personnel had still not been transferred to the provinces (The Rising Nepal 2022). By now, commentators considered it evidence of a deliberate delaying tactic by the federal government.
The lack of resources undoubtedly hampered the ability of the provincial government to deliver fully on its responsibilities under the 2015 constitution. This was particularly the case in regard to its exclusive right to govern the provincial police: this right would be moot if there were no provincial police to govern. However, creating the provincial police was only one part of the change in rules-on-paper: the provincial government also had broader responsibilities over peace and order. For example, the provincial government could have held the Nepal Police to account at the Provincial Security Coordination Committee or through the provincial Chief Attorney’s mandate to monitor human rights abuses, which also involves monitoring police stations. Delays and obstruction by federal bodies partly diluted the influence on policing that the provincial government acquired on paper, but could the provincial government have exerted influence regardless of its limited resources?
As already noted, the continued existence of strong Chief District Officers with extensive security powers alongside the provincial governments was incongruous. The provincial Chief Attorney regretted that the Chief District Officer and the Nepal Police still operated according to the old system.4 That said, we found no indication in our interviews, focus group discussions and newspaper review that the provincial government sought to alter this status quo on the ground. The provincial government was an important player at the federal level, but it only approached the Chief District Officer to manage the security of its compound in Janakpur and its officials when they were travelling. At that point, the provincial government’s focus on discussions held at the federal level over rules-on-paper, rather than the application of rules-in-use at the district level and below, allowed the Chief District Officer and the Superintendent of the Nepal Police to continue their privileged relationship, ignoring the provincial government.
Conversely, we found no evidence of an effort on the part of district-level actors to actively undermine the provincial government. To be sure, the Chief District Officer did not attend the meetings of the Provincial Security Coordination Committee, justifying his absence at these meetings by the fact that he represented the federal government. This could be expected, as the district is a smaller administrative unit than the province.5 But the justification is also further evidence that institutional change had proceeded through layering rather than displacement (Mahoney and Thelen 2010). In our interview, the Chief District Officer justified his absence at the Provincial Security Coordination Committee by his role as representative of the federal government, not by the fact that the district is a smaller administrative unit than the province. In other words, he answered to the federal government, not the province. As the position of Chief District Officer had not been explicitly challenged by the new rules-on-paper, the new constitution contained the possibility of both continuity and change, and ward-level dynamics were eventually decisive in shaping rules-in-use.
Local pragmatism
Viewed from Janakpur in early 2019, security governance had changed. Some ward chairs had gained substantial leverage over local police stations, influencing where and when the police would be conducting patrols. This was not universal — some ward chairs responded that they had had no contact with the Nepal Police — but represented a significant change since the 2017 elections. Intriguingly, this was not a change guided by the new legal framework, which gave wards only responsibilities for the delivery of local services.6
The change was largely nominal at the municipal level: after 2017, the Mayor took over from the Executive Officer, a civil servant who was managing the municipality since the mandate of the last elected officials had expired in 2002. The new rules-on-paper gave the Deputy Mayor — a woman — authority over dispute resolution, enforced by the municipal police, which the role previously did not entail.
But what mostly stood out was the continued pre-eminence of the Chief District Officer in the governance of security, in close collaboration with the Superintendent of the Nepal Police. This is intriguing not only because, formally, the 2015 constitution no longer recognised the district as a tier of government, but also because, historically and symbolically, the Chief District Officer was the most visible representative of the federal government in Janakpur. We noted above that the Chief District Officers, and before them the bada hakim, historically represented the authority of the king and, by extension, of Pahadi rule, in the districts. Could the continued relevance of the Chief District Officer in Janakpur indicate that the broad normative changes we outlined above had not reached Janakpur?
In reality, Janakpur actively participated in the popular movements that had driven these transformations. The people of Janakpur made their voices heard in 2006 to demand an end to the monarchy (Navlakha and Varma 2006). The mobilisation was particularly strong during the Madhes andolan a few months later (Hachhethu 2007). ‘We participated in the movement for autonomy to take the reins of governance into our own hands’, said a ward representative; ‘We didn’t want to be ruled by the Pahadis.’7 The people of Janakpur participated in the local consultations that took place during the first Constituent Assembly (Himalayan Times 2011), and demonstrations in town were fierce between 2012 and 2015, during the second Constituent Assembly and then around the promulgation of the constitution (My Republica 2015; Nepali Times 2015).
In interviews and focus group discussions, ward representatives concurred that security was the responsibility of the local and provincial government. They recognised the legitimacy of the provincial government, yet they ignored it when it came to security governance:
We have been functioning in the old way… sitting together and making decisions… the government, constitution, social context etc. have changed, but we still adopt the old policies. Complaints are being registered at the ward, but we are still dependent on the district-level authorities for dispute resolution.8
The above quote suggests that norms had not changed at the same pace as the rules-on-paper, and old habits died hard. Madhesi protests brought in federalism, local elections were held, autonomous municipal and provincial governments were constituted, but ward representatives still relied on the Chief District Officer and the Nepal Police district office for their security needs.
The ward representatives’ discourse also betrays a dissonance between a general condemnation of the Pahadi domination of instances of governance and the continued legitimacy of the individuals who represented this central government in town. One male ward representative denounced an administration controlled by the central government, including the Chief District Officer, and insisted that people’s concerns would only be heard once the provincial government got real power. However, the same representative later praised the Chief District Officer, saying he was ‘doing a good job’.9
From our interactions, we could sense that the ideological outrage manifested in Janakpur in public protests around 2007–2008 and between 2012 and 2015 had subsided since local elections had been held. Fierce debates around the future of the provincial police and the role of the Chief District Officer continued to be reported in national newspapers in 2019, in which the Province 2 government played a prominent role, yet in Janakpur these were mentioned only once by a male ward representative during the focus group discussions. The drivers of institutional change clearly differed across tiers of governance from the national to the local level.
Instead, our data suggests that, at the ward level, the Chief District Officer and the Nepal Police district office were seen pragmatically as the more effective intermediaries. The ward representative quoted above said they ‘depended on’ district-level authorities. Relying on the Chief District Officer was, therefore, not the result of a norm per se but rather a rational calculation of utility. Such effectiveness is also a source of legitimacy in governing security, despite the Chief District Officer’s lack of representativity and democratic accountability.
Effectiveness also underpins the relationship between ward committees and local police stations. During focus group discussions, women’s ward representatives decried ‘the administration’ for protecting criminals.10 When asked to whom they were referring as the ‘administration’, they mentioned the Chief District Officer and the Superintendent of the Nepal Police. However, others expressed their satisfaction with the work of the Nepal Police. ‘At present, the police has become quite vigilant,’ said one ward representative.11 She recounted how she had recently observed the police controlling young boys and girls travelling in border areas without a guardian because of the rise in elopement cases. ‘I liked it, ’ she concluded.
The perceived effectiveness of the police can explain why ward committee members turned to the police with their security concerns. They also had another reason to do so. ‘Ma’am, this is Madhesh, ’ said a women’s ward representative. ‘People do not listen to local representatives; they obey the police … administration … Unless we join hands with the police, people do not listen to local leaders like us. They say we exist because of their votes and would not obey our instructions.’12 Another women’s representative described a tenancy dispute where she had had to intervene: ‘The owner questioned my authority … that I, a local leader, cannot intervene. I then told him I would report to the police. I reported to [name of police station]. Later police personnel came and understood my point.’13
Elected ward representatives enjoyed some authority in the communities despite having only been in office for two years, as exemplified by their role in dispute mediation. Many representatives reported how the community congratulated them after their election. However, some felt that this initial respect had waned. ‘Yes, I am an elected member, ’ said one. ‘However, people complain that I cannot help maintain cleanliness. What am I good for? They say I was more efficient before being an elected member. How do I respond?.’ The sense emanating from their remarks is that these ward representatives benefitted from the initial capital of democratic legitimacy. However, this legitimacy was still fragile, dependent on their ability to deliver concrete services to their constituents. They had representativity and accountability, but they lacked effectiveness. Joining forces with the police enabled them to combine legitimacies and boost their authority.
In addition, the police had another asset: it could wield coercive power that acted as a deterrent. A male ward representative explained that security was not a problem in his ward due to the presence of several offices of security agencies.14 Ward representatives felt protected by their association with the police: ‘It is not possible for us to function without the police administration’, said a male ward representative, recounting how a drug abuser threatened him.15
Ward chairs were subverting the rules-on-paper when they exerted influence over local police stations, given that the 2015 constitution did not give them a role in security governance. However, rather than a deliberate subversion of formal rules, our evidence suggests that ward chairs generally only had a vague idea of their legal mandate. Most ward chairs and representatives lamented their lack of knowledge, referring both to a lack of general education and of understanding of specific laws and procedures, which left them disempowered and exposed to manipulation. One (educated) ward chair contended that this lack of experience explained the passivity of ward chairs when it came to security governance: he considered his peers could and should do more. However, ward chairs could also more easily ignore their lack of formal authority over security governance if they poorly understood formal rules in the first place. Only one ward chair acknowledged their lack of legal mandate regarding security; this did not stop him from approaching the local police station to request patrols in particular areas in his ward. Elinor Ostrom introduced the concept of rules-in-use as a recognition that rules-on-paper that are not known have no traction (Ostrom 2005: 138). The ward chairs’ influence over Nepal Police stations is a case in point.
While this explains why ward committees would seek out the Nepal Police, it does not explain why the Nepal Police would heed their requests. Had the Nepal Police changed its views on who could legitimately give it instructions? The openness that local police stations displayed toward engaging with elected local authorities — perhaps more so with the men than with the women — contrasts with the Nepal Police headquarters’ efforts to minimise change, indicating that multi-tiered dynamics were also at play within the Nepal Police structure.
Gaining influence over local police stations required a shared understanding and expectation with local Nepal Police officers that ward chairs could legitimately govern security. Indeed, all the Nepal Police officers we interviewed spoke positively about their relationship with ward chairs and representatives. However, they framed ward representatives’ contribution as information sharing, not governance and influence. They mostly referred to the disputes that ward chairs helped handle. They also indicated that they were enlisting the ward committees’ help in cases of domestic violence or to obtain information on ‘cases’. One police respondent ascribed their improved relationship with the wards since the advent of federalism to the diverse representation on ward committees, including women and minorities.16 This suggests that theirs was something more than a functional complementarity. If ward committee members sought the collaboration of the police to benefit from their ability to get things done and their power of deterrence, the Nepal Police equally sought the collaboration of ward committees to benefit from their democratic legitimacy and their ties to the local communities.
Conclusion: reverse engineering institutional change
An important question for the literature on institutional change is whether change can be engineered. If a change in the ways that institutions actually work — in the rules-in-use — was purely emergent, there would be little hope of designing policy interventions to make security more inclusive. This case study of institutional change in the security domain shows that in Nepal, the new rules-on-paper had a traceable impact on rules-in-use. The increased influence of ward-level actors suggests that security became more inclusive as a result of the constitutional change. However, reverse engineering this change also shows the complexity of the relationship between rules-on-paper and rules-in-use, raising points for consideration by policymakers and practitioners involved in reforming security sector governance.
First, when considering which formal rules had the most impact on making security more inclusive, the case study highlights the need to expand our focus beyond specific rules about security governance to include rules with an effect on security governance.17 What was critical in Janakpur were not the specific rules about security governance, which mainly sought to empower the provincial governments, but rather the resurrection of ward-level (and, to a lesser degree, municipal) elected authorities. Local security governance might be more about local governance than security sector governance, an important reminder in the current drive to localise security sector reform.
This cross-section of institutional change in security governance also highlighted the different forces at play across tiers of governance when negotiating the implementation of the new rules. In higher (federal and provincial) tiers of governance, actors focused on the letter of the rules, leveraging their resources and using their agency to orient their interpretation, and push for or hamper their implementation. The change in formal rules was not a single-point stimulus on the security system but rather a continuous process of contestation, akin to Mahoney’s (2000) reactive sequence. Meanwhile, in lower tiers (wards, municipality and district), ideational factors and pragmatic considerations came to the fore. Significantly, the boundary between higher- and lower-tier dynamics is not fixed. In Janakpur, in the period under consideration, the provincial government focused its efforts on the contestation and negotiation of formal rules, even though nothing prevented it from already assuming a greater role in governing policing. This points to the importance of the strategic choices that actors make not only between competing policy development priorities and approaches, but also between policy development and early implementation.
The vertical dimension of multi-tiered governance also invites a differentiated analysis of norms, values and beliefs across tiers of governance. Striking in Janakpur was the apparent contradiction between the idea prevailing in the Madhes that the region ought to be autonomous, which can be directly related to the introduction of federalism in the new constitution, and the importance that ward representatives nevertheless continued to afford the primary symbol of the central (federal) government in Janakpur: the Chief District Officer. This, we showed, was due to habits, which help preserve the influence of long-established security actors in processes of institutional change. It was also related to the different facets of legitimacy, premised on both the representativity and accountability of government institutions, and their effectiveness.
This finding offers an important complement to Levi et al’s (2009) examination of the legitimacy of government institutions. Their influential essay assumes formal authority, but this study shows that the relationship between formal authority and legitimacy is complex, requiring a distinction between democratic legitimacy, which may be enhanced by broad formal authority, and formal powers in the specific policy area under consideration. Neither is sufficient to ensure influence on security governance, but adequate consideration of pre-existing habits, democratic legitimacy and effectiveness in a particular policy domain may provide important pointers as to how formal rules could be crafted to redistribute influence.
Would the effect on rules-in-use have been more pronounced in Janakpur if the rules-on-paper had explicitly restricted the role of the Chief District Officers in the new security governance architecture? Counterfactual arguments are an important tool in process tracing, but not so easily applied. It is difficult to claim that the Constituent assembly could have agreed to abolish the Chief District Officers, or indeed that this was a matter to be addressed in the constitution. But it does feel significant that the future role of the Chief District Officers only started to be debated in Nepal around 2019, a decade after the beginning of the Constituent assembly, when these had been such pivotal players in the governance of security in the previous constitutional order. If formal rules can call new institutions into being, it is worth exploring whether they should also more deliberately address the fate of existing institutions, rather than assuming that these will naturally make way in a new constitutional order.
This case study has limitations. First, the complexity of institutional change limits the generalisability of findings. In addition, our data was collected at a time when formal rules guiding security governance were still being contested and shaped, and a significant re-organisation of rules-in-use was still possible. We cannot exclude the possibility that the two outcomes on which we chose to focus our analysis, namely the influence gained by some ward chairs, and the continuing pivotal role of the Chief District Officers in contrast to the provincial government, could yet change. Nevertheless, our study does identify features of a multi-tiered governance system whose ex-ante analysis, we contend, can and should guide the design of reform of a security governance architecture.
Future research could further unpack the causal significance of the elements identified in this study. The mediating role of formal rules on local security actors’ legitimacy, in particular, deserves closer examination. Legitimacy is relational: the attributes of the governance actors holding this legitimacy are as important as those of the actors accepting to comply with them. Consider the leverage that some ward chairs acquired on local police stations, despite their lack of formal authority to govern policing: they were able to use their democratic legitimacy as a resource, just as others used their financial or material resources, because this legitimacy was sought after by local sub-inspectors of police. Is the resulting influence equally true of all dyads of actors, or does the democratic legitimacy of the ‘influencer’ matter more for appointed state officials such as police officers? In addition, the local governance of security also features a gender dimension that needs to be further investigated. Future research could usefully examine whether and how a change in the participation of women in local governance influences the way security is governed.
Notes
[1] The Province received its permanent name of Madhes in January 2022. We keep its old name of Province 2 in this article as this was still used at the time of our data collection.
[2] All major Nepalese newspapers — the Kathmandu Post, Himalayan Times, My Republica and Rising Nepal — had extensive English editions available online. They also had online archives spanning back, in the case of the Kathmandu Post, to the years of the Madhes andolan in 2007–2008.
[3] The Nepal Police and Provincial Police (Operation, Coordination, Supervision) Act was eventually adopted in February 2020. This re-directed change to an interpretation more friendly to the provincial government’s view.
[5] Province 2 included eight districts, each with their own Chief District Officer; Chief District Officers did not attend meetings of the predecessor Regional Security Coordination Committee.
[6] The Constitution makes no provisions regarding the competencies of the wards, but the Local Government Law adopted in 2016 makes ward offices centres of service delivery (My Republica 2016a). Officially, ward committees have no security prerogatives.
[17] We owe this insight to Gains and Lowndes (2022), who similarly consider ‘rules with gendered effect’ rather than on ‘rules about gender’ in their study of police governance.
Competing Interests
The authors have no competing interests to declare.