Judicial Reform – A Neglected Dimension of SSR in El Salvador

ABSTRACT Security Sector Reform is an integral part of peace-building but the focus of international actors tends to be on formal state security providers. This article argues that reforms in the judicial system are key for the non-violent transformation of societal conflicts. Based on historical institutionalism a theoretical argument links justice and peace. Reforms of the judiciary need to be an integral part of SSR because otherwise reforms in the military and the police can easily be undermined or turned back. A case study on El Salvador provides empirical insights on the interrelation between reforms of these institutions.

2009; Bowen 2013). In the context of transitions out of war, these reforms target security institutions that have been part of the armed conflict. Processes of demobilization, vetting and professionalization aim at changing the structure, the numbers and the behaviour of the personnel. The aim of SSRas well as of demobilization, demilitarization, and reintegration (DDR)is the prevention of war recurrence and the establishment of security. This is important, but peace and conflict research is increasingly aware of the need to move beyond the minimalist approaches of peace (Regan 2014;Wallensteen 2015;Diehl 2016). Broader concepts of peace need to include the reduction of other manifestations of violence such as state-repression and 'criminal' violence. Under such a broader perspective of peace-building, reforms in the police and the judiciary at least theoretically are even more important because these institutions play a major role in non-violent conflict transformation and the reduction, control and sanctioning of illegal violence. 3 This article aims to contribute to the debate on post-war peace-building by examining the interaction between the reforms in core state security institutions (military and police) on the one hand and the judiciary on the other hand. Based on a case study on El Salvador the main argument developed here is that without reforms in the judiciary, reforms in the police and military are in danger of being undermined, instrumentalized or turned back. El Salvador is a critical case for post-war SSR reform for several reasons. First of all, El Salvador is considered one of the few 'success' stories of liberal peace-building (Karl 1992;Call 2007). Nevertheless, regarding homicide rates, El Salvador remains one of the most violent countries in the world (UNODC 2014). Most of the literature on post-war violence in El Salvador focuses on crime and youth gangs (Cruz 2010;Bruneau, Dammert, and Skinner 2011). The analysis of security sector and judicial reform is important to contextualize and explain how the dynamic of violence and its reproduction developed. Furthermore, El Salvador is deemed the blueprint case for the United Nation's liberal peace-building agenda: Alvaro de Soto served as a UN mediator in the Salvadorian peace process (1990)(1991) and then drafted the United Nation's Agenda for Peace (Boutros-Ghali 1992) probably based on his experience in the case.
Methodologically, I will use process tracing as an 'analytical tool for drawing descriptive and causal inferences' (Collier 2011, 824) triangulating three sources of data: academic literature, official documents and reports of national and international organizations (ONUSAL, human rights organizations) as well as semi-structured interviews. 4 I analyse the interaction between security and judiciary reforms in El Salvador at two points in time following Collier's (2011) assumption that process tracing should analyse causal processes via the observation and description of sequences using snapshots at specific moments. As a historical starting point, I first outline the role of security and judiciary institutions until the end of the war in 1992. Afterwards, the analysis will focus on two theoretical triggers or potential windows of opportunity for institutional reform: the comprehensive peace agreement and its implementation between 1993 and 1997 and the change in government when the candidate of the former guerrilla FMLN (Frente Farabundo Martí de Liberación Nacional) was elected president in 2009. The case study provides evidence that a lack of elite accountability and the development of elite accommodation undermines reforms and supports the path-dependent reproduction of public security approaches. Although El Salvador has not experienced war-recurrence since 1992 the high levels of interpersonal violence and increasing state repression of the alleged perpetrators (Cruz 2011;Aguilar 2016;INCIDE 2016) provide evidence for the limitations and deficits of more than 25 years of so-called peace-building.
The article proceeds as follows: The first section introduces the theoretical argument based on historical institutionalism and its relevance for institutional reform in the transition out of war under a liberal peace-building perspective (Capoccia 2016;Fioretos, Falleti, and Sheingate 2016;Ansorg and Kurtenbach 2017a). The analysis of reforms needs to include three crucial elements: the influence of contextual conditions, the importance of temporal processes and the different shapes of reforms. The second section investigates the interaction between reforms in the three core state institutions 5military, police and judiciarybased on a case study of El Salvador. The conclusion summarize the findings and discuss their relevance for future research.

A historical institutionalist approach to SSR
Liberal peace-building strategies after the end of the Cold War were based on the expectation that 'securing justice and human rights' would support 'social progress and better standards of life in larger freedom' (Boutros-Ghali 1992, 1). Along these lines, international cooperation emphasized the promotion of political and civil rights (democracy) during the 1990s (Paris 2004;Richmond 2006;Heathershaw 2008). The assumption underlying liberal peace-building is that democratization will enable societies to transform societal conflicts without the use of violence. Under this perspective, the security sector and judicial reforms are core elements aiming to strengthen or establish state institutions providing security and justice (Sedra 2017). In democratic systems, these institutions should be accessible to everybody without regard to the person as well as be bound to the rule of law (Luckham 2003).
The main critique of liberal peace-building rests on the argument that the reforms intend to reproduce a Western development model in societies living in very different cultural, historical and international contexts and as a consequence lack local ownership. 6 This produces conflicts, frictions, hybrid structures and grey zones characterized by 'neither war nor peace' contexts (Mac Ginty 2010; Richmond and Mitchell 2011;Berdal and Suhrke 2012). Post-war reforms do not happen in a political vacuum as there is rarely a new start from scratch. Despite high hopes that the end of war will bring fundamental change and open up space for a better future, empirical evidence shows that it is 'unlikely to see a clean break from violence to consent, from theft to production, from repression to democracy, or from impunity to accountability' (Keen 2000, 10). Even when fundamental reforms are envisioned in peace agreements, they interact with existing structures and behaviours shaped by history, culture and the experience of war and widespread violence (Wood 2008;Arjona 2014;Ansorg and Kurtenbach 2017b). Hence, the end of war is not necessarily a critical juncture where profound change happens in a short period of time (Capoccia 2016) but can also produce path-dependent developments or fortify existing structures. Historical institutionalism helps us to identify the relevant patterns of change and path-dependence. This is of utmost importance for the security and justice sector in post-war societies as these institutions were part of the war but are also important to provide security and reduce violence after its end as these are important markers in the transformation out of war.
Increasing the autonomy and independence of courts and human rights institutions is an important prerequisite for the control of military and police. Along these lines, this article follows the call of Jackson and Bakrania (2018, 21) for a non-linear approach recognizing 'that externally imposed liberal structures sit on top of the real underlying politics of states, rather than being neutral arbiters'. As a consequence, an analysis of the existing power relations is necessary, as these frame the conditions for reform and shape as well as affect the dimensions and outcome of reforms (Berg 2012). Thus, the analysis of security and judiciary sector reforms needs to take into account (a) the relation between these institutions and the political context because the latter shapes the degree of autonomy and/or control of the former; and (b) the nature and scope of a change as well as its contentious nature. This perspective links up with the main factors emphasized by historical institutionalism that embeds reform processes in the specific contextsincluding the 'social coalitions' (Hall 2016, 39-42) and that acknowledges that reform processes can take different shapes (gradual, incremental, or radical). 7 The interaction of different reform processes in the security sector and the judiciary is discussed if at all in the relation between peace and justice. Here two arguments come up: First, Schröder and Kode (2012) point out that the rationale underlying the support for reforms in the security institutions is strengthening their efficiency and capacity. Promotion of the rule of law and judiciary reforms on the other hand aim at the control and limitation of state capacities. Thus, there are conflicting goals that might produce deadlock on the ground. Second, there is a debate on the relevance of transitional justice and the rule of law reforms for peace (Buckley-Zistel et al. 2015;Vinjamuri and Snyder 2015;Sriram 2017). Theoretically transitional justice mechanisms can be seen as an important link between the violent past and the non-violent future. The way a society copes with the violence of the past may be an important indicator of the potential of change and existing power relations. This connection has rarely been made or investigated. 8 But what interactions between the different reforms might be relevant for the broader process of peace-building? Theoretically, reforms should introduce a clear distinction between the mandates of the military (external security) and the police (internal/public security). The judiciary should be responsible for sanctioning the illegal use of violence and monitor state institutions in the security sector and beyond. Independence and autonomy of the judiciary should promote the accountability of the police and the military to the rule of law. If implemented, these reforms are expected to contribute to a reduction in violence at least from the state institutions. Inversely, deficits in implementation might lead to the persistence of autonomy in the military and the police. To investigate the interaction under a perspective of historical institutionalism, the starting point is thus an analysis of the role of these institutions prior to the war termination. As reforms do not start from scratch, we need to know what the structure of the security and judiciary sector was before reforms began. This sort of benchmarking exercise anchors reforms in a specific historical process allowing for a comparison of the scope of changes envisioned and implemented. Questions to be answered are: Who controlled these institutions? What resources were available and where did they come from? Were these institutions part of an integrated system? Secondly, we need to identify windows of opportunity and options for reforms provided at the end of war through peace accords and their implementation. What reforms were envisioned? Were they implemented? Who were the actors driving or opposing these reforms?

Security sector and judicial reforms in El Salvador
Setting the stagewar, state repression and widespread violence Historically, El Salvador's security and justice system formed part of the ruling coalition between the economically powerful large landholders and the military. While there was a formal separation between the executive, the legislative and the judiciary, the latter was highly politicized and dependent (Ching 2014;Artiga-González 2015). The majority of the population was submitted to the elites by direct and indirect forms of violence such as repression but also vagrancy laws securing the availability of cheap labour on the coffee plantations. Until the 1970s, the most prominent event in the violent history of the country was the massacre of more than 30,000 protesting peasants in 1932 by the military, remembered as La Matanza. During the following decades, the military dominated the state administration and the government; mobilization for change and reform was answered by amplified violence (Stanley 1996;Holden 2004). Effective checks and balances between the executive, the legislative and the judiciary did not exist.
Protest and mobilization increased during the 1970s, leading to open war against the authoritarian and exclusionary regime from 1980 onwards. Until 1992 El Salvador experienced an internal war between five guerrilla groups united in the FMLN and a military government which was replaced by a civilian-led, but military-dominated 'revolutionary junta' later. 9 Although the war is mostly framed as a two-sided confrontation between the Salvadorian state and the FMLN, a variety of paramilitary forces fought alongside state institutions against the guerrilla groups and their real or perceived civil society supporters. The FMLN was the strongest guerrilla force in Latin America and would have won the war had the United States not significantly supported the government financially and militarily (Peceny and Stanley 2010).
Regional conflict dynamics influenced the course of the Salvadorian war. In neighbouring Nicaragua, the Sandinista National Liberation Front (Frente Sandinista de Liberación Nacional, FSLN) ousted the dictator Anastasio Somoza in 1979. With the election of Ronald Reagan as president in 1980, the United States pursued a strategy of contention and roll-back by financing the armed opposition against the Sandinista regime in Nicaragua as well as the Salvadorian government and its armed forces to prevent a second opposition victory (LaFeber 1993). However, an important element in this strategy was a limited set of reforms such as regular elections to provide evidence that the US was supporting democracy (against the revolutionary government in Nicaragua). In 1984, Christian Democrat Napoleón Duarte became Salvadorian president after elections with a very limited spectrum of political options. While promising reforms, he was clearly controlled by the military and the United States (LaFeber 1993;Montgomery 1995).
The armed forces, the military police, and paramilitary death squads were the main perpetrators of violence. Extra-legal executions, torture, massacres, and letting people 'disappear', mostly in the rural areas, were widespread (Peceny and Stanley 2010). Besides the military and the police, the judicial system was complicit in the bloodshed against civilians failing to sanction the illegal use of violence by state institutions and their paramilitary partners. The Truth Commission later summarized the co-responsibility of the judiciary: The inability of the courts to apply the law to acts of violence committed under the direct or indirect cover of the public authorities is part and parcel of the situation in which those acts took place and is inseparable from them. (Betancur, Planchart, and Buergenthal 1993, 168) Reports of human rights organizations such as Amnesty International and Human Rights Watch and even of the US State Department provide additional evidence for the state system of violence and the involvement of the judiciary. 10 According to the Truth Commission, government forces and their allies were responsible for more than 90 per cent of human rights violations. But the FMLN also committed serious human rights abuses such as murdering elected mayors and suspected traitors in the zones it controlled (Betancur, Planchart, and Buergenthal 1993, 24-30;Bourgois 2001).
As a consequence, post-war reforms needed not only to demobilize and disarm the FMLN and reduce the number of military personnel, but also to break up an integrated system of state repression that included the military, the police and the judiciary.
Envisioning comprehensive reforms in the peace accords In 1992, the comprehensive Chapultepec Peace Agreement was signed including provisions on profound reforms of the security sector. 11 The FMLN had demanded a complete dissolution of the state armed forces (including the three different active police forces) as a complement to its own disarmament (Buchanan and Chávez 2008) but was unable to achieve such a radical solution. The compromise was set down in the Mexico Agreement (1991) as part of the 1992 Chapultepec Agreement. Here the core reforms were (i) the establishment of a new civilian police force replacing the existing three police forces under military control; (ii) the reduction in number of the armed forces and a new mandate limited to secure the country's external borders; and (iii) an increased independence and professionalization of the judiciary.
The design of these reforms has been praised at the national and international level (Stanley 1995(Stanley , 1999Call 2007). 12 Theoretically, the reforms aimed at a fundamental change of the integrated repressive system of the past via an evaluation of past behaviour and the dismissal of those responsible for human rights violations (vetting), by changes in mandate and judicial independence. Military reform included the reduction in number although no specific threshold was given. A second element was the appraisal and possible dismissal of 2000 members of the armed forces by an Ad Hoc Commission of three independent experts and two officers with an impeccable professional record, an unprecedented development in Latin America. The old police forces were to be dissolved, and the new civilian police established under civilian control. A quota of 20 per cent for former members of the police as well as 20 per cent of former FMLN combatants could apply after being screened for human rights violations but had to pass through the courses of the new Public Security Academy. According to the peace accord (United Nation's Security Council 1992) the police's 'mission shall be to protect and safeguard the free exercise of the rights and freedoms of individuals, to prevent and combat all types of crimes, and to maintain internal peace, tranquillity, order and public security in both urban and rural areas'.
Reforms in the judiciary were less far-reaching and detailed as those on the police or the military but they also envisioned profound changes focusing on increasing the independence and professionalism of the judiciary and its personnel. The members of the Supreme Court of Justice were to be elected by a two-thirds majority in the parliament; the judiciary was supposed to receive a minimum financial basis of 6 per cent of the annual state budget; a national judicial council and a human rights ombudsperson office were established and judicial careers were professionalized.
At the same time, the peace accord formulated some transitory provisions. Until the new police force was able to operate in the whole territory (planned to be possible after 21 months, that is by the end of 1993), the old police forces, as well as the military, were allowed to function under the supervision of the United Nations Observer Mission (ONUSAL). A Truth Commission was established to investigate gross human rights violations and formulate recommendations on further reforms with binding character according to paragraph 10 of the Mexico Agreement ('The Parties undertake to carry out the Commission's recommendations.').

Contentious implementation
Analysing these reforms, the first question is if, and to what extent, they were formally implemented. 13 From the perspective of historical institutionalism, the second question is how existing structures and power relations shaped this process. This analysis will also provide evidence on interactions between different reforms.
To understand the implementation of these reforms in the aftermath of war, we need to take a closer look at the evolution of the political system. The political system of El Salvador had been democratized already during the war with the introduction of competitive elections, albeit within a limited spectrum of political parties. As general elections with the participation of the FMLN were only scheduled for 1994, the ARENA (Alianza Republicana Nacionalista) government and its majority in parliament took on the crucial role of implementing the peace accords. 14 Already in April of 1991, the ARENA-dominated parliament passed three important constitutional reforms regarding the separation of the mandate for external and public security (Art. 159, 168). Emphasis was put on the respect for human rights. Art. 174 of the constitution established that the judiciary should receive at least 6 per cent of the government's expenditure. Military jurisdiction was limited to strictly military offences (Art. 216). Reforms continued in 1992 despite some small delays. On 23rd January, a law on National Reconciliation enabled the transformation of the former guerrilla into a political party. The Ad Hoc Commission started to vet 2000 officers of the Armed Forces in May and handed in its report and recommendations to President Cristiani and the UN Secretary-General on 22nd September. On 1st December, the new law on the National Judicial Council installed an independent body of judicial oversight. The new Public Security Academy had started its work on 1st September. The government and the FMLN declared the official end of the war on 15th December 1992. This was a promising start, and there was much optimism that the peace process was on a positive track (Montgomery 1995;Stanley 1995;Popkin 2000).
However, first signs of resistance also became evident, as the government was unwilling to follow the recommendations of the Ad Hoc Commission and dismiss officers with a record of human rights violations by 31st December 1992. In early January 1993, the United Nations reported that the peace treaty had been violated as fifteen of the officers who were on the list, including the acting minister and vice-minister of defence, were not dismissed. The government also opposed the dismissal of the members of the old Supreme Court on formal grounds. The persistence of the judges and the officers responsible for severe human rights violations and the lack of investigation and prosecution of the perpetrators was a clear indicator that the government was either unwilling to implement important provisions of the peace agreement or that the military as an institution was putting pressure on the government (Stanley 1995;Popkin 2000;Call 2007).
The Truth Commission emphasized the lack of reform implementation in the judiciary and its potential negative effects for peace in its report on 15th March 1993: The Commission does not believe that a reliable solution can be found to the problems it has examined by tackling them in the context which is primarily responsible for them. The situation described in this report would not have occurred if the judicial system had functioned properly. Clearly, that system has still not changed enough to foster a feeling of justice, which could promote national reconciliation. On the contrary, a judicial debate in the current context, far from satisfying a legitimate desire for justice, could revive old frustrations, thereby impeding the achievement of that cardinal objective, reconciliation. That being the current situation, it is clear that, for now, the only judicial system which the Commission could trust to administer justice in a full and timely manner would be one which had been restructured in the light of the peace agreements. (Betancur, Planchart, and Buergenthal 1993, 169) Although the recommendations of the Truth Commission were supposed to be binding, President Alfredo Cristiani stated that the report was one-sided and that the past had to be laid to rest to promote reconciliation (Popkin 2000, 150). Only five days after the presentation of the Truth Commission's report the still ARENA-dominated parliament passed a second, this time sweeping amnesty law although the peace agreement did neither mention nor include any form of amnesty for gross human rights violations. A first amnesty limited to political offences had been included in the 'Reconciliation Law' (Ley de Reconciliación, 23rd January 1992) as a precondition to release political prisoners and to enable FMLN members to enter the country legally. The second amnesty law (Ley de Amnistia General para la Consolidación de la Paz, 22nd March 1993) applied to all political, common, and connected offences committed prior to 1st January 1992. 15 The promotion of a broader amnesty by the ARENA government is not surprising taking into account ARENA's close ties to the death squads and the war crimes committed by the military and police during the war. Its founder Roberto D'Aubuisson, who had died of cancer in 1992, was named in the Truth Commission's report as being responsible for gross human rights violations such as the assassination of Archbishop Romero in 1980 (Betancur, Planchart, and Buergenthal 1993, 119-23). Death squads and the Salvadorian military were also responsible for other atrocities such as the massacre of El Mozote (1981), where more than 900 civilians were slaughtered by the military, and the murder of six Jesuit priests of the Central American University, their caretaker and her daughter in 1989. The broad amnesty law of 1993 rendered irrelevant the recommendations of the Truth Commission regarding the dismissal of the responsible officers and judges as well as further fundamental reforms in the armed forces and other security institutions.
This had consequences for SSR as it led to high levels of military personnel continuity; the amnesty law was a decisive turning point for the peace process. However, neither the FMLN nor the United Nations opposed the amnesty openly. Segovia (2009) argues that the amnesty was the price for formal compliance to the overall peace agreement by the Salvadorian government. At the same time, the FMLN too was not fully following the peace agreement as hidden stockpiles of weapons inside El Salvador and in neighbouring countries were soon discovered. In May 1993, the detection marked a major crisis of the peace process weakening the FMLN's credibility, leverage and the power of contestation significantly. The following bargaining process brokered by the United Nations resolved the crisis. The FMLN handed over the hidden weapons. The government formally carried out the remaining vetting of the armed forces by 30th June 1993 when the last officer included in the report of the Ad Hoc Commission left the armed forces. However, these officers were not dismissed but had just completed their term of service and retired via normal military procedures (Segovia 2009, 26). 16 They were part of the so-called La Tandona, the class of military officers that had graduated in 1966, many of whom were promoted to the highest positions in the government and the military (Walter and Williams 1993).
Similar to the difficulties for profound reform in the military a judicial reform was also highly deficient. A comparable pattern emerged as some reforms were initiated according to the peace accord but stalled in the process of implementation. A major legal reform preceded the peace accord when the constitutional reform of 1991 changed the rules for the election of Supreme Court magistrates. These had been elected by a simple parliament majority for five years before (securing high levels of political influence). Starting in 1994, supreme court magistrates needed a two-third majority for a 9-year term with the renewal of a third of the magistrates every three years (Spence et al. 1997). 17 However, the government did neither start a vetting process of judges nor dismissed the acting Supreme Court membersboth measures had been recommended in the Truth Commission's report.
Across Latin America, Offices of Human Rights Ombudspersons were established during the 1990s as a means of strengthening horizontal accountability countering the weak and politically dependent judiciary institutions. In El Salvador, the fate of the Procuraduria para la Defensa de los Derechos Humanos (PDDH) established in 1992 illustrates the contentiousness of the reforms. The second Ombudsperson, Victoria Marina Velásquez de Àviles (1995)(1996)(1997)(1998), questioned the government's record regarding human rights throughout her term and tried to promote reforms of the PDDH. As a consequence, she got death threats and the office was subsequently 'punished' with a 10 per cent cut in the budget. After the end of her term, the Salvadorian parliament elected the Ombudsperson under political criteria reducing the impact the office could have had on accountability (IDEHUCA 1998;Dodson and Jackson 2004).

Undermining reforms
The backlash for reforms in the security sector became evident as early as 1993, when the government sent the military into the rural areas (Plan Oro) under the pretence to secure the coffee harvest (Interview San Salvador, 17 March 2017). Government and media discourses on violence and insecurity (Moodie 2010) served the ruling coalition of ARENA and the military as a pretence to limit reforms and use traditional repressive policies. As the new police was underfunded and understaffed, former members of the military and its military police were increasingly admitted into the police and became largely overrepresented especially in leading positions (Stanley 1995;Call 2007) (Interview San Salvador, 15 March 2017). In open violation of the peace accord the government established joint military-police patrols in 1995, the parliament passed emergency measures in 1996, a law allowed the private possession of heavy weapons in 1999. These measures violated provisions on the separate mandates of the police (public security) and the military (external security) as envisioned in the peace accords. In the early 2000s, various plans for a 'strong fist' or 'super strong fist' on crime followed (Cruz 2011;Aguilar 2016). These repressive strategies did not reduce violence but fuelled its escalation.
An analysis of the power relations and the 'social coalitions' promoting or undermining reform implementation helps to explain these interactions. Three actors were important: the government, the FMLN and the military. While some interesting power shifts became visible at the end of war, it was the dominance of ARENA for nearly two decades after war's end that shaped the implementation process of the peace agreement. With ARENA, the economic elite recovered its political and economic dominance (Montgomery 1995;Albiac 1998;de Zeeuw 2010). First elected in 1989, ARENA's control of the government and the parliament was decisive after the FMLN's disarmament. ARENA was able to use its majority in the assembly not only to pass the amnesty law but also to undermine other reforms. 18 Most significant were serious delays in the establishment and a lack of funding of the new police. Economically, the ARENA government reversed the nationalization policy of the preceding Christian Democratic government and followed a neoliberal approach with the privatization of state enterprises such as banks (de Soto and del Castillo 1994;Boyce 1995). These policies thwarted the spirit of the peace agreement as the preamble of the economic and social chapter had explicitly called for: 'One of the prerequisites for the democratic reunification of Salvadorian society is the sustained economic and social development of the country.' Changing the course? SSR and the government of the FMLN After the peace agreement, the second window of opportunity for substantial reforms in the security and judiciary appeared in 2009 when the FMLN won the presidential elections. During the peace negotiations and as the main opposition party, the FMLN had advocated for broad reform of military and police and favoured an approach to public security emphasizing social and economic participation as well as violence prevention. But despite winning the presidency and being the largest party in the parliament, the FMLN never held an absolute majority. It won the presidency in 2009 with a moderate candidate, Mauricio Funes. During the first years of his term, Funes (2009-2014) started some reform initiatives although in contradictory directions. Legislation regarding public security continued to call for the involvement of the military 'to support' the police to an unprecedented level (Aguilar 2016). Parliament prolonged this 'exceptional' possibility seven times between 2009 and 2011. A short period of dialogue between the two most violent youth gangs with the secret involvement of the government decreased the numbers of homicides significantly. But confronted with political opposition, unfavourable opinion polls and a media outcry, the truce ended and violence escalated to even higher levels (INCIDE 2016).
Regarding judiciary reform, the most important financial and technical support after the war had come from the US Agency of International Development (USAID) with a strong focus on training to increase the capacities of and professionalize judges and other judicial personnel as well as on modernizing the criminal and civil codes. While these measures were important, their impact is difficult to analyse. Nevertheless, a process of gradual change in the judiciary can be observed (Cerqueira and Arteaga 2016). The judicial handling of the amnesty law from 1993 is an interesting case in point as human rights groups never stopped trying to revoke the law. In 1993, the Constitutional Chamber of the Supreme Court declared itself incompetent referring to the amnesty as 'a political act'. Five years later, the chamber ruled: 'the law admitted interpretations but was not unconstitutional per se'. In 2002, the office of the Human Rights Ombudsman in a rare act of political autonomy declared that the law contradicted the Salvadorian constitution and the country's international obligations. In 2009, five independent magistrates were elected to the Supreme Court; four of them entered the constitutional chamber. 19 When a group of human rights organizations filed a new lawsuit in 2013, the Supreme Court accepted to hear the case declaring the amnesty law unconstitutional in 2016. As a consequence, a series of lawsuits against the perpetrators of gross human rights violations such as the massacre of El Mozote re-entered the courtrooms. While important at the symbolical level, it remains to be seen if this is a turning point of judicial independence and autonomy.
Mainstream media largely ignore the process. There is increasing evidence that a coalition between ARENA and FMLN is blocking the trials. Although the Truth Commission's report laid more responsibility on ARENA and its supporters than on the FMLN, ARENA politicians were quick in accusing the current president and former guerrilla commander Sánchez Cerén of gross human rights violations during the war. As a consequence, he faces the allegation of being responsible for war crimes committed under his command. 20 Most of the FMLN leadership, thus, does not have much interest in promoting reforms in the judiciary; the party is currently facing high levels of discontent and has lost a high number of seats in the parliament in the elections of March 2018.
While relations between ARENA and the FMLN are tense during elections, a system of elite accommodation or informal power-sharing has developed over the last 25 years. Both parties have an interest to preserve the social, economic and political status quo. During the last decade, the FMLN has become part of the elite through its access to state funding and the establishment of economic enterprises. The political economy of post-war El Salvador thus resembles the persistence of institutional policies and behaviours Robinson (2010) analysed for Liberia and South Africa among others.

Conclusion: no peace without justice
This article aimed at the analysis of the interaction between reforms in the military, the police and the judiciary. Regarding the reform process, only rather technical and procedural changes in the institutional set-up of the security and justice institutions prevailed despite the profound reform design of the peace accords. The gradual changes implemented were unable to substantially transform the institutional practices. Powerful former armed actors and their political allies successfully blocked effective reforms in the judiciary that then would have held them accountable for gross human rights violations. Despite fundamental reforms on paper, personnel continuity undermined and prohibited the implementation of reforms leading to a path-dependent development. Without an independent autonomous opposition, protest against the reproduction of repressive policies only came from international and national human rights organizations such as IDHUCA (Instituto de Derechos Humanos de la Universidad Centro Americana) or the Inter-American Human Rights Court. These actors were able to maintain some issues on the political agenda but were unable to press for more substantial changes. The latest report of the UN Human Rights Committee (May 2018) on civil and political rights in El Salvador repeats all the deficits in the legal and institutional setting from a lack of comprehensive evidence for two important relations: First, even when fundamental reforms in the security sector (here: police and military) are implemented formally, high levels of personnel continuity limit and undermine their impact. Despite the training or formal regulations to respect human rights, the predominance of these actors reproduces repressive practices. Second, due to the limited character of judicial reforms, the responsible entities for horizontal accountabilitythe courts or the PDDHhad no influence on the government's public security policies. This reproduced repressive strategies in public security.
The main explanation for path dependence rests on the relations between different elite factions in the immediate aftermath of war favouring the status quo. The dominance of the right-wing government before the first post-war election in 1994 and until 2009 limited and undermined reforms. ONUSAL and other international actors did not exert significant pressure to counter this development. Their focus was on the success of peace-building and democratization using a minimum definition of peace as no war recurrence and on formal democracy as largely free and fair elections. Under this minimalist approach, political violence in the immediate aftermath of war was downplayed and violence was defined as 'criminal' 21 . This supported the government's claim that there was a crisis of public security.
Despite a change in government in 2009, path-dependent repressive strategies prevailed due to a process of elite accommodation. While the FMLN supported some reforms such as the election of independent magistrates to the Supreme Court or a short period of talks with the gangs, it also resorted to electoral strategies regarding repressive public security after a backlash of public opinion and media reactions. As a consequence, the violent practices of the state security forces were reproduced and the cycle of violencerepressionviolence persisted. Until today, the justice system is unable to reduce or end impunity for prominent as well as everyday grave human rights violations.
What can we learn from Salvadorian experiences for the interaction between institutional reforms in the security and judiciary sector? After all El Salvador is considered a success story of liberal peace-building. Two general topics emerge: First of all, timing matters and a linear approach to the security sector, and judiciary reform is risky as impunity prevails and repressive strategies are not sanctioned. While most peace agreements confront the challenge of accommodation of powerful armed and non-armed actors, fundamental reforms may only be possible as long as the quest for peace and war termination is urgent and/or as long as external pressure is felt. Afterwards the reform impetus (if existent at all) is lost, international actors leave to other crisis regions, and local dynamics prevail. Second, elites need to be held accountable even if this takes time. The debate on past atrocities might not lead to legal punishment but is important to break the cycle of violence reproduction and to establish new parameters for non-violent behaviour. 4. I conducted 55 semi-structured interviews with members of different government institutions and ministries, experts, civil society activists, and international representatives working in the field of institutional reforms regarding security and violence. Focus group discussions were held in three different regions of the country with ex-combatants. The guiding questions during interviews and focus groups were related to (i) the nature of the reforms and their legal basis; (ii) the actors involved in the reform and those providing security and justice; (iii) international assistance and funding. Thanks go to Marlon Hernández for the coordination of the dense agenda. Results are rather inconclusive. 9. In October 1979, a coup by reform oriented civil-military coalition ousted the military government of General Romero. But hopes for reforms soon vanished as the civilian members resigned due to military pressure early 1980. The following repression and political violence was a major trigger for the onset of the civil war (Montgomery 1995;Stanley 1996). 10. Since 1978 the US State Department reports on human rights documented severe and widespread violence against civil society, see http://www.politicalterrorscale.org/Data/ Documentation.html. 11. The English version of the agreements can be accessed at https://peaceaccords.nd.edu/ accord/chapultepec-peace-agreement (last accessed 10 June 2018). 12. Even 25 years after the end of the war, most interviewees from different backgrounds, such as a former general and ex-combatants of the FMLN as well as civil society representatives, stated that these reforms were excellent on paper. However, the former general criticized the human rights focus of the reforms that he thought unsuitable for the Salvadorian context (Interview San Salvador 13 March 2017). 13. According to the Peace Accord Matrix project, military reforms were fully implemented in 1996, police reforms in 1997, while the level of implementation regarding judiciary reforms was only intermediate. See Peace Accords Matrix (Date of retrieval: 11 February 2018), https://peaceaccords.nd.edu/accord/chapultepec-peace-agreement, Kroc Institute for International Peace Studies, University of Notre Dame. 14. I thank Désirée Reder for research assistance regarding the timeline of reforms. 15. The texts of the laws are available in the government's archive: 23 January 1992 -Ley de Reconciliación Nacional Decreto No. 147: 1992 and 20 March 1993 -Ley de Amnistia General para la Consolidación de la Paz, Decreto No 486, D.O No 56, Tomo No 318, Publ, 22 March 1993. 16. Defence Minister René Emilio Ponce already offered his resignation in March 1993 but stayed in office till end of June 1993. 17. As a consequence a third of the magistrates elected in 1994 magistrates only served three years, another third six years. 18. In the first post-war elections (1994), the FMLN came second in the presidential and parliament elections with 24.9 per cent and 21.4 per cent respectively, while ARENA won the presidency with 49 per cent and was the strongest party in the parliament (45 per cent and 39 out of 81 seats) although it had no majority.

Disclosure statement
No potential conflict of interest was reported by the author.

Notes on contributor
Sabine Kurtenbach is a political scientist and Acting Director at the GIGA German Institute of Global and Area Studies and Honorary Professor at Philipps University Marburg. Her research has focused on peace, conflict transformation, postwar societies, youth, and institutions with a specific regional focus on Latin America but also beyond.