Does the 2015 Peace Accord in Mali Reflect the Priorities of Households and Local Leaders?

Using unique household level data collected in northern Mali in August and September 2015, this paper explores whether the content of the 2015 Peace Accord reflects the concerns and priorities of the citizens, local leaders, and refugees, or whether it is an exclusive agreement between elites who fail to represent the interests of the population at large. The study finds that local leaders? opinions are appropriate proxies for the citizens they represent, and most (but not all) of the measures stipulated in the Peace Accord align. The greatest divergence between citizen preferences and actions specified in the Peace Accord is the emphasis in the latter on road infrastructure. When different activities are costed, this disconnect becomes very apparent. The bulk of funds (45 percent) are apportioned to infrastructure construction, while the typical respondent prefers job creation; the poorest prefer investments in social services. However, most of infrastructure spending has been earmarked for the medium term. Short-term priorities reflect the preferences of the population in northern Mali much better. It is concluded that although the Peace Accord had all the elements to be an exclusive elite pact, its priorities largely reflect those of the population, granting it much higher probability of success.

Using unique household level data collected in northern Mali in August and September 2015, this paper explores whether the content of the 2015 Peace Accord reflects the concerns and priorities of the citizens, local leaders, and refugees, or whether it is an exclusive agreement between elites who fail to represent the interests of the population at large. The study finds that local leaders' opinions are appropriate proxies for the citizens they represent, and most (but not all) of the measures stipulated in the Peace Accord align. The greatest divergence between citizen preferences and actions specified in the Peace Accord is the emphasis in the latter on road infrastructure. When different activities are costed, this disconnect becomes very apparent. The bulk of funds (45 percent) are apportioned to infrastructure construction, while the typical respondent prefers job creation; the poorest prefer investments in social services. However, most of infrastructure spending has been earmarked for the medium term. Short-term priorities reflect the preferences of the population in northern Mali much better. It is concluded that although the Peace Accord had all the elements to be an exclusive elite pact, its priorities largely reflect those of the population, granting it much higher probability of success.

I. Introduction
After two decades of multiparty democracy, Mali was viewed as a democratic success story. The fifth presidential elections were scheduled to take place in March 2012 and another peaceful and democratic transfer of power was widely anticipated. Reality was different, however. A secessionist movement sparked by a Kel Tamasheq 1 rebellion led to a political and constitutional crisis culminating in a coup d'état in March 2012 and an attempt to take over the country by force. The three northern regions of Gao, Timbuktu and Kidal became occupied by various rebel and Islamist factions until early 2013, when a coalition composed of the Malian Army, French troops and the ECOWAS-led African-led International Support Missions to Mali (AFISMA) recaptured the occupied areas. 2 After months of insecurity in the North and two violent attacks in Bamako, a Peace Accord was signed in May and June 2015 between the government and different actors involved in the rebellion. The Accord established a joint vision for peace and prosperity predicated on demobilization and disarmament, the devolution of authority to local governments, and the establishment of conditions for restoring stability and economic recovery in northern Mali.
In spite of the Accord, the regions of Gao, Kidal and Timbuktu remain in a state of prolonged crisis, with high levels of insecurity and weak governance. Without army protection most parts of the North, especially Kidal remain inaccessible to those working for the central and local government. 3 Armed bandits are active and IED explosions as well as violent attacks on the MINUSMA peacekeeping forces are regular occurrences.
Under these circumstances, data collection is very difficult. INSTAT, the National Bureau of Statistics, has not been in a position to collect information from northern Mali since the beginning of the crisis. To our knowledge, our surveys implemented by a private survey entity, GISSE, are the only systematic and representative effort to collect data in north Mali since the crisis. They offer a unique database providing a crucial perspective that would otherwise not be reflected in academic analyses and policy level decision-making, and a perspective that is indispensable in any attempt at understanding the situation in Northern Mali. 4 Preceding the Accord on Peace and Reconciliation in Mali (Accord pour la paix et la réconciliation au Mali) (hereafter, the Peace Accord) of May and June 2015, four peace accords had been signed between the government and Toureg and Arab armed groups in Mali. 5 There are different interpretations as to why past peace agreements failed, ranging from the assumption that peace agreements had been used as cooptation strategies (Bøås 2012); that rebellions were mainly driven by internal rivalries as opposed to grievances against the state (Pezard & Shurkin 2015); that the government failed to follow through on its commitments (Wing 2013); that shortcomings of decentralization and democratization were at fault (Wing 2013;Pezard & Shurkin 2015); and the lack of representativeness at the negotiation table (Pezard & Shurkin 2015;Claudot-Hawad 1996). It is this latter issue that drives the research question of this paper.
The armed groups involved in Mali's rebellions are not necessarily representative of the entire northern population and could well represent minority interests. The popular mandate of those negotiating with the government on behalf of the armed groups is not self-evident. Telling is that each accord was signed with a different armed group, representing at most a few factions, the exception being the 1992 National Pact, which involved a broader coalition. Yet, even this broader coalition did by no means represent all or even the majority of the northern population. 6 This pattern was repeated in 2015 and the Peace Accord has been described as an elite bargain, securing lucrative roles for northern traditional leaders in to-be-formed local institutions and protecting the de facto autonomy of the most rebellious Tuareg clans, 7 though it is not immediately evident that such objectives have been achieved (at least not in the short run). What is evident is that certain groups, such as Gatia, (MINUSMA). 9 Reaching an accord with so many players and interests at stake can be challenging and concessions need to be made that may lead to a narrow representation of elite interests at the expense of those of the population.
The 2011 World Development Report argues that elite pacts can provide short-term security but that violence often recurs unless the pact broadens and is accompanied by institutional transformation. 10 Thus distinctions have been drawn between inclusive and exclusive elite bargains, depending on the degree of integration of broad coalitions of key elites versus narrow coalitions in terms of access to state structures (e.g. jobs in the government and the army). 11 Elite pacts and peace agreements can bring an end to conflict and violence in the short-term, but if such horizontal pacts are not accompanied by inclusiveness and responsiveness along a more vertical dimension, societies remain vulnerable to the same tensions that brought about violence and instability in the first place. 12 Using household level data collected in northern Mali in August and September 2015, we explore whether the content of the 2015 Peace Accord is inclusive, i.e. whether it reflects the concerns and priorities of the people of the north or whether it is an agreement between elites who fail to represent the interests of the population at large. We look at how the actions outlined in the Peace Accord align with the priorities of local leaders and citizens in the north. We also explore whether the financing priorities reflected in the report of the Joint Assessment Mission (Mission d'évaluation conjointe, MIEC), which operationalizes the Peace Accord, align with priorities identified by households and their leaders.

II. The Population of the North: Consequences of a Crisis
The 2012 rebellion and subsequent conflict in Mali resulted in large numbers of displaced. About 353,000 persons were internally displaced, while another 170,000 fled to neighboring countries. At the height of the crisis around 36% of the population in the north had left their homes. By October 2015, this number had decreased to less than 200,000.
Internally displaced returning has brought about the majority of this decrease, while an estimated 137,000 refugees continue to reside outside Mali. 13 Although it was the better educated and wealthier households as well as those less exposed to violence who fled the crisis, the negative impact of the crisis on the welfare of refugees, IDPs and returnees was significant. There was a notable shift towards feeling poorer corroborated by a reduction in income and employment and decreases in assets and livestock owned. Those who fled the crisis have seen a considerable decline in their welfare, including the loss of many of their possessions. Expressed in value terms, they lost 20% to 60% of their durable goods and 75% to 90% of their animals. 14 Households suffered several shocks including violence, insecurity and rising food prices. In Gao, more than half the households were victims of violence, more than 37% were displaced, more than 27% were victims of theft and almost a quarter experienced a serious accident with fatal consequences for one of its household members. In Kidal and Timbuktu, households were affected by violence and insecurity (47% and 63% respectively) and by displacement (22% and 47% respectively). Following the signing of the Peace Accord, households in northern Mali consider that the government is not managing reconciliation, decentralization (a cornerstone of the Accord) or infrastructure rehabilitation very well, nor is it creating a climate of trust between the different communities and ethnic groups or addressing the stigmatization of Tuareg and Arab groups. In Kidal, practically all households state that the government is poorly managing all of its problems with the exception of decentralization. In Gao, over half the households believe that the government is inadequately managing all of these problems except reconciliation and the fight against stigmatization of which 47% think management is poor. In Timbuktu, 48% of households believe that reconciliation is being poorly handled and 54% to 73% state that all other issues are being poorly managed.
Perceived security is low in northern Mali and citizens feel insecure during the day and at night. In this climate of uncertainty the overall level of confidence is also low. Only 74% of households trust people of other ethnic groups and only 70% trust strangers. The level of overall confidence is lower in Gao than in Kidal and Timbuktu. In addition, over 40% of households do not trust the judicial system and political parties. Nevertheless, in the three regions, over 90% of households trust neighbors, people of the same ethnic group, teachers and health workers, as well as traditional and religious leaders.  Following the signing of the 2015 Peace Accord, the African Development Bank, the Islamic Development Bank and the World Bank were asked to conduct a joint identification and evaluation mission to identify the priority needs in the North, taking the actions outlined in the Peace Accord as point of departure. The World Bank subsequently commissioned a study to evaluate the socioeconomic situation of the population in the north and their priorities in the areas of security, peace, governance and socioeconomic development. Data were collected among 500 households (3,673 individuals in total) and 50 authorities/leaders in rural and urban areas of Gao, Timbuktu and Kidal in August and September 2015. The objective of the survey was to analyze perceptions and priorities with regard to the Peace Accord, to analyze perceptions of security, to determine access to basic infrastructure and school attendance, to understand nutrition levels, and to measure household asset ownership.
In addition to this baseline survey, the World Bank was sponsoring a mobile phone survey among IDPs in Bamako, returnees in Gao, Kidal and Timbuktu town and refugees in camps in Niger and Mauritania. This survey interviewed 500 respondents on a monthly basis. In the August 2015 round of this survey, questions about perceptions and priorities with regard to the Peace Accord were included. This paper also makes use of a subset of the responses obtained from that survey, particularly those from refugees in Niger (n=80) and Mauritania (n=100) as these sub-populations who live outside Mali's borders are important stakeholders in the peace process whose opinions risk being ignored. 15 To select a household in a village or neighborhood for the baseline survey, random selection was used: the enumerator divided the locality into two parts and selected five households in each. To select a starting point the enumerator used a code of the day 16 and chose every second house in rural areas and every fifth house in urban areas. The selection of individuals within the household to answer the questionnaire was conducted as follows: the head of household (male or female) was selected to answer the first part of the questionnaire dealing with general questions about the households. Using the roster of household members which was compiled during the first part of the interview, another member of the household aged 18 or above was selected randomly to answer the second part of the questionnaire in which perception questions were asked. Alternation between male and female was ensured. The survey thus generated data that are reflective of the opinions of those aged 18 and above in northern Mali.
To assess the representativeness of the data, which were collected under rather challenging circumstances, the ethnic composition of the sample was compared with the ethnic composition in the North as reported by the 2009 Census. Similarly, regional representation of the sample was found to be in line with actual regional distribution of the population in the north. 16 The code of the day is the sum of the two figures of the date, i.e. if it's the 25 th of August the code of the day is 2+5=7. The enumerater will chose house number 7 as a starting point.

Figure 3: Ethnic composition of sample
In addition to the 500 households and 180 refugees, 50 local authorities were interviewed. Of those, 18 are based in Goa, 22 in Timbuktu and 10 in Kidal. Included were 38 village chiefs, 11 mayors and 1 local notable.

III. The 2015 Accord on Peace and Reconciliation in Mali
Preceding the Peace Accord's signature, talks were held in Algiers between six armed groups congregated in two broad coalitions -the CMA (Coordination des Mouvements de l'Azawad) and the Platform -and the Malian government. Al-Qaida in the Islamic Maghreb (AQIM) and the Movement for Unity and Jihad in West Africa (MUJAO), both foreign-origin jihadist groups, were excluded from the talks although they were controlling most of the North from June 2012 to January 2013. 17 Another security actor that emerged after the negotiations began, and which was not represented, is the Imghad Tuareg and Allies Self-Defense Group (GATIA), 18 a pro-government self-defense group, fighting those groups seeking greater autonomy and or an independent state of Azawad. Initially, the CMA did not sign the Peace Accord. The Platform coalition of armed groups took over Ménaka in Gao region in April 2015, prompting the CMA to refuse to join the signing ceremony on 15 May 2015. The CMA set as condition for signing the withdrawal by the Platform from Ménaka. Following collective international mediation initiatives, the Platform announced its immediate withdrawal from Ménaka on 18 June and on 19 June the Government of Mali lifted arrest warrants against 15 leaders of the CMA. Subsequently, on 20 June, Sidi Brahim Ould Sidatt from MAA-CMA signed the Peace Accord on behalf of the CMA. On 23 June, Mali's President Ibrahim Boubacar Keïta met with the leadership of CMA in Bamako to discuss the way forward and on 25 June, the National Assembly approved financial provisions to start the implementation of the Peace Accord. 20 Although six armed groups had been included in peace talks, it has been observed that none legitimately represented civil society in the north nor were civil society representatives invited to the talks. 21 In addition, the exclusion of GATIA from the peace talks has aroused questions whether reaching an agreement without GATIA even has the potential to create peace and order, when such an important actor is left out. The HCUA, MNLA and the anti-Bamako faction of the MAA have a strategic interest in marginalizing GATIA "in part because GATIA's very existence challenges their claim to be the legitimate voices of northern Mali's Tuareg and Arab communities." 22 This alludes to the fact that peace negotiations may have been an exclusive elite bargain, lacking critical representation. On the other hand, GATIA was only formed in August 2015 and is allied with the government, so its interests may have been reflected in the Accord nonetheless.
In addition to the local parties who negotiated the Peace Accord, a wide range of international stakeholders influenced (or tried to influence) the Peace Accord. Algeria, Morocco, Mauretania, Burkina Faso, Niger, Chad, Libya, the EU, OIC, ECOWAS and MINUSMA were all stakeholders and somehow involved with interests that varied from security in the Sahara, control over natural resources, border security or stemming the flow of migrants and illegal trade.
To explore the inclusiveness of the Peace Accord and the degree to which the Accord reflects the preferences of the population of north Mali it merits taking a closer look at the content of the Accord itself. The Peace Accord determines that the principles and foundations of sustainable conflict resolution are built on the respect of national unity and territorial integrity as well as a common understanding of the term Azawad, a Tamasheq word to describe the three regions of the North. The Peace Accord covers four substantive themes: 1.) Policy and Institutional Issues, 2.) Defense and Security, 3.) Socioeconomic and Cultural Development, and 4.) Reconciliation, Justice and Humanitarian Issues.
The section on Policy and Institutional Issues does not accord independence to northern Mali (a position the Government of Mali and all other countries facilitating the negotiations strongly defended), nor a federal solution but outlines the new institutional framework that does grant greater managerial autonomy to the regions that constitute Azawad. It is based on an institutional architecture structured along 'Collectivitees territoriales' (i.e. regions, districts and local communities) that allows the population of the north to administer their own affairs and to increase representation of the northern population in national institutions. The Accord further outlines that there will be an equitable distribution of power and responsibilities between the state and communities. Regions will have an elected regional assembly, with a significant transfer of competencies, resources and judicial power. Local communities will also have governing bodies elected through universal suffrage and will be granted administrative freedom. To ensure adequate financial means for the collectivitees territoriales, a financial transfer mechanism will be put in place whereby 30% of national budgetary revenue is transferred, with particular attention being paid to the North 'along criteria yet to be determined.' Most of this is not new and has already been specified in past peace accords in some form or another.
Defense and Security addresses four key areas. First, disarmament, demobilization and reintegration (DDR) of armed groups in the north, which will be undertaken with the support of MINUSMA. A national DDR commission comprising of government representatives and movement leaders -signatories to the peace accord -will be set up. Second, gradual redeployment of security and armed forces to the north will be ensured. A substantial number of those redeployed will be from the north. Third, the parties to the Accord agreed on Security Sector Reform (SSR) with the aid of Mali's international partners. To this end, a national council with an advisory function will be established and police forces under the control of the collectivitees territoriales will be put in place. Finally, all parties commit to fighting terrorism and transnational organized crime. The guiding principle of all activities under the Defense and Security section is the redeployment of the Malian army to the north and the representativeness and inclusiveness of all Malians in the armed forces.
The third substantive section in the Peace Accord is Socioeconomic and Cultural Development. One important decision in this section is to create a development zone comprising the three regions of the north. The objective is to improve development indicators to levels attained in the rest of the country in no more than 10-15 years. A development strategy will be designed and implemented and the northern development zone will have access to domestic and international financial resources to achieve its objectives. In addition, the following initiatives are proposed under this section: (i) improve schools and health facilities, (ii) revitalize the local economy, (iii) rural development, food security and the environment, (iv) job creation, (v) develop road infrastructure, (v) improve access to basic services (education, health, drinking water), (vi) develop access to electricity, and (vii) revitalize cultural activities in the north.
The last substantive theme is Reconciliation, Justice and Humanitarian Issues. National reconciliation will be based on the establishment of a national peace charter, putting in place mechanisms for a Truth and Reconciliation Commission, creating a commission to combat corruption and one which deals with war crimes, ensuring better coordination of judicial powers, and initiating a judicial reform, among others. In addition, the parties commit to creating the necessary conditions that will facilitate return, repatriation and reintegration of the displaced population.
Most of the measures proposed in the Peace Accord are not very specific, but Annexes to the Accord detail the measures to be considered in the immediate period after the signing of the Peace Accord, defined as 18 to 24 months. Education, health, hydraulics and revitalizing the local economy are indicated as urgent priority interventions to be implemented in the north in the immediate to short-term. Access to basic services, job creation and supporting rural development, food security and the environment -all top priority interventions for households and authorities -have been identified as measures to be implemented in the medium to long-term. In addition, there is a heavy focus on road construction, including several airports in the north.
Finally, a monitoring committee (Comitte suivi d'accord) will be put in place to oversee implementation of the Accord. It will be comprised of representatives of all signatory parties. Implementation remains the most important challenge and the role of the monitoring committee will be crucial in clarifying issues such as how many combatants from the Platform and the CMA can be integrated into Mali's security forces. In this sense the Peace Accord did not end the Malian crisis, but opened a new phase during which peace will have to be won. 23

IV.
Results: Governance, Security, and Socioeconomic Development The Peace Accord was well received by the population: 46% of households are satisfied and 18% very satisfied with its signature. Even in Kidal, where sentiments for an independent Azawad are generally believed to be strongest, a great majority of the population is satisfied with the Peace Accord. At the same time, only 17% of households believe that the Accord will be respected while 38% do not think it will be respected and 45% are not sure. Inquiring about the importance of the initiatives outlined in the Peace Accord, 24 we find that most households and authorities consider all proposed initiatives as necessary to bringing peace and security to Mali. The vast majority consider all initiatives as very important thus Figure 6 shows some variation within the category of 'very important'. We find that the most important governance initiative is the management of own affairs; the most important security initiative is the redeployment and the reorganization of the armed forces as well as the fight against terrorism, and; the most important socioeconomic development initiative is job creation.
The clearest differentiation occurs when households and authorities have to choose their preferred initiatives within and among all categories of governance, security and socioeconomic development. Among all categories we find that households and authorities show a preference for socioeconomic initiatives, followed by security and governance. We will now focus on preferences within each category and compare preferences of households with those of local authorities to assess whether preferences of households and authorities align.

Governance
For households, the management of their own affairs is the most important governance initiative. This holds across all three regions. In both rural and urban areas around onethird of all households prioritize this initiative. In Timbuktu, households attribute the same importance to the management of their own affairs as to greater representation of the northern population in national institutions. We find this to be consistent when disaggregating by ethnicity. 25 The second most important measure varies slightly as can been seen in table 7.

Kidal
Greater responsibility in managing our own affairs (35%) Strengthening the rule of law and accountability (23%)

Timbuktu
Greater responsibility in managing our own affairs (32%) Greater representation of the northern population in national institutions (32%)

Urban
Greater responsibility in managing our own affairs (33%) Greater representation of the northern population in national institutions (26%) 25 We are only considering Songhai and Tamashek who represent 47% and 37% respectively of the survey population.

Rural
Greater responsibility in managing our own affairs (33%) Greater representation of the northern population in national institutions (25%)

Songhai
Greater responsibility in managing our own affairs (35%) Establishment of communities with elected bodies through universal suffrage (21%)

Tamashek
Greater responsibility in managing our own affairs (32%) Greater representation of the northern population in national institutions (32%) Leaders and citizens share the same priority initiative for governance: both citizens (33%) and leaders (29%) prefer greater responsibility in managing their own affairs. Among leaders the second most important initiative is the establishment of communities with elected bodies (26%) and third, strengthening the rule of law and accountability (22%). Disaggregating be region we find that greater responsibility in managing their own affairs is the most important initiative for households and authorities in Gao and Kidal. In Timbuktu 36% of leaders prefer an equitable distribution of power and responsibilities between the state and communities. While there is a greater preference for communities with elected bodies (41%) and greater responsibility in managing their own affairs (41%) among refugees in Niger, we find that refugees in Mauritania have very different preferences. 85% of Mauritanian refugees prioritize an equitable distribution of power and responsibilities between the state and communities.
While urban citizens have a preference for greater representation of the northern population

Figure 8: Most important governance initiatives, by region (%)
Equitable distribution of power and responsibilities between the state and communities Strengthening the rule of law and accountability

Greater representation of the northern population in national institutions
Greater responsibility in managing our own affairs Establishment of communities with elected bodies through universal suffrage in national institutions (26%), urban leaders show no desire (3%) for this. One interesting aspect is that the majority of leaders who have no education (57%) have a preference for the governance initiative of greater responsibility in managing their own affairs. Those with a high school degree, however, prefer a greater representation of the northern population in national institutions. Citizens, by contrast, prefer greater responsibility in managing their own affairs, irrespective of the level of education.

Security
Insecurity remains high during the day and the night and affects all three regions. Over 40% of households in Gao and Kidal and 30% in Timbuktu do not feel safe. Households blame insecurity on bandits, thieves and criminals (92% of the households in Gao, 100% in Kidal and 97% in Timbuktu), armed movements like the MNLA, MAA or HCUA (66% of households in Gao, 39% in Kidal and 75% in Timbuktu), and the presence of traffickers (64% of households in Gao, 85% in Kidal and 58% in Timbuktu). Terrorists and jihadists are also stated as a source of insecurity by 74% of households in Gao, 92% in Kidal and 67% in Timbuktu.
Looking at the initiatives proposed under the rubric of security in the Peace Accord, we find that leaders and citizens share very similar preferences. Priority number one is the gradual redeployment of security and armed forces (36% households, 32% leaders), followed by the participation in the management of local security (25% for both) and finally the reorganization of the armed forces (18% households, 16% leaders). Sixteen percent of leaders also prioritize representation of the northern population in the armed forces. Disaggregating by ethnicity we find that for Songhai and Tamasheq the most important initiative is the gradual redeployment of the armed and security forces followed by an active and meaningful participation of the northern population in the management of local security. Disaggregating by education we find that all households prioritize the gradual redeployment of the army with the exception of those with a high school degree who prefer participation in the management of local security.  When breaking this down by region we find that Kidal and refugees in Mauritania stand out. In Kidal 89% of the leaders and 37% of the citizens prioritize the participation of the northern population in the management of local security. Outside Kidal, the gradual deployment of the armed and security forces is the measure deemed most important both for households and authorities. Kidal is the only region that does not prioritize this initiative where it is important to only 9% of households and none of the authorities. This may not come as a surprise if one realizes that much of the income earned in Kidal is from smuggling -an activity that may become more difficult or less profitable once the army is redeployed in the region. 26 They share this with refugees in Mauritania where only 1% support the return of the army to the north while this is the most important initiative for refugees in Niger (54%). 89% of refugees in Mauritania prioritize inclusive and meaningful representation of the northern population within the armed and security forces.

Socioeconomic Development
Socioeconomic development is the third substantial issue agreed upon in the Peace Accord. Leaders and citizens share the same three priority initiatives. Leaders have a preference for access to basic services (26%), followed by job creation (17%) and rural development, food security and environment (15%). Citizens have a preference for job creation (21%), followed by rural development, food security and environment (15%) and development of access to basic services (13%).
We find that households in Gao prioritize job creation, in Kidal it is the development of access to basic services and in Timbuktu supporting rural development, food security and the environment, but for both Kidal and Timbuktu job creation is a very close second. Disaggregating by urban and rural, job creation still stands out as the priority initiative for 26 Raballand, Gael (2015)   Preferences of citizens and leaders for socioeconomic development initiatives are more heterogeneous when disaggregating by region. In Gao the preferred initiative among citizens is job creation (24%) and for leaders it is access to basic services (28%). In Kidal both citizens (30%) and leaders (51%) prioritize access to services, while in Timbuktu 20% of citizen prioritize rural development, food security and environment and 23% of leaders prefer access to electricity. Once again the overwhelming majority of refugees in Mauritania share the same opinion: 90% prioritize the establishment of a northern development zone, while the priority intervention for refugees in Niger is job creation.  We also wanted to understand whether preferences vary by poverty levels and losses in assets as a consequence of the crisis. We calculated wealth quintiles based on livestock ownership, ownership of consumer durables and the number of times a household eats different kinds of animal protein. We find that the preferred initiative for all groups is socioeconomic development but that there is significantly more interest in security initiatives for the wealthiest as compared with the poorest. Those who lost least are almost double as interested in security as those who lost most, who care much more about socioeconomic development activities.
With regard to governance we find that while those who lost least as a consequence of the crisis care little about strengthening the rule of law (13%), those who lost most do care (22%). While all other groups care significantly about greater representation in national institutions, leaders care least (11%). Only 2% of the poorest and 8% of those who lost most prioritize the redeployment of the armed forces while this is a priority for 20% of   those who lost least.
We found that job creation was the number one priority for citizens and leaders, but it turns out that this does not reflect preferences by income. In fact, the poorest do not care much about job creation (6%) but prefer access to basic social services (16%) and rural development (22%), preferences they share with leaders but not with the wealthiest. Only 13% of the wealthiest prioritize rural development and 9% access to basic services. Despite the heterogeneity in preferences, there is not much evidence of leader preferences reflecting those of one specific group more than another i.e. they do not seem to side with the wealthy or the poor, with those who lost more or less. We ran a multinomial regression (see annex) to further assess whether there exist systematic differences between the preferences of leaders and citizens. Data from the authorities and household respondents were pooled to explain the choice between various governance, security and socioeconomic initiatives which had been regrouped in seven categories: a development strategy that puts the north at par with the rest of the country, social services, rural development and employment, infrastructure, security, governance and jobs. We find that explanatory power is low: i.e. it is hard to predict preferences based on observed characteristics. Authorities generally represent the preferences of the population (with some slight preference for social projects in urban areas), rural households in Kidal are less interested in rural development and employment and those in rural Gao are particularly interested in infrastructure and jobs. Based on this, the general conclusion is that local leaders' opinions are appropriate proxies for the citizens they represent.

V. Operationalizing the Priorities in the Peace Accord
As mentioned above, Mali's Peace and Reconciliation Accord called on the World Bank, the African Development Bank and the Islamic Development Bank to carry out a Joint Assessment Mission (Mission d'évaluation conjointe, MIEC) to operationalize the Peace Accord. Using the various measures mentioned in the Accord, the MIEC was expected to prioritize and cost needs with an eye to ensuring a rapid recovery, to addressing poverty and promoting development, and to allowing the northern regions to reach the same level of development as the rest of the country within the next 15 years.
A prioritization framework was thus developed over the course of 60 days, along three pillars: 1) Strengthen peace, social resiliency and decentralized governance; 2) Restore and expand basic social services and social protection; and 3) Promote economic recovery, employment and infrastructures. Within each pillar, priorities were categorized based on whether implementation occurs immediately (year 1), in the short-term (year 2), or in the medium-term (3 to 6 years). 27 We find that, in broad strokes, priorities of households and local authorities are reflected in the pillars of the Report of the MIEC, although the MIEC demonstrates a strong preference for financing socioeconomic development activities instead of security and governance ( Figure 12). Table 12 outlines financing priorities as defined in the MIEC report but rearranged to fit the three main components of the Peace Accord: Governance, Security and Socioeconomic Development. The MIEC report proposes that the government spends 86% of its budget for recovery and development activities on socioeconomic development -62% of citizens see this as a priority sector for intervention; 7% is planned for security -26% of citizens see this as a priority sector; and 6% is dedicated to governance -for 12% of citizens this represents the priority sector. The bulk of spending for governance (69%) and socioeconomic spending (75%) is planned for the medium term (3-6) years, while 59% of the budget allocated to security is assigned to the immediate to short-term (1-2 years).
When taking a closer look at the MIEC's subcomponents, greater discrepancies appear between what households and authorities deem important and the monetary priorities brought forward by the MIEC. The gradual redeployment of the armed forces is important to both citizens (11%) and leaders (8%) -see Table 12, but there is no mention of this in the MIEC report, for instance. Given the importance attached to this issue by households and leaders in the north, it would merit clarifying this point and emphasizing the activities that will be financed to ensure the redeployment of the armed and security forces in the short to medium term. Year 1 + 2 (Short term) Years 3-6 (medium term)

Figure 12: Priority initiatives for households and authorities compared with MIEC financing priorties (%)
Governance Security Socioeconomic Development Job creation is the third most preferred initiative among all initiatives for both households and authorities. Within socioeconomic development it is the preferred initiative for 21% of households and 17% of authorities, which makes it the top priority for households and the second choice for authorities. Nonetheless, out of a total budget of 1,781,488.4 CFAF, million budgeted for the Pillar on Promoting Economic and Infrastructure Recovery and Employment in the MIEC report, only 21,287.4 CFAF, million -no more than 1.2% -is dedicated to Private sector development and financing services, which includes an objective on employment creation. The importance of this issue would merit a stand-alone subcomponent dedicated to job or employment creation with a budgetary allocation that mirrors the importance of this initiative for peace and stability.
The bulk of monetary allocation under this pillar is apportioned to infrastructure rehabilitation and construction (71%), with the objective to rehabilitate and construct critical transportation, communication, and electricity infrastructure to connect the north to the national and regional economy. The MIEC report reflects all transportation projects proposed in the Peace Accord (roads and airports), disregarding popular preferences (see Table 13). Developing road infrastructure is of priority to 6% of the households and 5% of the authorities making it to place 8 and 9 respectively out of a total of 20 development initiatives important for establishing peace and stability. Developing access to electricity is a priority for 0.5% of households and to none of the authorities.  (2016) The subcomponent agriculture (food security, productive capacity and livelihoods) has been allocated 24% of the total budget, 37% even of the budget for years 1-2. This is important because rural development and food security is the priority intervention for 13% of households and 15% of authorities ranking it as first and second priority initiative (respectively) in its importance to promoting peace in the north.
Developing access to basic services is the priority initiative for 23% of leaders and 9% of households, ranking it as number one priority for authorities and number four for households. In the MIEC report the two subcomponents of rehabilitating social service infrastructure and improving the quality of basic service delivery are very much in line with the preferences of households and authorities. It would be worth rethinking budget allocation over time on the other hand, as it seems that too little budget is being allocated to the short-term. Only 14% of the budget allocated for rehabilitating social service infrastructure has been reserved for Years 1-2. It looks a bit better with regard to improving the quality of basic service delivery where 38% are allocated to Years 1-2. Given the importance of access to basic services for peace it may be advisable to allocate a greater share of the budget to the short to medium term interventions.
Overall, the budget allocation raises questions whose priorities are reflected in the Peace Accord and its operationalization in the MIEC costing. Road infrastructure receives the lions' share of the budget allocation while very little is dedicated to employment creation. The economic rationale is already relatively weak for the road to Timbuktu while the road to Kidal seems mostly informed by political and security objectives. It is noteworthy however, that the activities proposed to be financed during the first two years are much closer to popular preferences, while the MIEC has shifted those elements in the Peace Accord that have less popular support to the future. In short, both the MIEC Strategy and the Peace Accord do not fully reflect the preferences of the households and authorities in the north, but the MIEC strategy makes an attempt to make the Peace Accord more inclusive, and with respect to the immediate term does a reasonable job reflecting popular preferences, with the important exception of too little emphasis on job creation and too much on road infrastructure.
The MIEC would reflect popular preferences even more if more attention were given to job creation. Youth employment schemes and labor-intensive public work programs as short-term approaches have successfully been tried in numerous places including Côte d'Ivoire and the Central African Republic. The MIEC report does acknowledge this in its descriptive content but then does not reflect its importance in budgetary allocations. Bringing these activities to north Mali would not only help build confidence in the Peace Accord, it would also offer alternatives to earning a living through illicit activities. Such approaches have been found to be very successful. 30

VI. Conclusion
We find that the majority of the population of northern Mali supports the Peace Accord even though the Accord does not offer independence to Azawad or propose a federal solution. So, we note that while the 2015 Peace Accord had all the elements to be an exclusive elite bargain negotiated and signed by leaders of rebel group coalitions with an unclear mandate, the priorities outlined in the Peace Accord largely reflect those of the population. This is encouraging as an inclusive Peace Accord has a much greater probability of success.
One priority that is not shared by citizens is the strong emphasis on the construction of road infrastructure, the rationale for which is both security and political integration and economic development. The disconnect between priorities in the Peace Accord and preferences of citizens and local leaders becomes more evident in the costing of Peace priorities in the MIEC report. Forty-five percent of the cost of the implementation of the Peace Accord is accorded to road construction. Nonetheless, the MIEC report makes an attempt to rebalance the prioritization of development interventions brought forward in the Peace Accord with the priorities of citizens and local leaders in northern Mali as its costing for the first two years of implementation aligns closely with the preferences of citizens. The other aspect that remains under-represented is job creation.
Among citizens we find considerable heterogeneity of opinion. The wealthiest and those who lost least as a consequence of the crisis are much more interested in security initiatives than the poor and those who lost most. The poorest on the other hand care much more about socioeconomic activities. The average citizen in north Mali as well as refugees in Niger consider job creation as one of the most important socioeconomic initiatives. Refugees in Mauritania do not care much about job creation. They prefer a northern development zone.
Kidal is the only region where households and authorities do not prioritize the redeployment of the armed forces, a preference they share with refugees in Mauritania but not with those in Niger. The vast majority of leaders (89%) prioritize the participation of the northern population in the management of local security, while refugees in Mauritania prefer to participate in the national army.
The paper also looks into whether local leaders reflect the preferences of their citizens and finds that local leaders generally speak for the people they represent (a useful feature when it comes to planning development initiatives). Local leaders reflect the priorities of citizens regarding governance (greater responsibility in managing their own affairs and security (the gradual redeployment of the armed forces). Local leaders also prioritize the same three socioeconomic development initiatives as citizens: access to basic services, job creation and rural development, and food security and the environment.

Annex: Multi-nominal logit of development priorities
A development strategy that puts north at par with rest of the country (base outcome)

Rural Urban
Coeff.