Building a Safety Net: Explaining the Strength of Ex-Military Networks

Abstract The disarmament, demobilization, and reintegration of ex-combatants has become an integral part of peacebuilding. Although the main purpose of such interventions is to dissolve the military structures of armed groups, there is growing evidence that ex-combatant networks often remain intact. We investigate why such structures continue to thrive. We argue that ex-military networks are stronger when ex-commanders have weak links to elite patronage systems. Ex-combatants who are unable to rely on their former superiors for economic assistance must instead build denser ties to each other to gain access to a social safety net. To assess our argument, we conduct a comparative social network analysis (SNA) of two ex-military networks in Liberia. This innovative approach helps us uncover previously overlooked, but central, dynamics related to ex-combatant groups. We thereby show that SNA provides a range of underutilized tools and exact definitions that can increase our understanding of ex-military networks.

targeting a larger number of individuals, to prevent renewed hostilities. Put differently, while both strong and weak networks create challenges for peace-neither is intrinsically more peaceful or violent-they merely require different types of peacebuilding interventions.
Despite the challenge ex-military networks can pose to sustainable peace, there is a lack of theories regarding the resilience of ex-military structures. In fact, previous research has confined itself to describing the prevalence of ex-combatant networks without systematically investigating how they are sustained. The research question guiding this article is, therefore: Why are some ex-military networks stronger than others? The type of ex-military structures we focus on are those found in the postwar city, as ex-fighters are often drawn to such environments in search of employment, education, and contracts for new military missions. 9 As a consequence, ex-combatants networks often thrive in the city, where ex-MiLCs function as "hubs" sharing information with their ex-combatants about opportunities.
We argue that ex-military networks are stronger when ex-commanders have weak links to elite patronage systems. Ex-combatants are then unable to rely on their former superiors for economic assistance and must instead build denser ties to each other to gain access to a social safety net. Central to this argument is the desire of physical laborers-the most marginal group of exfighters-to address their socioeconomic vulnerability. Without the economic backing of their ex-commanders, physical laborers are obliged to invest more time in building multiple relations with more well-off ex-combatant peers, subsequently making the network stronger. Conversely, if ex-commanders possess strong elite connections, physical laborers prefer to interact with their former superiors, subsequently making the network weaker. To assess the explanatory value of our argument, we conduct a structured focused comparison of two ex-combatant networks in Liberia, drawing on unique individuallevel interview data and social network analysis (SNA). This article makes four major contributions. First, we highlight that exfighters predominantly "stick together" for economic reasons-either geared toward vertical patronage networks or horizontal exchange networksunderlining the productive role ex-military networks can have in stabilizing postwar societies. Hence, in many instances, it may be better to leave exmilitary networks intact rather than attempt to disrupt them. This finding underlines how international peacemakers must reconsider contemporary policies, which emphasize the need to uproot ex-military structures. Second, we offer a nuanced explanation for the strength of ex-military networks rooted in methodological individualism, which accounts for social relations at the group level. We thus demonstrate how individual incentives translate into particular network structures. Third, we use an innovative approach, in that we are, as far as we know, the first to use SNA to study exmilitary networks. We thereby show that SNA provides a range of underutilized tools and exact definitions of key concepts that can increase our understanding of ex-military networks. Particularly noteworthy is that we have acquired comprehensive relational data that allow us to analyze and compare two complete ego networks, composed of sixty-four individuals. Finally, our results highlight that even if ex-military networks look similar on the surface, they differ significantly when it comes to how horizontal and vertical ties are structured. Contrary to previous assumptions, it is not necessarily the networks with the most robust ex-command structures that are the strongest; due to lower levels of horizontal interactions, such networks tend to be less dense. Efforts by peacemakers to engage ex-combatant structures must, therefore, be based on an analysis that takes these variations into account.

Ex-Military Networks in Previous Research
Since the early 1990s, DDR has been an integral part of the peacebuilding template. 10 Whereas disarmament and demobilization have traditionally focused on the collection of arms and disbanding armed units, the objective of reintegration has been to increase the opportunity costs for violence by providing economic incentives (training, education, reinsertion money) to ex-fighters. Over time, practitioners and scholars have given greater emphasis to the social dimensions of reintegration. 11 This development is the result of a growing awareness of the need to mend relations between communities and ex-combatants and the role social capital has in generating economic opportunities. In addition, due to the increased complexity of many contemporary interventions-with continued fighting and a proliferation of armed groups (for example, in Mali and Somalia)-scholars increasingly stress the need for a more flexible approach to DDR. 12 Despite the attention given to DDR, there is ample evidence that ex-military networks continue to thrive long after the formal dissolution of armed groups. The prevalence of such ties has been observed in countries ranging from Columbia, Sierra Leone, and Burundi to Afghanistan, Indonesia (Aceh), and Mozambique. 13 Ex-combatant networks are usually composed of both horizontal ties (relations between ex-combatants) and vertical ties (links binding ex-combatants to their ex-commanders). In Namibia, horizontal ties flourished between ex-rebels who settled in the same urban areas after the war. 14 In such ex-combatant communities, ex-fighters habitually interact on a daily basis and meet at local hangouts. 15 Meanwhile, in countries such as Afghanistan and Indonesia (Aceh), the presence of vertical ties has allowed ex-fighters to retain contact with national elites. 16 Such structures provide ex-fighters with the channels needed to lobby political leaders.
In the previous literature, four key factors are assumed to affect the strength of ex-military networks. Several authors have stressed the central role ex-military networks have in providing socioeconomic goods, such as employment, food, loans, and access to elite patronage systems. 17 Another reason for maintaining relations with ex-military colleagues is that such relations provide security in what are often turbulent postwar societies. 18 Meanwhile, studies have also highlighted that ex-combatant networks constitute a vital source of friendship. Having access to former comrades provides ex-fighters with a vital coping mechanism during transitions from military to civilian life. 19 Finally, some authors point to the central role trust-generally a function of having fought side by side during war-plays in sustaining ex-military networks. 20 As such, levels of trust should be higher in strong networks.
Even though previous research has made a vital contribution by beginning to outline the inner workings of ex-military structures, they suffer from several shortcomings. First, most studies do not clearly define ex-military networks and do not systematically compare different structures. This impedes efforts to assess the explanatory value of the proposed variables. Second, studies that have a more systematic and comparative approach either employ the strength of ex-military networks as an independent, rather than dependent, variable, 21 or confine the analyses to explaining the resilience of ex-command structures. 22 The latter approach's problem is it obscures the possibility that there may be dense horizontal ties between exfighters and that lower-ranked individuals can challenge the typically more privileged positions of ex-commanders. Failing to acknowledge the central role horizontal relations have in sustaining ex-military networks can have severe implications for postwar stability. If peacemakers only focus on vertical ex-command structures, there is always the risk that spoiling elites align themselves with ex-fighters who have positioned themselves as key broker figures and co-opt the network for violent purposes. 23 Both horizontal and vertical relations can be employed for violent purposes, but a key question, which we examine in this article, is to what extent peacemakers must adjust their policies depending on whether they are faced with a network characterized by intense ex-combatant interactions or a network where ex-fighters gravitate toward their ex-commander.
Hence, this calls for a research approach that systematically compares different ex-military networks and includes both the vertical and horizontal dimensions of such structures. This will make it possible to unpack the nuts and bolts of ex-military networks.

The Strength of Ex-Military Networks
In this study, we focus on ex-MiLCs and the ex-combatants they continue to interact with on a regular basis. We define the former as "the military personnel that was previously situated between the rank-and-file combatants and the highest military leadership, and who personally led their subordinates in battle," 24 while ex-combatants are held as individuals who have taken direct part in previous hostilities on behalf of a warring party and have either been discharged from or voluntary left the military group in which they served. We focus on ex-MiLCs and their former subordinates for two reasons. First, cooperation in war tends to "generate a sense of belonging and comradeship among combatants." 25  can play a crucial role in giving meaning to wartime deeds and filtering peacetime experiences. 26 Hence, due to the presence of a collective identity, we can suspect relations between ex-fighters of the same unit are particularly dense. Second, ex-MiLCs are often central to the upkeep of ex-military networks. 27 Even if ex-MiLCs differ in the influence they possess (for example, as patrons or security providers) they generally function as "hubs" in ex-combatant networks. Known by both ex-fighters and elites, they can disseminate information and help link groupings of ex-combatants to different elites and other ex-military colleagues. Due to the clout ex-MiLCs develop over their subordinates during war, it is predominantly with the latter-rather than ex-combatants of other units or factions-that they tend to interact.
More specifically, we analyze ex-military networks in the context of the postwar city, as ex-combatants tend to flock to urban areas after demobilizing. 28 The allure of the city lies in its potential as a "launching pad" for individuals seeking to carve out new lives for themselves. 29 It is in the city that many ex-fighters, and others living in the margins, can gain access to education, capital, business opportunities, and elite patrons. 30 Melting into the city can be particularly attractive for ex-combatants who have committed wartime atrocities in their home communities. 31 As newcomers, excombatants often stick together and assist each other in finding ways to navigate the city. During these interactions, ex-commanders habitually play a crucial role. Ex-combatants tend to pay homage to their ex-commanders to get the latest news, receive help to interpret the meaning of political events, or gain information about economic opportunities. 32 In this sense, postwar urban milieus can best be described as "the city as barracks," where ex-combatant relations continue to thrive. 33 Even if shared war-and peacetime experiences in an urban environment can create a common sense of belonging, ex-combatants vary in how they relate to their former colleagues. While many ex-fighters continue to socialize with former comrades, others work hard to distance themselves from their military past. 34  and size. In this study, we try to capture these differences by analyzing the strength of ex-military networks. Here network strength refers to the level of personal ties-through regular meetings-between ex-combatants (horizontal) and between ex-fighters and the ex-commander (vertical). Hence, in a strong network, a large portion of the network members (ex-fighters and the ex-MiLC) are in direct contact with each other. Meanwhile, in a weak network, there is less direct contact between network members. Since ex-fighters generally outnumber ex-commanders in any given network, horizontal ties predominantly define the density, or strength, of an exmilitary network. 35 Figure 1 illustrates the key difference between strong and weak ex-military structures by depicting two ideal types of ex-military networks. The network to the left represents a weak network that completely lacks horizontal ex-combatant relations, and all interactions are geared toward the ex-commander. The network to the right, on the other hand, represents an extremely strong network where all ex-fighters, as well as the ex-MiLC, interact with each other. Note that real-world networks will most likely be found somewhere in between these two extremes. should be compared to the 30 potential ties between the ex-fighters and the ex-commander. The equation for calculating potential ties is: g(g-1). Another reason for why the number of horizontal ties predominantly affects network strength is that we define an ex-military network as "an ex-MiLC and those ex-fighters that the former meets on a regular basis." Consequently, all ex-fighters have at least some contact with their excommander. This entails that the number of vertical ties is generally more or less constant, as long as similar sized networks are compared. One exception is if "regular meetings" are operationalized very narrowly (for example, an actual tie is present only when ex-MiLCs and fighters interact at least once per day). In such situations, there may be a discrepancy between the amount of actual and possible vertical ties. See the "Methodological Considerations" section for information on how we operationalize and measure network strength. 36 The red nodes represent the ex-commanders and the grey nodes their ex-fighters. The edges represent personal interactions. Vertical ties dominate the weak network, whereas dense horizontal ties characterize the strong network. Clustering algorithm: Fruchterman Reingold.
The rationale for focusing on personal contact is that such interactions are generally seen as a defining component of social networks, and the density of interactions is widely used to measure network strength. 37 This does not mean alternative conceptualizations of networks-such as friendship, trust, or economic dependence-are unimportant. 38 However, since previous research has largely described such ties as influencing ex-combatants and commanders' willingness to "stick together," it makes sense to treat them as potential independent variables rather than as indicators for network strength.

Building a Safety Net
We argue that ex-military networks are stronger when ex-commanders have weak links to elite patronage systems. Ex-combatants who are unable to rely on their former superiors for economic assistance must instead build denser ties to each other to gain access to a social safety net. This is particularly true for physical laborers-the most exposed ex-combatantswho are obliged to seek favors from more well-off ex-fighters. As such, the individual utility maximization of physical laborers is central to the causal logic of our argument. Before further outlining the theoretical argument, it is first necessary to gain a deeper understanding of how ex-combatants use social relations to get by in the postwar city.
Ex-fighters opting for a life in the city must acquire the skills needed to navigate in a fluid and unwelcoming environment. 39 In many developing countries, rapid urbanization and poor governance leave inhabitants exposed to homelessness, unsanitary living conditions, criminality, and corrupt government officials. 40 Due to an excess of laborers, newcomers often end up in low-income jobs, such as informal vendors, car loaders, or carriers at work sites, making them vulnerable to chronic poverty and bodily harm. 41 As a result, ex-fighters, along with other downtrodden residents, often "operate without clearly delineated notions of how the city is to be inhabited and used." 42 Add to this ex-fighters also needing to cope with the challenges of transitioning from a military to a civilian life. Not only do ex-combatants need to acquire new social norms about acceptable peacetime behavior, they tend to lack the skills to succeed on the formal job market. 43 There is, furthermore, often a social stigma associated with being "outed" as an ex-combatant. 44 To manage the vulnerability of the postwar city, ex-combatants are often obliged to invest in social relations. Without access to formal welfare provision, kin and friends become a vital source of support. 45 Here it is possible to distinguish between two types of social relations that can be cultivatedvertical and horizontal. The former concerns relations to influential elites (national politicians or businessmen, army generals, senior government advisors, or bureaucrats) who, due to their position in informal or formal economic-political structures, accumulate substantial resources. In fact, in many developing countries, access to employment and education is a function of having elite connections. 46 The problem for many ex-fighters is that they have few elites-besides their ex-military leaders-to whom they can turn. However, ex-fighters can seldom rely on their factional leaders for assistance, since the latter do not have the resources to pay back the "debt" accrued to all wartime followers. It is, therefore, crucial that ex-combatants reach out to new elites, such as civilian politicians or businessmen. This is difficult, however, since many ex-fighters do not move in the same social circles as leaders. To bridge such "structural holes," ex-fighters are commonly dependent on their ex-MiLCs. 47 Since the latter often have relations with both ex-combatant communities and elites, they can "play a vital role as translators … allowing elites and grassroot groups to find enough common ground to trust each other." 48 Studies have also highlighted how state elites sometimes employ ex-MiLCs as government brokers to funnel employment, money, and food to ex-fighters and incorporate the latter into the state-building process. Hence, for many ex-combatants, investing in vertical relations is largely about cultivating ties to their ex-MiLCs. 49 Ex-fighters can also seek to overcome the challenges of urban vulnerability by building horizontal relations. A major finding in anthropology and sociology is that the urban poor tend to develop exchange networks to help each other. 50 48 Themn er and Utas, "Governance through Brokerage," 262. 49 Themn er, "Wealth in Ex-Combatants." 50 Doherty, "Life (and Limb) in the Fast-Lane"; Simone, "People as Infrastructure"; Stack, All Our Kin. serve as an occasion to cultivate an ethics of mutuality." 51 Via exchange networks, marginalized people can help each other with food, small exchanges of money, and the sharing of living expenses. 52 A central component of such economic interactions is principles of reciprocity and sharing. Even if ex-fighters also have relations with individuals who did not take part in the war-such as relatives, neighbors, and childhood friendsex-military colleagues often continue to be their closest peers. Approximately ten years after the demobilization process in South Africa, an estimated 53 percent of all ex-fighters met their former comrades on a daily basis. 53 Experiences have shown that ex-military-based exchange networks constitute a vital safety net for ex-fighters, providing them with food, loans, and assistance in times of need. 54 To address the everyday challenges of urban life, ex-fighters need to decide how much to invest in vertical versus horizontal relations. This form of utility maximization is particularly important for ex-fighters who are physical laborers-for example, carriers, loaders, domestic laborers, and prostitutes-whose only asset is their bodies. Physical laborers' dependence on their physique make them more vulnerable to injuries and illness, which can deprive them of their livelihood. 55 In addition, laborers often lack the capital and skills needed to engage in activities that can help them out of poverty. This is in contrast to informal traders (for example, individuals who sell used clothes and rice), whose invested capital has the potential to grow, and more gainful earners (here held as individuals such as skilled workers, formally employed individuals, and business owners) who generally have more stable incomes. In addition, since members of the two latter categories are less dependent on their physique, they are not as exposed to bodily harm. Hence, compared to ex-combatant physical laborers, informal traders and gainful earners are less dependent on social relations to get by.
When ex-commanders have weak elite connections, ex-combatants are obliged to give greater attention to building relations with each other to get by. Without access to elite patronage, ex-combatants cannot rely on their ex-MiLCs to carry them out of destitution. However, favor-seeking among ex-fighters is unlikely to be evenly distributed. Instead, interactions will be clustered around certain key individuals. 56 This is because "mutuality is not guaranteed" in exchange networks, resulting in an uneven distribution of resources between members. 57 Since physical laborers live more in the 51 Doherty, "Life (and Limb) in the Fast-Lane," 194. 52 Simone, "People as Infrastructure"; Stack, All Our Kin. 53 Guy Lamb, "South Africa: Reintegration into Civilian Life of Ex-Combatants," News from the Nordic Africa Institute 3 (2003): 11-13. 54 de Vries and Wiegink, "Breaking Up and Going Home?" 55 Doherty, "Life (and Limb) in the Fast-Lane"; Simone, "People as Infrastructure." 56 Doherty, "Life (and Limb) in the Fast-Lane"; Simone, "People as Infrastructure." 57 Doherty, "Life (and Limb) in the Fast-Lane," 194. margins than informal traders or gainful earners, they are more prone to approach ex-combatant peers for support. In this quest, they are particularly likely to reach out to ex-fighters who have positioned themselves as gainful earners, as the latter are more reputed to have a stable income. Informal traders often face the prospect of having their goods stolen by thieves or confiscated by police officers. 58 Physical laborers' desire to attach themselves to gainful earners can also elevate the latter into key network brokers; by building up their own circle of clients, they hold the potential to challenge the typically dominant position of ex-MiLCs. Since there are limits to how much assistance grassroot individuals-even gainful earners-can provide, physical laborers must build multiple horizontal ties to get by. 59 As a result, bonds between ex-combatants are likely to be denser in networks where ex-MiLCs have failed to attach themselves to elites, thus making the network stronger.
The situation is markedly different in networks headed by excommanders with strong ties to elite patrons. Under such circumstances, physical laborers-as well as informal traders and gainful earners-are likely to prioritize interacting with their ex-MiLC rather than with each other. This is because such interactions are likely to generate more coveted forms of assistance, such as employment and introductions to resourceful elites. 60 This should be compared to the petty money and food that horizontal relations with other ex-fighters can generate. Put differently, exchange networks do little to lift physical laborers out of poverty; they merely help them get by. 61 In this process, the value of ex-MiLCs' elite connections is not only determined by the quantity of resources excommanders distribute; it is also assessed by the credibility of future promises of assistance. 62 Such promises are, for instance, likely to be particularly reliable if ex-commanders are close to key elites working in government or standing as major contestants in national elections. 63 Above all, by attaching themselves to "connected" ex-commanders, physical laborers can keep the hope of social advancement alive. 64 Hence, due to the greater rewards of befriending their ex-commander-as compared to other ex-fightersphysical laborers will primarily gear their attention toward the former. This significantly impacts the structure of the network. Since physical laborers have fewer incentives to attach themselves to gainful earners, it will be 58 Simone, "People as Infrastructure," 423. more difficult for the latter to position themselves as key brokers in the network. Consequently, ex-commanders are more likely to retain control over the ex-military network, as there are limited horizontal interactions between ex-fighters, and the network will also become weaker.

Methodological Considerations
The wider population of cases this study speaks to is ex-military networks in fragile postwar urban environments, characterized by poverty and elite patronage politics. Cities such as Brazzaville, Dili, Kinshasa, Freetown, and Monrovia fit this profile. 65 For the purposes of this article, we focus on exmilitary networks situated in Monrovia (Liberia). We do this for two reasons. First, previous research has highlighted how ex-combatant networks have continued to thrive in Monrovia despite the cessation of hostilities in 2003. 66 Monrovia, therefore, offers a pool of relevant networks from which to choose. Second, one can largely see Monrovia as a typical fragile postwar city, struggling with problems of under/unemployment, a lack of formal housing, and unsanitary living conditions. Furthermore, state institutions are notoriously weak in Liberia; as a consequence, patronage politics not only determine the outcome of political processes but also ordinary people's access to work, contracts, and education. 67 More specifically, we analyze the experiences of two ex-commanders-Edwin and Paul 68 -and their ex-combatant networks. We selected these ex-MiLCs because they share many similarities: they both (a) fought for President Charles Taylor's Navy Division (largely composed of ex-fighters of National Patriotic Front of Liberia (NPFL)); (b) were considered courageous and capable commanders; (c) had similar-sized networks (Edwin had 30 and Paul 32 ex-combatants); 69 (d) sought to build relations with both elites and their ex-fighters; and (e) were, economically speaking, relatively well-off. In addition, evidence suggests Edwin and Paul are somewhat representative of the broader population of ex-commanders in Liberia; from a prior survey we know they share several key characteristics with other ex-military commanders in Liberia. 70  those surveyed were males; 58% were born between 1969 and 1978; 76% were at least partly dependent on the informal sector to make a living; 80% were politically active; 82% felt accepted by their community; and 87% lived in a family situation (with a spouse and at least one child). To determine the network boundary, we asked the two ex-MiLCs to provide a roster of all the ex-combatants with whom they were in regular contact (that is, not ex-fighters who they only met by chance). It was the excommanders who subsequently helped the project team set up interviews with all their respective ex-fighters. Both the ex-commanders and each excombatant were then interviewed individually.
To assess the explanatory value of the article's main argument, we predominantly make use of SNA. Sociologists and anthropologists have long used SNA, but the "network turn" in political science is more recent. Although SNA has recently proven its utility for increasing our understanding of interstate relations 71 and terrorist networks, 72 it has not made headway in the study of ex-combatants. One noteworthy exception is Zoe Marks and Patrycja Stys, who have employed SNA to analyze the reintegration of ex-fighters. 73 We are, however, to our knowledge first to apply SNA to ex-military networks. SNA has several advantages compared to other approaches. First, SNA provides a range of methodologically exact tools to assess and analyze networks, as the approach offers formal definitions and ways to measure relationships. 74 Second, SNA is ideally suited to explain social relations, and it can simultaneously pertain to different levels of analysis. 75 Alternative approaches often need to assume that units are independent of each other and focus on individual characteristics rather than relational properties.
"Nodes" and "edges" are two key terms in SNA. Where the former concerns the actors in a given network, the latter constitutes the ties connecting these actors. In this study, the nodes are the individual ex-fighters/ commanders and the edges the dyadic ties-based on, for example, personal interactions, economic transactions, trust, or friendship-binding excombatants/commanders together.
The strength of Edwin and Paul's networks is determined by measuring the levels of personal contact-in the form of meetings-between its members. A relationship between two ex-fighters/an ex-fighter and commander is defined as being present when a person states that they meet each other at least once per year. 76 To gain information on this indicator, we asked all ex-combatants and commanders, "How often do you meet each person on the list?" 77 More specifically, we rely on descriptive SNA and compare the density scores of the interactions in the selected networks. The higher the density score a network has, the stronger it is. Network density is based on the proportion of edges present in the network and is calculated as the number of edges (E) divided by the possible number of edges. Since an edge is an ordered pair of nodes, there are g (g À 1) possible edges. The density ranges from 0 (when there are no edges present) to 1 (when all possible edges are present). If the density score equals 1, the network is complete and all nodes state that they interact with each other. 78 Since we focus on ego networks, all ex-combatants automatically share a tie with their own ex-commander.
To assess our main explanatory factor, we employ a more qualitative indicator. Ex-commanders are seen as having strong links to elite patronage systems when they have close relations with relatively many key leaders (national politicians or businessmen, army generals, senior government advisors, or bureaucrats) who employ them to distribute patronage (employment, money, food, clothes, shoes, medicine), or promises thereof, to ex-fighters. Of particular importance is if these elites are presidents, key presidential candidates, or government ministers, as commitments to provide future assistance are likely to be particularly credible when coming from them.
To assess the claim that ex-fighters invest more in horizontal economic relations when their ex-MiLCs have weak ties to elite patronage networks, we compare how extensive economic interactions are between the two excombatant networks. For this purpose, we asked all ex-fighters, and their ex-commanders, who in the network they would go to if they needed a small sum of money. Based on our theoretical argument, ex-fighters who are members of a network with weak ties to elites should be more willing to ask each other for money. We capture this push for economic assistance by establishing the density scores of the two networks when operationalized as "who asks who for money" (to which extent all ex-fighters/commanders ask each other for assistance). Since ex-fighters in networks with limited 76 Research has shown that people are bad at recollecting how often they meet others (see Stephen P. Borgatti, Martin G. Everett, and Jeffrey C. Johnson, Analyzing Social Networks (London: Sage, 2018)). To minimize the risk that we omit relevant ties, we decided to set the threshold at once/year. It is important to note that the relative difference in network strength between the two ex-MiLCs-measured through density scores-is still significant with a once/week threshold (once/year: 0.605 (Edwin) and 0.229 (Paul); once/week: 0.392 (Edwin) and 0.171 (Paul). 77 The list being the roster their ex-MiLC provided. 78 Wasserman and Faust, Social Network Analysis, 129. elite connections should be more willing to ask others for money, we expect this type of network to have higher density scores.
To evaluate the claim that it is predominantly those ex-combatants who are physical laborers who spur horizontal economic interactions in networks with limited elite connections, we compare average in-and outdegree scores (when it comes to asking for money) of ex-fighters who belong to different work categories. In-degree refers to the number of edges that a particular node receives, whereas out-degree relates to how many edges that are being sent. 79 These are arguably the most widely used centrality measures in SNA. Prominent network actors tend to have high indegree (for example, other actors contact them/ask them for money), whereas actors with high out-degree tend to be dependent on others (for example, they contact other/ask other actors for money). More specifically, we compare in-and out-degree scores of three work categories: (1) Physical laborers: ex-fighters who wholly or partly work as car/truck/ship loaders, construction site carriers, car washers, domestic workers, or prostitutes; (2) Informal traders: ex-combatants who peddle goods (for example, used clothes, drinks, sand, or gasoline) as unregistered sellers; 80 (3) Gainful earners: ex-fighters who have a more stable income and do skilled work (carpenters, plumbers, auto mechanics, electricians, motorcycle taxi/car/ truck drivers), are formally employed (for example, security guards), own formally registered businesses, or live off of overseas remittances. Since we expect that it is predominantly physical laborers who ask other ex-combatants for money, they should, on average, have higher out-degree scores than other categories of ex-fighters. Conversely, because the former tend to target gainful earners when soliciting help, the latter should have higher indegree scores. Meanwhile, in a network where the ex-MiLC has strong elite connections, there should be relatively less of a difference in average outand in-degree scores between various work categories.
The empirical material is based on interviews conducted in Liberia during May-November 2017 and March-June 2018. 81 The interviews were semi-structured, following a list of preset questions that were sufficiently open to allow the respondents to develop their answers and the interviewer to pose follow-up questions. The two ex-commanders were interviewed on several occasions; each ex-fighter (sixty-two in total) was interviewed at least once (several ex-combatants were also interviewed two or three times to gain additional information). After providing a brief historical overview 79 Wasserman and Faust, Social Network Analysis. 80 Informal traders sometimes engage in activities falling under the gainful earner category (for example, carpentry). Commonly, these are individuals who do not receive enough contracts to live off their skilled work. Since they are more exposed economically than other gainful earners, we code them as belonging to the category "informal trader." 81 We also make use of interviews conducted with Edwin, Paul, and their ex-fighters in 2011. of post-civil war Liberia, we devote the following section to comparing the experiences of the two selected ex-military networks.

Post-Civil War Liberia
The second Liberian civil war ended in August 2003 with the signing of the Accra Comprehensive Peace Agreement. The accord called for the DDR of forces loyal to Charles Taylor (often referred to as NPFL), Liberians United for Reconciliation and Democracy, and Movement for Democracy in Liberia; the stationing of United Nations peacekeepers; and the creation of an interim government (2003-6). The 2005 presidential elections resulted in a victory for Ellen Johnson Sirleaf (Unity Party; UP). The success of UP, and other political parties not directly associated with the war, obliged many ex-fighters to find new ways to integrate themselves into the country's elite patronage system. This was particularly vital due to the inability of the DDR program-which included education and skillstraining programs-to generate employment for ex-fighters. As a consequence, many ex-fighters relocated to Monrovia. 82 Although Johnson Sirleaf and UP defended their hold on power during the 2011 elections, six years later the party lost to a coalition which, among others, included Congress for Democratic Change (CDC) and ex-Taylor loyalists of National Patriotic Party (NPP).
In the following, we compare the postwar trajectories of two ex-MiLCs and their networks to explain why some ex-combatant networks are stronger than others. break when he was employed as one of the chief security providers for Alex Cummings-the former executive vice president of Coca-Cola and standard-bearer for Alternative National Congress (ANC) for the upcoming 2017 presidential elections. After the war, many of Paul's ex-fighters established themselves in the same community as their former superior. By 2017 as many as two-thirds of the ex-fighters in Paul's network lived in the same area. By this time, the network was composed of thirty-two ex-fighters. Interestingly, there was relatively limited horizontal interaction between his ex-fighters (see Figure  2). In fact, their attention primarily focused on Paul. As a consequence, Paul's network was comparatively weak, which is also reflected in the low density score of the network (0.229). What is more, when comparing the node sizes in Figure 2, it becomes apparent that Paul was the undisputed head of the network. Since the small cliques of ex-fighters were rather isolated from each other, Paul was in position to control the flow of information within the ex-combatant community. Paul's role as an intermediary also manifested itself geographically, as his house and the ANC headquarters-where he often hung out-were the natural meeting points for exfighters who wanted to hear the latest gossip.  owned. After the latter was closed, Edwin's main source of income came from a tire repair shop he set up. In addition to these activities, Edwin, at times, also engaged in a range of other businesses-he owned two motorcycle taxis, sold marijuana, rented out rooms in his house, and took contracts to clear farmland.

The Strength of Two Ex-Military Networks
After leaving the Navy Division, many of Edwin's ex-fighters ended up in the same community as their former superior: in 2017 as many as threefourths of the ex-fighters in Edwin's network lived in the same suburb as him. In total, the network was composed of thirty ex-fighters. What is noteworthy about the network (see Figure 2) is the high level of interaction between ex-fighters. Due to these intense ties, Edwin's network was relatively strong, with a density score of 0.605. This did not mean that Edwin was an unimportant figure in the life of his ex-fighters. On the contrary, many made a habit of visiting Edwin's house or tire repair shop. However, when analyzing the distribution of personal interactions (by comparing the node sizes in Figure 2), it becomes clear that some ex-fighters also played a crucial role in upholding the network. As such, Edwin was not in a position to monopolize communication between his ex-fighters.

Paul: Embedded in Elite Networks
From the end of the war, Paul was able to embed himself in the patronage networks of multiple postwar leaders. This was particularly true for ex-Taylor stalwarts-such as Cyril Allen (businessman and NPP dignitary), Roland Duo (ex-NPFL general and Johnson Sirleaf's security advisor), John Gray (NPP vice president of the transitional government), Jewel Howard-Taylor (Taylor's ex-wife and NPP senator), and Benoni Urey (businessman and ex-commissioner of maritime)-who had accumulated substantial resources during the Taylor NPP regime. Thanks to these connections, Allen commissioned Paul to guard the real estate he owned, and Gray made him head of security for the Ministry of Public Works. Paul's elite connections also benefitted his ex-fighters. Individuals such as Duo, Howard-Taylor, and Urey asked Paul to distribute cash, rice, and scholarships to his ex-combatants on their behalf. In addition, many elites turned to Paul when they needed to mobilize ex-fighters to take part in rallies, clean local communities, and work on farms and construction sites.
With time, Paul was able to branch out to elites not associated with Taylor. This included politicians such as Joseph Boakai (vice president of Liberia), Cummings, and George Weah (head of CDC), as well as wealthy businessmen such as George Haddad and Musa Bility. Each employed Paul's services to mobilize his ex-fighters for various endeavors. Of particular importance was Paul's newfound friendship with Cummings. On several occasions, the latter tasked Paul to find people to do community projects, work at the ANC headquarters, and participate in ANC campaigns. When asked, Paul used the money he received from Cummings to employ his ex-fighters. In addition, thanks to his ANC connections, Paul was able to offer kickbacks-money, food, clothes, and shoes-to ex-fighters who joined the party, as well as extend promises of employment and education if Cummings became president.
Due to Paul's strong links to elites, his ex-fighters devoted much attention to their ex-commander. Of particular importance was Paul's aptitude in identifying job opportunities. According to one ex-combatant, "When it comes to jobs and security business in the community, Paul is the man … people know that Paul is a star, a big man, they usually go to him to ask 'Do you have a guy [somebody that they can employ]?'" 85 In addition, Paul also made it a habit to assist his ex-fighters with money, clothes, and food. The main benefactor in the network was also clear. While all ex-fighters declared that they would go to Paul if they needed money, Paul could only consider soliciting four (13 percent) of his ex-combatants. In addition, Paul's ex-fighters were willing to change political allegiances based on the anticipation that he would introduce them to new patrons. When Paul decided to leave NPP in favor of ANC in 2016, he made a declaration on Facebook and organized a march to ANC headquarters. Many of the rally participants were his ex-fighters, who later became members of ANC. In fact, as many as twenty-eight (88 percent) of the ex-fighters in his network joined the party (see Figure 3). The belief that their well-being was connected to Paul's was lucidly expressed by one ex-fighter: "We know that Paul knows big people and when he succeeds, we all succeed along with him." When asked why he had joined ANC, he simply responded, "Paul is there and things will happen." 86

Edwin: Shunned by Liberia's Postwar Elite
After the end of the civil war, Edwin struggled to gain access to elite patrons. This was, however, not due to a lack of trying. In fact, he habitually contacted key leaders such as Roland Duo, one of the main benefactors of ex-NPFL combatants. According to Edwin, Duo was not at all receptive: "No, even my former chief of staff Roland Duo does not do that [reach out to him]. Like, for instance, one day my deputies and I met at the junction and decided to visit the chief of staff, and luckily we met him there. But he told [Edwin and his deputies] that he did not have money and he was busy running his campaign, because he wanted to become a representative." 87 It was not until 2013-when a friend introduced him to George Solo, chairman of CDC-that Edwin was able to get "connected." Solo proved to be a valuable contact, willing to help Edwin with financial support when in need. When Solo fell out with CDC and joined the All Liberian Party (ALP)-a newly formed party headed by Urey-Edwin was obliged to follow him so as not to lose his newfound patron. Through this connection, Edwin and some of his ex-fighters were, on at least one occasion, paid to placard ALP posters during the 2017 elections. In addition, Edwin extended promises of employment-if ALP won the elections-to ex-fighters who joined the party. Edwin was, however, not in a position to distribute any other benefits to reward recruits.
In the end, Edwin's ALP connection proved problematic; it was difficult for him to cash in on his association with ALP, since he needed to go via Solo to access Urey. To some extent, this was tempered by Edwin and Duo reestablishing their relation during an NPP convention meeting in 2016. After the meeting, Duo reached out to Edwin on a few occasions to employ his ex-fighters to unload containers at the port and work on his farm. However, when asked about his rapport with Duo, Edwin responded, "Our relationship is not strong like before, when we used to visit one another. Edwin was still bitter that Duo had abandoned him after the war. Even if Edwin succeeded in building relations with a few additional elites, such as Sando Johnson (an NPP senator), it did little to address the core problem-Edwin lacked strong ties to high-profile elites.
Edwin's weak elite connections did not mean his ex-fighters lacked interest in him. On the contrary, since Edwin was, economically speaking, relatively well-off, many ex-combatants made it a habit to ask him for money, clothes, lodging, and recommendations for jobs in the community. There were, however, limits to the assistance Edwin could provide. It was not uncommon for Edwin to solicit his ex-combatants for help. When asked how many of his ex-fighters he could consider asking for money, Edwin mentioned twelve persons (40 percent). Edwin's willingness to turn to his ex-combatants for assistance blurred the patron-client distinction in "his" network. What's more, Edwin's ex-fighters did not rearrange their lives in anticipation that he would carry them out of destitution; despite his recruitment efforts, only four (13 percent) of Edwin's former fighters joined ALP (see Figure 3). This was, according to one ex-fighter, "because everybody [ex-fighters] had a special interest. Everybody feels the party you can join to get help, you go to that person who can help you." 89 Put differently, Edwin's ex-fighters did not assess that he was the person who could help them.  90 The red nodes represent the ex-commanders and the grey nodes their ex-fighters. The edges represent whom a node would turn to if they needed a small sum of money. The size of the nodes is determined by how many group members said they would turn to that particular person (that is, nodes are scaled by in-degree). Clustering algorithm: Fruchterman Reingold.

Paul: The Limited Returns of Horizontal Relations
Paul's strong relations with elites had an adverse effect on the horizontal relations between his ex-fighters. Many of the latter preferred to spend time with Paul rather than their ex-combatant counterparts. A key reason for this was, according to one ex-fighter, that "Paul is the first person we [ex-combatants] look up to when we have problems … so when he does not have it, then we can go to other friends [ex-combatants] to help, but he is the first choice." 91 Similarly, another ex-fighter declared that "Paul is good to the ex-fighters, which others [ex-fighters] can't do for us. Paul always opens avenues for us to hustle more than others." 92 These types of sentiments are also captured when analyzing the network in terms of "who asks who for money." As is depicted in Figure 4, economic interactions were relatively low in Paul's network (0.125 in density score). On average his ex-fighters were only willing to go to four other network members for money. This does not mean Paul's ex-fighters were opposed to the idea of soliciting assistance from wartime colleagues. On the contrary, several excombatants stressed the important role other ex-fighters had in helping them with money, food, clothes, footwear, lodging, and advice. However, most ex-fighters preferred to turn to a small clique of ex-combatants, or Paul, in times of need.

Edwin: Investing in Horizontal Relations
Edwin's lack of durable elite connections meant his ex-fighters had strong incentives to build relations with each other. When asked why he spent so much time with other ex-fighters, and not only with Edwin, one ex-combatant replied, "It is true that Edwin helps us, but not as compared to these guys [other ex-combatants]." 93 The horizontal economic dimension of the network becomes even more apparent when operationalizing it as "who asks who for money." As can be distilled from Figure 4, a large portion of Edwin's ex-fighters were part of an extensive exchange network where it was natural to ask each other for assistance. This is also reflected in the high density score (0.356) of the network. On average, Edwin's ex-combatants were willing to solicit money from as many as eleven of their counterparts. The importance of this type of assistance was stressed by one exfighter, who declared: "Yes, we help one another with different things. for help. They help with cash." 94 The exchange network was, however, not only centered on money; it also included assistance with job opportunities, food, clothes, footwear, and lodging. According to one ex-fighter, it also encompassed medical support when someone fell ill. On such occasions they would "help one another by assisting with transportations and food to the hospital." 95

Physical Laborers and Gainful Earners
Paul: The Low Appeal of Well-Off Peers Interestingly, there was no greater propensity among physical laborers in Paul's network to solicit assistance from other ex-fighters. When analyzing the network in terms of "who asks who for money," the average out-degree score was four for both physical and nonphysical laborers. What's more, gainful earners were only marginally more likely to be approached for money-while they had an average in-degree score of four, the same figure was three for non-gainful earners. This relative symmetry in horizontal economic interactions is apparent when comparing the node sizes, depicted based on in-and out-degrees, in Figures 5 and 6. The lack of interest in  96 The red nodes represent the ex-commanders and the other nodes their ex-fighters. Gainful earners are colored green. The edges represent whom a node would turn to if they needed a small sum of money. The size of the nodes is determined by how many group members said that they would turn to that person (that is, nodes are scaled by in-degree). Clustering algorithm: Fruchterman Reingold.
soliciting assistance from gainful earners was, however, not due to an absence of resourceful ex-fighters in the network. For instance, Frank 98the person who, besides Paul, most ex-fighters would ask for moneyworked as a plumber and was generally well liked by his peers. According to one ex-combatant, Frank was "always helping his ex-combatant brother. He always helps them when they are in problem." 99 Frank could, however, not compete with Paul as the network's main benefactor. When operationalizing the network as "who asks who for money," Paul had an in-degree score of thirty-two; the equivalent for Frank was thirteen. The reason why most ex-fighters preferred to take their problems to Paul, rather than gainful earners like Frank, was-in the words of one ex-combatant-because "he [Paul] is the first in command to us and also cares and helps us more than anybody else." 100 In the end, the disinterest among Paul's ex-combatants-in particular physical laborers-to reach out to each other in times of need made the network relatively weak.

Edwin: A Network with Multiple Benefactors
Even if horizontal economic interactions were extensive in Edwin's network, they were in no way randomly distributed. On the contrary, it is possible to discern some distinct patterns; while the average out-degree score (for "who asks who for money") for physical laborers was thirteen, it was only eight for nonlaborers (see Figures 5 and 6). As such, it was predominantly the former-many of whom were car loaders working at a local market-who solicited help from ex-military colleagues. The reason why car loaders were so keen to beg others for favors was, according to one exfighter, that "you know nothing much [money] is in the carloading and the job itself is a low-grade hustle." 101 For many of the latter, it was unthinkable to inquire other car loaders for money. This was, in the words of another ex-combatant loader, because "asking your friend on the carloading field ends up in confusion [a conflict], because the man knows that you know his condition and you are not supposed to ask him." 102 Instead, most downtrodden ex-fighters preferred to turn to the network's gainful earners for money. While the average in-degree score (for "who asks who for money") was twelve for the latter, it was as low as eight for non-gainful earners (see Figures 5 and 6). In this context, it is interesting to mention the example of Jacob, who was working as head of security at the previously mentioned market. In fact, many physical laborers made it a habit to call on him when they needed assistance. This was, according to one ex-fighter, because Jacob 103 "is free handed toward us and everybody that he knows. That is why we call him the 'Good Samaritan'" 104 For several physical laborers it was more natural to ask him-and other gainful earners-for help rather than Edwin. This tendency was lucidly captured by one ex-fighter, who professed that "going to Edwin [for money] is not so important … even if I ask him, he will not give something that can sustain me, but those guys [for example, Jacob] will do it." 105 As a consequence, Edwin was not the undisputed patron of the network. Not only was his in-degree score (for "who asks who for money") only eight degrees higher than that of Jacob (29 compared to 21), a number of other gainful earners also received high in-degree scores (for example, 19, 15 and 14). Hence, because of the physical laborers' desire to build denser economic ties with their more well-off peers, Edwin's network was strong.

Comparing the Two Networks
When comparing the ex-military networks of Edwin and Paul, there is a clear variation in strength. While the density of the former was 0.605, it was only 0.229 for the latter. Put differently, the ex-fighters in Edwin's network interacted almost three times as much with each other. What's more, while Paul was the undisputed "head" of his network, a number of Edwin's former subordinates were almost as popular as him. Having established himself as an influential intermediary, Paul was in a position to control interactions between his ex-fighters. This was in sharp contrast to Edwin, whose ex-combatants could easily gain access to other parts of the network by befriending alternative broker figures.
How can we understand these differences? Evidence suggests the variations in network strength were ultimately a function of how well integrated Edwin and Paul were in Liberia's elite patronage networks. Since the end of the war, Paul had succeeded in aligning himself with multiple key leaders. As such, Paul was not only able to distribute a significant amount of patronage; he was also in a position to extend credible promises of future rewards. The situation was markedly different for Edwin. With limited ties to key elites, Edwin struggled to provide economic opportunities to his excombatants. These differences had important repercussions for how the ex-MiLCs, and their ex-fighters, related to each other. First, while it was clear that Paul was the main benefactor in his network, patron-client relations were more blurred in Edwin's. Second, there was a general belief among Paul's ex-fighters that he could-through his elite connections-significantly improve their lives. In fact, as many as 88 percent of Paul's ex-combatants adhered his call to join ANC. There were no similar expectations among Edwin's ex-combatants-only 13 percent followed him to ALP.
Due to Paul and Edwin's varied access to key patrons, their ex-combatants developed diverse strategies to cope with the challenges of making it in the postwar city. Hoping to benefit from Paul's more entrenched position, his ex-fighters put greater emphasis on cultivating relations with their former superior. Meanwhile, for Edwin's ex-combatants it was strategic to address their economic troubles by also investing in horizontal relations. Put differently, it was simply too risky for them to hope that Edwin would one day carry them out of the margins. The density of the networks also reflects these differences when operationalized as "who asks who for money." While Paul's network only had a density of 0.125, Edwin's was as high as 0.356.
The fact that it was Paul, and not Edwin, who had close ties to elites, also had ramifications for how horizontal economic interactions were structured. When comparing physical laborers with nonphysical laborers in Edwin's network, the former were more likely to solicit money from other ex-military colleagues (13 compared to 8 in average out-degrees). Meanwhile, it was predominantly gainful earners who were approached for assistance (average in-degrees were 12 for the latter and 8 for non-gainful earners). Interestingly, no similar variances existed among Paul's ex-fighters. While the average out-degrees for physical and nonphysical laborers were both 4, the in-degrees for gainful and non-gainful earners were 4 and 3, respectively. Why were horizontal economic relations more symmetric in Paul's network? One explanation is Paul's capacity to deliver patronage presumably dampened the need for physical laborers in his network to ask gainful earners for assistance. This was in sharp contrast to the laborers in Edwin's network. Since there was no general expectation that Edwin could take care of them, they were obliged to turn to more well-off ex-military colleagues in times of need.
The observation that ex-combatants-and especially those who are physical laborers-adjust their economic behavior depending on how well integrated their ex-commanders are in elite patronage networks highlights the de facto agency of ex-fighters. Not only do such forms of utility maximization help ex-combatant laborers more efficiently navigate the uncertainties of urban life, it ultimately shapes the strength and shape of ex-military networks. In fact, by giving greater emphasis to building relations with gainful earners, physical laborers in Edwin's network not only made the network stronger but also propelled some gainful earners into becoming key broker figures. Gainful earners never attained similar positions in Paul's network; in the hope of gaining access to elite patronage, physical laborers preferred to attach themselves to their ex-MiLC. These findings feed into a burgeoning literature, which has begun to question the traditionally static view scholars have had of patronage networks. 106 Rather than being locked down in perpetual cycles of political obedience, clients can improve their position by playing various elites off against each other, switching patrons, taking payoffs from multiple leaders, or, as highlighted in this article, giving greater emphasis to horizontal exchange networks. commanders whom in the network they: (a) were friends with, and (b) would lend their cell phone to for one day (the latter being an indicator of trust).
When operationalizing the two networks as "who is friends with whom" and "who trusts whom," some interesting differences emerge. The relative difference in network density between money, friendship, and trust-based ties were much larger in Edwin's network (see Figure 7). While the former was 0.356 in the first network, it was as low as 0.220 and 0.124 in the two latter networks, respectively. Said another way, while Edwin's ex-combatants could consider asking eleven individuals for money, they were only friends with seven and trusted four in the network. This should be compared to Paul's network, where the density scores were 0.125 (money), 0.109 (friendship), and 0.090 (trust). Put differently, Paul's ex-fighters could only contemplate asking four ex-colleagues for money, while they were friends with and trusted three. These findings highlight that as we move from Paul to Edwin's network, it is, above all, economic interactions that increase, and seemingly push ex-combatants toward greater interaction.
How then do economic interactions, friendship, and trust relate to each other? One interpretation of the results is that, in networks headed by excommanders with strong ties to elites, ex-fighters are less obliged to beg mere acquaintances for assistance. For them, it may suffice to turn to their ex-MiLC and a small group of trusted friends. Ex-fighters whose excommanders have weak elite ties do not have the same luxury. Due to the latter's inability to distribute patronage, ex-fighters must solicit help from individuals who are not their friends, and whom they do not trust. Friendship dynamics may also help shed additional light on why horizontal economic relations were more symmetric in Paul's network. Since bonds based on friendship were relatively stronger among Paul's ex-combatants, gainful earners may have been more willing to solicit physical laborers for help. By turning to laborers who were friends for money, they could minimize the risk of being outed in the community as having economic problems. Seeing as gainful earners in Edwin's network were more likely to interact with laborers who were mere acquaintances, it may not have been as appealing for them to ask the latter for money.
Likewise, differences in access to security cannot explain the diverse outcomes. As part of the study, we asked both ex-MiLCs and their ex-fighters if they had recently been subjected to abuse or threats-for example, by the police, armed forces, or community members. Interestingly, while 33 percent in Paul's network stated that they had been mistreated, the corresponding figure for Edwin's was only 19 percent. Hence, according to the insecurity argument, it should have been Paul's, and not Edwin's, network that was stronger.
We also considered several other alternative explanations. First, there were no significant differences in levels of DDR participation between the two networks. In fact, while 97% in Paul's network had registered for DDR and 50% benefited from vocational training/education, the corresponding figures for Edwin's was 100% and 44%, respectively. Second, a key finding within SNA is that ties are generally less dense in large groups. Although Paul's network was indeed slightly larger (33) than Edwin's (31), this small difference in group size does not suffice to explain the large variation in network strength. Third, there is little evidence that the variation at hand can be explained by Paul possessing superior leadership skills. When asked to describe how their ex-MiLCs were as commanders, 88% of Paul's and 90% of Edwin's ex-fighters described them in a positive light. Finally, there is no significant variation between ex-fighters of the two networks when it comes to: ethnic/religious fragmentation, level of education, time spent in the ex-commander's armed unit, military rank, and social relations with non-ex-combatants.

Implications for Peace
Despite the focus peacemakers put on lingering ex-military networks, there is a lack of studies explaining why some are stronger than others. This article begins to fill this lacuna. We argue that ex-military networks are stronger when ex-commanders have weak links to elite patronage systems. Under such conditions, ex-combatants are unable to rely on their former superiors for economic assistance and must instead build denser ties to wartime peers to gain access to a social safety net. This is particularly true for physical laborers; because their bodies are their only asset, they are more exposed to the pressures of the postwar city. As such, the desire of physical laborers to address their socioeconomic vulnerability is central to the causal logic of the argument. A structured focused comparison of two ex-military networks in urban Liberia, based on SNA and in-depth interviews, lends support to this argument.
What implications do these findings have for our understanding of and work with ex-military networks? Irrespective of whether we are talking about weak or strong networks, the resilience of such structures is not necessarily a function of malign intentions by ex-military actors. It is, rather, a local response to the failure of top-down peacebuilding interventions in general, and DDR programs in particular, to address the economic plight of ex-fighters. Efforts to dismantle or reconfigure ex-combatant networks may, therefore, not only deprive ex-fighters of a much-needed social safety net but also generate new violence. Studies have shown that it is when informal economic structures, such as patronage networks, begin to crumble that individuals-such as, ex-fighters, former warlords, and politicians-are most likely to engage in violence. Due to the dire consequences of losing access to such structures, individuals and groups fight to get "reconnected." 109 Hence, in many war-ridden societies, it may be wiser to leave ex-military structures intact rather than attempt to uproot them. It is thus imperative for peacemakers to cease overtly "securitizing" these networks, and instead start acknowledging the productive role that such structures can play in postwar societies.
However, in some instances, it may be necessary for peacemakers to actively engage ex-military structures. This concerns situations when political tensions escalate between elites (for instance, during highly disputed elections, when peace processes come to a halt or civil wars rage in neighboring countries). There is otherwise a risk that elites attempt to mobilize ex-military networks for violent purposes. 110 In such instances, it is vital for peacemakers to know what kind of networks they face. Since ex-MiLCs tend to dominate weak networks, it will generally be easier for international actors and governing elites to engage and steer such structures in a peaceful direction. In fact, it may be possible to prevent ex-combatant violence by only negotiating with and offering private goods to the ex-commanders. By providing the latter with attractive employment opportunities (in the public or private sector) or utilizing their broker status to implement local development projects, ex-commanders can be convinced to mobilize their ex-combatants for peaceful purposes. This is particularly true if similar initiatives allow ex-commanders to distribute economic benefits or employment to their ex-fighters. Alternatively, peacemakers can seek to remove ex-MiLCs from weak networks by, for instance, providing education or business opportunities abroad. Under such circumstances, networks are likely to disintegrate into smaller cliques. Having said this, it is crucial to keep in mind that if peacemakers fail to "co-opt" ex-commanders in weak networks, remobilization is likely to go quicker than in strong networks. Since ex-fighters tend to be more dependent on their former superiors in such ex-military structures, ex-combatants are more likely to adhere calls to take to arms.
In stronger networks, it is unlikely to be enough to simply target excommanders. This is because well-connected ex-fighters-in particular, gainful earners with multiple horizontal ties to physical laborers-often challenge ex-MiLCs' control over the network. With multiple network brokers, it is difficult for peacemakers to know with which actor to negotiate. Such forms of structural ambiguity increase the risk that entrepreneurs of violence can convince some influential ex-fighters to co-opt the network for violent purposes. When facing strong networks, it is therefore imperative to provide more public goods of a broader scope, which target the wider network, or at least its key individuals.
It is vital to note that these forms of economic "co-option" strategies presumably work best in contexts where socioeconomic marginalization constituted a central reason for joining armed groups in the first place. In other networks consisting of ideologically committed ex-fighters, such as extreme nationalists, Islamists, or Maoists, it may be necessary for 110 Christensen and Utas, "Mercenaries of Democracy." peacemakers to put greater emphasis on other peacebuilding tools, such as engaging in political reintegration, policing, or breaking up ex-military networks. 111 The findings presented here are based on a comparison of two ex-military networks in Liberia. To assess and refine the theoretical framework, it is therefore vital that future research look beyond Liberia and study exmilitary networks-using SNA-in other post-civil war contexts. A particularly fruitful avenue would be to trace how ex-combatant structures change over time. We suspect that both the strength and composition of ex-military networks may shift over time. Network reconfigurations are probably associated with critical junctions in war-to-peace transitions, when national elites may have an interest in using the mobilization potential of ex-military networks. In fact, by employing ex-commanders-or well-entrenched ex-fighters-elites can quickly gather large numbers of ex-combatant demonstrators, voters, or fighters to defend their interests. Similar forms of elite interventions often entail an infusion of patronage into ex-military networks. According to the theoretical argument presented in this article, this should allow ex-commanders to consolidate control over their ex-fighters, and this would subsequently result in a weakening of the network. For this reason, future projects would do well to compare the strength of networks before and after events such as: national elections, the arrest or sentencing of ex-military leaders accused of war crimes, or the departure of peacekeepers. However, since it may be difficult to predict when major political events (besides national elections) occur, an alternative approach would be to measure network strength at predetermined time intervals. Such a research design could, for instance, compare the resilience and composition of ex-military structures right after the completion of a DDR process and a few years down the line.