Common enemies? Coups, insurgent strength and intra-elite competition: evidence from Latin America

ABSTRACT What is the relation between coups d’etat and civil wars? While a wide set of studies have traced the determinants of internal armed conflicts and coup attempts, the interplay between these contentious processes remains unexplored. Building on different strands of research, this article seeks to explain why, and under what conditions, some regimes experience coup attempts in the midst of civil wars while others do not. Concretely, I posit that coup attempts during internal armed conflicts are more likely to occur when two conditions converge: when insurgents reach a medium-level of strength in situations of intra-elite competition. Key military forces, elite outsiders and coalition insiders interpret this situation as a unique opportunity for changes in the distribution of power and potentially coalesce through the formation of alternative regime coalitions. This argument is tested with a novel dataset on 90 Latin American revolutionary socialist insurgencies active since 1950 and a qualitative case study of the dynamics leading to the 1976 coup d’etat in Argentina, with results supporting the theoretical expectations. These findings contribute to a more detailed understanding of the relation between coups and civil wars, opening the way for further studies on this burgeoning area of research.


Introduction
Why do some regimes experience coups in the midst of civil wars while others not?Research on the linkages between coups d'etat and internal armed conflicts sustains that the level of strength of insurgent groups is a core factor shaping coup attempts: the higher the level of insurgent power, the higher the chances of wartime coups. 1 Challenging this perception, this article argues that coups will be more likely to occur at mid-levels of rebel capacity, and only in scenarios of elite disunity.Under conditions of intra-elite competition, the presence of an insurgent organization with mid-level of strength creates the necessary incentives and ability for elites to coalesce with state military forces for the formation of alternative regime coalitions.If successful, this will materialize in coups d'etat against existent political structures.
I explore this argument in the context of the Latin American revolutionary insurgencies.For this, I rely on a novel database, the Latin American Guerrillas Dataset (LAGD), which captures the core features, dynamics and evolution of 90 revolutionary socialist armed groups operating in the region since 1950.Most importantly, this database disaggregates at the organization-year level the strength of insurgent groups in three ideal categories, allowing the comparison across cases with low, medium and high level of strength.I also rely on a qualitative case study of the dynamics between insurgent actions and elite choices leading to the 1976 coup d'etat in Argentina.The analysis supports the theoretical argument: under conditions of elite competition, coup attempts become more likely when insurgents reach a medium level of strength.
This article contributes to several literatures.First, it directly speaks to the scarce set of studies that explore the relationship between coups and internal armed conflicts.Roessler 2 provides the most compelling argument on this issue, showing how African leaders face a basic dilemma: if they aim to reduce the likelihood of civil war, they need to give access to state structures to politically excluded ethnic groups.However, power sharing increases the risk of coups.While the argument fits perfectly well to ethnically heterogeneous societies, it is necessary to develop novel theories for societies in which ethnicity is not a cross-cutting cleavage.Moreover, I also challenge one of the core findings of this area of research, that portrays the linearity of the relation between levels of rebel strength and coup attempts.
A second literature to which this study contributes relates to the vibrant research programme on the study of civil war.Despite the pivotal role of intra-elite dynamics in processes of democratization, state building or revolutions, the interactive relation between insurgents, incumbents and other elite forces during civil wars has been understudied. 3The reliance on an elitist-approach challenges existent perceptions about the ontological nature of the state in the civil war literature, which conceptualize the state as a cohesive and unitary institution.
Finally, this article is in direct connection with the wide set of studies tracing the determinants of coup processes.Several studies show that one of the factors that consistently shapes the prospects for coup occurrence is political instability. 4Its effects, however, are unclear and two different strands of research provide alternative implications.On the one hand, most cross-national studies argue that political instability facilitates coup attempts. 5But on the other hand, a different strand of research argues that it prompts a "rally around the flag" effect, leading elites to forge agreements that bind them together against the risk of revolution. 6While this research has provided fundamental advancements, certain gaps remain to be filled.First, few studies have delved into the heterogeneous effects of distinct contentious processes.Second, most of this literature neglects the fundamental role that civilian elites outside institutional structures play in the development of alternative regime coalitions.And third, overly deterministic arguments disregard the complex nature of elite strategic interactions during conflict.In direct contrast to the expectations of prior research, I side with recent works that sustain that the effects of violent processes will be contingent on the level of elite unity. 7hrough the combination of insights from a variety research programmes, this article aims to move the literature forward deepening into the intrinsic dynamics explaining the relation between coups d'etat and civil wars, through a particular focus on the understanding of elite choices during these contentious processes.

Theory
In the development of a theory of elite choices during civil wars, I rely on the key building blocks of selectorate theory 8 and social conflict approaches. 9From social conflict approaches, I use a conception of society formed by a variety of social actors with divergent preferences over institutions and policies.Also, I rely on an understanding of politics as a zero-sum game, in which the set of policies that benefit one group will necessarily be contrary to the interests of other social force. 10From selectorate theory, it is expected that each of these actors has an inherent preference for the maximization of their power within institutional structures.Also, it is conceived that it is not possible for any political leader to rule alone, they need to form a winning coalition, as "every leader answers to some group that retains her in power." 11These winning coalitions are conceived as the "subset of the selectorate of sufficient size such that the subset´s support endows the leadership with political power." 12 These coalitions are defined by the compromise and mutual benefits between the leaders and the set of actors that support them and guarantee their survival, 13 and are characterized by their fluid and elusive nature. 14
The incumbent is the leader of the political regime.It oversees the direction of the executive power, either in democratic or authoritarian settings.The main goal of incumbents is "to remain in office or to obtain greater power and autonomy for the institutions [they] represent." 16Coalition insiders are the set of social forces that are part of the winning coalition.Leaders rule in cooperation with other actors, as "governing coalitions […] form around the leader in a mutually beneficial relationship." 17hey retain important positions within governmental structures, have sufficient capacity to influence policy outcomes and their main goal is to maximize the prospects of being part of the ruling coalition.Key military players, mainly represented by top military officers, are driven by the fulfilment of their corporate interests, seeking to maximize their political power within state institutions through the holding of relevant office positions.Elite outsiders are the set of elite forces without access to governmental decisions and political power.Even though they could be part of the state apparatus, they are "excluded from key positions with power to influence policy and enjoy political privileges." 18 These actors' main aim is to enter state institutions with the goal of being part of a ruling coalition that represents their political, social and economic interests.Finally, insurgent organizations, conceived as a set of organized individuals that use violence for the seizure of political power. 19e argument How do civil war dynamics, and more concretely, insurgent strength, affect coup attempts in the midst of internal armed conflict?A comprehensive answer to this question requires a thorough analysis of the endogenous effects of rebel actions on elite preferences through a focus on the dynamic and interactive process of elite strategic choices during civil war.This approach shows how, under conditions of intra-elite strife, key military players, elite outsiders and a subset of coalition insiders, have the disposition and the ability to coordinate for the formation of an alternative regime coalition, due to the opportunities created by the presence of a violent actor with mid-levels of strength. 20The process of elite coordination, if effective, could eventually crystallize in a coup d'etat (Figure 1).
Historical evidence shows that the type and scope of agreements between elite groups vary across countries and evolve over time.Concretely, Higley and Gunther 21 distinguish three different types of national elites according to their level of unity: (1) ideologically unified elites, defined by low levels of disagreement on ideological and policy issues; (2) consensually unified elites, in which there is an underlying accord on the respect of existent institutions, despite disagreements on concrete policies; and (3) disunified elites, with high levels of distrust across social forces and lack of institutional and policy cooperation.
In situations of crisis, such as contexts of armed violence, elite collective action becomes activated.Indeed, "contentious politics […] powerfully and persistently shapes how political alliances and rivalries will be defined." 22Distinct social forces are confronted with sudden modifications in the political opportunity structure prompted by the presence of insurgent groups.The overt scenario of civil war facilitates the prospects of potential alignments across elite forces, as the threat on the monopoly of violence posed by guerrillas acts as a shock that opens a window of opportunity for changes in the distribution of political power. 23owever, the impact of contentious processes, and more particularly of civil war, on elite divisions will be contingent on the existent structure of elite alignments, whether this is one of elite unity (ideologically and consensually unified elites) or elite competition (disunified elites), as transitioning across each of these categories is of utterly difficulty.The stickiness of elite agreements is so entrenched in the nature of the regime and the distribution of political power existent in the country that it is rare, but not impossible, to see how countries transition across these categories.As argued by Higley and Gunther: "Modern history […] records relatively few elite transformations from one basic type to another.Such elite transformations appear to occur only in rare circumstances." 24Despite several civil war dynamics could affect elite preferences and strategic choices, such as the intensity of civil wars, counter-insurgency actions, the duration of the conflict or the level of insurgent strength, none of these dynamics is expected to make countries transition from a situation of elite unity to one of elite competition and viceversa, though certain elite groups may alter their preferences or these preferences could become highly polarized.How do elites respond to these contentious processes considering the type of elite structure existent in a country? 25

Elite unity
When elite cohesion is characterized by being either ideologically or consensually unified, I expect that, regardless of the level of insurgent threat, elites remain cohesive.The pact forged across elites reduces not only the incentives, but also the ability to coordinate elite preferences against the prevalent winning coalition.On the one hand, the "underlying consensus about rules of the game and the worth of existing political institutions" 26 reduces the disposition to intervene, and specially, of becoming the first mover in the coup process.The small odds of succeeding and the access to institutional decisions, together with the harsh consequences linked to failed coup attempts, should hamper the disposition of aggrieved social forces to promote profound regime changes through coups d'etat.
In addition to this, the lack of extant intra-elite competition exacerbates coordination problems, reducing the ability to intervene of coup instigators: potential first movers lack sufficient information about the degrees of loyalty of distinct elite groups towards the regime leadership. 27Under scenarios of elite agreement, it is difficult to grasp the actual preferences of these groups, given that interactions would be characterized by preference falsification, as no actor has incentives to deviate from the existent status quo.The combination of these characteristics hampers the development of strategic alignments across distinct elite groups.The immediate observable implication is that, under conditions of elite unity, the presence of insurgent groups produces a "rally around the flag" effect that strengthens the cohesiveness across elite forces.

Elite competition
However, under conditions of elite disunity, the situation of rampant insecurity, political uncertainty and weakened governmental legitimacy, opens the way for the contingent coordination across social actors for the development of an alternative regime coalition.
Scenarios of intra-elite competition facilitate the prospects for the disposition to intervene through coups d'etat.From a rationalist perspective, three distinct actors have incentives to coordinate to maximize the expected payoffs from a potentially successful coup attempt.First, military forces, for which the opening of the coup structure maximizes their capacity to promote their own corporate interests.This is mainly related to their potential advancement in positions of power and by the increased grievances of security forces produced by the effects of counterinsurgent actions against the guerrilla threat. 28Moreover, the incapacity of the government to quell the insurgents deepens their demands of autonomy from the executive power in order to increase the effectiveness of their actions.The combination of all these factors rises their disposition to intervene through coups d'etat.
The second set of actors, elite outsiders, face the arrival of a prominent insurgent threat as a real opportunity to maximize the promotion of their goals.In a scenario in which intra-elite competition promotes the development of policies and institutional arrangements that favour the interests of the existent winning coalition, their lack of political power lead them to conclude that the only way of promoting their preferences is through their access into state institutions.The presence of armed organizations destabilizing the political regime and weakening its legitimacy presents as an ideal opportunity.
Lastly, a subset of coalition insiders could potentially have incentives to coordinate with military forces and elite outsiders to gain prominence within the winning coalition.Under the elusive and fluid nature of coalition arrangements, coalition insiders aggrieved by their scarce voice in policy and institutional decisions, could interpret the existence of an insurgent threat also as an opportunity to expand their power through the forging of alliances with the potentially winning side in the competition over the control of state institutions.As argued in the selectorate theory developed by Bueno de Mesquita et al. "the incentive to defect from the incumbent to a challenger depends on the prospects of being included in the challenger's winning coalition if he should replace the incumbent." 29

Insurgent strength
While the presence of insurgent organizations shapes the incentives of elite groups in scenarios of overt intra-elite competition, I posit that coup attempts should only be expected in contexts of mid-levels of guerrilla strength.Four factors play a role here: the threat of revolution; the "guardianship dilemma"; the level of state legitimacy; and the level of insurgent political capacity.In the following subsection, I distinguish the potential impact of three distinct ideal types of insurgent levels of strength. 30w insurgent strength Insurgent organizations with low levels of strength are characterized by their lack of military and political capacity.The armed group is active but engages in scarce and non-influential attacks.It is formed by few members, it is highly sensitive to counter-insurgent operations and lacks any type of territorial control.At this stage, the group is particularly focused on the reinforcement of its operational capacity and survival becomes its most relevant task, 31 remaining as a marginal phenomenon.
At low levels of insurgent capacity, the threat of revolution is not credible enough, making increasingly difficult the coordination of elite interests.Moreover, weak insurgencies are most likely targeted by police forces and other state security agencies, not by the military, 32 toughening the penetration of military forces into regime structures.As well, levels of state legitimacy are unlikely to suffer a high level of decrease, given the scarce number and insignificant operations engaged by weak rebel organizations.Based on this, I do not expect coups to occur when incumbents face weak insurgent groups.

Medium insurgent strength
In scenarios of insurgencies with medium levels of strength, armed groups begin to be considered as increasing forces characterized by the expansion of their military capacity and by their level of political mobilization.There is an important shift from clandestine operations by underground groups under strict security measures, to groups operating in wider scale in public contexts, in some cases capable of holding considerable levels of territorial control.Despite their ability for expansion, groups do not pose a substantial threat and there are considerable levels of asymmetry between incumbents and insurgents.
I posit that, in situations of elite disunity, at mid-levels of guerrilla strength, the adequate contextual conditions for coup attempts are present: existence of a sufficient insurgent strength as to justify the need of profound institutional changes; the military forces play a pivotal role within regime structures; there are decreasing levels of government legitimacy; and insurgent groups have not reached high levels of political capacity.But how do mid-levels of insurgent strength facilitate the ability across elites to coordinate and organize a coup attempt?I argue that this process unfolds across three different layers.First, through the penetration of state military forces into political structures, leading to the so-called guardianship dilemma.The rising force of insurgent violence puts the state army as a pivotal political actor with strengthened power within decision structures, as it becomes the only agent in charge of the repression of guerrilla activities.However, the increasing power of military forces does not necessarily entail higher levels of political stability, as this heightens the risks of coups: "regimes that frequently face mass, organized, and violent opposition must integrate their militaries within their repressive apparatus […] Once soldiers attain a politically privileged position, they naturally attempt to preserve it." 33Already settled and with access to regime institutions, the military is in an advantageous position for the development of an alternative regime coalition.
Second, through a gradual process of social polarization.The heightened power of the military is accompanied by rising levels of conflict escalation: the new role of the army as the main repressive force leads to insurgent militarization, prompting moves towards clandestinity and shifts in armed strategies. 34The spiral of violence between guerrillas and state armed forces lead to a polarization of the political sphere, partitioning the political scenario between the "subversive" guerrillas and the counter-insurgent front led by the army. 35This situation sharply reduces the level of state legitimacy, specially prompted by a feeling of social chaos in which the military forces emerge as the only actor capable of restoring institutional authority, facilitating the coalescence of divergent elite groups' preferences.However, the level of effectiveness of the actions against armed groups will also be linked with the potential set of coup-proofing strategies implemented by the incumbent.Extensive quantitative and qualitative evidence shows how coup-proofed regimes are more prone to the onset and expansion of insurgencies: aiming to reduce the threat of insider threats, they become exposed to rebellions coming from actors outside the regime coalition. 36oup plotters aiming to form an alternative regime coalition are hampered by three informational asymmetries: uncertainty over regime capabilities; no transparency on the level of insurgent strength; and uncertainty about the strategic choices and preferences of other social forces.The actions engaged by insurgents in the context of civil war reduce each of these asymmetries. 37First, the ineffectiveness of the state in the deterrence of contentious processes serve as a potential signal of the fragile capacity of the regime coalition to resist a coup.If in a context of civil war, the regime is not capable of adequately ensuring the monopoly of violence in its national territory, coup plotters could interpret this as a signal of the potential success of a future coup attempt.Another path of action for coup plotters to reduce its lack of information on state capabilities is to target potentially aggrieved coalition insiders, as they could provide fundamental information on state strength and elite preferences.Moreover, this type of actors will be likely targeted by the alternative regime coalition in formation not only for its informational resources, but also because the defection of pivotal actors shapes the prospects of success of coup instigators.
The second informational issue relates to the lack of information on the level of insurgent strength.Given the asymmetry of power existent between government and insurgent forces, one of the core goals of the rebels is to develop an adequate level of secrecy to ensure its survival. 38However, the necessary publicity of insurgent actions will serve as the main signal of the approximate level of rebel strength.If rebels want to advance in their goals, they will need to expand their territorial control, mobilize civilians and get access to a wide set of human and financial resources that make their political project viable. 39Each of these actions will give an inexact, but sufficiently relevant, approximation to coup plotters on how strong are the rebels.Plotters can make use of the perceived level of insurgent support shown in anti-government demonstrations 40 for an approximation of rebels´political capacity; focus on the effectiveness, quality and quantity of rebels´armed actions and combatants, pointing their potential threat as a military force; or rely on the expansion of territorial control as a good approximation of the overall level of insurgent power.Based on these elements, and despite the clandestinity of a variety of rebel actions, plotters will be able to assess to a high degree whether insurgents are an irrelevant force, whether they have become a highly developed political and military organization, or if there the threat of revolution is imminent.
And third, there are informational asymmetries across the preferences of other elite groups.However, contentious processes, and more concretely, insurgent actions in the midst of civil war, lead to a reduction in the strategic uncertainty across distinct social forces, as elites update their believes about the potential preferences and expected paths of action of other social groups.Faced with a threatful insurgent organization, and mediated by the publicity of rebel actions, coup plotters expect that the weakening of the existent leadership opens the way for changes at the regime level, facilitating the prospects for elite coordination across groups with distinct preferences but with a common interest of ousting the current regime coalition. 41nce elites have reached a coordination point for the implementation of a new political structure, finding the adequate timing and contextual conditions is of paramount importance in the coup process.Given that coup engagement is a risky option, and the fate of coup plotters is directly linked to death, exile or imprisonment "coup conspirators will […] only attempt a coup when […] its probability of victory are high enough to offset the dire consequences of a failed putsch." 42This is expected to happen when a sufficiently cohesive alternative regime coalition is formed and when the legitimacy and strength of the current coalition is sufficiently low as to justify the coup towards internal and international audiences.

High insurgent strength
On the other extreme, insurgencies with high levels of strength have already become an established alternative to state power: the group has important levels of territorial control, it engages in highly complex military operations, and has a large number of armed combatants.The rebel organization is therefore conceived as an existential threat in a political and social landscape characterized by an implicit situation of dual sovereignty.
It could be argued that coups will be more likely to occur under conditions of high levels of insurgent strength: the threat of revolution is the highest; the military are embedded into state structures; and the legitimacy of the state has been severely weakened.However, I sustain that potential coup plotters will attempt a coup before rebels reach a high level of military and political strength.As argued by Bell and Sudduth 43 : If there is any possibility that rebels could emerge victorious […] then potential coup plotters must anticipate their welfare should the rebels win.Any possibility of rebel victory must decrease satisfaction with the status quo for anyone who would not expect to benefit from rebel victory.
But it is not only the risk of insurgent victory what drives the actions of coup plotters, but also the potential concessions to the rebels.According to the bargaining theory of war, the more powerful the threat against incumbents, the lower the barriers to grant concessions to insurgent groups, such as power sharing agreements on territorial, political or military issues. 44lso, at high levels of insurgent power, rebel groups engage in tactic diversification.Not only they attempt to achieve military victories in the battlefield, but they engage in widespread political actions in order to achieve concessions from the government.The capacity of insurgent groups to mobilize a considerable proportion of social forces, including civilians, grassroots local movements, religious organizations, political parties or international actors, convert rebel groups in a pivotal actor in the wartime political scenario. 45Beyond their military strength, strong insurgencies directly influence broader constituencies, which through their non-violent actions could expose the weakness of incumbents and may pose a risk to the stability of the current government. 46his high level of military and political power increases the bargaining power of insurgents, leading to negotiated settlements between them and the government.These negotiated settlements frequently include provisions that could pose an important risk to the position of power of a subset of coup plotters, and more particularly, of the military, which acquired important privileges within the regime during wartime.Agreements between insurgents and incumbents usually include provisions that engage in important reforms of the security sector, leading to reductions in the size or resources received by the military, or even lead to the integration of rebel forces into military structures.These heighten the tension between the government and the military, which could lose its privileged position.This set of settlements could also cover provisions that directly affect the interests of coalition insiders and coalition outsiders, including political, economic, territorial or power-sharing provisions, such as land reform, enforced political representation, access to economic resources or territorial autonomy, among others.
Given the high uncertainty on how the interests of coup plotters could be represented in a scenario of insurgent victory or in which negotiated settlement is implemented, I sustain that coups will be more likely to occur in situations in which insurgents have mid-levels of strength.Though the risk of revolution, the penetration of military forces into regime structures, the level of state legitimacy and the level of insurgent political capacity is lower in situations of medium level of strength than against strong insurgents, coup plotters are more willing to deal with the uncertainty of a coup that depend on their strategic ability than with the indeterminacy of a conflict whose outcome might not directly depend on their actions. 47herefore, the argument developed derives the following hypothesis: H1: Mid-levels of guerrilla strength, under scenarios of elite disunity, increase the likelihood of coup attempts in the midst of civil war.

Data and empirical strategy
In order to test this hypothesis, I rely on a cross-national analysis of the determinants of coup attempts in the midst of civil wars with a regional focus on Latin America.Concretely, I make use of time-series cross-sectional data using country-year as the main unit of analysis.The sample covers 19 countries of the Latin American region for the period 1950-2018.Given that this analysis traces the determinants of coups during civil wars, I rely on a sample of observations that includes only the countryyears in which any insurgent group was active.With the goal of testing the impact of rebel groups' actions in Latin America, I make use of a novel dataset on Latin American insurgencies, the Latin American Guerrillas Dataset, which captures the dynamics and features of 90 armed groups created in the region since 1950.Concretely, the scope of the sample only covers revolutionary socialist insurgencies, as they are the most common type of insurgency operating in the region during this period.They "constitute a particular type of the much broader category of revolutionary rebels, for whom the objective is not to just win power or just to create a new state, but to overthrow the existing social order and replace it with a new one." 48Other type of groups, such as paramilitary organizations, purely ethnic/religious insurgencies and organized crime groups are out of the sample.The dependent variable, Coup Attempt, relies on a novel collection of evidence of coups in the Latin American region. 49Given the considerable divergence between episodes of coup attempts collected in regional and global datasets, I bring together information from several databases in order to provide an accurate coverage of the set of coup attempts that occurred in the region.The outcome variable is a dichotomous variable that captures any coup attempt, defined as "illegal and overt attempts by the military or other elites within the state apparatus to unseat the sitting executive." 50ut of the 147 coups occurring in Latin America between 1950 and 2018, 24 took place in periods of active operation of insurgent groups, a 16.32% of the total sample. 51egarding the main independent variables, I rely on an interaction variable between the degree of elite unity and the level of insurgent strength.For this, I have relied on two different data sources.On the former, I make use of one of the most systematic and comprehensive measures of elite competition available, the one created by Albertus. 52n order to capture the impact of coalitional splits on levels of land redistribution in Latin America, and relying on rich historical evidence, he creates a dummy variable of intra elite competition between ruling political elites and landed elites.Concretely, he codes the existence of an elite split when none of these three situations are present: (1) when members of the cabinet are not members of landed elites; (2) when cabinet members are not directly appointed by landed elites; and (3) when landed elites do not provide with financial, repressive, logistical or any other type of material support to ruling political elites. 53Otherwise, if any of these situations is present, no split will occur between incumbents and landed elites. 54Despite this variable cannot capture the complexity on the degrees of elite unity, as it implies a partial coverage of the wide range of intra-elite strives, and that the other core variable insurgent strengthhas a territorial component that might be directly related to the interests of landed elites, it still captures important features of elite alliances and splits, especially in Latin America, where landed elites have played a pivotal role in the political dynamics of the region. 55Given the limitations of this index, I make use of two other measures of intra elite competition as robustness (see Appendix D2).
On levels of insurgent strength, the development of a quantitative measure presents several difficulties as well.The main problem is the lack of clarity on what insurgent strength is and what are the different factors that should come in to facilitate its measurement.Prior studies have made use of measures of territorial control; centralized commands; mobilization, arms and fighting capacity 56 ; number of absolute combatants and ratio of combatants vis-à-vis the state 57 ; number of fighting events, geographic areas of operation and average distance to capital 58 ; or government force fatalities. 59s a solution to this, I rely on the Latin American Guerrillas Dataset (LAGD), 60 which develops a novel analytical framework that conceptualizes civil wars as dynamic, recursive and non-linear processes that unfold across different stages, relying on the armed group-year as its main unit of analysis.Based on this framework, LAGD includes a variable on the levels of success achieved by guerrilla organizations, distinguishing three ideal categories: Minor Force (low strength), Break Out (medium strength) and Dual Power (high strength).This measure, which is based on an inductive approach of fined-grained evidence of insurgency trajectories in the region, captures the absolute level of strength of insurgents since their inception until their demise, relying on the use of a variety of relevant measures.
The use of a measure on absolute strength is adequate particularly from an empirical perspective.Though theoretically it is true that "the absolute strength of a belligerent is less important than its strength in relation to its opponent," 61 the use of measures of absolute insurgent strength facilitates not only the comparability across rebel groups beyond their geographic context, but also avoids the establishment of unclear cutoffs on the relation between insurgent and government features (e.g.ratio of government vs. insurgent troops).As a solution to the issue of capturing the relative level of insurgent power, I include a variety of control variables on government strength in the main models, as explained below.
The three stages of strength are as follows.Stage 1 (Minor Force) is the lowest level of strength.Usually held by groups at the onset of armed actions, it represents a phase of the insurgent life cycle in which the group lacks the capacity for engaging in consistent political mobilization and the military capacity to perform complex operations.During this stage, most of the actions are targeted to increasing the operational capacity of the group.Insurgent groups are included in this category when they begin their operations, when they have less than 1000 combatants and/or when they lack the capacity to engage in complex military operations.Stage 2 (Break Out) is defined by the ability of the armed group to successfully launch violent actions against harder targets, such as military barracks, or by its capability to mobilize a considerable number of armed combatants (≥ 1000).This type of measurement captures the increasing strength and capacity of the group, as the engagement in this type of actions require a solid organization both from a military and political perspective.Finally, Stage 3 (Dual Power) is defined by the capacity of the group to establish territorial control of a city of over 100,000 inhabitants, of at least 1/4 of the national territory, by the engagement in at least 500 armed attacks and 1000 battle-deaths per year, or when the structure of combatants is formed by over 5000 individuals.Following prior research, 62 I make use of the level of strength of the strongest insurgent group operating in every country-year.Moreover, to ensure the robustness of the results, I run a variety of additional tests on this variable, including an intercoder reliability test, the use of a variable on the number of active groups, the use of alternative measures of rebel strength or the inclusion of a composite measure on absolute insurgent strength, among others (see Appendices C and D).
Regarding control variables, I include a variety of factors that help isolate the potential impact of intra-elite competition and rebel strength on coup attempts within the context of civil war.As way to capture the level of strength of the state, and therefore facilitating the understanding of relative rebel strength, I include two variables.First, the natural logarithm of GDP per capita coming from Maddison, 63 which has been conceived as a good proxy of state strength. 64And second, a measure on the amount of Military Personnel, key for counterinsurgency actions, with data coming from the Correlates of War project.I also include two variables that could have an impact on coup processes and intra-elite dynamics.Concretely, I include a variable on Regime Type, using a continuous variable coming from Polity V, as well as a variable capturing the level of Ethnic Fractionalization from Fearon and Laitin, 65 as both have been found to be fundamental to understand elite structures and coup attempts 66 .Moreover, I make use of data coming from the Banks CNTS Data Archive capturing the number of Riots and Domestic Protests to unpack the potential heterogeneous effects of distinct contentious processes on coup attempts, as they also have been found to influence elite and civil war dynamics. 67Finally, I include a standard control on Population coming from Maddison.All these control variables are lagged by one year.
Considering the binary nature of the main dependent variable, I rely on the use of probit estimators.Robust standard errors clustered by country are used to deal with the potential heteroskedasticity of observations in a given country.

Empirical analysis
Before turning into the main model specifications, Table 1 shows the distributions of coup attempts according to the distinct levels of insurgent strength and intra-elite competition.Despite coups seem to be present at all levels of strength, they occur more prominently in two situations: (1) in conditions of elite unity and when insurgents are weak; and (2) when regimes face rebel organizations with mid-levels of strength under conditions of intra-elite competition.Table 2 and Figure 2 show the results of the core analysis.Stage 1 ("Minor Force") and No intra-elite competition serve as a reference categories for Models (1) and ( 2), while the interaction between No intra-elite competition and Stage 1 ("Minor Force") serves as a reference category from Models (3) to (6).While Models (1) and (2) show the lack or small statistical significance of either the variables on insurgent strength (p < 0.1) or intra-elite competition, the interaction variable in Model (3) follows the theoretical predictions: the combination of mid-levels of insurgent strength in the presence of intra-elite competition increases the likelihood of coup attempts when compared to situations of no intra-elite strife under the actioning of weak insurgent groups.Concretely, the variable is positive and highly significant (p < 0.01), even after the inclusion of all control variables.
These results stay in line with the argument that, if levels of elite disunity are high, certain social forces could take advantage of the presence of insurgent organizations in order to justify the need of bringing profound political changes.A high level of insurgent strength is positively related to the engagement in coup attempts, as found by Bell and Sudduth. 68However, the variable lacks statistical significance in all the models.These results point as well to the necessity of understanding the divergent determinants of coups under distinct contextual conditions, even within the analytical category of coups during civil wars.One result deserves further explanation.As shown in the descriptive table above (Table 1), more than one-third (37.5%) of the coups that took place in Latin America in periods of active insurgencies occurred when insurgents had low levels of strength.However, qualitative evidence casts doubt on the centrality of insurgent actions, as other political dynamics seem to have triggered these coups, including electoral fraud -El Salvador 1972 69 -, military factionalism -Argentina 1971 70 or democratic demands -Venezuela 1962. 71s a further test, I unpack the concept of "political instability" and trace the potential distinct impact of other contentious processes (Table 3). 72The interaction variable on mid-levels of insurgent strength and intra-elite competition remains positive and highly significant (p < 0.01).In line with prior research, 73 the only variable that remains significant across all the models is the one on number of strikes (p < 0.01p < 0.05), particularly under conditions of elite cohesion.Contrary to other studies, domestic protests do not seem to have played an important role. 74No impact is found either for political assassinations or riots.In sum, these results portray the distinct effect of contentious events in the determination of coups d'etat, reflecting the necessity of unpacking the heterogeneous impact of divergent contentious processes. 75

Robustness checks
These results hold to a wide set of robustness checks, all of them covered in Section D of the Appendix.These include the use of distinct measures of the dependent and independent variables (Subsections D1 to D4); the use of different empirical models (Subsection D5); testing for the potential presence of omitted variable bias (Subsection D6); controlling for the potential impact of US policies in Latin America (Subsection D7); and to the running of a battery of further tests (Subsection D8).

Alternative explanations
Other explanations could account for the observed results.First, a key driving force facilitating cooperation across elite forces could be ideology.In the highly polarized global context of the Cold War, alignments across social forces could be based on shared ideological preferences, rather than on a purely instrumentalist perspective of siding with the potential winning coalition.While further research should test the role of ideology with the development of fine-grained data on elites´ideological preferences, existent evidence casts doubt on the centrality of ideology as the core driving factor.A key example of this is the 1979 coup d'etat in El Salvador, where military forces ousted President Romero with the promise of conceding amnesty for political prisoners and the expansion of political rights.The new coalition was formed by a variety of actors with divergent political preferences, including leftist forces, purely technocrats and conservative military officers. 76 second alternative explanation relates to the level of cohesion of the ruling coalition, endogenously affected by levels of insurgent strength.While this might be a valid argument, I sustain that levels of elite competition will determine the way by which aggrieved social forces will seek its path to power.While in scenarios of elite unity "elite factions share an underlying consensus about rule of the game and the worth of existing political institutions," elites in overt competition "to protect and promote their own interests […] they must destroy or cripple the regime and the elites who operate it." 77Therefore, regardless of the weakness of the ruling coalition, social forces forming an alternative regime coalition will only engage in coups d'etat under situations of overt elite competition, while other paths to power, such as democratic elections, could be promoted by unified elites.

Case study: Argentina during the 70s
While the previous analyses provide systematic evidence on the relation between insurgent strength, intra-elite competition and coups, it is still unclear which are the set of mechanisms explaining this intriguing relation.In order to provide a partial answer to this question, I rely on a brief analysis of the contentious processes and elite dynamics taking place in Argentina in the 1970s. 78

Context
Argentina's history is marked by the footprint of political violence.However, few episodes attained the levels of atrocity reached from the late 1960s onwards.In 1966, a military coup led by Juan Carlos Onganía ended up with the constitutional presidency of Arturo Illia and avoided any possibility of ceasing the proscription of the majoritarian political force, the Peronism.In a regional scenario led by the advancement of revolutionary armed organizations, the newly Military junta established an authoritarian regime which repressed social protests and engaged in the promotion of a liberalizing economic model.
Two revolutionary guerrilla groups emerged in this context.On the one hand, Montoneros, a convergence of youth Peronist organizations and Christian groups.And on the other hand, the ERP, which acted as the armed branch of the Communist Partido Revolucionario de los Trabajadores (PRT). 79In this scenario of ebullient social turnoil, both Montoneros and the ERP capitalized on the existent dissatisfaction of middle-strata and the working class through an astute combination of targeted selective violence and powerful social mobilization, becoming key political actors prompting the breakdown of the military regime and the return of Perón in 1973.However, and contrary to the expectation of Perón and its closest allies, the comeback of the leader and the implementation of a democratic regime did not lead to the end of armed revolutionary activities.Indeed, the actions of guerrilla organizations reached a yearly average of 750 violent incidents between 1974 and 1976. 80How did elites react to the presence of insurgent groups in the country?

Elite competition and insurgent strength
Even before the onset of violent revolutionary actions in the early 70s, Argentina already had high levels of intra-elite disunity.The political and economic model of the Peronism widened the power of priorly excluded actors, mainly trade unions, the working class and certain sectors of the industrial elite, encompassing an impressive variety of forces with divergent ideological preferences within a unique political movement.This coalition promoted the interests of these groups, to the demise of those of elite outsiders, such as landholders and financial elites.However, the return of Perón in 1973 ended up the period of tactical alliance between the divergent ideologies within the movement.By early 1974, Montoneros was not capable of stopping the advancement of the rightist faction led by López Rega, which had the expulsion of Montoneros from the movement as the main collateral effect.The increasing levels of Montoneros' violence against members of the rightist Peronist wing facilitated the process of ideological purge within the movement.However, this process went beyond the strict boundaries of Peronism, specially after the death of Perón in July 1974 and the arrival to the Presidency of his wife, María Estela Martínez de Perón, which solidified the power of the rightist surge within the regime coalition.The increasing clandestine activities by the paramilitaries of the Triple A 81 led to a vicious spiral of violence between them and the two insurgencies, which already had mid-levels of strength.

Mechanisms
This rising scenario of violence facilitated the incursion of military forces within the government of Martínez de Perón, becoming a pivotal political actor since 1974.The army emerged as "the guardian of an order that civilians could no longer guarantee." 82From that moment onwards, the military not only became the sole actor capable of putting an end to the increasing disorder existent in the country, but also appeared as the only force capable of coalescing the interests of a divergent set of factions whose interests were not being represented by the Peronist movement.
The situation of chronic violence led to a profound crisis of leadership legitimacy for the government of Martínez de Perón, opening the coup opportunity structure.The context of social unrest and the incapacity of the government to tackle the actions of the guerrillas facilitated the engagement in discursive practices justifying the necessity of deep social, political and economic changes under the direction of the military forces.In a gradual but rapid process of profound political polarization, a clear-cut division across two distinct poles arise in the country: subversive and counter-insurgent forces. 83This facilitated the coordination across elites around the figure of the military, as several social forces interpreted this context of chaotic violence as an opportunity for promoting a social, political and economic project that would maximize their interests. 84The statements of the conglomerate APEGE, which represented the interests of landed elites, the alimentary industry, or the financial sector, are clear examples of this 85 : Only an energetic action and a profound change will be able to modify the attitude of distrust, disorientation, lack of authority, lack of security and order in which Argentines live.Once this fundamental condition has been fulfilled, a drastic modification of the economic-social scheme that is leading us to the collapse as a Nation must be faced. 86ese actions proved the lack of support and the decreasing legitimacy suffered by the Peronist government across the social spectrum, including civilians, for which "that perceived threat made it possible for them to accept, advance, and celebrate the drastic program supported by the traditional exporting sector allied with big export-import commerce and the financial sector." 87 The basis for the formation of an alternative regime coalition were set by late 1975, as "there was widespread support in the military, among businessmen, and even among politicians for removing Isabel from office." 88In addition to the rural elite and the financial groups, "moderate union leaders […] were urging the military to step in." 89The gradual forging of an alternative regime coalition marked the beginning of the end of the government of Martínez de Perón.
While multiple mechanisms could explain the process of elite convergence in the midst of civil war in Argentina, most of the historical evidence points to the instrumental exploitation of the subversive threat for the advancement of the interests of certain social groups.Franco 90 argues that, by 1975 "the military public presence became more frequent and visible through the denunciation of the advance of subversion […] The coup tensions were as real as they were instrumentally agitated by various civil and military sectors."In line with this argument, Muleiro 91 shows the instrumentality of violence for the promotion of a new political project: "the armed forces had the required instrument to invent their civil war […] The encompassing nickname of subversive operated as a magic word for [the imposition] of a concrete economic model." The 1976 coup By early 1976, the implementation of a rightist civil-military regime was seen as the only way out to the chaos prompted by guerrilla actions.While rumours of coups were a constant for months, the alternative regime coalition waited until the deepening of the governmental crisis.In the end the economic chaos of 1975, the crisis of authority, the factional struggles, the death present daily, the spectacular action of the guerrilla organizations, and the terror sown by the Triple A, all of this created the conditions for the acceptance of a coup d'état that promised to restore order and ensure the state monopoly of the force. 92 24 March 1976, the infamous coup occurred, setting in motion "the most significant and dramatic economic and social restructuring in the history of Argentina," where the emergence of a new power coalition "resulted in the transition from an industrial society to one based on the financial valorization of capital." 93One of the main goals of the new political project was "to restore the political and economic power of the upper agrarian bourgeoisie and the large economic and financial groups by undermining local industry and the domestic market." 94he arrival of the authoritarian regime brought the highest levels of state repression seen in the Argentinian society in the entire twentieth century.This systematic repression had a direct impact on the trajectory of both Montoneros and the PRT-ERP, which were incapable of sustaining their armed structures against the much stronger military capacity of the newly established civilian-military dictatorship.

Conclusion
Why do some regimes experience coups in the midst of civil wars while others not?Drawing upon the literatures on civil war, coups and social conflict theory, and relying on the use of a novel dataset on Latin American revolutionary socialist insurgencies, this article has argued that the level of strength of insurgent groups, combined with patterns of intra-elite competition, are fundamental factors explaining coup attempts in the midst of internal armed conflicts.Concretely, it has shown that, in scenarios of intra-elite strife, the presence of insurgents with mid-levels of strength facilitates the coordination of elites in order to promote coups against existent political structures.This finding provides an important contribution, both from a theoretical and empirical perspective to the underexplored relation between two contentious processes that, until very recently, have been studied in complete isolation.It also facilitates the understanding of how one of the most overlooked, but fundamental, civil war dynamics, such as insurgent strength, could explain recent wartime coup attempts, such as those in Ethiopia, Mali or Chad.
However, this article has its own limitations.First, the lack of comparable measures on distinct types of intra-elite competition.Second, it is unclear whether the findings would be applicable in distinct regional contexts and for religious and secessionist insurgencies.And third, I have not tested the potential heterogeneous effects of distinct types of coups.
Considering the effects of these events, future studies should build upon these findings and ascertain the complex, dynamic and interactive nature of civil wars and coups d'etat.Shedding light on these issues is of critical importance in the development of a holistic understanding of heterogeneous, but highly interlinked, processes of political violence.

Table 1 .
Coups by strength and intra elite unity.