Conflict externalization and the quest for peace: theory and case evidence from Colombia
CConflict externalization and the quest for peace: Theoryand case evidence from Colombia ∗ Hector Galindo-Silva † Pontificia Universidad JaverianaFirst Draft: June 2014This Draft: August 2020
Abstract
I study the relationship between the likelihood of a violent domestic conflict andthe risk that such a conflict “externalizes” (i.e. spreads to another country by creatingan international dispute). I consider a situation in which a domestic conflict between agovernment and a rebel group has the potential to externalize. I show that the risk of ex-ternalization increases the likelihood of a peaceful outcome, but only if the governmentis sufficiently powerful relative to the rebels, the risk of externalization is sufficientlyhigh, and the foreign actor who can intervene in the domestic conflict is sufficientlyuninterested in material costs and benefits. I show how this model helps to under-stand the recent and successful peace process between the Colombian government andthe country’s most powerful rebel group, the Revolutionary Armed Forces of Colombia(FARC).
Journal of Economic Literature
Classification Numbers: H72, D72
Keywords:
Conflict, externalization, peace talks ∗ I am grateful to Raphael Godefroy, David Karp, Alessandro Riboni and two anonymous referees for helpfulcomments and suggestions. All errors or omissions are mine. † Department of Economics, Pontificia Universidad Javeriana, E-mail: [email protected] a r X i v : . [ ec on . GN ] A ug Introduction
The existing theoretical literature on civil wars often assumes that a group’s decision ofwhether to fight depends exclusively on the domestic context, i.e. on other domestic groups’decisions. However, an increasing empirical literature on conflict is demonstrating that theregional context also plays an important role in domestic conflicts. Researchers are well awareof the regional dimensions of conflicts in Myanmar, Nicaragua, Kosovo, Sudan, Lebanon andIraq, for example. In this paper, I develop a simple model of conflict externalization, and provide new case-study evidence from Colombia. The objective of the model is to formalize an existing theory,and to propose a new mechanism through which the possibility that an external party inter-venes in an internal conflict could affect the outcome of the conflict. In addition to demon-strating the applicability of the model, the Colombian case provides new evidence on thistopic.In the model, a government and a group of rebels simultaneously choose whether to attackeach other. Attacking is costly, but can also decrease the opponent’s military resources, whichincreases the aggressor’s probability of victory. Crucially, the use of violence might cause athird (foreign) actor to join in the conflict. When this happens, I say that the domestic conflict“externalizes.”This externalization changes the power dynamic between the two domestic actors. I assumethat only the government’s use of violence can trigger such an externalization. For example,consider a group of rebels strategically located along a porous border, where the neighboringcountry shares an ideology or ethnicity with the rebels. Given the relationship between therebels and the neighboring country’s government, an aggression might be viewed as a violationof the neighboring country’s sovereignty and could motivate a military response, starting aconflict spiral that might lead to an international war. I focus on how the threat of such anexternal conflict affects the likelihood of peace between domestic actors.This paper’s first main contribution is to show that the risk of externalization increasesthe likelihood of peace, but that this only happens if a government is sufficiently powerfulrelative to rebels, if the risk of externalization is sufficiently high, and if a foreign actor issufficiently uninterested in material costs and benefits.This paper’s second main contribution is new case-study evidence from Colombia. Colom-bia has suffered one of the world’s longest-running internal conflicts. Although many armedgroups have participated in the conflict, the left-wing Revolutionary Armed Forces of Colom-bia (FARC) was always the largest non-government actor. In September 2012, the Colombian See Blattman and Miguel (2010) and Jackson and Morelli (2011) for reviews of this literature. See Mason et al. (1999), Hegre and Sambanis (2006), Gleditsch (2007), Balch-Lindsay et al. (2008) andCunningham et al. (2009, 2011) for evidence showing that the regional context (e.g. a conflict in a neighboringcountry, a highly autocratic region, trans-boundary ethnic groups, or direct intervention of external parties)matters for the onset, incidence and duration of civil wars. See South (2008) for the conflict in Myanmar; Gleditsch and Beardsley (2004) for the Nicaragua-Contrasconflict; Crawford (2001) and Kuperman (2008) for Kosovo; Ali et al. (2005) for Sudan; Bouckaert and Houry(2007) for Lebanon; and Gunter (2008) and Morelli and Pischedda (2013) for the Iraqi-Kurdish conflict. As I will argue later, this situation matches the recent dynamic of the Colombian internal conflict. Itis also consistent with Myanmar-Thailand border clashes prompted by Myanmar pursuit of Karen NationalLiberation Army rebels across the border into Thailand; see South (2008, 2012). As in Kydd and Straus (2013), I find that whether thepotential for an external intervention makes war less likely depends on whether power isbalanced, but unlike them, I find that peace is more likely when the government is stronger,even if an intervention strengthens the rebel group. The outline of the paper is as follows. Section 2 presents the model. Section 3 presentsand analyzes the case study based on the model. Section 4 concludes. Although the argument does not account for “moral hazard” — that an external intervention biased infavor of one party might make this party more belligerent — the model in this paper can be easily extendedto include this possibility, which does not change the key findings. However, this phenomenon does not seemto play an important role in the Colombian conflict, and despite its simplicity, the model is able to providenon-trivial and new and empirical predictions. This paper also shares similarities with Bakshi and Dasgupta (2019), who study how the balance of powerbetween groups within a country affects group conflict in another country. Like Bakshi and Dasgupta, thispaper also studies the relationship between cross-border spillovers and conflict. However, unlike Bakshi andDasgupta (2019), the results in this paper crucially depend on the possibility of interstate conflict. Model
In this section, I develop a simple model that illustrates how the risk of a domestic conflictexternalizing can affect the outcome of the conflict.Suppose that a government faces a rebel group. The government’s main goal is to defeatthe rebels. The rebels are not strong enough to defeat the government, but can harm thegovernment and thereby avoid being defeated. Both parties simultaneously choose whetherto attack the other or pursue peace. If either party attacks, the other party suffers a loss L in military resources, and both parties lose an amount of wealth, C . Importantly, if thegovernment attacks the rebels, then a foreign government F can intervene. This interventionis assumed to favor the rebel group, and always implies that the government is unable todefeat the rebel group. The model focuses on the role of the government’s initial resources in determining theoutcome of the game. I use G to denote these resources. I assume that G ∈ ( L, G ). Theprobability of the government and the rebels winning, for all outcomes in which F does notintervene, depends on G . To model these probabilities, I first define a function Z such that Z ( x ) = 0 for all x ≤
0, so that Z (cid:48) = 0 for all x < Z (cid:48) > Z (cid:48)(cid:48) ≤ x ∈ (0 , G ); and Z ( x ) = 1 for all x ≥ G . This function will be key in the analysis, and can interpreted as thegovernment’s probability of defeating the rebels if the group with the most resources wins, and if the resources of the rebels are random. The assumption that C and L are the same for both groups seem restrictive. However, the results arerobust to potential differences, within certain constraints. The assumption that an external intervention can occur only if the government attacks the rebels iscrucial for the main result (that, under certain conditions, peace is selected as the unique equilibrium). If anexternal intervention could also occur when the rebels attack the government, then conflict would always bean equilibrium, and the conditions for an additional peaceful equilibrium would be harder to achieve. Thus,this alternative scenario would imply weaker results. However, the mechanism would be the same. The detailsfor this alternative scenario are available upon request. For an intuition of the types of situations I try to model, consider a domestic conflict where some rebelsare located close to a border shared with F . Given the proximity of the rebels to F , and the ability of therebels to move between the two countries (because of either porous borders or sympathy from F ), an offensiveaction by the government in the rebels’s controlled territory might harm the citizens of F or be interpreted by F as a violation of its sovereignty. This implies a strong reaction by F to try and ensure that the governmentis unable to defeat the rebels. This intervention can be interpreted either as a direct attack on the government,or as military aid to the rebels. In Section 3, I argue that this scenario is consistent with what occurred priorto the peace process between the Colombian government and the FARC. The fact that
G > L implies that the harm caused by the government’s violence is limited by thegovernment’s resources. The condition
G > G defines an upper bound for G . This means that if the resources of the groups A and B are R A and R B , respectively, then the probabilityof victory for group A is 1 if R A > R B , 0 if R A < R B and 1 / R A = R B . This form of modeling thewinning probabilities is known as an “all-pay auction.” It has been used to study contests in which groupmembers exert effort that translate into “group effort” (see for instance Baik et al., 2001; Barbieri et al., 2014;Chowdhury et al., 2016; Chowdhury and Topolyan, 2016a,b). Under this interpretation, the scenarios in whichone group defends against another group’s attack (as in Chowdhury and Topolyan, 2016a), and where theaggregation technology is asymmetric among the contesting groups (as in Chowdhury and Topolyan, 2016b)are particularly important. To see this, define R as the resources of the rebels, and assume that R is a random variable on R witha cumulative distribution function Z such that Z ( x ) = 0 for all x ≤ Z (cid:48) > Z (cid:48)(cid:48) ≤ x ∈ (0 , G ),and Z ( x ) = 1 for all x ≥ G . Thus, if the winning probabilities are defined as an “all-pay auction,” we have Z ( G ) be the government’s probability of defeating the rebels when bothparties attack and F does not intervene. When the rebels attack but the government choosespeace, the government loses L in military resources, so its probability of winning is Z ( G − L ).The winning probabilities are defined analogously when the government attacks but neitherthe rebels nor the foreign country attack. When no group attacks, peace occurs; in thisscenario, the parties compete in an election in which each party’s probability of winningdepends on the relative initial resources. Finally, F ’s decision of whether to intervene depends negatively on the resources of thegovernment: the greater these resources, the less likely it is that F will intervene because F ’spotential losses are greater. However, and importantly, F may not be purely interested inmaterial costs and benefits. F may have ideological or religious motivations, which, insofaras they involve agents that can be difficult to negotiate with, can be viewed as being notmaterially based (in this respect, see Jackson and Morelli, 2011). Specifically, F ’s goals arepurely material with probability 1 − φ , in which case its decision to intervene depends on G . F ’s goals are not materially based with probability φ , in which case the probability of anintervention is assumed to be exogenous and normalized to one. Timing (1) The government and the rebels simultaneously decide whether to attack each other.(2) If either party decides to attack, a violent conflict occurs. If neither party attacks,peace occurs, and the parties compete in an election in which each party’s probabilityof winning depends on its relative initial resources.(3) If there is a violent conflict, both parties face exogenous costs
C >
0. In addition, themilitary resources of a party decrease by L if it is attacked.(4) F observes whether the government attacked. If the government attacked, F interveneswith probability φ , and makes a determination on whether or not to intervene based onG with probability (1 − φ ). If F intervenes, the government cannot defeat the rebels.(5) The violent conflict ends. Each group receives a payoff according to the outcome, whichis described in the next subsection. that the government’s probability of victory is P r [ G > R ] = Z ( G ) if both groups attack. The other winningprobabilities can be defined analogously. The randomness of R is consistent with the rebels being somewhatinformal. This may be particularly true if the rebels are located along the borders with F , if these borders areporous, and if there is uncertainty about how effectively F surveils its border (as the level of arms or troopsflowing in from F may be unknown). As previously mentioned, when F intervenes, the government’s probability of defeating the rebels is 0. This can be interpreted as an scenario in which the political parties that represent the interests of thegroups compete for seats in the national assembly under a proportional representation system. It is alsoconsistent with a number of seats in the national assembly being given to the rebels under the terms of apeace deal that guarantees proportional representation for the rebels. These scenarios are consistent withwhat occurred after the 2016 peace accord between the Colombian government and the FARC. As will be discussed in Section 3, ideology is crucial to understanding Venezuela’s motivation to intervenein the Colombian conflict. In addition to Jackson and Morelli (2011), see Maynard (2019), who discusses therole of ideology in armed conflicts, and Owen (2010), who provides examples in international politics. This normalization is without loss of generality; what matters is that the probability of an interventionwhen F ’s goals are not materially based is exogenous. ayoffs The preferences of the domestic groups are characterized by a Bernoulli utility function,where each party gets a payoff of 1 if it wins, and 0 if it loses. The payoff function of thegovernment is denoted by π , and the payoff function of the rebels is denoted by ρ . Note thatthese functions depend on what the domestic groups simultaneously decide (either attack ( a )or peace ( p )), as well as on the foreign government’s decision about whether to intervene ( a )or not ( p ).Since the government and rebels do not know whether F will intervene when deciding ontheir military actions, I use Φ to denote each group’s expectation that such an interventionwill occur (conditional on an attack by the government). Thus, given an attack by the rebels,the government’s expected payoff from attacking is π ( a, a ) = (1 − Φ) Z ( G ) − C . The expectedpayoffs for all cases, after rearranging and omitting some constants, are given in Table I.
Rebels ( a ) ( p ) Govt. ( a ) π = (1 − Φ) Z ( G ) − C π = (1 − Φ) Z ( G + L ) − Cρ = − (1 − Φ) Z ( G ) − C ρ = − (1 − Φ) Z ( G + L ) − C ( p ) π = Z ( G − L ) − C π = Z ( G ) ρ = − Z ( G − L ) − C ρ = − Z ( G ) Table I: Expected payoffs for government and rebelsI model F ’s decision about whether to intervene in a very simple and stylized way. First,I assume that a foreign intervention is less likely when the government is stronger. Then,I define W ( G ) as the probability of an intervention from the perspective of the governmentand rebels when F is interested in material benefits, and where W is a function such that W ( x ) = 0 for all x ≥ A > G , W ( x ) = 1 for all x ≤
0, and for all x ∈ (0 , A ), W (cid:48) < W (cid:48)(cid:48) ≤ Additional assumptions
Assumption 1.
C > Z ( L ) . Assumption 1 guarantees that the cost of using violence is high enough that a peaceagreement is always possible.
Assumption 2.
Let sup (cid:8) Z (cid:48) ( G ) W (cid:48) ( G ) | G ∈ ( L, G ) (cid:9) ≡ k . Then W ( G ) k < − . Note that, with some abuse of notation, I have defined π ( a, a ) ≡ E x [ π ( a, a, x )], where x ∈ { a, p } is F ’schoice, and E x is the government’s expectation about x . A very simple micro-foundation for W is the following. First, define F ’s payoff as equal to B − G if F intervenes, and 0 if it does not intervene, where B represents any material benefit derived from the intervention.Then, assume that B is unknown to the government and the rebels and that, from the point of view of theseparties, it has a cumulative distribution function V defined on R such that V ( x ) = 1 for all x ≥ A > G , V ( x ) = 0 for all x ≤
0, and for all x ∈ (0 , A ), V (cid:48) > V (cid:48)(cid:48) ≥
0. Finally, set W ( x ) ≡ − V ( x ). Assumption 3. (1 − W ( G )) > Z ( G − L ) . Assumption 3 guarantees that given an attack by the rebels, the government’s best re-sponse function dictates that the government should also attack, even if the probability of aforeign intervention is small.
I will now characterize the equilibrium of the game. I focus on the intuition for the mainequilibrium decisions, and leave the formal derivation to the Appendix.First, note that from the point of view of both the government and rebels, a foreignintervention will occur with probability Φ = φ + (1 − φ ) W ( G ) (conditional on the governmentattacking). Replacing this expression in the payoff functions of the government and the rebels,and computing each group’s best response function, it is possible to establish a set of non-trivial conditions under which a conflict occurs in equilibrium. These conditions focus on theexogenous component of the risk of a foreign intervention (i.e. φ ), and on the government’sresources (i.e. G ). They constitute the main result of this section, which is summarized inthe following proposition: Proposition 1.
Consider the above-described game. Let φ ≡ − Z ( G − L )(1 − W ( G )) . Given Assumptions1-3, the following hold:(i) Given any φ ∈ [0 , and any G ∈ ( L, G ) , ( p, p ) must constitute a Nash equilibrium.(ii) Whenever φ ≤ φ , ( a, a ) constitutes a Nash equilibrium for all G ∈ ( L, G ) .(iii) Whenever > φ > φ , there exists ˆ G ( φ ) ∈ ( L, G ) , with ˆ G (cid:48) ( φ ) < , such that ( a, a ) constitutes a Nash equilibrium if and only if G ≤ ˆ G ( φ ) .(iv) Whenever φ = 1 , ( p, p ) constitutes the unique Nash equilibrium.(v) Neither ( a, p ) nor ( p, a ) can constitute a Nash equilibrium.Proof. See Appendix.To understand the intuition behind Proposition 1, first note that if the risk of an externalintervention is low, a well-resourced government can easily defeat the rebels. Such a govern-ment will have a strong incentive to attack the rebels, so a violent conflict will constitute aNash equilibrium. In addition, note that this scenario is consistent with the assumption thatthe risk of an external intervention decreases with the strength of the government.However, and importantly, recall that the risk of an external intervention only partiallydepends on the government’s resources. How does this affect the mechanism described in the This assumption can be relaxed at the cost of much less general functions Z and W . Two functions Z and W that satisfy Assumption 2 are: Z ( x ) = x / G if x ∈ [0 , G ], Z ( x ) = 0 if x < Z ( x ) = 1 if x > G , and W ( x ) = (1 − x / A ) for x ∈ [0 , A ] and with A > G . More generally, when Z ( x ) = ( x / G ) β with β ∈ (0 , W ( x ) = (1 − x / A ) γ with γ ∈ (0 , βγ )( A − GG ) >
1. I thanka referee for suggesting this Assumption and example. p, p ) is selected as the unique Nash equilibrium).This mechanism is based on the idea that when the risk of an external intervention doesnot entirely depend on the material resources of the groups in conflict, a well-resourced gov-ernment may have less incentive to attack the rebels. Why? Because this government is betterable to tolerate a rebel attack, and by doing so, it can decrease the risk of an external inter-vention. Importantly, this may happen even if a well-resourced government faces a relativelylow probability of an external intervention. For a well-resourced government, the pro-peaceeffects of resources (a greater capacity to tolerate a rebel attack) can outweigh pro-war effects(a greater capacity to defeat the rebels).Proposition 1.(iii) establishes two conditions for the selection of the peaceful scenario de-scribed in the last paragraph: the foreign actor should be sufficiently uninterested in materialcosts and benefits (such that the strength of the government is not a big deterrent to interven-ing), and the government should be sufficiently resourced (such that it can tolerate an attackfrom the rebels without being seriously harmed). To the best of my knowledge, this resultis new in the literature. In the next section, I show that it is crucial to understanding theintriguing peace deal between the Colombian government and the FARC.
In this section, I show how the model proposed in the previous section explains case studyevidence from Colombia. I use information from the main actors in the Colombian conflict,as well as from the secondary academic literature.
The conflict between the Colombian government and the left-wing rebels dates back to thelate 1950s. Its origins have been associated with the founding of the FARC, which was alwaysColombia’s largest and best-equipped rebel group. Between 1958 and 2012, the conflictclaimed at least 220,000 lives (see GMH, 2013, p. 31).From the 1980s until its end, the Colombian conflict was accompanied by several negoti-ations, including three failed peace talks with the FARC (see Sanchez, 2001; Chernick, 2009;Nasi, 2009). This section focuses on the most recent (and successful) peace talks, which oc-curred from 2010 to 2016. These peace talks were preceded by the “Cagu´an” peace process,which occurred from 1998 to 2002, and which failed miserably. A relatively weak government,amateurish bargaining teams and spoilers (actors who use violence to undermine peace talks)are some of the explanations given for Cagu´an’s failure (see Kline, 2007; Nasi, 2009). Thebreakdown of the Cagu´an peace process led to the election, in 2002, of a hawkish and far-right If either of these two conditions are not satisfied, Proposition 1.(ii) and Proposition 1.(iii) imply that, inequilibrium, both the government and the rebels will attack (which means that ( a, a ) also constitutes a Nashequilibrium). A weak government will attack the rebels (regardless of the risk of foreign intervention) whenthe government has a limited ability to defend itself from a rebel attack if the government chooses peace. Other smaller rebel groups participated in Colombia’s conflict. These include other left-wing insurgentsand right-wing paramilitaries. The most important left-wing insurgent other than the FARC is the NationalLiberation Army (ELN). The main right-wing paramilitary group was the United Self-Defense Forces ofColombia (AUC), which officially demobilized in 2006.
During the 2000s and early 2010s, Colombia’s government pursued an all-out military effortagainst the FARC. Due to an increase in defense spending and a significant improvement inmilitary effectiveness, the Colombian government achieved relative success: during this period,the FARC suffered the worst blows in its history. According to the Colombian Ministry ofDefense, an average of 40 FARC members were captured or killed annually from 2002 and2011. In the same period, roughly 17 members deserted the FARC each year, and the numberof FARC combatants was halved. The government also killed the head of the FARC andseveral other leaders, in an action characterized by the Colombian president as “the mostdevastating blow that this group has suffered in its history.” These actions were deeplyresented by the FARC. During this period, the FARC retreated from key regions in the center of the country (i.e.the departments of Cundinamarca, Tolima, and Santander) to border areas with Venezuelaand Ecuador (i.e. to the departments of Nari˜no, Cauca, Caquet´a, Norte de Santander andArauca) (see IISS, 2011; Avila, 2013). The FARC’s decision to move to the periphery of thecountry was strategic, given ideological similarities between the FARC and the governmentsof Venezuela and Ecuador, and the porousness of the borders with these two countries (seeOwen, 2010, p. 257). The FARC’s decision, combined with a hawkish military strategy bythe Colombian government, brought the three countries to the brink of war. I argue that italso raised the likelihood of a peaceful solution.
In March 2008, Colombian security forces crossed into Ecuador to assault an outpost of theFARC. More than two dozen rebels were killed, including a high-ranking leader thought bymany to be FARC’s second-in-command. The Colombian government also captured computerswith documents indicating that Venezuela had been supporting the FARC (see IISS, 2011).The assault caused a serious diplomatic incident between Colombia and Ecuador. Ecuadorimmediately broke off diplomatic relations with Colombia. Venezuela, in solidarity withEcuador, expelled Colombia’s ambassador and other diplomats. Venezuela and Ecuadoralso sent troops to the Colombian border, advising that any additional violations of theirsovereignty would result in war. The tension reached a peak in July 2010 when, weeksbefore a change in Colombia’s government, the Colombian press secretary provided evidenceof a FARC presence in Venezuela to international authorities. See Ministerio de Defensa (2009) and . The exactnumber of FARC members who deserted or were captured or killed has been debated by analysts (Avila, 2013;Rico, 2013) and the FARC (FARC-EP, 2013b, Feb. 12), but analysts agree with the government that thenumber of combatants fell by roughly half during the 2000s. Translation by the author from . See . See The New York Times (2008, Mar. 4) See . According to the Colombian press secretary, “For six years the Colombian government sustained a patient
9n the following days, the Colombian Ambassador to the Organization of American States(OAS) presented photographs, maps, coordinates, and videos proving the presence of illegalarmed groups in Venezuelan territory. Venezuela reacted by breaking off diplomatic relationswith Colombia, sending more troops to the border and ordering them to be on full alert.Venezuela’s then-President Hugo Chavez said:
To maintain our dignity, we do not have any other option but to sever diplomatic ties withColombia [...] We will be on alert —I have ordered the maximum alert along our border[...] Uribe is a threat for peace. He is even able to establish a fake camp in our territoryand raid it to start a war. The Colombian government was aware of the high risk of war, particularly if it violated aneighbor’s sovereignty. The Colombian Minister of Defense at the time, Gabriel Silva, said:
I said privately to President Uribe, “If you give an authorization, I will bring back toColombia [all FARC leaders] who are there [in Venezuela]” [...] He did not authorize. Hesaid it was too risky for the country and for national security [...] I do believe that therewas very nearly a war with Venezuela. (Davila, 2014, p. 89-91; translation of the author).
In July 2010, Colombia severed diplomatic relations with Venezuela. Two weeks later, anew Colombian president, Juan Manuel Santos, took office. Known for his strategic prag-matism, Santos was closely associated with Uribe’s successful military campaigns against theFARC and elected with a mandate to continue Uribe’s hard-line policies. From his early days in office, Santos combined an extremely aggressive campaign againstthe FARC with efforts to improve diplomatic relations with Venezuela. Ten days after hewas sworn in, diplomatic relations with Venezuela were restored, and approximately onemonth later, Santos announced the death of the FARC’s second-in-command and leader of itsstrongest fighting division. One year later, the FARC’s top leader, who went by the nom deguerre of Alfonso Cano, was killed. The FARC’s choice for Cano’s replacement, whose nomde guerre was Timochenko, was also influential: Timochenko was known for operating alongthe border with Venezuela and for having lived there previously. In addition, many peoplehad raised concerns about the close ties between Timochenko and important figures in theVenezuelan government. In September 2012, almost one year after Timochenko had become the FARC’s leader,the Colombian president announced that his government and the FARC had agreed to starta peace process. The announcement surprised analysts and national leaders. These peace dialogue with the Venezuelan government, on various occasions providing it information on the location ofterrorists in that territory. All was unsuccessful with respect to terrorist leaders. We must once again considertaking the matter to international authorities.” See Presidencia (2010, July 16). See OAS (2010, July 22) and . Translation by the author from . As Serbin and Serbin(2017) say, “this mobilization of troops was not a minor factor [...] Since 1987 [....] there had been nomobilization of the Venezuelan military at this scale” (see Serbin and Serbin, 2017, p. 241). See Revista Semana (2010, June 12). See El Tiempo (2011b, Nov. 15) and El Tiempo (2011a, Nov. 19). Timochenko was called the FARC’s “ambassador” to Venezuela (see Caracol Noticias, 2010, Mar. 2). See Revista Semana (2012, Sept. 3) for initial reactions. In October 2016, Juan ManuelSantos was awarded the Nobel Peace Prize for his efforts to end the Colombian civil war. Why did Colombia and the FARC decide to pursue peace? I suggest that the risk ofexternalization of the Colombian conflict to Venezuela, the politicization of Venezuela’s foreignpolicy, and the military strength of the Colombian government are at the root of the peacedeal, creating what some literature on conflict has called a “ripe for resolution” situation (e.g.see Zartman, 2000).The FARC’s motivations for peace seem clear: it suffered significant setbacks between2002 and 2011, and its leaders seemed convinced that they had no chance of defeating theirenemy. Talking about the relevance of the setbacks, the FARC’s leader, Timochenko, said in2012: “I can’t deny we’ve received serious blows —and extremely painful ones. The deathsof four members of the National Secretariat can’t be minimized [...] it’s obvious that today’sconditions are not the same as a decade ago.” Asked why the FARC decided to negotiatewith Santos, Timochenko responded that the costs of continuing the conflict would have beenvery high: Whatever may come, persistent conflict will entail many more deaths and great destruc-tion, more sorrow and tears, more poverty and misery for some and greater wealth forothers. Imagine the lives that could have been saved over the past 10 years. That’s whywe seek negotiations, a solution without blood, and an understanding through politicalroutes. (Carlos Lozano’s website, 2012, Sept. 19; translation of the author).
The Colombian government’s motivations for negotiating with the FARC are less apparent.Analysts and the Colombian government have highlighted the increase in the power of Colom-bian forces relative to the FARC. But if the Colombian government was winning the war,why wouldn’t it continue fighting the FARC for a few more years, as advocated for by formerpresident Uribe? War is costly and unpredictable, so rational agents should have incentivesto reach peaceful settlements that all would prefer to war. However, from the point of viewof the majority of the Colombian population, media, and important leaders, just before thepeace talks were made public, the FARC were close to being defeated. Thus, the govern-ment’s cost of war could be perceived as being very low. In addition, at the time, a peace See The New York Times (2016a, Sept. 26). See The New York Times (2016b, Oct. 7). Carlos Lozano’s website (2012, Sept. 19), translation of the author. The FARC’s strategy may have always been respond to peace with peace. According to the FARC’sleader, the FARC “negotiate because a political solution has always been our objective, and also that of thepeople’s movement” (Carlos Lozano’s website, 2012, Sept. 19, translation of the author). This awareness wasalso recognized by FARC’s representative of the peace delegation, Pablo Catatumbo: “We are ready to startpreparing the way that will lead us towards the expression of our regret for what has happened ... No doubtthere has also been harshness and pain caused from our side.” (FARC-EP, 2013a, translation of the author). For instance, Colombian president Santos said in 2012, “If we can talk about peace now [...] it is becauseof the effectiveness of our armed forces.” (See Presidencia, 2012b, Oct. 25, translation of the author). See Caracol Noticias (2012, Oct. 28). I argue that themain reasons the government chose peace was the high risk of an international conflict withVenezuela, the extreme politicization of Venezuela’s foreign policy, and the Colombian gov-ernment’s significant military strength. According to the theory proposed in Section 2, theseconditions are sufficient for a peaceful equilibrium.Were these conditions satisfied just before the start of the peace talks between the Colom-bian government and the FARC? I already argued (in subsections 3.1.1 and 3.1.2) that twoof these conditions were likely satisfied. First, I showed that just before the start of thepeace talks, the government was as strong as ever. A military victory by the FARC (or anyother non-state armed group) was seen as so unlikely that the discussion was entirely focusedon when — not if — the government would defeat the guerrilla groups. Second, I showedthat just before Santos took power, the risk of an interstate conflict between Colombia andVenezuela was extremely high. This risk was directly related to the presence of the FARCin Venezuelan territory, the very likely possibility that Colombia might violate Venezuelansovereignty to pursue the rebels, and Venezuela’s determination to respond if this happened.The third condition — the extreme politicization of Venezuela’s foreign policy — wasalso met. According to Serbin and Serbin (2017), Venezuela’s oil-based foreign activism wasamplified under the presidency of Hugo Chavez, and, importantly, included: “the unbridled involvement of the president in foreign affairs and its extreme politicization[...] [and this trait was expressed in] the role assigned to the armed forces in dealing witha potential asymmetric conflict with the United States, in developing links and militaryexchanges with countries in the region and beyond, and in establishing ties with irregularforces such as the guerrilla movements” (Serbin and Serbin, 2017, p. 239)
The conflict with Colombia just before the start of the peace talks seems to be a perfect exam-ple of this activism: Colombian had Latin America’s most pro-American president, the Colom-bian armed forces were perceived by Chavez as being commanded by the United States, andthe FARC and Chavez were not only ideologically very close (i.e. both were left-wing adanti-American), but shared an explicit Pan-American “Bolivarian” discourse. In this paper, I develop a simple model of conflict externalization and provide new casestudy evidence from Colombia. The first main contribution of the paper is to show that therisk of externalization of a domestic conflict increases the likelihood of peace, but that thisonly happens if the domestic government is sufficiently powerful, if the risk of an externalintervention is sufficiently high, and if the foreign party that may intervene is sufficientlyuninterested in material costs and benefits. In the speech announcing the opening of peace talks, Santos said, “There comes a moment in historywhen you have to take risks to arrive at a solution [...] This is one of those moments.” (See Presidencia, 2012a,Sept. 4, translation of the author). Motivated by Colombia’s plans to allow the US to use seven of its military bases, Chavez said in 2009:“The Yankees are starting to command the Colombian armed forces. They are the ones who are in charge”(Financial Times, 2009, Agu. 10) See Revista Semana (2008, Jan. 19).
12n the second part of the paper, I use the model to examine the Colombian conflict. I focuson peace talks that occurred between 2010 and 2016 between the Colombian government andthe FARC. The second main contribution of the paper is to show how the risk of externalizationof the conflict to Venezuela played a crucial role in the success of these talks.Although the theory is inspired by the Colombian conflict, its application is not limited tothis case. The model can be applied to any internal conflict in which governments, undertakingcross-border counterinsurgency actions, initiate military actions against neighboring states.While other explanations exist to explain how Colombian peace talks evolved, such as lessideological extremism and less militarism from both parties, the evidence shows that thepossibility of externalization should be considered in any examination of the issue.13 ppendix
Proof of Proposition 1.
I solve the game by backward induction. Having examined F ’s decisionabout whether to intervene in the main text, I must now examine the other groups’ optimal decisionsabout whether to attack each other. I start by showing that, under Assumption 1, peace (i.e., ( p, p ))always constitutes a Nash equilibrium of the game. Then, I show that under Assumptions 2 and 3,there is a threshold value for φ that determines whether war (i.e., ( a, a )) is also a Nash equilibriumof the game. I conclude by showing that neither ( a, p ) nor ( p, a ) can constitute a Nash equilibrium.First, define π ( p, p ) − π ( a, p ) = Z ( G ) − (1 − φ )(1 − W ( G )) Z ( G + L ) + C , where π ( p, p ) and π ( a, p ) are from Table I, and where I have used the fact that Φ = φ + (1 − φ ) W ( G ). Now note thatif π ( p, p ) > π ( a, p ), then when the rebels seek peace, the government’s best response is also to seekpeace. In the last expression, note that π ( p, p ) > π ( a, p ) if and only if C > (1 − φ )(1 − W ( G )) Z ( G + L ) − Z ( G ) . (1)Now note that since Z (0) = 0, Z ( x ) = 1 for all x ≥ G , and Z (cid:48) > Z (cid:48)(cid:48) ≤ x ∈ (0 , G ),then Assumption 1 implies that C > Z ( x + L ) − Z ( x ) for x ≥
0. In particular, Assumption 1 impliesthat
C > Z ( G + L ) − Z ( G ) for G ∈ ( L, G ). Thus, since (1 − φ )(1 − W ( G )) ≤ G ∈ ( L, G ),we have that (1) holds for all G ∈ ( L, G ).Now define ρ ( p, p ) − ρ ( p, a ) = − Z ( G ) + Z ( G − L ) + C , where ρ ( p, p ) and ρ ( p, a ) are from Table I,and where I have used the fact that Φ = φ + (1 − φ ) W ( G ). Note that if ρ ( p, p ) > ρ ( p, a ), the rebels’best response is to seek peace, given that the government also seeks peace. In the last expression,note that ρ ( p, p ) > ρ ( p, a ) if and only if C > Z ( G ) − Z ( G − L ) . (2)Since C > Z ( x + L ) − Z ( x ) for all x ≥ x = G − L for G ∈ ( L, G ), we have that (2) holds for all G ∈ ( L, G ).We therefore have that peace (i.e., ( p, p )) constitutes a Nash equilibrium of the game. This provesProposition 1.(i).Now I establish the conditions under which war (i.e., ( a, a )) is also a Nash equilibrium of thegame. To do this, first define D ≡ π ( p, a ) − π ( a, a ), and note that if D >
0, then the government’sbest response to a rebel attack is to choose peace, and if
D <
0, the government’s best response isto attack. From Table I, note that D is equal to D = Z ( G − L ) − (1 − φ )(1 − W ( G )) Z ( G ) (3)where I have used the fact that Φ = φ + (1 − φ ) W ( G ). Differentiating (3) with respect to G , we get dDdG = Z (cid:48) ( G − L ) − (1 − φ )(1 − W ( G )) Z (cid:48) ( G ) + (1 − φ ) W (cid:48) ( G ) Z ( G ) (4)which, rearranging the terms, is equivalent to dDdG = [ Z (cid:48) ( G − L ) − (1 − φ ) Z (cid:48) ( G )] + (1 − φ ) W (cid:48) ( G ) Z ( G ) (cid:104) W ( G ) Z (cid:48) ( G ) W (cid:48) ( G ) Z ( G ) (cid:105) . (5)In (5), note that Z (cid:48) ( G − L ) − (1 − φ ) Z (cid:48) ( G ) ≥ Z (cid:48)(cid:48) ( x ) ≤ x ∈ ( L, G ). Thus, given any φ ∈ [0 , dDdG > W ( G ) Z (cid:48) ( G ) W (cid:48) ( G ) Z ( G ) < iven that W (cid:48) < x ∈ (0 , A ) where A > G . I will now show that Assumption 2 implies that(6) holds for all G ∈ ( L, G ). To see this, note that1 + W ( G ) Z ( G ) Z (cid:48) ( G ) W (cid:48) ( G ) ≤ W ( G ) Z ( G ) k (7)where k ≡ sup { Z (cid:48) ( G ) W (cid:48) ( G ) | G ∈ ( L, G ) } . In addition, note that since W (cid:48) ( x ) < Z (cid:48) ( x ) > x ∈ (0 , G ), we have that k ≤
0, and that1 + W ( G ) Z ( G ) k ≤ W ( G ) k ≤ W ( G ) k. (8)Combining (7) and (8), we have that1 + W ( G ) Z ( G ) Z (cid:48) ( G ) W (cid:48) ( G ) ≤ W ( G ) k. (9)So, if 1 + W ( G ) k < φ ∈ [0 ,
1] and for all G ∈ ( L, G ).Next, note that dDdG > L < G implies that D ∈ (cid:0) − (1 − φ )(1 − W ( L )) Z ( L ) , Z ( G − L ) − (1 − φ )(1 − W ( G )) Z ( G ) (cid:1) (10)for any G ∈ ( L, G ). Since − (1 − φ )(1 − W ( L )) Z ( L ) < φ <
1, then, for this case, thegovernment’s best response to a rebel attack is always to counterattack if Z ( G − L ) − (1 − φ )(1 − W ( G )) Z ( G ) <
0, or, equivalently, if Z ( G − L ) − (1 − φ )(1 − W ( G )) < Z ( G ) = 1. To identify the conditions under which (11) holds, define φ ≡ − Z ( G − L )(1 − W ( G )) (12)and note that under Assumption 3, φ >
0. From (12), note that when φ < φ , (11) holds. Thismeans that when φ < φ , the government’s best response to a rebel attack is to also attack, regardlessof the value of G . And since the rebel’s best response to a government attack is clearly to attack,then whenever φ ≤ φ and φ <
1, ( a, a ) there is a Nash equilibrium for all G ∈ ( L, G ). This provesProposition 1.(ii).From (12), note that φ <
1, so there are values of φ such that 1 > φ > φ . For these values, notethat (11) does not hold. Combining this observation with the fact that dDdG > G ∈ ( L, G )(and given any φ ∈ [0 , > φ > φ , there exists ˆ G ( φ ) ∈ ( L, G ), definedimplicitly by Z ( ˆ G − L ) = (1 − φ )(1 − W ( ˆ G )) Z ( ˆ G ) (13)such that ( a, a ) constitutes a Nash equilibrium if and only if G ≤ ˆ G ( φ ). To see that ˆ G (cid:48) ( φ ) < φ to get ˆ G (cid:48) ( dDdG | G = ˆ G ) + (1 − W ( ˆ G )) Z ( ˆ G ) = 0; then notethat the fact that for any φ ∈ [0 , dDdG > G ∈ ( L, G ), and that (1 − W ( ˆ G )) Z ( ˆ G ) >
0, implythat ˆ G (cid:48) <
0. This proves Proposition 1.(iii).Finally, recall from the analysis above that the best response of both the rebels and governmentis to seek peace when their opponent also seeks peace. This implies that neither ( a, p ) nor ( p, a )can constitute a Nash equilibrium (i.e., Proposition 1.(v)). This, combined with the fact that when φ = 1, D = Z ( G − L ) > G ∈ ( L, G ) proves Proposition 1.(iv). eferences Ali, Ali Abdel, Ibrahim A. Elbadawi, and Atta El-Batahani , “Sudan’s Civil War. Why HasIt Prevailed for So Long?,” in Paul Collier and Nicholas Sambanis, eds.,
Understanding Civil War.Evidence and Analysis. Volume 1: Africa , World Bank, 2005, pp. 193–219.
Amegashie, J. Atsu and Edward Kutsoati , “(Non)intervention in intra-state conflicts,”
Euro-pean Journal of Political Economy , September 2007, (3), 754–767. Avila, Ariel Fernando , “Del Cagu´an a la Habana,” Informe, Corporacion Nuevo Arco iris 2013.[Online; accessed 28 April 2014].
Baik, Kyung Hwan and Sanghack Lee , “Two-Stage Rent-Seeking Contests with Carryovers,”
Public Choice , 2000, (3/4), 285–296. , In-Gyu Kim, and Sunghyun Na , “Bidding for a group-specific public-good prize,”
Journalof Public Economics , 2001, (3), 415 – 429. Bakshi, Dripto and Indraneel Dasgupta , “Identity Conflict with Cross-Border Spillovers,”
Defence and Peace Economics , 2019, (0), 1–24. Balch-Lindsay, Dylan, Andrew J. Enterline, and Kyle A. Joyce , “Third-Party Interventionand the Civil War Process,”
Journal of Peace Research , 2008, (3), 345–363. Barbieri, Stefano, David A. Malueg, and Iryna Topolyan , “The best-shot all-pay (group)auction with complete information,”
Economic Theory , 2014, (3), 603–640. Blattman, Christopher and Edward Miguel , “Civil War,”
Journal of Economic Literature ,March 2010, (1), 3–57. Bouckaert, Peter and Nadim Houry , Why They Died: Civilian Casualties in Lebanon Duringthe 2006 War , Human Rights Watch, 2007.
Caracol Noticias , “Alias ‘Timochenko’ era el embajador de las Farc en Venezuela, seg´un computa-dores de ‘Ra´ul Reyes’,” March 2 2010. Online; consulted 1 April 2014., “Lea aqu´ı el discurso completo del expresidente Alvaro Uribe,” October 28 2012. Online;consulted 1 April 2014.
Carlos Lozano’s website , “Comandante Timole´on Jim´enez: “Hemos estado dispuestos a lab´usqueda de la Paz”,” september 19 2012. Online; consulted 1 April 2014.
Carment, David and Dane Rowlands , “Three’s Company: Evaluating Third-Party Interventionin Intrastate Conflict,”
Journal of Conflict Resolution , 1998, (5), 572–599. and , “Force and Bias: Towards a Predictive Model of Effective Third-Party Intervention,” Defence and Peace Economics , 2006, (5), 435–456. Chernick, Marc , “The FARC at the Negotiating Table,” in Virginia Marie Bouvier, ed.,
Colombia:Building Peace in a Time of War , United States Institute of Peace Press, 2009.
Choi, Jay Pil, Subhasish M. Chowdhury, and Jaesoo Kim , “Group Contests with InternalConflict and Power Asymmetry,”
The Scandinavian Journal of Economics , 2016, (4), 816–840. howdhury, Subhasish M. and Iryna Topolyan , “The Attack-And-Defense Group Contests:Best Shot Versus Weakest Link,” Economic Inquiry , January 2016, (1), 548–557. and , “Best-shot versus weakest-link in political lobbying: an application of group all-payauction,” Social Choice and Welfare , December 2016, (4), 959–971. , Dongryul Lee, and Iryna Topolyan , “The Max-Min Group Contest: Weakest-link (Group)All-Pay Auction,” Southern Economic Journal , 2016, (1), 105–125. Crawford, Timothy W. , “Pivotal Deterrence and the Kosovo War: Why the Holbrooke AgreementFailed,”
Political Science Quarterly , 2001, (4), pp. 499–523.,
Pivotal Deterrence: Third-party Statecraft and the Pursuit of Peace , Cornell University Press,2003.
Cunningham, David E., Kristian Skrede Gleditsch, and Idean Salehyan , “It Takes Two:A Dyadic Analysis of Civil War Duration and Outcome,”
Journal of Conflict Resolution , 2009, (4), 570–597. , Kristian Skrede Gleditsch, and Idean Salehyan , “Explaining External Support for Insur-gent Groups,” International Organization , 9 2011, , 709–744. Davila, Vicky , Enemigos: Santos y Uribe Por qu´e se odian? , Editorial CAMM, 2014.
El Tiempo , “B´usqueda de ‘Timochenko’, ‘papa caliente’ para Colombia y Venezuela,” November19 2011. Online; consulted 1 May 2014., “‘Timochenko’, otro radical en la jefatura de las Farc,” November 15 2011. Online; consulted 1May 2014.
FARC-EP , “The Conversation Table, a triumph of the national clamor for peace and the politicalsolution,” August 2013. Online; consulted 1 April 2014., “El ‘tope’ de las FARC-EP,” February 12 2013. Online; consulted 1 May 2014.
Fearon, James D. , “Commitment Problems and the Spread of Ethnic Conflict,” 1998, pp. 107–126.
Financial Times , “Chavez ups rhetoric in spat with Colombia,” August 2009. Online; consulted24 May 2020.
Gleditsch, Kristian Skrede , “Transnational Dimensions of Civil War,”
Journal of Peace Research ,2007, (3), 293–309. and Kyle Beardsley , “Nosy Neighbors: Third-Party Actors in Central American Conflicts,” The Journal of Conflict Resolution , 2004, (3), pp. 379–402. GMH , Grupo de Memoria Hist´orica: Basta ya! Colombia: memorias de guerra y dignidad. InformeGeneral , Imprenta Nacional, 2013.
Gunter, Michael M. , The Kurds Ascending: The Evolving Solution to the Kurdish Problem inIraq and Turkey , Palgrave Macmillan, 2008. ausken, Kjell , “Production and Conflict Models Versus Rent-Seeking Models,” Public Choice ,2005, (1), 59–93.
Hegre, Hvard and Nicholas Sambanis , “Sensitivity Analysis of Empirical Results on Civil WarOnset,”
Journal of Conflict Resolution , 2006, (4), 508–535. IISS , “The Farc Files: Venezuela, Ecuador and the Secret Archive of ‘Ra´ul Reyes’,” IISS StrategicDossier, International Institute for Strategic Studies (IISS) 2011. [Online; accessed 28 April 2014].
Jackson, Matthew O. and Massimo Morelli , “The Reasons for Wars - an Updated Survey,”in Chris Coyne and Rachel Mathers, eds.,
Handbook on the Political Economy of War , ElgarPublishing, 2011.
Kline, Harvey F. , Chronicle of a Failure Foretold: The Peace Process of Colombian PresidentAndr´es Pastrana , The University of Alabama Press, 2007.
Kuperman, Alan J. , “The Moral Hazard of Humanitarian Intervention: Lessons from theBalkans,”
International Studies Quarterly , 2008, (1), 49–80. Kydd, Andrew H. and Scott Straus , “The Road to Hell? Third-Party Intervention to PreventAtrocities,”
American Journal of Political Science , 2013, (3), 673–684. Mason, T. David, Joseph P Weingarten, and Patrick J. Fett , “Win, Lose, or Draw: Pre-dicting the Outcome of Civil Wars,”
Political Research Quarterly , 1999, (2), 239–268. Maynard, Jonathan Leader , “Ideology and armed conflict,”
Journal of Peace Research , 2019, (5), 635–649. Ministerio de Defensa , “Logros de la Politica de Consolidaci´on de la Seguridad Democr´atica,”July 2009.
Morelli, Massimo and Costantino Pischedda , “Oil, Federalism, and Third-Party Intervention:An Assessment of Conflict Risk in Iraqi Kurdistan,” Technical Report 2013.
M¨unster, Johannes , “Simultaneous inter- and intra-group conflicts,”
Economic Theory , 2007, (2), 333–352. M¨unster, Johannes and Klaas Staal , “War with Outsiders Makes Peace Inside,”
Conflict Man-agement and Peace Science , 2011, (2), 91–110. Nasi, Carlo , “Colombia’s Peace Processes, 1982-2002. Conditions, Strategies, and Outcomes,” inVirginia Marie Bouvier, ed.,
Colombia: Building Peace in a Time of War , United States Instituteof Peace Press, 2009.
OAS , “OAS Calls for Dialogue and Cooperation in Bilateral Relations between Colombia andVenezuela,” July 22 2010. Online; consulted 1 May 2014.
Owen, John M. , The Clash of Ideas in World Politics: Transnational Networks, States, and RegimeChange, 1510-2010 , Princeton University Press, 2010.
Presidencia , “Presidencia de Colombia: Secretar´ıa de Prensa,” July 16 2010. “Presidencia de Colombia: Alocuci´on del Presidente de la Rep´ublica, Juan Manuel Santos sobreel ‘Acuerdo General para la Terminaci´on del Conflicto’,” September 2012., “Presidencia de Colombia: Ni un solo dia ni un solo minuto vamos a bajar la guardia ante losactores violentos,” October 2012. Revista Semana , “El juego de Ch´avez,” January 19 2008. Online; consulted 24 May 2020., “Por qu´e Santos,” June 12 2010. Online; consulted 1 May 2014., “Qu´e se sabe del proceso de paz,” September 3 2012. Online; consulted 1 May 2014.
Rico, Daniel , “En qu´e estado militar est´an las FARC?,” April 2013. Online; posted 21 April 2013,consulted 1 April 2014.
Sanchez, Gonzalo , “Problems of Violence, Prospects for Peace,” in Charles W. Bergquist, Gon-zalo Sanchez, and Ricardo Penaranda, eds.,
Violence in Colombia, 1990-2000: Waging War andNegotiating Peace , Wilmington, Del.: SR Books, 2001.
Serbin, Andres and Andrei Serbin , “The Foreign Policy of the Bolivarian Republic of Venezuela:The Role and Legacy of Hugo Chvez,”
Latin American Policy , 2017, (2), 232–248. South, Ashley , Ethnic Politics in Burma: States of Conflict , Routledge, 2008., “The Politics of Protection in Burma: beyond the humanitarian mainstream,”
Critical AsianStudies , 2012, (2), 175–204. Stein, William E. and Amnon Rapoport , “Asymmetric Two-Stage Group Rent-Seeking: Com-parison of Two Contest Structures,”
Public Choice , 2004, (1), 151–167.
The New York Times , “Crisis at Colombia Border Spills Into Diplomatic Realm,” March 4 2008.Online; consulted 1 May 2014., “Colombia Signs Peace Agreement With FARC After 5 Decades of War,” September 26 2016.Online; consulted 1 Jun 2018., “Colombia’s President, Juan Manuel Santos, Is Awarded Nobel Peace Prize,” October 7 2016.Online; consulted 1 Jun 2018.
Werner, Suzanne , “Deterring Intervention: The Stakes of War and Third-Party Involvement,”
American Journal of Political Science , 2000, (4), pp. 720–732. Zartman, I William , “Ripeness: The Hurting Stalemate and Beyond,” in Paul Stern and DanielDruckman, eds.,
International Conflict Resolution After the Cold War , National Academy Press,2000., National Academy Press,2000.