Why Do Peace Negotiations Succeed or Fail? Legal Commitment, Transparency, and Inclusion during Peace Negotiations in Colombia (2012–2016) and Turkey (2012–2015)

Why do peace negotiations succeed or fail? Identifying the structural and contextual factors that pave the way for lasting peace is a central concern for scholars of intrastate wars. The realization that even successfully concluded peace agreements do not always prevent the renewal of violence has led scholars to focus on the relationship between peace agreements and the reemergence of violence, often to the exclusion of the negotiation process itself as an independent causal factor. This article explains the ways in which the structure of peace negotiations shapes the likelihood of concluding a peace agreement between two adversaries, comparing the recent peace negotiations to end intrastate wars in Colombia and Turkey. It examines the ways in which negotiation frameworks contribute to or hinder the conclusion of peace agreements.

The negotiation framework refers to the legal guarantees, information management mechanism, and scope of inclusivity in peace negotiations. The government, which usually is responsible for establishing the framework to negotiate an end to intrastate war, takes into account the perceived costs and benefits of promoting peace negotiations. The government also faces its own and the adversary's information asymmetry and commitment problems in deciding whether peace negotiations should have a codified legal basis, how much and through what channels the public should receive information about the negotiations, and which domestic and international actors should participate. Codifying the peace negotiations in law, publicizing information about the negotiations, and including mediators and civil society actors in the process will likely increase a government's short‐term costs because these steps signal an irreversible investment in conducting talks with the adversary, publicize negotiation setbacks and failures, increase the likelihood of disagreement between a large number of stakeholders, and finally, embolden spoilers who capitalize on disagreement, setback, and failure. Either strategy—one that is legalized, public, and inclusive or one that is not—may be in a government's self‐interest given political opportunity structures and constraints if the goal is to minimize dissent and maximize electoral success. However, concluding a peace agreement is more likely in the context of a legalized, public, and inclusive framework that alleviates the adversaries' information asymmetry and commitment problems, sets guidelines to protect the process from exogenous shocks (crises in other areas of political life and sporadic acts of violence), and increases the number of actors taking part in conflict management during the negotiations.

The peace negotiations in Colombia that occurred from 2012 to 2016 and those in Turkey that took place from 2012 to 2015 present dramatically different trajectories despite similar initial conditions. Dyads of negotiators in both countries—in Colombia, Juan Manuel Santos's government and the leadership of the Revolutionary Armed Forces of Colombia (FARC in Spanish acronym), and in Turkey, Recep Tayyip Erdoğan's government and the leadership of the Kurdistan Workers' Party (PKK in Kurdish acronym)¹—acknowledged the presence of a “mutually hurting stalemate” (Zartman 1989) in which state security forces weakened, but could not extinguish, insurgent armies in the context of a prolonged internal armed conflict. Both countries had already witnessed two sets of failed peace negotiations, in the 1980s and 1998–2002 in Colombia, and in 1993 and 2009–2011 in Turkey.² In addition, neither Santos nor Erdoğan had expressed interest in peace in their early political careers, or had a strong electoral mandate for negotiations in 2012.

Despite these initial similarities, however, the negotiation frameworks could not have been more different. The Colombian government legislated guidelines to identify the negotiating parties, the issues at stake, and guarantees available to the insurgents; publicized all partial agreements as well as some of the disagreements throughout the process; and included a large number of third‐party mediators and enabled nonbinding civil society participation. In contrast, the Turkish government legislated only immunity from prosecution for government negotiators, shared none of the negotiations with the public, and excluded all potential third parties. Despite similar initial conditions, and the ubiquity of spoilers (individuals or groups interested in sabotaging the peace process) and exogenous shocks in both countries, the differences in the negotiation frameworks resulted in divergent outcomes. In Colombia, a final agreement was concluded in September 2016; in Turkey, renewed violence erupted after July 2015.

This article is organized as follows. Following the introduction, the second section provides an overview of the scholarship on the success and failure of peace processes to explain why the negotiation framework matters. It defines the negotiation framework, identifies its components, and assesses the costs and benefits to a government of its choices regarding such components. The third section justifies the selection of Colombia and Turkey as case studies and discusses the data collection and analysis methods. The fourth section compares and contrasts the peace negotiations in Colombia and Turkey, documenting the ways in which different negotiation frameworks led to divergent processes and outcomes. The following two sections examine the ways in which the negotiation frameworks alleviate or aggravate the information asymmetry, commitment, spoiler, and exogenous shock problems, and assess alternative hypotheses for why peace negotiations fail. The conclusion highlights the policy implications of the findings and suggests areas of future research.

Influential rational choice models explain large‐scale violence with market failure logic. Rational actors are expected to achieve nonviolent outcomes that maximize the allocation of resources like financial assets and preferred policies for themselves under conditions of perfect information, as the stronger player (typically a state) can avoid a costly war by transferring some of its economic and/or political resources to a poorer and/or more aggressive state or nonstate actor.³ Information asymmetry (i.e., insufficient information about the adversary's ability and willingness to fight) is likely to drive states and nonstate actors to engage in violence, which serves as a mechanism to signal intentions and demonstrate comparative strength (Brito and Intriligator 1985; Fearon 1995). The commitment problem—i.e., when warring parties do not commit to peace because they perceive the balance of military power to be more favorable in the present than in the future, or because they fear an attack from their adversary—may be another source of aggression (Fearon 1995; Leventoğlu and Slantchev 2007). While much scholarship takes the cost of war as a given to simplify theory building, uncertainty about the cost of war may be an additional factor that explains not only the onset but also the continuation of war (Powell 2004).

According to this theoretical framework, intrastate conflicts may be more likely to lead to violence than interstate ones. Information about nonstate armed groups is more difficult to collect than information about governments and the rebels themselves might not know their own strength (Bapat 2005). In addition, the power asymmetry between the government and the rebels makes post‐conflict commitments less credible (Walter 1997), and the resources over which the parties are bargaining are often indivisible (Walter 2014).

Scholarship on peace processes closely follows scholarship on wars. Information asymmetry and commitment problems overshadow peaceful resolution efforts in interstate (Fortna 2004) and intrastate wars (Joshi and Mason 2011). Uncertainty about an adversary's willingness and ability to commit to peace may derail peace efforts. Aware of this possibility, spoilers test the resilience of the commitment to peace by engaging in violent acts or organizing against peace proposals (Stedman 2000; Kydd and Walter 2002). The commitment problem, which may fuel belligerent behavior if a change in the balance of power is anticipated, can invalidate future promises of peace. Thus, even when the warring parties prefer peace to war, they may prolong a war in the hope of attaining a decisive military victory or receiving additional information on the adversary (Filson and Werner 2002).

Variation in the success of peace processes is explained as a result of structural conditions or contextual factors, or as a function of peace agreements themselves. Structural explanations emphasize belligerents' preference for peace over violent solutions (Villaveces‐Niño 2003); accordingly, low expectations of immediate victory and/or the high cost of violence make peace more likely.⁴ Insurgent organizations are offered a seat at the table and are more likely to receive concessions if the government fails to secure early decisive victory (Bapat 2005; Cunningham, Gleditsch, and Salehyan 2009). The realization that peace agreements do not always result in lasting peace has led to a reappraisal of agreements themselves; their capacity to solve information asymmetry, commitment, and indivisibility problems, enable power sharing, and address the structural causes of violence has been scrutinized (Fortna 2004; Mattes and Savun 2009; Joshi and Darby 2013; Joshi and Quinn 2015; Westendorf 2015). A government's (and the international community's) ability to include former enemies in the post‐conflict political settlement is found to be an important determinant of whether a peace agreement will be durable (Call 2012).

What is missing from this picture is the negotiation framework itself, which is distinct from, yet intertwined with, the agreement and implementation phases of a peace process (Darby 2001; Findley 2013). Acknowledging that not all peace agreements contribute to lasting peace, it is nonetheless important to ask why some negotiation processes conclude in agreement, while others fail altogether. Stable talks that do not fall victim to spoilers or exogenous shocks are a subset of all peace talks (Heger and Jung 2017). On‐and‐off negotiations between the Cypriot government and Turkish Cypriots since 1974 (Michael 2007), the Indonesian government and the Free Aceh Movement from 2000 to 2003 (Aspinall et al. 2003), the Philippines government and the National Democratic Front (Bell and Farahnoosh 2015), the Angolan government and the National Union for the Total Independence of Angola in 1992, the Sudanese government and various rebel factions in the early 1990s, Mobutu Sese Seko and Laurent Kabila in the Democratic Republic of the Congo in May 1997 (Joshi and Mason 2011), the Israeli government and the Palestinian Liberation Organization in 2000, the Sri Lankan government and the Tamil Tigers in 2001 and 2003, the Nepalese government and the Maoists from 2003 to 2006 (Sapkota 2017), and more recently the Syrian government and the armed opposition in 2014 (Camp 2014) are cases in point. Peace negotiations may have been predestined to fail in some of these cases due to unfavorable structural and contextual factors or exogenous shocks, but this hypothesis should be assessed rather than assumed, keeping in mind that the negotiation framework may explain the variation in the success of peace negotiations.

As noted above, the negotiation framework encompasses the legal and policy decisions made in the prenegotiation and negotiation phases about legal rules, information sharing, and inclusion of stakeholders. The negotiation framework also influences how relevant parties come to the negotiation table, how negotiators are chosen, and how a framework for resolving substantive issues is shaped (Hartzell and Hoddie 2003; Lilja 2011). It determines who gets to sit at the negotiation table, the rights and obligations of the negotiators, how negotiations are to be carried out, and how much the public should know about the content of the negotiations. Thus, the negotiation framework has serious implications for managing disagreements and discontent in the negotiation process, and for building support for the peace process.

This article does not claim that the negotiation framework by itself can reverse unfavorable structural and contextual conditions to make peace agreements possible. It does argue, however, that variation in the success and failure of negotiations that take place within broadly similar structural and contextual parameters may be due to the negotiation framework. The absence of a mutually hurting stalemate and accompanying expectations of quick military victory, the willingness to endure the cost of violence, or negotiating in bad faith to placate the international community (Thrall 2017) may of course hamper peace efforts, but it is safe to assume that not all negotiation processes take place against impossible odds or in bad faith. Furthermore, the mere existence of negotiations signals that both sides perceive the cost of continuing violence to be high. Thus, the existence of a real or perceived stalemate is often a constant, rather than variable, factor during peace negotiations. Information asymmetries and commitment problems need to be addressed during negotiations, but the ubiquity of these problems means that by themselves they do not explain the differences between successful and failed negotiations. Rather, some negotiations may alleviate these problems more successfully than others. Finally, exogenous shocks, like cross‐over effects of other violent conflicts, political instability, and economic crisis, as well as spoiler activity, may derail peace negotiations by increasing the price of peace or reducing the expected benefits from peace, but the success with which they can be controlled varies across peace processes.

A number of design features in the negotiation framework can help to overcome or at least alleviate information asymmetry, commitment, and exogenous shock problems. Nonstate armed actors need security guarantees to conclude peace negotiations successfully (Walter 1997). Strong political and legal institutions can ensure peaceful settlement of disputes by alleviating the commitment problem (Walter 2014); conversely, weak institutions are likely to obstruct peace processes because they fail to alleviate the nonstate actors' commitment problem (DeRouen, Lea, and Wallensteen 2009). An independent and strong judiciary may build nonstate actors' trust in the peace process and ensure their safety and security in an eventual settlement (Villaveces‐Niño 2003). While secrecy is common to peace processes (Storholt 2001; Sapkota 2017), it is increasingly acknowledged that transparency between the negotiating parties (Albin and Druckman 2012) and sharing at least some of the content of peace negotiations with the broader public can strengthen the process. Third‐party intervention by individuals, states, regional organizations, and international organizations (Bercovitch and Gartner 2006) may affect the underlying bargaining structure and ultimately allow the negotiating parties to communicate more effectively and overcome commitment problems (Pugh 2009; Ruggeri, Gizelis, and Dorussen 2013; Joshi and Quinn 2015). Nonstate actors, especially, are likely to demand the presence of an outside party in the process. More intrusive forms of intervention may impose sanctions on the parties to keep them accountable to the peace process (Walter 2002; Terris and Maoz 2005; Walter 2014). Civil society groups may contribute to lasting peace by focusing on a violent conflict's underlying dynamics and by facilitating dialogue between adversaries (Barnes 2006), and are found to contribute to the durability of peace agreements (Kew and Wanis‐St. John 2008; Nilsson 2012). Inclusionary peace processes are hailed for their capacity to absorb the negative reactions from stakeholders and build a culture of peace (Bell 2017). Finally, civil society groups help to legitimize peace negotiations (Zanker 2014). However, it is important to note that civil society participation should not be seen in binary terms; the degree to which civil society actors can participate in negotiations through, for example, direct representation, observation, consultation, public decision‐making, and mass action, and the extent to which their demands and decisions have binding status, should also be taken into consideration (Paffenholz 2014).⁵

The idea of a broadly transparent and multi‐actor peace process in which both sides acknowledge legal rights and obligations may be appealing for the reasons listed above, but the framework‐setting government also considers the likely costs of such a process (Kaplow 2016). Agreeing to preconditions by codifying the legal framework for peace negotiations limits the maneuvering space for one or both parties (Kew and Wanis‐St. John 2008), as they may not concede more than what is legally allowed. Providing insurgents with legal immunity may provoke accusations of aiding and abetting criminals. Sharing information with the broader public increases the likelihood of disaffection among the population, publicizes the setbacks and failures of the process, and is likely to provide spoilers with a platform. Including more actors as stakeholders or opinion leaders increases the size of the constituency to satisfy. In a multi‐actor setting extremists can undermine moderates; spoilers can emerge more easily; and the inclusion of “zealots” in the process may rule out the possibility of reaching an agreement (Darby 2001; Findley 2013; see also Kydd and Walter 2002). Third‐party mediation may generate costs such as unforeseen concessions, less control over the outcome, and reputational costs (Bercovitch and Gartner 2006). Furthermore, the parties to the conflict can keep shifting alliances throughout the negotiation process. In addition, legal codification, publicity, and inclusivity all require the government to invest in the peace process; thus, a government that sets the negotiation framework knows that abandoning negotiations is costly (Heger and Jung 2017) and therefore may be interested in minimizing the costs associated with such a high degree of buy‐in. A government's short‐term interests, therefore, may be in conflict with long‐term planning for successful peace negotiations (Aspinall et al. 2003; for government calculations on the timing of concessions, see Rasler 2000).

A government that considers the costs and benefits of peace negotiations, therefore, may choose a legalized, transparent, and inclusive negotiation framework; a nonlegalized, opaque, and exclusionary one; or a framework with a combination of these features that complicates this binary. This article's analytical method treats the negotiation framework as a critical juncture and looks at cases in which the negotiation frameworks are extremely different—but the cases are otherwise similar—to demonstrate the framework's effect. I argue that multiple negotiation frameworks may maximize a government's short‐term goals, but a legalized, transparent, and inclusive framework is more likely to promote the conclusion of a peace agreement.

Both Colombia and Turkey initiated peace processes in mid‐ to late 2012 at the conclusion of secret exploratory talks with the representatives of rebels. Prior to these talks both countries had witnessed failed negotiations followed by increasing levels of violence: Colombia's peace talks with FARC in the second half of the 1980s and from 1998 to 2002 (“El fracaso de los diálogos de paz” 2010), and Turkey's talks with PKK in 1993 and from 2009 to 2011 (Yarali 2017). In both countries state security forces proved superior to the insurgent armies but failed to extinguish them. The immediate run‐up to the peace negotiations was marked in both countries by heavy fighting in which governments and insurgents tried to showcase their military strength. On November 4, 2011, the Colombian armed forces killed Alfonso Cano, one of the top commanders of FARC. Heavy fighting continued into early 2012. Likewise, the period between 2011 and 2012 marked the most significant escalation of violence since the mid‐1990s in Turkey (for details, see  Appendix: Timelines of Peace Negotiations in Colombia and Turkey). Thus, a mutually hurting stalemate existed in 2012.

The two cases were also similar in terms of spoiler activity, as critics of the peace process existed inside and outside the government but could not derail the talks. Both governments refused to declare a cease‐fire, thereby maintaining the threat of violence as a bargaining tool. Though reduced as a result of the negotiations, violent exchanges continued in Colombia throughout the peace talks.⁶ Likewise, Turkey reduced its use of violence dramatically; in 2014 the military asked permission to carry out 290 strikes against PKK, of which only 8 were approved by the government (Goksel and Mandiraci 2016). However, no cease‐fire was declared. Because the negotiations in both countries started around the same time, the global know‐how on conflict resolution techniques available to the negotiators was identical. Finally, in both countries the political leadership was known to have recently implemented counterinsurgency policies; therefore, the frameworks for peace negotiations were set up by leaders who were not perceived as “doves.” Both Juan Manuel Santos and Recep Tayyip Erdoğan drew part of their political capital from earlier commitments to fighting the rebels.

Yet, despite these broad similarities, the Colombian peace negotiations ended with a bilateral agreement in September 2016 while Turkey's peace process came to an unsuccessful end in July 2015. Thus, the cases represent an illustration of the most‐similar systems design: similar cases exhibit different outcomes as a result of divergence in an explanatory factor, which here is the negotiation framework. Two other differences need to be acknowledged. First, the spillover effects of the civil war in Syria strained Turkey's peace negotiations, especially after the PKK‐affiliated Democratic Union Party (PYD in Kurdish; also named after its militia, the People's Protection Units, YPG) asserted control over northern Syria. Second, the erosion of democracy and the rule of law in Turkey generated uncertainty over the prospects of the peace process. In what follows I show that these two factors, which developed exogenously to the peace negotiations, produced such deleterious effects on the negotiations only because the negotiation framework in Turkey lacked the robustness to absorb such shocks.

These case examinations are based on data collected primarily from press archives. Factual information about each peace process is complemented with speeches and press interviews by leading politicians, representatives of insurgents, and other social and political actors with a stake in the negotiations. The collected information identifies major decision points around the negotiation framework, the content of decisions, relevant actors' positive and negative signals to their adversaries, and key developments explaining the success or failure of the negotiations. In order to ensure diversity of opinion and coverage, data in both countries was collected from ideologically divergent press sources. In Colombia, the sources are Semana, known for its investigative journalism and considered liberal; the left‐leaning El Espectador; El Tiempo, owned by the Santos family and considered centrist; and La Silla Vacía, chosen for its consistent coverage of the negotiations. In Turkey, the sources are Radikal, considered mainstream; Bianet and T24, left‐leaning online sources with long histories of covering the Kurdish issue; and Yeni Şafak, which follows the official government line. International sources such as the BBC and Al Jazeera are also used to compile the timelines.

Process tracing is used to analyze the within‐case data to highlight similar initial conditions and divergent outcomes across cases. The issue at hand, that is, how the negotiation framework affects the success of peace negotiations, requires the identification of initial conditions, key decisions, divergent paths following the decisions, and divergent outcomes (Bennett and Checkel 2014). Furthermore, the analysis uses process tracing to adjudicate between competing hypotheses by validating the observable implications of those hypotheses (Bateson 2017). Since the article examines the influence of the negotiation framework on commitment, information asymmetry, and exogenous shock problems, negotiating actors' signals of satisfaction and dissatisfaction are documented in detail.

The Colombian peace negotiations started with a general accord that mapped the content of the proposed talks. The road map, called the Juridical Framework for Peace (Marco Jurídico para la Paz), was ratified as a constitutional amendment in June 2012 and provided a transitional justice model to enable negotiations with FARC (“Y la paz fue Ley” 2012). Turkey's peace process did not have a road map or legal framework (Hakyemez 2017). Despite calls from PKK leaders for legal guarantees, the first one‐and‐a‐half years of Turkey's peace negotiations were not covered by any legislation. Government spokesperson Bülent Arınç declared in April 2013 that the peace process should have no legislative basis, as it was under the executive's responsibility. The Turkish parliament amended a law so as to extend legal immunity to National Intelligence Agency negotiators and in July 2014 passed what was supposed to be a framework law, entitled the Draft Law on the Termination of Terror and Strengthening of Social Integration (Terörün Sona Erdirilmesi ve Toplumsal Bütünleşmenin Güçlendirilmesine Dair Kanun, Resmi Gazete/Official Gazette, Law no. 6551, July 16, 2014). The law did not contain information about the content or proposed timeline of the negotiation process, provided no legal guarantees for the parties, and did not stipulate any obligations on either side. Thus, it did not set a framework for the peace talks.⁷

The Colombian process was remarkably transparent. Initially it was not the government's intention to publicize peace negotiations, but after former President Álvaro Uribe leaked the secret negotiations between the Santos government and FARC, Santos had no other option but to announce the peace talks. On August 26, 2012, a general accord was signed by the government and FARC representatives,⁸ as the two sides agreed to negotiate the following issues: ending the conflict, land distribution, political participation, drug production and trafficking, attention to victims, and the implementation of an eventual peace agreement (“Esto es lo que negociaron” 2016). Santos confirmed the existence of peace negotiations the day after the accord was signed. The government and FARC drew up lists of negotiators, and by October 2012 the peace talks were underway. The Colombian government allowed FARC negotiators to travel to Oslo and Havana to carry out negotiations (Gómez Giraldo 2014). Thus, the stakeholders did not face major communication problems. The public was informed through frequent updates about the state of negotiations. Every agreed‐upon item was publicized, and by corollary, the public knew about unresolved issues, as well. While the specific content of the negotiations was not publicized, every breakthrough in the process was communicated to the public. Partial agreements were made public in September 2014 and August 2015. Disagreements and setbacks were discussed in Congress, the media, and other public arenas.⁹

As in Colombia, the peace process in Turkey started with secret negotiations that were later leaked. However, in Turkey the announcement of negotiations did not improve the transparency of the process, as it had in Colombia. The Turkish government had initiated secret talks with PKK leadership through Hakan Fidan, undersecretary of the National Intelligence Organization, in early 2012.¹⁰ This initiative was leaked to the press and the government suspended the talks, fearing a public backlash for making concessions to so‐called terrorists.¹¹ Another round of talks was announced in late 2012 but the government merely publicized their existence; the announcements did not provide information about the content of the negotiations. The Turkish government limited the information that circulated between the stakeholders as well as the information available to the public. Members of the Turkish National Intelligence Agency talked directly to Abdullah Öcalan, the imprisoned leader of PKK; the PKK leaders in northern Iraq had no way to communicate directly with him. The government only infrequently allowed members of parliament who belonged to the Kurdish political movement¹² to communicate with Öcalan; what is more, the government retained the right to disallow visits to Öcalan, thereby threatening to cut off communication between Öcalan, PKK's commanders in northern Iraq, and the Kurdish political movement in general. This aggravated the insurgents' information asymmetry and commitment problems. The government allowed no visitors to Öcalan between April and July 2015, the critical period that sealed the fate of the peace talks. In sharp contrast to the Colombian peace process, where every item discussed in the negotiations was publicly known and any progress or lack thereof was reported regularly, in Turkey the content of the talks remained unknown to the public.

Another contrast is the relatively large number of third‐party participants and alliance‐building efforts in the negotiations between the Colombian government and FARC, compared to the exclusionary nature of the talks in Turkey. The government of Colombia and FARC agreed to bring a number of foreign governments into the peace talks in order to diversify support for the negotiations and avoid unilateral defection by either party. The governments of Venezuela and Chile were observers during the talks, while those of Cuba and Norway were hosts and guarantors (Maldonado 2016). The United Nations was present through its various bodies to provide relief for victims and to shape the peace agenda (Rincón, Bautista, and Pugh 2019).

Furthermore, the civic sector's participation was one of the hard‐won achievements of nongovernmental actors in Colombia (Herbolzheimer 2019). The lesson learned from past failed negotiations, especially those between 1998 and 2002, was that issues of interest to citizens (such as attention to victims' demands) would go unaddressed if representatives of the civic sector were not allowed to communicate with negotiators. In the face of government reluctance to open the talks to civil society actors, civic groups pushed for participation. Over one million citizens marched for peace on April 9, 2013, not only to show their support for the initiative but also to make their presence felt as stakeholders. Civil society representatives were not negotiators, but their platforms had to be taken seriously in the negotiation process. For example, advocacy by victims' groups, which were allowed to visit Havana to communicate their demands to the negotiators, appears to be key to the final agreement's incorporation of criminal accountability for human rights violations (Arenas 2014). Representatives of the Colombian military—important stakeholders—were kept informed about the negotiations.

This could not be more different from the situation in Turkey, where the government and Öcalan sought to limit the number of negotiators and third‐party mediators. The government established that Prime Minister Erdoğan would have the last word in the peace talks and the process would be managed by his trusted aides. The insurgency exhibited more diversity, not least because PKK's active organizational leadership resided in northern Iraq while Öcalan, the government's preferred negotiator, was in prison. The PKK leadership accepted Öcalan's status as the undisputed representative of the insurgency¹³ but frequently warned the government against disregarding their agency. Despite repeated calls from PKK (“Cemil Bayık” 2014), no foreign government or international organization was allowed to take part as an official negotiator, facilitator, or guarantor of the process.

Furthermore, Turkey's peace negotiations were marked by the remarkable absence of civil society representation. A number of nongovernmental actors endorsed the process in their individual or organizational capacity, and there was enormous activism by women's groups, human rights organizations, and other peace advocates that sought to make positive contributions to the talks, but the setup of the peace process excluded them.¹⁴ A civic initiative called the Wise People's Committee (Akil İnsanlar Heyeti), which brought together public intellectuals and celebrities in an effort to improve the public relations aspect of the peace process, was entirely sidelined by the government after leftist and liberal members of the Committee criticized the government for its conduct during the 2013 Gezi protests. Other venues for articulating the public's demands were absent (Rumelili and Çelik 2017).

Thus, the Colombian process allowed for consultation with, and direct representation by, civil society groups, including human rights and victims' organizations. In contrast, at the start of the Turkish peace process the government allowed only for elite‐level consultation and disregarded even this limited public involvement soon after the peace talks began. Communication with the military followed a similar pattern. The Colombian military was kept informed of the negotiations; in Turkey, the Chief of the General Staff complained that he only learned about developments in the peace talks from the media (“Özel'den hükümete ‘çözüm' sitemi” 2014).

In summary, Colombia's peace negotiations took place under legal guarantees, and were relatively transparent and inclusionary, while peace negotiations in Turkey had insufficient legal basis, and were opaque and exclusionary. The Colombian government's choice of negotiation framework reflected its interest in alleviating information asymmetry and spoiler problems and in encouraging FARC to make credible commitments to peace. The Turkish government ignored calls for legal guarantees, information sharing, and inclusion, in the belief that agreement between Prime Minister Erdoğan and jailed PKK leader Öcalan would be sufficient to ensure peace.

Colombia and Turkey both faced similar impediments to successful negotiations. The governments and insurgents in both countries had insufficient information about the adversary's willingness and ability to choose peace over war and all sides suspected that peace negotiations allowed the adversary breathing space to reinitiate hostilities. Insurgents in both countries could not trust government calls to disarm and demobilize. In addition, public criticism of negotiations and violent attacks by spoilers were anticipated; and economic downturns, corruption scandals, violence across borders, and a host of other political exigencies not caused by the peace talks threatened the negotiations. The peace process was on the brink of collapse in Colombia in June 2013, November–December 2014, and June–July 2015; the same was true in Turkey in October–November 2014, and tensions in July 2015 brought the process there to an end. The negotiation frameworks were key factors in shaping the extent to which these challenges could be overcome or alleviated in each country.

Both dyads faced the risk of renewed violence due to information asymmetries. Governments and insurgents accused one another of unwillingness to commit to peace and of taking advantage of the cessation of hostilities to prepare for war. Both governments' refusal to declare a cease‐fire and both insurgencies' refusal to disarm in the absence of a peace agreement aggravated mutual uncertainty about adversaries' willingness and ability to choose peace over war. In the case of Colombia, the international observance of both sides' commitments and the codification of the commitments in law increased the cost of renewed fighting dramatically. Sharing information with the public gave both sides a chance to assess the costs and benefits of the negotiation process as measured by public opinion. Thus, periods of escalating tension did not result in all‐out war.

In contrast, the Turkish government intentionally demonstrated its indifference to peace or war; its spokesperson announced that while the peace process was important, they were not obliged to continue it (“Arınç: Çözüm sürecine mecbur ve mahkûm değiliz” 2014). There were no efforts at building trust between the parties, as shown by the exclusion of the PKK leadership in northern Iraq. Thus, both sides emerged from the negotiation convinced that their adversary was more invested in fighting than in seeking peace. In April 2013, in the early phase of the peace negotiations, the Turkish government began construction of high‐security military headquarters, provoking protests and lethal police violence in June (“Medeni Yıldırım binlerce kişiyle toprağa verildi” 2013). Time and again PKK leaders referred to the construction as a declaration of war. Likewise, when PKK leaders suspended the cease‐fire (“KCK: Çekilmekle saflık ettik” 2015) and a PKK‐affiliated youth militia unilaterally declared autonomy in July 2015, the Turkish government used such declaration to justify reinitiating hostilities.

The Colombian government acknowledged that the rebels would not disarm and demobilize before a peace agreement was signed. Having suffered massacres of its unarmed sympathizers in the late 1980s and 1990s,¹⁵ FARC was adamant that it would commit to peace only if security guarantees were in place. In fact, the inclusion of the implementation phase as a negotiation item was meant to ensure both sides' confidence in the enforceability of a peace agreement. The participation of third‐party actors such as U.N. inspectors to monitor the demobilization process enhanced the government's and FARC's trust in implementation.

In Turkey, by contrast, the government asked PKK to lay down arms as a precondition of further negotiations, which the insurgents refused to do. Öcalan's call for PKK to “end armed struggle”—in a letter publicized on March 21, 2013—was interpreted by the PKK leadership in northern Iraq as a call to withdraw forces from Turkey, not for full demobilization. In the words of Cemil Bayık, a PKK leader, “The guerrilla cannot take one step back without legal guarantees” (“PKK'lı Bayık” 2013). Withdrawal started on May 8, 2013 but the PKK leadership brought it to a halt four months later, accusing the government of failing to take steps to deepen the peace process.¹⁶ The government's refusal to engage third‐party actors further aggravated the commitment problem. The Kurdish political movement's repeated calls to establish a monitoring council (izleme kurulu) were likewise ignored (“Baluken: Sürecin gidişatı 3 adıma bağlı” 2014). From the early days of the peace process all the way to its collapse, both sides continued to accuse each other of breaking promises (Çiçek and Coşkun 2016).

Both peace negotiations faced influential spoilers interested in derailing the process, as well as a number of unrelated or indirectly related events that jeopardized the negotiations. In Colombia, former President Uribe managed to build a coalition of the rural landowning and cattle‐ranching elites, some sections of the military, and a number of legislators to oppose the peace process (Bouvier 2013). One of his allies, Oscar Iván Zuluaga, ran for president in 2014 with the explicit objective of halting the peace talks and lost narrowly. The Uribe coalition even managed to halt the agreement's ratification when it succeeded in swaying public opinion in a referendum in October 2016, though this did not derail the bilateral agreement between Santos's government and FARC. The main negotiation provisions against which the Uribe coalition mobilized were those facilitating former FARC members' participation in politics and those setting forth a transitional justice model that stipulated immunity from prosecution for most rank‐and‐file FARC members and lighter sentences for FARC members who committed serious crimes.¹⁷ In addition to the open campaign against the peace process, a number of wiretapping scandals, reportedly directed by powerful military interests, tested the robustness of the peace process, but failed to undermine it.¹⁸ Despite all these setbacks, the Colombian government managed to reach an agreement with the insurgents, which was eventually ratified by the legislature (Casey 2016).

In Turkey the potential spoilers were identified as disgruntled members of the military, Turkish ultranationalist groups, the Gülenists, and dissatisfied PKK sympathizers. Armed spoilers were conspicuously absent for most of the period. Much of the opposition to the peace process took the form of ultranationalist protests on the street or during pro‐peace meetings (Oran 2014). Pro‐Gülen prosecutors have long been accused of sabotaging the preliminary secret talks in February 2012 by attempting to arrest Erdoğan's confidant, Hakan Fidan. Nonetheless, when the process unraveled in mid‐2015, neither side accused the Gülenists or ultranationalists of orchestrating the downfall of peace talks. They were at times considered spoilers but did not play a key role in undermining the peace process. In a way, the process in Turkey was more favorable for a peace agreement than the one in Colombia because Turkey's spoiler activity was much more restrained. Thus, the level of spoiler activity is not the determining factor in the outcome of the two countries' peace processes.

A number of exogenous political and military developments jeopardized both peace negotiations. Public opinion polls show that the Colombian government's approval ratings consistently lagged behind popular support for the peace process.¹⁹ In addition, the peace negotiations started at a time when the government faced diplomatic crises with neighboring Ecuador and Venezuela (Maldonado 2016) stemming from successive Colombian governments' suspicions that its neighbors hosted and supported FARC operatives. None of these exogenous factors, however, stopped the successful negotiation of a peace agreement. The government chose to double down, rather than backtrack, on its commitment to the process when its popularity was at stake because the negotiation framework had already forced the government to bear the cost of negotiating. Cross‐border risks were absorbed into the process. President Santos asked the Venezuelan government under Hugo Chávez, which FARC viewed in a sympathetic light, to act as a guarantor of the process, thus neutralizing hostility from Colombia's neighbor.

Turkey was engulfed in multiple crises between 2012 and 2015. The brutal crackdown during the 2013 Gezi protests signaled the regime's lack of commitment to peaceful conflict resolution; the fallout between the Gülen community (an erstwhile government ally) and Erdoğan unveiled the most serious corruption scandal in the nation's history; and finally, the government's response to the multiple political crises was to erode the foundations of constitutional checks and balances (Esen and Gumuscu 2016). Furthermore, the internationalization of the civil war in Syria increased the risk of cross‐border violence and provided PKK with the opportunity to extend, through its affiliate, into northern Syria. Facing these domestic and international crises, the government found its interest best served by backtracking on the peace process because the cost of abandoning peace talks was limited—the negotiation framework did not impose legal obligations, the public knew little about the status of the negotiations, and there was no third party to hold the actors in the process accountable.

It is possible to conjecture that a democratic government with a system of checks and balances, including an independent judiciary, is more likely to achieve a peace agreement because liberal democratic institutions solve the commitment problem and safeguard against exogenous shocks. Colombia remained a relatively stable democracy with functioning courts between 2012 and 2016,²⁰ whereas Turkey's democracy and the rule of law kept sliding in the same time period.²¹ Thus, it is possible to attribute differences in the outcome of peace talks to regime type, rather than to a country's negotiation framework.

Regime type is an important causal factor in explaining the difference in outcomes, but only through its mediated impact on the negotiation framework, and not independently. The Turkish government and PKK had agreed to negotiate in 2012, when the regime was already moving in an authoritarian direction, and throughout the process there was no indication that democracy and the rule of law would be strengthened in the foreseeable future. Thus, regime type did not influence the decision to negotiate or to stay in negotiations. Öcalan's public statements during the Gezi protests had a neutral tone, and he condemned efforts to prosecute Erdoğan and his family on corruption charges in late 2013 (“Öcalan” 2014). In fact, pro‐government pundits expressed optimism that the model of “strongmen” negotiations between Erdoğan and Öcalan would limit spoiler activity and thus facilitate a peace agreement.²² Therefore, the absence of democratic checks and balances was an inherent part of, rather than an exogenous obstacle to, the negotiation framework, which was based on lack of legal commitments, secrecy, and exclusion. In other words, the regime type influenced the type of negotiation framework each country adopted, thus exerting an indirect impact on the outcome, but this influence was only exerted through the intervening variable, that is, the negotiation framework.

The absence of a system of checks and balances aggravated the information asymmetry and commitment problems caused by the negotiation framework. In a milieu of mutual mistrust, the insurgents realized that they could not place their trust in courts or other accountability mechanisms. It is perhaps telling that the ruling Justice and Development Party government reinitiated violence soon after the Kurdish political movement's unexpected success in the June 2015 general election denied the party a majority in the parliament. The government treated the democratic process as a threat to its control over the negotiations, rather than an opportunity to overcome information asymmetry and commitment problems—a view consistent with its negotiation framework's overall philosophy.

On a related note, commentators critical of Erdoğan viewed his move to consolidate political power as the main reason for the failure of peace negotiations. A preliminary agreement between government representatives and the Kurdish political movement on February 28, 2015 (“Ortak açıklamanın tam metni” 2015) was ignored by the government itself when Erdoğan declared that he did not endorse it. The Kurdish political movement's objection to Erdoğan's efforts to entrench himself as a powerful president is said to have eliminated the common ground. While Erdoğan's strategic moves indeed appear to be the immediate cause of failure in 2015, the talks' reliance on the will of one person is proof of their fragility. Personal and organizational agendas were as prominent in Colombia as in Turkey—former President Uribe built a coalition opposing the peace negotiations, and parts of the military tried to discredit the process. However, the negotiation framework was sufficiently robust to withstand the criticisms and sabotage attempts.

It is also possible to conjecture that the civil war in Syria created such strong spillover effects that Turkey's peace process could not be maintained (Ozkahraman 2017), while there was no such conflict on Colombia's border. The Turkish government made efforts to put diplomatic pressure on YPG, a PKK affiliate, to reverse its territorial gains in northern Syria. The siege of the YPG‐held town of Kobanê by the Islamic State (IS) in the fall of 2014, and terrorist attacks by IS on peaceful Kurdish demonstrations in Turkey in June and July 2015, led PKK, YPG, and Turkey's Kurdish political movement to accuse the Turkish government of being indifferent toward, if not complicit with, IS violence, while the government argued that it indeed saw no difference between PKK and IS and was committed to eliminating both (“IŞİD ne ise PKK da odur” 2014). The fact that the peace process collapsed three days after an IS attack killed thirty‐three youth activists in Suruç, which the government used as its official excuse to attack IS and PKK simultaneously, shows the extent to which the dynamics of the Syrian civil war may have affected events inside Turkey. Likewise, PKK's decision to suspend peace talks in July 2015 may have been influenced by its expectation of region‐wide success after the fighting in Syria enhanced the organization's military position and international legitimacy.

However, the spillover from the Syrian civil war could assume such a determinant role inside Turkey only because the negotiation process lacked robustness. Statements by Turkish government and PKK spokespersons demonstrate the extent to which they mistrusted the adversary's motivations (“'Tezkere savaş ilanıdır'” 2014). The government portrayed gains by PKK and its affiliates in zero‐sum terms even while it was negotiating with PKK. What is more, many of the negative signals were evident before, and not after, the civil war in Syria opened a new front for the Turkish government and PKK affiliates. For example, in June 2013, PKK accused the government of building new military outposts in the Kurdish region, and in September 2013, mutual accusations were made when PKK stopped the withdrawal of its units inside Turkey (see  Appendix). The perception that the adversaries would go back to fighting eventually—a perception reflecting the depth of information asymmetry and commitment problems—could not be overcome through the negotiations.

Thus, a host of political and military developments may be conjectured to have set Colombia and Turkey apart. Perhaps a more accountable government facing fewer international crises is likely to bring peace, all else being equal. What this analysis shows is that regime type and international crises do not shape the outcome of negotiations independent of the negotiation framework itself. The impact of regime type is mediated by the negotiation framework, and international crises undermine peace negotiations when the framework is too fragile to contain them.

This article is an effort to rethink the roles of the prenegotiation and negotiation phases in the resolution of intrastate conflict. There is enormous variation in the actors, procedures, substantive discussions, and outcomes of peace negotiations. A government may choose to control the talks by making no legal commitments, maintaining a monopoly over information, and excluding third‐party mediators and civil society. Such an approach minimizes the costs associated with making concessions to the other side, tolerating spoiler activity, and satisfying diverse constituencies. However, failing to provide the nonstate armed actor with any guarantees and limiting the ownership of the peace process to a small number of actors creates a highly fragile framework that is vulnerable to external shocks and changing political opportunity structures. Conversely, a government that makes legally binding commitments, communicates key aspects of the negotiations, and incorporates third‐party mediation and civil society participation is likely to bear a high cost throughout the peace negotiations; yet, this approach is likely to make the negotiations robust to external shocks, spoilers, and changing political contexts. This is what explains the divergent outcomes of peace negotiations in Colombia and Turkey.

The key policy implication of this research is that a transparent and multi‐actor peace process with legal guarantees and obligations is more robust than one that is opaque and exclusionary. The Colombian government's approach confirms the validity of state‐of‐the‐art conflict resolution techniques that emphasize multi‐stakeholder participation and buy‐in from the public in the context of an intrastate war. However, another implication of the findings is that a government's perceived self‐interest is not necessarily aligned with peaceful conflict resolution. The risk‐averse strategy of making a minimal commitment to peace sets a peace process up for failure but is still likely to be adopted by governments that are interested in short‐term gains.

This article treats the selection of a negotiation framework as a critical juncture in the peace process. The negotiation framework itself may be shaped by factors such as regime type, a government's political ideology, and political and legal constraints faced by negotiating actors. The comparison of Colombia and Turkey suggests that none of these factors imposed inevitable choices, that is, the political leaders in both countries could have chosen alternative negotiation frameworks. Thus, the negotiation framework is an influential causal factor. Future research and practice should consider the factors that shape a government's decision‐making calculus at the onset of peace negotiations.

This analysis of the peace processes in Turkey and Colombia presents cases that are similar in many ways but with extremely divergent negotiation frameworks and, consequently, divergent outcomes. I have chosen these cases in order to showcase the causal effect of the negotiation framework. Arguably, Colombia and Turkey could not be more different in terms of legal guarantees, information management, and the inclusion of third‐party actors. Future research should allow for the possibility that the framework would take continuous, rather than binary, values in some contexts. The causal mechanisms highlighted in this article should be confirmed in light of other small‐n comparisons as well as large‐n analyses.

Civil society input deserves more detailed attention because it relies as much on an inclusionary institutional setup as on meaningful interactions between civil society actors and the parties to the conflict. This article contrasts a peace process with almost no civil society input, namely Turkey's, with another that includes civil society, at least minimally, in representation and consultation, namely Colombia's. Future research should delve deeper into the quality of civil society participation (Paffenholz 2014; Paffenholz 2015).

Finally, future research should assess the relationship between the negotiation framework and lasting peace. The case of Colombia suggests that implementing peace is an uphill battle even after an agreement is signed and ratified. The case of Turkey, however, delivers a simpler message: the negotiation framework should be at least minimally robust to give peace a chance.