Los nadies y las nadies: The Effect of Peacebuilding on Political Behavior in Colombia

Abstract

How do peacebuilding institutions affect political behavior? This article studies the historic victory of the Colombian left in the 2022 presidential elections in light of the implementation of local peacebuilding programs through the 2016 Peace Accords. Using a quasi-experimental design, we show that the Development Plans with a Territorial Focus (PDET), a central component of the 2016 Peace Accords between the government and the FARC, increased the vote share for the leftist coalition, Pacto Histórico, in the 2022 elections by increasing voter turnout in PDET regions. In a departure from existing literature, we find that the explanatory effect of violence on vote share is significantly reduced when we include an indicator for PDET implementation and additional covariates. While there is a substantial body of work examining the effects of conflict violence and the presence of armed actors on elections, there has been relatively little focus on how the peacebuilding has affected vote choice and political behavior. We see our project as a bridge to fill this gap in the literature.

Los nadies: los hijos de nadie, los dueños de nada. Los nadies: los ningunos, los ninguneados, corriendo la liebre, muriendo la vida, jodidos, rejodidos: que no son, aunque sean…

—Eduardo Galeano, 1989

On June 19, 2022, Gustavo Petro, a leftist politician and former guerrilla member, and Francia Márquez, an environmental activist and Black feminist social leader, won the Colombian presidential election. Their victory was historic and has received global attention due to Colombia’s long history of political conservatism, enduring political violence, and structural racism. Vice President Márquez claimed the electoral result as a victory for the nadies (nobodies) who have been ravaged by the conflict and decades of state neglect. The numbers speak for themselves. Turnout in the second round of the election reached 58 percent, a high-water mark for the century. Petro and Márquez were able to gain more than 2.7 million votes between the first and second rounds, and hundreds of thousands of these crucial second-round votes came from peripheral areas facing substantial risks of violence during the election (López 2022; Galindo 2022a).

This article studies the victory for the Colombian left in light of two different but complementary factors: political violence and the implementation of local peacebuilding programs through the 2016 Peace Accords. Several studies over the past decade have demonstrated that civil conflict violence has affected voting behavior. Armed actors have used violence to coerce votes, and this pattern has typically favored politicians and parties on the right (Acemoglu et al. 2013; Gutiérrez Sanín and Vargas Reina 2016; Bandiera 2021). Intense episodes of conflict violence in peripheral regions of the country have tipped some elections in favor of hawkish politicians—especially where voters are farther away from violence (Weintraub et al. 2015). In contrast, localities that experience more intense violence have tended to vote more strongly for peace (Branton et al. 2019; Dávalos et al. 2018).

Political violence remains stubbornly relevant in Colombia (Arjona 2016a; Gallego 2018). But the government has also made important strides toward peacebuilding. In late November 2016, the Colombian government and the Revolutionary Armed Forces of Colombia (FARC) signed a peace agreement to end the longest-running conflict in the Western Hemisphere. ¹ In the two presidential elections since the signing of the accords, we have seen a secular decrease in the centrality of the armed conflict (Gamboa Gutiérrez 2019). The existing literature has focused much attention on the relationship between violence and voting (Taylor 2009; Gillooly 2022). Yet we know relatively little about how peacebuilding fits into this story. How has peacebuilding affected political behavior in Colombia? In particular, what were the impacts of peacebuilding institutions on political participation and the Petro-Márquez vote in 2022? To address this important gap in the literature, we focus on one of the most encompassing peacebuilding institutions in the 2016 Peace Accords, the Development Plans with a Territorial Focus (Planes de Desarrollo con Enfoque Territorial, hereafter PDET) in the context of the 2022 national election.

PDET is a bottom-up policy designed to provide public goods and services. In the words of the peace deal, its objective is “to achieve the structural transformation of the countryside and the rural environment and to promote an equitable relationship between rural and urban areas” (Gobierno de Colombia and FARC-EP 2016, 22). The program design prioritized the municipalities with high levels of unsatisfied basic needs, presence of coca crops, historical conflict, and low administrative and management capabilities. There are 170 municipalities in the program, covering 36 percent of the Colombian territory and including 6.6 million people (Gobierno de Colombia 2022) (see figure 1).

Figure 1. PDET Map

Figure 2. Petro-Márquez Vote Share in the First Round

We argue that PDET municipalities voted for Petro and the Pacto Histórico at higher rates compared to non-PDET municipalities. ² Many PDET municipalities were electoral strongholds during Petro’s failed presidential bid in 2018. PDET created a set of new institutions at the local level that guarantee civic participation in local-level decisionmaking (Velásquez and Londoño 2022; Shenk 2023). These forums bring together actors that have been historically marginalized by the conflict, such as community organizations and former FARC members, and a series of state actors, like national government representatives, the Office of the Inspector General, the Office of the Attorney General, the Office of the Comptroller General, and the Office of the Ombudsman (Gobierno de Colombia and FARC-EP 2016). Building on existing scholarship that shows that local civic engagement provides the “mobilizational infrastructure” for voter turnout (Boulding and Holzner 2021; Holland and Palmer-Rubin 2015), we argue that PDET’s participatory design boosted turnout substantially in the municipalities that were already favorable to Petro and Márquez.

We use municipal data from multiple rounds of the 2022 Colombian presidential election, in addition to municipal data on conflict violence and PDET implementation. to test the marginal and interactive relationship between violence and peacebuilding and the vote share for Petro and Márquez. PDET communities were historically the most affected by the conflict and are the communities most at risk for continued violence. They were and continue to be the most economically marginalized and to suffer from the underprovision of public goods. In order to distinguish the effects of PDET on vote choice and turnout from the conditions that made these municipalities eligible for PDET in the first place, we simulate a quasi-experimental design through nearest-neighbor propensity score matching (PSM) to estimate the effect of PDET on municipal vote share for the Pacto Histórico. We use the propensity scores to simulate a differences-indifferences design to measure the effect of PDET on voter turnout. To our knowledge, our study is one of the first to attempt to causally identify the effect of the program on elections. ³

We find that in the 2022 presidential elections, higher levels of historical violence did not predict support for the Petro-Márquez coalition. This finding is consistent with several recent studies on the reduced salience of the conflict to Colombian electoral politics (Gamboa 2018; Botero et al. 2023). Instead, we identify a significant effect of PDET on vote share that coincides with a boost in turnout. Local PDET implementation increased Petro’s vote share by 1.2 to 2.3 percentage points, controlling for the previous election. Although we cannot make too strong a generalization from the analysis of one election and a complex program like PDET, our evidence suggests that PDET municipalities experienced a steeper increase in their rate of voter turnout relative to matched municipalities. We take this as evidence that PDET implementation addressed some of the structural limitations to voter turnout under the threat of violence in the 2022 election (Ley 2018; Albarracín et al. 2023). This finding makes an important contribution to the literature on postconflict elections in places where violence or the risk of violence persists (Albarracín et al. 2023). Our analysis also has implications for the literature on Colombian electoral politics, in the context of waning conflict salience (Gallego 2018; Gillooly 2022).

This article proceeds as follows. The next section presents the context of the 2022 presidential elections in broad strokes. Section 3 discusses the design, implementation, and challenges of the PDET program. Section 4 uses the existing literature on violence, voting, and the evolution of the conflict cleavage in Colombia to develop our expectations for the relationship between voting and violence in this changing context. Then we discuss PDET and its significance to political participation to spell out our core argument. The data employed and the empirical strategies utilized are described in section 6. The results of our regression analysis and quasiexperimental investigation are presented in section 7. Section 8 concludes with a discussion of the implications of our study for the literature on elections and peacebuilding in Colombia and in comparative perspective.

The 2022 Presidential Election

In the two decades prior to the 2022 presidential elections in Colombia, the counter-insurgency promulgated by President Alvaro Uribe and the hardline discourse he used to justify state and paramilitary violence, constituted a powerful force in the electoral arena (Gamboa Gutiérrez 2019). The 2022 elections marked a substantial departure from the status quo (Botero et al. 2023). The election was framed as a rejection of traditional political parties and Álvaro Uribe’s hardline approach to violence and insecurity (Long et al. 2022). Colombians also voiced their dissatisfaction with the outgoing Duque government during the Paro Nacional (National Strike) in 2021. Dissatisfaction continued to mount in response to the unequal economic burden of the government’s policies to prevent the spread of COVID-19 (ECLAC 2022).

Uribe’s party, Centro Democrático, lost congressional seats in the legislative elections and failed to field a candidate in the first round of the presidential elections. Moreover, Federico Gutiérrez, the candidate of a coalition of right-wing political parties, Equipo por Colombia, was expected to take second place (Semana 2022). But Gutiérrez came in third (23.94 percent), and a relatively unknown candidate, Rodolfo Hernández (28.17 percent), managed to qualify for the second round. Hernández emerged as a surprising favorite in the election after securing a spot in the runoff (Botero et al. 2023). He championed anticorruption efforts, which resonated with citizens weary of political gamesmanship. Some even compared him to former US president Donald Trump for his establishment-defying attitude and resilience against controversies (Turkewitz 2022).

The strong performance of Francia Márquez in the primary elections and her historic vice presidential run mobilized decades of voter frustration and used Colombian culture and art, along with specific appeals to Black and Indigenous identity, to make visible the voices and preferences of communities that had been silenced throughout Colombia’s history (Ramírez-Botero 2022). Gustavo Petro won a historic victory in the second-round runoff against Hernández, taking 50.4 percent of the vote to Hernández’s 47.3 percent. In his historic victory speech, Petro assured citizens that the demands of the National Strike, as well as the Havana Peace Agreement, would be implemented during his government (CNN 2022).

In line with this antiestablishment view of the 2022 elections, another important factor explaining the shift in the electorate was the strong turnout by members of marginalized Black, Indigenous, and peasant communities. The continued wave of violence on the Colombian periphery since the 2016 Peace Deal was expected to pose a real threat to turnout and electoral competition, especially in the PDET municipalities (Colombia+20 2022; Nygård et al. 2020). Voter turnout increased to 58 percent in the June second-round presidential elections, up 13 percentage points from the 2006 presidential elections in which Uribe was reelected. The high levels of participation by the most marginalized sectors of the Colombian electorate provided an important counterweight to the strong conservative bastion in the Andean region (see figure 2). ⁴ As López (2022) explains, “Abstention decreased precisely in the territories where Petro and Márquez have the most influence: on the Pacific and Caribbean coasts [, the Amazon,] and in big cities.” To further understand the surprising results of the election, we turn our focus to PDET communities, where historically marginalized voters provided critical electoral support to the Pacto Histórico.

Development Plan with a Territorial Focus (PDET)

Through the 2016 Peace Agreement, the government has overseen the rollout of several different programs designed to stabilize institutions and improve living conditions in the communities that have been most directly affected by the violence. The implementation of the different chapters of the peace agreement has been characterized by heterogeneous advances (Rettberg 2020; Angarita and Gelvez 2022). As of our writing, nearly 30 percent of the 578 stipulations in the Peace Accords have been completed (Kroc 2023). Out of those 578 stipulations, 130 focus on issues and rights related to gender and 80 on issues and rights related to ethnic groups, but only 11.54 percent and 12.5 percent of these stipulations, respectively, have been completed. In our analysis we focus on one of the most encompassing peacebuilding programs, the PDET. Relative to the overall agreement, the government has made substantial progress on completing the PDET program (50 percent of PDET stipulations have been implemented at the time of this writing). ⁵

PDET is a bottom-up policy designed to provide public goods and services. In the words of the peace agreement, its objective is “to achieve the structural transformation of the countryside and the rural environment and to promote an equitable relationship between rural and urban areas” (Gobierno de Colombia and FARC-EP 2016). PDET zones were created in areas that met four characteristics: high exposure to conflict, coca cultivation, low state capacity, and high levels of poverty (unsatisfied basic needs). ⁶ There are 170 municipalities in the program, covering 36 percent of the Colombian territory and including 6.6 million people, as shown in figure 1 (Gobierno de Colombia 2022).

PDET comprises several interconnected components, spanning from comprehensive rural reform to addressing historical disparities in land distribution, as well as improving infrastructure and social services like education, healthcare, and transportation (Gobierno de Colombia and FARC-EP 2016). Collectively, PDET places a strong emphasis on promoting economic and social development initiatives, particularly targeting projects that generate employment opportunities for vulnerable populations (Valencia and Chaverra 2022). Additionally, it recognizes the crucial role of local governance and institutions by focusing on building capacity and fostering participatory decisionmaking processes (Escobar Arango 2017). The program places significant importance on social inclusion and active participation, with a specific focus on empowering marginalized groups and addressing gender and ethnic disparities (Tovar et al. 2023). Furthermore, PDET incorporates measures to promote environmental sustainability, advocating responsible management of natural resources (Dávila et al. 2019).

The Comprehensive Rural Reform Program, of which PDET is a part, aims to “contribute to the reversal of the effects of the conflict and to change the conditions that have facilitated the persistence of violence” in these historically affected zones of the country (Gobierno de Colombia and FARC-EP 2016, 8). However, evidence from two survey waves (2019 and 2021) of 12,000 Colombians living in the 16 PDET areas finds that perceptions of overall security in PDET municipalities are quite low and have worsened over time (Weintraub et al. 2023). In 2021, respondents in 13 of the 16 PDET municipalities perceived that their security situation had worsened since 2019; 35 percent of respondents reported that conflict continued in their municipality, in comparison to 17 percent in 2019. While guerrilla violence in PDET municipalities has decreased steadily, the rate of targeted assassinations of social leaders has increased by 481 percent since the implementation of the Peace Accords began. The majority of these assassinations (more than one thousand) have occurred within PDET zones and have disproportionately impacted women social leaders (Llanes 2022; MOE 2018). By the time of the 2022 elections, many PDET municipalities continued to face significant threats to election security due to violence (Abierta 2022). So it is unlikely that PDET proxies for the effect of peace. ⁷

The absence of the state in Colombia’s peripheral regions is an often-used explanation for cycles of poverty, violence, corruption, and institutional weakness (Holmes and Amin 2014). According to Escobar Arango (2017), PDET proposes a solution to the “endemic problem” of centralized decisionmaking by the national government for peripheral regions to the exclusion of local community members. PDET provides a bottom-up approach to territorial reform, in which problems are identified and solutions are developed by community members themselves, first at the smallest administrative unit (veredas), then at the municipal level, and finally integrated into a regional action plan. According to official statistics collected by Ramírez Sarmiento (2021), between 2018 and 2020, nearly 1,300 PDET projects were completed through this community consultation process, leading to a projected $4.1 billion Colombian pesos of investments in PDET communities.

Despite PDET’s lofty goals for “transformative participation,” understood as institutional change with tangible effects on economic and physical security and full participation of citizens in policy and implementation, Ramírez Sarmiento (2021) argues that the reality often falls short. She credits poor program design, limited political will (under the Duque administration), and limited program implementation (beyond infrastructural development) for the program’s shortcomings with respect to transformative participation. Nevertheless, between 2018 and 2020, there were more than 33,000 programs involving more than 200,000 direct participants across the 170 PDET municipalities (Ramírez Sarmiento 2021). As one PDET participant in the south of Tolima explained, “PDET is like a life jacket that was thrown to us to save our territories and our families … now we have a voice that we can use to demand that peace be implemented” (Ramírez Sarmiento 2021, 81). In a series of testimonials gathered from PDET participants and members of PDET communities by the Agencia de Renovación del Territorio - ART, participation is one of the key words that people use when they are asked to describe what PDET is. ⁸

In sum, the communities that have been most affected by the civil conflict historically have been specifically targeted with a bundle of policies designed to improve their living conditions by expanding the terms of community participation and representation in collective decisionmaking. As a result, we consider the electoral implications for expanding the terms of political participation in these historically marginalized areas.

The Changing Salience of Violence to Voting

Prior to the ratification of the peace deal with the FARC, the civil conflict was the central cleavage in Colombian electoral politics (Gamboa 2018). Targeted political violence, particularly from paramilitary groups, aimed to influence elections in favor of right-wing parties and to exclude leftist social leaders and former combatants from political participation and elected office (Acemoglu et al. 2013; Fergusson et al. 2021; Steele 2011). Guerrilla groups, often lacking the strategic alliances with local elites, due to former president Álvaro Uribe’s counterinsurgency campaign (2002–10), used violence as a strategy to decrease voter turnout (Gallego 2018). The political polarization of peace supporters and hardliners reached its zenith during Juan Manuel Santos’s 2014 reelection campaign and the 2016 Peace Deal referendum, when a razor-thin majority voted “No” to reject the Santos administration’s proposed peace deal with the FARC (Rodríguez-Raga 2017).

The mobilization of the United Self-Defense Forces of Colombia (AUC by its Spanish acronym) in the mid-1990s increased the intensity of the conflict throughout the country and especially in the municipalities that would later become PDET zones. The election of Álvaro Uribe in 2002 on his platform of mano dura centered the conflict as an electoral cleavage (Wills-Otero 2014). In this environment of heightened conflict salience, exposure to violence was a strong predictor of vote choice. Acemoglu et al. (2013) find that Uribe won a larger share of the vote in the presidential elections in 2002 and 2006 in municipalities that had experienced a greater number of paramilitary attacks. They find a similar relationship between paramilitary violence and support for third parties running for legislative office during this same period. Weintraub et al.’s examination of the 2014 presidential contest between Juan Manuel Santos and Oscar Iván Zuluaga (2015) shows that exposure to guerrilla violence also affected vote choice in what they refer to as an election that boiled down to a referendum on the peace deal. As Colombian voters, especially voters in the most affected areas, continued to grow weary of the violence, many grew more supportive of peace talks generally and the final peace agreement (Branton et al. 2019), as voters living in conflict zones were more willing to make concessions to armed actors if this meant bringing an end to the conflict (Arjona 2016b; Branton et al. 2019; Tellez 2019).

Armed groups use political violence to “achieve tangible political or military objectives” (Valentino 2014, 91). Paramilitaries and regional elites share interests in eradicating the presence of guerrillas and establishing a “wartime social order” within their territory that is favorable to the interests of large landowners, extractive industry, and foreign investment (Potter 2020). When it comes to elections, paramilitaries have used violence as an electoral tool to coerce voters to support right-wing parties and candidates (Arjona 2016a; García Sánchez 2016). This electoral relationship between paramilitaries and political elites has been possible where paramilitaries have achieved stable territorial control (Dube and Naidu 2015; Gallego 2018). Guerrillas, in contrast, have typically consolidated territorial control in peripheral areas where it is more difficult to build relationships with local elites. As a result, they have been more likely to “sabotage elections” by driving down turnout (Gallego 2018). Thus, political violence can affect electoral returns not only by affecting voters’ preferences but also by constraining their ability to freely exercise their suffrage.

Despite the central importance of violence to the literature on Colombian electoral politics in the last two decades, the electoral salience of the conflict has decreased significantly since the final ratification of the Peace Accords with the FARC. ⁹ While most elections before 2016 centered on the armed conflict (Wills-Otero 2014), by the second round of the 2018 presidential elections the principal candidates were no longer competing over hawkish and dovish positions toward the guerrillas. Both candidates had pledged to respect the terms of the peace agreement and were primarily competing over their contrasting positions on Venezuela, rising levels of poverty, and the stagnant economy (Gamboa 2018; Gamboa Gutiérrez 2019). The civil conflict continued to be a marginal issue on the national stage in the 2022 presidential elections (Botero et al. 2023). It is unlikely that we would observe the same effects of exposure to violence on voters’ electoral preferences that we observed in previous elections because the candidates were not polarized along this dimension at the national level.

Hypothesis 1 Violence perpetrated by leftist guerrillas and right-wing paramilitaries did not affect the municipal vote share for the Pacto Histórico.

We might expect that violence would be more salient in places that were disproportionately affected by the conflict at the height of its political salience and where insurgent groups and paramilitaries have continued to use violence against civilians. PDET communities exist at this intersection. Even though PDET has contributed to some reduction in violence, 40 percent or more of people surveyed reported the continued presence and governance of armed actors in PDET municipalities: “18 percent of respondents said ‘the conflict still persists here.’… for just over a million Colombians, of the 6.6 million living in PDET areas, the benefits of peace are still hardly visible” (PRIO 2020, 10). ¹⁰ This percentage increased in 2021 to 38.5 percent (MOE 2018). In anticipation of the 2022 elections, MOE reported that approximately 58 percent of the 167 municipalities with Circunscripciones Transitorias Especiales para la Paz (Special Transitory Peace Seats) (all of which are PDET municipalities) faced high electoral risks due to violence. The alarming rise in social leader assassinations has hit PDET municipalities especially hard. Albarracín et al. (2023) argue that paramilitary groups have particularly used violence to sabotage the implementation of “pro-peace policies” and reduce electoral competition.

Gustavo Petro and Francia Márquez were particularly attentive to PDET zones. Unlike Hernández, they proposed to improve the implementation of PDET, bring the The National Liberation Army (Spanish: Ejército de Liberación Nacional, ELN) and other dissident actors to the negotiating table, and improve community participation in this process (Estupiñán 2022). We argue that it is still unlikely that violence would have had much effect on voters’ preferences because the principal candidates in the first and second round were not polarized on the peace process. But to the extent that we would observe any impact of violence on voting in PDET zones, we would expect a positive boost for Pacto Histórico.

The Effects of PDET on Political Participation

We argue that PDET municipalities were critical sites for Petro and Márquez’s electoral support because of PDET’s positive impact on voter turnout. PDET is a multidimensional institution that spans a range of issues, from land reform to education, healthcare, and infrastructure. As explained above, municipal assignment to the PDET program was not random. PDET municipalities have experienced high levels of conflict violence, illicit crops are cultivated there, and historically these municipalities have had low levels of state presence and high levels of poverty.

We do not argue that political participation is the only mechanism that would predict distinctive patterns of voting in PDET areas relative to non-PDET areas, but we do argue that the participation mechanism is important for a few key reasons. First, it is counterintuitive that electoral participation has continued to increase in PDET zones, given the continuing issues with violence and insecurity since the Peace Accords. Second, while Gustavo Petro’s electoral strongholds remained stable comparing the outcomes in 2018 and 2022 and across the first and second rounds of the 2022 election, voters favoring Petro in these strongholds turned out in significantly higher rates compared to 2018. And third, while PDET’s track record on bringing peace and rural transformation has been mixed, these programs have mobilized hundreds of thousands of community members through their bottom-up approach to governance (Shenk 2023; Tovar et al. 2023). We argue that community mobilization through PDET has probably served as a counterweight to the structural obstacles to turnout in PDET regions.

Like the relationship between violence and vote choice, the relationship between political violence and voter turnout is contextual. Preelectoral violence can be used to turn out voters or suppress turnout, “congruent with the goals of the perpetrators and electoral environment” (Bekoe and Burchard 2017, 89). Gallego (2018) finds that FARC violence had a suppressive effect on electoral turnout from the mid-1990s until the early 2000s, while paramilitary violence affected competitiveness instead of turnout. This is consistent with the contrasting goals of paramilitaries and guerrilla groups: as paramilitaries continued to consolidate territorial power under Uribismo, the FARC turned to using violence to sabotage elections in the areas that they no longer controlled. This territorial mechanism is not unique to guerrillas; Steele (2011) finds that paramilitaries used violence to displace potential left-wing voters in territories traditionally controlled by the FARC. Albarracín et al. (2023) find that post-accord social leader assassinations have been more frequent in places where elections have been moderately competitive and leftist parties have traditionally been more successful. Thus, targeted violence by right-wing actors has been used to discourage voter turnout and other threatening forms of civic engagement in recent years.

In anticipation of the 2022 elections, the Mission for Electoral Observation (MOE) reported that approximately 58 percent of the 167 municipalities with Circunscripciones Transitorias Especiales para la Paz (all of which are PDET municipalities) faced significant electoral vulnerabilities as a result of the prevalence of violence (MOE 2018). Moreover, MOE reported that 162 of these municipalities were at risk of forced displacement, which would have the anticipated effect of suppressing turnout (Barreto 2022). ¹¹ Based on the threats of violence, particularly right-wing violence in PDET areas, we would have expected to see suppressed turnout in PDET areas, but we actually observed higher turnout in many of these areas in comparison to 2018 and even between the first and second rounds of the 2022 elections (Galindo 2022b). ¹²

The change in electoral turnout from 2018 to 2022 contrasts with the stable concentration of votes across the two elections, especially for Gustavo Petro. In the majority of cases, Petro did not flip municipalities that opposed him in 2018. Instead, he continued to earn a strong concentration of votes in the municipalities that he had carried in the previous election, but with higher levels of turnout (Albarracín and Milanese 2022a). Likewise, the same municipalities that rejected Duque in 2018 voted strongly in opposition to Hernández in 2022. Generally, the most marginalized municipalities in Colombia have cast pro-peace and leftist votes (Arjona 2016b; Branton et al. 2019; Gillooly 2022). ¹³ While it is possible that progress on municipal development projects, increased social spending, and improvements in education and health may have changed some voters’ preferences in favor of Petro, the consistent concentration of votes for the left casts doubt on this alternative explanation.

We argue that the PDET program probably boosted support for the left in 2022 by reducing the material barriers to turnout in highly marginalized municipalities. One important barrier to participation in PDET areas is poverty. Generally, there is a deficit in electoral participation between the rich and the poor in Latin America and other global regions. But community organizations can counter these structural disadvantages because “[they] do much of the hard work of mobilizing people into political activity by providing basic … infrastructure that reduces the costs of individual and collective action” (Boulding and Holzner 2021, 11). Community organizations provide information about candidates and policies that facilitates collective action, and they provide a link between voters and the candidates who seek to mobilize them (Holland and Palmer-Rubin 2015).

PDET is designed to facilitate community input and participation, from the design to the implementation of community development projects. As it was mentioned before, from 2018 to 2020, more than 33,000 programs engaged with more than 200,000 direct participants across the 170 PDET municipalities. The program rests on the foundational interests, input, and oversight of los nadies, the Indigenous and Black communities, campesinos, and women “that permit [the] speaking of everyone, for everyone” (Velásquez and Londoño 2022, 19–20). It is possible that this process of community mobilization boosted participation by increasing faith and trust in the state, although recent studies suggest that participation increases trust and satisfaction with government and not the other way around (Kostelka and Blais 2018; Norris 2011; Pacek et al. 2009). In addition, members of PDET communities have complained that they have been excluded from consultation during the implementation of development projects (Tovar et al. 2023). On average, the level of satisfaction with security and public works in PDET communities has decreased over time (Weintraub et al. 2023). For these reasons, we argue that PDET’s effects on voter turnout are a function of the “mobilizational infrastructure” that community participation in PDET has produced (Boulding and Holzner 2021), rather than its effects on political preferences, on trust in government, or its impact on violence.

In summary, we expect to find that voters in PDET municipalities cast a higher share of votes for the leftist Pacto Histórico coalition in the 2022 presidential election. We argue that PDET boosted the leftist vote share by increasing voter turnout in areas that were already favorable to Pacto Histórico. Still, there are valid reasons to expect that the characteristics of marginalized municipalities that make them eligible for participation in PDET, rather than the actual PDET program, are also responsible for an increase in support for the left in the 2022 presidential elections. Municipal assignment to the PDET program was not random. Specifically, all PDET municipalities have experienced high levels of conflict violence, illicit crops are cultivated there, state capacity has been historically low and poverty rates have been much higher than the national average. Marginalized communities in the Colombian periphery, such as PDET municipalities, showed strong support for Gustavo Petro in 2018, for Santos (in opposition to Óscar Zuluaga) in 2014, and for the Peace Deal referendum in 2016. These highly marginalized communities already exhibited political preferences distinct from those of less marginalized municipalities. In terms of our analysis, if the explanatory effect of PDET on left coalition support is reduced once controls for PDET characteristics are introduced to the model (selection effect), it would reduce our confidence in the participation effect argument that we have outlined above. The following hypotheses lay out the observable implications of the two possible ways that PDET would predict stronger electoral support for the left, the selection effect (hypothesis 2.1) and the participation effect (hypothesis 2.2).

Hypothesis 2 PDET municipalities will be more likely to support the leftist coalition compared to non-PDET municipalities.

Hypothesis 2.1 Left coalition support will be similar in PDET municipalities to non-PDET municipalities with similar characteristics prior to the implementation of PDET (historical violence, coca cultivation, low state capacity, economic vulnerability). ¹⁴

Hypothesis 2.2 Left coalition support will be greater in PDET municipalities than in non-PDET municipalities with similar characteristics prior to the implementation of PDET (historical violence, coca cultivation, low state capacity, economic vulnerability), as a function of higher turnout in PDET municipalities.

Data and Methodology

To assess the impact of peace implementation and conflict violence on vote choice in the Colombian 2022 presidential elections, we utilize municipal-level data, the most micro-level unit of publicly available information. The municipal data provide a rich trove of information regarding the spatial distribution of violence, Peace Accords implementation, and support for Petro-Márquez. Our dependent variables, Petro-Márquez Vote Share Round 1 and Petro-Márquez Vote Share Round 2, are the percentage of support for Gustavo Petro and Francia Márquez at the municipal level in the two rounds of the presidential elections. These data were collected from the Registraduría Nacional del Estado Civil, the government body that administers elections.

Our key independent variables are municipal-level measures of violence and PDET implementation. For rebel groups’ violence, we use Acciones subversivas, which is the sum of the number of military interactions between guerrillas and Public Forces in the municipality from 2003 to 2020. ¹⁵ This includes FARC actions, ELN, and three small guerrilla group attacks (ERP, ERG, and EPL). ¹⁶ To proxy for paramilitary violence, we use Social leaders, the sum of social leaders killed in a municipality between 2005 and 2020, compiled by Llanes (2022). ¹⁷ As a robustness check, we also used the VIPAA dataset, which counts the violent presence of armed actors (paramilitaries and guerrillas) in Colombia between 1988 and 2019, collected by Osorio et al. (2019). ¹⁸ We use a dichotomous variable, PDET, that equals 1 if the municipality is part of the peace program, 0 otherwise. The 170 PDET municipalities are displayed above in figure 1. Data on PDET zones were obtained from la Agencia de Renovación del Territorio (the Agency of Territorial Renovation), the government body that implements the Peace Accords.

Figure 3 presents two graphs reflecting the relationship between the support for Petro-Márquez in the first round and the two key independent variables, Subversive actions and PDET implementation. The left panel displays the relationship between Vote Share Round 1 and Acciones subversivas, while the right panel displays the relationship between Vote Share Round 1 and PDET. Both panels reflect a positive relationship; municipalities with a greater number of subversive actions over the last two decades voted more for Petro-Márquez in the first round. Likewise, PDET municipalities went for Petro-Márquez at a higher rate compared to non-PDET municipalities. We also conducted similar exploratory correlation analyses involving other forms of violence (i.e., Social leaders), and we identified a positive relationship with the vote share of Petro and Márquez in this context.

Figure 3. Plots of Petro-Márquez Vote Share in the First Round Across Key Variables

In addition to the primary variables of interest, our models and quasi-experiment include several potentially confounding municipal-level social and economic demographic factors. We use data obtained from a databank hosted by the Economics Department at the Universidad de Los Andes. ¹⁹ Those control variables include the level of poverty as the average of the index of unsatisfied basic needs (NBI in Spanish) and municipal demographic and natural characteristics, such as a log of municipal population, altitude, and distance to the capital city of the department, all of which condition rebel recruitment and the effectiveness of state policing (Fearon and Laitin 2003). We also use data culled from the Departamento Nacional de Planeación to measure subnational government quality and national government monetary transfers to the local government.

Likewise, we control for the hectares of coca crops culled from the Ministerio de Justicia and the United Nations Office on Drugs and Crime, and electoral variables, such as previous presidential electoral results, turnout, and political alignment from the Registraduría Nacional del Estado Civil and published by the Misión de Observación Electoral and Pachón et al. (2014). ²⁰ We control for the ratio of the slave population in the middle of the nineteenth century because the historical geography of slavery has continued to predict support for parties, especially to the left of center (Ahmed et al. 2021). Also, in order to control for potential punishment of the incumbent, we control for the rate of deaths due to COVID-19 published by Datos Abiertos Colombia. ²¹ The definition and main descriptive statistics of the mentioned variables can be seen in the appendix, table A1. Finally, department-fixed effects are used to account for unobserved heterogeneity.

We test our hypothesis in three different ways. First, we use an ordinary least squares estimator and control for the factors explained above, which might influence armed violence by groups on the left and the right, policy implementation, and electoral support. ²² Second, we take advantage of the PDET’s policy design to create a propensity score matching (PSM) design. This quasi-experiment allows us to compare the effects of violence and PDET on the Petro-Márquez vote share in municipalities with similar characteristics before PDET implementation. Through PSM we use these pre-PDET covariates to estimate a maximum likelihood model of the conditional probability that a municipality would be designated to participate in PDET (Rosenbaum and Rubin 1983; Cunningham 2021). Taking into account the four main selection criteria for PDET, we utilize four municipal characteristics: high levels of unsatisfied basic needs, NBI; average hectares of coca crops from 1998 to 2017, Coca crops; Acciones subversivas; and Local capacities, low local administrative and management capabilities. ²³ Finally, using the propensity scores, we simulated a difference-in-difference design to test our mechanism.

Results

Table 1 indicates that Acciones subversivas are not significantly related to Vote Share Round 1 and Vote Share Round 2. Although there is an initial correlation between Acciones subversivas and Vote Share (model 1 and model 4), this relationship decreases in magnitude and significance when we introduce a full set of controls to the model. In particular, when PDET interacts with Acciones subversivas, the uninteracted violence measure loses significance, and we observe a relatively small, negative, and nonsignificant relationship between Vote Share in PDET municipalities. In other words, we find that conflict violence is unrelated to Petro-Márquez vote share, both in non-PDET and PDET municipalities. Consistent with hypothesis 1, municipalities that experience more civil conflict violence were not more likely to vote in favor of the leftist coalition candidate. ²⁴

Table 1. Petro-Márquez Vote Shares and Peace Implementation

*** p<0.01, ** p<0.05, * p<0.1

Standard errors in parentheses

The results in table 1 reveal that PDET municipalities have a positive and significant correlation with the vote share for Petro-Marquez, as predicted by hypothesis 2. Importantly, we find a consistent, positive relationship between PDET and Vote Share Round 1 and Round 2, challenging hypothesis 2.1. Our conservative models suggest that in PDET municipalities, the left-wing coalition experienced an increase in vote share, ranging from 1.2 to 2.3 percentage points compared to non-PDET municipalities. As discussed above, initial models show that violence among PDET municipalities was a determining factor initially, but this negative interactive relationship does not reach statistical significance in models with control variables (see tables A2 and A3). Figure 4 provides a substantive interpretation of these results, based on models 2 and 5 in table 1. In sum, the Petro-Márquez vote share increased by an additional two points in both rounds in PDET municipalities. ²⁵

Figure 4. Marginal effects of key variables on Petro- Márquez victories

As part of a robustness check, we also examined hypotheses 1 and 2 using the VIPAA dataset (Osorio et al. 2019). The results, as shown in table A4, indicate that the variable PDET is statistically significant and correlates significantly with Petro-Márquez’s vote share in both the first and second rounds. To further investigate the relationship, we conducted a detailed analysis by disaggregating the violence perpetrated by paramilitary groups, which are the main right-wing rebel actors in Colombia (see table A5). Even after controlling for the aforementioned factors and considering violence perpetuated by left and right-wing rebel groups, our findings indicate that violence does not significantly explain the 2022 presidential electoral outcomes. These results are consistent with hypothesis 1. On the other hand, in line with hypothesis 2, the analysis confirms that PDET municipalities do exhibit a positive and statistically significant correlation with the vote share for Petro-Márquez.

The OLS analysis cannot fully rule out hypothesis 2.1. Despite the robustness of our results with multiple specifications and controls, selection into PDET was based on municipal characteristics, and these criteria, rather than PDET itself, may be the actual cause of the relationship that we observe. To account for selection bias, we employed a propensity score matching (PSM) approach. PSM enables us to quantify the effect of the treatment, in our case the implementation of PDET, by matching each PDET municipality (“treated unit” in the language of PSM) with a non-PDET municipality (“untreated unit”) that had similar characteristics, effectively creating an artificial control group (Rosenbaum and Rubin, 1983; Cunningham 2021). Figure A6 shows the result of a logit model, where the outcome variable is Treatment status (PDET) and includes the four covariates: unsatisfied basic needs, the log of hectares of coca crops, number of subversive actions, and an administrative and management capabilities score. The estimated propensity scores by treatment status can be seen in figure A6.

In order to match PDET municipalities to non-PDET municipalities with highly similar characteristics (propensity scores, and thus potential for inclusion in PDET), we use nearest-neighbor matching. In this method, a distance is calculated between each treated unit and several control units, and each treated unit is paired with at least one control unit (Greifer 2022). Nearest-neighbor matching is particularly advantageous in our context, due to its simplicity and transparency, making it easier to communicate and replicate our matching procedure. Additionally, it allows us to make efficient use of available control units, maximizing the sample size of the matched dataset and enhancing the statistical power of our subsequent analyses (Thoemmes and Kim 2011). ²⁶

Once the matching was completed, we analyzed the impact of PDET on Vote Share Round 1 and Vote Share Round 2. Model 1 and model 2 in table 2 indicate that PDET had a positive and significant impact on Petro-Márquez’s vote share in both the first and run-off rounds. This suggests that being part of the PDET program, a key component of the peace agreement, led to increased support for the left-wing coalition by 1.5 in the first round and 2.4 percentage points in the second round, compared to the 2018 election. These findings strongly refute hypothesis 2.1, which established that support for the left coalition was similar in PDET municipalities and in non-PDET municipalities with similar pre-implementation characteristics (historical violence, coca cultivation, low state capacity, and economic vulnerability). Instead, our results suggest that the local implementation of PDET provided an electoral boost to the Petro-Márquez coalition.

Table 2. The Effect of PDET on Petro-Márquez’s Victory

*** p<0.001, ** p<0.01, * p<0.05

Standard errors in parentheses

So far, the evidence presented is support works of hypothesis 2.2, but we have not yet established the link between PDET and increased voter turnout—a core part of our argument in this article. The strong predictive effect of the 2018 Petro vote share is a step in this direction, as it shows the stability of the Petro vote from 2018 to 2022. Consequently, evidence of a significant boost to turnout in our PSM analysis would provide a strong test of our proposed mechanism. We therefore use our propensity score matching to estimate a difference-in-difference (DD) design, with PDET municipalities serving as the treatment group, and the control group consisting of their previously matched neighbors in the PSM. Substantively, this allows us to estimate the effect of PDET on turnout by comparing differences in the rate of change in turnout for PDET municipalities between the 2018 and 2022 presidential elections and the rate of change in non-PDET municipalities. Using PSM with DD provides a more comprehensive assessment of causality, reducing the potential for bias and increasing the precision of estimates (Stuart et al. 2014).

The results of the 2022 elections, as demonstrated in table 3 and figure A8, indicate an overall increase in voter turnout compared to the 2018 elections. Moreover, the increase was more significant in PDET municipalities than in their non-PDET matched counterparts. PDET municipalities saw an additional 5.3 percentage point increase in voter turnout in 2022. This supports hypothesis 2.2 and suggests that Petro-Márquez’s support was stronger in PDET municipalities with similar historical characteristics (such as historical violence, coca cultivation, low state capacity, and economic vulnerability), due to higher voter participation. While the ideological leanings of these new voters cannot be determined, the evidence suggests that the positive effect of PDET on turnout largely captures the consistent strength of the left in historically marginalized areas and the mobilizing effects of PDET.

Table 3. Difference-in-Difference: Effect of PDET on Voter Turnout

*** p<0.001, ** p<0.01, * p<0.05

Standard errors in parentheses

Conclusions

Our research studies the effect of peacebuilding institutions on political behavior. In particular, we use the case of Colombia to underscore the robust support for the left coalition in the context of peacebuilding, during both rounds of the 2022 presidential elections. We find compelling evidence to suggest that PDET had a positive impact on the leftist vote share, primarily through its influence on voter turnout. Furthermore, our findings shed light on the evolving dynamics of conflict violence in the electoral landscape. The 2022 elections witnessed a notable shift, with all major parties and candidates endorsing the Peace Accords, thereby mitigating any directional effect of exposure to conflict violence on municipal vote share.

Our study carries significant implications for the broader field of peacebuilding institutions and electoral politics. We highlight that electoral violence, often employed as a deliberate strategy by armed groups to suppress competition (see, for example, Gallego 2018), can be countered through citizen engagement facilitated by participatory institutions. In other words, our findings suggest that citizen engagement through participatory institutions can play an important role to counteract the suppressive effects of violence.

However, it is crucial to exercise caution in generalizing our results. Questions remain about the future support of PDET communities for the left, especially given the uneven progress of the Peace Agreement and the PDET program (Weintraub et al. 2023). Violence continues to impact politics in many PDET regions, raising concerns about the potential consequences of failing to address safety, public service, and resource disparities in these areas in future elections.

The overall conditions in 2022 were highly favorable to the leftist coalition. High levels of frustration with the incumbent right-wing government, due to the social and economic burdens of the COVID-19 pandemic, the highly unpopular proposal to reform the pension system by the incumbent Duque administration, and the same government’s violent response to the protests, provided broad support for an antiestablishment candidate (Botero et al. 2023). That the leftist coalition provided a more programmatic and transparent appeal to voters is, no doubt, an important explanation for the historic victory of the left over mainstream and right-wing alternatives.

Given the mix of factors that make 2022 so unique, will PDET communities continue to support the left? The progress of the Peace Agreement, and specifically the PDET program, has been extremely uneven. Violence still matters to politics in most of the 16 PDET regions (Weintraub et al. 2023). Social leaders in these areas are killed at higher rates, and armed groups still occupy territory in a number of PDET municipalities (Albarracín et al. 2023). Could a failure to systematically address safety, public service, and resource gaps in these areas lead voters to disengage or even punish the Petro-Márquez government in future elections?

Future work on the electoral, developmental, and security effects of peacebuilding institutions should explore the variation in program implementation and conflict violence reduction to better understand the implications of the varying degrees of satisfaction with peace programs for political behavior. This article analyzes a snapshot of the PDET effect aggregated across 170 municipalities with varied experiences and outcomes. Follow-up studies will also need to investigate whether the direction and strength of this effect will remain consistent across time. Our municipal-level analysis provides rigorous evidence of an effect on vote share and turnout, but more micro-level qualitative and quantitative data are needed to better understand how PDET aligns with political preferences, trust in institutions, and political efficacy. Especially under the shadow of continued political violence, the success of PDET in bringing peace will be critical to our analysis of the political implications of peacebuilding.