Rejecting Peace? Legacies of Violence in Colombia

Falling Short of Peace

Popular referendums to approve or reject civil war-ending peace agreements have been adopted in a number of countries as conflict parties seek to politically legitimate complex agreements. In a few cases, including in Guatemala and Cyprus, citizens ultimately opted to reject negotiated exits from war, producing uncertainty and confusion about popular commitments to peace.

On October 2, 2016, following four years of negotiations between the Colombian government and Fuerzas Armadas Revolucionarias de Colombia (FARC), the country’s largest and oldest rebel group, Colombian citizens were asked via a plebiscite whether they wished to accept or reject the final agreement. By a razor-thin margin, voters opted to reject the final agreement. What explains the decision by Colombian voters to reject the best hope for ending a five decades-long conflict?

Some point to the extremely high levels of abstention, with less than 40% of eligible those eligible casting their vote, due to apathy, to the fact that this was an unusual election that was not on the traditional election calendar, or to difficulties that voters on the Caribbean coast faced in reaching the polls due to the effects of Hurricane Matthew. Others point to failures of the campaign itself: claims of misinformation and lies on the part of the “no” campaign and claims of exaggerating the benefits of peace on the part of the “yes” campaign. These are certainly valid explanations, yet we lack hard evidence to be able to evaluate a number of these hypotheses.

Legacies of War

In this Policy Brief we assess whether legacies of the armed conflict, in addition to various sociodemographic and economic indicators, can help explain why some communities in Colombia voted in favor of ratifying the peace accords while others opted to reject the deal. To do so we take municipalities as our unit of analysis.

We use electoral data obtained at the municipal level from the Registraduría Nacional de Estado Civil, the governmental body that implements elections.

As the map in Figure 1 indicates, support for the agreement negotiated between the Colombian government and the FARC was primarily concentrated in peripheral regions of the country and along the coasts, while the central region (including Antioquia, home to Medellín and opposition leader ex-president Álvaro Uribe) voted to reject the deal. In the map, darker shades of blue indicate higher support for the agreement, white regions indicate rough parity of the vote, and darker shades of red indicate lower support.

We take our analysis further by combining the electoral data with data on violence taken from the Uppsala Conflict Data Program’s Georeferenced Event Dataset (UCDP GED) and the Human Rights Observatory Database, compiled by the Presidency of Colombia and based on daily bulletins internally compiled by the Colombian government’s security agency, including data on economic, political, and geographic factors (population size, rainfall, area of the municipality, distance from Bogotá, coca cultivation, royalties from natural resources, state presence, voting patterns in the 2014 presidential election, histories of different kinds of victimization by armed groups, poverty, and others). We estimate a set of statistical models, which seek to uncover why some municipalities voted for and others voted against the peace agreement, control for potential confounders that might affect both our chosen independent variables (such as past violence and poverty) and the dependent variable (percent of votes in a municipality that voted “yes”).

As we know from a large body of literature, victimization affects political behavior, including but not limited to decisions about citizens’ vote. We show that even after controlling for the factors mentioned above, the total number of FARC attacks perpetrated between 1988 and 2015 had an important effect on communities’ decisions about whether to support or reject the peace agreement. Communities that were not touched by FARC violence tended to vote no, whereas communities most heavily affected by FARC violence tended to vote yes.

As a first indication of this, in the map in Figure 1, we have added bubbles that are proportional in size to the number of battle-related deaths that UCDP GED have recorded in the district over the 2005-2015 period for all conflicts in the Colombia in the period. The UCDP GED measure of battle deaths are a conservative measure of victimization and only include direct battle deaths. This measure of victimizations therefore does not take into account the additional burden of conflict incurred through indirect deaths resulting from, for instance, a breakdown in state services, increased rates of poverty and hunger, and forced displacement.

Nevertheless, the map clearly indicates a relationship between battle deaths and the plebiscite vote. There appears to be a strong clustering of communities that have seen more battle deaths and communities with a higher degree of support for the peace agreement.

Using the longer data series of 1988–2015 on the number of FARC attacks in a community from the Human Rights Observatory Database, Figure 2 shows that for every additional FARC attack over this time period, we expect to see an additional 0.21 percentage point increase in yes votes in a given municipality.

The opposite pattern holds for cumulative paramilitary attacks: communities where paramilitaries engaged in high levels of violence over this period were less likely to vote to ratify the peace deal. As seen in Figure 3, an additional attack between 1988 and 2015 likewise results in a reduction of .22 percentage points for the yes vote.

Direct attacks by guerilla and paramilitary forces by themselves do not come close to giving a complete picture of how Colombian communities have been victimized by the more than 50-year long conflict. Indeed, the consequences of armed conflict extend far beyond attacks and direct deaths in battle. Armed conflict often leads to forced migration, long-term refugee problems, and the destruction of infrastructure. Social, political, and economic institutions can be permanently damaged.

The conflict in Colombia is no exception. Forced displacements have been a grave problem for Colombia, with more than 7 million victims of this form of violence over the course of the conflict. A similar pattern as the one described above holds for forced displacements. Illustrating this, Figure 4 shows a strong tendency for communities to be much more favorable to the agreement if those same communities had experiences of forced displacement. More generally, both communities that have received large numbers of displaced persons and those that have seen large numbers of people expulsed were far more likely to vote in favor of the agreement than those that saw fewer displacements.

A Divided Society

Underscoring the divided nature of Colombian society, economic factors also played a role in how communities voted in the plebiscite. As seen in Figure 5, poorer communities tended to vote more overwhelmingly in favor of the peace agreement than those that are comparatively more affluent. The figure uses the NBI index, a comprehensive measure of the extent to which individuals have unsatisfied basic needs, such as lack of access to clear water, education, and nutrition. The effect of poverty is easily on par with the effect of guerilla attacks. This may be because poorer communities were more likely to be affected by conflict and therefore sought to be free from violence, or because the peace agreement would have brought much-needed economic development to poor (especially rural) areas.