by Leopoldo Fergusson* and Juan F. Vargas**
Democracy and conflict: a hard question
Civil conflict is one of the major threats to long-run economic development, and causes enormous human suffering. However, we still do not have a clear understanding of the factors that exacerbate or help mitigate the risk of conflict and, in particular, about the role of democracy in shaping violence.
In many theories democracy and conflict are interrelated, but testing the effect of democracy or democratic transitions on conflict is extremely challenging. On the one hand, democracy is a difficult concept to define and measure. On the other, there are many factors in society that may simultaneously affect the risk of conflict and the degree of democracy, making it hard to interpret any correlation between democracy and violence as evidence of a causal relationship.
To address these challenges and make progress in understanding whether democracy reduces or exacerbates the risk of conflict, in a recent paper we studied an experiment of history that allows us to evaluate the impact of one simple, measurable dimension of democracy: the size of the franchise.
An experiment from history
In 1853, Colombia enacted a Constitution that abolished slavery as well as literacy and wealth requirements for voting, effectively establishing universal male suffrage. This historical episode provides us with an opportunity for studying the impact of extending voting rights on the incidence of civil conflict.
A comprehensive population census administered in 1851, two years before the constitutional reform, allows us to build a proxy for the proportion of enfranchised citizens in each municipality. As it turns out, the overall increase in voting rights was very large since the preexisting wealth and literacy requirements, together with the existence of slavery, disenfranchised the majority of the population. We also compiled a novel detailed dataset of the number of political battles fought in each municipality each year over the course of the XIX century, a period marked by intense violence and frequent civil wars in Colombia.
Taken together, this creates an ideal setting to investigate the effect of franchise extension on civil conflict. We assess whether places where more people gained the right to vote experienced more or less violence following the new Constitution, and we* asked ourselves.
Is there less violence if more people can vote?
There is less violence where more people can vote
Our basic strategy and main results are illustrated in Figure 1. The figure divides XIX-century Colombian municipalities between those with a high proportion of enfranchised new voters and those with a low proportion of new voters. For each subsample we compute the average number of battles that were fought every year by municipality.
Figure 1: XIXth-century Battles in Colombia
Yearly average of municipal number of battles, by level of new voters
The circles in the figure represent this number for the municipalities with a high proportion of new voters. The triangles, on the other hand, are the corresponding averages for municipalities with a low proportion of new voters. Finally, the smoothed lines correspond to a regression that tries to `fit’ the data (the solid line for the municipalities with high proportion of new voters, and the dashed is for the places with a low proportion of such measure).
Vertical lines are shown for the year 1853 (when the Constitution extending the franchise was enacted) and for 1856 (when the first presidential election with universal male suffrage and direct voting took place). These are the dashed and solid lines respectively.
A number of important messages emerge from the figure:
1) Prior to 1853, municipalities with a high and low proportion of new voters do not differ in terms of the incidence of battles. This is very important for the validity of our approach. It means that municipalities with different proportion of new voters were pretty comparable before the reform took place. Hence, we can interpret what happened after the reform as a result of the increase in voting rights, and not as the consequence of pre-existing differences across different types of municipalities.
2) Starting around 1853, both groups of municipalities begin to diverge. In particular, in line with the idea that more democracy (here more voting rights for a larger share of the population) is a pacifying force, we observe that the incidence of battles in the low-voters municipalities is higher than in high-voters municipalities.
3) The difference seems to be maximized just after 1856, when the presidential elections under the new rule took place.
Interestingly, the violence gap between the two types of municipalities starts to close in 1863. This is a meaningful year, as Colombia adopted a new (ultra-federal) constitution with individual states changing voting rights. Most of the states reversed part of the extension of voting rights that the previous (1853) Constitution had enacted.
In short, Figure 1 shows that, while in effect, the extension of voting rights of the 1853 Constitution decreased violence in places that benefited more from the enfranchisement relative to places that benefitted less.
The interpretation of our results is however a bit more subtle than this.
As Figure 1 shows, the 1853 Constitution period was one of escalating violence. We then find that municipalities where more voters were enfranchised relative to their population experienced a lower incidence of violent battles relative to the rest of the country. That is, given there are elections, there is less violence if more people can participate in them.
However, compared to other years, the overall level of violence during the 1853 Constitution period was unusually high. While this increase could well have been fueled partly by the 1853 Constitution itself, we are not able to separate the effect of the constitution from the myriad of other changes in circumstances that could have occurred at this time in Colombia.
In fact, we do not rule out that elections may exacerbate violence, but we cannot directly test this possibility. What we can say is that, in our particular setting, giving voting rights to more citizens reduces violence.
What drives these results?
In the paper, we verify that the message from Figure 1 survives statistical scrutiny and is robust to a number of specification checks.
More importantly, we try to understand the potential mechanisms that cause the decline of violence in places with a higher proportion of new voters. In particular, we find that the impact of new voters is stronger in places with stronger state institutions and with more political competition.
We interpret both of these results as supporting two key ideas:
1) Violence in nineteenth-century Colombia (as in many other places) was essentially a technology for political elites to compete for the rents from power.
2) Elections with a larger franchise can undermine and substitute this technology, reducing violence.
That there should be more violence where the spoils are larger and the power of the parties is more evenly distributed is intuitive, present in many discussions by historians, and common in many theories of conflict. And elections with a larger franchise may substitute violence to the extent that they increase the legitimacy of democratic institutions, enticing groups in society to peacefully accept the rules of the game.
What is new in all of this?
The relationship between democracy and conflict is the subject matter of a large literature with a very distinguished pedigree. One line of research, the long-standing democratic peace hypothesis, suggests that democracies are unlikely to go to war with each other.
This idea goes far back: its origin can be traced to Immanuel Kant’s 1795 essay on perpetual peace. Among the arguments that have been used to support this idea is the fact that democracies face institutional constraints: in a democracy, the rulers are accountable to the population and the population at large wants to avoid the calamities of war.
Another argument is that democracies develop a set of “democratic norms” that preclude war from taking place. These types of arguments have been also made for intra-state conflicts. More generally, it has been argued that democracy allows for a peaceful allocation of power among the constituent groups that otherwise may resort to violence to seize power.
What is perhaps more striking is that the bulk of the empirical evidence contradicts this hypothesis. For instance, the existing research suggests that in the context of ethnically divided societies with weak institutions, a rapid democratization in the sense of introduction of elections may exacerbate violence. Among the possible reasons are that democratization strengthens political mobilization along ethnic lines, and elections create losers who may be unwilling to peacefully relinquish power.
So, despite some long-standing arguments in the opposite direction, there is considerable skepticism that democracy or democratization may reduce violence.
However, democracy is a multi-faceted concept, and progress can be made by focusing on its different ingredients to understand exactly how, when, and why it may lead to violence or peace. In fact, it is very unlikely that democracy unavoidably leads to war or peace.
Our paper thus contributes to the current state of knowledge by showing that in the case of a long-standing and violent political struggle, democratization in the sense of extending voting rights may reduce violence.
Our findings are consistent with the view that, with elections already in place, the increase in the size of the franchise strengthens the legitimacy of institutions reducing the incentives to engage in violence.