After the optimism that reigned during the self-declared ‘end of history’ in the early 1990s, it turns out that political violence has been far from eradicated in recent years. What has changed, however, is the face of conflict. Several recent civil conflicts increasingly take the form of ‘complex warfare’, blurring traditional distinctions between classic military conflict (think of the US civil war), insurgency (e.g. the Cuban revolution), sectarian violence (the Rwandan genocide), and riots (recent food riots in Venezuela).
A substantial part of current political violence is directed against civilians (see Figure 1). Most prominently, Syria has now experienced more than six years of civil conflict, involving dozens of parties and overlapping allegiances, and resulting in shocking levels of human suffering. But other hotspots such as Ukraine, Yemen, Mali, Iraq, and Afghanistan also exhibit traditional guerrilla warfare combined with elements of ethnic violence between various groups, leading to many civilian fatalities.
Figure 1 The importance of political violence and protest events
Source: Armed Conflict Location and Event Data.
There are a number of burning questions that arise in this context. Under what circumstances can local interventions help to disrupt violence? How should resources be allocated to minimise the probability of future conflict? And what kinds of policies can get at the roots of violence or of peace?
It is a feature of the complex wars we are facing that larger, strategic or ideological goals are pursued with sectarian support, i.e. with the reliance on parts of the local population. It is therefore no surprise that military planners and academics agree that understanding this local support is key to furthering peace (US Army 2006). Here is where the increasing availability of local data poses a great promise. With local data we will, so is the hope, understand what drives violence locally.
Several findings are emerging from this micro approach already. Dube and Vargas (2013) and Berman et al. (2017), for example, link the availability of natural resources to the extent of violence. König et al. (2017) show the importance of networks in conflict, and Berman et al. (2011) analyse the impact of public policies on violence in Iraq. In addition, academic research has identified the configuration of groups across space as an important determinant of conflict. For example, Michalopoulos and Papaioannou (2016) link violence to the fact that some ethnic groups in Africa are divided by national boundaries. Rogall (2014) finds that the travel cost of armed groups to instigate violence during the Rwandan genocide was an important determinant of the number of deaths in different parts of the country. In other work, Dell (2015) finds that violence in Mexico occurred more often in locations that happen to lie on optimal drug trafficking routes of Mexican cartels. Based on this result, she has developed techniques to optimise the allocation of law enforcement resources to disrupt the trafficking networks.
While recent advances in this literature have been impressive, many challenges remain. In particular, there is an important pitfall when using spatially disaggregated data to analyse the drivers of local support. Depending on the military technology used and the spatial distribution of the population, we might not observe the violence where it originates. For example, if the political and economic exclusion of an ethnic group motivates it to raid other areas, then we will observe violence that is motivated by exclusion but observe it in areas that are relatively rich. This does not matter very much for the aggregate analysis at the country level but the more detailed the data the more the distinction between origin and target of attack matters.
Our study contributes to this new research by investigating localised conflict based on the spatial configuration of groups in Northern Ireland (Müller et al. 2017). The conflict between Catholics and Protestants in Northern Ireland was one of worst instances of political violence in Western Europe between the late 1960s and the early 1990s, resulting in more than 3,000 casualties. We geo-locate the occurrence of all casualties, often to a precision level of below 100 metres, and combine this information with fine-grained information from the census on the distribution of Protestants and Catholics both at the moment of the outbreak of the conflict and towards the end.
Our innovation is to model conflict not as a process that takes place in a single location but rather as the interaction of a source and a destination – if perpetrators and victims live close to one another, violence is much more likely than if they are physically separated by distance or barriers. Using this premise, we develop a new methodology to estimate the extent of violence in a location using the spatial distribution of potential perpetrators and victims. In our game-theoretic model, the proximity to potential victims affects the attack likelihood directly (by easier access) and indirectly (by the ease of recruiting). When structurally estimating the parameters of the model, we find that indeed distance between potential victims and perpetrators strongly curbs the number of attacks. This methodology turns out to be a powerful predictor of the number of casualties across wards in Northern Ireland, also allowing us to estimate the flows of violence from a given point A to a point B. This means the strategic model can be used to back out the origins of attacks. In Figure 2 we show the number of attacks that originate and target every ward in Northern Ireland.
Figure 2. Origin and destination of attacks in Northern Ireland in the 1970s
Source: Müller et al. 2017.
If valid, such a model can help to target policies at the origin of attacks and attempt to change the interaction between local groups, for example, through initiatives that improve dialogue or power sharing in local government. This may, in many circumstances, be equally as effective as trying to address conflict at the macro level.
We test the validity of our model by looking at the placement of artificial barriers, so called peace walls, in Northern Ireland. We find that ward boundaries which, according to our model, were potentially particularly violent interfaces were also targeted by the UK government with barriers. In addition, we use the changes of violence from the 1970s to the 1980s to evaluate the share in the changes that can be explained by the composition of population. Our results suggest that changes in the geographical distribution of groups across Northern Ireland have significantly contributed to conflict abatement. If we interpret the correlation between changes in population and changes in predicted violence as causal, the population movement saved over 600 lives.
Do these findings imply that keeping rival groups apart is a long-run solution for peace? No, not by any means. In the long run, namely during post-war reconstruction, economic integration and the fostering of inter-group trade can be a strong promoter of trust and peace (Rohner et al. 2013). However, in the short run, when war is still raging, temporarily separating groups may help to protect civilians. What is of course important (and tricky) is to decide when is the right moment to start to tear down the walls and get together again.
Another lesson offered by Northern Ireland is the potential importance of power-sharing. The biggest and most persistent drop in sectarian violence has occurred after the 1998 ‘Good Friday Agreement’ brokered by Tony Blair, which brought large-scale power-sharing by Catholics and Protestants in Northern Ireland. A recent paper (Müller and Rohner 2017) shows that exogenous variation in local power-sharing in the decades leading up to this agreement had an important impact on local violence and may have therefore probably paved the way for peace in Northern Ireland.
Berman, E, J Shapiro, and J Felter (2011), "Can Hearts and Minds Be Bought? The Economics of Counterinsurgency in Iraq", Journal of Political Economy 119: 766-819.
Berman, N, M Couttenier, D Rohner, and M Thoenig (2017), "This mine is mine! How minerals fuel conflicts in Africa", American Economic Review 107: 1564-1610.
Dell, M (2015), "Trafficking Networks and the Mexican Drug War", American Economic Review 105: 1738-79.
Dube, O, and J Vargas (2013), "Commodity Price Shocks and Civil Conflict: Evidence from Colombia", Review of Economics Studies 80: 1384-1421.
König, M, D Rohner, M Thoenig, and F Zilibotti (2017), "Networks in Conflict: Theory and Evidence from the Great War of Africa", Econometrica, forthcoming.
Michalopoulos, S, and Elias Papaioannou (2016), "The Long-Run Effects of the Scramble for Africa", American Economic Review 106: 1802-1848.
Müller, H, D Rohner and D Schönholzer (2017), "The Peace Dividend of Distance: Violence as Interaction Across Space", CEPR Discussion Paper 11897.
Müller, H and D Rohner (2017), "Can Power-Sharing Foster Peace: Evidence from Northern Ireland", mimeo, IAE and University of Lausanne.
Rogall, T (2014), “Mobilizing the Masses for Genocide”, mimeo, UBC.
Rohner, D, M Thoenig, and F Zilibotti (2013), "War Signals: A Theory of Trade, Trust and Conflict", Review of Economic Studies 80: 1114-1147.
US Army (2006), "Counterinsurgency. Field Manual No. 3-24", Headquarters Department of the Army.