Syndicate content

Measuring the cost of conflict and violence

Gary Milante's picture

I've just returned from a very interesting conference on measuring violence hosted by UNDP and the Small Arms Survey/Geneva Declaration.  Measurement issues are always vexing in this field and the people that work on conflict are well aware of the challenges.  It turns out that there are very few statisticians running around battlefields, tracking gangs and trafficking or monitoring coups and political instability.

First, let me say, we met for two days.  The first included 12 (twelve!) presentations on challenges and approaches for measuring armed violence, and during the second day participants broke in to working groups to develop a list of "Millennium Development Goal (MDG)-like" indicators for violence reduction. 

Both days were really helpful to our work on the World Development Report 2011.  The first validated the measurement techniques we are using to quantify violence (core indicators being direct battle deaths, one-sided violence and homicides, complemented by other sources), and the second day challenged us in group exercises to try to come up with some MDG-style indicators on violence.

Measuring Armed Violence: This confirmed many of the approaches (and limitations) we face in collecting data on this topic.  In addition to people who work with us on the WDRAndy Mack (Human Security Report), Keith Krause/Rob Muggah (Geneva Declaration/Small Arms Survey), Debby Sapir (CRED) and Steven Malby (UNODC)we heard presentations from WHO and BCPR representatives, as well as research institutions, including FAFO, STATaid, DIW and the Brazilian Forum on Public Safety.

Their presentations were particularly helpful because they confirmed our approach for the WDR. Basically everyone agreed that direct battle deaths and one-sided violence data from PRIO and homicides from UNODC are the best available sources, though minor improvements can be made at the margin through additional input.  A few relevant and interesting tidbits out of related work presented:

  • Steven Malby (UNODC) presented on homicide data and had two very compelling graphs.  The first was on the relationship between governance and homicides, showing not only a negative relationship, but a clustering of types of countries.  The second figure was on disaggregated homicides in the Philippines, which showed spikes in homicides in Manila around the time of elections.  This is really fascinating stuff, why are those spikes there?
  • Rodrigo Soares presented some compelling comparisons on crime rates (from victimization surveys) and reported crime rates (from police records), showing that reporting rates are much lower in countries with high corruption.  This is kind of the core part of the expectations and institutions argument we are exploring in the WDRcorruption serves as a tax on crime reporting, broad perceptions of corruption in government in general can lead to less trust of police which results in lower rates in general.
   Photo © Arne Hoel/World Bank

MDG-like Indicators: On the second day, we broke in to three groups and were tasked with brainstorming on what MDG-like goals, targets and indicators would look like on armed violence.  This exercise and the presentations and discussion that followed were very interesting, reflecting various perspectives of disciplines and agencies.  We found consensus on only one clear goal: 'Reduce armed violence'measured basically by battle deaths, homicides and indirect deaths (where possible).  Other goals suggested by the group included reducing indirect impacts of armed violence (costs), limiting the enabling environment for conflict and violence (risk reduction), and a process goal of developing violence reduction programming.  However, all of these last three had significant disagreement around the table.

As an economist, I have to wonder whether reducing the costs of armed violence is a good goalno matter how you interpret 'costs'from realized deadweight loss to actual expenditures on conflict (or avoiding it).  Shouldn't we be lowering the number of incidence and not the costs associated with conflict?


Submitted by Jon Rudy on
Thanks for this interesting report. Exposing the true cost of violence may be one more way to convince people and institutions that violence is a bad idea. I don't see a lot of economists making direct impacts on lowering the incidence of violence, but by exposing the costs, they can add their voice to many others-- CR specialists, policy makers, educators and others. For peacebuilding, we need every contribution.

Jon, thanks for the comment. It is a challenging field but there is lots of good work being done on this. Two first good starts are the Geneva Declaration's Global Burden of Armed Violence: and the work by the Bank unit SDV on measuring costs of violence: There are always, of course, issues with comparability across countries and over time, but these sources provide some well formulated statistics, helpful for peacebuilding policymakers.

Submitted by jennorins on
Thanks for sharing this summary of the conference on measuring the cost of violence and conflict. Was there any discussion about the licit or illicit trade of arms? I would imagine that tracking illicit flows of (small) arms would be challenging, but it should be an important factor in the incidence of violence and conflict. And documenting the legal transfers of arms would (and the number of arms "missing") also would be a valuable measure of incidence. If we are truly serious about decreasing the extent of direct or indirect violence, the financial gains from arms trade have to be addressed. I am also curious if there was discussion on demining?

Jen, thanks for the comment. Yes, actually arms trade was discussed as an indicator. You may be interested in the text of the Geneva Declaration which has now been signed by more than 100 countries: It specifically explores the relation between armed violence and development. As I understand it, one of their challenges is including text on arms trade in the declaration for some of the signatories. We didn't discuss demining as much, primarily I think because of the paucity of the data. In related research in DEC, we are working with UNMAS and the International Campaign to Ban Landmines to help collect better quality data on landmine contamination, casualties and clearance.

Submitted by Rex Brynen on
At the risk of being provocative, does it make much sense to address the costs of armed violence without addressing the benefits--or, more accurately--the *perceived* benefits of such violence? One doesn't have to be a rational choice theorist (which I'm certainly not) to recognize that collective violence occurs, in part, because participants feel that it is in some way "cost effective." Here, much would seem to hinge on the particular combination of motivating factors at work. Is violence being driven by key elites who benefit from the war economy, or who hope to garner additional future resources through violence? If so, efforts to limit violence need to be focused on reshaping the political economy of elite gain (through limiting gains or offering side payoffs) and/or paramilitary recruitment (by assuring the presence of economic alternatives, etc). Conversely, if violence is driven by a widespread sense of grievance--grievance that can't be addressed through alternative lower-cost channels--then the policy imperative becomes more of a diplomatic-political one, whereby we hope to create alternative nonviolent channels for dispute resolution. This, of course, is at the root of much of the "greed and grievance" debate about the dynamics of civil conflict. It is also much of the basis for past and present peacebuilding efforts. It is complicated in practice, however, by two confounding dynamics: 1) We often see both sets of (greed and grievance) motivations at play, entwined in complex ways and various vicious circles (for example, grievance > conflict > war economy > predatory rent-seeking > greater social inequality > the emergence of entrenched interests associated with continued war, as well as a deepening of grievances). 2) Once violence erupts, elites and social groups find themselves trapped within classic security dilemmas. Paramilitary preparation and even violent preemption appear less risky than cooperative behaviours. Confidence is destroyed, rather than built. Ideologies of hatred and social differentiation become more deeply entrenched. Functional interrelationship and common interests are fractured. In the first case, the very fact of conflict alters the fundamental terrain of material cost-benefit analysis by local actors. In the second case, internalized perceptions of costs and benefits are similarly transformed. It seems to me that the latter point highlights an important caveat in any effort to measure the "costs of armed conflict." What is important in shaping the conflict is not so much any objective measure of cost (human, economic, or otherwise), but rather the cost perceived by the actual parties themselves.

Rex, we find ourselves again at the intersection of policy and academia. All of what you say is true if we want to understand the underlying dynamics of the conflict, however, for policy and advocacy purposes, aren't the indicators themselves enough? Recall that this exercise was an effort to identify MDG-like indicators on violence - where is it, how intense is it, how do we know when we have prevented or mitigated it? I agree, to ADDRESS the conflict policymakers have to know who stands to gain, how these costs are internalized and what other levers can be pulled to shape/alter/transform the incentives. If we are just trying to understand the scale, scope and presence of conflict, aren't costs enough?

Submitted by Rex Brynen on
I'm just being provocative ;) Imagine two possible scenarios, one of them counterfactual: 1) Conflict intensifies in a country, leading to growing armed clashes between angry marginalized shantytown youth and the authorities. As many as 500 die, and the stability of the political order is increasingly undermined. 2) Development efforts reduce off youth alienation. No violence erupts. The political order remains intact. MDG-type violence indicators would, presumably, score the first as the "worse" outcome, and the second as "better." The first outcome is the 1976 Soweto Uprising, which helped end apartheid in South Africa. The second implies a longer continuation of white minority rule. Is #2 still better? The difference between measuring violence and a MDG indicator is that things like infant mortality or literacy have unequivocal "better" and "worse" values. It isn't clear to me the same is true of violence, where the costs of human suffering may be counterweighed by gains in justice, freedom, etc.

Submitted by Anonymous on
If we want to look at "costs" of violence as a way to get attention to the need for prevention or mitigation, then we also need to understand the "benefits" of the violence - identifying who reaps the benefits and how and why. I think that Gary's response to Rex's first comment didn't take this issue seriously enough!

I found your blog and discussion very interesting. I have a couple of comments/suggestions: First, I think that a lot of what is missing on "Measuring the cost of conflict and violence" is actually related to analysing the impacts of conflicts (and violence, depending on what is meant and how is measured). A lot can be done by using the existing MDG indicators (national ones and global ones). For example, assessing how conflicts affect malnutrition or schooling of children. Second, there is now a substantial amount of evidence within the applied micro area, but also from medical literature on the effects of conflict on health-related outcomes (nutrition, morbidity, mortality, PTSD). What I think is missing is more evidence on the effects on other indicators such as on education and labour. Here, I agree that not all impacts need to be negative; there might be "positive impacts" such as the reduction of child labour in armed conflicts, to the extent that we think there is a trade-off between schooling and labour. Third, evidence-based policy draws from knowing the mechanisms through which conflict/violence affects people, institutions and the economy. Here, empirical evidence is poor and knowing that there are effects does not necessarily help humanitarian workers or policymakers, we ought to also know the channels and how the conflict has translated into such and such. My final comment is on your case studies, I find it surprising that the WDR 2011 does not envisage having a case study on Iraq, despite the extent of public attention that it has had, not to mention the costs on human lives.

Your report is compelling. Can you provide more detail on Steven Malby's work? In particular which countries show up in the homicide clustering - and how do they co relate to other country rankings, corruption, poverty, conflict, governance etc. The Manila report also cries out for further reporting and would make for some an interesting journalism in the light of the November 23 killings of 57 people, including 32 journalists and media workers in Maguindanao province. Those killings were allegedly carried out to prevent Esmael Mangudadatu from challenging the the powerful local politician and political clan leader Andal Ampatuan Sr’s control of the province in May’s elections. Your report on Rodrigo Soares presentation on levels of corruption and crime/victimization reporting seems a good jumping off point for a WDR analysis. Would investigative reporting that put a spotlight on the costs of entrenched corruption ameliorate the situation or simply lead to media repression? Beyond logging the cause and effect in the WDR what prescriptions do you have in mind to improve the situation?

Add new comment