When Do Interventions Work?

by Erica Chenoweth on August 28, 2013 · 9 comments

in Violence,War

Jon Western is right to point out that there are certainly cases of intervention success. Indeed, those arguing in favor of intervention in Syria will surely draw on cases like Bosnia and Kosovo to make their case for intervention in Syria.

Setting aside the possibility that cases like Bosnia and Kosovo are not great examples of intervention success, there is certainly some persuasive evidence that military interventions can work. The cumulative research from people like Page Fortna, Barb Walter, and Michael Doyle & Nicholas Sambanis suggests there are some important qualifications to this.

First, the interventions need to take the form of peacekeeping missions—with well-resourced boots on the ground to protect civilians and enforce the peace.

Second, the missions must be multilateral, providing further credibility to enforcement and legitimacy to the cause.

Third, the peacekeeping missions must be multidimensional. This means they aren’t just military missions, but that they also involve extensive efforts at state capacity-building, humanitarian assistance, refugee resettlement, economic development, election monitoring, and the like.

Fourth, the combatants are ready to negotiate and consent to the intervention. Although there is some controversy about this (Fortna finds that Chapter VII enforcement missions are just as effective as consent-based ones), Doyle & Sambanis suggest that multilateral peacekeeping missions aren’t too successful at ending civil wars; rather, they help to enforce the peace once hostilities have ceased.

All of these conditions being present, there have been some peacekeeping successes (e.g. East Timor, El Salvador, etc.). Any of these being absent, the results are more mixed. All of them being absent, the outcomes of international intervention are much less favorable in both strategic and humanitarian terms. See several of my previous posts (here, here, and here) for more on this.


Andrew Gelman August 28, 2013 at 3:27 am


You express the results as a deterministic checklist (you need A, B, C, and D for all to work). But I assume that the research results are more of a continuous nature (i.e., more of A + B + C + D increases the likelihood of success)?

Erica Chenoweth August 28, 2013 at 2:59 pm

My reading of Doyle & Sambanis is that in the quantitative section they treat these as separate covariates influencing the outcomes. However in their case study section it seems to be the case that absent all four of these features, PKOs have a very mixed track record without full success. So I guess they would see these as necessary conditions. But someone can correct me if I’m wrong here.

Andrew Gelman August 30, 2013 at 2:18 am


Sure, but you also write, “All of these conditions being present, there have been some peacekeeping successes . . . Any of these being absent, the results are more mixed.” This suggests a statistical analysis which will be more complex than merely saying that these are necessary conditions.

Scott Monje August 28, 2013 at 9:52 am

If the belligerents have given up fighting and invited outsiders to patrol the streets and rebuild the country, is it still meaningful to call it intervention?

jonathan August 28, 2013 at 10:13 am

I agree with Scott. Peacekeeping is intervention in a different sense.

Erica Chenoweth August 28, 2013 at 3:02 pm

I believe that militarized peacekeeping/peace enforcement missions are technically considered interventions–especially Chapter VII missions (e.g. Libya 2011)–although they can vary as to the timing of the intervention (early-, mid-, and post-conflict). Regan’s 2002 study finds no meaningful difference between the timing of the intervention and the duration of the conflict, if I remember correctly.

Jon Western August 28, 2013 at 11:15 am

Erica, I agree with your post and I’m glad to see the discussion opened up beyond the simply dichotomy of whether not intervention is good or bad. We need a better understanding of the conditions under which various types of the use of force are effective, not effective, or harmful in controlling and mitigating violence; more extensive understanding of the internal dynamics and patterns of violence in particular conflicts; and, how legitimacy and legal elements influence outcomes. These are inherently difficult research questions because of the complexity of the conflicts involved, but also because we are often dealing extensively in the realm of counterfactuals — in cases of intervention, what would have happened in the absence of such intervention (Kosovo, Libya) or in cases of non-intervention, what would have happened if there had been an intervention (Rwanda). I think Fortna, Howard, Walter, Doyle and Sambanis are all good starts — and I am generally persuaded about the importance of the conditions they specify, but we clearly need more.

Erica Chenoweth August 28, 2013 at 2:56 pm

Jon: agreed! We especially need a better baseline understanding to deal with the counterfactual–did these interventions improve the situation relative to what would have happened if there had been none? That is the key question that, to my mind, no one has satisfactorily answered.

TGary August 30, 2013 at 8:04 am

I may be misreading the dialogue, but my understanding treats Chapter VI operations (peacekeeping) very differently from Chapter VII (peace-enforcement).

The first one, UNSCR 82 on Korea, is the prototypical breach-of-peace enforcement model that was probably envisioned by UN Charter writers. Later Resolutions, on Afghanistan, Gulf War, and Libya made the issues much more muddy, but were still premised on a peace-enforcement model where at least one of the belligerents did not agree to the thrid-party intervention.

UN Forces in Golan Heights, Kashmir, Sinai, and the like seem more appropriately placed under Chapter VI authority since they have the consent of the belligerents to be present, conceivably minimizing the likelihood of either side seeking mission failure.

In our blending of the language of “peacekeeping” with “peace-enforcement” operations, we may be making more difficult the measurement of the efficacy of each.

Comments on this entry are closed.

Previous post:

Next post: