In lieu of an abstract, here is a brief excerpt of the content:

  • Civil War Settlements and the Prospects for Peace
  • Laurie Nathan (bio) and Monica Duffy Toft (bio)

To the Editors (Laurie Nathan writes)

In "Ending Civil Wars: A Case for Rebel Victory?" Monica Duffy Toft questions whether policymakers are correct to have a strong preference for terminating civil wars through negotiated settlements.1 Her main endeavor is to undertake a statistical analysis that compares the effect of negotiated settlements, military victories, and ceasefires/stalemates on war recurrence and on the state's level of democracy and economic growth. With respect to war recurrence, which is the focus of this letter, she finds that negotiated settlements are largely ineffective, that civil wars ending in military victory by one side are less likely to recur, and that rebel victories produce the most durable settlements (pp. 7–8). Toft's purpose is not only to understand these phenomena but also to provide policymakers with guidance (p. 22). She recommends that third parties should pay greater attention to security-sector reform (SSR) during negotiations, leading to settlements that can credibly guarantee both benefits from cooperation and harm from defection; failing that, support in pursuit of victory, especially rebel victory, may be a worthy objective for policymakers (p. 36).

Given that the stakes and risks of external intervention in a civil war are high, it is worth reflecting on the merits of Toft's results and recommendations. I suggest below that her findings on war recurrence and the stability of rebel victories are not dependable and helpful for policy purposes. Advice to policymakers on war termination ought to be based on case studies that explore causal relationships and examine the role actually played by policymakers in seeking to end civil wars.

Contesting the Statistical Findings

Toft finds that in the period 1940–2000, war recurrence was 22 percent after negotiated settlements (5 of 23 cases), 12 percent when wars ended in military victory (10 of 81 cases), and only 6 percent when wars ended in rebel victory (2 of 33 cases) (p. 16). These findings are questionable because they rest on contestable judgments about some of the cases. A small number of classification changes based on plausible interpretations that differ from those of Toft can have significant statistical implications.

In this regard, Toft notes that scholars disagree on "matters of interpretation about how to aggregate or disaggregate certain conflicts."2 She points out that Michael Doyle [End Page 202] and Nicholas Sambanis consider Afghanistan to have had three separate wars from 1978 to 2001; Guatemala to have undergone separate civil wars in 1966–72 and 1978–94; Nicaragua to have experienced separate conflicts with the Sandinista and Contra uprisings; and the Democratic Republic of Congo (DRC) to have had separate wars in 1996 and 1998.3 Toft combines the conflicts in each of these countries on the grounds that their combatants were essentially the same, they were fighting over the same issues, and the war could not be considered "terminated" in the interim.4

What is the statistical effect of these different interpretations? The rate of war recurrence after rebel victories would rise from 6 percent to 16 percent (6 of 37 cases) if the following four cases, omitted from Toft's data set as separate conflicts, were included and coded as recurring wars: the leftist victory in Afghanistan in 1978; the mujahideen victory in Afghanistan in 1992; the Sandinista victory in 1979; and the Kabila victory in the DRC in 1996. Furthermore, whereas Toft codes the Liberian civil war in 1989–97 as a rebel victory and as nonrecurring,5 the subsequent war in 2002 can plausibly be seen as a recurrence of the earlier war.6 A coding change in this respect would raise the war recurrence rate of rebel victories to 19 percent (7 of 37 cases), which is close to that of negotiated settlements.

For policymakers the crucial point is not whether Toft or some other scholar is correct on the disputed cases. The point is rather that statistical findings that can fluctuate notably as a result of a few legitimate differences of opinion are not a dependable guide to decisionmaking on the most appropriate means of ending civil wars.

It is...

pdf

Share