PART THREE

Case Studies

The quantitative analysis presented in chapters 4 and 5 found that third-party security guarantees and power-sharing pacts significantly increase the likelihood that a civil war will end in a successfully implemented peace settlement. Statistical analysis, however, only confirms that these factors are consistently associated with the outcome of civil wars. It says nothing about why third-party security guarantees and power-sharing pacts tend to cause combatants to sign and implement such treaties, or if the fear of post-treaty exploitation really drives behavior. A more detailed historical analysis of individual cases is needed to uncover whether the desire to make credible commitments to a peace treaty is the mechanism causing combatants to seek outside guarantees and power-sharing pacts.

The following two chapters trace the day-to-day negotiating processes in Zimbabwe and Rwanda. I chose these two cases from the seventy-two civil wars listed in Appendix 1 because they were as similar as possible on all the key independent variables except third-party security guarantees. Zimbabwe’s and Rwanda’s wars were neither particularly long nor short conflicts. The median duration of wars listed in appendix 1 was seventy-four months; in Zimbabwe and Rwanda they lasted eighty-four and fifty-one months respectively. Both wars produced less than the median number of battle deaths (ninety-two thousand per war).¹ Neither war experienced a military stalemate. In fact, at the time of negotiations, the rebels in both cases were advancing on the battlefield and were expected to win the war given additional time. The rebels in both cases fought for limited rather than total goals and were not interested in obtaining territorial autonomy. In Zimbabwe, the Patriotic Front demanded one-man, one-vote majority rule for the country’s predominantly black population, while the Rwandan rebels demanded multiparty rule and the repatriation of Tutsi refugees. Both wars had a clear ethnic component, with Zimbabwe’s war fought between blacks and whites, and Rwanda’s fought between Tutsis and Hutus. Combatants in both wars initiated numerous formal attempts to negotiate during which mediators were present, and both signed comprehensive peace plans that included specific quotas of power. The only causal factor that differed between the two cases was the presence of a third-party security guarantee. Zimbabwe’s Lancaster House agreement was underwritten by a third-party security guarantee; Rwanda’s Arusha accords were not.

Seven other civil wars were considered for case study analysis but were passed over in favor of the two cases mentioned above. Tajikistan, Liberia, Uganda, Chad, and Burundi met the selection criteria and would have worked equally well. I chose to investigate Zimbabwe and Rwanda, however, for three reasons. First, the peace processes in Zimbabwe and Rwanda have been documented to a much greater extent than any of the four alternate cases, allowing for richer and more comprehensive analyses. Second, the civil wars in Zimbabwe and Rwanda have both ended, allowing me to analyze the peace process in its entirety. Finally, there is a pressing desire among policymakers and the public to understand how a peace settlement signed in Rwanda could have ended in genocide. Rwandans paid a far higher price for the failure of the Arusha accords than citizens in any of the other civil wars under consideration. Two additional wars (Angola and Guatemala) would have worked well as a pair but were discounted because negotiations in both those wars occurred too recently to collect sufficient historical data.

I analyzed these cases with two goals in mind. The first was to tease out the causal links between third-party security guarantees, power-sharing pacts, and the resolution of civil wars, as discussed above. The second and perhaps more important goal was to ensure that other factors equally critical to our understanding of settlements not be overlooked. Have any essential variables been omitted from the analysis?

I used three strategies to meet these goals. First, I paid careful attention to any condition or event that appeared to play an important role in any stage of the peace process even if this factor had not been considered previously. It is quite likely, for example, that combatants are sensitive to a variety of costs in a war beyond its duration or battle deaths, or that other economic, political, or international variables better explain observed behavior.² A closer look at individual cases, therefore, serves to inform both the author and the reader of other potentially significant factors.

The second strategy was to carefully track when and how decisions to cooperate were made. If the success of negotiations really does rest on resolving difficult issues of post-treaty security, then one expects to observe the same three patterns discussed in chapter 4. First, combatants should have the greatest difficulty resolving the issue of post-treaty security, and this issue should be the last on the table before combatants sign or reject an agreement. In fact, since it will be easier for combatants to resolve the causes of war than to credibly commit to these solutions, successful negotiations should experience a lag time between the point at which a deal is reached on these issues and a settlement is signed. This is the period when combatants are likely to seek reassurance and attempt to design credible commitments to the terms. Second, the combatants should foresee the risks and dangers of implementation and actively seek outside security guarantees to neutralize their negative effects. Third, the timing of implementation should correspond to the arrival and departure of outside monitors or enforcers, thus revealing a direct connection between safety guarantees and the decision to cooperate.

The credible commitment theory would be disconfirmed if (1) security issues (such as the disposition of forces, army reorganization, and military integration and training) did not dominate negotiations, (2) combatants were willing to sign and implement peace settlement without an outside guarantor, or (3) treaty implementation was unrelated to the arrival or departure of a third party. Any one of these patterns would suggest that security fears and enforcement were not the greatest barriers to successful cooperation and would indicate that other more important mechanisms drove behavior.

We will see in chapters 6 and 7 how much of the negotiations are dominated by attempts by both sides to protect themselves against post-treaty exploitation. Combatants in both wars voiced the same fears and insecurity for any upcoming transition period. They either hesitated or refused to sign a peace treaty until sufficient arrangements were made to guarantee their safety during the transition, and implementation of the terms of each agreement coincided exactly with the arrival (or absence) of the third party. In short, a direct causal link does appear to exist between combatants’ fear of post-treaty exploitation and their demand for outside guarantees and power-sharing pacts.

¹ The war in Rwanda did produce an exceptionally large number of deaths in 1994, but this was after the Arusha peace accords were signed.

² Elisabeth Wood, for example, argues that El Salvador ended in a negotiated settlement because the conflict was fought between groups that were economically interdependent, a variable not included in the quantitative analysis in chapter 4. See Elisabeth Jean Wood, Forging Democracy from Below: Insurgent Transitions in South Africa and El Salvador (Cambridge: Cambridge University Press, 2000).