Conflict&Security

A Lost Chance for Peace The Bicesse Accords in Virginia Page Fortna

In 1991, the combatants in Angola’s longstanding civil Virginia Page Fortna is Assistant war signed a peace agreement. But the country was soon Professor of Political back at war in a conflict that would grind on for another Science at Columbia University. She is cur- decade. Three particular problems in the peace process rently a Visiting Schol- led to the failure of Angola’s first “best chance for ar at the American 1 Academy of Arts and p e a c e . ” First, the peacekeeping mission was extre m e l y Sciences. limited, partly because the international community hoped to keep peace as cheaply as possible. Second, issues of “spoiler” management were handled poorly due to the assumption that União Nacional para a Independencia Total de Angola (UN I TA) would win the elections and that the Movimento Popular de Libertacão de Angola (MPLA) would not be able to contest the win by forc e . T h i rd, elections were held before demobilization, era d i- cating any incentive that the electoral loser, Jonas Savim- bi, had to end the war, thereby squandering the interna- tional community’s only effective leverage. This combi- nation of mistakes proved disastrous for Angola and added substantial difficulty to subsequent attempts to achieve peace. Eleven years later, the international com- munity and the Angolan parties should reflect upon the failed peace of the past in order to take advantage of Sav- imbi’s death and further the peace process today.

Winter/Spring 2003 [7 3 ] A LOST CHANCE FOR PEAC E

Background. Co n t e m p o ra r y An g o l a T he Peace Agreement . Th re e has never known stable peace; civil war tenets comprised the Bicesse Accords: a erupted on the heels of the war of liber- ce a s e - f i r e supervised jointly by the two ation from the Portuguese. What started Angolan parties; the demobilization of as a three-way fight among rival national- MPLA and UNI T A forces in order to ist groups ended as a war between the cr eate an integrated national army; and governing MPLA, led since 1979 by Pres - multi-party elections. A Joint Pol i t i c a l - ident José Eduardo dos Santos, and Military Commission (CCPM) was UNI T A, led by Jonas Savimbi. The con- established to oversee the peace process. flict was primarily a struggle for control It consisted of rep r esentatives from the of the state and the country’s rich oil, MPLA and UNI T A, with the UN and the diamond, and mineral res o u r ces, rat h e r “Troika”—Portugal, the United States, than an ethnic or identity conflict.2 Th e and the Soviet Union—participating as conflict was expressed in ideo- observers. Unlike the arrangements in logical terms: the MPLA espoused social- many other peacekeeping cases, imple- ism, and was backed by both the Soviet mentation of the Accords was primari- Union and Cuba, while UNI T A took an ly the responsibility of the belligerents. anti-communist line, winning support The UN peacekeeping observation mis- from the United States and South sion, UNAVEM II, was mandated only Af r i c a . 3 In 1988, as Cold War tensions to monitor the progress of the Angolan waned, Cuba and agreed to process. Although the UN was eventu- wi t h d r aw troops and cease military activ- ally called on to play a more substantial ity in Angola. Within a few years, the role, it was severely limited by having to stalemate on the battlefield, U.S. and work through the unwieldy CCPM. Soviet pres s u r e, and Portuguese media- The Bicesse agreement was a tall tion induced dos Santos and Savimbi to order on a very tight timetable, allow- sign the Accordos de Paz para Angola, ing only 16 months between the signing often called the Bicesse Accords . 4 of the Accords and elections. In that When the Accords were signed in May time the country, whose infrastructure 1991, Angola was not viewed as a partic- was in shambles, had to demobilize ularly difficult peacekeeping case. This 150,000 troops, form a new army, overly optimistic assessment was a res u l t extend government authority to the of the previous success of the United o n e - t h i rd of the country held by Nations during Cuba’s withdrawal from UNITA, and implement voter registra- Angola, external support for peace by tion and polling. Furthermore, major the superpowers whose rivalry had dri- components of the agreement had not ven the war, and the non-ethnic, non- been fully resolved. A telling section of secessionist character of the war. As the Accords entitled “Concepts for Angola was rich in natural res o u r ces, it Issues Still Pending between the Gov- was hoped that they would pay for much ernment and UN I TA” enumera t e d of the peace implementation. However, demobilization logistics, specifics of this initial underestimation of the situa- creating a neutral police force, and tion coupled with the failure of Angola’s details about who would setup the first real chance for peace, made subse- army. Quarrels over these details, espe- quent attempts much more difficult. cially the creation of the police force,

[7 4] Georgetown Journal of International Affairs F O R T NA Conflict&Security

quickly undermined trust and cooper- ther the United States nor the Soviet ation within the CCPM.5 Me a n w h i l e , Union viewed Angola as a high priori- technical difficulties, resulting from ty. The primary work of maintaining woefully inadequate time and res o u rc e s , peace fell to the UN, even though it provided excuses for deliberate non- had played a minimal role in the nego- compliance by both parties. tiations. Both the peace accords and

The Bicesse Agreement was a tall orde r on a very tight timetable.

The president was to be elected by a the UN Security Council gave majority of the people, while the Nation- UNAVEM II a restricted mandate: al Assembly was to be elected through a strictly to observe the work of the system of proportional rep re s e n t a t i o n . CCPM, not to keep peace dire c t l y . The Accords established a National Elec- Because the belligerents had protested to r al Commission consisting of UNI T A against extensive outside involvement, and MPLA rep r esentatives to oversee the UNAVEM’s mandate was overly limited. elections. UNAVEM was later asked to The MPLA was particularly touchy about help monitor these elections. so v e r eignty issues and presumably feared In retrospect, many have argued that a a pro-UNI T A bias on the part of the major flaw of the Bicesse Accords was the U.S.-dominated international commu- “w i n n e r - t a k e-all” presidential contest, as nity. UN I TA, on the other hand, opposed to a power-sharing arran g e - seemed to want to avoid a strong pres - me n t . 6 Both sides, however, saw the elec- ence that would enforce the demobiliza- tions as a chance to win at the ballot box tion provisions of the agreement. The what they could not win on the battle- international community, feeling the field: legitimate power to rule Angola st r ain that new missions would impose, alone. Neither party desired to share was all too happy with preserving peace as power, and the United States, believing inexpensively as possible.8 its proxy would win, did not push for it. The res o u r ces provided for imple- Only when it appeared that UNI T A menting the Bicesse Accords were might lose did outside mediators begin to in c r edibly limited given Angola’s vast size pr ess for power-sharing arran g e m e n t s . 7 and the inaccessibility of most of the country. The initial 18-month allocation Implementing Bicesse. External for UNAVEM II was $132.3 million for a involvement prolonged the war for country of 12 million people.9 The mis- years, but when it came time to imple- sion consisted of 350 military observers, ment the peace, Angolans were left 126 police observers, an electoral divi- largely to their own devices. While sion of 400 poll monitors, and a civilian there was support in principle from the staff of 242.10 Co m p a r e this with 4,650 Troika, its role on the ground was lim- troops, 1,500 police monitors, an elec- ited. Portugal was less active in imple- to r al team of 900, and the $368.5 mil- mentation than in mediation, and nei- lion spent in about 1 year in , a

Winter/Spring 2003 [7 5] A LOST CHANCE FOR PEAC E country with 1/8 the population of Estimates indicate that the MPLA kept Angola. This lack of spending ref l e c t e d 10,000 troops mobilized, while the reluctance of those paying the bill UNITA kept 30,000.12 Two days before rather than a genuine or realistic assess- the elections, 65 percent of MPLA’s ment of what Angola needed. troops had demobilized, but only 26 Implementation was behind schedule p e rcent of UN I TA’s had. UN I TA from the beginning. Beset with logistical tipped the security equation further in and technical problems, the process was its favor by sending its personnel to often deadlocked by disagre e m e n t s areas formerly under MPLA control between the MPLA and UNI T A and fur- and hiding arms caches to support an ther disrupted by the unwieldy machin- offensive. While intentions are difficult ery of the CCPM. Lack of accurate infor- to gauge ex ante, UNITA’s actions should mation on the number of troops in each have set off alarm bells. Why didn’t the army hampered both planning for international community react earlier demobilization and its verification. The to deal with UNITA?13 difficulties of deploying soldiers to The implementers’ assessment of pol- remote areas delayed the cantonment of itics in Angola was based on the Cold War the military forces. Lack of adequate assumption that Savimbi would win in a food, tents, and supplies made it difficult de m o c r atic election. Believing its own to keep troops assembled. Many MPLA rhetoric, the United States justified sup- troops simply left the camps. Secure stor- porting UN I TA “freedom fighters” age of weapons was also a problem in struggling for democracy, and assumed remote areas, where most buildings were that, without strong Soviet backing, the gr ass huts. Demobilization teams, sprea d MPLA would simply wither away.1 4 thin and without essential res o u rc e s , Under this assumption, there was no rea - delayed the creation of the new army. A son to question Savimbi’s intent: he small, purely symbolic force was hastily would have no reason to upset a peace sworn in the day before elections.11 process that would put him in power. In Even with more res o u r ces, time, and 1991, with the Soviet Union disintegrat - planning, technical and logistical delays ing, American views and assumptions could not have been surmounted in became the dominant influence shaping implementing the Accords. The real the peace process. Meanwhile, a combi- problem was deliberate non-compli- nation of limited res o u r ces for gathering ance. Technical difficulties provided information, including a lack of Por - cover for deception, although it soon t u g u e s e - s p e a kers, and the desire to became apparent that both sides were remain impartial made it difficult for holding troops in reserve to contest an UNAVEM to brand Savimbi a spoiler. e l e c t o ral defeat or in case the war Fur t h e r m o r e, growing evidence of resumed. Non-compliance by both UNI T A’s plans was also downplayed or parties fueled the other’s suspicions, ig n o r ed. There were plenty of indicators but UNITA was the worse offender. that the militaries kept forces in res e r v e , The government had trouble getting especially those of UNI T A. UNAVE M and keeping troops at the new army was aware of the discrepancy in demobi- camps, in part because UNITA troops lization rates, and of claims by UNI T A remained mobilized and ready to fight. defectors of 20,000 troops and arms

[ 7 6] Georgetown Journal of International Affairs F O R T NA Conflict&Security caches hidden in the bush.15 Be f o r e the the cease-fire or to accept defeat. Ye t , elections Savimbi stated that “if UNI T A rather than taking action against Sav- does not win, then it means frau d . ” 16 It imbi, UNAVEM reduced its troop was clear that he planned to contest the s t rength and hunke red down in Luan- elections by force if he lost.17 But the d a .1 9 By October, Angola had plunged international community did nothing to back into a full-scale war. While the alter this incentive to use force . elections were the only part of the Meanwhile, the electoral exerc i s e peace implementation that ra n in Angola was extremely well moni- smoothly and on schedule, they clearly t o red and implemented. Given the should have been delayed to ensure short timetable and logistical obsta- post-election stability. cles, both the re g i s t ration of about Although they could have postponed 4.86 million Angolans (92 percent of elections, both Angolan parties were anx- those eligible) and the polling itself ious to hold a contest each thought it w e re minor miracles. Elections took would win. Third parties did not pres s place without major incident on Sep- for an extension, nor was strong pres s u r e tember 29-30, 1992, with 91 perc e n t applied to either side for failing to demo- of those re g i s t e red voting.1 8 bilize. Both UNAVEM and the Tro i k a Because priority was placed on elec- we r e suffering from the “Tin k erbell syn- tions rather than on demobilization or drome,” as one British diplomat called it, other aspects of the peace agree m e n t , hoping “that if they believed hard enough implementers stuck to the original dead- in the process and avoided criticizing the line for elections rather than the parties…the country’s political tran s i t i o n sequence on which the timetable was would be successful.”20 The United States based: complete demobilization with a had little desire to point out UNI TA ’ s single integrated army in place before the non-compliance, while the UN pref e r re d election. The purely symbolic creation of making “even-handed” criticisms to a new army the day before elections is avoid the perception of partiality.21 Th i s indicative of the extent to which wishful combination of wishful thinking, the thinking and a rigid timetable for elec- de s i r e to be seen as impartial, and U.S. tions took precedence over the substance bias, added up to a peacekeeping mission of what was needed for stable peace. that failed to condemn flagrant violations Th e MPLA won the legislative elec- of the peace agreement, which resulted in tions, with dos Santos taking just outright aggres s i o n . under 50 percent of the pre s i d e n t i a l In retrospect, the policy looks like vote and Savimbi 40 percent. Electora l appeasement. The international com- law re q u i red a second round if no one munity tried to cajole Savimbi into a achieved 50 percent, but it was never series of negotiations between October held. As soon as he realized he had 1992 and March 1993. The United lost, Savimbi took up arms. His adher- States and the UN eventually imposed ence to the cease-fire had hinged on sanctions on UNI T A, but only after Sav- the belief that UN I TA would win the imbi had completely destroyed the peace elections. Knowing the outcome, and process and overrun two-thirds of the with his army still intact, there was no country. Sanctions eventually helped the reason for him to continue to re s p e c t MPLA reverse UNI T A’s military gains.

Winter/Spring 2003 [7 7] A LOST CHANCE FOR PEAC E

Even so, mediators held out the prospect insisting on demobilization before of a power-sharing agreement to induce elections, as was done in Mozambique, Savimbi back to the negotiating table. the international community would With the military tide turning against have given the belligerents a choice: dis- him, the promise of a role in the gov- arm and participate in elections, or face ernment coerced Savimbi into signing international condemnation for failure the Lusaka Protocol in 1994. The Lusa- to demobilize. Instead they were allowed ka agreement “fixed” a number of the to have both a shot at winning the elec- problems of the Bicesse Accords, mov- tion, and a military hedge. ing away from winner-takes-all elections Be f o r e the elections, peacekeepers had and deploying a much stronger UN le v e r age. Had the United States been fo r ce. If this agreement had been imple- willing to pres s u r e UNI T A, this leverag e mented in 1991 it might have worked . would have been improved. Each side But the failure of Bicesse made peace in had an incentive to cooperate since both 1994 much more difficult to obtain, and thought they could win legitimate power. within a few years the Lusaka peace Although Savimbi would probably still ag r eement fell apart. have thrown a tantrum upon losing the vote, had he been pushed to demobilize, Conclusion. Blame for the res u m p - he would not have been in a position to ti on of war in 1992 rests squarely on destroy the peace so easily. Instead, once Jonas Savimbi’s shoulders, but the Savimbi knew the outcome of the elec- international community could have tions, he had nothing to lose by fighting. managed the situation much more effec- It became impossible to induce UNI T A tively. The pathetic peacekeeping mis- to demobilize peacefully or respect the sion, biased assumptions about Savim- ce a s e - f i r e. By the time a stronger peace- bi’s electoral prospects, and failure to tie keeping force and a power-sharing elections to full demobilization squan- ag r eement were arranged in 1994, it was de r ed Angola’s best chance for peace. too late. The failure of the Bicesse Some would argue that the subsequent Ac c o r ds condemned Angola to another fa i l u r e of the Lusaka Accords shows that ten years of war. The war would not end Savimbi was an intransigent spoiler; that until Savimbi’s death and UNI TA ’ s nothing short of his electoral victory, a defeat in 2002. “Liberia solution,” would have crea t e d While generalizing from a single spe- pe a c e . 22 Nevertheless, things could have cific case is imperfect, the failure of the been improved in the first round. Had Bicesse accords suggests several broader th e r e been a serious verification pres - policy lessons. First, attempts at under- ence with more res o u r ces, it would have funding and overly-hurrying peace are been harder for the parties not to li k ely to backfire. The costs of failure, demobilize and the results of the first both the physical costs of later attempts at round of Accords could have been quite peace in Angola and the damage to UN di f f e r ent. There would still have been cr edibility, outweighed the money saved an incentive for both sides to hedge by sending a skimpy mission. The costs to their bets by maintaining mobilized Angolans of this failure were devastating. f o rces, but their non-compliance A broader lesson can be derived would have been much more evident. By from the Cold War legacy that blinded

[7 8 ] Georgetown Journal of International Affairs F O R T NA Conflict&Security the mission to the spoiler problem with Finally, elections should not be held Savimbi. Strict impartiality may not be be f o r e demobilization is complete. Elec- necessary for successful peacekeeping; tions held when both sides maintain the in fact the UN’s attempt to maintain a ability to fight is a recipe for disaster. semblance of impartiality at all costs They remove the peacekeeper’s most probably contributed to the problem, effective leverage and leave the loser little but initial biases toward the parties incentive not to fight. A bigger, more prohibited objective and re a l i s t i c adequately financed peacekeeping mis- assessments of the potential risks to sion with a realistic timeline, a more peace. Assumptions about which side objective assessment of the incentives fac- will win elections, and how the losing ing the parties, and an insistence that side will react need to be tested careful- demobilization precede elections might ly. Anticipating potential spoiler prob- well have created stable peace in Angola. lems ahead of time is crucial to tackling Instead, the country has suffered war for them effectively. the past decade. N OT E S 1 Paul Hare’s book on the Lusaka Accords of 8 Various officials at the U.S. Department of 1994 is entitled Angola's Last Best Chance for Pe a c e (Wa s h - State, Washington DC, 6 August 1998 and 11 ington, DC: United States Institute of Pe a c e , December 1991. Also see UN Document S/22627, 1 9 9 8 ) . 20 May 1991. 2 William Minter, 's Contras: An Inquiry into 9 UN Document S/22627 Addendum 1, 29 May the Roots of War in Angola and Mozambique ( J o h a n n e s b u r g : 1991. All told UNAVEM II cost $175.8 million over Wi t w a t e r s rand University Press, 1994), 103-5. 4 years. United Nations, The Blue Helmets: A Review of 3 Fernando Andresen Guimarães, The Origins of the United Nations Pe a c e - K e e p i n g, 3rd ed. (New York: Unit- : Foreign Intervention and Domestic Po l i t i c a l ed Nations, 1996). C o n f l i c t (New York: St. Martin's Press, 1998). 10 United Nations, The Blue Helmets, 254. 4 José Manuel Durão Barroso, "CSIS Study 11 Vines, 17; Venâncio, 34. Group on Angola," (Washington, DC: Pre s e n t a t i o n 12 Various officials at the U.S. Department of at the Center for Strategic and International Stud- State, Washington DC, interview with author, 6 ies, 20 February, 1991). August 1998; Vines, 17. 5 Margaret Joan Anstee, Orphan of the Cold War: The 13 Stephen John Stedman, "Spoiler Problems in inside Story of the Collapse of the Angolan Peace Pr o c e s s, 1 9 9 2 - Peace Processes," International Security 22, no. 2 9 3 (London: Macmillan, 1996), 69-71; Alex Vi n e s , ( 1 9 9 7 ) . One Hand Tied: Angola and the UN, CIIR Briefing Pa p e r s 14 Various officials at the U.S. Department of (London: Catholic Institute for International State and the Washington Office on Africa, inter- Relations, 1993), 5; Moisés Venâncio, The United view with author, Washington, DC, 6 August 1998. Nations, Peace and Transition: Lessons from Angola, vol. 3, See also Minter, "Sorting out Lessons". Luminar Papers (Lisbon: Instituto de Estudos 15 Anstee, 52-3. E s t ratégicos e Internacionais, 1994), 35. 16 Quoted in Anstee, 151. 6 See for example, Anthony W. Pe re i ra, "The 17 Victoria Brittain, Death of Dignity: Angola's Civil Neglected Tragedy: The Return to War in Angola, Wa r (London: Pluto Press, 1998), 49. 1992-3," The Journal of Modern African Studies 32, no. 1 18 United Nations, 243. (1994): 1-28. For an interesting critique of power- 19 Anstee, 235. sharing see William M. Minter, "Sorting out 20 Quoted in Venâncio, 46. Lessons from the Angolan Elections," (Presented at 21 Minter, Apartheid's Contras, 9; Anstee, 377; the African Studies Association, November 1994), UN, 246. See also, Vladimír Krpka, "Pe a c e ke e p i n g 4, 9. in Angola (UNAVEM I and II)," International Pe a c e- 7 Various officials at the United States Depart- k e e p i n g 4, no. 1 (1997): 90. ment of State and United States-Angola Chamber 22 In 1997, Liberians elected rebel leader of Commerce, interview by author, Wa s h i n g t o n , Charles Taylor president to prevent him from con- DC, 6 August 1998. Also see Anstee, 149. tinuing the civil war.

Winter/Spring 2003 [7 9]