The flurry of democratization that the world has seen since the late 1980s has engendered a running debate--some of it conducted in the pages of this journal--about how fledgling democracies can best design constitutions that give rise to political harmony and stability. 1 In the context of this debate about "constitutional engineering," electoral-system design is increasingly being recognized as a key lever that can be used to promote political accommodation and stability in ethnically divided societies.
Most scholars agree that the choice of an electoral system has powerful political consequences; where they differ is in their basic prescriptions for newly democratizing states. The evidence from the emerging democracies of southern Africa strongly suggests that divided societies need proportional representation (PR) rather than plurality elections, and a parliamentary rather than a presidential form of government. A simple parliamentary-PR system, however, is not enough: these fragile democracies are better served by a type of PR that maximizes the geographic representativeness of MPs, as well as their accountability to the voters. Although appropriate electoral laws are insufficient to ensure stability and good governance in divided societies, poorly designed laws can entrench societal divisions and exacerbate preexisting conflict.
The prevailing academic wind is clearly blowing in favor of PR--and against plurality elections--for ethnically divided states. As W. Arthur Lewis asserted in his seminal Whidden Lecture of 1965, "The surest [End Page 86] way to kill the idea of democracy in a plural society is to adopt the Anglo-American system of first-past-the-post." According to Lewis, the vagaries of plurality elections would produce racially exclusive and geographically parochial governments that would exploit a "mandate" from a plurality of the electorate in order to discriminate systematically against minorities: "If you belong to a minority in a new state, and are being asked to accept parliamentary democracy, you can hardly build much faith in the system if you win 30 percent of those votes and get only 20 percent of the seats, or no seats at all. If minorities are to accept Parliament, they must be adequately represented in Parliament." 2
Supporting this view is Arend Lijphart, who argues not only that divided societies need PR in order to protect minority interests, but that PR systems (in conjunction with parliamentarism) "almost invariably post the best records, particularly with respect to representation . . . voter participation, and control of unemployment." 3 Timothy Sisk also identifies a "consensus on PR" in the debate over electoral-system design for South Africa, a consensus born of that country's overwhelming need for governmental structures that facilitate political accommodation: "Threatened as they are by the centrifugal politics of extremist outbidding, divided societies like South Africa need institutions that pave the way for moderation and compromise." 4
A number of prominent scholars, however, come down on the other side of the debate. Guy Lardeyret argues that the best way to counteract the tendency for electoral competition to follow ethnic lines is to "oblige members of each group to run against one another on (transethnic) political and ideological grounds in single-member districts," while "the worst way is to adopt PR, which tends to reproduce ethnic cleavages in the legislature." Lardeyret believes that South Africa's prospects will be grim without the development of "big and moderate multiethnic parties" and posits that plurality elections would be most likely to give rise to such parties. 5 His reasoning is echoed by Donald Horowitz, who argues that "vote-pooling," in which parties are encouraged to appeal across societal divides, "lies at the heart of intergroup compromise in severely divided societies." 6 Vote-pooling, argues Horowitz, is best achieved through the "alternative-vote" electoral system, a modification of single-member-district plurality that involves reallocation of voters' second and subsequent preferences until a candidate attains a majority.
In the last five years, southern Africa has been the scene of constitutional changes unmatched since the end of European colonialism in the 1960s. Namibia became a multiparty democracy in 1989, with independence from South Africa formally declared in 1990; Zambia moved to a democratic system in 1991. South Africa and Malawi [End Page 87] completed their remarkable democratic transitions in 1994, and Mozambique held what were generally considered to be free and fair elections at the end of that same year. Of the remaining southern African states, Zimbabwe, Botswana, and Lesotho have experienced varying degrees of political competition, while Angola has taken tentative steps toward a new constitutional arrangement and an end to civil strife. Only President Mobutu Sese Seko of Zaire continues to buck the trend of failing authoritarian rulers, maintaining power despite his legalization of opposition parties in 1990.
At present, the most "electorally democratic" states in southern Africa include Malawi, Namibia, South Africa, and Zambia. Zimbabwe has also had relatively free, fair, and regular elections since 1980, but electoral competition has been truncated since 1987. Of these five, Malawi, Zimbabwe, and Zambia can be categorized as presidential-plurality systems under Lijphart's "basic types of democracy," while South Africa is a parliamentary-PR system. 7 President Nelson Mandela's title is actually a misnomer: even though the 1994 elections for the South African National Assembly were strongly reminiscent of a presidential campaign, Mandela was not directly elected, and has classic prime-ministerial powers derived from his leadership of the largest party in parliament. Namibia began its democratic life in a similar situation, with "President" Sam Nujoma leading the nation, but moved to a genuine presidential-PR system in December 1994.
The argument that plurality elections exacerbate regional and ethnic polarization draws considerable strength from the experiences of the fledgling democracies of southern Africa.
Malawi. The results of Malawi's May 1994 multiparty elections showed each of the three main political parties to be overwhelmingly dominant in its core region, and exceedingly weak in the remaining two regions. The number of seats that each party won in a given region, however, did not closely reflect the percentage of votes that it polled there. The Alliance for Democracy (AFORD), led by Chakufwa Chihana, won every seat in the northern region even though non-AFORD candidates managed to capture nearly 15 percent of the vote there. The Malawi Congress Party (MCP) of then-President Hastings Kamuzu Banda won fully 75 percent of the seats in the central region while receiving slightly more than 60 percent of the vote. In the southern region, the United Democratic Front (UDF) of successful presidential candidate Bakili Muluzi won 95 percent of the seats with just three-quarters of the popular vote. The overall proportion of seats that each party gained in the National Assembly was surprisingly close to the percentage of the popular vote that it captured nationwide, but the distribution of seats created a false picture of homogeneous regional bastions that could derail efforts to build political unity across ethnic groups. [End Page 88]
Zimbabwe and Zambia. The virtual disappearance of a vigorous political opposition in Zimbabwe and Zambia in recent years provides ample evidence of the threat that plurality elections can pose to the emergence of new parties in fledgling democracies. Jonathan Moyo maintains that the danger to democracy in Zimbabwe was revealed most clearly in 1990, when the opposition parties were rewarded with only two parliamentary seats despite polling 20 percent of the popular vote. According to Moyo's calculations, had the elections been conducted under a list form of PR, the opposition could have expected to win 23 parliamentary seats--a far more solid foundation from which to critique government policy. 8 In Zambia's 1991 plurality elections, Frederick Chiluba's Movement for Multiparty Democracy (MMD) won 125 of the 150 parliamentary seats--enough to rule unhindered as well as to amend the constitution unilaterally. In the words of journalist Jowie Mwiinga, "there has been little organized pressure on the President to clean up his government, thanks to the absence of either a meaningful opposition or an independent press." 9
South Africa. Use of a plurality system in the 1994 elections for the South African National Assembly would have had a number of negative consequences. 10 The immediate result would have been a denial of parliamentary representation to minority interests. Despite their geographic pockets of electoral support, the Freedom Front (which won 9 seats in the actual election), Democratic Party (7 seats), Pan-Africanist Congress (5 seats), and African Christian Democratic Party (2 seats) would have failed to win a single parliamentary seat. Although all these parties together constitute only 6 percent of the new South African National Assembly, they have an importance within the structures of government that is disproportionate to their numerical strength. General Constand Viljoen's Freedom Front represents a highly volatile Afrikaner-heartland constituency that could easily have fallen into the hands of white-supremacist demagogues such as Eugene Terre'blanche had its representatives been shut out of the political process. Similarly, the predominantly Xhosa Pan-Africanist Congress of Azania (PAC) is a party torn between the ballot box and the AK-47. Had the PAC been denied parliamentary representation, its armed terrorist wing might have gained the upper hand, posing a severe threat to stability in South Africa in the immediate postelection period. The significance of the place obtained in parliament by the small, liberal Democratic Party (DP) is also not to be underestimated. Its leaders were crucial to the crafting of the liberal elements of the interim Constitution, and its presence in the new National Assembly is necessary to ensure that such tolerance remains the official creed of the land.
The list-PR system actually used in the South African elections was probably neither an advantage nor a disadvantage to the midsized National Party (NP) and Inkatha Freedom Party (IFP), for the "national [End Page 89] referendum" nature of the campaign led to a two-party battle between the old and the new: the African National Congress (ANC) versus the IFP in the KwaZulu-Natal province, and the ANC versus the NP in the rest of the country. Given the ethnically homogeneous nature of constituencies and the strong geographic concentrations of popular support in South Africa, the NP would have won only slightly fewer, and the IFP slightly more, seats under a single-member-constituency system than under PR. Yet the anomalies of the Anglo-American electoral system--summed up by Vernon Bogdanor's statement that "the number of seats which a party wins in a general election depends not only upon how many votes the party receives but also upon where these votes are cast"--would have led the ANC to win all of the parliamentary seats in the Northern and Eastern Transvaal, North-West, and Eastern Cape and all but one of the seats in the Orange Free State. 11 In the Western Cape, the NP would have captured over 80 percent of the seats with just over half of the popular vote. Clearly, these results would have exacerbated ethnic and regional polarization, jeopardizing the development of a stable political system in South Africa. Plurality might have done its greatest harm in precluding the emergence of nonracial or smaller minority parties without roots in the old constitutional arrangement.
Proportional Representation in Action
Proportional representation has undoubtedly been an integral part of efforts at power-sharing and ethnic accommodation in both South Africa and Namibia. Yet PR alone would not have ensured multiparty government in South Africa, as a dominant single party could have formed an administration unilaterally had the interim Constitution not exhibited consociational features. 12 What PR did was realize Article 88 of that Constitution, which guaranteed all parties with more than 5 percent of the seats in the National Assembly a proportionate share of the cabinet portfolios. List PR thus helped F.W. de Klerk's NP win 6 of the 27 cabinet portfolios, and the Zulu-based IFP, 3 portfolios. Real sharing of power was ensured when an NP member was named to head the finance ministry and Chief Mangosuthu Buthelezi, head of the IFP, was appointed minister of home affairs. Thus in South Africa PR helped fulfill one of Lijphart's basic principles of consociational democracy--a "government by 'grand coalition,' that is, by a broadly representative coalition of all significant groups." 13
In the Namibian elections of 1989 as in the 1994 South African voting, PR ensured that there were few "wasted votes." In Namibia, 99 percent of the votes cast went toward electing a representative; the figure was even higher in South Africa. This high level of efficiency was due primarily to the low threshold of representation needed to win [End Page 90] a parliamentary seat under the national list-PR system (0.7 percent in Namibia, 0.2 percent in South Africa). Lijphart has argued that, for new democracies, "moderate PR [with its high threshold] and moderate multipartism, as in Germany and Sweden, offer more attractive models than the extreme PR [with its low threshold] and multiparty systems of Italy and the Netherlands," reasoning that a highly fragmented party system would be detrimental to the formation of coalitions and subsequent governmental stability. 14 The evidence from southern Africa, however, suggests that the costs of excluding small but significant parties through a high (say, 5 percent) threshold far outweigh the benefits. Even a 3 percent threshold would have excluded some of the most important political factions from the first South African National Assembly, and would have denied representation to all but the three largest parties in Namibia. In the earliest stages of a new democracy, the need for all significant groups to feel included in the system outweighs concern about the obstacles that may face coalition-builders in subsequently elected parliaments.
Proportional representation has also allowed the South African parliament to be fairly reflective of South African society as a whole. The national lists, which were "closed" in the sense that the electorate was unable to alter or reorder them, allowed parties to present ethnically heterogeneous groups of candidates with anticipated cross-cutting appeal. The resulting National Assembly was 52 percent Black, 32 percent White, 8 percent Indian, and 7 percent Coloured, as compared to an electorate that was estimated to be 73 percent Black, 15 percent White, 3 percent Indian, and 9 percent Coloured. South Africans of Indian extraction fared particularly well, gaining more than 30 MPs and 4 cabinet ministers. Moreover, an influx of female MPs representing the ANC--they now constitute almost a third of their parliamentary party 15 --brought the percentage of women in the new National Assembly to 24 percent of the total, one of the highest percentages in the world. 16 On the basis of evidence from the United Kingdom and other countries with winner-take-all electoral systems, it seems fair to surmise that use of plurality elections in South Africa would have resulted in a much less heterogeneous and more polarized National Assembly, with whites of various parties representing majority white constituencies, Xhosas representing Xhosas, and Zulus representing Zulus.
A Few Criticisms--and Rebuttals
The strongest recent criticism of the PR prescription for emerging democracies has come from Lardeyret, who lists five ways in which PR supposedly impedes conflict resolution in a divided society. 17 First, he argues, PR tends to allow the representation of extremist groups (he cites as an example the French National Front), which in turn breed [End Page 91] opposing extremist parties that may also gain representation. In fact, however, PR merely helps to fulfill minority parties' desire for representation, rather than promoting new parties, and reflects the actual size of, rather than exaggerates, such parties' support base. Moreover, as noted above, in the most severely divided societies, allowing fringe parties to gain parliamentary representation may actually help the democratic polity to coopt extremists, giving them an incentive to press their case with ballots rather than bullets.
Lardeyret's second criticism is that the experience of Belgium indicates that PR is dangerous for countries faced with ethnic or cultural divisions. He claims that in Belgium, "linguistic parties sprang up after PR was introduced early in this century," making Belgian politics "little more than a feud between the Flemings and the French-speaking Walloons." 18 Here again, Lardeyret is mistaking reflection for causation: Belgian politics would have evolved along linguistic and, to a lesser extent, religious lines regardless of the particular electoral system chosen. The early evidence from southern Africa indicates that PR has exerted pressure for parties to be less rather than more ethnically exclusive. In the new South African National Assembly, almost a third of the 130 White members of parliament represent the ANC, and F.W. de Klerk's NP gained more votes from non-Whites than from Whites.
Lardeyret's third criticism of PR for new democracies is that the multiparty coalition governments it usually produces run the risks of instability and inability to implement coherent policies. This argument is perhaps the most convincing. South Africa's new government carries the burden of huge expectations, and the factionalized nature of Mandela's "cabinet of national unity" may indeed hobble decision making. Yet here as elsewhere in southern Africa, the alternative of control by a single party, with its majoritarian excesses, would be worse. No government, not even a single-party government, would be able to "deliver the goods" if civil society were perpetually on the verge of permanent breakdown. Furthermore, Lijphart has shown that governments elected through PR are no less effective than those elected by other systems. Indeed, parliamentary-PR systems do better in terms of governmental longevity, voter participation, and economic performance. 19
Another criticism offered by Lardeyret is that PR not only allows small extremist parties into parliament, but also gives them too much leverage, as they often control the "swing" seats that can make or break a would-be majority coalition. This may be true, but as the cases of [End Page 92] South Africa and Namibia clearly show, including minorities--no matter how extremist--in the democratic political order is a better alternative than banishing them to the fringes of the system.
Lardeyret claims that plurality elections are more likely than PR to give rise to moderate parties, for under such a system parties are forced to compete constantly for undecided centrist voters. Not only do Britain's Thatcher-led Tory governments (1979-90) and South Africa's NP governments (1948-94) fly in the face of such reasoning, but one would not expect to see much moderation in a plurality system where a dominant party can count on easily winning an absolute majority, as is currently the situation in South Africa, Zimbabwe, Zambia, and Namibia. In southern Africa, only Malawi lacks a single dominant party; and, as mentioned above, the Malawian electorate is divided along regional and ethnic lines, leaving few undecided voters floating in the center.
The high level of geographic representativeness and accountability to the voters that single-member constituencies can provide has long been invoked as an advantage of plurality elections over PR. It is true that the large size of multimember list-PR constituencies can virtually destroy the link between a representative and a specific geographic area (in the South African case, each 20 members represent an average of 2.5 million electors). Roelf Meyer, the minister for constitutional development, claimed that this was a "serious weakness," and that "any new constitution must make provision for clearly designated representatives." 20 The use of closed party lists in South Africa and Namibia is indeed problematic, for the MPs that emerge are far more beholden to party bosses in national party headquarters than to local communities or society as a whole. In order to remedy this without violating the basic proportionality of parliament, the constitutional engineer can use smaller multimember constituencies (with perhaps between 5 and 12 members) designed to dovetail with traditional geographic and societal boundaries, as well as open lists, in which voters choose among candidates as well as parties. In South Africa, the principal objection to such a system was made by members of the ANC, who argued that voting for individuals rather than parties would be too complicated for many of the country's illiterate voters. Yet such a system would still only entail placing one mark by one name and symbol, and the low level of spoiled ballots (less than 1 percent) in the 1994 elections suggests that educational illiteracy is not necessarily synonymous with political illiteracy.
While increasing geographical representativeness and accountability, such an adaptation of simple list PR would also maintain the overall benefits that PR offers to diverse societies. Multimember constituencies would still be able to provide a choice of representative for citizens to approach concerning constituency casework problems. In practice, this [End Page 93] would mean that constituents might be able to approach either an Ndebele or a Shona MP in Matabeleland, Zimbabwe; a Chewa, Lomwe, or Yao MP in Lilongwe, Malawi; or a Zulu, Xhosa, White, or Indian MP in southern KwaZulu, South Africa. Finally, such a constituency-based list-PR system would serve as a further incentive for parties to be ethnically accommodating--that is, to present a diverse slate of candidates as a means of maximizing both their total vote and their number of parliamentary seats.
After the choice of electoral system, the second fundamental choice facing new democracies is that between a presidential and a parliamentary form of government. As this debate is conducted in the context of southern Africa, opinions may be too easily colored by the performance of former African presidents, whether they were democratically elected or not. Prior to the most recent wave of democratization, the political reputation of African presidents ranged from poor to atrocious. All of the countries discussed here have endured painful presidential histories, which one might think would make them eager to avoid vesting one individual with much power. Yet in 1985 Zimbabwe went to a directly elected presidency, as did Malawi nine years later. Namibia moved from an indirectly elected president to a directly elected one in 1994, and South Africa, by naming what is really its prime minister the "executive state president," has created the appearance of presidential control if not the reality.
Good constitutional design for divided societies dictates against directly elected presidents for the emerging democracies of southern Africa. Lijphart has warned that while "the combination of parliamentarism with proportional representation should be an especially attractive one to newly democratic and democratizing countries," the Latin American experience shows that a presidentialist-PR system should be avoided. 21 Matthew Shugart and John Carey have identified three key traits of presidentialism that often have negative consequences: temporal rigidity, majoritarianism, and dual democratic legitimacy. 22 Of these three, majoritarianism is by far the most damaging to stability in ethnically divided societies. The winner-take-all nature of a directly elected executive office in itself militates against presidentialism in southern Africa. In a divided society without a history of stable democracy, there is no assurance that the loser or losers of a presidential race will accept defeat in what amounts to a zero-sum game. As the recent experiences of Angola and Nigeria show, there is little hope for elaborate constitutional mechanisms designed to promote power-sharing if a fragile new democracy is shattered at the first hurdle. Ann Reid of the U.S. State Department laid the blame for the collapse of [End Page 94] peace plans in Angola and the bloody conflict that ensued largely on the country's presidential system: Given that "both [José Eduardo] Dos Santos and [Jonas] Savimbi were vying for the only prize worth having," it was inevitable that Savimbi would resume his violent struggle after losing the election. 23 In Nigeria, the all-or-nothing structure of the 1993 presidential race made it easier than it otherwise would have been for the military to succeed in annulling the election before the final results had been officially announced: unsuccessful candidates and political factions had no immediate stake in the political outcome, and many readily acquiesced in the annulment in the hope of being able to contest again. Ian Campbell claims, moreover, that Nigerian presidentialism caused a marked increase in electoral corruption in the run-up to the vote. "It was suggested that the problem was the 'size of the jackpot,' with the selection process [for presidential candidates] being seen as an 'investment opportunity' and the presidency as the source of instant wealth." 24
Another danger of a presidential system is that a directly elected president tends to be pressured into ethnic or regional exclusivity. Such presidents have a great incentive to offer special privileges to their own ethnic or regional group as a means of ensuring reelection through a simple majority or plurality of votes. The risk to democratic stability is particularly great in Malawi, Namibia, and Zimbabwe. In Malawi, Bakili Muluzi won the 1994 presidential election with 47.2 percent of the vote, much of it coming in the form of a huge vote share in his native southern region. In the South, which accounts for half of the voting population, Muluzi won 78 percent of the votes cast, while in the central and northern regions he polled only 27.8 percent and 4.5 percent, respectively. As mentioned above, Malawian politics is sharply divided along regional lines, and any constitutional structure that allows one region to be permanently excluded from power will destabilize the state as a whole. President Muluzi may well turn out to embrace a strategy of nation-building, distributing resources fairly across the country, but such a positive outcome will have been due to the virtue of the man rather than the institutional structure of the office. The dynamics of Malawi's presidential system invite Muluzi to pamper the South, do only enough to maintain a third of the votes from the central region, and largely ignore the voters in the North.
In Namibia, Dirk Mudge, former leader of the opposition Democratic Turnhalle Alliance (DTA), articulated similar fears about presidentialism: "Politically, the proposal for a directly elected president is unsound and dangerous, because it denies the existence of a multiparty system. A parliamentary head of state who is mandated by parliament becomes a symbol of unity and conciliation, since in the exercise of his executive powers he needs to follow the wishes of the representatives of the people." 25 In fact, in Namibia's first democratic elections in 1989, a full [End Page 95] 70 percent of Sam Nujoma's national vote of 57.3 percent came from his ethnic base of Ovamboland. In the 22 electoral districts outside of Ovamboland, the DTA outpolled Nujoma's party, the South West Africa People's Organization (SWAPO), by 181,000 to 159,000 votes. In the presidential election of December 1994, Nujoma again rode to victory on the back of an overwhelming share of the Ovambo vote, with his 72 percent of the popular vote being strongly concentrated in the heavily Ovambo-populated northern regions of the country.
The actions of Robert Mugabe in Zimbabwe have clearly shown how a powerful directly elected president can politically marginalize a given ethnic group. For seven years after its first accession to power in 1980, Mugabe's Shona-based Zimbabwe African National Union-Patriotic Front (ZANU-PF) effectively excluded Joshua Nkomo's Ndebele-based Patriotic Front-Zimbabwe African People's Union (PF-ZAPU) from political power. In 1987, they signed a unity agreement giving three PF-ZAPU leaders cabinet posts in return for acquiescence to one-party rule. In fact, Zimbabwe's presidential system, combined with the numerical dominance of the Shona, would have allowed Mugabe to maintain his ethnically exclusionary government even without the accord with Nkomo.
If we accept Lijphart's hypothesis that consensual rather than majoritarian democracy is what is needed in divided (or plural) societies, then vesting a single individual with all executive power is clearly detrimental to constructive power-sharing arrangements. The interim South African Constitution has partly overcome this problem by creating a "triumvirate" of leaders from parties representing over 80 percent of the electorate. President Mandela is assisted in carrying out his executive functions by two deputy presidents--Thabo Mbeki of the ANC and F.W. de Klerk of the opposition NP. Such creative power-sharing mechanisms--available in a parliamentary system--can make the difference between stable democracy and spiraling ethnic conflict.
Shugart and Carey offer "premier-presidentialism" as a persuasive alternative to presidentialism. In such systems, the presence of a prime minister tempers the president's "exaggerated sense of mandate." Moreover, the president can dissolve parliament and call new elections when crises arise, which mitigates the problem of rigid terms. Because of its majoritarianism, however, premier-presidentialism is as inappropriate to the southern African democracies as straightforward presidentialism. As Shugart and Carey note, "Because the cabinet is subject to parliamentary confidence, it will not be as narrowly representative of the president's interests as will a presidential cabinet, unless, of course, there is majority support in parliament for the president's narrow interests." 26
When the results of elections in Malawi, Namibia, Zimbabwe, and South Africa are reinterpreted under a hypothetical premier-presidentialist system, it is clear that such a system would still allow one party to [End Page 96] dominate both the executive and legislative branches of government and govern in an ethnically exclusive way. The presence of a statutory multiparty cabinet, as now exists in South Africa, might ease the problem, but as Scott Mainwaring has observed, this too is unattractive in practice: "Multiparty presidentialism is more likely to produce immobilizing executive/legislative deadlock than either parliamentary systems or two-party presidentialism." 27 Such deadlock carries with it the danger of popular discontent with the new power-sharing government, which might well overwhelm any executive achievements in the realm of ethnic accommodation.
Perhaps the South African practice of electing a parliamentary government, headed by a prime minister, and then bestowing upon that leader the title of state president is the most constructive route to follow. Along with inclusive proportional representation and minority-majority power-sharing arrangements within the cabinet, the country thereby also gains a national figurehead who can serve as a rallying point. President Mandela's legitimacy is based on his leadership of a majority parliamentary party and an executive branch that represents over 90 percent of the country's voters--a far stronger foundation than the simple plurality of the electorate that legitimates a directly elected president.
Finding the Least Imperfect System
No electoral system or any other constitutional mechanism is a panacea. The task of the constitutional engineer is to find the least imperfect system and then adapt it to the needs of an emerging democracy, achieving a balance between a representative parliament and a coherent government; an understandable ballot and a broad range of voter choice; and a proportional parliament and solid links between representatives and specific geographic areas. Although electoral-system design is only one cog in the intricate constitutional machine, a misshapen cog may cause the whole structure to grind to a halt.
Divided societies and those without solid democratic histories may be particularly susceptible to instability if the electoral-system design fails to meet any of the following criteria:
Representativeness. If an electoral system is to maximize legitimacy and ensure a degree of electoral consent to any government formation, it must carry out the will of the entire electorate, not just the majority.
Accessibility. The legislature will be representative only to the extent that people feel included in the political process. When people feel that their vote makes a difference, they are more likely to work for change within the system instead of supporting antisystem parties and organizations that feed on societal instability.
Provision of incentives for conciliatory behavior. The electoral [End Page 97] system must promote an environment of compromise and reconciliation, rather than exacerbate existing conflict. This does not mean enforced consensus, but rather the mutual recognition of opposing viewpoints.
Accountability. The electoral system must promote a high degree of accountability of government and MPs to their constituents. The amount of influence that voters have over their representatives depends on both the size of geographic constituencies and the extent to which voters can choose among candidates, as opposed to parties.
Encouragement of cross-cutting parties. Prospects for stability are maximized by an electoral system that engenders parties rooted in shared perspectives on issues and political values rather than ethnic, linguistic, or geographic factors.
The early evidence from southern Africa indicates that PR systems uniformly outperform their plurality counterparts with respect to the first three criteria listed above. Clearly, accountability is a problem in the large-district, closed-list form of PR used in South Africa and Namibia; the use of smaller multimember PR constituencies in conjunction with open lists could alleviate this problem. To date there is little evidence of the encouragement of cross-cutting parties under either plurality or PR in southern Africa, but presumably the evolution of such parties is best facilitated by the broad-based party lists necessary under PR. Finally, the experiences of Malawi, Zimbabwe, and Nigeria show that investing substantial executive control in a directly elected president is detrimental to stability in a multiparty government. For these reasons, parliamentary-PR systems of government are best for the fragile new democracies of southern Africa as well as for other African states--such as Angola, Mozambique, and Zaire--that have yet to make the transition to democracy.
Andrew Reynolds is a doctoral candidate at the University of California, San Diego, specializing in constitutional design and democratization in southern Africa. In April 1994, he served as an observer of the first nonracial South African elections. He is the editor of Election '94 South Africa: The Campaigns, Results and Future Prospects (1994), and the author of Voting for a New South Africa (1993).
This essay is based on research supported by a grant from the National Science Foundation of the United States (SBR-9321864).
1. See Arend Lijphart, "Constitutional Choices for New Democracies," Journal of Democracy 2 (Winter 1991): 72-84; Guy Lardeyret, "The Problem with PR," Journal of Democracy 2 (Summer 1991): 30-35; Quentin L. Quade, "PR and Democratic Statecraft," Journal of Democracy 2 (Summer 1991): 36-41; Ken Gladdish, "The Primacy of the Particular," Journal of Democracy 4 (January 1993): 53-65; Vernon Bogdanor, "Israel Debates Reform," Journal of Democracy 4 (January 1993): 66-78; and Timothy Sisk, "South Africa Seeks New Ground Rules," Journal of Democracy 4 (January 1993): 79-91.
2. W. Arthur Lewis, Politics in West Africa (London: Allen and Unwin, 1965), 71, 72.
3. Lijphart, "Constitutional Choices for New Democracies," 81.
4. Sisk, "South Africa Seeks New Ground Rules," 88.
5. Lardeyret, "The Problem with PR," 35.
6. Donald Horowitz, A Democratic South Africa? Constitutional Engineering in a Divided Society (Berkeley: University of California Press, 1991), 167.
7. Lijphart, "Constitutional Choices for New Democracies," 74.
8. Jonathan N. Moyo, Voting for Democracy: Electoral Politics in Zimbabwe (Harare: University of Zimbabwe Publications, 1992), 158.
9. Jowie Mwiinga, "Chill for Chiluba," Africa Report 39 (March-April 1994): 60.
10. The analysis that follows is presented in more detail in Andrew Reynolds, "Re-running the South African and Malawian Parliamentary Elections Under Alternative Electoral System Formulae" (paper to be presented at a conference on "Elections in Africa" sponsored by the United States Institute of Peace, Washington, D.C., May 1995). The results presented here were calculated on the basis of equal-sized single-member districts within the nine provinces, and voting data from the 343 counting districts used in the actual election. It was assumed that voting preferences would not have been significantly affected by the varying incentives provided by different electoral systems.
11. Vernon Bogdanor, What Is Proportional Representation? (Oxford: Martin Robertson, 1984), 44.
12. The extent to which the interim South African Constitution is consociational could be the focus of an important debate. Although the method of constituting the "government of national unity" is classically consociational, minority parties do not have a formal blocking veto, and the absence of executive power-sharing in any new constitution would make the government considerably more majoritarian in nature.
13. Arend Lijphart, "Prospects for Power-Sharing in the New South Africa," in Andrew Reynolds, ed., Election '94 South Africa: The Campaigns, Results and Future Prospects (New York: St. Martin's, 1994), 222.
14. Lijphart, "Constitutional Choices for New Democracies," 81.
15. In drawing up its national and provincial lists, the ANC used a quota system that entitled women to a third of the places.
16. Among established democracies, only the four Scandinavian nations and the Netherlands have a higher percentage of female MPs than South Africa. See Wilma Rule, "Women's Underrepresentation and Electoral Systems," PS: Political Science and Politics 27 (December 1994): 689-92.
17. Lardeyret, "The Problem with PR," 32-33.
18. Ibid., 32.
19. Lijphart, "Constitutional Choices for New Democracies," 81.
20. Editorial, Natal Witness (Pietermaritzburg), 18 August 1994.
21. Lijphart, "Constitutional Choices for New Democracies," 72.
22. Matthew Shugart and John Carey, Presidents and Assemblies: Constitutional Design and Electoral Dynamics (New York: Cambridge University Press, 1992), 28-43.
23. Ann Reid, "Conflict Resolution in Africa: Lessons from Angola," INR Foreign Affairs Brief (Bureau of Intelligence and Research, U.S. Department of State, Washington, D.C., 6 April 1993), 2.
24. Ian Campbell, "Nigeria's Failed Transition: The 1993 Presidential Election," Journal of Contemporary African Studies 12 (1994): 182.
25. Quoted in Lionel Cliffe et al., The Transition to Independence in Namibia (Boulder, Colo.: Lynne Rienner Publishers, 1994), 208.
26. Shugart and Carey, Presidents and Assemblies, 49-51 (emphasis added).
27. Scott Mainwaring, "Presidentialism, Multipartism, and Democracy: The Difficult Combination," Comparative Political Studies 6 (July 1993): 200.