13 - REJOINDER: THE CASE FOR PROPORTIONALITY (Phương Hoàng)

REJOINDER: THE CASE FOR

PROPORTIONALITY

Andrew Reynolds

Andrew Reynolds is associate professor of political science at the Uni-versity of North Carolina at Chapel Hill. He has consulted on issues of electoral and constitutional design for a wide array of countries. His latest book is The Architecture of Democracy: Constitutional Design, Conflict Management, and Democracy (2002). This essay originally appeared in the October 1995 issue of the Journal of Democracy.

In “Elections in Agrarian Societies” (pp. 135–45 above), Joel Barkan offers a strong challenge to the newly emerging conventional wisdom that proportional representation (PR) is the best electoral formula for the fledgling—and often highly divided—democracies of Africa. First, he criticizes PR for weakening (or even severing) the link between indi-vidual MPs and constituents. This hinders the development of the “vertical” dimension of democracy (that is, the representative relation-ship between elites and nonelites with a common political interest), “greatly reduc[ing] the prospects for the consolidation of democratic rule” (p. 136). For Barkan, the relationship between representative and voter that obtains within a single-member district (SMD) best reflects the nature of agrarian societies in Africa, in which the strongest ties are those of kinship, neighborhood, and land, and people “define their [po-litical] interests . . . on the basis of where they live” (p. 136). While accepting that some degree of proportionality (indicating the protec-tion of minority interests) is a normative good, he argues that the patterns of geographically polarized voting seen in agrarian societies enable SMD plurality systems to produce parliaments that are reasonably re-flective of the distribution of the nationwide popular vote. Thus one can retain one of the underpinnings of consociational government (a proportionally constituted parliament) while avoiding the main draw-back of PR: the detachment and lack of accountability of representatives elected from party lists.

It is true that when voting patterns closely follow cleavages among

Andrew Reynolds

147

groups defined by ascriptive traits (such as race, ethnicity, language, or religion), and when different groups cluster in different areas, elections held under SMD plurality can produce highly proportional results. As Barkan notes, the 1994 general election in Malawi gave rise to a parlia-ment that closely mirrored the distribution of the national vote among the three main parties. An election’s Index of Disproportionality (ID) measures the degree to which the distribution of parliamentary seats among parties diverges from the distribution of votes, with zero repre-senting a perfectly proportional outcome. The score on that index for the Malawian election (2.1) was lower than the figures for plurality elections in all but three of the established democracies for which data are presented in Arend Lijphart’s comprehensive Electoral Systems and Party Systems, which covers the period from 1945 to 1990.1 Similarly, the regionally polarized parliamentary election of 1985 in Zimbabwe produced an ID of 2.5, and my own “re-running” of the 1994 parliamen-tary elections in South Africa and Namibia (which were held under PR systems) indicated that SMD plurality would have produced relatively proportional results, with IDs of 6.7 and 4.0, respectively.2

The ID for a given election tells us much about parliamentary com-position. High IDs indicate a high likelihood that: 1) minority parties are receiving little or no representation; 2) larger parties are gaining “seat bonuses” over and above their share of the popular vote; 3) gov-ernments with 100 percent of the executive power are being catapulted into office with less than 50 percent of the popular vote; and 4) govern-ments based on a simple majority of the popular vote are being awarded supermajority powers (as in South Africa, where the use of a plurality system would have provided the African National Congress [ANC], which received 62 percent of the vote, with two-thirds of the seats in the National Assembly—enough to write the new constitution unfettered).

Atypical Cases

The evidence from Malawi in 1994 and Zimbabwe in 1985 (and the hypothetical evidence from South Africa and Namibia) seems to indi-cate that, in southern Africa, SMD plurality can provide “the best of both worlds”—a proportionally constituted parliament, along with the representational advantages of a district-based system. Barkan claims that “in agrarian societies . . . PR does not produce electoral results that are significantly ‘fairer’ or more inclusive than plurality elections based on SMDs” (p. 136). Yet the results from Malawi and Zimbabwe have not been mirrored in established democracies, in Africa as a whole, or within southern Africa itself. At the end of the day, these are atypical cases, and we must take care not to be blinded by them.

The five countries using SMD plurality for democratic elections in southern Africa since 1965 have, on average, experienced IDs in line

148		Rejoinder: The Case for Proportionality
TABLE 1—AVERAGE IDS OF SMD PLURALITY & PR ELECTIONS
	SOUTHERN AFRICA		AFRICA	ESTABLISHED
				DEMOCRACIES
SMD Plurality	11.0a		12.9c	9.8e
PR	3.7b		4.8d	2.9f

a Thirteen elections: Botswana, 1965–94; Lesotho, 1993; Malawi, 1994; Zambia, 1991; Zimbabwe, 1985 (common roll), 1990, 1995.

b Six elections: Angola, 1992; Mozambique, 1994; Namibia, 1989, 1994; South Africa, 1994, Zim-babwe 1980.

c Twenty elections: Botswana, 1965–94; Gambia, 1966–92; Kenya, 1992; Lesotho, 1993, Malawi, 1994, Zambia, 1991; Zimbabwe, 1985–95.

d Thirteen elections: Angola, 1992; Benin, 1991; Burkina Faso, 1992; Burundi, 1993; Cape Verde, 1991; Madagascar, 1993; Mozambique, 1994; Namibia, 1989, 1994; Niger, 1993, 1995; South Af-rica, 1994; Zimbabwe, 1980.

e Calculated from data for 78 elections held under plurality in five countries: Canada, 1945–88; India, 1952–84; New Zealand, 1946–90; United Kingdom, 1945–92; United States, 1946–94.

f Calculated from data for 212 elections held under various PR formulas in 17 countries: Austria, 1945–90; Belgium, 1946–87; Costa Rica, 1953–90; Denmark, 1945–88; Finland, 1945–87; Ger-many, 1945–87; Iceland, 1946–87; Ireland, 1948–89; Israel, 1949–88; Italy, 1946–87; Luxem-bourg, 1945–89; Malta, 1947–87; Netherlands, 1946–89; Norway, 1945–89; Portugal, 1975–89; Sweden, 1948–88; Switzerland, 1947–87.

Note: Method of calculation: least squares index developed by Michael Gallagher. See his “Propor-tionality, Disproportionality and Electoral Systems,” Electoral Studies 10 (1991): 33–51. Sources: For Africa as a whole, author’s calculations based on data presented in Shaheen Mozaffar, “The Political Originas and Consequences of Electoral Systems in Africa: A Preliminary Analysis” (paper presented at a conference on “Comparative Democratic Elections,” Kennedy School of Gov-ernment, Harvard University, May 1995); for established democracies, calculations based on data presented in Arend Lijphart, Electoral Systems and Party Systems (New York: Oxford University Press, 1994), and updated by Reynolds.

with plurality elections in the rest of the world (Table 1). The results have been slightly more proportional than those for Africa as a whole but slightly less proportional than those for the established democra-cies—Britain and the inheritors of her “first-past-the-post” electoral system. Furthermore, the figures for PR elections show that, despite the low IDs for plurality in Malawi and Zimbabwe, when it comes to trans-lating votes into seats, PR is not just marginally superior to SMD plurality but substantially so. The evidence from throughout Africa suggests that the use of an electoral system based on SMDs does not ensure the pro-portionality of electoral results. This is true even in agrarian societies, where voting patterns are geographically concentrated. It follows that if proportionality is a necessary feature of consociational democracy, plu-rality cannot be relied on to secure it.

Even if plurality elections in fledgling democracies produced rea-sonably proportional results across the board, there would still be worrying threats to democratic consolidation. First, the experience of South Africa illustrates that the inclusion within parliament of small minority parties can play a crucial stabilizing role in the early years of democratization of a divided society. Clearly, the presence of the Afrikaner Freedom Front (FF), Pan-Africanist Congress (PAC), and Demo-cratic Party (DP) in the first nonracial South African parliament has been conducive to an atmosphere of reconciliation. As I noted in “Con-

Andrew Reynolds

149

stitutional Engineering in Southern Africa,” this measure of political inclusiveness would not have been achieved had the April 1994 elec-tions been run on a plurality system rather than a PR system.

Second, if over the longer term economic and class mobility increases and ethnic and regional divisions fade, as most people hope, the result-ing integration of neighborhoods and reduction in the political importance of primordial ties will likely bring about a decline in the geographic polarization of voting. This weakening of regional bloc voting will give rise to all the vagaries of plurality that more homoge-neous societies such as Britain and the United States have been forced to endure. Such vagaries include the exclusion of substantially sup-ported third parties from representation and the winning of parliamentary majorities by parties with fewer votes than their opponents. It is often claimed that while the more stable and established Western democra-cies can tolerate these representational anomalies, they could prove fatal in Africa, where democracy is—and will be for many years to come—fragile at best. If the Malawian Congress Party (MCP) of former president Hastings Kamuzu Banda had won a parliamentary majority in 1994 with only 34 percent of the vote, against the 46 percent won by the United Democratic Front (UDF), one can only imagine the civil strife that would have erupted. Nor is this wild speculation without historical foundation. In the whites-only general election in South Af-rica in 1948, D.F. Malan’s Nationalist–Afrikaner coalition came to power with 79 of the 150 seats in parliament despite polling only 42 percent of the vote. Jan Smuts’s United–Labour coalition gained 52 percent of the popular vote but only 71 parliamentary seats.

Perhaps most important, the use of plurality in southern Africa freezes the party system to such a degree that the alternation of parties in gov-ernment and opposition is not perceived as a likely or natural occurrence. This is true even when levels of disproportionality are not high. As Table 2 illustrates, elections held under SMD plurality in southern Af-rica have provided a paucity of truly competitive seats—the lifeblood of the Westminster system of government.

All five southern African plurality elections described in Table 2 reveal the classic elements of a de facto one-party state, in which gov-erning parties are insulated from electoral challenges. In Malawi, Zambia, and Zimbabwe, the largest parties won their seats with huge segments of the vote, making them largely invulnerable. Marginal seats are those won with a margin of victory (whether a plurality or a major-ity) of less than 10 percent of the valid vote. These are considered “battleground seats”—which way they go determines which party forms a government. In the elections of the five established democracies, be-tween one-fifth and one-third of the seats were considered marginal. In southern Africa, however, competitive seats have never amounted to more than a tiny fraction of the legislative body. In Zimbabwe, no seat

150		Rejoinder: The Case for Proportionality
TABLE 2—MARGINAL SEATS IN RECENT SMD PLURALITY
	ELECTIONS
	ELECTION	NUMBER OF		% OF	LARGEST	AVERAGE
		MARGINALS		TOTAL	PARTY	VOTE IN
				SEATS		SEATS WON
Southern Africa	Zimbabwe, 1995	1		0.8	ZANU-PF	82%
	Zimbabwe, 1990	4		3.3	ZANU-PF	81%
	Zimbabwe, 1985	3		3.7	ZANU-PF	91%
	Malawi, 1994	7		3.9	UDF	72%
	Zambia, 1991	6		4.0	MMD	81%
Established	USA, 1994	88		20.2	Republican	65%
Democracies
	Canada, 1993	63		21.4	Liberal	56%
	UK, 1992	171		26.3	Conservative	52%
	India, 1980	157		29.7	Congress	52%
	New Zealand, 1993	30		30.3	National	44%

Note: Marginal seats are defined as those seats won with a margin of victory of less than 10 percent of the total valid vote. Figures are for lower houses of parliament only.

defended by a candidate of the incumbent party has ever changed hands. In effect, the use of plurality systems in heterogeneous societies freezes the number of seats won, creating an environment in which only a po-litical earthquake can jar the patterns of party-vote concentration.

This finding is a serious blow to the case for plurality in Africa, for such majoritarian prescriptions rely on at least the perception that power can change hands from election to election. If plurality leads to a de facto one-party state, there are no incentives for losing parties to remain loyal opposition parties. In the deeply divided societies of Africa, this poses the greatest possible threat to democracy in its initial stages. In contrast, PR systems may display sensitivity to evolving vote patterns by facilitating shifting government coalitions or forcing a single-party cabinet to include other parties when the ruling party’s support falls below an absolute majority of the population.

Remedying the Problems of PR

As Barkan points out, accountability and dialogue between repre-sentatives and constituents are crucial to any properly functioning democracy. And it is true that the type of national, large-district PR used in South Africa and Namibia has weakened the link between elites and nonelites, giving rise to fears of a “‘suspended state’ . . . discon-nected from the population” (p. 137). Yet instituting an electoral system based on SMDs may not always be the best way to overcome such ob-stacles to the fulfillment of the vertical dimension of democracy. First, the overwhelming regional concentration of voting patterns through-out Africa gives rise to SMDs that are little more than pocket boroughs of this or that party. One intuitive hypothesis is that the less competi-

Andrew Reynolds

151

tive the seat, the lower the quality of the candidates, and therefore the less responsive the elected MP will be to his or her constituents’ needs. This “yellow dog” syndrome (so called because voters would sooner elect a yellow dog than someone from an opposing party) has been noted throughout the democracies that traditionally have used SMD plurality, especially the United States (where the term was coined) and the United Kingdom.

Second, adequate constituency representation is not simply a matter of advocating the interests of 50, 60, or even 80 percent of a given community; it is about allowing supporters of both majority and minor-ity parties within a certain area to have their views articulated in parliament. In Africa, SMD plurality has accentuated regional fiefdoms of party dominance to the extent that 25 percent of the voters of Zambia’s Eastern Province (who happened not to support the United National Independence Party [UNIP]), 15 percent of Malawi’s Northern Region (who voted against the Alliance for Democracy [AFORD]), and virtu-ally the entire opposition vote of Zimbabwe (which hovered around 19 percent in both 1990 and 1995) are prevented from having a direct say in legislative affairs. The constitutional engineer who seeks to craft a dynamic and inclusive representative democracy must look somewhere in between the extremes of the remote and unaccountable representa-tion that characterizes national-list PR and the exclusionary and all too often complacent representation provided by MPs elected from “safe” SMDs.

The debate currently taking place within South Africa’s Constitu-tional Assembly highlights both the dissatisfaction with large-district list PR and the ways in which PR systems can be adapted to provide more accountable and responsive representation. In South Africa, two options are now on the table. The first is to introduce SMDs for electing a proportion of the National Assembly (perhaps half), with the rest of the MPs being elected from regionally based lists. This would resemble the German system in that, overall, the National Assembly would reflect near proportionality between votes cast and seats won, but a single person would represent each district.

Although this “mixed-member” system might seem attractive, it has two key flaws. First, because of plurality’s tendency to exaggerate re-gional fiefdoms, it is quite probable that certain parties would dominate all of the SMD seats, leaving minority-party MPs to be elected only from the “top-up” pool of members from regional lists. For example, throughout South Africa’s Orange Free State, urban and rural constitu-encies would be represented solely by ANC members, while minority representatives would be elected from the more detached party lists. This leads to the second problem of the mixed-member system: the creation of two classes of MPs, the first group elected from districts tied to local parties and local voters and responsive to constituents’ con-

152 Rejoinder: The Case for Proportionality

cerns, and the second group elected from the lists and therefore ac-countable chiefly to party bosses in national headquarters.

The second option is to reduce the size of the multimember districts (MMDs) of list PR and provide voters with more sophisticated choices on the ballot. In Voting for a New South Africa, I advocated a formula, akin to the Finnish system, by which 300 members of the South African National Assembly would be elected from MMDs ranging from 5 to 12 members in size, and voters could choose among candidates as well as parties.3 Parties would win seats in proportion to the number of votes they received, but those seats would be filled by the parties’ most popu-lar individual candidates. To ensure overall proportionality, 100 seats would be reserved to “top up” each party’s share of the parliamentary seats; in order to remedy the problem of detachment, however, these additional members would be drawn from each party’s most successful losing candidates.

This system is by no means perfect. Yet it retains overall proportion-ality in parliament and simplicity for the voter, while at the same time enhancing geographic accountability and diversity of representation within regions. An added benefit is that minority votes are not “wasted,” as they are in SMD plurality; this might encourage minority parties in regions where one ethnic or linguistic group predominates. Minority representation and party organization within regions are particularly important when provincial powers are strengthened through federalism, as regional minorities are then less able to rely on the intervention of the national state to protect their interests.

Joel Barkan and I agree on a number of important issues. We agree that choosing an electoral system involves a number of trade-offs among consequences that we attempt to predict by assessing evidence from previous elections. We concur that we should not seek a “perfect” elec-toral system, nor should we propagate one method as the single best system for any context. Rather, electoral rules must be tailored to the specific needs and desires of each individual society. Furthermore, we agree that the inclusion of both minority and majority interests, and therefore some degree of electoral proportionality, is a normative good in fledgling democracies that are divided along ethnic or regional lines.

I would argue, however, that the results of SMD-plurality elections throughout Africa and the rest of the world indicate that this system cannot be relied on to produce a distribution of parliamentary seats that closely mirrors the distribution of the popular vote, and thus will not necessarily facilitate a diverse and inclusive legislative body. Further-more, plurality’s potential to produce extremely disproportional results and legislative anomalies makes it especially unsuitable for the fragile new democracies of Africa. Barkan argues that electoral rules could be written so as to mandate culturally homogeneous districts (`a la the U.S. Justice Department’s guidelines for racially representative districts),

Andrew Reynolds

153

but my guess is that these legal gerrymanders would spark the same tension and controversy that have arisen in the United States.

Certainly, the list-PR systems currently in effect in Africa exhibit deficits of legitimacy and accountability. But there is no need to throw the baby out with the bathwater. With a little imagination and innova-tion, PR systems in Africa can be made to provide a solid link between representatives and constituents, thus strengthening the prospects for democratic consolidation. If South Africa, in devising its new constitu-tion, finds a way to elect not only a proportional National Assembly but an accountable and responsive parliament representing all the people, it may again serve as a beacon for the rest of Africa.

NOTES

This essay is based on research supported by the National Science Foundation of the United States (grant SBR-9321864) and the Institute on Global Conflict and Cooperation.

1. Arend Lijphart, Electoral Systems and Party Systems (New York: Oxford University Press, 1994).

2. See Andrew Reynolds, “Re-running the 1994 South African and Malawian Parliamentary Elections Under Alternative Electoral System Formulae,” in Timo-thy Sisk and Andrew Reynolds, eds., Elections and Conflict Resolution in Africa

(Washington, D.C.: U.S. Institute of Peace Press, 1998). The 4.0 figure for Namibia is slightly lower than Barkan’s calculation for plurality on the basis of the 95 counting districts. My assessment is based on a combination of the 95 counting districts to provide 72 SMDs of roughly equal size. That calculation leaves the South West People’s Organisation (SWAPO) with 55 seats and the Democratic Turnhalle Alliance (DTA) with 17. (I am very grateful to Joel Barkan for sharing his counting-district data with me.)

Andrew Reynolds, Voting for a New South Africa (Cape Town: Maskew Miller Longman, 1993), 66–103.