The South African Electoral System

Unusual Features and Prospects for Reform

Arend Lijphart

        Probably the most surprising aspect of the April 1994 parliamentary election in South Africa is that it took place at all. Although I was one of a tiny minority of cautious optimists, most observers of the South African political scene in the 1970s and 1980s did not believe that a negotiated transition to democracy was possible.
        The election was also unusual in the sense that it was the first democratic election in the country -- and hence not just an election but an affirmation of democratic liberation and a referendum, supported overwhelmingly, on the new democratic system.

Highly Proportional List PR

        I should like to call attention to yet another highly unusual feature of the election: the almost purely proportional PR system that was used. In fact, I believe that it can be called unique in this respect. To my knowledge, no national parliamentary election has ever been held under an equally or more thoroughly proportional system, with the exception of the short-lived East German democratic parliament in March 1990. (PR itself is not unusual, of course; most of the world's democracies use some form of it.)
        The proportionality of PR systems depends mainly on four factors: the electoral formula (such as plurality, different forms of PR, etc.); district magnitude (the number of representatives elected per district), electoral threshold (the legal minimum required for representation), and the size of the assembly to be elected.
        South Africa opted for maximum proportionality: one huge, nationwide district for the conversion of votes into seats, no electoral threshold at all and a very large assembly with 400 seats. The one small exception was use of the slightly disproportional Droop formula (for the first five remaining seats before switching to a modified version of d'Hondt) instead of the purely proportional Hare formula.
        A quick glance at the election results for the
seven parties that won representation in the National Assembly shows the high degree of correspondence between seat percentages and vote percentages. The African National Congress (ANC) won 62.65% of the votes and 63.00% of the seats, while the respective figures for the National Party (NP) were 20.39% and 20.50%, for the Inkatha Freedom Party (IFP) 10.54% and 10.75%, for the Freedom Front 2.17% and 2.25%, for the Democratic Party 1.73% and 1.75%, for the Pan Africanist Congress 1.25% and 1.25% and for the African Christian Democratic Party 0.45% and 0.50%.
        A good way of measuring the overall degree of proportionality or disproportionality of an election is the commonly used Loosemore-Hanby index, which reveals the total percentage of over-representation (that is, the total percentage by which the "over-represented parties" are over-represented in terms of seats won compared to votes won).
        According to this index, the disproportionality of the South African election result was only 0.82% -- lower than in all but three of the 384 elections that I examined in my Electoral Systems and Party Systems: A Study of Twenty-Seven Democracies, 1945-1990 (Oxford University Press, 1994). If the more proportional Hare system had been applied, only two seats would have changed hands: two small parties would have received one seat each at the expense of the ANC and the IFP. This would have lowered the index of disproportionality a bit further: 0.53%. Nevertheless, while not perfectly proportional, South African PR came extremely close to perfect proportionality.

Party List PR

        An additional unusual feature of South Africa's PR was that it was a party list system -- unusual because in English-speaking countries and former British dependencies PR normally takes the form of preference voting (e.g., choice voting or the single transferable vote), in which voters vote preferentially for individual candidates. (Party list PR is used in all non-English-speaking countries with PR systems.) The main exceptions are Guyana, Zimbabwe (in the 1980 election), Sri Lanka (from 1989 on), Cyprus (the Greek part of the island) and New Zealand (the electoral system adopted in 1993, but not yet used).
        Moreover, South Africa's list PR system used closed lists, in which the voters did not have an opportunity to indicate preferences for individual candidates. Being able to vote for individuals as well as parties tends to be considered highly desirable in democracies with a British political heritage, although it is worth noting here that the prevalent electoral system in these countries -- plurality in single-member districts -- does not offer a choice of individuals within parties either, because parties typically nominate only one candidate each.

Moderate Multi-Partism

        Because PR is associated with multi-partism and because enemies of PR tend to worry about the dangers of extreme multi-partism, it is worth emphasizing that South Africa's highly proportional PR system did not lead to an extreme multi-party system. As indicated above, only seven parties won seats; one party, the ANC, won an absolute majority of both votes and seats; and the three largest parties (ANC, NP, and IFP) captured 94.25% of the seats (377 out of 400).
        The best comprehensive and widely used measure of the degree of multi-partism is the "effective number of parties," which weights parties according to size. For instance, in a two-party system with two exactly equal parties, the index is 2.0. With three exactly equal parties, the index is 3.0. For three unequal parties, the index is less than 3.0; for example, in a three-party system with parties holding 45%, 40% and 15% of the seats, the index is 2.6. The effective number of parties resulting from the 1994 election in South Africa was 2.2 parties -- only slightly more than in a pure two-party system! -- mainly, of course, because of the commanding majority won by the ANC.
        Finally it is worth noting that the list PR system provided a strong incentive to the parties to be moderate and inclusive rather than divisive. In order to appeal to as many voters as possible, the main parties made strong efforts to nominate racially balanced lists of candidates. Somewhat ironically, this made the NP -- the most racially exclusive party in the apartheid era -- into the party with the most diverse voter support.

Prospects for Reform

        The electoral system used in April 1994 is part of the interim constitution. In principle, therefore, it could be changed drastically when the permanent constitution is adopted, but because the electoral system is universally judged to have worked very well, wholesale changes appear to be out of the question. However, a few minor reforms may well be introduced: in particular, the two features listed above as unusual PR rules -- extreme proportionality and closed lists -- are sure to be considered for modification.
        The high degree of proportionality can be decreased by introducing an electoral threshold and/or reducing the district magnitude. But anything except the most modest measures of this kind would immediately have a strong impact and would be seen as unduly punitive to small parties. For instance, if there had been a relatively low 2.5% threshold in the April 1994 election, four of the seven parties that actually won seats would have been denied any representation.
        The closed lists may well become partially open lists, allowing voters to express a preference for an individual candidate on the list, as in the Belgian and Dutch forms of PR. New Zealand's new PR system, fashioned after the German model, also offers attractive possibilities for strengthening the element of individual candidate choice: it combines plurality single-member district elections with national list PR and over-all national proportionality.
        A less far-reaching change would be to combine national proportionality with PR in relatively small multi-member districts. Actually, the South African system used in April 1994 was already such a system: while the over-all election result was determined by converting the parties' national vote totals into the 400 seats in the National Assembly, 200 of the representatives were elected from separate provincial lists in each of the nine provinces. This means that, on average, each province has 22 of its "own" representatives. This number can easily be reduced -- and hence closer voter-representative ties fostered -- by dividing the larger provinces into separate election districts.
        It is clear that all of the above possible or likely reforms are rather minor. South Africa's highly proportional list PR may be moderated to some extent, but my prediction is that the electoral system used in the next parliamentary election, scheduled to be held in 1999, will still be a list PR system with a high degree of proportional purity.

        Arend Lijphart is a professor of political science at the University of California at San Diego and President-Elect of the American Political Science Association. This article is part of a National Science Foundation-funded research project with Bernard Grofman and Andrew Reynolds on Electoral Laws, Electoral Lists and Campaigning in the First Non-Racial South African Elections.

Table of Contents
Spotlight Four