SHOULD THE 2021 LOCAL GOVERNMENT SYSTEM SERVE AS A TEMPLATE FOR A NEW NATIONAL ELECTORAL SYSTEM?
Electoral politics in democratic South Africa is on a knife-edge. Polls indicate that the ruling party may lose its absolute majority in Parliament by 2024, the outcome of which should introduce coalition politics at the national level, ending almost three decades of one-party dominance by the ANC.
Despite the healthy democratic turnover in rulers that such as shift would introduce, this prospect is not without complications. For one, the electoral system under which the votes are to be counted in 2024 has yet to be finalized, the details of which can decisively shape election results.
Second, the inherent weakness of coalitions is that of chronic instability, and this is already present in some of the elected metropolitan councils, especially Tshwane, Ekurhuleni, and Nelson-Mandela Bay. Nonetheless, there is a persistent case being made that the electoral system used at the local level be taken as a template for constructing a new one at the national level.
If this argument deserves serious consideration, then it is worth scrutinizing some of the contentious aspects of our local electoral system.
The basic design, familiar to all voters, is that of a mixed system of proportional representation (PR), with two ballot papers, one for electing a councillor from a ward, and another for councillors from political parties according to a closed party-list. The party-list acts as a compensatory device for securing party proportionality in the (inevitable) event of distortion that may arise from the first-past-the-post (FPTP) electoral design used for electing ward representatives.
A number of concerns can and have been raised against a full replication of this aspect of the system at the national level.
Winner-takes-all
First, the standard objection to the FPTP system is that it allows for a candidate to win with only a relative majority of votes (i.e. just one more vote than the candidate that gets the second most votes). This is aptly described as a winner-takes-all system. This adversely affects the legitimacy of the winner, who may then have, in total, more votes cast against her/him, than for her/him.
These concerns are valid, and the FPTP method should be avoided in a new national Electoral Act.
Following from the above, with only a relative majority of one vote needed, all other votes become immaterial to the outcome of the election in that ward, they thus end up being so-called “wasted” votes.
When using FPTP in a mixed system, as we do at local government level, the last objection falls away, as those votes are not “wasted” but are all added to the total number of votes (votes for independents excluded) which enters into the calculation of the quota for PR seats. They can thus contribute to seats gained from the party-list for losers in wards. This attribute of a mixed system is worthy of incorporation into a new national Electoral Act.
Independents
Second, proportionality is achieved in this system for political parties only, with the second ballot paper only presenting party choice to the voters. Votes for independents are discarded from the total number of votes for the purpose of calculating the PR seat quota, and seats won by independents are subtracted from the total number of seats used in calculating the quota.
This feature raises strong objections from many advocates of electoral reform in South Africa. The original rationale for the challenge to the Constitutional Court on the constitutionality of the current Electoral Act was that the national party-list system does not allow independent candidates to enter the electoral arena, and therefore violates aspects of section 19 of the Constitution.
Some have argued that independents must somehow gain access to the second ballot paper should we import a mixed system of PR at the national level. If not, the argument goes, all votes for independents (both for candidates that won and those who lost in ward elections) end up as being “wasted”, having been deleted from the overall calculation for achieving proportionality in the election result.
In that sense then, it is argued, all votes do not carry identical weight in shaping election outcomes, thus violating the constitutional norm of equality.
One counter to this is that discarding the votes of independents in this way is best practice. This applies to South Africa’s local government electoral system (which has not been challenged in court) and also at the national level in New Zealand.
It can also be objected that independents, by their intrinsic nature, cannot be represented proportionally. The argument is that parties are composed of individuals who carry two identities with them into the election: their individual attributes and also their collective party identity.
In the wards they are elected as individual members of the party, and the party also gains collective representation through the compensatory list. Independents carry only one identity - themselves. Voters can select them as singular entities, but these unique and distinctive independents cannot again be re-represented as a collective because they do not have a shared collective identity.
Independents also cannot be collectively represented because they are by definition autonomous/distinct and organizationally separate from all other contenders, including from other independents.
They are not a group or an organization, but only a category of individuals. According to this view independents also cannot gain access by forming associations prior to elections. If you are an independent you cannot also be a member of an electoral association, and if you are a member of such an association you cannot be considered to be an independent.
There are ways in which the votes that are cast for independent candidates can feature more prominently in the computation of election results (i.e. carry more weight, so to speak) at the level of parliamentary elections.
These all entail systems that allow for preferential voting, i.e. where results are determined not only by the first choice of voters, but also by their second or even third choices. If the mixed system uses single-member wards, then either the alternative vote or the 2nd ballot systems allow for preferential voting. If multi-member wards are used then the Single Transferable Vote (STV) method as used in Ireland can be applied.
Quotas
Third, the formula for calculating quotas in the current local government electoral system is highly problematic. There are various ways of calculating quotas for the conversion of votes gained into seats allocated to political parties.
These can be broadly placed into two categories. The first is known as the “highest average” method, and the second, which concerns us here, the “highest remainder” method, with the Hare quota and the Droop quota as two of the most used methods.
The Hare quota is calculated by dividing the total number of valid votes by the total number of valid seats being contested in the particular instance:
total valid votes
Hare quota = ---------------------
total valid seats
The Droop quota is calculated by dividing the total number of valid votes by the total number of valid seats plus 1, with 1 being added to the product thereof.
total valid votes
Droop quota = --------------------- +1
(total valid seats +1)
The aim of a quota is to calculate the smallest number of votes needed for a candidate to ensure/guarantee that it wins a seat. At first glance the Hare formula seems the most sensible, uncomplicated and obvious way to go about achieving this, while the Droop formula appears to be overly intricate. This is not necessarily the case. The crucial variable here is the number of seats being contested.
The difference in the size of quotas achieved through these two methods recedes as the number of seats being contested rises. Any given number of valid votes, let us say 100,000, divided by 20 yields a Hare quota of 5,000, and a Droop quota of 4,762, a difference of 238. If the number of seats is 30, the Hare quota is 3,333, and the Droop quota is 3,226, a difference of only 107.
As the number of valid seats decline, the difference in quotas increase, and it is at its most stark when the formulas are applied to a single-member ward, which is the test case. For our 100,000 voters competing for one representative (in a single ward) the Hare quota is 100,000, and the Droop quota is 50,001.
More generally, for any number of voters in a single member ward the Droop quota is always 50% of the votes plus 1. This is the minimum absolute majority, and guarantees victory for any candidate who gains it.
The Hare quota therefore does not meet the criterion of finding the smallest number of votes that ensure a winning result, but the Droop quota does, and is therefore the superior method. The Hare quota in any single-member ward is also non-sensical, even with the mathematically correct calculation being executed.
The formula for establishing quotas in our local government system is neither the Hare nor the Droop method. When the Local Government Municipal Structures Act was first published in the Government Gazette no. 19614 of 18 December 1998, it contained a formula (schedule 1, paragraph 12), which has been retained in every subsequent amended version of the Act:
Total number of valid votes cast for parties only
Quota = --------------------------------------------------------------- + 1
(total number of seats in the metro or local council, minus seats won by independents)
In our imaginary test case of a single member ward with 100,000 votes cast the quota derived by this method (let’s call it the official formula), would be 100,001. This result is even more non-sensical than that produced through the Hare quota, and is wholly illogical.
As it stands in writing in the Local Government: Municipal Structures ACT, no. 117 of 1998, as amended, the formula for establishing quotas, is therefore incorrect, and not the Droop quota formula (see Schedule 1, paragraph 12, 16 and 17 of the Act). This incorrect formula has also been dutifully used by the Independent Electoral Commission, up to and including the 2021 elections ( see here):
The 2021 elections illustrate a number of the features described above. First, the difference in results calculated by the Hare method, the official method, and the Droop method is small in the case of large metros. From the data sets of the IEC, for example, the results of the election in the Nelson Mandela Bay metro are as follows:
A total of 523,343 valid votes were cast for all parties, 120 seats were contested, no independent candidate won a seat, and no ward seats were won by parties without PR list. The official quota is then calculated as a straightforward procedure: 523,343 divided by 120, with the result +1 = 4362. The Droop quota entails dividing 523,343 by (120+1), with the result +1 = 4326. The different quota value produced through the Droop formula did not have a large enough impact on subsequent calculations to affect the allocation of seats to any party in this metro.
As expected, in smaller municipalities the quota differences were relatively larger. In Prins Albert, for example, 7 seats were contested by 11,202 voters. The official quota was calculated at 1 601, and the Droop quota at 1 400 (fraction discarded). Twelve parties took part in the contest, with three of them gaining seats in the first round of allocation. The procedure for seat allocation goes as follows: the number of votes for each party is divided by the quota, with the votes remaining, (the “remainder”) being expressed as a fraction of 1.
The Karoo Gemeenskapsparty, for example, drew 2,467 votes, divided by 1601 gave them one seat, with a remainder of 0.5409. In a similar calculation for every other party, the DA with 4 194 votes took two seats, with a remainder of 0.6196, and the Patriotic Alliance’s 1 607 votes converted into 1 seat with a remainder of 0.0037.
In the second round of allocation, the ANC, with the largest remainder (0.7583) gained its first seat in the council, with the Democratic Alliance and the Karoo Gemeenskapsparty also each gaining one more seat.
Applying the Droop quota does change the rank order within the top three partys’ remainders, but this does not change the seat allocation, as these top three parties concentrated the bulk of the votes. (The ANC took the fourth most votes, 1214, and the party with the fifth most votes, the Karoo Democratic Force, took only 520 votes).
Only a full audit of all the 2021 local government election results throughout the country will show if the official quota formula produced a seat allocation that is different to that of the Droop quota. As it stands, however, the current election is run under a quota formula that is suspect, embarrassing and without merit.
To conclude, the Droop quota needs to be applied correctly at the local government level to ensure the validity of results. The Droop quota is currently in force at the national level, and there is no rationale for maintaining this deviant official formula at the local level.
The immediate remedy is not to declare the 2021 local election results invalid. Nor does the election need to be rerun. The current result can stand, and need not even to be recounted, but the apportionment of seats has to be recalculated according to the correct Droop quota.
In order to make this possible, an appropriate amendment to the Local Government: Municipal Structures Act needs to be made as a matter of urgency.
Mutual accommodation and cooperation.
The remaining issue is one that should be of the uppermost concern to all stakeholders, but is rarely even mentioned: that the electoral system should provide rules that create incentives for parties to cooperate with one another prior to elections, in such a way that they can potentially gain seats.
The rule of apparentement is used in some democracies, notably Israel and the Netherlands. It entails that (two or more parties) can reach a formal agreement (which is printed on the ballot paper) to pool the votes they receive as remainders, in order to increase their combined vote for capturing another seat on the second distribution.
In the case of Prins Albert, had the Karoo Democratic Force been able to combine their remainder under an apparentement rule (520 votes, 0.3248 remainder, using the official quota) with that of the United Community Front (512 votes, 0.3198 remainder, again using the official quota), they would have topped the Karoo Gemeenskapsparty (0.5409 remainder) in the rank order of remainders, and would have captured one seat between them, with implications for the distribution of power in this town.
To incorporate this electoral device into a new national electoral system provides at least some incentive for parties, many of whom may end up in the same ruling coalition, to temper their election campaigns, something sorely needed in our present political arena.
Pierre du Toit
Emeritus Professor
Department of Political Science
Stellenbosch University.
