The 2016 Local Government Elections and the Metros - Part III: Swings and Roundabouts

Before proceeding, it is important to make clear the difference between a prediction and a projection.  Nothing in these three briefs should be interpreted as a prediction of what is going to happen.  Publicly available polls of party support are as rare as hen’s teeth in South Africa.  Moreover, it is standard for polls internationally to be precise only up to about 3% either way.  And patterns of support can change rapidly in the period immediately prior to an election, as the recent United Kingdom elections demonstrated.  In the movie A Night at the Opera, one of the Marx brothers observed that ‘it’s not over until the fat lady sings.’  The same is true of elections, the role of the fat lady in South Africa being taken by the Independent Electoral Commission (IEC) and the singing the final announcement of results.

Nonetheless, projections based on clearly defined assumptions can be helpful in exploring the logic of a situation and as a basis for considering qualitatively the impact of a wider range of circumstances.  Both parts of the exercise are presented here.

Quantitative Projections

A baseline scenario was presented in the second brief.  The approach taken here to modelling changes in party support is as follows:

• Start with the ANC, the majority party in the country as a whole.

• Consider the effect of transferring votes (relative to the no swing baseline) towards and away from the ANC.  If votes are transferred towards the ANC, assume that these votes come from the DA and EFF in proportion to their support in the baseline.  If votes are transferred away from the ANC, assume that the DA and EFF benefit in the same proportions.  The support for other parties is assumed to remain constant.  

Along with ‘no independents’ assumption, these assumptions are sufficient to yield determinate outcomes. Fifteen assumptions can then be modelled: the consequences of a 2%, 4%, 6%, 8%, 10% and 12% increase in ANC support relative to the baseline, a 2%, 4%, 6%, 8%, 10% and 12% decrease in ANC support, and the baseline scenario.

The results of these projections are presented as follows: the possible outcomes are an ANC majority, a DA majority or the necessity for coalition government, since neither the ANC nor the DA would command a majority.  On the projection assumptions, and with the range of swings selected, there is no chance that the EFF will be able to command a majority in any metro, and the same is true a fortiori for smaller parties.  Reported also, is the total number of seats in the nine metros which would be commanded by the ANC, DA and EFF.  

The outcomes projected are as follows:

On the projection assumptions and the swing ranges, the projected results are that the ANC takes four metros – Buffalo City, Ethekwini, Mangaung and Midvaal-Emfuleni.  The DA takes one – Cape Town.  Outcomes in the remaining four metros are subject to the swing assumptions.  In the baseline case, and if the ANC loses support, Nelson Mandela will be governed by a coalition.  If the decrease in ANC support reaches 2%, Tshwane will be governed by a coalition.  If the decrease in ANC support reaches 8%, Johannesburg will be governed by a coalition.  If the decrease in ANC support reaches 12%, Ekuruhleni will be governed by a coalition.  Under other swing assumptions, Nelson Mandela, Tshwane, Johannesburg and Ekuruhleni will be ANC governed.

The allocation of the 1579 seats in all the metros would be as follows:

Suppose, for the sake of argument, that a former DA supporter becomes unwilling to vote for the DA.  In this case, she has two options:

• Registering a vote for another party, which is the assumption behind the projection above.

• Deciding that while she cannot vote for the DA, she cannot bring herself, for historical or cultural or whatever reasons, to vote for the ANC, EFF or a smaller party.  In this case, she will abstain from voting.  Her abstention hurts the DA, but not as much as if she shifted her vote to another party.  

There will be people who voted for a party in the national elections of 2014, but who will be disenchanted with all the options on the table in 2016.  Parties may be differentially affected by this disenchantment.  Aggregate disenchantment will lower the voter turnout, already considerably lower in municipal elections than in national elections, because the stakes are lower.

Coalitions will form around the ANC and the DA.  There is no reason for grand coalitions between the ANC and the DA anywhere.  The party with the larger number of seats will have an advantage in forming the core of a coalition, since it will need to reach agreements with fewer partners.  In all cases apart from a 12% decrease in the ANC vote in Nelson Mandela, the projection assumptions yield the result that the ANC will have the greater number of seats.   However, the table above also indicates that the EFF will be the elephant in the coalition room, especially in the three existing Gauteng metros – Ekuruhleni, Johannesburg and Tshwane – where they established themselves most strongly in the 2014 national election.  Three possibilities present themselves.  Either the EFF starts to fade, or it maintains its present support base, or it strengthens it, particularly outside Gauteng.  In the latter two cases, the EFF would have to decide what it wants to do wherever coalition government becomes necessary.  The basis for such decisions are not currently visible, but they may be critical in some metros.  

Charles Simkins

Senior Researcher

charles@hsf.org.za