Johannesburg - The election in Johannesburg is shaping up as a tight affair. In 2014, the ANC received a historically-low 52% of the vote in Johannesburg, setting the metro up as a major swing area for the 2016 local government elections.
So what will happen this year?
The only credible public polling on Johannesburg is IPSOS’ running tracking poll, which currently has the DA leading (36% for the DA to 31% for the ANC). The poll includes a very large number of undecided former ANC voters – suggesting that the outcome could look markedly different than the current polling average.
Another way to assess potential election results in Johannesburg is to look at previous voting patterns. An analysis of the latest registration figures and the voting patterns from 2014 and 2011 suggests that the ANC could be on the verge of being pushed below 50% in Johannesburg.
The Johannesburg Electorate
After the latest round of registration weekends, Johannesburg has a total of 2.1 million registered voters and about 835 voting districts (VDs). These voting districts can be divided into 5 distinct sub groups:
1. Solid ANC VDs (where the ANC receives more than 70% of the vote)
2. VDs that lean towards the ANC (where the ANC receives 50-70% of the vote)
3. Marginal VDs (where a party wins with less than 50% of the vote)
4. VDs that lean towards the DA (where the DA receives 50-70% of the vote)
5. Solid DA VDs (where the DA receives more than 70% of the vote)
These five groups of VDs have the following characteristics:
But despite this advantage, the ANC faces some major problems that threaten its ability to reach 50% in the metro.
The ANC’s 3 big problems in Johannesburg
Problem 1: Voter registration
The DA significantly out-registered the ANC during the 2016 registration weekends. The total voters roll grew by roughly 30 000 new voters in the DA base areas, whilst the total number of registered voters in ANC areas was essentially unchanged.
This outcome is significant for two reasons: Firstly, it means that the DA base grew relative to the ANC base, creating a slightly more favourable overall electorate for the DA. But secondly, it could indicate that DA supporters are much more motivated to get to the polls in this election than ANC supporters are.
Problem 2: The EFF
The ANC’s got a very serious problem on their hands with the EFF in Johannesburg. Comparing the 2011 to 2014 results, we can see that the EFF siphoned off a huge number of votes from the ANC in ‘solid ANC’ and ‘lean ANC’ VDs. Consider these results:
In both the ‘solid ANC’ and ‘lean ANC’ areas, the ANC suffered significant losses, almost exclusively to the EFF (the DA did make minor inroads in each case). Although the ANC still won in these areas, it won them by smaller margins - reducing the overall ANC vote count significantly.
The EFF’s impact on the ANC also has an important geographic element to it. The ‘solid ANC’ VDs (where the ANC only lost about 10% to the EFF) are almost exclusive in Johannesburg’s Southern townships (Soweto and Orange Farm). It is interesting to note that the ANC is more resilient in these Southern townships.
The ‘lean ANC’ VDs (where the EFF had a massive impact on the ANC by taking about 16% of the vote from the ANC) are almost exclusively in the Johannesburg CBD and Johannesburg’s Northern townships (Diepsloot, Alexandra and Ivory Park).
The divergence in the ANC performance between these regions is a very important development, and it suggests that both the EFF and the DA’s best growth prospects are in Johannesburg’s CBD and Northern townships.
Problem 3: Turnout
For the last two elections, the ANC has struggled to get voters in its strongest areas to the polls, whilst the DA has done relatively well to get its supporters to the polls. This creates a significant ‘turnout gap’ between the DA base and the ANC base, inflating the DA’s overall percentage of the vote and deflating the ANC’s overall share. Here is the evidence:
Turnout in DA areas far exceeded turnout in ANC areas in both elections, but the ‘turnout gap’ was particularly pronounced in 2011 (the last local government election). If the pattern repeats itself, it could spell trouble for the ANC.
- Election map: Previous ANC results
Expectations for August
So the ANC faces some challenges this August: The turnout gap is likely to look similar to the one in 2011, the makeup of the overall electorate is slightly more favourable to the DA, the EFF is set to again take a chunk out of the ANC vote and the DA is set to grow its share of the vote in the townships if the IPSOS poll is to be believed.
Simulating an election outcome using the latest registration figures, the 2011 turnout gap and the 2014 voting pattern (so for the moment assuming no new growth for the DA or EFF), produces the following result:
The result shows how close the ANC is to the 50% mark. Even without any DA or EFF growth, the party is just barely hovering over 50%. The critical question of course is whether the 2011 turnout pattern will change and whether the DA or EFF can make further inroads in ANC areas.
I modelled a plausible scenario using the following assumptions:
1. Turnout in ANC areas remains the same as in 2011
2. Turnout in DA areas increases by 5%
3. The DA grows marginally in the townships whilst the EFF remains stable (a conservative assumption given the IPSOS polling)
4. The DA consolidates the opposition vote further
That simulation produced the following result:
This is a conservatively favourable scenario for the DA. Any number of other changes could take place that amplifies the DA’s growth (like huge DA base turnout or significant DA growth in strong ANC areas). Conversely, there are also plausible developments that could occur in favour of the ANC. ANC base turnout in particular is an important variable with upside potential for the ANC given that it was very low in 2011 (55%) and that it could edge higher this time around as voters will be aware of the high level of contestation in the metro.
The model suggests that the ANC is very likely to be the largest party in Johannesburg, but that it could easily be pushed below 50% (and even well below 50% in certain scenarios). The conservative scenarios that were tested generally produced results where the ANC falls just below or just over 50% in a range from 45% to 52%.
It is important to remember though that the simulations described above are based on (conservative) assumptions. Dramatic and unexpected things can happen on Election Day, especially if turnout looks markedly different than 2011.
- Dawie Scholtz holds an MBA from UCT and is currently studying towards a Master's Degree in advanced management at Yale University. He is a former DA employee.
- Find everything you need to know about the 2016 Local Government Elections at our News24 Elections site, including the latest news and detailed, interactive maps for how South Africa has voted over the past 3 elections, or download the app for iOS and Android.