Tshwane: When factionalism leads to fracture. Is the ANC's decline in the Metro due to a voter stay-away or changing voter support?


Ebrahim Fakir and Elnari Potgieter | 18 August 2016

Prior to the 2016 South Africa local elections, conflict related to the ANC's mayoral candidate led to serious protests in particular townships, all traditionally strong ANC supporting areas, in the Tshwane metro. This was precipitated by serious internecine ANC internal factionalism in which the ANC's regional chair was pitted as rival against his deputy. Both were nominated by the different regional branch structures of the ANC as mayoral candidate nominees'. In a bid to avoid a factional battle within its own ranks - both in terms of electoral support, but also in any likely post-election governance regime, the ANC's purported solution was to find a "compromise" mayoral candidate nominee.

The party settled on a black woman, Thoko Didiza with sufficient administrative and management experience (a former cabinet Minister) who would both suite the candidate criteria the party defined as standards for its nominee's, but more importantly, was someone who rose above the factionalism that appears to have become endemic within the regional structures of the Tshwane ANC. This did not sit well with the regional party structures - who in open rebellion against a set of nomination rules mutually agreed upon by the party - embarked on a five day protest culminating in millions of Rands of damage to public and private property, and five unfortunate deaths. For the ANC, and the regional ANC in particular, these events had a disastrous impact on its electoral performance.

Did this cost the ANC its clear majority in the Tshwane Metro Council?

Evidently, in the lead up to the elections, the ANC's own members and core support base in Tshwane did not accept what they viewed as the imposition of a "neutral" or compromise candidate as Mayor nominee. They suggested that they would "punish the ANC" through a "stay-away" from the polls or for voting for an opposition party. This was no idle threat. It is what they did.

But the stay away factor has perhaps, been overstated especially in the traditionally ANC supporting areas (what we call the five hotspots) of Atteridgeville, Soshanguve, Mabopane, Mamelodi and Hammanskraal, where voter turnout remained similar to turnout levels from 2011, and in some cases even increased. This is counter to the pre-election intuitive expectation that there would be dramatically lower voter turnout in these areas compared to 2011, due to the ANC's factionalism and the threat of ANC voters to stay away.

Voter Turnout


Figure 1: Tshwane Voter Turnout (2011 compared to 2016)
Figure 1: Tshwane Voter Turnout (2011 compared to 2016) [Click for larger image]
Source: IEC results accessed through News24 election website and IEC results website. Please endnotes 1 & 2 on how data for the five areas and for "DA Supportive" was calculated.

Figure 1 shows an increase of 2.31% in Voter Turnout in the five identified areas between the 2011 and 2016 local elections. This is less than the increase in Voter Turnout for the metro as a whole over the same period (3.88%), and consistent with the increase in voter turnout in the Gauteng Province (2.20%) as a whole. Of the five areas, Hammanskraal and Atteridgeville saw the largest increases in the Voter Turnout for this period (3.68% and 4.71% respectively).

Contrary to expectations, more residents in the five identified traditionally ANC dominated areas did not "stay-away" during the 2016 elections as a result of the internal ANC conflicts regarding the ANC's mayoral candidate, but rather more turned out to vote than in the previous local elections. The voter turnout in these areas, however, is lower in both the 2011 and 2016 local elections than the metro as a whole.

In areas traditionally more supportive of the DA, the average Voter Turnout in 2011 was 64.17%. All of the wards included had a Voter Turnout of more than 64%, except ward 56 (with a Voter Turnout in 2011 of just under 50%). The average Voter Turnout for DA supportive areas increased by 7.6% to 71,77% - a bigger increase than for the metro as a whole (3.88%). Most wards included here had a Voter Turnout of more than 72%, except ward 83 and 56 (the latter of which had a Voter Turnout percentage of just over 58% in the 2016 elections). The average Voter Turnout for these areas is 12.57% higher than the Voter Turnout for Tshwane as a whole, and 15-20% higher than the Voter Turnout in the five traditionally ANC supporting township areas.

Ward 56 of Tshwane is an interesting case (see table one below under Tshwane East) , given that the ward covers areas of transient spaces of the city - such as Sunnyside and Hatfield (where immigrants, internal migrants and students often temporarily reside), as well as more middle-class and affluent residential areas such as Muckleneuk, Brooklyn and Groenkloof. The Voter Turnout for this ward for both the 2011 and 2016 elections was at a level lower than that for the Metro as a whole.

Party Support Patterns

In the 2016 local government election, a trend analyses shows that there has been an incremental but sustained shift in votes towards the DA and the EFF, in particular in wards of the five areas we closely examine. This has not had an impact on the ANC's ability to win the wards in these specific areas, but it did result in a higher percentage of aggregate proportionate support for opposition parties (in particular the DA) in the city as a whole - which in turn will have an impact on the total number of seats the ANC acquires given the PR top-up system. In other words, the DA - while it has made incremental and sustained gains in Tshwane Metro - has not won (m)any wards in the traditional township areas of Tshwane. Wards which the DA did win in traditional township areas are those in voting districts which spatially border other areas (in the below wards shaded - ie: 3 and 55 in Atterdigveille and 60, 80 and 81 in Central and 100 in Nokeng), which have very high levels of DA support. In African townships - the DA managed to win two wards in Soshanguve, two in Atteridgeville and none in Mabopane or Mamelodi).

The tables below amply demonstrates this.

Table 1: Ward Winners, Tshwane 2016
Table 1: Ward Winners, Tshwane 2016 Source: Ward Results as tabulated from the IEC.
Table Two: Number of wards won per party - Tshwane Metro
Table Two: Number of wards won per party - Tshwane Metro
Source: Ward Results as tabulated from the IEC.
Figure 2: Tshwane, Party Support 2011 compared to 2016
Figure 2: Tshwane, Party Support 2011 compared to 2016 [Click for larger image]
Source: IEC results accessed through News24 election website, Wazimap website and IEC results website. Please see endnotes 1-3 on how data was calculated.

Support for the DA in traditionally DA supporting areas was 46.56% more than in the Metro as a whole in the 2011 local elections (at 85.21%), and 43.50% more than for the metro as a whole in the 2016 local elections (at 86.61%). There was a slight increase (1.4%) in support for the DA in these areas. Average residual ANC support in the DA supportive areas dropped by 3.26% from 9.89% in the 2011 local elections to 6.63% in the 2016 local elections. Both of these are less than the support for the ANC in the metro as a whole. The average support for the EFF (2-3%) in these areas is much lower than support for the party in the metro as a whole (around 11% in the 2014 national and 2016 local elections).

Ward 56 is an interesting case, with a slight decrease in support for the DA from 2011 (81.48%) to 2016 (79.03%), and more support for the EFF than other DA supportive wards (6.21% in the 2016 local elections), but remains a DA area with a high level of support for the party despite lower turnout than is the case in other DA supportive areas. This might be attributable to the mixture of areas the ward covers, and in which voters turned-out (i.e. DA supporters in the area turned-up to vote more than ANC / EFF and other party supporters).

ANC support in the five areas dropped on average with 20.34% from the 2011 to the 2016 local elections. This is much more than the drop in support for the ANC in the metro as a whole (14.10%), Gauteng (13.81%) and South Africa (8.04%).

Support for the DA increased in the five areas with an average of 4.21% from the 2011 to the 2016 local elections. This almost consistent with the incremental and sustained increase in support for the DA in the metro over the same period (4.46%), and more than the increase in support for the DA in the province (3.78%) and South Africa (2.95%) as a whole. Hammanskraal has the highest levels of support for the DA among the five areas, but support remained static at just over 15% between the 2011 and 2016 elections. Support for the DA grew moderately with just over 2% in Atteridgeville, and with over 5% in Mabopane and Mamelodi. Soshanguve shows the most growth in support for the DA, with a 7.69% increase in votes from the 2011 to 2016 local elections.

Support for the EFF in the 2016 local elections could only be compared to support for the EFF in the 2014 national elections (given that the party was founded in 2013). Some valuable insights can be derived from using the available data. Support for the EFF increased between 1-2% in the five areas from the 2014 to 2016 local elections. This is more than the increase in support for the EFF in Tshwane over the same period (0.7%), and consistent with the increase in support of the EFF in Gauteng (just over 1%) and South Africa (2%).

The decrease in support for the ANC in the five identified areas (on average 20%) can partly be explained by the increase in support to opposition parties - mainly the DA (on average 4%) and EFF (between 1-2%) - but not fully. The remainder of the drop in support for the ANC can possibly be explained by ANC supporters staying away on voting day, and opposition supporters turning-up to vote. Further inspection of the number of votes cast in support of each party, as well as in total, in the five areas during the 2011 and 2016 local elections provides more clarity in this regard.

Figure 3: Number of Votes cast in Five Identified Areas - DA, ANC and EFF (2011 compared with 2016 local elections)
Figure 3: Number of Votes cast in Five Identified Areas - DA, ANC and EFF [Click for larger image]
Note: All figures in the table above represent the number of valid votes cast in the respective year and in support of the respective party per category. *Ward and PR votes. Source: IEC election results website. Please see endnote 1 on how data for the five identified areas was calculated. Please remember that ward boundaries changed from 2011 to 2016, but the combination of wards constituting the five respective areas were comparable.

In all five areas the number of votes cast (PR and Ward) increased from 2011-2016, except in Hammanskraal where votes cast decreased. This is also the only area where DA support did not increase significantly from the 2011 to 2016 local elections.

Interestingly though - the whole number of votes cast for the DA shows an almost two fold (and in the case of Soshanguve three-fold) increase for the DA, though in each one of the township areas the EFF receives more votes than the DA (except in Hammanskraal).

In the context of stable levels of overall voter turnout in these townships compared to 2011, with marginal increases in voter turnout in some areas, these changes are attributable to three possible factors, all of which should be worrying for the ANC.

First, even accounting for new or first time voters on the voters roll, the whole number of votes cast show increased levels of support for both the DA and EFF. Second, the EFF appears to be growing at a faster rate in township areas compared to the DA. Thirdly, along with this trend - the whole number of votes cast for the ANC demonstrates a definitive and consistent decline in each one of the township areas examined.

Tshwane Metro Party Support and Assigned Council Seats

The figures above show that the five identified areas remain as "ANC areas" and wards, but the drop in ANC support and stay-away votes in these traditionally ANC areas had an impact on the proportion of votes for the ANC in the metro as a whole. This, coupled with a higher turnout in more suburban areas - most of these with more support for the DA (and evidence of a drop in residual ANC support in these wards), coupled with support for the EFF in both traditional ANC and DA wards - contributed to the DA's marginal win over the ANC in the metro.

The results and distribution of seats based on the hybrid electoral system where ward ballots cast for a party are counted towards its PR tabulation (ie: adding the ward votes to the PR votes cast for that party - a fair system that reflects the total aggregate proportion of support that a party receives) - bulks up the DA's ward wins with the additional seats gained from its PR tally. This is reflected in figure 4 below.

Figure 4: Tshwane Assigned Council Seats (2011, 2016)
Figure 4: Tshwane Assigned Council Seats (2011, 2016)
Source: IEC election results website.

Conclusions

Though there was a stay away in Tshwane - its impact has been overstated. Turnout levels in the traditionally ANC supporting areas (what we call the five hotspots) of Atteridgeville, Soshanguve, Mabopane, Mamelodi and Hammanskraal remained similar to turnout levels from 2011 and in some cases, even increased. The suburban areas, traditionally DA supporting - however saw a spike in voter turnout. Data analysis from previous elections showed that if there was a stay away in the areas that would traditionally boost the ANC's percentage share of support in the metro - even without the opposition parties necessarily making big inroads into the ANC vote - the effect on the ANC is that opposition votes would carry a proportionately higher weight. This is what occurred in this election.

Coupled with very high turnouts and sustained consolidated spikes in support in traditional DA wards, consistent drops in support for the ANC as well as sustained but dispersed, support for the EFF - are what saw the marginal eclipse of the ANC.

The EFF basically received its seats primarily on the PR allocation and won no wards in the Metro. Though there are incremental but sustained marginal shifts away from the ANC - much of its loss is attributable to its internecine internal factional problems leading to voter stay-away, aggregate proportionate average gains for the DA due to spikes in voter turnout in traditional DA supporting wards. While the EFF won no wards, its average aggregate proportionate increase of 2% comes off the back of a 20% lower turnout in the local elections compared to the national election. But the inability of opposition parties to capitalise on the ANC weaknesses and factional in fighting shows that while the DA has sustained incremental increases, may yet see the ANC be able to mount a recovery, failing which the EFF is likely able to capitalise - depending on its performance between this election and the next. Perhaps the most telling story is that the DA has been unable, to any significant degree, capture wards in traditional African townships in spite of its sustained increases in the face of the ANC's factional disunity.

The ANC on the other hand, appears not to have made any significant outreach, and inroads into the suburban areas. Current results patterns may provide a disincentive for the ANC to try and reach out to voters in these area's - but a lack of influence amongst high net worth, highly sophisticated and mobile voters connected with global networks of commerce, industry and government can prove detrimental for a national governing party. It may also point to weaknesses in the ANC's ability to instrumentalise one of its key principles - that of "non-racialism", or failures in its governance strategies of catering to the needs of this constituency. One view might also attribute the ANC's lack of penetration in these voting constituencies as a result of its national governance failures and ethical scandals, while others migh attribute this to the lingering mistrust of a black majority Government. Whatever the real reasons - these ought to be concerns that a national governing party needs to urgently address.

The Local Government 2016 elections in Tshwane, thus tells a story of continuity and change.

Suburban voters in Tshwane remain relatively homogenous in their voting preferences, while much more interesting swings and dynamic voting patterns appear to be emerging in the Townships. While renewed and vigorous voter turnout levels in suburban areas is to be celebrated, since they indicate a continued interest in the political and governance system. It is worth noting however, the relatively limited range of their political preferences - split primarily between the DA, sustained continued support for niche special interest parties like the Freedom Front Plus (FF+) and evidence of some residual support for other parties, including the ANC (though this was in decline in this election).

Metro voters in the townships however, provided a much more interesting demographic - demonstrating a much more sophisticated calculus - showing some continuities in support for the ANC (they won almost all the wards in the Township areas after all), with new dynamics and shifts in voting patterns for both the main opposition the DA and the emergent EFF, with residual but declining support for other special interest parties with a (small) portion also exercising the preference not to vote.

Endnotes

  1. The values for Atteridgeville, Soshanguve, Hammanskraal, Mabopane and Mamelodi are averages and were calculated by using the results for the wards in the areas. Ward boundaries do not necessarily follow the borders of places / communities / townships and ward boundaries may change from one local election to another. For this reason, a combination of wards - as with the 2011 South Africa municipal elections - were identified, which roughly represents the respective places / areas. These were then compared with the wards of the 2016 municipal elections to ensure compatibility. Please take note that some of the identified wards may partially fall within a specific area, but that the same ward may also fall within another area. The figures here thus roughly represent the % Voter Turnout and % Support in specific areas this study investigated, but not perfectly. The sample of Tshwane wards included in the calculations (which we referred to as hotspots) for each of the five areas are: Atteridgeville - wards 51, 62, 63, 68, 71, 72; Hammanskraal - wards 49, 73, 74; Mamelodi - wards 6, 10, 15, 16, 17, 18, 23, 28, 38, 40, 67, 97; Mabopane - wards 12, 19, 20, 21, 22, 29; Soshanguve - wards 11, 13, 25, 26, 32, 33, 34, 35, 36, 37, 39, 88, 89, 90, 94.
  2. The DA supportive areas included the wards covering the following areas (or part of the following areas): Waterkloof, Groenkloof, Lynnwood, Monumentpark, Faerie Glen, Garsfontein, Erasmuskloof, Elarduspark, Moreletapark, Die Wilgers (to the South-East of the city centre), and Eldoraigne and Wierda Park (to the South/South-West of the city centre). The wards included in the calculations include: 42, 44, 46, 47, 56, 69, 70, 82, 83.
  3. The EFF was not yet formed during the 2011 local elections. We have thus used support for the EFF during the 2014 national elections in Figure 2 to identify possible trends in support.