GARETH VAN ONSELEN: What the ANC’s 2016 election performance means for 2019

Warning: This column is primarily targeted at number junkies. If you would like to skip all the electoral math, head straight to the conclusion at the end.

The 2016 election, in which the ANC’s vote share collapsed to 53.9%, has been widely used as the backdrop to the upcoming 2019 election and as the frame of reference for the ANC’s fortunes.

The argument goes that in his second term, the many problems with Jacob Zuma and his administration came together in the form of an internal and external crisis for the ANC and, as a result of that pressure, the ANC’s vote imploded and the party lost control of three key metros. Thus, if the ANC were to prevent itself from falling below 50% in 2019, Jacob Zuma had to go. This view was widely shared, inside and outside the party.

ANC chief whip Jackson Mthembu put it like this in March 2017: “If we fail to show the people of SA that we are able to self-correct, and that from December up to 2019 we will do things differently in their interest … I am afraid come 2019 we will also lose at national level.”

But is it true that, had Jacob Zuma stayed, the ANC would have inevitably lost in 2019? Let us look at what the hard numbers say.

Before we do that, a quick methodological note: first, this analysis will exclude the 1994 national and 1995 local government elections, as the way voter registration took place over that period was subsequently reconstituted and, from the 1999 national election onwards, we are better able to compare and contrast results.

Second, with regard to the opposition, the votes of all opposition parties have been amalgamated into a single number throughout this analysis.

Third, all numbers in this analysis are drawn from the Electoral Commission of SA's (IEC's) website and, where local government results are concerned, only the proportional representation (PR) ballot was used as this is the best indicator of national party political support (it differs slightly from the final result — in 2016 for example, the ANC finished with 53.9% but on the PR ballot alone, it got 54.5%).

Understanding turnout

Part of the problem with using the 2016 election as a frame of reference, is that voting behaviour in local government elections is profoundly different to voting behaviour in national elections.

Here are the turnout levels for each election since 1999:

Table 1: Turnout

As you can see, voters simply don’t vote in local elections (between 47% and 57% of registered voters turn out on election day) to nearly the same extent they do in national elections (where between 73% and 89% of registered voters turn out). But — and this is an important but — lower turnout in local government elections does not affect all parties the same way, and the opposition uses this to its advantage.

To illustrate this, let us look at the differential, in terms of absolute votes, for the ANC, between national elections and local elections.

The ANC’s improving local government performance

Table 2: The ANC’s vote share

A few patterns are immediately obvious: the first is that, in absolute terms (that is, in terms of actual votes received) the ANC fares far better in national elections than it does in local government elections: 1999: 10.6-million votes; 2004: 10.8-million; 2009: 11.6-million and 2014: 11.4-million. These numbers are far, far higher than the number of votes it manages in local government elections: 2000: 5.2-million; 2006: 6.4-million, 2011: 8.4-million and 2016: 8.1-million.

If you subtract the number of votes the ANC receives in a local government election from the respective national election that preceded it, you get the following differential: In the 2000 local government elections, 5.3-million fewer ANC voters voted for the party than did in 1999. In 2006, 4.4-million fewer; in 2011, 3.2-million fewer and in 2016, 3.3-million fewer. You could call this, the ANC’s “apathy gap”.

This should already be of some interest. The ANC’s 2016 result was actually fairly typical and by no means an aberration in this regard. In fact, the ANC had lost more votes between 1999 and 2000 (5.3-million) and between 2004 and 2006 (4.4-million) than it did between 2014 and 2016. Indeed, it was, over time, actually closing the gap between its performance in national and local government elections.

This is best illustrated in percentage terms. The 5.3-million ANC voters who stayed away in the 2000 local elections constituted 50.2% of the 10.6-million votes the ANC won in 1999 national elections. By 2006, the 4.4-million ANC voters who stayed away constituted 40.5% of the 10.8-million votes the ANC won in 2004.

In 2011, the 3.4-million ANC voters who stayed away constituted 27.8% of the 11.6-million votes the ANC won in 2009. And, in 2016, although slightly up, the 3.3-million ANC voters who stayed away then constituted just 28.9% of the 11.4-million votes the ANC won in 2004.

So, actually, in historical terms, its 2016 result was not that bad, at least in terms getting ANC voters to vote in local government elections — something the party has historically been much worse at.

How does this compare to the opposition?

The opposition’s turnout brilliance

Table 3: The opposition vote share

Again, some patterns are obvious. First and foremost, the inverse of that applies to the ANC: the opposition performs better in local government elections than it does in national elections. Not in terms of absolute numbers, but in terms of its percentage of the vote.

How does it do this? The answer lies in the fact that proportionally, far fewer opposition voters stay away between national and local government elections.

Whereas 52.2% of the ANC voters in 1999 stayed away in 2000, for the opposition the comparative percentage is 35.9% (two-million opposition voters stayed away). That represents a 16.3 percentage points advantage for the opposition.

In 2006, 40.5% of ANC 2004 voters stayed away; the comparative percentage for the opposition was 32.1% (1.6-million voters). That’s an 8.4 percentage point advantage for the opposition. And in 2011, 27.8% of 2009 ANC voters stayed away; the comparative percentage for the opposition was 21% (1.3-million) — a 6.8 percentage points advantage (the smallest it has ever been).

But, in 2016, the opposition outdid itself. For the ANC, in 2016, 28.9% of 2014 ANC voters stayed away but for the opposition, a only a remarkable 5.9% (431,256 voters) stayed away, giving it a huge 23 percentage points advantage over the ANC.

And that is what made all the difference. In simple terms: the opposition — led primarily by the DA — was able to get its national support base to vote in far greater numbers than the ANC was able to in the 2016 local government election. And even though the ANC had been getting better at closing the gap between its ability to so between national and local elections, in 2016 the opposition simply blew it away.

The case is even better made in Gauteng.

Table 4: Gauteng – ANC vs DA

The same national pattern is evident: the ANC was closing the gap between its national and local government election performances — 46.2% (2000) down to 37% (2006) down to 30.3% (2011). But, as happened nationally, the opposition was still managing to get more of its national voters to the polls in local government elections — 27.4% (2000) down to 21.1% (2006) down to 18.9% (2011). However, its advantage over the ANC was shrinking: from 18 percentage points (2000), down to 15.9 (2006), down to 11.4 (2011).

All of that changed in 2016. The ANC managed to hold its differentiation at around 30% but the opposition differentiation was next to negligible (6%). And so, the whopping 24.5 percentage points difference meant the opposition vote surged in urban centres.

The ANC’s vote share is declining

Importantly, none of this means the ANC’s vote, in and of itself, is not declining. You will notice both from the national and Gauteng numbers, the ANC’s share of the vote remains relatively consistent. It fluctuates between about 10.6-million and 11.6-million in national elections. In Gauteng it is more static still, hovering between 2.3-million and 2.5-million votes. Likewise, it fluctuates only in a small band of between 1.3-million and 1.8-million votes in local government elections, in that province.

But staying still in politics is a problem. Every year the voters' roll increases (the number of voters registered to vote). And if you stay still in terms of votes, you are inevitably losing, because more and more people are voting each year and the electoral pool getting bigger and bigger. Thus your static number of votes becomes worth less and less.

Here are two tables that illustrates that:

Table 5: How the opposition out-registers the ANC

Table 6: Relative decline of the ANC’s vote share

Table 5 demonstrates how the ANC’s vote share fails to grow in national elections compared to that of the opposition. Between national elections, the voters' roll grows by more or less 2.5-million new voters every five years. While the ANC did manage to turn some of those new voters into new support in 2004 (279,585 new votes) and 2009 (769,833 new votes), in 2014 it actually lost some hard support (down 213,825 votes).

By comparison, the opposition, after a poor performance in 2004 (down 644,489 votes), has relatively surged in the last two elections. Up in 2009 (by 1,286,575 new votes) and 2014 (948,632 votes). That differential means the ANC’s relatively static vote share has dropped: from 69% to 65% to 62% in national elections, while the opposition’s has grown: from 30% to 34% to 38%.

Table 6 demonstrates the point in a different way. It shows quite beautifully how the ANC’s vote share, as a percentage of all registered voters (as opposed just those who vote — “valid votes” cast) is systematically shrinking: from 58% to 53% to 50% to 45%. In the other direction, how the opposition’s vote share has increased over the last three national elections: from 24% to 27% to 28%.

(As an aside, Table 6 also demonstrates how remarkable the 2016 election was. Converse to what is happening to the ANC national vote share, in local elections it was systematically increasing its proportion of all registered voters between 2000 and 2011: from 28% to 30% to 35%. But, in 2016 that all went south. It dropped back to 30% and the opposition grew to 25% — from 21% in 2011 — and as a result of the turnout differential set out above.)

Conclusions

All of this leads to the following conclusions:

• The 2016 election bucked a number of long-term trends for the ANC. Most importantly, its ability to close the “apathy gap” between its national and local government election performance stalled.
• Up to that election, the ANC had made good progress closing the “apathy gap” — from a difference of 5.3-million fewer votes in 2000 (a loss of 50.2% of its 1999 vote share); to a 4.4-million vote loss in 2006 (a loss of 40.5% from 2004); to 3.2-million loss in 2011 (a loss of 27.8% from 2009) — between itself and the collective vote of the opposition.
• In the 2016 local government elections, the ANC lost 3.3-million votes from the 2014 national elections (28.9% of its 2014 vote share). This failure to further reduce the “apathy gap” cost it enormously as the opposition not only further extended its turnout differential, but dropped its own “apathy gap” to no more than a staggeringly low 5.9%.
• This produced the biggest turnout advantage for the opposition in South African voting history: a 23 percentage points differential. In urban centres such as Gauteng, the differential grew to as much as 24.5 percentage points.
• Thus, if history is to be repeated, the ANC’s 2019 election performance will see the bulk of those 3.3-million ANC voters who stayed away in the 2016 local government elections, return to vote in the 2019 national elections, regardless of any change in ANC leadership.
• All that, however, does not mean the ANC’s vote share is declining. The ANC has failed to grow its proportion of registered voters as the voters' roll has grown over the years. In 2009 and 2014 the opposition grew its number of new voters to a far greater degree than the ANC.
• The generally static nature of the ANC’s share of registered voters means slowly, over time, the opposition is making steady inroads into its majority. And there is every reason to believe that pattern will continue to play out in 2019. The ANC still has a majority but that majority is getting relatively smaller and smaller every national election.
• That fact, coupled with the ANC’s weak turnout in local government elections and the opposition’s ability to turn out its voters more effectively, came together in the perfect storm in 2016, and cost the ANC heavily.

If one takes a long-term view, the ANC’s prospects over the next decade or two appear dire. On the one hand, it has been able to rely on the support of between 10.5-million and 11.5-million ANC voters who, come rain or shine, Mbeki or Zuma (and likely Ramaphosa), turn out to vote in national elections.

However, it shows little ability to be able to grow its vote share by registering new ANC voters. And that problem will have been compounded by the advent of the EFF — effectively the ANC Youth League — which has taken away from the ANC an entire generation of younger, potential ANC voters. Without that, the problem that is its inability to grow its share of registered voters will being to manifest in real electoral decline over the next 10-20 years.

One could argue that if the ANC has any long-term analysis of its prospects, this reason alone — and more than any other — could be driving its desire to reincorporate the EFF back into the party.

In SA, the electoral machine turns very slowly, but it would certainly seem to be turning in one direction: the ANC is in decline. That decline will manifest in a change of power in local government elections before it does in national elections, because the opposition has become so brilliant at closing its “apathy gap” between the two election types.

The ANC seemed to be making some progress in improving on this front itself, but it has always performed worse than the opposition and what progress there was, was arrested in 2016. Whatever the result in 2019, if the opposition doesn’t fracture or reintegrate into the ANC, the next set of local government elections could be the event that sees the ANC drop below 50% nationally and result in widespread change in municipalities and metros across the country.

Ultimately, it is likely the ANC would have won with Jacob Zuma in charge. But the longer-term consequences would have been even more profound.

What Zuma did was arrest the ANC’s ability to increasingly neutralise opposition turnout in local government elections. In fact, he motivated opposition voters to turn out in higher numbers. But he doesn’t seem to have any real effect on the ANC’s actual support base.

This effect would have cost the ANC relatively more votes in 2019, but probably not a majority.

Jackson Mthembu was right, only he had the wrong election. The local government elections in 2021 will be the election to watch.

• Van Onselen is the head of politics and governance at the South African Institute of Race Relations.