How does municipal voting actually work?

14 May 2015

There’s an article by Prof. Steven Friedman in the Business Day positing that the ANC will need to fall significantly below 50 per cent in next year’s elections in certain metros if it is to lose its majority in these metros. The three Gauteng metros were mentioned but the focus was on the Nelson Mandela Bay metro, where Prof. Friedman has calculated that the ruling party will need to fall to about 40 per cent before it risks losing the metro.

Prof. Friedman’s analysis is based heavily on the maths behind constituency-based votes, or ‘first-past-the-post’ voting as it is also known. Under such a system, a party can get a plurality of the vote without needing a simple majority to win a constituency-based seat. Put more simply, and in terms of our municipal elections, a party can win significantly less than 50 per cent of the votes in a ward and still end up winning the ward if it gets more votes than any other party.

Now, the professor is right about a few things in his analysis, but he’s a bit off on some other things. It’s true that the ANC can fall below 50 per cent of the vote and still win about half the seats – but it won’t win much more than that. It’s true that a party could win each and every ward without getting above 50 per cent in any of them – but what’s missed is that our municipal voting system has in-built checks and balance to prevent a party winning much more representation than its actual share of the vote.

Again, in simple terms, if a party manages to win all the ward seats (about half of all seats) but only achieves 40 per cent of the overall vote, our system of seat allocation will cap the number of total seats for the party, and it won’t win a single other seat.

Making sense of the vote-counting and seat-allocation rules is not easy, so let’s try to keep it simple. Two votes are cast in every municipal election (not including by-elections): a ward vote and a proportional representation (PR) vote. In general, voters vote the same way for both votes, choosing the same party on each ballot. This means that, in a particular municipality, most parties can expect their share of the PR vote to be more or less the same as their share of the ward vote.

This is an important detail. It’s possible to construct scenarios where a party’s share of the total ward vote in a municipality is vastly different from its share of the PR vote. In reality, the shares are usually similar and they don’t differ by more than 3 per cent in the most extreme cases.

These differences can be attributed to two things. Firstly, there are independent ward candidates and party candidates whose parties have no PR list (these candidates don’t qualify for a PR vote). Secondly, some voters who might have a personal preference for a particular ward candidate who is not from their party of choice, but will still vote for their party on the PR ballot, thereby splitting their vote.

Overall, a party’s percentage of the ward vote should be about equal to its percentage of the PR vote, and these should in turn be equal to the party’s percentage of the total vote and also its percentage of the total number of seats.

This works because of the legislation around voting and seat allocation. Prof. Friedman is right that we have a mixed electoral system in municipalities – that is, we have both constituency/ward seats and PR seats. But what isn’t clear to most people is the relationship between these two sets of votes and a party’s overall seat allocation.

Here’s the link to the Municipal Structures Act. The laws around votes and seats are in Sections 12 and 13 of Schedule 1, near the end of the Act. It’s not an easy bit of maths to digest, but here’s a layperson’s summary of what happens in a municipality:

– All the votes (ward and PR) are counted in one pile. A party’s share of the total vote is then recalculated as the party’s share of total seats. So, for example, if a party received 100 votes out of 1000 votes cast, its share of the total seats is 10 per cent.

This doesn’t mean it will get exactly 10 per cent of the seats, but it shouldn’t receive a significantly different share of the seats. It also means that the breakdown of this total share isn’t significant – the party could (for argument’s sake) receive 20 PR votes and 80 ward votes, or any other combination of the two types of votes.

– Ward seats are then allocated based on the first-past-the-post system.

– PR seats are then allocated to all parties based on the parties’ share of the total votes, and if there’s a big discrepancy between a party’s share of ward seats and its share of the ward vote then the PR seats are allocated in a way to correct this imbalance as much as possible.

Let’s look at a real-world example, specifically the Nelson Mandela Bay metro. The table below gives the ward and PR seats for each party in the 2011 municipal elections, plus the percentage of each sub-total:

A breakdown of votes and seat allocations in Nelson Mandela Bay, 2011 municipal elections

A breakdown of votes and seat allocations in Nelson Mandela Bay, 2011 municipal elections

Only the ANC and DA won ward seats in the metro, and it can be seen that the parties’ share of the ward seats is higher than their respective shares of the vote. The ANC only received 52 per cent of the ward vote but 57 per cent of the seats, while the DA received 40 per cent of the vote but 43 per cent of the seats. COPE was the biggest loser, winning 5 per cent of the ward vote but not receiving a single ward seat.

This is a good illustration of the downside of constituency-based voting, but look at the next four columns. The ward vote giveth to the ANC and DA, and the PR vote taketh away. The ANC received 52 per cent of the PR vote but just 48 per cent of the PR seats. The DA’s share of the PR vote (40 per cent) is similarly higher than its share of PR seats (37 per cent). But look at COPE now: just 5 per cent of the PR vote but 10 per cent of the PR seats.

The allocation of the PR seats is essentially the correcting factor. You can see from the final columns that the percentage of the total vote for a party correlates quite nicely with the share of total seats received by that party.

Prof. Friedman is right that in every mixed system there is a bias towards one type of vote or another, and he’s right that the bias is to the wards seats, but I think he’s overestimating the size of the bias, and therefore the margin of safety that the ANC has.

In theory, even with the system’s in-built corrective factor, a large party could win every ward seat and receive far less than 50 per cent of the vote. But it would not receive a single PR seat under our system; its total number of seats would be capped.

The biases in the system are most glaring for councils with fewer seats: there are Northern Cape municipalities with just a handful of wards where a dominant party can win far more power than it deserves, based on its share of the vote. But that’s unlikely to happen in large metros which each have well over a hundred council seats.

Prof. Friedman is right about a number of other things. A simple, single-seat constituency-based system will not necessarily make politicians more accountable. It definitely won’t dilute the power of the big two parties. But the municipal seat-allocation system is much more robust and fair than he credits it.


2 thoughts on “How does municipal voting actually work?

  1. Markus

    Excellent explanation. I knew something wasn’t quite right about Friedman’s article. However, there are some councils with a very small number of seats, and in those the final outcome will only be very roughly proportional.

  2. Pingback: Nelson Mandela Bay metro Part 2: The EFF factor | paul berkowitz

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>