South Australia is now the third state government to announce plans to tinker with the voting system for local councils this year, joining Queensland and Victoria. I’ve given cautious support for parts of the Queensland plan which would bring in proportional representation (PR) for some councils, while severely criticising Victoria’s proposal to strip back PR in most councils in favour of single-member electorates. Unfortunately the South Australian plan is in the latter category.
The South Australian government has proposed a series of reforms, but one in particular would completely change the voting system back to a simplistic and unfair voting system used in South Australia until the late 1990s, one that people worked to reform, with the main justification being to make it simpler and cheaper for local councils to run their elections.
South Australian councils are currently elected by a variant of Single Transferable Vote proportional representation (STV-PR), similar to those systems used for the Senate, mainland upper houses, the Tasmanian House of Assembly, the ACT Legislative Assembly, and local councils in New South Wales, Victoria, Tasmania and the Northern Territory.
The discussion paper proposes a shift to a system known as ‘bottoms up’. Voters still cast their votes as preferences, and then the lowest-polling candidate is eliminated until the number of remaining candidates is equal to the number of vacancies to be filled.
The critical difference between the current STV system and bottoms up is that it does not feature any use of a quota. Under STV, a candidate must reach a quota to be elected. The quota is set as the smallest number of votes which could only be achieved by as many candidates as there are to be elected. So if you are electing five seats, the quota is 1/6 + 1. If a candidate’s vote exceeds the quota, their surplus votes are then passed on as preferences.
This ensures that each candidate who wins gets approximately the same number of votes, and means that voters for a candidate who wins by a big margin still have their votes counted in the same proportion. There is little need for tactical voting under such a system, since a vote for a candidate who already has enough votes to win is not wasted, but is instead passed on. The quota also ensures that a candidate who already has enough votes to win can’t accumulate any more, and any extra preferences are instead passed on to someone who needs them.
Bottoms up instead creates a great deal of potential for tactical voting. Let’s take this scenario which Kevin Bonham outlined on Twitter:
There are 3 candidates for 2 vacancies and 1000 voters. 901 vote A-B-C. 49 vote B-A-C. 50 vote C-A-B.
B is excluded. A and C win although 95% preferred B to C.
— Kevin Bonham (@kevinbonham) August 5, 2019
If one candidate accumulates a large proportion of the vote, none of their preferences will play a role in electing another candidate. You can get 90% of the vote and still only influence the election of one of the two candidates. It simply cannot be proportional, and means voters have an incentive to try and pick and choose which candidates they think have already won.
This voting system was previously used in South Australian council elections. This edition of Quota Notes (the newsletter of the Proportional Representation Society) from 1992 discusses efforts to replace bottoms-up with STV (which had been an option since 1985). The system was finally abolished in 1999.
So why make the change? At times it has been advocated for as a way to prevent candidates organising as party tickets, since a concentration of votes with a single candidate is punished, but there are other ways to do this (no above the line voting, no party groupings, local conventions about party involvement in councils).
The reason given in the discussion paper is that the proposed method “will remove the requirement for complex counting software, and reduces the risk posed by any lack of access to this software.” In other words, this is about cost.
Running fair STV elections and producing a proportional result is more expensive, since it often requires staff with a greater level of knowledge, or at least more time spent data-entering ballots into software.
New South Wales ended up allowing local councils to use privatised election providers in 2012 after councils were shocked by the bills they received from the NSW Electoral Commission for the conduct of the 2008 elections.
Ultimately the cost of functional and competent democracy is relatively small compared to the overall budgets of local government, and if councils are struggling to cover the costs the state government should chip in the cash needed, not give them an out by moving to a simpler but less fair voting system.
For now the proposal is in the form of a discussion paper, and you can make a submission, with submissions closing on November 1.
Thanks to Casey Briggs and Kevin Bonham for bringing this to my attention, and to Andrew James for his knowledge about the history of this voting system.
The argument for a new local government electoral system in the SA Government’s discussion paper would have to be one of the weakest pieces of “analysis” I have ever seen in such a document.
The comparison it offers between single transferable vote and the proposed system amounts to little more than slogans: “fair” vs. “simpler, faster”. Its attempt to assess how the system might work on the basis of only six recounts of earlier elections is simply risible: anyone with the slightest knowledge of electoral systems knows that how a system will perform in practice is a function not just of its intrinsic representational properties (which aren’t examined at all in the paper), but of how political players adjust their strategies to reflect the opportunities the new system creates – with the rise of preference harvesting at elections using group ticket voting being a prime example. Re-running old counts provides no insights at all into how the system might work once players get their teeth into it.
In fact, the proposed system is similar in some respects to, and shares most of the defects of, the single non-transferable vote system which has been such a disaster in, of all places, Afghanistan.
The authors of the paper might question whether analysis based on representation of parties is relevant to local government. But what is relevant, even where there are no parties, is representation of points of view.
Suppose the community is broadly in favour of development rather against development, and there are 6 candidates – three pro-development and three anti-development – for three vacancies; and they all direct preferences to their like-minded candidates. The most prominent pro-development candidate – perhaps the incumbent Chairperson (“Mayor”) – gets 52% of the vote and the other two pro-development candidates get 6% and 4%. The three pro-conservation candidates get 15%, 12% and 11%. On these figures, the anti-development candidates will win two out of the three seats even though the pro-development position is favoured by the community 62-38. Worse, if the pro-development candidate who won with 52% resigns, under the proposals in the discussion paper his or her seat would go to the third of the anti-development candidates. This arrangement would clearly create an incentive for external players to seek to manipulate the composition of councils, for example by offering a council member a job which would require him or her to give up his or her position.
The argument that the paper makes about the alleged comparative complexity of single transferable vote counting is also, really, pathetic. STV elections have been run in this country for over 100 years, including for the Senate since 1949, and for the SA Upper House for several decades. The use of computer software to conduct them is a comparatively recent development. Provided you understand the rules, all that they involve is primary school arithmetic, supplemented with a calculator. If South Australian authorities have reached the point where they are incapable of coping with this, then the State has a much bigger problem than just its local government electoral system.
Finally, it’s worth mentioning that this counting method was proposed for NT local elections about 10 years ago, but wasn’t adopted, not least because of defects pointed out in a paper prepared by Professor Ben Reilly.
Comments are closed.