How people elect parliaments
In the ACT, the Labor-Greens loose coalition goes into the 2016 election attempting to secure a 5th term of government (well, for Labor, at least…the coalition has existed for only the last two terms). Can they do it?
Canberra isn’t naturally a Liberal-voting town. Only two of the eight ACT elections has seen the Liberals form government, and both times involved winning just 7 of 17 seats and finding two more votes from the cross-bench*. There’s no doubt that the sum of Labor and Greens supporters in the city well outpolls the number of Liberal supporters, and will do for the foreseeable future.
But the last election saw Labor and the Liberals very close, in fact the Liberals actually won the popularity race, beating Labor by 41 votes (out of 220,000). Both Labor and Liberal won 8 seats, and the 17th seat went to the Green, giving Labor-Greens a 1-seat win.
In this years coming election, the Liberals would have been in dire trouble with Tony Abbott still as Prime Minister; the impact of federal leadership on Canberra’s mood will be bit calmer now.
But there is a problem. And it’s all of Labor’s own making.
Last year Labor and the Liberals jointly supported new legislation to expand the Assembly to 25 seats, and more significantly to make all 5 of the new electoral divisions have 5 seats each. And it’s these 5-seat divisions which give the Liberals a chance.
The problem goes like this: in all five electoral divisions, both Labor and the Liberals are almost guaranteed two seats each, because they both inevitably win enough votes for two quotas (sometimes the Liberals are a little short in the most progressive central suburbs, but they can make up any slight gap.)
So the election turns on which combination of parties win the 5th seat in every division.
If no-one gets enough quotas for seats 3, 4 and 5, they are awarded to the candidates with the highest votes anyway. That’s fair enough.
The issue is that what matters in these scenarios is a technical rule that bears little relationship to voter views: the order of elimination during counting.
Imagine this scenario in a five-member seat: the Liberal candidates have won 2.44 quotas, Labor’s 2.39 and the Greens’ 0.65. You can probably see that when the candidate preferences are shared out, the two main parties will win two seats each. But who wins the 5th? The Green candidate, with the largest remainder? Or at least, wouldn’t Greens and Labor preference flows mean that one of those two will win the last seat somehow?
It all depends, it turns out, on how well each major party’s 3rd candidate has performed; specifically, how close they are to the 2nd-placed one.
Consider the following graphic. Each of Liberal and Labor’s vote has been distributed after their first candidate has won a seat, so each of those two has exactly one quota of votes against their name. The remaining party vote favours their 2nd candidates, with their 3rd some way behind. That’s a pretty normal outcome.
Under the elimination order rule, what happens next is that the candidate with the lowest vote is eliminated. In this case, the 3rd Labor candidate will go out. Some preferences might flow to the Green – more than flow to the Liberals anyway – and yes, the Green candidate wins the 5th seat. That’s not an unreasonable result, since the combined Labor-Green vote was 3.04 quotas to the Liberals 2.44.
Now consider the next graphic, which spreads exactly the same total party votes in a different way. Again, each of Liberal and Labor’s vote is distributed after their first candidate has won a seat. But in this example their remaining votes are spread almost equally between their 2nd and 3rd candidates.
Under the elimination order rule, what happens next is that the Green candidate goes out.
Then each of Labor and Liberal will win a second seat (in one way or another), and the final seat will also go to one of them.
In the second graphic above Labor is behind, but they will hope that Green preferences might help them snatch the 5th seat.
Because otherwise, the 3rd Liberal candidate will win it.
And you may have spotted that I’m ignoring preferences from other minor parties and independents, which usually flow a bit stronger to Liberals than to Labor.
It gets worse if the Liberals have their 2nd and 3rd candidate close together, while Labor has them far apart. Then the latter’s 3rd candidate actually goes out first. That leaves 6 candidates, and since the Green is placed 6th, it’s them that loses out, and the Liberals also win three seats.
(This problem isn’t partisan, by the way; it disadvantages whichever major party is relying on a semi-allied minor party, which in the ACT at this point in history happens to be the parties outlined in this discussion. It’s a classic case of how voting numbers work to the disadvantage of two-party coalitions against one solid party, even if the collective public support of the two outnumbers that of the one.)
Now, it so happens that in the ACT’s five electoral divisions, the Liberals traditionally poll very well in one (Brindabella), and normally win the 5th seat there. The Greens will have one strong seat (the new seat of Kurrajong) and can be expected to win that 5th seat.
But out of the remaining three seats, if the Liberals can pull off two wins, they win government.
Is this likely? Well, the numbers in the example above are drawn precisely from an extrapolation of hypothetical results in Murrumbidgee, the closest of the remaining three seats – using the 2012 election results.
No-one can tell you how likely a Liberal win is, but it is entirely possible even if the community’s overall Liberal vote is behind their opponents’ combined total.
Is this unfair? The vote counting system itself is not to blame (although if a way to avoid using sequential elimination could be agreed and implemented, such anomalies would be removed).
In fact, the old 17-seat system also had two divisions of five seats, but the larger seven-seat division sort of anchored the whole system. Had the expanded Assembly been made up of divisions containing more than 5 seats, it would be have been more fundamentally representative, and the odds of overall wins against the political grain of the community would be much reduced.
The STV voting system that is used in Canberra (known locally as “Hare-Clark”) is a fine one, but its benefits are diminished when legislators create small-sized electoral divisions. The phenomenon described above happens no matter how many seats an electoral division has, but at higher numbers the impact on overall election results is much less.
Come October, the probability of against-the-grain election results will ultimately depend on how many votes everyone wins. Obviously, if both Labor and the Greens can improve their votes, they improve their chances.
But there are two good lessons of political strategy here.
One, while it’s pretty difficult to control how close a party’s 2nd and 3rd candidates will run, parties should always make sure that even their third best candidate is still pretty attractive! Nominating two strong candidates followed by weak or unknown new candidates can trigger this syndrome.
Secondly, Labor and the Greens can most effectively forestall this problem if they can persuade their supporters to fill out their ballot papers with additional preferences flowing between their two parties. Achieve that, and the prospect of a Liberal government is almost entirely prevented.
Of course, the imperfect relationship between the Labor Party and the Greens – and their grassroots movements – is a significant issue all over Australia at the moment. They don’t always fully cooperate….
Labor went along with the 5-member-seat decision in part in the hope of driving down the number of Greens elected, hoping to take those seats themselves. That move was not really a good way of building up trust.
In October this year, Labor will find out the result of its decision.
21 January 2016
(* The author spent 7 years working at the Assembly as an adviser to one of those independent cross-bench MLAs.)