How people elect parliaments
Voters in the Australian state of Western Australia are about to elect a Legislative Council using a built-in mechanism to bias the election in favour of non-urban voters.
Voters go to the polls this Saturday to elect both houses of their state Parliament.
Held every four years, the state election is being watched anxiously because polls indicate the Liberal party government will be ousted by the Labor party, although perhaps only narrowly.
Of particular fascination will be the performance of nativist populist party One Nation, which is attracting very significant support across the nation against a background of voter frustration with the nation’s two long-established major political parties.
Earlier polls have indicated that One Nation would attract up to 13% support across the state on election day, more than enough to influence results in many lower house (Legislative Assembly) electoral divisions, where results are decided by Australia’s preferential voting system.
The Legislative Assembly of Western Australia – the lower house of the state Parliament
Polls in the final week of the campaign show One Nation support a little lower, at around 9%, although polls attempting to read voter break-aways from previous patterns have been notoriously unreliable in recent years. Evan at 9%, the impact of voter preferences will be important to the make-up of the chamber.
There has much public debate over the ‘preference deal’ between the Liberals and One Nation. But in the elections for the 59 Legislative Assembly seats parties can only make recommendations to their supporters – published as ‘how-to-vote’ leaflets at polling places.
Past analysis – such as by the ABC’s Antony Green – has suggested that minor party preference recommendations have very little impact at shifting what voters already intend to do, and then only if the party’s polling place presence is very strong on election day.
Where the Liberal-One Nation preference agreement will certainly matter is in electing the state’s upper house, the Legislative Council.
The Council’s 36 members are elected using the single transferable vote (STV) system, with the 6 most favoured candidates chosen in each of 6 electoral divisions.
The Legislative Council of Western Australia
A similar system is used for a number of elections across Australia including those for the national Senate, for four state upper houses, and for the lower houses in Tasmania and the Australian Capital Territory. (The New South Wales and South Australian cases use a single, state-wide electoral division.)
But in Western Australia the Council ballot papers still use a special ‘group voting ticket’ system, developed for elections for the Senate in 1984 and subsequently adopted for use at some of these state elections.
Under this system instead of marking preferences for each individual candidate, voters may mark their ballot paper to adopt a full order of preferences selected by the one political party they nominate. Each party submits their ticket to the electoral commission in advance of the voting, and they are then used during the counting of votes.
The technique allows parties to strategise and negotiate over the content of their voting tickets. The result is increasingly large blocks of preferences flowing through the vote counting process as the candidates of minor parties are eliminated, leading to near-random effects on the final seats decided in each election.
Constitutional scholars ponder whether this approach meets the Australian constitutional standard that representatives must be “directly chosen by the people”.
With an increasing number of minor parties emerging across Australia in recent years, the system has started to malfunction visibly, leading to widespread criticism.
The GVT system was created somewhat hastily in 1983 for use in national Senate elections, ostensibly as a solution to the increasingly burdensome demand that voters fill out preferences for every single candidate.
NSW abandoned the GVT system in 2003, replacing it with one where voters specified themselves the order of political parties through which they wanted their preference votes to flow. A similar system was also adopted for the Senate in advance of last year’s elections.
Voters using the STV system in NSW, Tasmania, the ACT and (as of 2016) for the Senate are also no longer obliged to mark preferences for every candidate, but need only number a fairly modest minimum number.
Only the states of Victoria, South Australia and Western Australia still use the GVT system for their upper house elections. Western Australia has had nearly a year to consider following the abandonment of the use of GVTs for the Senate, but did not legislate. Elections are due in the other two states in 2018.
With a quota of 14.3% of the total vote needed to win a Council seat in each electoral division, One Nation is likely to be a live contender for the final seat in most divisions, and the flow of preferences caused by the GVTs of both One Nation and the Liberal Party may significantly advantage either of them.
The plan has however irritated supporters of both parties, with public comments disowning each party for its co-operation with the other, potentially causing direct loss of votes on election day.
The Western Australian Legislative Council is also the last Australian parliamentary chamber to be elected on the basis of weightings of vote value in favour of rural areas.
Previously common in several states and nationally, rural vote weighting was largely dismantled in the 1960s and ‘70s.
Of the six Council electoral regions in which voters must vote, the three regions based on Perth have voter enrolments a few weeks before the election of 392,000, 396,000 and 410,000 people.
But the three other regions have much smaller pools of voters: 226,000, 103,000 and just 68,000 in the ‘Mining and Pastoral’ region.
Worse still, while the state electorate has grown as a whole in the past decade, enrolments in the Mining and Pastoral region have actually fallen, from 76,000 to 68,000, over the past two elections.
Mining and pastoral voters also have the lowest voter turnout, in the mid-70%s, compared to around 85% turnout in the other five divisions, yet they still elect six Councilors.
The impact of this is that just 7,500 voters in the Mining and Pastoral Region can elect a representative to the Council, while it takes over 49,000 Perth voters to elect the same single representative.
The effect of the regional vote-biasing is generally to create three seats for the regional-based conservative parties that they would not otherwise have been won.
In recent elections the biasing has resulted in the National Party, which has won state-wide voter support that should yield just less than two seats in the Council, winning three seats in 2008 and five in 2013. In the 2008 elections two other seats were also won by the Liberal Party above the number its state-wide vote should have earned it.
In this year’s election the expectation is that the biasing may assist the One Nation party to win the bonus seats. While that is likely to come at the expense of the conservative parties, it may still create a One Nation balance of power in the chamber that would not naturally occur.
Given that One Nation has a good chance to win seats in any case, it may be difficult to determine whether biasing specifically made the difference.
As usual, the ABC’s Antony Green has a detailed analysis of the history and impact of regional biasing in electing Western Australia’s parliament.
Tasmanian Kevin Bonham also had a detailed discussion of the continuing use of the Group Voting Ticket technique in Western Australia.
images: State Parliament of Western Australia
Update: this post has been revised; the comment that “NSW abandoned the GVT system in the 1990s” has been revised to read “in 2003”.
How would you design the electoral system for WA’s Legislative Council?
Thanks Jeremy. This requires a whole post! But in brief, I’d go back to first principles and ask what is the role of the chamber, and of the associated lower house, and of the related chambers of the national parliament, and select an electoral system aligned to each such defined role.
The lower houses should be places where a highly representative selection of voter opinions – or political parties- carry out the main ‘political’ contests over legislation, and over the choice of an executive. For that reason I’d lean towards STV-based lower houses. Noting that that would mean sacrificing the ‘local member’ aspect of single-member electoral divisions.
The upper houses are meant to be distinct houses of review, so you want to choose again a broad representation, including geographical representation. STV in divisions of 6 or more would broadly achieve that. But really I’d also argue for other structural arrangements, such as not having ministers from upper houses (or at least, only having in the upper houses special ministers responsible for scrutiny of governance functions such as auditors, electoral commissions, the justice system, the public service, an Attorney-General, and so on). That would change the dynamic of the sort of nominees who would seek upper house careers. I’d also consider whether the registered political parties which were used as vehicles for upper house election had to be distinct from those which sought lower house election.
That’s where I think constitutional and electoral law design debates should go. At present the incumbent parties (and new entrants) only debate which systems will win them which numbers of seats – the iron law of partisan self-interest.