Western Australia does (mostly) the right thing

After winning an almighty landslide in the state election last March, the re-elected Labor government lost no time in setting up an inquiry into reform of Western Australia’s upper house, the Legislative Council. The Committee’s final report was released yesterday, and the government has announced that it will legislate promptly to implement its recommendations.

There were two big problems with the upper house, and the Committee has adopted the right solution to both of them. One was the bizarre system of automatic preferencing via group voting tickets (GVTs), which delivered seats to parties with negligible public support via backroom preference deals. GVTs will be abolished, and voters permitted to allocate their own preferences either above or below the line – that is, either for party tickets or for individual candidates.

This isn’t rocket science; New South Wales led the way with this reform more than twenty years ago, later followed federally and in South Australia. But Labor opposed the change at federal level in 2016 and has refused to make it in Victoria (where the GVTs produce equally weird results), so it is good to see the Western Australian comrades showing more enlightenment.

The second and more serious problem was the gross malapportionment. The Council was elected on the basis of six geographical regions, which each returned six members even though they had wildly different numbers of voters: South Metropolitan had about 450,000 while Mining & Pastoral had less than 70,000.

There were a number of proposals for how the regions could be reorganised to provide for approximate equality, but the Committee decided that it was easier and fairer to scrap them altogether and have the whole of the state vote as a single electorate. This too, in my view, is clearly the right decision. In Victoria, which votes by region, there is a serious failure of proportionality even without any malapportionment.

Western Australia already has a lower house elected from geographical districts; there seems no reason why local representation should [require] a geographical-based upper house as well. Rather than suggest that some people’s votes should be worth more than others’, the Committee has a useful set of recommendations for encouraging MPs to locate their offices in rural or remote areas.

The big advantage of a single electorate model is the much lower quota for election. An upper house of 37 (an increase of one, to avoid the problem of deadlock with an even number, although this was not mentioned in the actual report) will have a quota of one-thirty-eighth, or 2.63% of the vote; due to exhaustion of preferences, candidates in practice will probably be elected with about 1.5%, providing for real diversity of representation without the lottery of GVTs.

In New South Wales and South Australia, which also vote as a single electorate, Legislative Council terms are staggered so that only half are elected every four years. I think this is a good system, since it prevents governments winning an upper house majority on the basis of a single landslide result. But since Western Australia already has four-year terms, the Committee, perhaps wisely, declined to recommend it. As Ben Raue comments, “I don’t see how it could be politically feasible to introduce eight-year terms in a chamber currently elected for four-year terms.”

So within its terms of reference, I think the Committee has done the right thing. The big outstanding problem is one that it was not given power to consider, since it was told to recommend reforms to the preferential system. But the changes to be made raise the question of whether preferences for the upper house are necessary at all.

To understand why, it’s best to read Antony Green’s explanation. As he puts it, “In every other country, electing 36 [or 37] members by proportional representation would be done by a form of List PR without preferences.” The more members you elect at once, the less likely the preferences are to make a difference to the outcome, but the more work is involved both for the voters and for those who have to count them.

At some point – and I suggest that point is well short of the 21 at a time that New South Wales elects – the gain in voter control by allowing preferences is more than outweighed by the cost, inconvenience and delay that it causes. It would be enormously simpler to give voters a list of parties and ask them to tick one, then use a Sainte-Laguë calculation to allocate seats to the parties’ lists of candidates.

The Western Australian government decided not to venture in this direction, possibly due to fears about its constitutionality (the constitution requires that “members [be] chosen directly by the people”) and possibly in the belief that, in Green’s words, “Australian voters distrust an electoral system that does not allow choice of candidate and preferences.” I confess that I am sceptical of the latter assertion, since I think public engagement with electoral systems is too low to inspire either trust or distrust, but for now at least it will not be tested.

Despite that reservation, Labor in Western Australia should be congratulated for moving quickly and decisively on reform. The pressure must now be on its Victorian counterpart to follow suit.


6 thoughts on “Western Australia does (mostly) the right thing

  1. As a relatively recent Wozzie (April 2020) I’m still “reading in” to Western Australia’s politics and governance, having spent the bulk of my time in Australia on the other side of the rock, and in Queensland at that, which dispensed altogether with its legislative council in 1922. So I thank you for this interesting analysis and agree with your assessment.


  2. As a long-time STV supporter I am aghast at this bizarre recommendation. You (Charles) and Antony Green are correct about STV: it’s not well-suited for filling more than 15 seats at most (and only works well with 9 seats or fewer). Rather than seeing that as a reason to scrap all districts and elect 37 MLCs at large by any form of PR (whether list or STV), i would see that as a reason to keep districts but reduce the malapportionment — or at least the political advantage from malapportionment.
    One way to achieve that would have been to have (say) five seven-seat regions plus one four-seater for the vast underpopulated outback of WA (total 39 MLCs). The trade-off would be that — instead of the four-seater containing around 10.25% of WA’s residents or voters, and each seven-seater containing about 17.9% (plus or minus 10% tolerance) — as you would expect in a per-capita basis — the ratio should instead be 19% per seven-seater and 5% for the four-seater. So the voters outside the populated areas would have about twice as many MLCs as a population basis would indicate — to ensure their parish-pump concerns are not overlooked — but the seats would divide 2-all between Labor and Liberals/ Nationals in partisan terms. The MLCs could well gang up 3-1 or even 4-0 if Perth tried to close down all rural fire services or site a nuclear waste dump at Halls Creek. Also, given the… interesting state of Liberal/ National relations in WA, one MLC from each might not always mean both vote as a block. But still, like the ACT and NT Senators, their over-representation would not distort the party balance in the chamber.
    Electing 37 MLCs state-wide by STV would mean a ridiculously low quota and an unworkably huge ballot-paper — for a tiny marginal gain in proportionality and minor-party diversity over what 7- seaters would give — even if it were adopted in Tasmania. Doing so State-wide in Western Australia — the world’s second-largest subnational polity in area, with more square kilometres than 141 of the UN’s 150 member States — is insane. This is going to be seized upon to discredit STV the way that the 1989 ACT election is Exhibit A for “why at-large open-list PR with preferential voting is a bad idea”.
    (yes, I suppose I should have put in a submission while the Committee was still taking them… I just assumed the result would be something sensible)
    (I also think 21 State-wide for NSW is silly too… the Failed State, too, should have seven-MLC regions, three if rotation is retained, or better five if both Houses move to concurrent four-year terms. South Australia’s 11 per election is within the range of tolerability, and besides, “State-wide means immune to gerrymandering” was a very big selling point for Dunstan-era Labor to accept extensively-reformed bicameralism).


    1. Thanks Elizabeth – we agree on the problem of statewide STV, but draw opposing morals from it. I’d be perfectly happy with STV on a regional basis if we were designing a reformed lower house. But we’re not; there’s already a lower house based on geographical units with preferential voting, so I don’t feel any need to preserve either of those things in the upper house. A simple party list system, as used successfully in dozens of countries, seems to me the obvious way to go.
      I don’t think the gain in proportionality is “tiny”: a party with a 3% or 4% ceiling on its support (of which there seem to be quite a lot) has virtually no chance of election even in a 7-member region, but can reasonably hope to get a statewide member up – perhaps even two. Given the way minor parties are shut out of the lower house, I think that’s an important goal. It also draws the sting of some of the opposition to the abolition of GVTs.
      Even if we were forced back to a regional system, I’d still be totally against giving up the principle of one-vote-one-value. Sure, people in remote areas have lots of disadvantages when it comes to representation, but so in different ways do city people. I’d address both by giving additional resources to MPs’ offices (“give them all lear jets if we have to,” as I once said), not by diluting representation.
      As to exactly how serious the problem of a single 37-member STV election is going to be, I guess we’ll have to wait & see. I agree that the Committee has underestimated it, but I’m not sure it’ll be quite as bad as you think. I could be wrong.


      1. Agreeing to disagree, Charles. I respect where you’re coming from, but I would still see considerable daylight between 57 single-seat “geographical units” and, say, five, six or seven multi-seat “geographical units”. The latter would represent geography in quite a different way, as State and local governments do. I agree there seems little point in the US model in most States (not all the 49 bicameral ones, interestingly) where they elect N State Senators and N+X State Representatives/ Delegates all by first-past-the-post from single-member districts, with no discernible difference other than size of the chambers (admittedly quite dramatic for, say, New Hampshire) or rotating terms (and even that isn’t universal). But the PR/ multi-seat vs AV/ single-seat difference seems quite dramatic enough to mean two chambers cover different dimensions even if neither, to be sure, covers an entire State at large.
        J-D, I think open or flexible lists would overcome a lot of the opposition to list-PR in general – even among many whose first choice would be STV. Antony Green and Colin Hughes have both made the point that open lists could well be easier for voters to re-arrange electorally than STV with party-ranking on the ballot-paper.. especially, but NOT only, with “one in the box” GVT voting ATL. This might then overcome the “tablecloth ballot” problem because you could have a McDonalds placemat on the wall of the polling booth with all the candidates, arranged by list, and the ballot-paper itself need only be the size of a House of Reps slip, with only squares for the parties and then a write-in space for a candidate of that party. The rules for formality would need to be extremely lenient, eg either the candidate’s list number, or their full name, or even just the first three letters (“RIC” for Charles, etc) would be sufficient so long as it was unambiguous. If someone writes “LATHAM”, you can decide between John or Mark according to which list they supported, but they voted only for a candidate – or if they voted Labor but wrote in “FERGUSON” or “THEOPHANOUS” or “DICK” – the ballot might be void for fatal ambiguity. The 1997 ConCon postal
        Arguably, specifying that a vote only for a candidate – with no party list indicated – counts only for that candidate, might get around constitutional objections based on “directly chosen by the people”.. Such a vote would help that candidate move up the ladder on their own party list, but would not boost the total for their running mates.


  3. Sorry, should say “would not distort the left/ right balance in the chamber”.
    Due respect to Ben Raue but am note sure I follow his logic with “But once you are electing MLCs to represent a region stretching from the Kimberley to the south-west, you might as well elect MLCs at large.” The issue is not just one of (a) lots of seats, low quota + (b) very likely lots of candidates (correlates broadly to number of seats but can be accelerated or dampened by other factors such as deposits [monetary amount and refund threshold], optional vs compulsory preferences, and all-candidates vs number-teams-separately ATL voting) + (c) sheer geographical size + (d) absolute population size (although this is less of an issue for WA voting at large than, say, the Russian Federation voting at large in 2007 and 2011).
    A “Rural and Remote WA” region with (say) 4/39 or even 7/35 MLCs might look, from a distance on a map, to colour nearly as much of the State Of Excitement pink on the map as a statewide at-large constituency would, but there would be significant differences. Only one-fifth (and in a model like the one I proposed, one-twentieth) as many voters for the MLCs to constituency-service – 100,000 to 300,000 instead of 1.5 million. Only – assuming the Colin Hughes empirical average of around four candidates per vacancy (based on the era of compulsory preferences, invalidated for some time by use of one-in-the-box GVTs, but relevant again with semi-optional group preferencing) – two or three dozen candidates instead of nearly two hundred. A ballot paper that can actually be read by the human eye. Removal of the need to take away the voter’s right to number candidates individually. No more Trojan horse entree point for party lists.


  4. Have you considered the option of a system which: puts on the ballot paper a list of candidates grouped by party; requires each voter to vote for exactly one candidate; divides the total seats between the parties (the vote for each party being the total of all the votes cast for candidates of that party); and allocates each party’s seats to its candidates in order of individual votes received?

    I’m sure this would have problems, because _every_ system has problems. But would it be any more complicated than having voters simply vote for a party list (I don’t think it would)? Wouldn’t it create an incentive for parties to nominate candidates who have some individual voter appeal, and wouldn’t that be a good thing (among other things, because they’d want candidates with a diversity of appeal, including but not limited to geographical diversity)? Wouldn’t it reduce the power of party machines, to a degree that would also be beneficial?


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.