Thursday, December 10, 2015

Mackerras Piece Misleading On Senate Reform

As noted in the last part of my multi-volume series about people being Wrong On The Internet about Senate reform, nothing has happened publicly on this issue for some time.  But erroneous op-eds attacking the JSCEM-proposed model continue to appear in the media now and then, and the latest to muddy the waters (again) is Malcolm Mackerras in the Canberra Times.

While the consensus of psephologists Australia-wide favours scrapping the current Group Ticket preferencing system that has been gamed to death by preference-harvesters and other exploiters of confusing ballot papers (while retaining above-the-line voting for one or more parties), Mackerras has held out against this from the start.  Initially he argued that it would be acceptable to allow voters to stop after filling in 15 boxes below the line.  As the debate has progressed he has shifted to supporting a requirement for a minimum of six boxes below the line, which he now describes as "the easy and right thing" to do.  

I commented on this concept - essentially the Victorian system - in the Minimum Acceptable Reform section (scroll down) of a previous article.  Adopting the Victorian system would solve one major problem - the way in which the current system coerces voters to vote above the line by making it too onerous for many voters to vote any other way without risking casting an informal vote. 

It would not, however, rescue parties from the dilemmas they face when having to decide whether to do preference deals with ideologically hostile parties in order to try to get elected.  This harms parties with unusually high integrity levels (I'm sure they're out there somewhere) or whose followers expect high standards of integrity.  It would also not stop the election of largely unaccountable (since they have so little chance of re-election) micro-parties as a result of a combination of confusing and obscure preference deals and luck.  Nor would it eliminate the potential for a Senate contest to be voided in much the same way as happened in WA last time.

There being much to still dislike about the Victorian model, the question is why you wouldn't want to go further.  Many objections have been raised to doing so (and see other pieces on the "senate reform" tab for my comments on several of them).  In Mackerras' case, working out precisely what he has against the NSW alternative from this latest piece is rather difficult.

Mackerras' Objections To The NSW System

In the NSW system voters can vote for a party above the line, for multiple parties above the line, or for multiple (at least 15) candidates below the line.  Mackerras describes this system as "proportional representation by means of party list", and contrasts this to "proportional representation by means of the single transferrable vote" (PR-STV) in other Australian houses with multi-seat electorates.

It's misleading to call the NSW system a party-list system in the first place, especially in an article for popular consumption.  The concept of a party list system applies to European democracies in which a voter typically just votes 1 for a party.  The allocation of seats to parties is determined from the primary votes by mathematical formulas (examples include the D'Hondt and Sainte-LaguĂ« methods), which are often made subject to thresholds.  There are no preferences.

In contrast, while a 1 vote above the line in the NSW system functions the same way it does in a European party-list system (it is a vote for that party which assists no others), the voter in NSW has the choice to direct their own preferences if they want to, and the flexibility of doing so by party if they wish.  They also have the choice to just vote for their preferred party and no-one else.  The only inflexibility in the NSW system is the requirement that voters voting below the line for candidates must number at least 15 squares.  

After the seventeen of 21 seats awarded on raw quotas in the 2011 NSW contest, Pauline Hanson would have won the third of four remaining seats had the same votes been cast under a true party-list system.  Instead Hanson was overtaken by not one but two other parties after preferences, and lost.  This happened again in 2015, when the No Land Tax Party, which would have won the last of four non-quota seats, fell behind the Animal Justice Party and was defeated.  So the differences between the NSW system and party-list systems are not just matters of token choice, but impact upon who gets elected.  

Based on the incorrect classification of the NSW system as a party-list system, Mackerras then maintains that party-list systems are unconstitutional, although the use of group ticket preferences survived a constitutional challenge from an ungrouped candidate in 1984.  It appears (though I'm not quite sure) that Mackerras sets himself above the High Court in knowing what the Constitution means  here, but in any case his objection seems selective.  If McKenzie v Commonwealth 1984 was decided incorrectly, then that counts against the Victorian and current Senate systems as much as the NSW system.  All these systems allow voters to use certain shortcuts in casting their vote - it just happens that in the NSW system the shortest shortcut sends your vote to one party only then exhausts it, while in the Senate and Victorian systems it sends your vote into lost dimensions of electoral spaghetti.  The NSW system allows a voter to very easily direct their preferences between multiple parties of their choice, which the Senate system, especially, does not.

Mackerras on Senate cf. Victoria

Mackerras highlights that the Victorian Legislative Council election saw a large number of "odd results" from preference-harvesting successes, while claiming that the Senate election saw only one (Ricky Muir).  This is not the first time I have seen Mackerras making the false claim that Muir was the only anomalous winner under the present system.  However of the many online musings of his I have read on this subject, I am yet to find one where he discusses or justifies the election of Bob Day from less than 4% of the South Australian vote in any detail.

If we look at the raw quotas in South Australia, the Coalition, Nick Xenophon and Labor each secured a seat off their own vote.  Their remainders based on primary votes were then .922, .742 and .587 quotas, with .496 quotas for the Greens and .263 for Family First (Day).  Of the parties contesting these final three places, Family First started in fifth, with less than half the votes of the party in third.  It turned out that the parties in first, fourth and fifth place on primaries won seats, while those in second and third missed out.  As a result, the Nick Xenophon ticket won just one seat while polling more than twice the combined vote of the Greens and Family First, who both got up.  That is ridiculous.

I have chronicled before how Day, who stands well to the right of the Coalition (and was known to do so before the election), was eventually elected on yet another dud Labor preference strategy, though not for want of trying by the Greens to be the culprits instead.  However, to even get ahead of Labor and benefit from their inability to learn, Day needed to bridge a 4.9% gap, picking up more votes than he started with by means of preference-harvesting.  In the process, Day obtained above-the-line preferences from eleven micro-parties (some ideologically similar, some more or less opposite) while the ALP obtained only a one-third share from one (the Socialist Equity Party, whatever that is).  And this is the system that Sam Dastyari is defending, because the Labor backroom boys are clearly so good at using it?  

Noting that a third winner (Dropulich in WA) was initially declared under this system only for his election to be voided, and also that Lambie's win in Tasmania relied partly on preference-harvesting, it's not so clear that there's all that much to explain in the difference between the numbers of micro-party Senators elected in 2013 and micro-party MLCs elected in Victoria in 2014.  Mackerras does seem to think however that the 5/40 rate for preference-harvesters in Victoria means something to do with the system, but he doesn't spell out what.

The  most likely real cause of the difference is that in the 2013 Senate election some of the Others vote was soaked up by the Palmer United Party, whose wins in Queensland and the WA rerun were not caused by preference harvesting.  Excluding PUP, Xenophon and most of the confused NSW Liberal Democrat voters, the share of the vote left to true micros nationwide was only 13.7%.  In the Victorian Legislative Council, however, micros (including PUP) polled 19.7%.  That, and not the small differences in system between the Senate and Victoria, is the main reason so many micros won in Victoria at state level.

What is interesting about Victoria 2014 is that it utterly disproved the following Mackerras claim from the JSCEM process earlier that year:

When you give people a reasonable option to vote below the line, while there will be people who will try and game their way into parliament, this gaming will be overwhelmingly unsuccessful. Even under the present system it is not nearly as successful as many people make out. 

In fact in the 2014 Victorian election, preference-gaming won five of the sixteen seats not won on primary vote quotas.  The reasonable below-the-line option foiled just one otherwise winning preference snowball.  At one stage it was possible two micro-parties might win in the same electorate!


Oooh, it was a cunning plot by the big bad pseph cabal:

"My reaction to these pieces of election news was to say to myself: "My rivals in psephology will press the politicians from the big parties to install party-list systems designed to benefit the machines of their big parties. I must thwart them.""

Perhaps this bit is all just drama for the sake of style, but I found it still a little odd.  Psephology does have its competitive moments sometimes, but overall I think of fellow psephs both popular and little known as colleagues, not competitors.  I think most of us are out there doing our best in various ways to make electoral predictions as accurate as possible, and to defend and improve electoral analysis as some kind of science - since if that doesn't work, then all that's left is punditry and betting.  We mostly aim to get stuff right and learn from how others get stuff right so we can all get more stuff right.

When it comes to electoral reform, those of us who pushed for the JSCEM model did so not because we wanted to give big parties an unfair advantage (indeed, some of us don't even like the "big three" all that much!), but rather because the current system is broken.  It is confusing to voters, unfair to voters, and a hazard to the integrity of its own results.  It is also unreflective of real voter choice, as opposed to some simplistic model of choice in which any voter who likes one micro-party must be assumed to love them all.  

If the cost of fixing the current system is a party polling just a few percent of the vote in every state wins no seats, then so be it.  The price - a price commonly paid in PR systems the world over for a different reason (to prevent unstable coalitions) would be worth it.  But in any case, small parties will adapt to the new system, just as they adapt to threshholds in the many countries that have them.  Many likeminded micros will merge into broader flank parties, many will disappear, and this thinning of the forest should make it easier for any new force that might really be good for 8-10% of the vote in a state to stand out.


  1. Kevin, I agree with you. Professor Mackerras' comments are a bit hard to follow. While I strongly agree with him that party list systems are undesirable, and that Australian constitutional provisions most likely make it impossible for Australian legislatures to enact pure party list systems for our country, his claim that the NSW LC actually is a party list system clearly isn't right. My guess is that it simply seems so to him because of its large magnitude, which significantly reduces the impact of preferences (vis-a-vis full quotas achieved) in deciding results, but that doesn't change its nature. Perhaps the unhappy experience of the ACT's modified-D'Hondt exercise in '89 and '92 influences his opinion - I don't know.

    The present 'cause' of improving senate-style electoral systems involves shades of grey, proposals for partial improvements, and compromises. There's a large majority of the psephologically-minded in agreement that GVTickets are obnoxious, but holding out for STV purism (if that's what MM is saying) can be a barrier to improvements. The JSCEM proposals are partial improvements, and yes they include an inappropriate way of dealing with 'too many' small parties, but the wise course of action would be to bank what improvements parliament might pass in the near future and keep trying.

    STV purity - by which I mean highly individual-candidate based rules, no party boxes, no auto-tickets, etc - is conceptually very attractive, but it tends to work optimally with manageable numbers of candidates, as in the simplistic illustrations with 'candidates ABCDE', or at least with fewer than around 25-30, as in ACT and Tasmanian elections. For real elections with 50+ candidates some accomodation can be allowed for the needs of real voters and the conveniences they might desire, provided that such compromises should not have biased impacts between different parties and candidates.

    NSWLC-style options for 'above the line' voting are clearly more democratic than the present Senate version, and a case can be made that many voters are actually comfortable bundling their preference sequences by party in that manner.

    Similarly, optional preferencing with a minimum number of preferences is less damaging than full compulsory preferencing (both these rules operate to invalidate some people's votes for no good reason!), even if the 'minimum number of preferences = the same number as there are seats to fill' notion is an irrational brain worm (with a few real negatives).

    It's regrettable (but entirely predictable?) that many commentators - and the current major parties as well as most of the Senate cross bench - are entangling issues of fair voting design with hopes/expectations of partisan outcomes, including views on whether or not seeing diverse non-major-party senators get elected is desirable. As you've argued before, minor party candidates are completely legitimate - but they should succeed or fail within sensible rules.

    Malcolm Baalman

    1. My major concerns with pure STV as well as those mentioned above are:

      1. The practicality (or not) of counting it - data entry costs would skyrocket.

      2. Some voters accustomed to the current system will just vote 1. Either these votes become informal or they are allowed to exhaust. In both cases (especially the first) parties prone to receiving such votes suffer.

      I agree totally with the "irrational brain worm" comment. The combination of that notion for an election with 12 vacancies and a lack of adequate savings provisions caused a ridiculous informal rate of 7.5% in the non-compulsory postal vote for Hobart City Council last year.