Friday, August 12, 2016

Majors Stitch Up Senate Term Lengths, Film At 11

When the Senate resumes at the end of the month, one of its first items of business will be to allocate the Senators to six-year and three-year terms.  Today both major parties indicated that they will use the order-of-election method, under which the first six Senators who were elected in each state's cutup will receive six-year terms and the remainder three-year terms.  Special Minister of State Matthias Cormann has stated:

 "The important point is obviously this is a function of how many votes and how many preferences you are able to attract. If you are elected in the first six out of 12 then it stands to reason that you were elected earlier and as such you qualify for the longer period."

Labor's Penny Wong has issued a statement saying Labor will support the order-of-election method and saying that it "reflects the will of the voters".

The Electoral Act requires the AEC to conduct a "Section 282 recount" to simulate which Senators would have filled the first six places at a half-Senate election based on the same votes, as an alternative to using the order-of-election method.  However there is no requirement that the Senate use the recount to decide the order.  While the Senate affirmed in both 1998 and 2010 that the recount should be used, those decisions were not binding on the current Senate, and were under the old system anyway.



The Senate can in fact allocate the three and six year terms however it likes.  There is nothing in the Constitution to prevent the Senate from, for instance, giving all the crossbenchers three-year terms, or giving three-year terms to known Tony Abbott supporters or all Senators who opposed Senate reform (for example).  But there would probably be a massive outcry if the Senate did something that was obviously undemocratic.

It happens that both major parties benefit from the agreement to use the order-of-election method rather than S 282.  Senators Lee Rhiannon (Greens NSW) and Derryn Hinch (Vic) are both given three year terms while Scott Ryan (Liberal, Victoria) and Deborah O'Neill (Labor, NSW) will not have to look for work again until at least 2022 (unless there is another double dissolution).  This difference had previously been found by Grahame Bowland and the AEC has advised the Clerk of the Senate that Rhiannon and Hinch would get six-year terms if the Section 282 recount was the deciding factor.

Problems with the order of election method

Section 282 was brought in because the order-of-election method can lead to irregularities.  One of them is that if a minor party gets just over two double dissolution quotas, the minor party will win two six-year terms.  Then with the same vote it would win a third Senator at the next half-Senate election and hence hold a quarter of the state's seats off only 15.4% of the vote.  

When we look at the two different options in the 2016 case, it isn't immediately obvious that either causes such a problem.  Both Hinch and Rhiannon fell short of a DD quota and were the only elected candidate for their party in their state.  It doesn't seem to really matter much whether they are given three year or six year terms.

Yet there are some fairness-based problems with applying the order-of-election method even in this case.

I'll use Rhiannon's case as an example.  Her primary vote was 0.947 quotas (including 1 Greens ticket votes).  Because she was short of a quota on primary votes, she had to wait until the cascading surpluses at the start of the count elected eight major party candidates.  But had she polled a quota in her own right (just another 0.41% of the primary vote) she would have been elected third and won a six-year term.  

On the other side of that coin, Jacqui Lambie polled a primary vote of 1.068 quotas in her own right (including ticket votes).  This meant Lambie was elected fourth by the order-of-election method, and therefore gets a six-year term.  Yet the massive surpluses of the Tasmanian major party ticket leaders Abetz and Urquhart actually meant that seven Senators who won either on primaries or on cascading surpluses had more "support" than Lambie did.  Thus, as a sign of her support relative to these major party candidates, one might say that her having polled a DD quota on primaries was far from conclusive. 

Not only is polling a quota or not on primaries therefore a very arbitrary test of whether a rather easily-elected minor party candidate should get six years or three, but in a really borderline case it is worth bearing in mind that it is the candidate whose primary vote (including ticket votes) determines the term length, not the party.  So a minor party is disadvantaged if it runs a good support candidate who pulls primary votes from the lead candidate.  It is also disadvantaged if minor parties that are similar to it take primary votes from it, even if those minor parties are quickly excluded in the cutup with all votes they have taken from the winning party returned as preferences.

The order of election system is also absurdly granular.  In Tasmania the two major parties have about the same vote (Labor 33.6 Liberal 32.5) and get two six-year terms apiece.  But this would also happen if the figures were Labor 46 Liberal 20, or the other way around.

These issues arise because a double dissolution is an election for twelve Senators in a state.  The order in which one is elected in an election for twelve Senators says nothing conclusive about whether one is one of the six candidates voters would deem most worthy of a long term, because the bar to get over for the latter should rightly be higher.  

Problems with the Section 282 method

The Section 282 method seems like a better option.  It uses the votes cast in the election to simulate what could have happened had the same votes been cast at a half-Senate election.  In this way, it uses the election votes to conduct an election for six places, using the appropriate quota rather than the clearly inappropriate quota for an election for twelve.

There are issues, however, and a lot of them are created by the unexpected interface between the new Senate system and Section 282, which was written under a different system. Because the government left the implementation of Senate reform very late in its previous term (whether to avoid provoking the non-Green crossbench, in the hope of bipartisan support or simply because it was distracted by leadership fighting) a number of finer details were glossed over for the first run of the new system.  The matter of the fairest term-allocation system, which had long been on the back-burner anyway, was one that was so obscure that it wasn't even on most psephologists' shopping lists for things to fix for the new system.

Under the old group ticket system almost every vote was a full ranking of candidates - albeit by a voter who typically had no idea that they were ranking candidates in that way (and with very rare exceptions created by the old below-the-line savings provisions) - but now very few voters rank all the competitive candidates.  In NSW, only about 0.3% of voters went effectively all the way either above or below the line.  Most voters numbered just 1-6 above the line or 1-12 below, which means they tend not to have expressed a preference between all the candidates who were actually elected.  Some votes exhausted without reaching anyone who won.

These problems aren't as acute as they might appear, since parties that get elected tend to receive strong preference flows from voters generally.  In South Australia, where five different parties won at least one seat each, about 30.5% of above-the-line votes included at least four of the winning parties in their top six.  (Thanks again to David Barry's preference explorer.)

(Note added: Indeed, the proportion of votes that are not only usable but remain in the recount seems quite high.  Percentages of votes excluded from the recount based on Grahame Bowland's simulations appear to have been around: 4.6% Victoria, 4.1% Queensland, 3.7% NSW, 3.2% South Australia, 2.5% Western Australia, 1.7% Tasmania.  Exhaust rates during the recount are then estimated at: 4.5% NSW, 4.4% SA and Victoria, 2.3% for Queensland and Tasmania and 0.1% for WA.  So over 90% of vote-values were used.)

Still, when votes cast under the new system are processed in a Section 282 recount, some of them will carry no usable preferences, some will carry a few and some will carry preferences between all the elected candidates.  The effect is that if a voter's vote didn't reach any elected candidate then that voter has no say in which of the elected candidates get a six-year term.  But had the voter been asked to vote in an election between the elected candidates for that specific purpose, they may well have had a view on that.  Given that they weren't informed that they should vote all the way through if they wanted to have a say on term length, it doesn't seem ideal to exclude their vote from the assessment. 

That's not to say S 282 is less fair than the order-of-election system.  I think it's still fairer - just that it's hardly brilliant and it is worth a serious think about whether there could be a better approach than either for the new voting system.

There are also many ambiguities in applying S 282 to the new system.  Some of these existed in the old system as well (The AEC recommended fixing them in 2011 but it never happened).  I'll repost what I wrote on an earlier thread:

------------------------------------------------------------------------------------------------------------------
There is a detailed post up by Dean Ashley (Alaric) dealing with the ambiguity of the Section 282 rules to prepare a recommendation for which Senators get 3 and 6 year terms.  The drafting of these rules is quite a mess and this was a hot topic of discussions I had with a few people following the declaration of the polls today.  There are ambiguities in the wording of Section 282 concerning, for example:

* whether votes that were informal in the original count (eg they had two 1s) can become formal in the S 282 recount (I would think not)
* whether votes that were partially formal in the original count (because of duplications of numbers) can become more formal in the recount if numbers are no longer duplicated (again I would think not but this has never been clarified)
* whether votes that were fully formal in the original count can become informal in the S 282 recount because they do not have enough preferences (again I would think not, hardly the voter's fault)
* whether a vote that has zero usable preferences in the recount because it is entirely for excluded candidates counts towards quota although it in fact exhausts immediately (this is largely a quirk of applying S282 the new system, although it could have happened in the old system too; the AEC's view as of 2011 was no; Antony Green also believes such a vote is excluded).

(PS Another one has come up, and it's a cracker:

* whether a vote that was originally formal both above and below the line, but has zero usable preferences in the recount below the line can have its above the line preferences used in the recount instead.  I think the legislation has not anticipated this application of the new system at all!)
--------------------------------------------------------------------------------------------------------------------

I have received a copy of the AEC document sent to the Senate and it confirms that informal votes do not become formal and that votes with no usable preferences are excluded from the recount quota.  It sheds no light on the more esoteric points.

A myth about the Section 282 system has been that it disadvantages smaller parties.  It does so in Antony Green's example (where a minor party has two DD quotas but parties other than the "big three" have 5.6% of the vote between them).  However these days it is common for both major parties to poll not that much more than two half-Senate quotas in a state.  That creates a situation in which a minor party with just under one DD quota can win the fifth or sixth seat at a Section 282 recount, despite only winning one of the last few seats in the original election.  Thus at this election, it is actually the order-of-election method that disadvantages the smaller parties.

Alternative Proposal

Senator Hinch (who has been all over the place on this one, probably because he didn't initially realise the S282 recount would advantage him) has come up with a suggestion that every party that wins a seat should have at least one six-year term.

I think this is giving the micro-parties too much.  It means that the LDP and Family First, which have only snuck into the last seat in one state each because it was a double-dissolution, would get a six-year term off a vote that would not have got near winning one at a half-Senate election.  Hinch's case is quite different to theirs and S 282 recognises this.

The proposal is also open to rorting; for instance the Greens could deliberately devolve to separate state parties, thus winning themselves a six-year term in every state in which they won a seat.

Major Party Dealings

What is interesting here is that the AFR originally mooted (I am unsure on what evidence) that the major parties were looking at using Section 282 to disadvantage the crossbench.  Whoever had crunched possible numbers for this based on the results even as they stood at that time had absolutely no idea what they were doing. It has turned out that Section 282 actually advantages the crossbench, and, at this point, we have the major parties proposing to ignore it.

The charge of having an eye for the main chance is especially acute here in Labor's case.   Labor attacked the Greens and NXT for dealing with the Coalition to fix massive problems with Senate voting, accusing them of doing so for self-advantage and helping the Coalition in the process.  Yet now we have Labor doing what looks a lot like a deal with the Coalition to fix the Senate term lengths to its own almost indisputable advantage, and helping the Coalition in so doing. (I say "almost" because while the Greens grapple with their own savage case of NSW Disease, there's a case for Labor keeping Senator Rhiannon where she is as long as possible.) 

I doubt the expected decision is going to cause any great outrage, though it may produce some parliamentary theatre.  Those disadvantaged will complain, but it's a fact that the disadvantaged candidates polled less than a DD quota each, and would only have won six-year terms via the recount because of the weakness of the major parties, with exhaust perhaps a contributing factor.  It doesn't have the public-consumption no-go value of a Senator winning off half a percent of a vote or an election being voided and rerun, and by the next Senate election almost everyone outside the pseph community will have forgotten all about it.

Note added August 30:  See also the press release (PDF) by the Proportional Representation Society of Australia, which draws attention to more issues with the order-of-election method.  Some of these were covered above but, for instance, it is possible for a party to win more than 50% of the vote and hence seven out of 12 Senators, yet have five of them up for election again in three years.

September 1: It's Done:  The motion to approve the election-order method passed with the Coalition, Labor, One Nation and Lambie in favour and the Greens, NXT, Leyonhjelm, Day and Hinch against.

25 comments:

  1. Do we know how far off Larissa Waters and Sarah Hanson-Young were from getting longterm places under 282?

    Would preference instructions (written and verbal) adivising voters that they could vote beyond 1-6 ATL or 1-12 BTL have changed that?

    ReplyDelete
    Replies
    1. I don't know the official margin but it would be close to Grahame Bowland's simulations (and I didn't realise how much detail there was there til I looked for the answer to this point, so another free plug for https://angrygoats.net).

      Hanson-Young came in seventh in the SA simulation but over 58,000 votes behind Griff. Waters also seventh but with a margin of about 26,000 votes.

      I greatly doubt that fuller instructions to voters would have affected either of these results. If anything they would have increased the margins as Greens voters are more knowledgeable about election systems and hence more likely to vote through.

      Delete
  2. Whether or not one system is 'fairer' than the other is debatable (and an interesting debate that I thank for you exploring).

    Allowing everyone to see what the results would be under a variety of systems and THEN choose which to use just makes for another unedifying spectacle of each and every party once again falling over themselves trying to post-hoc justify following their naked self-interest.

    It's an absolute disgrace, and having that choice obviously favors the major parties heavily, so I don't imagine we'll have any hope of seeing it changed. Shameful.

    ReplyDelete
    Replies
    1. The prospects for improvement are indeed grim. Really a referendum is required, but it's hardly something that would be taken to a referendum by itself. So it would have to be tacked onto another referendum, and even then public understanding of the issue is so low that a lot of people would just reflexively vote no.

      Delete
  3. Some misinformation in this article in The Australian today. Claims regarding s282: "it is based only on the votes for the 12 elected senators and ignores the preferences of people who voted for other candidates". The last bit being complete nonsense of course.

    ReplyDelete
  4. Can't say I can get terribly excited about this issue. The s 282 method may be maaarginally fairer (attempt at Hawkie's pronunication there) but as you say there are problems in trying to apply it. And the quota's almost twice as easy to achieve in a DD (easier by 13:7 to be pedantic). Lambie and Hanson got quotas and Xeno got 3, and Hinch and H-Y didn't. So they get a chance to get re-elected in 3 years. Not really a lot to see here folks... (Personally I think we should amend the Const so they all get 3-year terms - keep them answerable I say!)

    ReplyDelete
  5. Oops, Xeno got 2.8, and his first 2 got long terms. Seems fairish, if not perfectly fair, to me...

    ReplyDelete
  6. Very bad new for the Greens in the Senate. The DD meant that this election was at best about drawing even in the Senate and more likely conceding a spot in SA. Looks like next election will also be about drawing even with no opportunity to get ahead in NSW, QLD or SA as they would with a six year term.

    ReplyDelete
  7. "In South Australia, where five different parties won at least one seat each, about 54% of above-the-line votes included at least four of the winning parties in their top six."

    I can't reproduce this; I get about 30.5%.

    Lib & ALP & NXT & Grn: 17.23%
    Lib & ALP & NXT & FF: 12.62%
    Lib & ALP & Grn & FF: 10.50%
    Lib & NXT & Grn & FF: 10.40%
    ALP & NXT & Grn & FF: 12.83%

    That sums to 63.58%. But each of those five components is including votes with all 5 preferenced in the first 6.

    Lib & ALP & NXT & Grn & FF: 8.28%

    To avoid double-counting, four times that 8.28% should be subtracted off, giving 63.58% - 4*8.58% = 30.46% of ATL votes preferencing at least four of the winning parties in the first six. (Plus or minus some round-off error.)

    ReplyDelete
    Replies
    1. Thanks; I've identified a source of double-counting in my estimate so I've adopted your figure.

      Delete
  8. It seems clear to me that the fairest/most democratic way to determine which of the Senators elected should get the 6 year term are those that would have got elected if it had been an election for half the number of places.

    Technical arguments about how this should be most accurately assessed under the new vastly more democratic voting system are important, but they shouldn't obscure the importance of that fundamental principle.

    Although one could be forgiven for wondering whether it's worth the bother - the two times it has been clear what the most democratic option has been, simple party political self-interest has applied anyway.

    ReplyDelete
  9. It would be interesting to see a list of who (and which party) would get a 6 year term under one of the two main options but a 3 year term under the other.

    ReplyDelete
    Replies
    1. It's just the two cases mentioned. In NSW Deborah O'Neill (ALP) gets the six year term under the order-of-election method while Lee Rhiannon (Greens) gets it under the Section 282 method. In Victoria Scott Ryan (Lib) gets the six year term under the order-of-election method while Derryn Hinch gets it under the Section 282 method. All the others stay the same.

      Delete
    2. Thanks Kevin. Does this indicate that the OOE method favours the majors and section 282 favours the minors or is that just coincidence?

      Delete
    3. Whether it favours one or the other depends a lot on the kinds of votes seen at the election. In 1987 OOE favoured the Democrats who got two extra six-year-terms at the expense of the Coalition. I think it's specifically the current combination of high total minor/micro party vote and relatively even minor party preferences that makes S 282 favour the minors/micros in the current case. A minor party candidate can have 0.4 of a quota in the recount and not be caught, which would have been impossible when major party shares were higher.

      Delete
    4. All of which is a key reason why the mechanism to be used needs to be agreed in advance, as no one knows for sure until after each election who will benefit from which mechanism. but given the wording of the Constitution, I guess there is no way this can be done.

      The farcical attacks on the changes to hugely improve the democratic system have now been shown to be totally false, but this seems to be of zero interest even to journos who are paid to inform people about politics, some of who now seem just as happy to retail attacks from the same people complaining about a result that is the exact opposite of their earlier, equally non-founded scare campaign.

      To expect them to comprehend or care about a post-election deal by the same people to negate the more democratic outcome would be a clear example of putting hope before experience.

      It's only about who ends up being the people who decide on the laws that impact on every person in the country, so I guess it doesn't matter much.

      Delete
    5. Unfortunately the only way to stop post-hoc decisions about who will serve 6-year vs 3-year terms is to change the Constitution. It would be good to do this if a chance arose, but it probably won't unless someone abuses the system in an especially scandalous way. I agree with Rowan above that the fact that a decision between systems is being made and justified post hoc is the worst thing here. Parties don't know where they stand in advance.

      There has been some media interest in the fact that scaremongering about Senate reform was proved wrong - especially Adam Morton here: http://www.theage.com.au/comment/election-2016-why-democracy-won-in-the-senate--even-if-you-dont-like-one-nation-20160809-gqobfc.html

      There have also been some journalists who while not writing on the subject have at least helped share articles about it on social media. Overall though little of the post-election discussion of reform has risen above he-said-she-said reporting of the comments of politicians.

      Delete
    6. It's not the biggest issue in the world, and I'd have to agree that holding a referendum on this issue would be absurd, but I think ANY system with the perverse incentives of this post-hoc decision making that effectively forces politicians into self-serving and dishonest behaviour is a problem. After they justify that self-interest and why they are more deserving than their political opponents to themselves mentally, it becomes that little bit easier next time to justify a bit of gerrymandering here, a bit of holding inquiries into opposition parties there, a spot of favoring donors in policy decisions on the side...

      The next best choice would be to publicly insist each party take a pre-election position to favoring one mechanism or the other, such that it would be too embarrassing to pretend the other one is better afterwards, but I don't think it would make a difference to the behavioural incentives and it's hardly a sexy media topic during an election campaign.

      Without any changes, I'd just appreciate a little more honesty from the politicians: "We chose this system because it gives us another seat in 3 years time. Obviously it wouldn't be fair to just split the 6 year terms with the [coalition/labor party] so we haven't done that, but given that we don't think one mechanism is inherently fairer or less fair with respect to section 282 or first elected, so we've made the choice that gives us more seats." Then they could at least have my respect.

      Delete
    7. I don't see it as a huge issue. There likely wont be another DD for a long time (20 years?).

      Perhaps it could be part of a suite of voting changes for a referendum over that time frame.

      Delete
  10. This is an argument about nothing. Neither method is intrinsically better than the other, and any conceivable method will produce winners and losers. We've had six DDs in 115 years so it's not like it's going to be an issue very often. Of course Andrew Bartlett supports the other method because in this case it would have gained the Greens an extra six-year seat. (Although since the beneficiary would have been Senator Rhiannon, who has done so much damage the Greens brand in NSW, perhaps he should be careful what he wishes for). AC

    ReplyDelete
    Replies
    1. Well yes, any method will have winners and losers, but we generally aim for whichever method best reflects the will of the voters. Quota based STV systems are designed to select a certain number of candidates, and the order elected is not representative of the proportion of the electorate that supports a candidate. It overweights first preferences, and dilutes the preferences of those who voted for a micro party first.

      To make the maths simple, in a DD, 12/13ths of the votes contribute to electing a senator. In deciding who gets six year terms, 6/7ths of the votes impact the result in the 282 method, but a few as 6/13ths impact the result in the order of election method.

      Delete
    2. Yes, as Matt said, the argument as to why the Section 282 is fairer and more representative is simple and clear.

      That's why I said repeatedly before the election that this is the method that should be used. I presume it is also why both Labor & Liberal also voted for motions in the Senate on two separate occasions in previous years stating this is the best method to use

      Such motions have no impact after the election has been held of course (other than to show the hypocrisy of party that will vote in support of a basic principle before an election and then against it after the election when self-interest demands it.

      Delete
  11. Hi Kevin,

    Is there any news as to whether there will be a court challenge to the Herbert result? The media seems to have gone quiet on this.

    ReplyDelete
    Replies
    1. I've heard nothing. They still have a few weeks to decide.

      Delete
  12. I've now uploaded my tabulation of the NSW Senate election preference distribution - all 151 counts of it. (The other states will follow.) The most interesting aspect is how random much of it is. Very few of the minor parties managed to deliver even 20% of their preferences to their preferred candidate. Without automatic preference allocation, and at most booths without a how-to-vote card to help them, most minor-party voters allocated their preferences more-or-less at random, often to the nearest parties on the ballot paper, regardless of ideology. This shows that, under the new Senate election system, parties which don't have enough people to staff all the booths, all day, will not be able either to be elected themselves or to help anyone else get elected. That will be even more the case at the next half-Senate election, when the quota will be 14.3%, not 7.7% as at a double dissolution. Once this fact sinks in, I think we will see a lot fewer minor parties at the next election.
    http://psephos.adam-carr.net/countries/a/australia/2016/2016senatensw.txt

    ReplyDelete