Friday, December 25, 2020

Do Green Preferences Matter At Tasmanian State Elections?

Secular seasons greetings to all, except the IPA, who get a great big HAPPY HOLIDAYS for the grinchy effort on the crosstabs in their latest scientifically unsound poll-shaped object.  It is a nearly annual tradition on this site to release something every Christmas Day.   I've now added a "Xmas" tag to all the Christmas Day articles so anyone sufficiently bored can see how varied these offerings have been.  

This year's Christmas Day article idea arose in mid-November following reactions to Tasmanian Labor's decision to letterbox fliers attacking the Greens and promising never to work with them in government again, a promise that I expect to expire immediately the next time Labor gets a majority then loses it.  Whatever might be said of the authenticity of Labor's messaging, I can't help thinking of the whole thing as something that you might see in a vintage British comedy skit:

"The Greens!  They're holding Tasmania back!  They're leaving people behind!"
"That's very unfortunate.  How many of these, er,  Greens are there?"

In online commentary about this, a theme sprang up that Labor was biting the hand that feeds it preferences, and that this (perhaps together with Labor's junking of its 2018 poker machines policy) could very easily backfire.  However, my own experience is that Greens preferences are not as big a deal in Hare-Clark as they are in other elections.  I thought it might be worthwhile to examine the history of Green preferences in Tasmanian state elections since the rise of the Greens in the late 1980s, and look at how much difference they have actually made and might make, or not make, in the future.

Why Green Preferences Matter Less in Hare-Clark

In federal House of Reps elections nearly all Green preferences are distributed between the major parties.  They are only not distributed in seats where the Greens finish in the top two.  At the 2019 election, Labor's advantage on Greens preferences nationwide on a 2PP basis was 6.7%.  Not quite all of this was actually distributed, since 9% of the Greens' votes were polled in five seats where they made the final two, but this still leaves Labor with a leg-up of about 6% of the national total vote.  Fifteen seats were won by Labor (see details here) that it would have lost had the split of Greens preferences between the major parties been even.  Also, this was nothing exceptional - Greens preferences have been splitting around 80% or slightly higher to Labor for some time, and Labor routinely wins many seats on them.

In comparison, under Hare-Clark in Tasmania:

* The Greens often end up competing for the final seat in a division, in which case their preferences are not distributed whether they win or lose.
* When the Greens win a seat more easily than that, much of their vote is soaked up by their quota, leaving a relatively small amount to flow on as preferences to other parties.
* By the time Greens preferences are distributed, it is sometimes the case that all candidates for one major party have been elected or defeated, so the preference throw may only be contributing to internal contests on another party's ticket, or (in rare cases) contests between one of the major parties and somebody else.
* Voters are only required to fill five boxes, which is the number of candidates the Greens typically run.  This creates a similar effect at party level to optional preferential voting in NSW - about a third of the Greens votes available as preferences exhaust from the count.

However, a small number of preferences from Greens candidates are always available in the form of leakage - a vote that flows from a Green candidate to someone else even though there are still other Greens it could go to instead.  Some of these votes weren't originally Greens votes and just result from a minority of voters voting across party columns for various reasons.  Tasmanian votes are manually counted and distributed so we only know who a vote is transferred from in the preference distribution, and not necessarily who it was originally for (the same applies to preferences).  

History of Greens Preference Flows

For each election from 1989 onwards, I present a table that shows:

* the breakdown by seat of leaked votes from the Greens into Labor, Liberal, others and exhaust/loss due to fractions.  
* the same breakdown for preference votes in cases where all Greens were elected and excluded and there was a transfer from them that could still go to both major parties
* the statewide total and percentage breakdown for all votes above
* the total gain by Labor over the Liberals statewide as a result of preferences from the Greens (whether they were leaks or final preferences when the Greens left the count)
* the two-party preferred share for Labor vs the Liberals of all votes leaving the Greens.  

The tables below were compiled by hand and are very likely to contain some errors, but any errors probably won't affect the overall patterns.  


1989 was the first year the proto-Greens, then known as the "Green Independents", ran in every seat.  However they did not run full slates in any seat, so very few of their preferences exhausted.  The Greens won a seat in every electorate.  Greens preferences in excess of their first quota were distributed between the major parties in Bass and Lyons.  They were also distributed in Braddon (a whole one vote's worth) and Franklin, but in both cases there were no Liberals remaining in the count by that stage.  In Denison, the second Green finished sixth so there were no preferences available, but there was a leak from Bob Brown's surplus.  Overall the flow to Labor was quite strong.  A fight between two Labor candidates in Braddon was decided by seven votes but so few Greens votes were distributed in Braddon that they did not (at least directly) cause the result.  Likewise in Franklin where Paul Lennon very narrowly lost, Greens preferences had actually favoured him over his Labor rival.


The 1992 election followed the Labor-Green accord, which became the Labor-Green minority government, which collapsed largely over logging quotas.  For the first time the Greens ran full slates in each of the seven-seat electorates, and this led to a high percentage of Greens votes exhausting (more than half in the cases where Greens preferences were distributed).  The Green vote declined, but the Greens still won seats in every electorate.  However only in Denison and Franklin did they have enough left over for preferences to flow to the major parties, and not that many.  As a result Labor's gain statewide was small.  Greens preferences did again - amusingly given later developments - shore up Paul Lennon's position in Franklin.  Lennon would still have retained his seat had the Franklin Green votes split 50-50 between the majors, but not had they substantially favoured the Liberals.


In 1996, Labor and the Greens had been in opposition to Ray Groom's Liberals for four years, but Labor went to the election with its candidates signing a pledge to govern in majority or not at all.  The Greens suffered a slight swing against them, narrowly losing Bass.  Greens preferences were only distributed in Denison, where the exhaust rate was lower than in 1992 (probably because they were distributed earlier in the count).  However the 2PP flow to Labor was slightly weaker.


Prior to the 1998 election, the major parties agreed to reduce the size of the House of Assembly from 35 seats to 25 seats.  Contentiously this reduced the number of seats per division from seven to five, which disadvantages the Greens when their vote is low.  This was widely seen as the purpose of the legislation, but the Greens were not successful in getting a sympathy vote out of it.  They lost everywhere except Denison, and their preferences were distributed via exclusion of their lead candidates in Bass and Braddon.

Between 1996 and 1998 the Liberals had governed in minority with minimal Greens support, and this - plus possibly a perception that Labor was more to blame than the Liberals for the choice of five-member divisions - contributed to a weakening of the preference flow to Labor overall.  The 60.7% 2PP share was the second weakest so far.


This election saw the Greens resurge to their highest vote to that stage.  The Bacon Labor government won with a massive primary vote lead over Bob Cheek's hapless, hopeless and divided Liberals, Cheek losing his seat.  However, all the seat gains were made by the Greens, who had votes to spare mainly in Denison and Franklin plus a little bit in Bass and Lyons.  The 2PP preference flow to Labor went back to above 70% and Labor made its biggest gain on Greens votes to this stage.


At this election there was a slight swing against the Greens, but they retained their four seats (very narrowly in the case of Bass).  Greens preferences were distributed in Denison and Franklin only.  This election saw the re-election of Labor under Paul Lennon although the party had been polling poorly leading many people to prematurely conclude that it could not retain its majority.  A decision to save habitat at Recherche Bay probably neutralised anti-Labor sentiment among Green voters ... this time ...


This was one of the more remarkable elections in terms of preference flow.  By this stage Labor had been in majority for three terms, and the 2006-10 term had seen the fall of Lennon as Premier, the Gunns Bell Bay pulp (non-)mill saga and numerous scandals.  Lennon's replacement, David Bartlett, had taken a strong stance against working with the Greens, and some of the Liberal candidates appeared to be more environmentally friendly to Greens voters, at least in the area of old growth logging.  

The Greens polled 21.6% of the vote in 2010.  They had preferences to spare in every seat except Braddon, but in Denison those preferences were thrown between Andrew Wilkie (Ind) and Elise Archer (Liberal), with Archer just hanging on despite a strong flow to Wilkie.  The preference flow to Labor in Labor-Liberal contests was very weak (only 52.6-47.4) and in Lyons the Liberals were preferred to Labor on both Green preferences and Green leakage.  Labor's statewide gain on Greens preferences was the lowest ever.


From 2010 to 2014 two Greens served as cabinet ministers in a Labor government.  This arrangement was artificially ended by Labor just before the 2014 election, but had been the closest relationship between either major party and the Greens to date.  Not surprisingly Labor's share of Greens preferences not only rebounded, but reached its highest level so far on a two-party basis (almost 80%).  

The Greens lost Braddon, where they were excluded and their preferences were distributed.  They also lost Lyons and won Bass without preferences passing on in either case, and had surplus votes in Denison and Franklin, albeit not many in the latter.  This election also saw the only clear case I can find of Greens preferences determining the winner of a within-party contest: Madeleine Ogilvie (ALP) defeated Julian Amos (ALP) by virtue of getting more Greens preferences than him; an even split of Greens votes between the two would have seen Amos win.  A curious choice by Greens voters in light of Ogilvie's socially conservative voting record and later career as an independent who votes mainly with the Liberal government, but none of this was known to them then.

2014 was also a case where an uneven split within a party had an impact.  Paul O'Halloran's initial throw saw 1507 votes go to Bryan Green (ALP) and 807 to Brenton Best (ALP), with 746 to the Liberals and 2016 exhausting.  Best - who had criticised the Greens heavily during the term - ended up losing by 451 votes, so his seat would have been saved had fewer Greens voters exhausted their votes.  An even split between Green and Best on Greens preferences wouldn't have saved Best alone, but it would also have helped him.  Even an overall gain rate for Labor that was as high as in Denison would have been likely to save Best's seat - however both the gain rate and the 2PP flow in Clark (formerly Denison) are normally higher than in other seats.  

(Note: for Braddon a case could be made for including 938 votes from Bryan Green's surplus.  These were votes that had come from Paul O'Halloran (Greens) - not necessarily as primary votes for the Greens - then flowed to Green.  These votes split 659-132 to Labor, with 147 exhausted or lost due to fractions.)


The 2018 election saw Labor in Opposition, but with a relatively young, left-wing female leader and a Greens-friendly policy of limiting poker machines.  Labor not only successfully competed for the Greens vote in the inner suburbs - contributing to a very weak result for the Greens - but also received enthusiastic preference support from Greens voters.  The 2PP split of Greens votes (86.2% to Labor) and the Labor gain against the Liberals on Greens votes statewide (1.4%) were at their highest levels ever.  However, this swung no seats, since the closest final seat between-party contests were Liberal vs Green in Franklin and Labor vs Green in Bass.  The seats where Greens preferences were distributed (Braddon and Lyons where they were excluded - the latter for the first time - and Denison where they had a little bit left over quota) weren't close at party level.  Lyons was close for the second Labor seat and Greens preferences widened Jen Butler's margin over Janet Lambert, though Butler would have also won with an even split.  

Greens leakage did contribute to Labor narrowly avoiding elimination in Bass - had the Greens votes that leaked done so overwhelmingly to the Liberals, then Labor would have been excluded and the Greens would have retained the seat.  


One of the things I was looking for here was any case in the last nine elections where votes coming from the Greens delivered a seat to Labor that they would have lost on an even Labor-Liberal split.  I couldn't even find one such seat, despite Greens preferences having been distributed 22 times and favouring Labor in 21 of those cases.  So the history suggests that it won't be common for Labor to need Greens preferences to win a seat.  It could happen though - it happened the other way round in Franklin 2018, where the Greens won a seat on Labor preferences, and would have lost it with an even flow.  And there is at least the one case (Braddon 2014) of Labor losing because the flow from the Greens wasn't strong enough.

There are a few conclusions to be drawn from these numbers.  Firstly, the strength of Greens preferences to Labor changes a lot from election to election, unlike in federal politics where nothing seems to move it all that much.  However, states that have had single-seat optional preferencing systems (NSW and Queensland) have also seen a lot of variation in how strongly Green preferences flow to Labor.  It's hard to say if this is an innate feature of optional preferencing systems or whether it just reflects the weirdness of Tasmanian politics.  Sometimes the major parties are quite similar to each other with the Greens the odd party out, but sometimes Labor and the Greens are quite similar to each other.  

Secondly though, because of the factors that make preferencing not a big deal in Hare-Clark, even if there is a big swing away from Labor in preference flow, it doesn't really do that much.  Labor only closed the gap to the Liberals by 1.4% statewide, and it's not likely that all of that would disappear in one election.  So if Labor think they see a tactical advantage in some ploy to recover lost voters from the Liberals, it's highly unlikely they are going to let the possible loss of Greens preferences stop them.  Whether anyone is actually swung by such promises is another story.

2021 Update (20 May 2021)

Here's the table for 2021:

The 2021 election saw Labor still with the same leader, but without the pokies policy that had attracted Greens voters in 2018.  The election also saw a greatly increased fourth party vote with independent Kristie Johnston elected in Clark, and independents Sue Hickey and Craig Garland going deep into the cutups in Clark and Braddon respectively.  As a result the share of Greens preferences and leakage that went to Others was much greater than ever before, and also the exhaust rate on departing preferences was very low.  

The Greens were excluded in Braddon and had votes had to leave the ticket when their second candidate was excluded in Clark and Franklin.  

The Labor 2PP split over Liberals weakened from 86.2% to 81.3% but this is still the second highest to date.  It's not clear whether the Green votes that flowed to independents would have been different to this, so the split might have been stronger or weaker had these independents not stood.  One other factor that could have weakened the split was the fact that three Liberal candidates were still in the count in Braddon when the final Green went out, compared to two in 2018.  However, in spite of the number of votes leaving the Greens increasing from 10151 to 11820, the Greens' net gain on Labor preferences fell from 1.4% of the statewide vote to 1.2%, because so many preferences went to indies.  Some of these would have come back to Labor later in Braddon in the form of Greens votes that went to Garland.

Greens preferences again did not decide any seat at party or candidate level, and were not close to doing so.  

1 comment:

  1. Nothing to do with the article, just want to wish you a merry festive season and thank you for the posts you've made during the year, even the head scratching ones where the wonk factor goes above one. Makes my meagre donations very worth while. Now to go back and actually read the post.....