Tasmania 2024: Is This Hare-Clark's New Normal?

Before and after ...

TASMANIA 2024: LIB 14 ALP 10 GRN 5 JLN 3 IND 3
Changes from 2021-based notional result: LIB -3 ALP -1 GRN +1 JLN +3

(2021 election for 25 seats LIB 13 ALP 9 GRN 2 IND 1)

(Before 2024 election LIB 11 ALP 8 GRN 2 IND 4)

Counting is over for the remarkable 2024 Tasmanian election and now come the negotiations.  The Jacqui Lambie Network yesterday announced it was expecting to release a confidence and supply agreement within days and independents are also being consulted.  Premier Jeremy Rockliff has stated he intends to request to be sworn back in, agreement to which would be automatic by precedent just to give him a chance to test his numbers even if the Parliament did intend to remove him.  But with Labor seemingly not interested in governing if it relies on the Greens in any fashion, the remaining crossbenchers' choice is to find some way to back the Liberals (at least on confidence votes when they happen) or else back the sort of instability that could see them defending their seats again within months.  If what the crossbenchers actually extract from the government right away (if anything) seems modest or embarrassing, that is one of the reasons for that.

In some ways, the surprise about this Tasmanian election is not that this multi-party magic quilt has happened, but that it did not happen sooner.  In federal House of Reps elections, the proportion of Tasmanians not voting for the major parties has been rising rapidly: 26.7% in 2016, 31.8% in 2019, 39.8% in 2022.  Yet a popular state Liberal government had managed to stem the tide at state level with just 17.1% in 2018 and 23.0% in 2021 voting for non-major-party candidates.  At this election, with that government now ten years old and on the nose over a proposed AFL stadium and constant personal chaos, a Labor opposition still recovering from its own problems and with no federal Coalition government for voters to rail against was unable to make much impression.  Over a third of voters did not vote for the major parties - not really a remarkable level anymore - but the state's electoral system did the rest.

The broad outlines of this likely result had been visible in polling for months.  But it was still always possible it wasn't true.  Maybe there would be a big rally to the party closest to forming a majority as there was in 2006 and 2018.  Maybe the non-major candidates would falter (the polls did overestimate them as it turned out, but not by all that much.)  Was the crossbench winning as many or more seats than Labor just too weird to be true? At the tally room on election night it soon became clear that the polls were more or less right.  The icing on the cake came in following days as the Liberals failed to pick up on postals in Braddon and became at much clearer risk of losing to the Greens or Craig Garland.  (It turned out it was the latter, and in the end it wasn't all that close.)

A proportional election

Hare-Clark isn't a purely proportional system.  Across the state it has segmentation into divisions (which makes life hard for parties with low but fairly uniform statewide support), it has preferences (though they don't change many party-level outcomes), and it also has some arguably anti-proportional candidate effects.  Nonetheless this election has delivered a strikingly proportional result:

It would have been quite proportional under 25 seats too, though slightly better for Labor at the expense of the Greens - I get it as 10-8-3-2-2, so 40%-32%-12%-8%-8%.  

Tasmanians have a parliament that is, in proportion to primary vote support at least, what they voted for.  We shouldn't always expect independent support to be proportionally represented because there are independents and there are independents (the flow between different ones is far weaker than between candidates from the same party) but it happened anyway, thanks to the preference flows in Braddon discussed below.

Many records fall

This election saw many records broken.  Obviously, it's the largest crossbench in the history of the system and the first Tasmanian case where the crossbench outnumbers a major party.  This is something that nationally is extremely rare in recent decades, reserved for wipe-out elections like WA 2021 and NT 2016, but in Tasmania it can happen in an election that is actually relatively close between the majors.  Indeed under any other electoral system in the country Labor would probably have won this election, and under compulsory preferential single-seat systems the Liberals would have lost heavily in seat terms.  (As noted in my earlier summary piece the 2PP might come out around 50-50 for an optional preferential system, which Hare-Clark in essence is, but more like 53-47 to Labor if the same primary votes were cast under compulsory preferential.)

The election saw the lowest combined major party vote share in Tasmanian history at just 65.7%. It saw the highest vote for candidates outside the "big three" (Labor, Liberal or proto-Liberal parties and Greens) ever, at 20.4%.  It saw the highest vote for classical independents ever, 9.6%.  (If one includes "Independent Labor" and "Independent Liberal" then the last election with a higher independent total was in 1948).  

The election saw the sixth highest swing (it finished at 12.05%) against an incumbent Tasmanian government party.  The swing was uneven, reaching 21.9% in Bass in the absence of Premier Peter Gutwein and with football parochialism kicking in against the proposed southern AFL stadium; in the south the swing was single figures.  The government even had swings to it in the odd booth in far southern Clark, but nowhere else of any size.  Labor had swings to it in Clark (they could only go forwards from 2021, especially with Josh Willie bringing them back across the northern suburbs) and Bass (those Liberal votes couldn't all go to minor parties) but were badly hurt in Franklin by competition from their dumped ex-leader David O'Byrne, losing a seat to him compared to the notional 2021 result for 35 seats.  Like this:

In the end Labor finished up as the latest addition to this ignominious list, albeit with an asterisk.

Three of the 25 recontesting MPs were defeated - ex-Liberal defectors John Tucker and Lara Alexander, and Liberal recount winner Dean Young (who lost to within-party competition).  There are thirteen "new" MPs (four Liberal, two Labor, three JLN, three Green and one Independent) but of these one of the Liberals is returning to the House a few years after resigning (Jacquie Petrusma), one has transferred from the Legislative Council (Jane Howlett), and one was a Senator for 28 years (Eric Abetz).  Only one (Rob Fairs) is new to politics.  One of Labor's two is also a Legislative Council transfer (Josh Willie) leaving Meg Brown as Labor's only new talent in parliament from the expansion.  Mostly the chances the expansion offered to build party rooms ended up being grabbed by the crossbench.  The Liberals are now entitled to increase their Cabinet to 11, but at this stage haven't increased their numbers of MPs across the two houses at all.  (They will by one if they can win Prosser, I think I can safely assume they will not be winning the other two.)

Although some seats looked close on election night, none in the end were.  The closest outcome was that Craig Garland escaped elimination by 730 votes (just over 1%).  The final margin in Franklin was Liberal over Green by 2.2% at candidate level, Braddon's final outcome was 1.9% Garland vs Liberal, Lyons in the end was 3.7% Greens over Labor, and Clark and Bass ended up so non-close that the final margins were within-party contests (and those, for what it's worth, were not close either.)  There was an extremely close battle between Liberals Giovanna Simpson and Vonette Mead to see who would escape elimination in Braddon, but it was irrelevant as the winner was drawing dead on Greens preferences anyway. 

The Greens finally won two seats in Clark; there had been several times this had looked possible but not transpired in both the 25 seat and 35 seat systems.  On my modelling this would also have happened in 2002 and 2010 had there been 35 seats at those times, but it didn't happen in 1989 (35 seats) when the Green Independents polled 23.5% in the seat but lucked out because both majors landed more or less on three quotas.  In 2021 a seven-seat system would have returned one Green and two independents in Clark but there were significant swings against the top two independents Kristie Johnston and Sue Hickey (part of this going back to Labor but also to Greens and other independents).  There was a very concentrated vote for anti-UTAS-move independent Ben Lohberger around the area of the University of Tasmania.  As I previously noted, the vast majority of independents running at this election failed.

The Jacqui Lambie Network ran a rather more organised campaign than its first attempt in 2018 and was rewarded with seats in Bass, Braddon and Lyons.  Had JLN not run, the Braddon seat would have gone to the Liberals while the other two would probably have gone to Labor, so the government would not have been near a majority anyway.  A curious feature of the Lambie campaign was that a lot of its voters voted for the party without taking candidate cues into account, leading to some remarkably even primary vote and preference distribution splits, with a high degree of linear voting (1-2-3 down the JLN column) and similar preferencing.

A record that did not fall is that no Tasmanian government has won four majorities in a row.  Four governments have had their streak end at three (Ogilvie-Cosgrove-Cosgrove (ALP) 1937-41-46, Reece-Neilson-Lowe (ALP) 1972-76-79, Bacon-Bacon-Lennon (ALP) 1998-2002-2006, Hodgman-Hodgman-Gutwein (Lib) 2014-18-21).  

Female representation in the two Houses combined is back to 50% (pending the Legislative Council elections at which there will be one male resignation as well as two enforced vacancies).  It got as high as 60% before dropping slightly below 50% with a string of female resignations in the previous term.  

The Garland charge - how did he reel them in?

Craig Garland's surge to victory from a primary vote of 5.1% was a remarkable Hare-Clark event in that no previous lone independent or party ticket in Tasmania has won from less than 7.2% (Benjamin Pearsall in 1934) or from a smaller percentage of quota (40% or 0.4 quotas).  Also, primary vote gaps of 3% are normally not closed down.  Garland's win was the only come-from-behind win this election, with the other 34 seats all going to the parties or independents with whole quotas or the highest remainders.   But this could be a commoner thing in the future - indeed had this election been held under the 25-seat system, my modelling suggests that independents Johnston and O'Byrne and also JLN's Andrew Jenner in Lyons would all have staged similar charges to likely victory (while Garland would have missed out).  

Garland's win going from 3% behind the Liberals to beat them by 1.9% was a story in two parts.  Firstly he had to outlast the Greens' Darren Briggs despite the Greens having over a thousand more primary votes than him.  Here is my breakdown of where they got votes from to the point where Briggs was eliminated:

 

I use the term "leak" to refer to any case where a vote being transferred from a party fails to go to another candidate from that party, when there is one still in the count for it to go to.  This was a high leakage election, perhaps because there were simply more options that a voter could send their vote across party lines to, both at party and candidate level.  So it was a problem for the Greens that their vote was not very concentrated in their lead candidate, and also that when their minor candidates were excluded, often most of the flow went to other minor candidates meaning that votes could leak again. I think it's also possible that the Greens' very limited promotion of their support candidates increased the Green leakage rates.  Outside of Clark it was common to see leakage rates around 20% on Greens transfers, whereas at previous elections low teens was more common.  

Lone independents do not leak, they continually gain and continually take leakage from other parties, and so they are competitive from somewhat lower quota totals than party groups.  In this case leakage out of the Greens ticket (including to Garland) itself erased most of the Greens' lead.  But even with no leakage from the Greens at all, Garland would still have beaten them (just!) because he outperformed them on leakage from other parties (particularly JLN) but also on preferences, especially from the Shooters, Fishers and Farmers party and also from fellow independent Peter Freshney.  I thought the flow from SFF to Garland would have been reduced by also having JLN in the mix, and it was reduced a little bit, but not that much.  Garland also had a peach of a ballot draw for picking up the preferences he needed (given that voters have some tendency to preference nearby columns over far away ones) - he was surrounded by the Shooters, Animal Justice and ungrouped (including Freshney) columns.  The main reason for the preference flow from SFF is simple - the "Fishers" part of the SFF support base, as Garland is himself a fisherman who has been very prominent on a range of recreational fishing related issues.  There's also a degree of outsider/rebel appeal to both forces.

In retrospect the Greens never had a chance of keeping Craig Garland behind them - any one component of his survival at that stage can be removed and he still passes them.  This then is his comparison with the Liberals:

Garland started over 2000 behind but the leakage effects off Liberal votes nearly erased that by themselves.  That said, the Liberals were better at accumulating leaks through the count (because there were at most stages more Liberals to get them) so some of the leaks off the Liberal ticket would not have actually started as Liberal votes.  Garland did slightly better than the Liberals off preferences leaving other groups or candidates but he was still just behind prior to the throw of Green preferences, which overwhelmingly favoured him.  Although parts of the Greens support base have become estranged from Garland over COVID controversies, 41.4% of the vote-values leaving Briggs (some of them via a surplus for Miriam Beswick) finished with Garland, compared to only 8.9% landing in the Liberal pile.  

This seat was actually the only case where Greens preferences had any impact at this election, since generally the Green votes were all used electing or failing to elect Greens.  In Clark and Bass the Greens hit their second and first quotas on external transfers and passed on no surplus of their own, in Franklin their last standing candidate finished eighth so was never excluded, and in Lyons they were elected without quota.  Labor at one stage publicised this article which pointed out that Greens preferences have little impact on major-party contests at Tasmanian elections; at this election they had even less than ever before!  But instead of Greens preferences causing Labor to win seats from behind as elsewhere in the country, Green votes helped Labor and the Greens to win one more seat than the Liberals.  However, Labor chose not to do anything with that.  

Informal voting and exhaust

The informal vote of 6.31% (up from 5.13% in 2021) was too high.  This rate reflects poorly on the current government, which ignored my call to consider temporary savings provisions to, for instance, ensure that votes that were formal in 2021 would be formal in 2024 too.  We will have to wait a while to see what the causes of the informal rise were, but many scrutineers including me saw some votes that would have been formal in 2021 be not counted in 2024.  It is a simple change management failure.  In the absence of a regular JSCEM-style process an inquiry should be held into informal voting at this election, including why the government did not include savings provisions and whose and what advice that decision was taken on.  The informal rate is the second highest in Tasmanian history (behind a 10.09% rate in 1946 caused by the introduction of ballot columns, accompanied by some rather confusing instructions).   It's also the highest in any Australian state or federal election since NSW 1991.  Fortunately, as there were no close results, it's highly unlikely any winners would have been different with a lower informal rate.  

The statewide exhaust rate (the proportion of vote-values ending up leaving the count) was 5.5%, with Bass the highest with 7.4%.  As a quota share (0.44 quotas average) this is very similar to 2014 and 2018, but much higher than 2021 which had very low exhaust rates (3.7% or 0.22 quotas average).  The structure of the preference distributions can have a large effect on exhaust rates, since major party voters often vote for their own party then stop, and contests that end up between candidates from the same party often see a lot of voters not numbering that column.  It's important to bear in mind that 5.5% of vote values exhausting does not mean 5.5% of voters' votes left the count entirely - exhaust includes votes coming from surpluses where part of a voter's vote value had helped elect somebody but part of it had later left the count.  

The polls were pretty good

Polls by five pollsters were released this year or nearly so - YouGov, EMRS, Redbridge, uComms and Freshwater.  However the commissioning source for two polls (one known to be Freshwater and one believed to be) did not cause them to be published with full details, instead just giving them to media sources where they were somewhat incompletely reported.  Not good enough on the part of the commissioning source; voting intention polling should be a public data science not a "psst journalist wanna poll on the sly" exercise. If such results are to be reported then the commissioning source should release them.   To do otherwise is unTasmanian, or all too Tasmanian.  

The YouGov poll in January was too early to be considered a final poll and also had some eccentricity in the handling of independent/other votes that meant it wasn't really comparable with the other campaign polls.  But it was still a useful modelling input.  Of the campaign polls that I had full data for, this is how they scrubbed up compared to the final results:

While uComms performed worse than the other two on average error and major party gap error (this mainly a result of having the Labor primary wrong by 6% after having the Liberal primary wrong by 7.4% in 2021), the uComms was the only poll to provide a remotely accurate estimate of the non-Independent Others vote.  Of the two "mystery polls", the first which I believe to have been by Freshwater appears to have been very accurate indeed (with perhaps a slightly higher average error than Redbridge but getting the major party gap dead right), but the later one which was reported as Freshwater had the Labor primary too low compared to the final results.  This poll plus the uComms being the last two to come out fuelled fears that Labor was on for a diabolically bad result, but actually Labor landed where Redbridge had them at the top end of the polling range. Redbridge did have the Liberal primary 3.7 points low compared to the final result but on the whole a pretty good debut especially with a relatively small sample size.  Redbridge were also closer than I thought with their projection that the Greens would win about 6 seats off 14% - the Greens in fact won five off about that, and narrowly missed two more.

One issue with all these polls is that none were taken in the last two weeks.  This means we can't really know whether voting intention changed or whether some of the polls were significantly wrong when taken.  I for one doubt that the Labor primary started in the mid to high 20s, dipped down to the low 20s then sprang up again, as there was nothing obvious to drive this big a change.  But I suspect also that all along the Lambie Network vote was soft as populist party votes often are, and that the Liberals' attack on JLN in the final weeks may have genuinely driven some voters back to Labor.  Just not 6% worth.

At this election I as usual published a polling aggregate, albeit very late in the piece.  I have been doing polling aggregate based projection for six Tasmanian elections (although the 2021 effort barely deserved the name with so little polling) and so far it has always got the seat result either exactly right or within one seat of the total for each party; this year was no exception.  My projection off the published aggregate was 15-10-4-3-3 and the result was 14-10-5-3-3.  Overall the vote shares projected in the polling aggregation had Labor too low and JLN too high across the board, had the Liberals close enough to right everywhere, and had independents mucb too high in two specific seats (Clark and Lyons - the latter being partly my fault for not allocating a higher Others vote in Lyons).  The polling aggregate's one clear miss in Clark was caused by the estimated Independent vote being too high (and the Green vote also too low in that specific seat).  The case of Craig Garland in Braddon was one which I did flag as a possible scenario (including that he might win from 5%) but it was hard to know how to allocate the independent vote in polling between him, Freshney and wishful thinking.  The higher Labor primary than projected in Braddon also helped Garland win because Green votes were no longer being used up putting Labor over the line.

Anyway great to see diverse polling back in Tasmania and that it did a good job of capturing the overall picture.  

The paradoxes of PR

It's very probable that a narrow majority of Tasmanians would have preferred some sort of Labor government to the Liberals.  And yet, because of Labor's commitments on the campaign trail to not give away ministries or do any deal with the Greens, and because Labor finished up so deeply in minority then decided to declare the election lost rather than see if the Greens would give them confidence and supply for nothing, it seems the Liberal government will continue.  This is not the only time a Tasmanian election has ended up with a result the voters apparently didn't vote for - in 1989 Robin Gray's Liberals would have won a 2PP vote easily, but because the Greens as a block decided to deal with Labor, a Labor-Green government that only a minority would have wanted was elected.  Because of the decisions that voting blocks make in the parliament, proportional representation sometimes does not equal proportional power.   Indeed if one wanted to vote for a party or prominent independent that was clearly committed to throwing out the Liberal Government in any hung parliament, no such voting option existed.

The challenge for the major parties here (assuming Hare-Clark can survive results like this!) is that it is likely about 45% of the vote will continue to be the target for majority government.  But if the non-major vote continues to be this high, that will require knocking the other major party down to a very low level of support (around the low 20s or less).  So unless this one is a bubble in non-major support, even very lopsided elections might not quite deliver majorities.  Perhaps this is actually a bubble: JLN could implode, David O'Byrne could be reintegrated into Labor, and governments of this age have gne totally off the rails and been routed before.  But absolutist majority/minority rhetoric will continue to bite major party leaders who engage in it.  It needs to be temopered with a reality that, as other states have shown (eg Steve Bracks' first term), some minority governments are very workable and successful.  Whether whatever this is we've just elected can be one of them, only time will tell.