"Safe Seats" Falling Is Nothing New

The Australia Institute and its director have been putting out a lot of its usual Hung Parliament Club type stuff about how "power sharing parliaments" are the new normal, how there are no safe seats anymore and so on.  They've been trying to claim that the rise of teals and the decline of major parties means the traditional 2PP swing-based model is more or less dead, although I actually nipped that view in the bud days after the election.  If major party vote shares keep declining we may sooner or later get to a point where 2PP swing-based models cease to be of much use, but 2022 wasn't even close.  See also here, where I point out that the Coalition didn't actually get a raw deal in the "non-classic" seats and what actually caused it to lose so heavily in 2022 was that Labor beat it on 2PP and thumped it on 2PP distribution in the classic Labor vs Coalition seats.

The "power sharing parliaments" analysis misleadingly lumps stable Coalition majority governments and non-majority upper houses in with the sort of thing we saw in 2010.  They're totally different: a true minority parliament involves a government that must make a fresh negotiation for supply and confidence and that continually depends on the crossbench for those things.  (Yes the Coalition has its own internal arrangement but it's a long time since there's been the slightest doubt that the Nationals or their precursors would continue to support a Coalition government).  When there is a "hung Senate" the passing of legislation is often at stake, but except in the most extreme cases supply is not, confidence is not, the composition of the Executive is not.  Hung Senates aren't generally perceived as causing potential stability issues, and the ability of governments to send them to double dissolutions if they keep blocking things can make it easier to browbeat them than it is to browbeat minority Reps crossbenchers.  The most successful governments use Senate obstruction, where it happens, to extend their own lifespans, by being able to signal to their base without having to put up with the consequences of policy their base likes being passed unamended.  A government majority in both houses can easily go to a government's head - cf Howard 2005-7 and Workchoices.

This article concerns the claim that the 2022 election marks some kind of radical dismantling of the concept of a "safe seat".  The AEC defines seats as "marginal", "fairly safe", or "safe" according to whether the 2PP winner's margin is 0-6% (50-56% 2PP), 6-10% (56-60% 2PP) or over 10% (60+% 2PP) of the two-party preferred vote at the previous election.  The AEC document simply excludes seats that finished as non-classic at the previous election altogether, so even a seat like New England isn't listed as "safe" although its 2019 margins were 14.4% vs IND, 17.6% vs ALP.  For the purposes of this analysis, I'll also include those sorts of seats.  

Safe Seats Have Been Falling For Decades

In 2022, three on paper "safe" seats on 2PP (and 2CP) fell to independents.  These were Mackellar (Lib 13.2% vs ALP, lost by Jason Falinski to Sophie Scamps), Curtin (Lib 14.0% vs ALP, lost by Celia Hammond to Kate Chaney) and Fowler (ALP 14.0% vs Lib, vacant on the retirement of Chris Hayes; Kristina Keneally (ALP) lost to Dai Le).  Mackellar and Curtin were part of the general teal trend that also picked up four seats on lower 2PP margins, while Fowler was an (almost, see below) isolated case of bad preselection.

That's an unusually high number, but another 61 2CP-"safe" major party seats and 3 "safe" crossbench seats didn't fall.  And the odd case of "safe" seats falling is not a new thing.  Here is a list of the 2CP-safe seats that have fallen since 1996:

1996 Lindsay (NSW, ALP 10.2%): Lindsay fell on a 11.8% swing to Jackie Kelly (Lib), unseating incumbent Ross Free.  Kelly was later ruled ineligible but won the resultant by-election with a swing to her.

1996 Oxley (Qld, ALP 12.6%): Oxley was famously lost to Pauline Hanson, who was deselected as a Liberal candidate after the ballot papers were printed (for contentious comments against First Nations government assistance programs) and ran as an independent while still listed as a Liberal on the ballot papers.  

1998 No classic safe seats fell but the Liberal Party recovered Kalgoorlie (IND vs ALP 10.4%) and Moore (IND vs ALP 13.3%) where it had failed to even make the 2CP against defecting ALP and Liberal incumbents respectively in 1996.

2001 Kennedy (Nat 11.2%) was retained by Bob Katter who defected to sit as an independent.

2001 New England (Nat 13.7%) was lost by the Nationals' Stuart St Clair to independent Tony Windsor.  Windsor had been state MP for Tamworth for ten years prior.

2004 No safe seats fell.

2007 Vacant seats Forde (Lib, Qld 11.5%) and Leichhardt (Lib, Qld 10.3%) fell to Labor off the national swing, vacant status and the popularity of Kevin Rudd in his home state.  

2010 Denison (Tas, ALP 15.3%) fell to independent Andrew Wilkie after Labor, on the retirement of 23-year incumbent Duncan Kerr, preselected a little-known candidate who was the son of a recent state MP.  Wilkie had narrowly missed election to state parliament in the same seat earlier that year, and narrowly won from third on Green and Liberal preferences.   This loss was something of a precursor for Fowler.   (A case can also be made for including the loss of O'Connor (WA, Lib 16.6%) to the WA Nationals who were not at that time strictly part of the Coalition.)

2013 Lyons (Tas, ALP 12.2%) was lost by Labor's Dick Adams to the Liberals' Eric Hutchinson off the national swing fuelled by forestry issues and the unpopularity of the Tasmanian Labor-Green coalition government.  (The seats of New England and Lyne which had been won by "safe" margins by independents in 2010 returned to the Coalition when the independents did not recontest them.)

2016 Mayo (SA, Lib 12.5%) was lost by the Liberals' Jamie Briggs to the then Nick Xenophon Team candidate Rebekha Sharkie.   Briggs had had a scandal-ridden term.  

2019 Warringah (NSW, Lib vs Grn 11.6%, Lib vs ALP 11.1%).  An attempt to bump off out-of-touch ex-PM Tony Abbott had failed in 2016 after opponents failed to settle on a convincing lead candidate but in 2019 there was no such trouble and independent Zali Steggall thrashed Abbott in a prototype of the teal campaigns of 2022.

All up then there were at least ten examples of major parties losing 2CP-safe seats in nine elections from 1996 to 2019.  Only four of these were straight-up two-party swaps and the rest were to some sort of crossbencher.  

Prior to 1996, safe seats really were safe for a while.  I found three that fell in 1961, one in 1966, two in 1969, Franklin (Tas) in 1975 - surprising it was only one that year - and then what seems to be a 21-year drought.  All these were two-party swings.  

Teals are not random winners

Not only were the 2022 "safe seat" losses precedented, they were also far from random.  Going into the election it was pretty obvious there was a class of seats where the Coalition was coming under major challenge, even if it was not clear how many of them would fall.  These were in general its most socio-economically priveleged seats.  Ranked by SEIFA score, the Coalition dropped seats 1, 2, 3, 6, 8 and 12 in the nation to teals, having also lost number 4 at the previous election.  What happened with the teals was unusual; previously independents had for a long time only gained seats where they were a defecting incumbent, the seat was vacant, the incumbent was unpopular or the independent had a huge profile.   

Richard Denniss claims that it is now the case that there are " [..] no seats that are safe against a strong candidate with a well-funded community campaign focused on the issues of concern to voters in that electorate." But what of all the other indies who failed to win, some of them despite spending hundreds of thousands of dollars?  Were they not "strong" enough candidates?  Were the voters not "concerned" enough?  It's easy here to engage in annotation by result when the demographics are telling a very obvious story about who won and who didn't.

Teal independent campaigns winning very wealthy seats does not prove that random rural, regional or  suburban incumbents will lose "safe" seats to indies just as soon as they have a good enough opponent.  What has happened with the teals in 2019 and 2022 is a new class of competitive seats, but it isn't any more proof than before that any old seat will now drop.  Rural indies have been trying to win federal seats (2PP safe or otherwise) for decades and the only pickups have been by extremely high profile candidates (Windsor, Oakeshott and Andren) and by Voices for Indi at the expense of a particularly disliked Liberal incumbent.   

There has been a lot of speculation about "Muslim vote" indies as the potential teals of 2025.  Could they dislodge incumbents in 2PP-safe Labor seats?  Perhaps they could if the UK election is anything to go by, though it will be challenging for such candidates to openly signal against Labor's position on Gaza and yet still get Coalition preferences.  Perhaps they will rely on endorsing groups to do all the signalling for them.  But say it happens, this will again not be proof that any old seat will go; it will be another new class of trouble that will affect a certain range of seats.  

It should be noted also that while Denniss correctly points to the standard 2PP pendulum doing a very poor job of predicting which specific seats would fall in 2022, it was never designed to do that.  The claim made on its behalf has been that it predicts how many seats will fall.  It didn't do a good job of that either from the Coalition's side, but there were reasons for that besides the Coalition losing six seats to independents.  Specifically, while the pendulum predicted the Coalition would lose the 2PP in eight of its seats, this in fact happened in twelve (two of which were won by the Greens).  This is an old problem, also seen in 1998 for example, that sometimes the two-party swing is uneven in a way that matters when measured against seat margins.  (1998 was a spectacular example of that, but 2019 also cited by Denniss was not at all - the swing that year was uneven too but the Coalition did nothing special out of that in seat terms, because of a conservative marginals strategy).  

The argument that parties and the media should downplay marginal seats because the 2PP swingometer did a poor job of predicting which seats actually fell is a silly one, because nobody actually knows for sure just what the swing will be.  And in fact there still was a pretty strong relationship between 2CP marginality and results in 2022.  The Coalition lost six out of nine seats (66.7%) on 4% or less.  They lost 5 out of 12 (41.7%) on 4% to 6%.  They lost 5 out of 17 (29.4%) on 6% to 10%.  They lost 2 out of 38 (5.3%) on more than 10%.  While this does suggest seats up the tree shouldn't be ignored, it's still the case that the most 2CP-marginal seats are the seats most likely to fall; in this case 11 of the 18 seats the government lost were its own marginals, and just over half of its marginals fell.  It's also wrong to suggest that either the media or the parties ignored the seats on higher margins - in the case of those that were actually competitive (and even some that weren't) far from it.   Media articles of the form 'Generic Indie Announces Run For Boring Seat Blah OMG They Might Win!' have been a dime a dozen at recent elections state and federal.  

And one more thing from Denniss's latest effort: yes the swing difference between two states (WA and Tasmania) was unusually high, but so what? 2022 had the greatest spread and the greatest standard deviation in state swings since 1943!  But this was not because "different cohorts of Australians now consume radically different news online, and care about radically different issues" or any such other nonsense plucked out of thin air.  It was because (i) there was an exceptionally good result for Labor in Western Australia as a result of COVID politics and probably the interrelated popularity of then-Premier Mark McGowan (ii) in Tasmania there was a poor swing result that partly reflected demographic transition in the northern Tasmanian seats but also reflected personal vote factors in the 2019 and 2022 elections.  (In 2019 the Liberals had had to disendorse their Lyons candidate while in 2022 Labor's candidates had adverse publicity in both Braddon and Lyons and two new Coalition incumbents had personal vote benefits.)

Federal elections are becoming more complex and both major parties now must fight on multiple fronts and not just one.  Despite this, the winner of the major party battle in each given seat still won nearly 90% of seats in 2022.  If either side chose to downplay the classic marginal battles in pursuit of fights in the minority of non-classic seats, that would be an extremely foolish move.