WA 2021 – the shape of the swing


By now we all know to expect a landslide victory for WA Labor in this month’s state election. The polls predict it, and even the Liberal leader has acknowledged it.

But I want to put what might happen in context by looking at what happened in 2017, which seats flipped, and how that compared to previous elections.

There hasn’t been a single election since 1980 which has seen a swing anywhere near as big as we saw in Western Australia in 2017.

The following chart shows the two-party-preferred vote for Labor and the Liberals and Nationals at every state election since 1980.

Two things jump out: the last election produced the biggest win for Labor over this period, and the previous election in 2013 produced the biggest win ever for the Liberal and Nationals parties.

But the 2013 result wasn’t due to one big surge in the vote. The conservative parties had gained swings at three successive elections, slightly gaining support at the 2005 election, when Labor won a second term in power, before gaining larger swings in 2008 and 2013. The biggest swing to the conservatives in this recent era was 5.4% at the 2013 election. So they built their way up to a large majority.

The Liberal/National government also won with a large 55%+ majority in 1996 before losing in 2001, with Labor gaining a swing of 8.1%, which had been the record over recent decades prior to the 2017 result.

In this forty-year period, a party has held on to power while losing the 2PP just once, in 1989.

This tells me a few things:

  • Considering that Labor won the last election with a huge swing and with their biggest ever result, it’s remarkable to think that we are now working on the assumption that Labor will gain further ground, perhaps quite a lot of ground, in 2021.
  • A landslide victory to the Liberals and Nationals in 2013 produced a large upper house majority, with 22 out of 36 seats, while Labor only won 14 out of 36 with a result almost as strong in 2017. That’s a topic I will return to in the coming weeks.

Next up, let’s look at where those swings landed in 2017.

This map (the first one I’m doing with some new technology, so bear with me) shows the swing to Labor at the 2017 state election. This is based on two-party-preferred so it ignores any Liberal vs Nationals contests and simply looks at Labor’s vote against the stronger of those two parties. There was only one seat where Labor went backwards. That was Central Wheatbelt, where Labor came a very distant second with less than 20% of the vote.

The map has highlighted seats that Labor gained with black outlines.

The map is zoomed out to show the whole state, but as you zoom into Perth you see the swings become much bigger. Labor gained a swing of at least 6.7% in every seat across the Perth metropolitan area, with swings of at least 11.5% or more across the East Metropolitan region.

The average swing to Labor was 12.6% in the northern suburbs, 13% in the southern suburbs and almost 15% in the eastern suburbs. This compares to just 6.6% in the Agricultural region and 8.5% in Mining and Pastoral, and 11.2% in the south-west.

This translated into 21 seats flipping to Labor (including Collie-Preston and West Swan which Labor had held in 2013 but had been redrawn as notional Liberal seats in the redistribution).

Labor gained eight seats in the eastern suburbs, giving them all fourteen seats in the region. They also gained six in the north, three in the south and three in south-western WA. There was also one gain in the Pilbara.

These results mean there are a large number of seats with first-term Labor MPs who are now up for election for the first time. It wouldn’t be surprising if the Labor swing was concentrated in those seats. We will see how much of a swing is available and how many more seats Labor could add to that tally later this month.

Liked it? Take a second to support the Tally Room on Patreon!


  1. It wouldn’t be surprising if the Labor swing was concentrated in the seats with first-term Labor MPs? Prepare yourself for a surprise Ben. The swing will be biggest in the currently-Liberal seats. The Westie’s straw-poll showed that, and it’s consistent with the feeling that the swing is because people, especially older people, feel protected by Mark.

  2. I think it will simply be proportional to the Liberal vote. McGowan seems to have captured a demographic that isn’t the typical “swing voter” demo. I predict huge swings in once safe Liberal seats, tiny swings in safe Labor seats.

    There will be large swings in seats with first term MPs, and swings in Liberal held target seats, but I think that is a consequence of the above.

  3. The Labor swing in the last election was larger in areas with more middle-to-working class young people, especially families, living in new estates. I suspect Labor has nearly maxed out their possible support in some of these places, whereas, as Jack points out, that may not be so among different demographics.

  4. Could the large swings to Labor in the previous election, despite McGowan’s current apparent popularity, mean that they could slide slightly in this election?

  5. ^^Despite the large margin of victory at the last election, we have the unprecedented scenario of McGowan being approved by ~88% of voters. Given that, and the general polling we have seen it would seem impossible that Labor will go backwards, and it is hard to see a scenario where Labor gets less than 60% of the 2PP vote. We are likely to see a 5-10% swing to Labor which would be remarkable given the high base they start from.

  6. I think the only seats Labor could slide in are Albany and Pilbara. Regionals always behave different to city.

    Albany because we don’t know how large the effect of Watson’s personal vote, which has held that seat for Labor for several elections longer than thought.

    Pilbara is a low chance, just because last election was weird with the Big Miners being anti-Nats/Grylls specifically. So I am not sure what the campaign is like up there but could be pushed back to being a Nats seat.

  7. The 2017 swing to Labor was partly a correction for the swing against them in 2013. That election was at the tail end of the Rudd/Gillard govt, and plenty of people were doing what Uncle Rupert told them to and voted against Labor. Seats like Forrestfield (a Labor marginal they were trying hard to keep) and Swan Hills (with the Libs’ broken promise of the train to Ellenbrook) are a good indicator of what WA 2013 could’ve been without federal interference.

    That 1989 election was a lucky escape for Labor. WA Inc had broken by then, and there’d just been a major redistribution cooked up between Labor and the Nats to freeze out the Libs (this was when Labor had a lot more country seats than they do now). They lost the 2pp by 0.5% worse than in 2008, but they got away with it.

  8. 10 days to go… while I obviously don’t expect my side to win (I’m a card carrying Liberal, voted AC in 2016 in protest of Turnbull but otherwise first preference Liberal since my first vote in 2010), I have enjoyed helping out on the Liberal campaign, it’s helped me get through the pandemic.

    And of course, I’ll be watching the returns on the night – will the swing be like NSW in 1981 (when Neville Wran’s Labor won 69 out of 99 seats, an increase of 6 seats)? Time will tell…

  9. As somebody who’d otherwise be voting in Churchlands if I still lived in Western Australia, I seriously doubt the swing will be big enough for it to change hands even with the Porter situation, the border backflip is probably enough to shave a couple % off the swing to Labor there and in Cottesloe, Nedlands and Bateman, which are the four metro seats I don’t see budging at all.

  10. There seems to be some indicators that the projected swing is in one area the ALP have never performed well – the over 65s – and that can be sheeted home to the way that Covid has been handled.

    What this means though is that the swing is likely to be big in areas with older populations. I’d actually expect there to be little swing to the ALP in their traditional areas, as most of that would have been gained at the last election, meaning it has to come from Liberal/marginal areas. I wouldn’t be surprised to see the Libs hold nothing by more than 5% with big swings in the “golden triangle”, and it bodes badly for Kirkup in Dawesville as well.

    What I’d also expect is that we’ll get a number of no-chance-of-more-than-one-term members if the ALP do pick up seats like Nedlands or South Perth that they *never* hold.

  11. Matt

    Indeed in a normal election Labor could nominate any old no hoper in the safest Liberal seats knowing they’re no chnace at an election (or a yoing chancer needing their first experience at a campaign).

    What I’d hope is that WA Labor had seen the swing is on, and have nominated stronger candidates in these seats. Certainly the Nedlands candidate looks pretty good and certainly well credentialed.

    And there’s no guarantee they’d be one-termers either. Ronan Lee won Indooroopilly for Labor in 2001 and was reelected twice in what had previously been safe Liberal territory.

  12. May I also add another seat from what Matt said. Port Darwin in NT

    Port Darwin was considered rock solid for the CLP previously but the incumbent Labor MP managed to win again while other traditional CLP seats (Katherine and Braitling come to mind) switched back to the CLP

    While I think most of the Freshmen MP’s will be 1 termers not all of them will be. And I wouldn’t be surprised if the 2025 election results in a similar result to 2017 and it might take 8 years (2028/29) for the Liberal party to regain power if the swing is as bad as the polls say. Nevertheless the reason older blokes are typically chosen for safe seats is because if they do win they don’t expect to be in politics long (Like how John Kennedy (Victorian politician) won Hawthorn for Labor in 2018.

  13. Daniel I can assure you that not a soul thought Kennedy was a hope in Hawthorn. It was certainly Labor’s greatest scalp as it robbed the Liberals of a highly credible leadership contender in Pesutto.

    Do I think Labor will win Hawthorn again in 2022? Not for a second.

    But there is a real danger for the Libs in WA if the predictions of demographically older electorates swinging to Labor comes true. A good member in such an electorate can build a substantial personal following with voters who like them, but not their party.

    The LNP in Queensland (if it survives as a merged party) will, for this reason, have an absolute shit of a time winning back Hervey Bay, Caboolture and Nicklin – expending far too many resources that could be better spent on other seats.

  14. The ALP will probably win South Perth easily, primarily because John McGrath, the popular local member is retiring and not endorsing the new Liberal candidate. There was a 3-4% swing against Steve Irons in South Perth last federal election, and considering the lack of an incumbent, an unpopular Liberal Party and a small “l” Liberal demographic that favours Labor at this election. McGrath’s personal vote is upwards of 4% IMO, and with the margin being 7.2%, that sizeable reduction means a swing of any more than 3.2% sees Labor win the seat. The swing will be mostly centered on generally non-ALP seats, like South Perth (as opposed to Armadale). South Perth is unchanged demographically, and the Libs will likely win it back in 2025. If they don’t, hell may as well of frozen over….

  15. Oh and also the Labor candidate is unfortunately typical of ALP candidates in LP safe seats, basically being a bumbling staffer to the Premier…

  16. There are a series of seats which have never voted labor. Eg Nedlands and Cottesloe.. South Perth. They are very wealthy…Hard to see they could shift
    Also rural seats like vasse and at least 3 np held seats are not areas of labor support…. Then again if 70/30 % 2pp the swing must go some where

  17. If the infamous Newspoll is correct (even within the lower MOE), then I’d expect the ALP to pick up Hillarys, Darling Range, South Perth, Riverton, Bateman, Scarborough, Nedlands and Dawesville, while Churchlands and Carine would be flip a coin. Cottesloe would be the final holdout if it was really bad, and the Nats would be the formal opposition. I don’t think it could be though.

    None of these seats would even be in discussion normally other than perennial Labor target Riverton.

    That’s before we even talk about regional seats: Gero would likely be a pickup, and Kalgoorlie would be a lucky dip as it often is, along with North-West Central. It’s a good time for the ALP to change members in Collie-Preston and Albany too, where the incumbency factor is not likely to be decisive.

    If you split the difference between Newspoll and 2017, then I’d expect the ALP gains to be Hillarys, Scarborough, Riverton, and Dawesville, with Bateman, South Perth, Nedlands and Darling Range in coin flip territory, and the other mentioned seats possible big swing outliers.

    TBH, the status quo would be a massive result for the ALP as 2017 was a historic level result. Anything on top of that and we’re looking at high water marks for the ALP Australia wide.

    My gut feel is that the end result will be the ALP picking up Dawesville, Hillarys, Riverton, Geraldton and any two others from the list. The only truly safe Lib seats are Cottesloe and Vasse.

  18. I think we can all agree that Mia Davies seat of Central Wheatbelt will remain the safest seat after this election. She will likely remain the leader of the WA nationals even if she loses 1 or 2 colleagues. The question is will Labor get a swing to them in Central Wheatbelt and if so how much? I predict maybe a small 2-3% swing to Labor in CW but way under the state average

  19. Also got to remember there is no coalition in WA. The nationals sit on the crossbench, so if they form the opposition would likely be the first time in Australian history that the Liberal party has sat on the crossbench. Could be a historic result!

  20. Half of central wheat belt is what used to be Avon which was Northam and near by had an alp mp as late of 1983 may be the swing could be there but still does not. Change what was said

  21. Northam’s a railway town – it was Labor back in the day because of all the unionised railway workers. Those jobs largely don’t exist any more, thanks to mechanisation and containerisation, etc. Labor held it into the 80’s because their MP, Ken McIver, was a train driver before politics – similar to Mick Murray in Collie. Labor can still win West Northam and Avonvale booths in a good year (the state housing parts of town), but that’s about it.

    Strange to think Labor held all of Avon, Warren, Kalgoorlie and Geraldton within my lifetime. The country’s certainly changed since then.

  22. The big change was the one-vote-one-value changes. What that did was enlarge a lot of the electorates for regional centers to have to include a bigger non-urban hinterland. Which screws the ALP in those seats.

    Geraldton for instance now contains a lot of the nearby Greenough area – were it still the old core electorate, the ALP would likely have won it in 2017, and it would normally be a marginal, but the non-urban component usually puts it out of the ALP’s reach. Labor might hold it occasionally, but only in high-watermark elections.

    This is what makes the ALP somehow still holding Albany so remarkable – OVOV should have put it out of the ALP’s reach, but Watson’s personal vote has held it in the ALP column. I’d expect that once the Libs regain it – if not this election which seems unlikely, it wouldn’t shock in 2025/2029, I’d expect the ALP to very rarely hold it again.

    Bunbury is probably the only exception in this – it’s the only regional town with a high enough population to keep an electorate entirely in it’s core, therefore being a viable seat for the ALP to win. The other exceptions are the far northern seats, Pilbara and Kimberley, but Pilbara is essentially based on Karratha and Port Hedland, and Kimberly has a large Indigenous vote, which is very strongly ALP. And both have large remote-electorate allowances.

  23. OVOV certainly didn’t help, but demographic change was already happening before then. There’s no way Labor would ever hold a seat based on Manjimup these days, for example.

    Bunbury’s big enough it could probably have two mostly urban electorates (the 90’s seat of Leschenault would work), but then you run into the problem of what to do with all the other towns currently in Murray-Wellington and Collie-Preston. One very big rural electorate stretching from Pinjarra to Donnybrook via Collie, probably.

    As to this from upthread:

    “the first time in Australian history that the Liberal party has sat on the crossbench”

    The Libs were on the crossbench in Queensland for the last two terms of the Bjelke-Petersen govt. Very different circumstances, obviously.

  24. Well Ben, now that it’s nearly all over I’ve done a plot of the swing in 2021 against the Lib or Nat vote in 2017 – swing according to Antony’s latest (incomplete) swing figures and the 2017 vote adjusted for the redistribution according to Antony’s guesstimates. No surprise – fairly wide scatter but if a curve can be threaded through it it starts and finishes down at the safe-ALP and safe-conservative ends and curves up in the middle-ish seats. When I’ve learned more about graphs and plots in Excel I hope to mark the ALP sophomore seats – as you predicted they’re generally higher up the plot. How do I send you a copy?


Please enter your comment!
Please enter your name here