2021 Federal Redistribution – Draft Boundaries for Western Australia

UPDATE 4 June – the AEC finalised the boundaries today with some minor nips and tucks. Data files have yet to be published, but the changes described in the final report do not suggests any significant changes to the new margins set out in this post. The new boundaries will be gazetted on 2 August, which perhaps gives a hint that the Prime Minister won’t be calling an election for Augsut.

Last year’s review of state representation in the House of Representatives recommended that Western Australia lose a seat, its representation falling from 16 to 15 members.

The draft boundaries are released at noon eastern time and I will update this post through the day with information on the new boundaries and estimated new margins.

In summary, the Liberal seat of Stirling is abolished and there is not much shift in margins for other seats.

The change in margins shown in the table below don’t show much shift in margins, but in going from 16 to 15 seats there are substantial changes in boundaries.

As expected, there has been major surgery on Pearce. It has lost its rural districts to Durack and O’Connor and lost Ellenbrook to Hasluck. Pearce gains Wanneroo and the suburbs east of Lake Joondalup and north of Hepburn Avenue from Cowan, so the seat is now entirely in the northern suburbs of Perth.

The abolition of Stirling has been easily achieved, the Labor end of the electorate east of the Mitchell Freeway was been transferred to Cowan and Perth, the areas to the west divided between Curtin and Moore.

Most other changes are consequences of abolishing Stirling and the major movement of Pearce. Hasluck gains Ellenbrook from Pearce and sheds Forrestfield to Swan and Kenwick and Maddington to Burt.

The table below summarise the estimated new margins. I’ve made no distinction between Liberal and National seats.

Division Old Margin New Margin Change
Brand ALP 6.7% ALP 6.7% 0.0 to ALP
Burt ALP 5.0% ALP 5.4% 0.4 to ALP
Canning LIB 11.6% LIB 11.3% 0.3 to ALP
Cowan ALP 0.8% ALP 0.9% 0.1 to ALP
Curtin LIB 14.3% LIB 13.9% 0.4 to ALP
Durack LIB 14.8% LIB 13.5% 1.3 to ALP
Forrest LIB 14.6% LIB 14.6% 0.0 to ALP
Fremantle ALP 6.9% ALP 6.9% 0.0 to ALP
Hasluck LIB 5.4% LIB 5.8% 0.4 to L/NP
Moore LIB 11.7% LIB 11.6% 0.1 to ALP
O’Connor LIB 14.5% LIB 15.4% 0.9 to L/NP
Pearce LIB 7.5% LIB 5.2% 2.3 to ALP
Perth ALP 4.9% ALP 3.2% 1.7 to L/NP
Stirling (abolished) LIB 5.6% .. ..
Swan LIB 2.7% LIB 3.3% 0.6 to L/NP
Tangney LIB 11.5% LIB 9.5% 2.0 to ALP


There were whispers before the redistribution that Pearce would be abolished. In my view this was never likely as Sir George Pearce, Australia’s longest serving Senator and longest serving Cabinet Minister, is a significant figure in Australia politics that did not deserve to be removed from the electoral map.

The certainty was that the new Pearce would look nothing like the old. It was way over quota, contained two of Perth’s population growth areas in the northern beaches and Ellenbrook districts, and combined them with a disparate collection of rural areas east and north of Perth.

So it came to pass that the new Pearce includes only one part of its former spread, covering only the outer northern beach of Perth, with Wanneroo and an area east of Lake Joondalup from Cowan. Taken together the boundary changes cut the margin of Liberal MP Christian Porter from 7.5% to 5.2%.


Hasluck loses Labor voting territory in the south to Burt but the major change is to gain the expanding Swan Valley suburbs around Ellenbrook from Pearce. This area makes up nearly a third of the population of the new Hasluck and Liberal MP Ken Wyatt will need to campaign in an area that has swung strongly against the Liberal Party at two state elections in a row. Major boundary changes but only a small change in the Liberal margin, up from 5.4% to 5.8%.


Cowan is made up of almost equal numbers from the old Cowan and the abolished seat of Stirling. But one week and 100 years after Edith Cowan became the first woman elected to an Australian Parliament, it was no contest whether to retain a seat named after her, or continue to commemorate the first Governor of Western Australia.

As the map shows, the boundary changes are extensive. The loss of Labor voting areas in the north to the new Pearce is balanced by the gain of Labor voting territory from the eastern end of Stirling, leaving Anne Aly’s margin in Cowan largely intact.


Loses Labor voting territory around Noranda to Cowan while gaining more mixed territory from the abolished seat of Stirling. On paper the margin for Labor MP Patrick Gorman declines from 4.9% to 3.2%.

27 thoughts on “2021 Federal Redistribution – Draft Boundaries for Western Australia”

    1. Had no idea or forgot that we are now in O’Connor, maybe that is why the member of Durack never answers my e mails,other than 1 flyer which arrived 5 days before the election I would have not known who is running , I feel quite insulated here,and I’m sure I’m not alone ,virtually zero coverage of any candidate ,in this town,

  1. I must admit that I am surprised that the Labor margin only increases in Cowan by 0.1%? I’d have thought uniting all of Balga-Mirrabooka and losing some of its better Liberal areas would make it much safer for Labor.

  2. The Liberals might be a big pissed about losing a seat – but in the light of last weekend, that’s the least of their worries. It’s pretty hard to see how anyone has much to complain about with this – though I’m sure there will be some loud and viscerial complaints.

    My only quibble is Durack reaching the outer edges of the Metro area – I’m not sure how the numbers could stack up, but a configuration more like the old Kalgoorlie/O’Connor split would seem to have better community of interest rather than the mishmash of leftover bits that Durack is now composed of.

  3. Tasmania retains its 5 seats and continues to be over represented. When will the other States get sick of this and remove the 5 seat minimum?

    COMMENTS: Not something the states can do anything about. It takes a constitutional referendum, and it would require a triple majority to cut Tasmania’s representation from five seats, a majority of the national vote, a majority in a majority of states, and a majority in Tasmania itself.

    1. Tom the first and best

      And since Tasmanian voters are extremely unlikely vote in a referendum to reduce the power of their vote in the House of Reps, the only way to achieve a one vote, one value outcome for the House of Reps is to expand Parliament so that the size of a seat is low enough Tasmania has the population for 5 seats. On current population numbers (as of the 2020 determination of House of Reps seat), with the Nexus Clause linking the Size of the House and Senate, that would require 18 Senators per original state.

      1. And if you adopt the Cube Root Rule, you can get anything up to a full doubling of Parliament.

        Full doubling: Take the C.R. for the population of the states [presently 292]. Then divide by 2 for the nexus and 6 for the number of states to get the number of senators per state [24, as rounded]. This is of course double what we have today; just follow the existing rules from there.

        Alternatively, you could argue that the Senate should be included in the C.R. total (though this is not what’s intended by the rule). In that case, take 1/3rd off 292 to get 195, then divide by 2 and then 6 to get 16 Senators per state.

      2. A better way to deal with that issue is to increase the size of the House so that Tasmania gets 5 seats on population anyway. Given population growth per MP over the decades since the last House size increase in 1984, it’s about time.

      3. If the senate was expanded to 9 seats per half election, then Tasmania’s 5 would be proportional in the resulting lower house expansion. There are other valid reasons to do this, or at least expand each half senate election to 7.

        COMMENT: I’m not sure the problem of Tasmania being over-represented is so serious that it warrants creating 100 extra Federal politicians to fix it.

        1. Of course resolving Tasmania’s lower house malapportionment would be well down the list of reasons to expand parliament. Odd numbers would be fairer for the upper house, lower house MPs have to represent far more people than previously, and it’s been 40 years since the last expansion.

      4. Pearce and Christian Porter… Will be interesting to see what happens. Any priors on how you see new voters (e.g. no loyalty to the incumbent) respond to controversial players?

        COMMENT: The electorate is largely formed from the old Pearce so there are not many new voters in the seat.

      5. Stirling has only an old margin… also misspelled…

        COMMENT: Yes, Stirling has an old margin. It is the margin for the seat at the 2019 election.

      6. I think the best part of this proposal is the contraction of Pearce to be a purely metropolitan based seat. It was a bit of a hybrid urban/rural abomination prior, containing communities with no real connection to each other. The metro seats largely seem to make good sense.

        Durack and O’Connor remain less than desirable as they split communities of interest, it would make much more sense if there was a wheatbelt seat and a mining and pastoral seat like O’Connor and Kalgoorlie of old, but I assume there are not enough electors outside the wheatbelt to draw coherent borders and still meet the population quotas for this to occur.

        1. The current arrangement was adopted because the old wheatbelt seat of O’Connor was turning into a dumbbell as Kalgoorlie grew ever larger.

          1. The current representation dating from 2008 makes far greater representational sense. It is much better to combine Wyndham with Bullsbrook, given communication links, than Wyndham with Kalgoorlie. The vast distances are rightly shared by two seats. The creation of Durack was in effect a revival of the seat of Dampier that existed 1913-22 and extended from the North to the northern fringes of Perth.

        2. Possibly was a copy/paste job from the Vic page, but not sure Lib/Nat grouping in the table is appropriate. Leaving aside political aspects (absence of a formal coalition in WA) it’s a 2PP metric, so a small chunk of Nats voters from three-cornered contests would be captured under the ALP side.

          COMMENT: Yes, that’s called two-party preferred and it measures the contest between Labor and the final Liberal/National/LNP/CLP candidate in each district. I used it in preference to listing them all as LIB.

        3. Good to see the names Cowan and Tangney are being retained, and that they’re adding in Sadie Canning for co-naming rights for Canning. The amount of men who lodged submissions suggesting Cowan be abolished was… a worry. Edith Cowan was the first woman elected to any parliament in Australia. It would be a slap in the face for women to have the seat named in her honour abolished. Ditto for Tangney, named after Dorothy Tangney, the first woman elected to the Senate.

          1. Agreed that these two names are valuable, not least because they were women serving in Parliament. However the Liberal Party didn’t propose the abolition of Cowan because it was named after a woman. Names are of infinitely less importance than boundaries and Cowan was one of several north-of-the-River seats to be well short of the 2025 projected enrolment quota.
            There have been instances of valued names being translated to other, unrelated seats. In 1968 in Victoria inner suburban seats named Scullin and Isaacs were abolished. The seat of Darebin was renamed Scullin and the name Isaacs given to a new seat in SE Melbourne.

            COMMENT: The new boundaries of northern Perth seats are unusually clear and well defined. I suspect they were drawn first and then the decision was made on names. By criteria such as area transferred or voters transferred, the seat now called Cowan could equally have been called Stirling. The electorate is exactly the same, the question is what name to attach to it.

            1. surely the Libs would want Cowan abolished as it’s a marginal seat with a popular member who looks to be fairly difficult to knock off. As for the other four Labor seats, abolishing them would be completely impractical given their locations (Burt, Freo, Perth, Brand)

              1. COMMENT: Exactly the same seats could have been drawn and the Commissioners could have chosen to name Cowan as Stirling, and then Stirling would have been a marginal Labor seat at the next election. The drawing of the boundaries which removed the part of Stirling west of the Mitchell Freeway created a seat that could have been named Cowan or Stirling.

              2. Unless I’m reading page 71 of the report incorrectly there were 4 comments and suggestions to abolish Cowan and only two of those were men, one was the Liberals (I’ve counted their comment and suggestion as a single one) and an anonymous one.

                Hardly a huge number.

                I really do think it’s a very long stretch to suggest that anyone is proposing to abolish an entire division because they don’t like the fact it’s named after a woman.

            2. Anthony, I get that the Maths involved in this are complicated. But somehow on these figures the abolition of a LIB +5.6 seat is leading to the net margins of all of the other seats in Western Australia increasing for Labor by 3.3%.

              Where did all those Liberal voters go?

              COMMENT: You are quoting a figure that has no meaning. The two party preferred’s on the old seats add up to the same value as the two-party preferreds for the new seats. There is no law of mathematics that requires the plus and minus changes in margin should add to zero which is what you are saying they should. The 55.6% two-party preferred vote in Stirling ends up re-distributed across Cowan, Curtin and Moore, but that is only one change across many seats. The changes won’t add to zero.

              1. I’m not sure I understand your response. Because based on my back of the envelope calculations the 2PP’s on the old seats explicitly do not add up to the same values as the 2PP’s for the new seats.

                I’ll put it another way. Going by the figures in that table. The average margin in a WA HoR seat at the last election was LIB + 5.6. (pretty much bang on the statewide 2PP. On these new figures, the average margin is LIB + 5.4.

                I get that’s an imperfect yardstick (because not all the WA electorates had the same number of people), but nevertheless, it’s a yardstick.

                I suspect the discrepancy between those figures (which is not large) is because the AEC rebalanced the numbers in each seat and/or rounding issues (because those 2PP results are rounded to 1 decimal place, which is going to cause a data loss).

                But, it’s also consistent with a small suburb somewhere being left out of the Excel spreadsheet.

                COMMENT: There is nothing that would make the 2PP percentages for the old seats add to the same value as the 2PP percentages for the new seats. Nor is there anything that make the change in percentage figures add to zero. You are adding up things that have no meaning when you add them together.

                Behind those percentages are a set of 2PP votes that are used to calculate the 2PP percentages. If you add the 2PP votes up for the old electorate, they add to the same value as the new electorates, and to the same value as the total 2PP for WA at the 2019 election.

                What you might be finding in your calculation is there is a difference between the swing needed for a majority of the state 2PP and the swing needed for a majority of seats. It is entirely normal that the difference between the two might vary when you change boundaries.

Leave a Reply