Projected Enrolment Data released for the Victorian Federal Redistribution

This post was prepared last year based on an initial release of projected enrolment numbers. The data proved to have a major fault and a completely new data set was released at the end of January 2024. I have written an entirely new post based on the corrected data that you can read at this link. The old post with the wrong data has been retained for posterity.

The next step in the re-draw of Victoria’s federal electoral boundaries has begun today with a call for submissions and the release of base enrolment data.

Victoria is losing a seat at the next Federal election, the state’s representation reduced from 39 to 38 seats.

Victoria gained a 39th seat ahead of the 2022 election, but a decline in the state’s relative population compared to other states will see Victoria revert to 38 seats. One of the state’s 39 seats is to be abolished, but it is unlikely to be the new seat of Hawke first contested in 2022.

The two year immigration halt caused by Covid, combined with on-going internal migration by Victorians to other states, is why the state is losing a seat.

Of the state’s 39 seats, 14 are below the permitted variation from projected enrolment quota. Thirteen of these seats cover parts of Melbourne and all will need to gain voters. Seats in regional Victoria are mostly within the permitted variation and will be largely untouched by the redistribution.

(I’m happy to publish comments on how people think the redistribution will unfold. Read on into the post for my thoughts.)

Setting the Quotas

The abolition of a seat will cause an increase in average enrolment of around 3,000 electors per division.

Two quotas apply for the redistribution, a current enrolment quota set for 9 August 2023, and a projected enrolment quota for 17 April 2028.

On current enrolments, the 38 new districts must have enrolments within 10% of the current enrolment quota set at 116,894 electors. All new divisions must have enrolments between 105,204 and 128,583. All districts currently sit within the allowed variation, but one seat must still be abolished.

More important to the redistribution is the tighter projected quota rules, with all districts required to be within 3.5% of quota. The projected enrolment quota is 127,238. All new divisions must have a projected enrolment in the narrower band between 122,785 and 131,691.

There are no districts above the 3.5% projected upper enrolment bound but 15 below the lower bound. Of the 15, 12 are in the metropolitan area. The 13th is McEwen, which extends beyond Melbourne’s northern metropolitan boundary, while further west the regional seats of Ballarat and Bendigo are under quota.

With all seats within the allowed variation from current enrolment quota, where new boundaries will be drawn is going to be determined entirely by the projected quota. All re-drawn seats that meet the projected quota will also be within the permitted variation from current enrolment quota.

With regional Victoria needing minimal change, it is certain that a seat will be abolished in Melbourne. The question is whether a seat will go east of the Yarra, or to the west and north of the Yarra.

Whichever side of the Yarra a seat is abolished, it will have major flow on consequences. There is a strong possibility that the Yarra will have to be significantly crossed somewhere along its length.

Regional Victoria

The 2020/21 redistribution set many regional seats above quota based on projected slow growth. Some remain above quota in 2023 even with the increased quota created by abolishing a seat, though all are under the permitted variation.

The 10 seats that can be classed as regional and rural, running clockwise from Corio to Monash, contain 9.87 quotas worth of voters. It may be that the edges of Melbourne are nibbled at to re-balance the country, but only Ballarat, Bendigo and the peri-metropolitan seat of McEwen have to change.

The map below colour codes all 39 divisions based on percentage above or below the 38 seat projected quota. Dark brown seats are more than 3.5% under quota and must gain electors. Yellow divisions lie in the permitted 3.5% under quota range, while the light green seats are over quota but within the 3.5% permitted variation.

You can use this map to drill down and look at Melbourne seats, but a larger Melbourne-only map is included and discussed further down the page.

When Victorian electoral boundaries are drawn, the first seats drawn tend to be those in the state's corners, Gippsland, Indi, Mallee and Wannon. As the map shows, none of these seats require change though there may be some tweaks. All the seats in rural and regional Victoria can be drawn before larger changes are required closer to Melbourne.

Most attention will be given to the regional seats of Ballarat and Bendigo, and McEwen on Melbourne's edge. There are enough numbers in surrounding seats to fix Bendigo, and a number of options to increase enrolment in Ballarat and McEwen by absorbing areas from the neighbouring metropolitan seat of Hawke. Taking voters from Hawke will increase the chances of a seat being abolished in Western Melbourne.

Monash to the east of Melbourne and covering the La Trobe Valley and west Gippsland may also see some boundary adjustments, perhaps absorbing semi-rural areas lost to La Trobe in the last redistribution.

Abolish a Melbourne seat - east or west of the Yarra?

The map below makes clear that the new quota leaves most Melbourne seats under quota. In total 12 seats must gain voters to reach the lower bound 3.5% variation from quota. Only three seats are above quota and do not require change.

The redistribution is going to be a cross between a jigsaw puzzle and playing a giant game of pass the parcel. From wherever the Redistribution Commissioners start drawing boundaries, every seat will need to increase its enrolment, cannabalising neighbouring seats until at some point an entire seat will disappear.

There are 16 seats east and south of the Yarra that are in total half a quota short of retaining 16 seats. North and west of the Yarra there are 13 seats and in total they are 0.4 quotas short of retaining 13 seats.

Wherever a seat is abolished, it is likely a seat will need to cross the Yarra to balance the population trends. But where to do it?

Will Macnamara be split with Southbank and Port Melbourne added to a seat across the river? Will the Yarra be crossed in it upper reaches, currently the boundary between Jagajaga and Menzies?

The map below again highlights below permitted quota variation seats in dark brown. The geographic size of Hawke and McEwen makes it appear that a seat must go west of the Yarra. The numbers also point to a western Melbourne seat going, with six of 13 seats in Melbourne's west and north below the allowed 3.5% variation compared to only five of 16 to the east.

Port Phillip Bay and the lower course of the Yarra tends to be the starting points for drawing boundaries. East of the Yarra, this sees Macnamara, Higgins, Kooyong and Goldstein drawn first. Low current enrolment and slow enrolment growth will cause these seats to grow eastward into neighbouring below quota seats.

There would be a similar pattern west of the Yarra, starting with Gellibrand and Melbourne, then working north and west.

But with enrolments so evenly balanced on either side of the Yarra, just drawing boundaries outwards from inner-Melbourne will be harder at this redistribution. All the major shortfalls are in outer Melbourne. The Redistribution Commissioners are going to need a strategy for drawing whole regions rather than simply cannibalising seats while moving outwards towards Melbourne's edge.

Working out that strategy makes submissions to the Redistribution Commission more important than normal.

If a seat is to be abolished west of the Yarra, then the seats to watch are Wills and Maribyrnong. Hawke may be under quota, but a newly created electorate named after Australia's third longest serving Prime Minister is not going to be abolished, though its boundaries could undergo radical surgery. Burke was abolished two decades ago so why not Wills as well? Maribyrnong as one of the few remaining Melbourne seats with a geographic name could also attract attention.

Other seats may undergo major changes but retain an existing name. Fraser, Scullin and Gorton are named after former Prime Minister while Cooper and Jagajaga are named after significant Indigenous Australians.

Gellibrand might be up for a name review, but being based on Williamstown gives the seat a geographic anchor that makes it unlikely to be abolished. Seats that could be abolished include Calwell and McEwen, neither having a clear geographic anchor and both named after lesser known politicians. (OK, no more messages. Jack McEwen briefly served as Prime Minister.)

If a seat is abolished west and north of the Yarra, a seat to the east of the Yarra will have to cross the river to take in eastern parts of McEwen and Jagajaga. There is a high likelihood the redistribution will produce a new upper Yarra seat on both sides of the river. Menzies and Casey are the seats most likely to slip into the void created by the abolition of a western Melbourne seat.

If a seat is abolished east rather than west of the Yarra, there is a strong chance the same upper Yarra seat could be drawn but end up being called McEwen. The abolition of an eastern Melbourne seat would most likely see either Chisholm or Hotham abolished.

Assuming seats like Macnamara, Higgins, Goldstein and Kooyong grow to reach the new quota, changes to seats further east will be amplified. The need for seats in inner Melbourne and outer south-east Melbourne to grow will push outer eastern Melbourne seats north. Unless an eastern Melbourne seat is abolished, the likelihood is that outer eastern and south-eastern Melbourne seats will shift north, and one electorate will end up straddling the Yarra in Melbourne's outer north-east.

In the end the redistribution looks set to make a choice between abolishing a western seat and rotating electorates clockwise, or abolishing a seat in the east and rotating seats anti-clockwise.

16 thoughts on “Projected Enrolment Data released for the Victorian Federal Redistribution”

    1. Black Jack McEwen was caretaker PM for three weeks after Holt disappeared in the ocean. I’m not sure that’s grounds for keeping the electorate…

      1. We have electorates named after every former prime minister who has died, including Forde and Page whose tenures were shorter than McEwen’s. The AEC may alter the electorate significantly, but I think the name will be retained.

    2. A very good summary of the choices the redistribution commissioners will face. I would just add two points.

      1. The decision on where to cross the Yarra imo is not an either/or choice. In terms of community of interest, there are good reasons why Southbank (and maybe adjacent areas) should be moved into Melbourne. And there is no good reason why the Yarra should be treated as a barrier in drawing the boundaries for Menzies. It would make the job easier if the redistribution tackled both.

      2. For a generation or more, most of Melbourne’s population growth has been to the north-west of the Yarra. I think the odds are that this will continue; it seems developers and governments still find it easier to get higher- & medium-density developments adopted there than by taking on the more NIMBY residents of the south-east. If so, it would be a step backwards to abolish a seat in the north-west, where there are no obvious candidates. The odds are that, if a seat in the west goes this time, the next redistribution would have to restore it.

      There seem to me a number of seats in the south-east that would be better candidates. Chisholm’s suburbs have grown very little for decades. Higgins had the slowest growth in recent years. And in the middle outer suburbs, one of Aston, Bruce, Hotham, even Isaacs (or Chisholm) could be given up and the voters rearranged piece by piece into larger but reasonably coherent seats.

      But to close, imo your two key points are very true. This redistribution must see seats crossing the Yarra – and groups of them moving clockwise or anti-clockwise to make the numbers fit.

      1. ps The Age has just reported new official state govt estimates of population growth 2021-2036. Nine of the ten fastest-growing councils predicted to be north-west of the Yarra. Eight of the ten slowest-growing predicted to be south-east.

    3. I think if a seat in the east were to be abolished, Higgins is a logical choice, although I don’t know if the fact that it produced 2 Prime Ministers would be a deterrent.

      The reasons for this are not only that it’s well below quote and has had the slowest population growth, and a VERY strong NIMBY element in most of the seat (almost everything east of Williams Road) as well as surrounding seats like Kooyong & Goldstein ensures this is likely to continue, but that of all seats it probably has the least cohesive “community of interest”.

      The western quarter of the seat is extremely inner-city, dominated by apartments, renters and a young demographic. The entire middle half of the seat is extremely affluent, “old money” territory. The southeastern quarter of the seat is very much middle class suburbia.

      What makes this even more appropriate to abolish, is that each of these 3 distinctly different sections of the seat actually borders another seat that those areas are better suited to, and in most cases need the extra population:

      – The “heart” of Higgins, its middle half, borders Kooyong to the north and Goldstein to the south (albeit with a slice of Macnamara in between, but more on that below). Both areas are under quota, both have strong NIMBY elements and slow population growth, and both are similarly affluent areas with “old money” suburbs and similar communities of interest.

      – The eastern part of Higgins, more suburban, is well suited to seats like Chisholm (under quota) and Hotham. For example, Murrumbeena is in Higgins but Hughesdale is in Hotham, having grown up there, they are essentially the same suburb and the new Hughesdale Station is even in Murrumbeena. Meanwhile areas like Ashburton fit well into Chisholm.

      – Most importantly, the western quarter of Higgins (the Chapel St corridor) has always been far more suited to Macnamara. It’s a large chunk of population, but that allows Macnamara to either send Southbank north to Melbourne, or the Caulfield area south to Goldstein.

      It also may not get very much opposition from the major parties either. Labor only just won Higgins for the first time but know they probably won’t keep it. The Liberals have already lost it and sending Liberal heartland suburbs like Toorak & Malvern into seats they recently lost to teal independents would probably appeal to them.

      So if the AEC can overcome Higgins’ history of producing Prime Ministers, abolishing it and pushing its suburbs outwards into Kooyong, Chisholm, Macnamara, Hotham and Goldstein would probably be a great place to start, and in particular moving a huge chunk of population into Macnamara provides a logical place to breach the Yarra by forcing Macnamara to send places like South Wharf and Southbank into Melbourne.

      Then, the north-western seats can expand inwards towards the city, it makes less sense to abolish a seat where most of the population growth occurs.

    4. Antony, what are your thoughts on the quality of the SA1 projections used for this redistribution?

      It appears that almost every SA1 is projected to grow at a rate of roughly 9.5% which doesn’t match common intuition or recent trends. It’s looking likely that a statistical error has slipped through or some really awful assumptions have been made. In short, this would bias the drawing of seats towards the slow-growing Eastern Suburbs over the high-growth areas on the suburban fringe. That’s not a good look for the AEC if so.

      If the numbers provided are proven to be defective, is there a mechanism to restart the redistribution attempt so that appropriate ones can be used?

      COMMENT: I can’t comment on how the calculations are done.

      1. It looks like you are correct, and the AEC have acknowledged that they received ‘bad’ data from the ABS. As a result, the AEC has published revised projected enrolment figures that are more realistic showing outer suburban areas growing faster than the inner suburban and rural districts.

    5. Hey Antony,
      I was just having a look back through the Victorian State Election coverage where, as you know there was quite a lot of discussions about pre-polls’ influence. Do you, in your matched polling place results, take the increase of vote-share of pre-poll into account?

      P.S. I know this is a bit of a misplaced comment I just thought you might not see it if I placed it on one of the Victorian election posts.

      COMMENT: You try to but in Victoria you do not know how many pre-polls you will receive on the night and you don’t know where they are from. There was discussion of it in the 2022 coverage because there was a significant narrowing of the results on pre-polls in 2018. It ended up not happening in 2022.

      1. Fair enough, but I think someone (I don’t quite remember who, maybe David Davis) said, there was a verifiable drop in the number of votes, maybe you could project that fall in polling place votes as well, take into account the postals and absents etc. you get figures on in VIC and get then give out the pre-poll votes as they were at the last election. And then take the new numbers into account when using matched polling places.

        Have you tried this before? If not I could cook you up a testing set and see if it works (when making the same assumptions federally I don’t have the Victorian figures).

        COMMENT: Postals are not the problem. They are a relatively stable proportion of the electorate, have a known political lean, and in most jurisdictions they are not counted on election night. So you leave your formulas with enough wriggle room to cope with switching to post election analysis.

        The problem in Victoria isn’t so much the number of votes. The problem is whether the pre-polls behave differently to the previous election in their swing. This is a particular complicated in Victoria because you cannot identify the source for pre-poll votes. Victorian pre-poll votes are the only vote type in any state that is transmitted as a progressively updated figure rather than as discrete counting centre totals. Because these figures come late and you don’t know how they will behave and you don’t know where they are from, you have to leave yourself wriggle room in your predictions in case there is a shift.

        In both Victoria and NSW we had problems caused by the proliferation of pre-poll centres. This was a problem in Victoria because we were not supplied with individual pre-poll returns. In NSW a strange decision on how to count pre-polls meant we only received one of several pre-poll centres in each electorate, and some of those in key seats were new pre-poll centres where you needed to work out history to match against. And as it turned out, the negligible difference between pre-poll and polling day votes in 2019 was replaced by a significant difference between the two in 2023, aggravated by new pre-polls and a redistribution.

        The biggest issue in election prediction is being able to identify where votes are from, and having a reliable historical figure to match against it. The AEC supplies both history and results in its feed, but for state elections you have to compile your own history.

        In both NSW and Victoria I am negotiating with the Commissions to try and improve reporting for future elections. With three years before the next state elections, they should be able to implement the required changes.

        1. Thanks for the comment Antony. You’re right, unsurprisingly, there is a major problem with how pre-polls are reported, not knowing where figures are from is annoying, but maybe this is not the whole problem. Because it seems to me that you only have one main form of vote analysis, matched polling places, have you thought of any other way? Because the electoral commissions are not just going to stop releasing new polling places first and trying to Frankenstein a new historical figure is not always going to be pretty.

          COMMENT: New polling places are not released first. New polling places are created in areas of population growth and arrive later. New joint polling places create problems, especially after redistributions. The issue that is developing is that all polling places report before you receive any pre-polls. How much allowance do you build into your model for possible difference in swing for pre-polls. In the case of Victoria, if we can get pre-poll centres to report individually, as occurs at Federal, NSW and Queensland elections, then you have something to work with as re-polls progressively arrive. This is an ongoing discussion I’m having with electoral commissions and parliamentary committees in NSW, Victoria, WA and SA. It requires changes in electoral commission computer systems so takes time to implement.

      2. Thanks Antony! Just to echo what Arthur said above: those Victorian projections are garbage. They’ve just applied a 9.5% growth rate across the board, with very few exceptions. The result is that most divisions are projected to move in the opposite direction relative to the mean than they did over the last three years. This is a major problem: they’re going to have to withdraw those and re-do them (and I don’t know the legality of how you can do that), which I presume will push the timetable out.
        (It might not be just a Victorian problem – on a quick look the WA ones look dodgy as well.)

      3. Rivers divide communities far less than freeways do. I’ll be making a submission urging consideration of different geographies for the drawing of borders in the west. And a rejection of the pizza cutting of Melbourne’s north and west electorates that is caused by creating electorates hemmed by watercourses. Perhaps worth pondering, Traditonal Owners of what is now Melbourne reportedly did not see their country borders defined by the rivers currently used to create electoral boundaries. Rather it has been accepted that the first people of what is now greater Melbourne were distinguished as river people or salt water people.

        COMMENT: That many local government boundaries also follow the rivers is a significant factor that the Commission takes into account in assessing community of interest. Attempts to run Wills and Cooper east-west rather than their current north-south alignment have usually foundered on such an alignment splitting two local government areas that don’t have to be split by the current north-south alignment. That doesn’t mean the current redistribution won’t divide the councils differently, but it does mean a good argument for change has to be put.

      4. I’ve taken the AEC’s Victorian population projections, added them up and divided by 38. I get 126,601. You get 127,238. What have I misunderstood?

        COMMENT: I don’t know. My calculation and the one by the AEC produces 127,238.

Leave a Reply