New South Wales legislation sets a schedule for redistributions of Legislative Assembly electorates to be held once every two terms. We have had redistributions before the 1991, 1999, 2007, 2015 and 2023 elections. Yet there is a clause in the constitution that allows for one to be triggered sooner, if enough electorates deviate far enough away from the average for two straight months.
And the current boundaries, which were finalised in 2021, are already showing signs of significant deviations that put lots of seats on track to fall out of the permitted range.
Redistributions must meet two population metrics when drawing electorates. Seats must fall within 10% of the average enrolment based on a current-day enrolment statistic – in the case of the 2020 redistribution they used enrolment data as of March 2020. Seats must also fall within 10% of the average based on a projected enrolment three years into the future (April 2023). The rule had previously required no more than 3% deviation on the projection, but it was loosened by the O’Farrell government prior to the redistribution before the 2015 election.
We’ve now almost reached that forward projection point, but this story is not about those projections being inaccurate, unlike in the case of the federal redistribution. They did understate the growth in Riverstone and Londonderry, but both these seats were already drawn at the absolute minimum size.
There is one further rule to restrict malapportionment. Section 28A of the constitution requires that if one quarter of all seats have an enrolment that deviates from the average by more than 5% for two straight months, and no election is scheduled for at least one year, then a redistribution is triggered. And by my analysis, current population trends put us on track for 30% of all seats to fall outside this 5% rule by March 2026, in time to trigger another redistribution prior to the next election.
Of course, this may not come to pass – population trends could change. But at the moment the projections are not even close.
We have enrolment data from March 2020 (the data used to draw the boundaries) and now have monthly enrolment data from every month from November 2022 to March 2023. I have performed a simple projection for each seat – taking the growth from March 2020 to 2023 and extrapolating out to March 2026. I should note that I also tried a projection just from November 2022 to March 2023 and that also produced similar results to March 2026.
That pink zone needs to contain at least three quarters of seats (70 seats), but my projection leaves only 65 seats in that zone.
I've highlighted the two most extreme cases in red: Barwon is rapidly falling behind, while Riverstone is rapidly shooting ahead.
So what is going on here? The answer appears to be a combination of fast growth in a small part of the state leading inevitably to seats exceeding the 5% rule despite best efforts, but also a bunch of other seats where the Commission was modest in making changes. A more dramatic redistribution that drew slow-growing seats as larger and fast-growing seats as smaller may have given us a better chance of avoiding a premature redistribution.
Next up, I've produced a list of the 28 seats I project as falling outside the 5% range.
It's interesting to look at the seats on this list.
For some of them, it's hard to see what could have been done. Riverstone, Londonderry and Leppington are all seats on the outer fringe of Sydney, and are subject to enormous amounts of housing development. They were all drawn as close to the minimum enrolment as of March 2020, and have already exceeded the average. They are projected to exceed the seat quota by 54% of a seat by March 2026. A number of other seats projected for massive growth were also drawn under quota: Camden, Hawkesbury, Kellyville, Macquarie Fields and Badgerys Creek are all on the outer fringe of Western Sydney.
But there's a number of other over-quota seats that were already drawn over-quota, and aren't actually projected to grow much faster than the rest of the state: Ryde, Granville, Upper Hunter, Sydney and Oxley.
This failure is even more obvious looking at the seats projected to fall 5% below the average.
The most extreme example is Barwon, which was drawn 5.4% under quota, and was expected to be 7.5% under quota by April 2023, but in reality has already hit 9.5% under by March 2023. It's ridiculous that a seat expected to fall behind the state population growth should've been drawn on the smaller end. It's worth noting that it borders Upper Hunter, which was drawn over quota.
Less dramatic examples include Cootamundra, North Shore and Wyong.
I wanted to examine the original redistribution decisions based on the data that existed at the time. 58 seats were expected to have slower growth than the state (and thus fall backwards relative to the quota), with 35 seats having faster growth. This reflects how the fastest growth areas are in a small number of outer suburban seats, while the slower growth is spread over a larger area - both rural and inner city seats.
The Commission drew nine seats over quota despite having faster growth, 14 under quota despite having slower growth. Barwon is the only seat drawn more than 5% away from the quota in the direction it was expected to shift.
That's 23 seats. It doesn't seem like that many, but it could easily make a difference when 28 seats are expected to fall outside the permitted range, and the permissible limit is 23 seats.
The Commission probably should have considered being more willing to draw seats closer to the 10% limit if they were expected to shift in the other direction. The Commission drew just twelve seats in the 5-10% range (six above the quota, six below). These seats tend to be successes. Winston Hills, Cabramatta, Mount Druitt, Castle Hill, Sydney and Hornsby were drawn substantially over quota. Five of these six seats are on track to be close to the quota in 2026 - Sydney has kept up with population growth.
Kellyville, Parramatta, Riverstone, Heffron, Londonderry and Leppington were all drawn well under quota. Three of these seats are already above average and expected to go well above the average, despite the best efforts to draw them small. Kellyville and Parramatta have also drawn much closer to the average, with Heffron the sole exception again.
I've produced a map showing which seats deviate from the average by more than 5% as of 2020, 2023 and 2026.
So it seems like there's a few stories here. In a few cases, population growth has not progressed as projected, but that isn't the main story. There are also seats like Londonderry, Leppington and Riverstone that were always going to shoot right past the average and become larger seats. This is a reflection that so much population growth is concentrated in a relatively small part of the state, and eight years is a long time for a set of boundaries to apply. But in addition, there were opportunities for the Commission to draw more seats projected to grow at the smallest end, and to draw more seats projected to fall behind at the largest end. Instead they chose to be more conservative in preserving boundaries. And that could potentially lead to the boundaries they drew reaching their use-by date four years ahead of schedule.
When you combine the projected excess quotas from Riverstone, Londonderry, Leppington, Camden, Hawkesbury, Kellyville, Macquarie Fields and Badgerys Creek you get to nearly 0.95 of a quota. This would likely justify the creation of a new seat in Western Sydney. I presume that somewhere in Northern NSW would have to lose a seat in exchange.
The tricky question here is where to draw that new seat. Without looking too closely at the numbers, I’m inclined to spilt Londonderry and Riverstone into three seats – a St Mary’s based seat, a Marsden Park based seat and a Quakers Hill based seat.
The St Mary’s seat would push south of the M4, and also take in territory from Penrith, pushing it south of the M4. This would allow Badgery’s Creek, Camden, Leppington and Macquarie Fields all to be drawn smaller and further south.
The Quakers Hill based seat would also allow for a southward push, forcing Blacktown or Winston Hills into Prospect and then into Badgery’s Creek.
The Marsden Park seat would likely be drawn under quota due to the further future growth in the North West corridor. It could also take some territory from Hawkesbery, reducing the overflow there.
It seems like they need to spilt up the growth areas. Box Hill and Gables in different seats. Riverstone and Schofields etc.
In my contributions to the last redistribution, I railed against the requirement for seats to be within 10% of a quota on current enrolment. Of course, I didn’t expect the redistribution panel to do anything about it as they aren’t entitled to – but I felt it was a point that had to be made. In attempting to satisfy these dual quotas, many suggestions made a point of splitting growth areas, leading to worse boundaries from a community of interest perspective. A prime example is Leppington district containing the Raby area instead of Edmondson Park and Bardia.
The figures provided in this article for Leppington, Londonderry, and Riverstone illustrate why this requirement is so silly. These seats were drawn to be as small as they could be, and yet at the time of the election, these seats are already over quota!
If it were up to me, the projection date would be the midpoint of the dates of the next two elections (e.g. March 2025 for the previous election), and there would be no numerical requirement in relation to current enrolment.
* (for the previous redistribution)
I think it’s absurd to ditch all numerical requirements in relation to current enrolment. Population change may well eventuate, but any structural malapportionment is here and now.
Excellent analysis, Ben. I think this bit is really key:
> [S]o much population growth is concentrated in a relatively small part of the state, and eight years is a long time for a set of boundaries to apply.
The solution here is obvious: New South Wales ought to have boundary redistributions every term, as is done in SA and WA.
Agree Alex J – the projections are only a rough guide, and it is harder to forecast growth the further you make the projections. Ideally redistributions should occur every election cycle, that way projections only need to last until the next election. You would be able to satisfy both numerical criteria and community of interest without having districts deviate too much from quota.
An alternative is to use a larger variation for the current enrolment (say 20%), that way it makes it easier to meet the 10% projected variation for a long-term date.
Elections are not held at the time a redistribution is held. The current enrolments are therefore hardly meaningful, and any significance they may have in relation to “structural malapportionment” is reflected (and indeed better reflected) in projected enrolments. Relying on projected enrolments is less likely to produce malapportionment throughout the cycle than relying on current enrolments. Just look at what has happened in Leppington, Londonderry, and Riverstone. And if we had relied solely on current enrolment, the malapportionment would have been even greater!
We all believe that electorates should have roughly the number of electors because we believe in “one vote, one value”. Votes are cast at an election. So that is the point in time at which electorates should have roughly the same number of electors. Projections might not be perfect, but there are far better than using on current enrolment to this end.
* but they are far better than using current enrolment to this end.
Seats like Barwon are kind of explainable in that they are already geographically enormous. There might be a sort of deliberate trade-off, keeping these seats at lower enrolment as a recognition of the difficulty caused by their size.
It might be time to bite the bullet and introduce a Large District Allowance or lower enrolment for seats above a certain size, like they do in QLD and WA.
Agree Mark, that also is a good idea that would work for almost all states. SA with its large outback divisions could also do with an LDA factor to better aid with community of interest.
I hate to sound like a contrarian, but I respectfully disagree with that idea too on the basis it violates “one vote, one value”. If there are districts that are “too large” by any measurement considered legitimate, the number of seats should be increased.
Perhaps I’m drawing a long tangent here, but I think it also begs the question of why we have states that unite such diametrically opposite parts of the country. The necessity of such large rural electoral districts is a symptom and a consequence of this.
I do agree on your last point Nicholas, it would be best to split the current states into smaller ‘region’ like structures similar to the size of US states or even countries like Italy.
Not only would that benefit in terms of creating manageable electoral districts, but things like Daylight Savings could also be agreed upon easier for these region-like states.
I strongly believe that NSW should go to 100 seats next redistribution. Would help ensure seats are not too large to be represented (e.g Barwon) and provide new seats for population growth.
I think we only have 93 because a previous government chose that number as they believed it was their best chance of victory.
The next NSW government should seriously look at expanding the size of the lower house to at least a minimum of 99 but preferably higher. It is ridiculous that there are fewer MPs now (93) than there were in 1950 (94). People will squeal about more politicians – but the politicians squeal that they need more staff to service more voters. At least MPs are elected – support staff are definitely need – but staffers also are a way for the poltical parties to milk the public teat. Take away one staff member from every MP and that cost pays for a certain number of new MPs and staff – cost neutral and better for democracy.
Increasing the number of seats from 93 to 99 reduces the quota by around 6% ((1/99)/(1/93) = 93/99 ≈ 0.94), so even in that case, Barwon would probably need to expand at the following redistribution! You’d need close to 110 seats just for Barwon to not change.
Ben,
interesting that Mid and north coast seats don’t figure here. I would have thought with increasing seachangers they would have grown more.
Also knock on effect of U Hunter, Cessnock and Maitland growth and under numbers in Wyong, Terrigal will see those seats (W& T) shift north with knock on effects. end result will be U hunter losing rural areas in its west and north and gaining areas in the east and south thereby becoming more urbanised and more marginal, particularly as the Huntlee development continues
I think the most logical move north-side of Sydney is Wakehurst, which will take the northern beaches part of Davidson. There is developments expected to occur along mona vale road in Pittwater so the only other seat movement would be between Manly and Wakehurst at the south, potentially taking sections of Manly Vale.
As for Davidson, this will be forced west. I’d imagine we could see the return of the seat of Gordon. This will force Wahroonga slightly south into Ryde and eat up some of the excess quota.
The last seat to get abolished was Lakemba. Will the next abolished seat be in the regions?
If so, the obvious choice for an abolished seat would be in the west or north. If the northern rivers and northern tablelands seats expand, they would all expand westwards and/or southwards. Barwon would take up large-ish towns in Murray or its eastern neighbours, whilst losing parts of the east and north-east. I don’t wanna see Barwon keep on expanding geographically.
Most logical choice for a new seat should be in north-western Sydney (between SW Sydney and the Hunter). SW Sydney only just got a new seat.
Went into a bit more depth with this and he is my very uneducated view on this:
Northern Beaches – Wakehurst takes the rest of Belrose and Davidson from Davidson, loses Collaroy to Pittwater.
North Shore – Davidson shifts west, gets renamed Gordon. Lane Cove shifts west to Lane Cove Road. Ryde gets renamed to Eastwood
Central Sydney – Newtown takes from Sydney up to Foveaux Street and South Dowling Street. Balmain extends to take Pyrmont and the rest of Ultimo from Sydney. Summer Hill takes Lewisham from Newtown and extends to Foster Street from Balmain
North West Sydney (between Hawkesbury and the M4) – There will need to be a new seat that takes segments of Hawkesbury, Londonderry and Riverstone. Suggestion is that this new seat be centered around Marsden Park. This would push Hawkesbury North, Londondery south and Riverstone East.
South West Sydney – You will see Seats expand west. Leppington, Camden, Macquarie Field and Badgerys Creek all gets smaller, which will pull the rest of South West Sydney West. The seats this could impact the most would be Holsworthy (loses Barden Ridge and Lucas Heights, gains past Prestons out to Cowpasture Road. The flow-on effect is that it could drag Heathcote back north and flip the previous redistribution. The other problem in the area is Oatley/East Hills, as you have very hard geographical boundaries, similar to the situation where Cook was forced to jump the Georges River.
The seat that will have to go will clearly be from North or Western NSW, to save Barwon. The problem you have is that the seats around Barwon (Murray, Cootamundra, Orange, Dubbo and Northern Tablelands) are all projected to decline. I’m going to go out on a limb and suggest that the seat that could go is Upper Hunter, which would pull a lot of these seats east and pull Cessnock and Hawkesbury North, while also giving some extra space for Wyong.
Comments are closed.