The recent NSW state election saw a high number of retirements in the lower house. The Coalition had more retirements than at any other election in the last thirty years.
Meanwhile the election produced inconsistent swings, with some seats producing large swings while neighbouring seats barely swung at all.
One theory that could explain some of this variety (although by no means all of it) is the impact of retiring MPs, and the subsequent loss of personal votes for a party in a particular seat.
I’m not going to try and quantify the relative value of one MP’s personal vote or another, but I’m going to apply a simplified model to assess where there is a change in the incumbency of a seat.
The theory is that personal votes are easiest to detect when the sitting MP changes. If an MP retires, their party no longer has their personal vote at the following election. A new MP then gains a new personal vote when they are facing their first re-election. There can be a double effect when a candidate wins a seat off a sitting MP from a different party. At the following election, the new MP has a new personal vote, and the former MP’s personal vote is lost.
I’ve calculated this trend for each of the 93 seats on the old 2015-2019 boundaries on a scale from -2 to 2. A score of zero means that the same MP was re-elected in 2019 and ran in 2023. A score of 1 means that the personal vote favours Labor – either the Coalition MP has retired in 2023, or the Labor MP retired in 2019 and the new Labor MP was up for re-election for the first time in 2023. A score of -1 favours the Coalition.
A score of 2 means that Labor won the seat off a sitting Coalition MP in 2019 (the only case being Coogee).
There were relatively few seat changes in 2019, and Coogee was the only classic race where a sitting MP was defeated, so it’s the only case that scores the maximum at either end of the scale.
This theory and scale is based on some ideas from Peter Brent, but I’ve done my own calculations and the model is not exactly the same.
I calculated these scores on the pre-redistribution electorates, and then I weight them to the redistribution changes. So, for example, the seat of Epping was significantly redrawn to take in parts of Parramatta. So while the sitting Epping MP (the former premier) ran for re-election, he did not have an existing personal vote in those areas. And what do you know, the swing was much larger in that area. Epping scores a 0.5, which is roughly halfway between being neutral and favouring Labor in terms of personal vote changes.
I applied this model to the 71 classic races which are Labor vs Coalition. This is both because they are the only seats which have a two-party-preferred calculated, but also because questions of who is favoured by a personal vote gets more complex when you bring in other parties.
Having said that, there’s reasons to think retirements significantly impacted on the non-classic races. The eight sitting crossbenchers elected in 2019 who were up for re-election all won re-election with increased margins, but the retirement of Greens MP Jamie Parker in Balmain led to a close race there. An independent also won Wakehurst and came close in Pittwater, both seats where longstanding Liberal ministers had retired.
So what did we find? This first chart shows the personal vote score (remembering that a +2 is favourable for Labor, and -2 is favourable for the Coalition) mapped against the 2PP swing.
The picture is messy, but there is a clear trend - Labor tends to get bigger swings in areas with higher personal vote scores.
Labor suffered negative swings in just two seats they hold, and both had a retiring local MP - Cabramatta and Liverpool.
This can partly explain some of the differences in swings. Labor gained very small swings in East Hills and Goulburn, both Liberal seats where the sitting MP had first been elected in 2019, and thus had a +1 bonus for the Coalition. Meanwhile they gained huge swings in Parramatta, Drummoyne, South Coast and even Castle Hill, where there was no sitting Liberal MP.
I also grouped these 71 seats into round numbers based on the personal vote score, and looked at the average 2PP swing to Labor. Bear in mind that 46/71 cases had a score of zero, with the rest split roughly evenly between +1 and -1, with only two seats scoring +2.
In seats with a -1 score (roughly meaning a retiring Labor MP or a new Coalition MP) there was barely any swing. Then the swing was higher for the groups with more of a favourable personal vote effect.
This is by far from the only factor in the variety of swings - the scatterplot does not show a tight explanation. But I do think it explains some things, and is useful context for the impact of an individual MP on an electorate.
UPDATE: I realised I made an error in how I calculated the by-election effects, which flips Monaro from -0.5 to +0.5, flips Upper Hunter from -0.58 to +0.27, and flips Strathfield from +0.34 to -0.55. I have corrected the numbers in this document to correct these errors. It slightly reduces the swing to Labor in -1 and 0 districts and slightly increases the swing in +1 districts.
interesting insight. so how did this work out for ‘safe’ liberal pittwater and wakehurst retirements?
They’re not part of the model but they both score a 1 if they were a classic race. And we do know they ended up producing big swings.
This is obvious.. miss out on upper Hunter and Goulburn but win south Coast and Monaro. The vote in Cootamundra and Bathurst exceeded the demographics of the electorates
What score did you give Willoughby, Ben? Was the swing measured against the 2019 result or the by-election. Much of the swing against Liberal (in 2023 and 2022 by election) can be put down to loss of Glady’s personal vote.
The Labor vote went up 5%, even though IND vote went up 17% 9bit against 2019). There’s going to be a large 2PP swing there when we see the final preferences counts between all candidates
The old Willoughby electorate was scored 0.5. Basically losing Gladys adds +1 to the anti-Liberal side, and adding Tim James subtracts 0.5. Once you factor in the redistribution it becomes +0.54.
I’d have to say the Seat of Oatley may be indicative of your mixed feelings on this, as the local member there had improved his margin – and just as well given the results- but also in the context of the overlapping Federal seats he had enough personal support to survive.
Clearly if he hadn’t stayed on it would have been another lost seat for the Libs.
Comments are closed.