Modelling The Seat Of Pascoe Vale

PASCOE VALE (LABOR VS LIB 22.7, EST LABOR VS GREEN 12.5 based on 2018 preferences)

Why is this even here? Vacancy, change in Liberal preferencing

Does this market know something that I don't?

One of the interesting ALP-Green seat contests in this year's Victorian election is Pascoe Vale, which seems to be between experienced political advisor Anthony Cianflone (Labor) and Merri-bek Councillor, campaigning director and volunteer Angelica Panopoulos (Green).   This seat has seen a lot of debate with some people saying the Greens are seriously in the mix and others saying Labor will win it easily.  I've been looking at this seat a lot and I kept taking into account all the fancy stuff then making basic errors, so I thought I'd try to do it justice, post a full projection and see where that ended up.  In 2018 this seat saw a contest between Labor and independent Oscar Yildiz, which ended up not terribly close with Labor winning 58.58-41.42.  On a two-party basis it was uncompetitive with Labor winning 68.32-31.68, while the Greens finished third on primaries on a mere 12.94%, not far above the Liberals.  With Yildiz having gone to the Victorians Party (which was then a non-starter) and the 2PP margin now above 20% this sounds utterly boring.  However, the Liberals' decision to preference the Greens, plus a very Greens-friendly redistribution, appears to make it interesting.

The first point is the impact of the redistribution.  The redistribution may seem from the 2PP to have been Labor's friend, but no.  It kicked 18829 true believers in the north of the district into Broadmeadows while importing 14731 communists from the People's Republic of Brunswick (which was won by the Greens).  This had a massive impact on the primary vote baseline.  The Tally Room has the following estimates for Pascoe Vale, with the 2018 results shown in brackets: Labor 38.2 (37.7), Greens 20.7 (12.9), Liberal 11.5 (11.4), IND Yildiz 16.3 (23.5), Animal Justice 2.1 (2.0),  Victorian Socialists 2.4 (3.0), other Pascoe Vale independents (mostly John Kavanagh) 5.3 (9.5) and others who only contested the Brunswick part 3.6 (0).  

Unless one assumes that most of the Yildiz votes were at the expense of Liberals, the Greens start in second on primaries and the seat is a Labor vs Green contest, especially in the absence of major independents.  (In the sort of thing that could only happen in northern Melbourne the sole independent is a quasi-endorsed Socialist Alliance councillor running although the Victorian Socialists are also running. The Socialist Alliance are not registered.)  

The first challenge here is to allocate Yildiz' remaining 16.4%.  Did his votes come from Labor, Liberal or Greens?  My attempt to get a handle on this has been to look at how 2018 edition Pascoe Vale voted for the upper house and then compare that to adjacent and broadly similar districts, and see what happened in them in the lower house.  

There are five districts surrounding Pascoe Vale, of which Northcote and Brunswick are too green to be useful upper/lower house models and Broadmeadows is not green enough.  However Essendon and Preston are reasonable comparisons as these had only lowish-profile independents and have similar levels of Green support, so the upper/lower house comparisons for them should give some insight into what Pascoe Vale might have looked like had Yildiz not contested and assuming a generic Labor incumbent.  

In Essendon, the 2018 Labor incumbent was first-term incumbent Danny Pearson, and Labor polled 6.2% higher in the Lower House than the Legislative Council, Liberals 1.9% higher, Greens 2.1% higher.  In Preston, Labor's 2018 (since somewhat tarnished) champion was third-term incumbent Robin Scott, and the differences were 7.0%, 1.4% and 0.5%.  These are pretty consistent across the two divisions.  In Pascoe Vale the Upper House primaries were Labor 41.1, Liberal 17.8, Greens 14.7.   This suggests that if Yildiz hadn't stood and ignoring candidate effects, the baseline could have been something like Labor 47.7 Liberal 19.5 Greens 16.4.  Compared with the actual primaries of 37.7, 11.4 and 12.9, the suggestion is that nearly all Yildiz' votes came off the major parties.  The reservation this is open to is that Labor's (then) first-term incumbent Lizzie Blandthorn underperformed on 2PP swing compared to Pearson and Scott, so with an MP as popular as them the Pascoe Vale Labor primary could have been somewhat higher, the Liberal and Green primaries somewhat lower, so more of Yildiz' vote could have come off the Greens and Liberals.  But this only goes so far, and I'll adjust for it shortly.

Happily, Yildiz' vote in the part of Pascoe Vale that remained in Pascoe Vale was more or less exactly his vote for the division as a whole (the Greens did a few points better in the remaining part while other candidates were generally the same or slightly weaker).  If the above follows through then reallocating Yildiz' 16.4% for the new Pascoe Vale based on the above model (46.3-16.2-37.5) converts the 2018 results in Pascoe Vale 2022 to Labor 45.8, Greens 23.4, Liberal 17.7 leaving 13.1% for the others.  

From what can be seen of the preferences of the others who stood in 2018, those that were distributed between Labor and the Greens split slightly in the Greens' favour, giving the Greens about a 7.2%-5.9% legup if those splits were representative (they may not have been as many votes split to other minor parties or the Liberals or Yildiz).  Add that in and it puts Labor on a 51.7-30.6 lead before considering the Liberals.  Ignoring shifts in Liberal preferencing this would be something like 62.5 to ALP.  

The Liberals are preferencing the Greens at this election, including on their lodged Pascoe Vale how to vote card which has Labor last.  It's a long time since they've done that in Victoria.  In 2006 in the state seat of Melbourne their preferences split 74.4% to the Greens, but that included votes from other candidates.  In 2010 in Melbourne (federal) the split was 80%; after that the Liberals stopped doing it.  In South Brisbane (Queensland) in 2020 the LNP preferenced against Jackie Trad and the split including votes from other candidates was 63.8% to Greens.  Perhaps with a general movement away from how to vote card usage the flow will be weaker, and also preferences tend to flow more strongly where a party does well on its primary vote, so 80% seems like too high.  If I assume that it is 75% then the underlying 2CP (Labor vs Green) comes out at about 56.1-43.9 to ALP.

But that's based on an assumption that Blandthorn was as successful a vote-getting incumbent as Pearson or Scott, which she probably wasn't.  In any case even if she had been such an incumbent, she isn't the candidate this time so the baseline for Pascoe Vale (given the redistribution, the Liberal preferences and the vacancy) looks like something like 54.5-45.5 assuming the Liberal flow to Greens is reasonably strong (or if it is weaker then the flow from others could be stronger).  Throw in the state polling expectation of a swing against the ALP and to the Greens and it seems on paper like Pascoe Vale could be really competitive, or at least still marginal even if some assumptions above turn out wrong.  There's a risk of the Liberals running second and eliminating the Greens, but that doesn't seem that likely (Kavanagh's preferences were no help to them.) 

 Of course this does not take into accounts the usual unpredictables - the impacts of local candidates, the quality of campaigns as visible to people on the ground (have heard Labor's not taking it for granted), the face that the Greens got no 2CP swing to speak of in Wills at the federal election and so on.  Seems to me both candidates are rather good.   The people who frame and bet on markets may know such things that I do not and all this could be nonsense.   But are the Greens really only a slightly better rank longshot than the Liberals in a seat on 22.3% two-party preferred where the Liberals will struggle to even make the final two?