I’ve got other things to be doing today, but just time for a quick post about the Greens.
The occasion is the forthcoming by-election for the seat of Batman, vacated by the resignation of its Labor MP, David Feeney, on the grounds of dual citizenship. As some readers will remember, the Greens candidate for the seat, Alex Bhathal (who is favorite to win it), is a close friend, so I don’t plan to write much about it; I am not an impartial observer.
If you’re interested, though, there’s a very good preview last week from Antony Green, who explains the background and draws attention to some differences between the Greens in Victoria and New South Wales. There’s also an interesting piece yesterday on it by Tim Colebatch at Inside Story.
Colebatch’s thesis is that while the Greens’ vote nationally is in decline, they are continuing to make gains in terms of seats because they are targeting a small number of seats effectively, particularly in inner-city Melbourne. There is clearly some truth in this, but overall I think it’s misleading, so I want to try to explain why.
Let’s start with this claim: “Back in 2010–11, the Greens’ parliamentary numbers peaked at thirty-five. Since then, their vote in most of Australia has fallen.” Is that true?
About seven years ago I produced a table showing the progression of the Greens vote; it’s obviously time to do it again. Here is their vote at the last six elections in all jurisdictions (figures from the UWA elections database):
So it’s true that the Greens’ current vote is below its peak federally, as well as in Western Australia, Tasmania and the territories. But in the other four states – surely “most of Australia” – it’s consistently risen. And even to describe the federal vote as having fallen is somewhat misleading, since the most recent movement is upwards; just not enough to counteract the fall from 2010 to 2013.
Colebatch’s table headed “The Shifting Base of Greens Support” breaks the federal vote down into two periods, 2004-10 and 2010-16, from which he gets significant results about the change in its distribution. The contrast between Melbourne and Sydney is particularly striking: in the latter period, the Sydney Greens gave back almost all the gains they’d made in the first period, while in Melbourne they kept growing.
When he comes to analyse that support by electorate, however, Colebatch changes his measurement. The table headed “Swings in Key Seats 2010-16” gives us “Greens vote at the three-party stage of the count” – in other words, the numbers are three-party-preferred votes, not primary votes.
Now, if you’re looking at a party’s prospects of winning seats, that’s a perfectly sensible measure to use. But if we do, it makes everything look different, not just the inner city. That’s because, as Colebatch mentions, the number of minor parties has increased in recent years, so major parties (which in this context includes the Greens) may suffer in their primary vote even if their position at the end of the count, where it really matters, is no worse.
To demonstrate, I downloaded the electoral commission’s detailed preference data and calculated a three-party-preferred House of Representatives vote for the whole of Victoria. In primary votes, the Greens improved only slightly between 2010 and 2016, from 12.7% to 13.1% (it’s the only state where they improved at all). But three-party-preferred, they jumped from 14.5% to 17.1%.*
At some point I hope to get time to do the same calculation for other states and other elections, but I suspect that they will show the same thing: that in at least one sense, the post-2010 dip in the Green vote is illusory, because the lost votes are coming back to them as preferences from smaller parties.
Just like two-party-preferred vote, three-party-preferred has to be used carefully. If a party forgets about its primary vote and only looks at its position after preferences, it can strike some unpleasant surprises. But fundamentally ours is a preferential system, and where the votes come from is less important than where they end up.
* Technical note: there was one seat in 2010 (Wannon) and three in 2016 (Holt, Indi and Murray) where the Greens were not in the last three, so I left them out to make the figures comparable. If you include them, the totals are 14.1% and 17.0%. All four, of course, are poor Greens territory, so my three-party-preferred figure is slightly overstating the level of Greens support, but that doesn’t matter because it’s the swing that we’re interested in.
ADDENDUM (same day, 8pm)
I’ve now done the three-party-preferred calculation for 2013, and it’s a real surprise. I was expecting that, as I said, it would show “the post-2010 dip in the Green vote [to be] illusory”; it does that, and more. The raw three-party-preferreds are like this:
2010 14.1% 2013 15.2% 2016 17.0%
Factoring out the seats where for at least one of the three elections the Greens didn’t make the top three (of which 2013 adds another three – Calwell, Mallee and McEwen), you get this:
2010 14.7% 2013 15.4% 2016 17.8%
Remember, these were the primary votes:
2010 12.7% 2013 10.8% 2016 13.1%
So the fall in the Green vote in 2013, much discussed and analysed, isn’t there at all on preferences – at least not in Victoria.