I hadn’t been planning to write a substantive blog post for election day – I usually just post an open thread for discussion during the day. But yesterday I had an idea for one more bit of analysis that seems relevant to consider before tonight.
I have heard a few smart people point out that there is a history of a party winning the two-party-preferred vote but losing the election, and that is true. It happened twice in the 1960s, and again in 1990 and 1998. But this also made me think of a theory I have been explaining when asked about the chances of a hung parliament. A hung parliament depends on two factors: the closeness of the result between the major parties, and the size of the crossbench. A tiny crossbench requires a very close result to produce a hung parliament. A much larger crossbench makes a hung parliament more likely.
So I thought I would examine these two interrelated topics: how does the size of the crossbench affect the chances of a hung parliament, and do those increased chances make a “wrong winner” result less possible? I’m defining a “wrong winner” as a major party winning a majority in the House despite losing the two-party-preferred vote, not simply forming government in a hung parliament.
First up, I gathered data since the 1984 election on the number of seats that would have needed to change hands for the newly-elected government to not win a majority, and what uniform swing would have been needed for those seats to change hands, based on the post-election pendulum. I also then calculated the uniform swing needed for enough seats to change hands for the loser of the election (the new opposition) to win a majority. In effect this creates a range within which a hung parliament would have been produced, or the “hung parliament range”.
This chart also shows the number of crossbenchers elected to the House at that election (in brackets after the year), and the national two-party-preferred margin marked in red.
You can see that the national two-party-preferred margin often falls outside of the hung parliament zonw, which implies that a uniform swing that would have changed the national 2PP winner would not have changed the outcome. This is true for all but one election between 1984 and 2007. Only in 1996 did the national two-party-preferred margin fall within the hung parliament range. Incidentally, the 1996 election produced the most crossbenchers of any election prior to 2010.
Since 2010, three out of four elections have produced a hung parliament range that overlaps with the national two-party-preferred margin, and it was just outside the range in 2016.
This suggests that a uniform swing could have created a "wrong winner" result in 1984, 1987, 1993, 2001 or 2004, although 2004 was only just outside of the hung parliament range. Indeed the Labor two-party-preferred vote in 1993 was quite a bit stronger than their seat count, and a uniform swing of just 0.4% would have flipped seven seats and put them into hung parliament, and a uniform swing of just 0.61% would have flipped nine and given the Coalition a majority.
So why the change over the last decade? Well, I think it's largely because the crossbench has grown.
There were five crossbenchers in 1996, but otherwise there were never more than three prior to the 2010 election. Since 2010, we have never had less than five.
I think these first two charts show that there is a relationship between the size of the hung parliament range and the size of the crossbench, but the next scatter plot makes that very clear.
The two elections with the widest hung parliament ranges were 2010 and 2019, in which six crossbenchers had been elected.
In contrast, the only "wrong winner" results in the last half century both took place in a political environment where the House was almost entirely dominated by the major parties, even if both elections were famous for a surge in the minor party vote. Indeed, the two "wrong winner" results in the 1960s took place with no crossbench at all.
We don't know what will happen tonight. It seems unlikely any of the sitting crossbenchers will lose, and there are numerous prospects for crossbench gains. For all I know we could see ten or more crossbenchers in the House. That makes it very likely that the range of two-party-preferred vote figures that would produce a hung parliament would be wider than it has been for a very long time.
That doesn't mean we will have a hung parliament. If Labor wins the two-party-preferred vote comfortably they could well clear that range.
As long as we have a large crossbench, that will structurally change the nature of close results. A major party can win a majority if they win comfortably, but a close election will likely result in a hung parliament. That may not be permanent, but it will likely be with us for quite some time.
We may well not know the outcome of the election when we go to bed tonight, but it seems very unlikely that both a Labor majority and a Coalition majority will remain options at that time. As the major party vote has declined and the crossbench has expanded, the chasm between these two majority options is getting wider, and I think it's very unlikely that enough seats will be in play at the end of tonight to leave both those options open.