Everyone seems to agree: Labour are in for a pummelling at the upcoming general election. The opinion polls, the local election results and the anonymous comments from politicians of all parties on the campaign trail all point in the same direction. Even the newspaper pundits, constantly looking for a new angle, are unanimously predicting a Conservative landslide. Few, however, have tried to put numbers on the eventual outcome.
This fool is going to rush in. I’ve been looking at the Labour defence and trying to work out what will be left after the general election. Settle down; this is complicated.
Polling consistently shows that the Conservatives are doing especially well among Leave voters, gaining a much greater swing among them than among Remain voters. You can argue which is cause and which is effect but for our purposes it doesn’t matter. Wherever you find more Leave voters we can expect to find more swing to the Conservatives.
So we shouldn’t try to interpret the polls through uniform national swing without first considering how Leaveworthy the current Labour seats are. Chris Hanretty has produced estimates of Leave percentages in each constituency and while these are not going to be perfect, they should be near enough for present purposes.
Any dividing line is going to be arbitrary. A seat that divided for Leave 51:49 is unlikely to behave very differently from a seat that just came up short 49:51. So initially I divided the Labour seats into three groups: those that were clearly Remain (which I defined as under 46%); those that were fairly evenly divided (which I defined as between 46% and 54%); and those that were clearly Leave (which I defined as over 54%). You can argue with my definitions and no doubt some will.
And immediately we see a major problem for Labour. 130 of their seats won at the last election were clearly won by Leave. We can expect the Conservatives to do better than usual in these.
But there’s a more subtle problem. 33 of the 63 Labour seats that were clearly won by Remain form part of the 100 safest Labour seats, while a further 11 are among the 25 most marginal Labour seats. Very few seats clearly won by Remain are in the zone where Remain voters’ resistance to Conservative charms is likely to make a difference.
I then looked at the seats that were at one extreme or another. And just seven of the 32 Labour seats where Leave scored under 35% of the vote fall outside the 25 most marginal Labour seats and the 100 safest Labour seats. Four of those seven seats have the Lib Dems or the Greens in second place. As chance would have it, the type of seat where Labour is best placed to hold off the Conservatives looks as though it isn’t going to be that relevant in this election.
At the other extreme, there are 75 Labour seats where Leave won more than 60% of the vote (yes, that’s asymmetric with the “extreme Remain” banding). That’s more than all the Labour seats where Leave scored under 46%. 20 of the seats in the 51st to 100th most vulnerable Labour seats are in this category. Another 10 seats in this category are in the 101st to the 125th most vulnerable Labour seats.
Anecdote, the tenth wave of the British Election Study (as analysed by Chris Hanretty) and ICM subsamples all suggest that the Conservatives are doing particularly well in Labour-held seats. The distribution of the Labour Leave-leaning and Remain-leaving seats might well explain that.
What does that mean in terms of seat losses? First, we need to work out what the likely swing from Labour to the Conservatives is likely to be and then adjust. Working on the basis that often the polling at the beginning of the election campaign is what the final result ends up as, I’m assuming a swing of roughly 7.5%.
Then we need to decide how this will differ in different bands of seats. I’m working on the basis that the swing will be 4% lower in extreme Remain seats, 2% lower in less extreme but clearly Remain seats, 2% higher in clearly Leave seats and 3.5% higher in extreme Leave seats. This would result in 79 seat losses to the Conservatives. On this model, Labour would hold Exeter and lose Don Valley. Allowing for other losses to the Lib Dems and other parties, this would leave Labour with just under 150 seats.
Of course, this is all pretty arbitrary, with assumptions galore. The polling might be a lot tighter (if the swing is only 5%, Labour would end up with somewhere around the 170 mark). Or Labour might collapse much further (if the swing is 10%, Labour would end up with somewhere around 130). I may have radically underestimated or overestimated the swing differential in different types of seats. But I expect the result to be in this ballpark.
I note that Bet 365 offers odds of 9/4 on Labour getting 126 to 150 seats and 5/2 on them getting 151 to 175 seats, implying they will get in this band roughly 60% of the time. I think the chances are more like 85%. Or if you want to back just a single band, Betfair Sportsbook’s and Paddy Power’s 7/4 on 120-159 seats looks generous to me. I’m on.