Looking at Labour’s share of the vote in the polls and what it means

May 19th, 2017

Wikipedia GE2017 poll chart

All the mood music is pointing one way. Both Labour and Conservative sources suggest a meltdown in Labour’s heartlands. The Conservatives have put out rumours that they are trying to take seats such as Leeds East, West Bromwich East and Bolsover. If they were to succeed, Labour would be reduced to a rump.

Conventional wisdom has it that the Conservatives are hoovering up the UKIP vote in Labour heartlands and taking a fair chunk of Labour leavers with them.  Political bettors have bought into this, with the Conservative over/under benchmark currently being set around the 400 mark, implying something like 70 seat gains.

There’s only one problem with this picture.  Since the election was called, Labour have risen quite markedly in the polls.  When the election was initially called, the Conservatives were polling between 45 and 50% while Labour were polling around the 25% mark.  Currently the Conservatives have stabilised at 47-48% while Labour are polling around the 30% mark.  Only ICM have them lower at 28% and even they have seen a substantial rise from 25% a couple of weeks ago.

The reaction to this movement among commentators and bettors has been largely to dismiss this news.  Even the pollsters themselves are openly sceptical that they are finding anything significant.  The general view is that either the polls are wrong – remember Brexit and Trump? (though actually the polls performed respectably at a national level on both occasions) – or that public opinion will shift back before the election.  Either way, public discourse is taking place on the basis of the polls at the outset of the campaign and assuming a swing of 7 or 8% that is non-uniform and favours the Conservatives heavily.

This might be the correct reaction.  Heck, I’ve made similar arguments myself.  But before deciding that it is, we should at least first seriously consider the possibility that the polls are telling us something useful.  They are striking in their consistency.

The first thing not to do is to make the opposite mistake of assuming that the Conservatives aren’t going to poll in the high 40s.  Some anti-Conservative commentators readily identify any number of targets that the Conservatives aren’t going to take on the basis that they simply won’t pick up enough support there.  But the Conservatives are going to pick up a lot of new votes somewhere (unless their support wanes, which it is showing no signs of doing just yet).  With those new votes they can expect a lot of new seats.

And then we run into the next piece of conventional wisdom, which is that the swing will be non-uniform because of Brexit.  This contains not one but two very dubious assumptions.  First, it is far from clear that Brexit is the driver of the Conservative rise in the polls.  The decision to vote for or against Brexit is more likely to be an effect of characteristics of individual voters that carry over into this election rather than a cause of future voting habits (I accept that there is an element of feedback here).  An urban metropolitan liberal professional was highly likely to vote Remain and for the same reasons is likely to find anti-immigration social conservatism and old-fashioned corporatism unattractive.  A working class northern pensioner is likely to react very differently to both prompts.  The pre-existing characteristics seem more relevant than the EU referendum vote for determining likely polling intentions.  Brexit choice may be a useful shorthand for these characteristics in some cases but a staunch socialist who voted Leave is unlikely to have a Damascene conversion to the Conservatives.

The second very dubious assumption (and in my view the more significant one) is that the swing would have been uniform but for the complicating consideration of Brexit.  When a party has a large rise in the polls, it is inherently unlikely to be distributed evenly.  Uniform national swing works best with small movements in the polls.  Large rises are driven by a new coalition being built.  That new coalition is likely to be concentrated in certain areas and of minor relevance elsewhere.

So the key questions for this election are the paired questions: where is Labour support holding up?  Where are the Conservatives building their new coalition?  In Hull, Halifax and Huddersfield, hurricanes hardly ever happen.  Are we about to see one?  Perhaps.  But it seems to me that most of the value in the constituency markets now is in identifying seats where the Conservatives’ provincialist corporatism is going to meet a cooler reception and backing Labour, because the polling is currently suggesting that Labour’s coalition is holding up pretty well in some places.  For example, that’s what I’ve done in Hove actually at 5/2 with Ladbrokes.  There are plenty of others if you go looking for them.  I suggest that’s where the betting effort is most usefully focused right now.

Alastair Meeks