Tuesday, 26 April 2016

Keeping track of Labour's performance - part three

Every time we've come to look at the UK Labour Party's polling and by-election performance in recent months, it's not been a pretty sight. We've basically had to shake our heads sadly and move away from the prone and stricken figure of a once-great party. If you're a Labour member or supporter, historically-informed and data-driven projections have for months painted a very ugly picture indeed: of Labour losing scores of seats at the next General Election, and perhaps being comprehensively routed in a manner it hasn't suffered since 1931 and 1935. Those numbers seemed to stagnate, rather than improving, as time went by.

Now things might be looking a little bit better, because the governing Conservatives' implosion over a very ill-received Budget, Tata Steel's difficulties in South Wales (and the Government's handling of that issue), party splits over the forthcoming European Union referendum and revelations about Prime Minister David Cameron's own personal finances have all blown up in one spectacular - and rather epic - month of crises. The Conservatives large poll leads suddenly seemed to hit a wall in March, heading down from eight or nine percentage points to just two or three. Not exactly a transformation, as we'll discuss in a moment, but certainly a break from their polling walkovers of the last year.

At the same time, Labour leader Jeremy Corbyn has started to look rather more comfortable in his own skin. Since the nadir of failing to mention Iain Duncan Smith's name when questioning Mr Cameron about Mr Duncan Smith's resignation, he's sharped up his act. A forensic list of questions about the Government's ludicrous forced academisation of English schools gave him his best Prime Minister's Questions to date (admittedly quite a low bar); a dry and self-deprecating seam of humour has begun to emerge, as it did when Mr Corbyn praised the Queen on the occasion of her ninetieth birthday. Although his personal ratings are still pretty bad, they've improved, mostly off the back of (the remaining ranks of) Labour voters revising their opinions of him a little: his numbers are still worse than many past Leaders of the Opposition, but they are no longer particularly atrocious.

But how to see this in context? What does this gap actually mean in terms of past history - and the next General Election? Well, we promised last time to keep an eye on just this question, and that's what we're doing now - while Labour look a little bit more chipper, and the Conservatives more downcast, than they have since Ed Miliband looked as if he was a realistic challenger for the keys to No. 10.

Start here: Labour's poll ratings are still awful. They are worse than those for any past period of modern Labour Opposition apart from Neil Kinnock's travails in 1984 and 1988 (above). To be doing about as well as Mr Kinnock, busy at the time trying to hold his party together as Mrs Thatcher's ideological revolution rolled over Labour - threatening to crush the life out of it - is not exactly a triumph. As usual, we do have to acknowledge here that polling methods have changed out of all recognition since the 1980s, and even since Ed Miliband's period as Labour's leader (though by nowhere near as much as you might think, at least in the latter case). Even so, the direction of travel and the orders of magnitude aren't particularly in dispute: Labour is now doing better than during the winter, but still not very well. It's at about the level it reached in the 1980s - competitive in midterm, perhaps, but apparently unlikely to win any national elections. 

Now, okay, polling's under a cloud at the moment. If pollsters do well at the EU referendum, perhaps they'll be let out of the doghouse, but for now a lot of people will just say 'ah, the polls', and switch off. They would be very unwise to do so. That's because we also have a good check on the polls in the case of local authority by-elections, held most Thursdays and giving us a handy check on polling via tens of thousands of votes being placed in real ballot boxes every quarter. 

Here there's still more difficult news for Labour to absorb. In this calendar year, there has been almost no swing at all in these, as measured by vote changes since the last Parliament. There has, in fact, been a very small (0.5% or so) swing from Labour towards the Conservatives in those twenty contests from which we can measure Labour and Conservative vote changes, and which were last fought in 2015. Since the local election share of the vote in 2015 (as opposed to the Parliamentary election numbers) saw the Conservatives ahead at 36% to Labour's 32%, that puts Labour about five points behind - by no means way out of line with voting surveys at all (and actually helping to suggest that polls are still overstating Labour, but that's another story). 

Just to give you another yardstick, during the same four month period in 2011, Labour secured about a five per cent swing in their favour, a figure that went up to seven per cent when we just look at contests that had last been fought in 2010 (you can play with all the numbers here if you like). Now we do know that Labour were coming from a lower base at that point, but it wasn't vastly lower because both 2010 and 2015 were General Election years when Labour could rely on getting out their vote. The by-election swing in early 2011 suggested that Labour were five or six points ahead of the Conservatives; they actually ended up a point behind them in the local elections held that year. So our chosen performance indicator here again, in the end, flattered them. 

All this is why local election gurus Colin Rallings and Michael Thrasher think that Labour will do quite badly in the upcoming local elections. They might lose 150 or so seats and go backwards in Opposition - something that's only happened twice (again, in the 1980s) in recent non-General Election years. Rallings and Thrasher by the way also utilise their key wards methodology to argue that Labour are progressively falling away in these local by-elections, not picking up - as they once did, during 2011, under Mr Miliband. Other respected analysts back up their case. 

Labour are way behind their 2011 polling scores in terms of the Welsh Assembly elections (though their Westminster numbers are not quite as bad), and are really, really struggling to keep their heads above water in Scotland. If all of this analysis means anything at all, they are in retreat pretty much everywhere. Only in London are they doing well, though apparently there are last-minute jitters over Sadiq Khan's chances even here (we think he'll win easily). The retreat is looking rather less rapid and ragged at the moment, but it's still a retreat: the very small number of competitive London marginal seats will do very little to buoy Labour in May 2020.

So let's update our seats analysis. In November, we thought that Labour might win only somewhere between 145 and 190 seats (on new boundaries) at the next General Election. In January, that number was not particularly changed: it had got to between 153 and 178. Now? Well, if we look at the likely local election results that Rallings and Thrasher project, we can probably expect a ten or eleven-point deficit between Labour and the Conservatives in 2020 (£). That might see Labour defeated, on a uniform swing, by something like 40% to 29%. On old boundaries, that would give us an overall Conservative advantage in the House of Commons totalling about 50, and Labour losing perhaps 20 seats to fall to a range of 210-215. On new boundaries, that looks a bit worse: a Conservative majority of perhaps between 80 and 90, with Labour returning only 190 MPs.

If, on the other hand, we apply the extent to which Labour's vote has fallen between this point in previous Parliaments to the next national contest (and adjusting for poll error, on the same basis as our previous entries in this series), the numbers suggest that there will be a fall of 7.9% from here (Labour's ratings have never risen from this point in a Parliament to polling day), with a range of 16.9% (between 1980 and the 1983 Conservative landslide) and just 0.4% (between 1993 and 1997, a time during which Labour's numbers were supported by the ascent to power of a certain Tony Blair). Let's subtract 1.7% off all those numbers, as we did last time - the average amount by which the polls have overstated Labour on election day itself, and thus a necessary correction to get at something like the 'true' polling number that actually reflected its standing between elections. So, if the past is anything to go on, there might be a 6.2% decline from here: a polling average of 33.6% right now would become 27.4%. If we do no more than give the Conservatives just one point of this ex-Labour share on top of their 2015 General Election score - a very limited concession to them, given the history of government recoveries from mid-term - this suggests very similar results to our rough forecast from local election numbers: Conservative majorities of about 60 or 90 on the two different electoral maps, with Labour hovering just above, or some way under, 200 seats.

That's why you know in your heart that - if nothing changes, as change it might well - Labour still looks very likely to be defeated in 2020. It's hard to imagine, right now, how the Government could be in more disarray. But the Conservatives are still, on average, ahead, while the bookmakers will at the time of writing quote you only a 15% change of a Labour government. That doesn't seem too far away from reality.

So Labour's doing a bit better. They've moved on from the previous absolutely horrific state of affairs, and a predicted electoral slaughter, to just a really bad and resounding defeat. Our sustained look at historic statistics now moves them up 35-40 or so seats. Their losses on these numbers might now range between 20 and 50. But losses it still seems likely that there will be, even if their better polling at the moment doesn't prove to be a bubble born only of the Government's sixth-year unpopularity.

Maybe we're wrong. Maybe all of this is just a figment of statistical reality. But we now have a real test to hand: a huge number of local, national, Police Commissioner and Mayoral elections that will tell us if all this is right. Maybe Labour will do well - especially in the English local elections - pushing ahead, winning seats and outstripping the Conservative vote share by more than just a couple of points. We're about to get a really strong signal. Reader, we can't wait.