Predicting the 2017 UK General Election
Watch how wrong I am.
June 8th, 2017, a day previously insignificant for anyone whose birthday isn’t June 8th, is now suddenly significant (and indeed, threatening to ruin several thousand birthdays — not least the June 9th ones).
But it also provides the fun of pointless speculation. Two years ago, on a different blog, I predicted the election and was about as wrong as it’s possible to be without predicting a 200 seat majority for Sinn Fein. It was hopeless. Which is why, of course, I have to do this again.
This time, I’m going to be both more and less scientific, in the pretence that UK general elections can ever be predicted. One of the difficulties that throws everyone, of course, is how the British system is really closer to being 650 elections instead of one. Uniform National Swing — always a little dubious prior to 2010 — tends to nowadays be completely hopeless. Polling failures in both the last general elections also make everything more difficult.
But I’m starting with those polls. At present, Wikipedia’s gathered polls — which extend to those conducted yesterday — are pretty fixed about most parties. Since the declaration of a general election:
Labour has, with the exception of one potential outlier, polled at 24–26%. The Liberal Democrats are at 10–12%; the SNP are at 4–6% (a useless measure, apart from affirming that they’re still a major party). The Greens are at 2–3%, apart from one poll at 4%.
The difficulty is with the Conservatives and UKIP. With the exception of the Opinium poll, their combined score is at around 53–57%. But within that share, UKIP swings between 5% and 11% (although significantly, they haven’t yet beaten the Lib Dems). Taking away those outliers and averaging the rest suggests UKIP are on 8%, giving the Conservatives around 47%.
To confirm, the polling average, as of April 23rd: CON 47%, LAB 25%, LD 11%, UKIP 8%, SNP 5%, GRN 3%, OTH 1%.
This seems terrifying for any non-Conservative. Luckily for them, there’s a long history of UK election polling being wrong — although this has often come at the expense of Labour. There has also, however, been a tendency for polls suggesting a landslide (as in 1983 and 1997) to overstate that landslide, in a way that benefits either soon-to-be-opposition party. For these reasons, I’m tempted to cut the Conservative lead, by at least five percentage points. Labour probably won’t benefit from this exclusively — indeed, the suggestion from the polling movements is that UKIP will benefit. As a result, I’m going with a final vote of CON 42%, LAB 27%, LD 12%, UKIP 10%, SNP 5%, GRN 3%, OTH 1%.
This results in a swing, Labour to Tory, of about 4.5%. This sounds small, but it’s a worse swing against Labour than in 1983 (4.07%), and only a little worse than 2010 (5.17%) and 1979 (5.29%). In other words, I’m effectively suggesting 1983 redux and delayed, Labour having managed to more-or-less stand still in the Miliband era.
The good news for Labour is that, on a uniform swing, this suggests the loss of no more than 41 seats to the Tories, one of which (Copeland) has been lost already. Labour would, in fact, be very unfortunate indeed to face the 100+ Tory majority. For what it’s worth, of these 40 constituencies, 14 voted Remain last year, and many of those Remain constituencies are among the more marginal. It would be a little surprising, for example, for Rupa Huq (majority 274, 1% swing) to be kicked out by a 71% Remain constituency, although of course, it’s not impossible.
At this point, I think our guide is Copeland. Trudy Harrison’s 6.9% majority is perhaps exaggerated by local circumstances (Corbyn’s views on the nuclear industry) and low turnout, but its 59% Leave figure gives us a crude yardstick by which to measure potential losses. Given this, and other weakening factors such as incumbency (some of these seats are now vacant, and lack the boost an incumbent provides), I’m guessing at 26 losses, Labour to Conservative.
The next question is whether the Lib Dems can steal anything off Labour. The swing in my prediction above is around 4%, and assuming a heavily Remain seat will fall, this provides no more, in reality, than 3 losses, Labour to Liberal Democrat.
The SNP will attempt to gain Edinburgh South. Curiously, Edinburgh Southern, the equivalent Scottish Parliament seat, was regained in 2016, with the SNP losing the seat. Even so, I’m going to posit that, with the Independence question still important, and Edinburgh South heavily against Independence, Labour will just about survive this one.
My prediction posits a swing, Lib Dem to Conservative, of around 0.5%, although with both parties gaining support and given both margins of error and individual seat circumstances, it seems likely that this amounts to no real change. Given what we know about Leave/Remain votes, however, I think some strong Remain seats could fall, and predict 3 losses, Conservative to Liberal Democrat.
The SNP is the final remaining question. Current polling suggests a slight decline in support, with some excitable predictions of the Conservatives gaining 12 seats in Scotland. I can’t see it happening, as the Tories are moving up from a long way back, but Berwickshire, Roxburgh and Selkirk presents an obvious target, and hence 1 loss, SNP to Conservative. Otherwise, Scotland does not seem likely to change, especially given the alignment of the SNP to Scotland’s almost uniform vote for Remain.
I reckon there’s enough of a shift to Plaid Cymru for them to acquire Ynys Mon, given the results of elections last year, which means 1 loss, Labour to Other.
Finally, Clacton is no longer Douglas Carswell’s seat, and Carswell’s own majority got sliced in 2015. I reckon it’ll be 1 loss, UKIP to Conservative, although this will reflect as a Tory gain more than a UKIP loss — it’s already gone for them.
As a result, we get the following changes: CON +25, LAB -30, LD +6, UKIP 0, SNP -1, GRN 0, OTH 0.
The results from this would be: CON 355, LAB 202, LD 15, UKIP 0, SNP 55, GRN 1, OTH 22, with a resultant CON MAJORITY 62.
If true, then this isn’t quite an impending disaster for Labour. The main difficulty, if everything I’m saying is correct (note: it probably isn’t), is with the subcutaneous numbers — namely, that there will be very few marginal Tory seats. There are already very few Conservative majorities lower than 3,000, especially amongst Labour targets, as a result of 2015.
A more immediate difficulty, should this prove accurate, is that the Corbyn camp have set the bar at 200 seats. The anti-Corbyn faction within Labour, however, would find any number of seat losses unacceptable. If I’m right, the Labour infighting would prove ferocious.
The Liberal Democrats would be perhaps a little disappointed by this result — an improvement, without being a breakthrough. But the losses in 2015 were never going to take a single election to recover, and Tim Farron’s position would likely be safe in this scenario.
The Conservatives would celebrate an actual majority under their own steam — but just like falling short of a majority in 2010, or feeding off Lib Dem losses to get said majority in 2015, so a non-landslide in 2017 would point to an underlying weakness. Scotland is more Tory than it’s been in decades, but that’s probably still not very; and in many major cities, such as Liverpool, Sheffield and Manchester, the Conservatives are a minor party. It is possible that they will continue to weaken in London. Their ‘urban problem’ is not a huge one, but it holds them back from utter dominance — maybe.
