David Herdson says “LAB most votes – CON most seats” is a good bet at 66-1
Why the “impossible” could happen
Labour has won most votes at a general election before and come out behind on seats. It happened in 1951, when Attlee’s Labour polled over 13.9m votes: around a quarter of a million more than Churchill’s Conservatives, who nonetheless formed a majority government (and for that matter, more than Labour has ever polled in any other election). A lot has changed since then and the conventional wisdom is that such an outcome is no longer possible. And indeed, without great change, it’s not – but great change is happening and greater change still could be just around the corner.
The reasons why it shouldn’t be possible are simple enough: the turnout in Labour’s safe seats is much lower than equivalent Conservative ones, Labour’s seats are on average slightly smaller, and Labour has fewer seats where it is a moderate third, winning a decent number of votes without being in sight of winning. There is a fourth reason: Scotland. In 2010, Labour won one MP for every 25,000 votes north of the border; the Tories won one for over 400,000 votes.
For the bet to come in, three things would need to happen: firstly, Yes would have to win the referendum on Thursday; secondly, Lab would have to have a narrow lead on polling day; and thirdly, there’d need to be a differential impact in the two big polling shifts during this parliament. Taken together, I think the chances of that happening are better than the 66/1 Ladbrokes are offering.
To take Scotland first, I’m surprised at how long the odds on Yes are. True, the polls mostly have No ahead but only by a few points. I’ve written before about why I’m not putting too much faith in the polling and the sudden clustering around 52-48 doesn’t give me any reassurance: are they all more-or-less accurate or are they finding comfort in each other’s company?  Even if they are right today, it wouldn’t take much of a swing for Yes to be ahead on Thursday.
Of course, should Yes does win, Scotland will still be a part of the UK in May 2015 but we could expect very different election results from a soon-to-be-independent Scotland compared with one with a future in the UK. The three GB-wide parties are likely to be in a state of shock and the SNP riding a wave, Labour’s battle-cry, to ‘protect’ against evil English Tories would be obsolete, and most of all, the voters would be unlikely to wish to send mixed messages about the independence negotiations. In 2011, the SNP won 53 constituency seats to Labour’s 15 and the Tories’ 3. That’s probably not a bad guide to what might be expected should Yes prevail.
The second condition, of Labour being narrowly ahead can be quickly dealt with: it’s where we are now and although a Scottish Yes might shake up politics in England and Wales, there’s no reason to assume that it would substantially benefit either Labour or the Conservatives’ relative to the other. There may be Black Swans between now and then, including possible leadership changes, but this is a probabilities game and the probability to my mind is for little change in that lead.
The third factor is more interesting. For Labour to win most votes but fewer seats, their vote would have to become less efficient, even if the Scottish effect is substantially reduced. That might well happen too. We know that the two main polling movements since 2010 have been a big shift from Lib Dem to Lab, and the rise of UKIP at the expense of all the others but principally the Tories. The question here is whether those movements will occur disproportionately in safe seats as against marginals.
In theory, this is what you’d expect to happen. Where the results were close in 2010, you’d expect tactical votes to have already been substantially squeezed so there’d be fewer Lib Dems to move across. Likewise, although UKIP voters as a whole may be relatively ambivalent about the alternative prospects of Miliband and Cameron as PM, there are still plenty to be pressured in the marginals, whereas ‘safe’ seats (I write this in advance of Clacton!), may offer a more cost-free protest vote, reducing Tory majorities and votes while still returning MPs. In reality, the constituency polls don’t particularly bear this out but there are still nearly eight months to polling day and a lot of hard campaigning to go between now and then.
As with all long-odds bets, I’m not suggesting that this scenario is going to happen. What I do think is that there’s a better than 1 in 67 chance that it will – and on that basis, there’s value.