UK 2015 forecasts

With the UK general election three weeks away, Chris Hanretty of the electionforecast.co.uk team offers a comparison of their forecast with those of two other academic teams, ElectionsEtc.com and Polling Observatory.

The forecasts must estimate a nationwide vote share for each party, and then devise a means of projecting these figures on to the 650 individual plurality (FPTP) contests that make up a UK House of Commons election. Given these two stages for any forecast, Chris’s comparison includes running their own seat calculator on other projections’ vote shares.

One thing all the projections agree on is that no party will be close to half the seats, although the confidence interval on the ElectionsEtc and ElectionForecast figures for the Tories include the majority mark (barely). All agree that Labour has almost no chance of winning a majority.  Another point of agreement is that the leading party in votes will be the Conservatives, but on only 34%, with Labour just a percentage point or two behind. In each projection, the confidence intervals on vote shares overlap. So, yes, the race remains more or less tied.

For those of us who enjoy anomaly watches, the PollingObservatory forecast (as of 1 April) has Labour ahead in seats, 276-271, despite being just behind in votes. The other two put Conservatives ahead on seats as well as votes, although again with overlapping confidence intervals.

On the eve of the 2010 elections, I ran some numbers from FPTP elections around the world up to that time, asking how uncommon it was for the largest party have less than 36% and the third party have more than 25%. The answer, in my sample of 210 FPTP elections, was one (Nova Scotia 1998). There was at least one other that was missing in my data, Quebec 2007, where all three leading parties were within a range of 33.1 to 28.3. The question arose, because it was the consensus of the final polls that the UK was going to have such an election. However, it did not quite get there. The Conservatives made it just over 36% and the third-place Liberal Democrats slipped to 23.0. In other words, there may have been some last minute tactical (strategic) voting by just enough voters to make the result just a little more “Duvergerian”.

If we think of Duverger’s Law as a baseline expectation for an electoral system like the UK, the projections for this election might be said to conform, if all one cares about is the relative dominance of the top two over the rest. Thanks to the fall of the Liberal Democrats since entering the coalition in 2010, all projections agree that there will be a big gap between second and third. Of course, the top two on around 33% each is not exactly what we normally think of as a Duvergerian outcome, regardless of the gap between second and third. The interesting thing to watch is if there is sufficient late desertion of the UKIP (on 10-13%) and Greens (less consequentially, as they are under 5%) to push the leading party up over 36% again. If that is the Conservatives, there might be an outside chance of a majority government, depending on the constituencies where such tactical voting takes place. UKIP leader, Nigel Farage, recently called on supporters to vote Conservative in seats UKIP can’t win, and there is some evidence of UKIP slippage in recent Ashcroft polls of marginal constituencies.

For now, let’s suppose that the seat projections are about right. Are there any combinations of two (other than Labour-Tory) that would control a majority? The House contains 650 members, but Sinn Fein (of Northern Ireland) does not take the seats they win, which were five in 2010. That makes the majority threshold probably at 323. A combine of Labour and Scottish National Party (SNP) gets there, according to Polling Observatory–barely: 325 (or 324 if using ElectionForecast’s seat model).

How likely is Labour-SNP cooperation? Nicola Sturgeon, SNP leader, has been appealing for it–not for a formal coalition, but for blocking a Tory government. An exchange in the final debate (opposition parties only; video) was telling. Labour leader Ed Miliband forcefully refused, saying he had fundamental disagreements with the SNP, mainly over the possibility of a second Scottish independence referendum. He said no to a “coalition”. However, he pointedly did not directly say there could be no cooperation short of a governing coalition, although he shook his head dramatically when Sturgeon appealed for him to join her to block the Tories. He is in a bind, for sure. Sturgeon is probably right when she says left-leaning voters would never forgive him if he refused cooperation and allowed a Tory minority government to form. On the other hand, he certainly has to be careful not to signal intention of working with a party that would break up the UK.

A path to a stable Conservative-led minority or coalition government is hard to see, on these projected numbers. However, two of the projections have Conservatives and Liberal Democrats at 311. I find it hard to believe the Liberal Democrats would enter a coalition again with the Tories, but less hard to believe they could enter a looser arrangement. However, such a combination would still be a around a dozen seats short, and even the Northern Ireland Democratic Unionists (not in the projection models, but 8 seats in 2010) would not quite get them there.

Absent a late surge for the Tories, it still looks like a government led by the party with the second largest number of seats, and needing support from Scottish separatists, is the most likely result. Let me close with an understatement: This will be interesting.