CBbZfgg

In this post (first published on Friday 27th February), the team at May2015.com compare five general election forecasts, and find that they are in broad agreement: Labour & Tories are likely to fall at least 40 seats short of a majority, and a hung parliament seems inevitable. 

The Guardian has joined the polling and prediction business. Since September May2015 has been tracking the polls and making election predictions. Earlier this year we started to track other models, and the Guardian have now been added to our page of forecasters.

May2015 now tracks five forecasts – our own model, a pair of academic ones (Election Forecast and Elections Etc), the Guardian’s, and the bookies’ (Ladbrokes). They are remarkably similar. The five forecasts range from 270 to 285 seats for the Tories, and 271 to 283 seats for Labour. An average of the five models gives 278.1 seats for the former and 275.3 for the latter.

Screen Shot 2015-03-05 at 16.11.53

May2015’s 270-seat prediction is the most pessimistic for the Tories. Election Forecast are behind the more favourable 285-seat prediction. We also offer the lowest estimate for Labour (271), along with the Guardian. Elections Etc suggest the party will win as many as 283.

Will the SNP win more than 50 seats?

Our forecasts are the lowest for both main parties because our model is so favourable to the SNP. That’s because we used Lord Ashcroft’s 16 Scottish seat polls to predict the whole of Scotland (his polls account for 26 per cent of Scotland’s seats).

We feel comfortable doing so because the national vote share his polls implied – SNP 48 per cent, Labour 23 per cent – are largely in-line with more than a dozen Scotland-wide polls released since the referendum. As new Scottish seat polls are released, our forecast may change (Ashcroft polled seats that were more likely to vote “Yes” – and therefore SNP – in the Scottish referendum). Our forecasts are the lowest for both main parties because our model is so favourable to the SNP.

The Guardian are the only other forecaster predicting more than 50 seats for the SNP (51). The way we make our predictions are very similar – we both rely on a mixture of national and local polls.

The two academic models and the bookies offer less sensational predictions for the SNP. But they are now far more nationalist than they have been. All three now predict the SNP will win around 40 seats. Two months ago the bookies suggested the SNP would only win 25. (An average of the five forecasters gives a prediction of 44.9 seats.)

Can the Lib Dems hold onto more than half of their seats?

We recently launched the New Statesman’s Political Index, which suggested the Lib Dems could win at least 30 seats in May. We revealed how private polling is offering hope to the coalition’s minor party. We will shortly be launching the Index as a standalone prediction (and ranking every seat in the UK), but none of the five forecasts we currently track suggest the Lib Dems will do so well.

These models are all based on public polling, which doesn’t take into account how well the Lib Dems say they are doing in seats like St Austell & Newquay, Cardiff Central, Solihull, Bermondsey, Leeds North West and St Ives. Until these polls are made public, or Lord Ashcroft returns to these seats and confirms the Lib Dems are doing well (or national polls improve for the party), Lib Dem seat predictions will remain below 30. An average of the five forecasters suggests they will win 25.9 (that’s down from 57 in 2010).

All roads lead to a very hung parliament

The most important part of these predictions are their forecasts for Labour and the Tories. And they all agree: both parties will fall at least 40 seats short of a majority. The Guardian has some whizzy graphics that detail how many seats various coalition or confidence & supply deals could have.

The key number is 326. If a group of parties can get to 326 they will have a majority in the Commons (there are 650 seats). But the magic number is really 323; the Speaker is apolitical and Sinn Fein’s five MPs don’t take their seats. With that in mind, here’s a rundown of how many seats various groupings could have:

  • Labour (275.3) + SNP (44.9) = 320 (3 seats short; figures rounded)
  • Tory (278.1) + Lib Dem (25.9) + Ukip (3.3) + DUP (8) = 315 (8 seats short)
  • Tory (278.1) + Lib Dem (25.9) = 304 (19 seats short)
  • Labour (275.3) + Lib Dem (25.9) + Green (1) = 302 (21 seats short)

Who will form the next government?

If either party forms a coalition just with the Lib Dems, they will be around 20 seats short of a majority. The only two feasible majority scenarios are a Labour-SNP deal, or a Tory-Lib Dem deal propped up by Ukip and Northern Ireland’s DUP. If either party forms a coalition just with the Lib Dems, they will be 20 seats short.

There are reasons to doubt both outcomes. The SNP and Labour are currently fighting a fierce battle in Scotland – are they going to easily come together after the SNP wins dozens of Labour’s Scottish seats? The SNP’s Westminster leader also recently told May2015 he is not spending any time preparing for a coalition; any Lab-SNP deal would only be a confidence & supply deal.

A Tory-led, four-party coalition is also hard to envisage. A Tory-Lib Dem deal, with occasional Ukip and DUP support, is easier to imagine, but accommodating four parties – as well as Tory backbenchers, the most rebellious MPs in the House – will be extremely complicated.

Note: The Financial Times have started to publish a prediction by Populus, the pollster, and Hanover, the PR firm. It is not yet updated regularly enough for us to track, but its latest predictions are very much in line with those we do track. They are predicting 277 seats for Labour, 272 for the Tories and 44 for the SNP. We have stopped tracking Electoral Calculus’ model because it is too irregularly updated. This post was originally published on the May2015 website.

About the Authors

CBbZfgg

Harry Lambert is Editor of the New Statesman’s May2015 site.

Tom Monk is Head of Data and Development at the New Statesman’s May2015 site.

Print Friendly