This week I attended the talk given by Peter Kellner, President of YouGov. It was fascinating to hear one of the leading lights of the political polling community explain his thoughts and opinions on the coming election, where issues such as the rise of the SNP and UKIP has made this a uniquely difficult election to call. Previously, I had taken the approach of using council election results to gauge the strength of opinion in each constituency. This time I have taken the opinion of those who have “skin in the game” by having a financial interest in the winner of each seat; the bookmakers. I was also interested to see how tactical voting might have a role to play in determining the winner of the election, and whether the current fragmented state of the political landscape is likely to help or hinder Ed Miliband’s attempts to unseat David Cameron as Prime Minister.
I compiled a dataset containing the best odds available on each candidate for each seat in the Election. I was able to use this data to evaluate the implied probability for each candidate for each constituency. Of course, the sum of the probabilities for each candidate for each seat do not sum to 100%. This is because, for every event that a bookmaker will take bets on, there will be an overround, which is how the bookmaker makes his or her profit. For example, if there was a horse race with two horses, both priced at 4/5, then each horse would have an implied probability of winning of 555⁄9%. The book is (555⁄9% x 2) 1111⁄9%, and the overround is 111⁄9%, so if the bookmaker took two bets totalling £111.11, split evenly on each horse (the punters each have a halfpenny), then the profit to the bookmaker is guaranteed to be £11.11. The higher the overround, the better the deal for the bookie, and the worse the deal for the punter.
Interestingly, of all the seats one can wager on, all of the worst 10 in terms of overround are in Northern Ireland. The least generous in the country is the market in East Belfast, with an overround of 29%. So not only is Northern Ireland the worst place in the UK to seek emergency medical care, but it’s the worst place to have a flutter on politics too. Typical.
I have taken any seat that has a candidate with odds of 1/20 or lower to be a safe seat. It is striking that, even given the unprecedented uncertainly surrounding May’s election, most of the seats are considered safe, uncompetitive races. The electoral map, if you only look at safe seats, looks like this.
In terms of safe seats, only one seat separates Labour and the Conservatives, who have 162 and 161 respectively. The SNP have six, the DUP five, Sinn Féin have 4. The SDLP, Plaid Cymru, Independent Sylvia Hermon, and the Speaker John Bercow all have one apiece. Neither UKIP nor the Liberal Democrats have any seats that are considered foregone conclusions by the betting markets. A spreadsheet containing all of the safe seats, as well as all the other data used to create this analysis, can be found in a Google Spreadsheet here.
So with Conservatives and Labour essentially tied in terms of safe seats, both David Cameron and Ed Miliband are almost exactly halfway to either retaining or capturing the Premiership. Both need a further 161 seats to claim the 322 needed for a parliamentary majority, assuming Sinn Féin do not take their seats. If every seat in the country goes the way the bookies think they will, what would the final tally be in terms of seats? I took the implied probabilities from the bookmakers’ odds, and adjusted for the overround in each seat to get probabilities for each candidate.
Labour are the betting favourites in 289 of the constituencies, whilst the Conservatives are favoured in 273. The Liberal Democrats are expected to win 30, with the SNP fourth on 25. After that the DUP are favourites in nine constituencies, with UKIP and Sinn Féin following them with five seats each. Plaid Cymru and the SDLP are expected to win three seats each. Naturally the Speaker is expected to hold his seat, and Sylvia Hermon is safe in North Down. Despite the surge in their polling numbers in recent weeks, the Greens are not expected to advance beyond the single seat that they hold in the current parliament. Three Scottish seats, Cambridge, and Birmingham Yardley are all considered too close to call, with two parties tied on exactly the same odds.
Given the gains the Conservatives have been making recently, this would probably be a good outcome for Labour. They would be able to govern with the support of the Liberal Democrats and any one of Plaid Cymru, the SDLP or the DUP. The Conservatives would have no real route to putting together a coalition, as they would still be five short even if they obtained the support of UKIP, the Liberal Democrats and the DUP.
Looking at betting odds gives an insight into how tactical voting and vote splitting could play a part in determining the winner. I placed all the parties with a greater than 5% chance of winning in one of three camps. The first group consists of parties who are left-leaning in outlook and would prefer to see Ed Miliband as Prime Minister; these parties are (of course) Labour themselves, the SNP, Plaid Cymru, the SDLP, the Alliance Party, the Greens and National Health Action. The second group consists of those who would prefer to see David Cameron as PM, which consist of UKIP and the UUP in addition to the Conservatives. The third group consists of the two parties that could go either way, the Liberal Democrats and the DUP.
I have assumed that, from a tactical voting perspective, voters would prefer to see one of the parties from their “team” elected as a first preference, but if there is no viable candidate from their group, then they would rather see a party from the “could go either way” group elected to keep out a party from the opposing team. For example, a Green supporter would obviously like to see a Green MP elected, but Labour would be their second preference, and if there was no viable Labour or Green candidate, they would prefer to see a Liberal Democrat elected if the alternative was for the Tories or UKIP to take the seat. I appreciate that most Northern Ireland voters don’t vote this way, but it seems a reasonable approach to take in Great Britain.
Of the 310 seats where the outcome is any sort of doubt, 98 are Tory-Labour marginals, with Labour favoured in 55 of these. There are 36 seats that are considered a straight fight between the Tories and the Lib Dems, of which the Conservatives are favoured in 22. There are 115 seats that are less likely to play a part in determining who can form a government, as they are between parties that would be likely to club together if there were coalition negotiations. Examples would include scraps between the Conservatives and UKIP, where there is no viable left-leaning candidate, or a straight battle between the SNP and Labour, who might be expected to club together if it was necessary to form a coalition or working arrangement.
There are eight seats where the only viable candidates are either Labour or UKIP. This includes Ed Miliband’s seat of Doncaster North, which the bookies reckon has an 8% chance of falling to UKIP. There are four wide open seats, where all three camps have a viable candidate. This leaves 49 marginal seats where a split vote, on the right or the left, could let in a candidate from the other team. When you look at the different categories of marginal, with the betting favourite in each one, the problem posed by UKIP to the Conservatives becomes apparent.
There are only five seats where a split on the left could lead to the Tories or UKIP winning, and a further four where a split between Labour and the Lib Dems could lead to UKIP winning. However, there are 40 constituencies where a split on the right could lead to Labour winning the seat, and Labour are favoured in 16 of these. It now becomes apparent why David Cameron is so desperate to include left leaning parties in the debates; the split voting effect is harming his chances of victory a lot more than it is harming Ed Miliband’s prospects. A rise in Green support would help David Cameron enormously by splitting the left vote in much the way that UKIP are splitting the right.
Peter Kellner made an excellent point. Had the Alternative Vote electoral system been introduced, then the split vote problem would evaporate, and the Tories may well have been able to win an outright majority. The Conservatives’ decision to vociferously campaign for a “No” vote in the AV referendum may well prove to be one of the most short-sighted and foolish decisions in modern British political history.