Figuring out the next election

I have bad news for those who complain, as I’m told Aneurin Bevan once did, that statistics take the poetry out of politics. The analysis that follows is mainly about numbers. There is a reason for this. The outcome of the next election—who will win, and how well—will depend on the way thirty-plus million votes translate into 650 seats in the new House of Commons. On election night, computers will outrank poets.

Moreover, the computers may surprise the unwary, for the relationship between votes and seats has changed markedly in recent years. In 2005, the last time Labour won a general election, it secured a 66-seat majority with a lead of three per cent in the popular vote. Today, that lead would leave it well short of a majority; it might not even be the largest party.

Or consider Labour’s biggest victory: the Blair landslide in 1997, when it won a majority of 179. Today, the same lead in the popular vote, 13 per cent, might see it merely scrape home. We shall come shortly to the reasons for these big changes. First, the numbers that will determine the shape of British politics after the next election. The graphic sets out the main staging posts on the way to the result. It converts different Conservative and Labour leads from votes into seats, and in two ways: (a) if we assume that marginal seats swing much the same way as safe seats, and (b) if there is tactical voting in marginal seats on the same scale as in 1997.[1]

Start with the uniform swing column, and what the Conservatives need to do to remain in office. They won the last election with a 12-point lead in the popular vote[1]. They need a four-point lead to retain a working majority and a three-point lead for a bare overall majority. However, they could remain the largest party even if they lag five points behind Labour in the popular vote.

These figures are based on the new boundaries on which the next election will be fought; I have used the data on the Electoral Calculus website to explore the scenarios. As with all such calculations, the numbers of seats are inevitably best estimates. The actual numbers on the night are likely to be slightly different, but not different enough to change the conclusions.

Why is the mountain Labour must climb so much higher than in the past? Here are four reasons.

Scotland is by far the biggest factor. In 1997, Labour won 56 seats north of the border. At the last election, it held only one. On a uniform Britain-wide swing a 13-point lead at the next election would give it just three. Unless support for the SNP collapses, It’s hard to see Labour climbing out of single figures. Every lost seat reduces Labour’s majority by two; so Scotland’s fifty-plus lost seats reduces its majority by hundred-double-plus

Second, a small but certain factor. In 1997 Labour won 34 out of Wales’s 40 seats. At the next election Wales will send 32 MPs to Westminster. Labour’s likely maximum will be 29. That’s at least ten more lopped of Labour’s majority.

Third, two sets of boundary changes since 1997—the latest coming into effect later this year—have helped the Conservatives. This is not the product of an establishment conspiracy, but a consequence of steady, long-term population shifts, which have tended to see the electorates of Conservative-inclined shires growing, and those of traditional Labour-voting areas declining. Had the 2019 election been fought on the new boundaries, the Conservatives would have won around ten more seats and a majority of 100.

Those three factors are baked into the next election. The fourth isn’t. The uniform swing calculations assume that the shift from Tory to Labour is much the same in all kinds of seat.  The same kind of projection in 1997 would still have given Tony Blair a big victory, but not as big as he actually achieved. His majority was enhanced by tactical voting. The most detailed statistical analysis[2] estimates that Labour gained an extra twenty seats by attracting Liberal Democrat supporters in Con-Lab marginals, while the Lib Dems attracted enough Labour supporters in their target seats to end up with ten extra MPs.

In 2019, when Jeremy Corbyn and Jo Swinson led their respective parties, there was little tactical voting. But opinion polls, by-elections and local elections from recent months tell a consistent story: with Keir Starmer and Ed Davey, tactical voting is back.

The most dramatic confirmation of this came last June, when the Tories lost two seats on the same day, Wakefield to Labour, and Tiverton & Honiton to the Lib Dems. Both seats saw ferocious tactical voting, which helped to sink the Conservatives.

The graphic shows what would happen if the impact of tactical voting at the next election matches what happened in 1997. This is not so much a firm prediction; more a best-guess illustration of what could happen.

The effect of tactical voting on this scale would be significant:

  • On a uniform swing, the Conservatives need a lead in the popular vote of just three points to win an overall majority. If tactical voting deprives them of 30 seats, they need a seven-point lead.
  • Without tactical voting Labour needs a six point lead in order to become the largest party in Parliament. With tactical voting, a lead of just one point is likely to be enough.
  • For an overall majority, Labour’s target is a useful three points closer: a 13-point lead on a uniform swing, 10 points with tactical voting.

Put another way, tactical voting would go a long way to removing the Conservative bias in Britain’s electoral geography. This reverses the bias towards Labour for some years before 2010.

With or without tactical voting, one feature of our voting system these days is the high likelihood of a hung parliament, in which neither Labour nor the Conservatives win the 326 seats they need for an overall majority. The reason is simple. In the 1950s and 1960s, almost all MPs were Labour or Conservative; there were no hung parliaments. Since 2010, each election has sent at least 70 “minor” party MPs to Westminster: Lib Dem, SNP, Plaid Cymru, Green and from Northern Ireland. As a result, two of the four most recent elections have produced a hung parliament.

With a uniform swing, a hung parliament is the likely outcome of anything between a two-point Conservative lead and a 12-point Labour lead. With tactical voting the numbers are different but the range just as great: from a six-point Tory lead to a nine-point Labour lead.

What then? The Conservatives have started reprising their tune from the 2015 campaign, when they warned of a “coalition of chaos”, if Ed Miliband had to negotiate with the SNP to win power. There is, however a huge difference between then and now. In 2015, the fixed-term parliament Act was in force. This reduced the power of the Prime Minister. They could not call an early election to break a deadlock—for example if they lost a no confidence vote. They would have to resign and hand over to someone else. So, yes, it was perfectly reasonable to suggest that Ed Miliband might end up beholden to the SNP in order to stay in office.

That’s not the position today. Boris Johnson’s administration repealed the fixed-term parliament Act. We are back to the position in which the leader of a minority government could in practice call a fresh election more-or-less whenever they wanted.

In March 1974, Harold Wilson returned to Number Ten when Labour edged ahead of the Conservatives in seats, but remained 17 short of a majority. He received an assurance from Buckingham Palace that if his Queen’s Speech was voted down, he would be granted a dissolution and a fresh election. Edward Heath’s Conservatives were terrified that if they helped to defeat Wilson so soon, voters would punish them severely for provoking—what’s the word?—chaos and give Labour a thumping majority. So the Tories abstained, and carried on sitting on the political sidelines until Wilson called a second election seven months later, when Labour did manage to achieve an overall majority, albeit a narrow one.

The crucial point is that Starmer, leading a minority government, would have the power that Wilson had in 1974, but Miliband would have been denied in 2015. That very power would ensure Tory caution, at least for a while; and with the Tories declining to bring down a new Labour administration, the SNP would have no leverage. This in turn means that Starmer would have no need—again, at least for a while—to seek a coalition.

The news for Labour gets better. Labour would not even need to be the largest party to form a minority government. Think back to 2017, when Theresa May had to win over the DUP to remain in office. Without their support, she would have been unable to secure a majority in the House of Commons, even though the Conservatives won 56 seats more than Labour.

Bearing that in mind, look at the “votes level” line in the graphic. On a uniform swing, the 310 Tory MPs would comfortably outnumber Labour’s 253. But on the generous assumption that the DUP retain eight seats, the total supporting Rishi Sunak’s Queen’s Speech would be 318. Every other party says they want the Tories out of office. Together, their 325 MPs would be able to bring Sunak down (again assuming Sinn Fein’s continuing absence from Parliament). He would have to resign. Indeed, he might well resign ahead of the vote, once his defeat became inevitable—as Edward Heath did in 1974.

So Starmer would become Prime Minister, as the leader of the second largest party. He would be able to survive without any deals with the SNP or Lib Dems, at least for a few months. Like Wilson in 1974, his destination would be a second election, not a coalition.

The precise way this would play out would depend on the numbers. If Labour were comfortably the largest party, it might well be able to carry on for two or three years. If it were just a few seats short of a majority, it might even last a full term with the aid of the Lib Dems—probably, not a formal coalition, but a looser arrangement, such as the Lib-Lab pact that sustained James Callaghan’s government in the late 1970s.

The broader point is this. The great majority of hung parliament scenarios lead to a minority Labour government not needing a coalition. Only if the Conservatives come very close to a majority would they be able to stay in office. And with tactical voting, it is perfectly possible for the Tories to end up with a four-point lead over Labour, and getting on for 1.5 million more votes, and still be ejected from office.

After more than a decade of Tory rule and its impact on the kind of people that he fought for, perhaps Aneurin Bevan would call that poetic justice.

[1] The amount of tactical voting will, of course, vary from seat to seat. But the overall numbers will be close to the figures shown, as long as the number of seats that stay Tory because of local below-average swings broadly match the number of “extra” gains that Labour makes where the swings are above average. Likewise, uniform swing estimates do not need the swings everywhere to be identical, but for the variations around the average to be reasonably symmetrical and so cancel each other out.

2 All percentages in this analysis relate to Great Britain; they exclude Northern Ireland. This is the normal practice among pollsters, whose nationwide surveys reported in the media also exclude Northern Ireland. However, again in line with normal practice, the figures in the text and graphic for seats in Parliament do take account of MPs from Northern Ireland

3 John Curtice and Michael Steed’s appendix to The British General Election of 1997, by David Butler and Dennis Kavanagh, Macmillan, 1997.

This analysis was first published by Prospect