The local elections told us that Labour's lead is smaller than national polls suggest

Labour’s lead over the Conservatives in the BBC’s Projected National Vote (PNV) share was nine points in the local elections, with Keir Starmer’s party on 34% and Rishi Sunak’s on 25%. This is a much smaller lead than estimated by national polls. The most recent YouGov national poll, for example, gave Labour a 27-point lead.

However, local election PNVs are not a particularly useful metric for assessing how the country might vote in a general election. There is very little relationship between the PNV in the set of local elections preceding a general election and the subsequent result. Our analysis of the last 10 general elections shows no provable relationship between local election PNV lead and final popular vote margin.

That being said, we think that the PNV is likely to be much closer to Labour’s ‘true’ lead, albeit for different reasons. Current national vote intention polls are ‘nowcasts’ — glassy snapshots of current opinion — rather than forecasts of actual voting behaviour in a general election. Many assumptions that pollsters make to turn their polls into forecasts, including how they treat ‘don’t knows’ and handle turnout effects, could be playing up Labour’s lead. In the month leading up to the local elections for example, different pollster methodologies accounted for a full four percentage point difference in Labour’s vote share.

To get a better view of Labour’s ‘true’ lead, we took the local election results as our baseline and compared them to historic data on local-to-general election voting behaviour from the British Election Study (inspired by the work of Dylan Difford). For the seats where we did not have a complete picture from the locals, we used the results from the latest public MRP polls.

As you can see from the table above, a significant chunk of Liberal Democrat and Green local election voters said they intended to vote for Labour in a general election. We thus allocated 30% of Lib Dem voters and 34% of Greens to the Labour column in our projection.

This locals-to-general election transition model produced a Labour lead of 12 points in Great Britain, with the party on course for a majority of around 140 seats (394 seats for Labour vs. 160 for the Conservatives). At the national level, our projection puts Labour on 38%, the Conservatives on 26%, the Lib Dems on 13%, Reform on 7% and the Greens on 5%. We think these results are closer to the expected party vote shares come the general election, with a rise in the aggregate vote share for the two main parties at the expense of the Greens and Reform.

Note that while the current vote share gap in national opinion polls is larger than Labour's 'true' lead, we expect Labour's lead in the polls to shrink, and the final result to fall in line with this 12-point estimate. Using PollBasePro, an archive of British opinion polling since 1955, we have built a simple model to predict the government and opposition vote shares. The model is based on three inputs: current polling averages, the progress through the parliamentary term and the number of years the government has been in power.

Applying this model to current polling averages and an a hypothetical election date in December, the model produces a Labour general election of 10 points -- much smaller than the 20+ point gaps in current polls. On a uniform swing, a 10-point lead would result in a hung parliament. Yet Labour's remarkable vote efficiency (a measure of how well-distributed voters are) means it actually needs to achieve a much lower vote share lead to win a large majority.

Labour's vote is increasingly efficient

The second important story that emerged from the local elections relates to vote efficiency. If we take our local-to-general election projection and group Conservative- and Labour-held seats based on their partisanship compared to the national average, we find that the Conservative-to-Labour swing is much higher in Conservative-held seats. Labour are winning votes exactly in the places it needs to form a majority government.

To examine vote efficiency in more detail, we started with Colin Rallings and Michael Thrasher’s estimates of the notional results for the 2019 general election on new constituency boundaries. All seats not held by the Conservatives or Labour were removed and each constituency was adjusted via a uniform national swing to estimate the outcome if the national popular vote was tied between the two main parties. Seats with a more than 25-point gap between the two parties were designated as ‘safe’ and those with a 10–25 point gap were designated as ‘lean’, with all others marked as battleground seats. As you can see below, Labour appears to be winning over voters exactly where it needs to, with proportionally larger swings projected in the safest Conservative seats.

An important side effect of better-distributed voters will be a huge increase in the number of marginal seats at the next election. The chart below shows all seats ordered by marginality in 2019 and our 2024 locals projection. The number of seats with a winning margin smaller than five points has more than trebled, from 56 in 2019 to an estimated 175 in 2024. While we expect some of this tight distribution to be an artifact of MRP modelling and how they estimate vote shares, there is no doubt that this election will see seats in play that have not been in more than twenty years.

The implication of this is that is that any movement in the polls between now and the election is likely to have an outsized effect on the number of seats won by each party. While we project a sizeable majority for Labour, the window of possible outcomes is wide.

What does Labour's lead need to be?

As a result of this increased vote efficiency, it is clear that Labour will not need a double-figure popular-vote lead for a majority based on the electoral geography of the local elections. But what the Conservative-Labour gap needs to be for Labour to win has been a point of extreme disagreement amongst analysts and commentators; and with good cause. The popular-vote lead needed for a majority — for either party — has changed dramatically over the last 45 years.

So much could be written about the graph above. Before going into more detail, I wanted to outline how you should read it. Looking at the 1992 election, for example, we calculate that in Great Britain (i.e. excluding Northern Ireland) Labour needed a popular-vote lead of only 0.6 points to govern (which is what the polls were showing), but the Conservatives needed a large 6.7-point voting gap. By contrast, in 2001, Labour could have governed the country with a majority despite being three points behind the Conservatives. The Blair years saw some of the most unequal, unrepresentative governments to have been elected in Britain, on increasingly depressed turnout levels.

The story of Labour's current vote efficiency is a complex scheme based on three or four main factors. There are the factors that converged to dramatic effect during the Blair years, although in many ways they preceded them.

1. Labour maintaining a decent vote share for suburban voters, non-graduates and C2 voters (see Ipsos data here). These voters are distributed very advantageously in historically marginal seats. Also critical is a constrained age curve (Gordon Brown won 31% of over 65s vs 17% in Jeremy Corbyn’s 2019 outing).
   • The Conservatives potentially losing a substantial number of votes amongst the old, the Leave voting, the non-university educated, and the rural could crater voting efficiency at the next election.
2. Inroads in Scotland — specifically the block of Central Belt seats Labour held — effectively lowered Labour’s required vote gap by 2–4 points.
3. Liberal Democrats taking 30–40 seats off the Conservatives in seats that were otherwise demographically-Conservative raised the bar for Conservatives to govern by another 2–4 points.
4. Tactical voting, and whether parties of the centre-left disliked the Conservatives more than they disliked each other.

On tactical voting in particular, there appears to be much more Labour/Liberal Democrat tactical voting under Keir Starmer's leadership than there was under Jeremy Corbyn.  This is understandable when we consider that 2019 Lib Dem voters rated Starmer an average of 5.4 out of 10 in the most recent British Election Study survey, versus 2.6 for Corbyn – the highest gap between the two men of any party. In fact, 2019 Liberal Democrats are basically indistinguishable from 2019 Labour voters in their opinions on the current Labour leader (the latter rated Starmer an average of 5.5 out of 10).

The exhibit below shows the level of efficiency of the Labour/Liberal Democrat vote share in the same set of unchanged local government wards won by the Conservatives in 2019. From our analysis, there is an increasing pattern of the two parties’ voters aligning themselves with the strongest anti-Conservative challenger in subsequent elections. This pattern has also emerged in by-elections over the course of this parliament, and would reduce Labour’s majority threshold even further if replicated at a general election.

Despite this highly-optimistic outlook for Labour, we must declare a word of warning: as mentioned, the window of probable electoral outcomes is large, and is all the dynamics which are likely to propel Labour to a sizeable majority could become unstuck very quickly. Any combination of the Conservatives appealing to cultural Conservatives and older voters again, an SNP recovery (roadmapped by a 2026 Holyrood election), and Southern free-market Liberal Democrat voters unable to cope with a period of high-tax Labour government, could cause significant problems for Keir Starmer. In today’s day and age, big majorities assembled at warp speed seem to look more like sandcastles than skyscrapers.