Throw the rascals out!

Evidence of anti-Conservative voting in the 2023 local elections

“When sorrows come, they come not single spies but in battalions”. For Conservatives, the 2023 local elections brought battalions of sorrow indeed. The headline results were dire, but detailed analysis brings even worse news. As we saw last week, there is evidence that the Brexit realignment which drove Conservative success in 2019 is now beginning to unravel. This week, I show how anti-Conservative voting magnified Conservative losses in 2023. The Conservatives did worst where they were defending seats. They lost most ground where they started strongest. And, in the last year, all of the Tories’ opponents did best when they started as the local opposition. These are damaging effects in isolation. Taken together, they could be devastating. An anti-incumbent mood is rising.

The 2019 locals, when the seats contested this year were last up, may not provide a good baseline for assessing anti-incumbent voting because this was also an anti-politics local election cycle. Voters turned against both main parties in 2019, fought during the middle of the Brexit deadlock at the end of Theresa May’s premiership. Many of these wards have changed hands since, in 2021 and 2022 local elections which saw a lot of change, as the Brexit shifts seen in 2019 worked they way into local voting patterns. Voters looking to oust an unpopular incumbent may also be more likely to recall the most recent results, encouraged by local challengers who will emphasise recent strength in campaign materials. For all these reasons, it makes sense to look at two baselines: 2019, when the councillors fighting for re-election this year were last elected, and 2022, the most recent local contests which will be freshest in voter memories and often most prominent in opposition parties’ campaign materials.

We start by looking at whether the Conservatives did worse when defending a seat. Here are the changes in vote shares for Conservatives and Labour in seats won by each party in 2019:

Conservative and Labour performance 2019-23 by previous winning party

Source: BBC Key Wards dataset (all charts are based on wards in the dataset where Con, Lab, LD and at least one other party stood in both elections)

The anti-Conservative effect really leaps out here: the Tory vote share slumped in Conservative held seats, falling by nearly 6 points, while it rose in seats won by anyone else. Labour were also up substantially more in Tory held seats than they were anywhere else.

This is even worse than it looks for the Conservatives, because 2019 was already a really awful set of local elections for them, with over 1,300 council seats lost. The party might have hoped that incumbents strong enough to survive the previous cull would hold up better this time. Instead, it seems voters were determined to drive out Conservative incumbents wherever they remained.

However, a lot has changed since 2019, which was before Boris Johnson’s “Get Brexit Done” campaign realigned voting patterns (as discussed in last week’s post). Perhaps what looks like anti-incumbent voting is instead an artifact of these broader shifts ? If so, the pattern should weaken if we shift our baseline to last year’s local elections, when the Brexit alignment was already in place. On the other hand, voters might focus (mistakenly) on the most recent results when identifying the incumbents they want to remove.

Conservative and Labour performance 2022-23 by previous winning party

Source: BBC Key Wards dataset (all wards where Con, Lab, LD and at least one other party stood in both elections)

The anti-incumbent pattern does not weaken much when we shift our focus to change on last year. The Conservative share is down by 5.6 points in the seats where Tory councillors won election last year, nearly three points worse than the party’s showing in Labour won seats, and over 4 points worse than its showing in Lib Dem won seats. Labour's vote share is up substantially on last year in seats the Conservatives won in 2022, but flat in Lib Dem won seats, and slightly down in seats Labour or the Greens won last year.

The story is thus the same in both years: much larger swings away from the Conservatives in Tory won wards. Unless we are to believe that Conservative councillors across England have simultaneously angered their constituents, this pattern suggests an electorate looking to take out their anger at national Conservative incumbents on unfortunate local Conservative incumbents.

However, there is a second possibility. An anti-incumbent vote pattern can also emerge because a party is losing more support where it starts strongest - known as proportional swing.¹ This is an unusual pattern. Normally, we have uniform swing, or something close to it, with the electoral tide rising and falling by similar amounts from one seat to the next. Proportional swings were most recently seen, twice, in the 2015 election: against the Liberal Democrats in 2015 at the end of the Coalition and against Scottish Labour following the SNP’s post-independence surge. As these two examples illustrate, a strong proportional swing against a party can be dire news indeed in a first past the post election. Even big majorities offer little security when the pattern is “the more you have, the more you lose”.

Conservative and Labour performance 2019-23, and Conservative to Labour swing, by Conservative strength in 2019

Source: BBC Key Wards dataset (all wards where Con, Lab, LD and at least one other party stood in both elections)

Here is some more bad news for the Conservatives - the swing against them is indeed proportional. The better the Tories did in 2019, the worse they did in 2023. The differences are quite stark - the Conservative vote was actually up in wards where the party got 30% or less in 2019, though with Labour up more there was still a modest swing to the opposition. But in seats where the Conservatives started on 30-40% they were down nearly three points, with Labour up nearly 6 points - a 5 point swing. But the big collapse is in the true Tory strongholds: in wards where the Conservative vote won over 40% in 2019, they were down an average of over 7 points, while Labour were up over 8 points, producing an 8 point swing to the opposition. Again, we may wonder if this pattern is an artifact of the unusual 2019 baseline and the turbulence since. And again we can check this by looking at changes since last year:

Conservative and Labour performance 2022-23, and Conservative to Labour swing, by Conservative strength in 2022

Source: BBC Key Wards dataset (all wards where Con, Lab, LD and at least one other party stood in both elections)

The story is the same - the better the Conservative starting point in 2022, the bigger the drop in their vote, and the bigger the swing to Labour. Indeed, the pattern is even more clear cut over the past year - with each move up the scale of starting strength, Tory performance gets worse and Labour performance gets better. The Conservative to Labour swing is zero in the weakest wards for the Conservatives, two points in the middle of the range, and 4.6 points in the Tories 2022 strongholds.

Bigger swings against Tory incumbents, and bigger swings in Tory strongholds paint a bleak picture already. But there is still a third way in which anti-incumbent sentiment could manifest. Voters motivated to oust Conservatives may seek to coalesce behind whichever challenger is best placed locally. If this kind of tactical voting is going on, we should expect each opposition party - Labour, Lib Dems and Greens - to do best in the wards where they came second to the Conservatives last time. “Last time” may again mean 2019, when councillors up for re-election were last elected, or 2022, when many seats last held a contest. Either or both may feature on party leaflets and influence voters’ calculations.

Change in Con, Lab, LD and Green vote shares since 2019 by 2019 winning party and second party

Source: BBC Key Wards dataset (all wards where Con, Lab, LD and at least one other party stood in both elections)

The 2019-2023 figures are mixed. In yet further bad news for the Conservatives, Labour do post better perfomrances in wards where they start second - gaining an average of nearly 10 points, compared to an average of seven points overall. But, a shred of comfort for Conservatives, there isn’t any evidence here of the Lib Dems or Greens doing better in 2023 in Tory wards where they were the local challenger on the basis of 2019 results. Quite the opposite in fact - both parties do worse than average in wards where they started second. And the Lib Dems do best in wards where the Greens came second in 2019, while the Greens do best in wards where the Lib Dems were second, which is hardly evidence of great tactical co-ordination.

However, the big recent shifts in local elections mean the local challenger in many wards was different in the most recent election to when these seats were last up four years ago. Voters looking to vote tactically, and parties looking to encourage them, may rationally focus on these more recent results as a better indicator of the local balance of strength. And, when we look at the pattern of vote changes since 2022 in Tory wards, a rather different picture emerges:

Change in Con, Lab, LD and Green vote shares since 2022 by 2022 winning party and second party

Source: BBC Key Wards dataset (all wards where Con, Lab, LD and at least one other party stood in both elections)

This looks more ominous for the Conservatives. Each challenger party has done better in wards where they start second to the Tories in 2022. Labour do 3 points better in Con vs Lab wards; the Lib Dems also do 3 points better in Con vs Lib Dem wards, and the Greens do 7 points better in Con vs Green wards (this last figure is based on a very small sample though, so should be treated with caution). This is exactly the kind of pattern we would expect if local voters were tactically coalescing behind the strongest anti-Tory challenger.

Time for a change?

One secret to Tory success in 2019 was geography - the Conservatives’ vote was better distributed across seats than their opponents’, enabling a very efficient conversion of votes to seats. What geography gives, geography can take away.² Anti-incumbent voting means the Conservatives lose more support where they are defending seats. Proportional swing means they lose more ground where they start strong. And anti-Tory tactical co-ordination mean the Tories’ rivals do best where they are most threatening. Each of these effects can generate bigger seat losses on the same vote loss. And in the 2023 locals, there is clear evidence for two of them and at least some evidence for the third as well.

These changes may also reflect a deeper shift in the public mood. Four words recur across studies of elections when incumbents are defeated: “throw the rascals out!” Once an anti-incumbent mood takes hold, survival for any government is difficult. And such a mood certainly looks in evidence now - in polling, in by-elections and this month in local elections too. If such anti-incumbent sentiment manifests in a similar way next year, the political map could be reshaped to a greater extent than polling suggests. The Conservatives need to find a way to turn things around with voters, and quickly, if they want to avoid a long and dark election night next year.

Edited 9.30 am 18th May - graph labelling changed for greater clarity and again 9.30am 19th May to correct typos

1. A proportional swing pattern will inevitably produce larger swings in seats the party previously won as these are, amost by definition, places where its vote is relatively high.

2. For more on the implications of shifts in electoral geography, check out this excellent piece by Stephen Fisher: https://www.prospectmagazine.co.uk/politics/61357/keir-starmer-labour-party-electoral-geography-polling-locals