What does Labour’s victory in the Wakefield by-election tell us about “Red Wall” seats more generally, and about Labour’s prospects in the next General Election?
1. Where We’re Coming From
Drawing lessons for Labour from previous elections is horrifically difficult, for (at least) four reasons. The first is the fact that the whole field is bitterly contested, so that discussions are likely to be distorted by partisanship. There are some (among whom I’m sure I wouldn’t count you, dear reader) who turn a deaf ear to any attempt to identify positives about the Corbyn period, typically pointing out in a discussion-ending tone that Labour didn’t win in either 2017 or 2019, whereas we did win in 1997, 2001 and 2005. (Good to know. I’ll make a note.) On the other hand (secondly), 2017 shouldn’t be taken as a model, either: some elements of the 2017 campaign went well and could usefully be repeated, but other elements we could do without. Polarisation, for example. Opponents of Corbyn will sometimes say that the only reason Labour did so well in 2017 was that Theresa May’s Conservatives were so weak, but if anything the opposite is the case. The only reason Labour didn’t do even better (and, e.g., win) was that the Conservative result was so strong: in 2017 the Conservative Party got its highest vote share since 1987, and took more votes than Labour did in their landslide victory of 1997. So it was a big success for mobilisation of the Labour vote; another time we just need to work out how to get our voters to turn out and theirs to stay at home. Demobilisation of the other side seems more like a Blair-era achievement, but – thirdly – if the Corbyn period isn’t to be discounted, the Blair victories shouldn’t simply be celebrated. You’d think it’d be reasonably easy to draw lessons from the most recent period when Labour did win general elections – three in a row, I’m reliably informed – but examined in detail the New Labour electoral record is rather qualified, not to say flukey. If we want to improve on 2017’s “40% of the vote and a hung parliament”, saying that we should be pitching for 2005’s “35% and a majority of 66” isn’t very helpful. And that leads on to the fourth reason – but I’ll save that for later. (It’s a corker.)
Here, anyway, is how the last few elections look, UK-wide.
Dark blue: far Right; purple: UKIP/BXP; green: Green/independent Left. %s shown when 3% or above.
Red line: Lab – Con vote %, right-hand axis.
We can see some trends here.
2010: Labour vote drops by 6% in the wake of the 2008 crash; some of those votes go to UKIP, the far Right and the Lib Dems (“I agree with Nick”), but most go to the Tories. Hung Parliament leading to Tory/Lib Dem coalition.
2015: Lib Dem vote drops by 15% after the Coalition; Labour and Tories pick up some of their vote and the Greens pick up more, but by far the biggest beneficiary is UKIP. Given the lack of common ground between the Lib Dem and UKIP manifestos, it seems that quite a lot of people had been voting Lib Dem on general “sod the lot of ’em” principles. Either that or the Tories have gained the adherence of a lot of former Lib Dems, while losing an equally large number of their own voters to UKIP. In any case, the result is a Tory government.
2017: The previous election’s Green and – more dramatically – UKIP votes vanish like melting snow in a heavily polarised election. The Tories’ vote share rises by 5.5%; Labour’s share rises by 9.6%. Tory minority government.
2019: Almost 8% of that Labour increase disappears again, to the benefit of the Lib Dems, the Greens, the Brexit Party and the Tories. Between people who didn’t trust Labour to deliver a second referendum, people who didn’t trust Labour to deliver Brexit and people who thought that if they wished hard enough they could vote for a hung parliament, Labour’s vote got driven down in many different places, for many different reasons and entirely to the benefit of the Conservative Party. Tory government.
2. The Fall of the Not Actually A Wall Actually
So that’s the baseline. Were things different in the Red Wall, though?
Hmm. [FX: Grits teeth] Red Wall, is it? I find it hard to believe how firmly that term’s ensconced in the political lexicon, having being coined less than three years ago and repeatedly redefined since then. But it looks as if we’re stuck with it now. In practice it means a whole variety of things, but usually when people talk about the “Red Wall” they’re talking about long-term Labour seats that went to the Conservatives in 2019, generally in post-industrial or semi-rural areas, generally in England but north of Luton, and which are thought to have gone to the Tories for primarily ‘cultural’ reasons. Either that, or seats that Labour still hold but which qualify on the other criteria and might turn out to be vulnerable next time round – a decidedly malleable definition.
Even if we confine ourselves to seats Labour actually lost, though, there are problems. What do we mean by a “long-term” Labour seat – one that’s been Labour since 2010, say? (Twelve years is a long time for most people.) Should we look for continuous holds since 1997? Since 1983? Longer? And what do we do about constituencies whose boundaries have been redrawn in that time? (The longer we go back the more likely that is to have happened, making “long-term” Labour seats harder to identify than you might think.) As for seats “going to the Conservatives”, surely we want to focus on seats where there have been substantial changes in absolute and relative vote share, not just a change in who tops the list. A seat that’s 39% Labour and 38% Tory at one election and 38%/39% at the next has “gone to the Conservatives”, after all, but hardly anyone there has actually changed their mind. The picture’s even more complicated when a seat has a strong Nationalist presence, and show me a Scottish seat that hasn’t; even Plaid Cymru’s ~10% presence in Wales makes them a bit of a wild card. So for simplicity we really need to confine ourselves to England.
With all of this in mind, in an earlier post I looked at all Labour’s losses in England in 2019 and categorised them as
- big wins (Labour continuously since 1979, decisive win in 2019)
- narrow wins (Labour continuously since 1979, but not a decisive win in 2019)
- marginals (had changed hands at least twice since 1979)
I defined a “decisive win” as one where
the Labour vote in 2019 fell 10% or more relative to 2017 and was lower than at any time since (and including) 2001, and where the eventual Tory majority was 5% or more
I then took some averages of the three groups, which worked out like this:
| 1 | 2 | 3 | Diff 1 | Diff 2 | |
| Lowest Labour vote %, 2019 | 24.4 | 39.3 | 36.9 | 12.5 | 2.4 |
| Highest Labour vote %, 2019 | 39.8 | 44.5 | 46 | 4.7 | 1.5 |
| Average Labour vote %, 2019 | 34.8 | 41.7 | 41.1 | 6.3 | 0.4 |
| Biggest Tory majority in %, 2019 | 31.4 | 12.6 | 15.7 | 15.7 | 3.1 |
| Biggest drop in Labour vote %, 2017-19 | 24.9 | 15.7 | 18.1 | 6.8 | 2.4 |
Diff 1: difference between 1 and either 2 or 3, whichever is closer
Diff 2: difference between 2 and 3
Essentially there’s not a huge amount of difference between the ‘marginals’ and the ‘narrow wins’ (Diff 2), whereas they’re both quite distinct from the ‘big wins’ (Diff 1). Which makes sense: even in a seat that’s been held by one party for a long time, you can score a narrow win by putting on a few hundred votes and driving the other side’s vote down by 5-6%. That – like a marginal changing hands – is the kind of thing that happens in any good election campaign; and the Tories in 2019 had a very good election campaign.
The big wins, though, really are different. So that’s my first finding, in terms of the reality of the Red Wall: yes, there was a group of long-term Labour seats that fell to the Tories in 2019 by really big margins, with really big drops in the Labour vote. And yes, they were Up North, if you define that term loosely enough (Derby and Durham are both in the North, but you wouldn’t want to walk it). There were only about 16 of them, though, meaning that if Labour had held every one of them and lost all the other seats they lost, the Tories would still have gained a big majority.
(Wakefield wasn’t in this group, although arguably it’s on the borderline: the Labour vote in 2019 fell by 9.9% relative to 2017, and Mary Creagh’s losing vote share of 39.8% was higher than her winning share of 39.3% in 2010.)
The analysis of big Tory wins ended up overlapping with something I wrote some time ago, analysing trends in Labour vote share in General Elections since 2005 in Labour seats in England. Having heard some Labour MPs bemoaning Corbyn’s leadership as electoral poison, and having seen Labour surge pretty well everywhere in 2017, I wanted to know whether there was any truth to that negative perception even at a local level. Was there a substantial group of seats where Labour’s vote share, relative to the Tories, had actually fallen in 2017? (Clearly there were some, as we lost six seats (while gaining 36) – but a handful isn’t really enough to generalise from; weird things do happen in elections.)
I compared 186 English Labour seats (as of the 2017 election) across the last five elections. The metric was simple: if we compared this election’s comparative vote share – “Labour % – Tory %” – with the previous election, had it gone Up or Down? I identified 50 seats where comparative vote share had in fact gone down in 2017, in all of which it had also gone down in 2010. In 39 of the 50, relative vote share had gone up in 2015 (so Down, Up, Down or DUD); in the other 11, vote share had gone down at all three elections (=DDD).
Remember those “big wins”? On inspection, almost all of them fitted the DUD pattern, and the remainder were all DDD. So there genuinely was a group of long-term Labour seats – a minority, but a substantial minority; around one in five – where the election campaign led by Ed Miliband seemed to deliver the goods in a way that 2017 didn’t; where the end result of that amazing May of campaigning was a sitting MP holding on with a smaller majority. And a lot of those seats were vulnerable, given a hearty shove in 2019, to going Tory in a big way.
But how?
3. None Of The Above: The True Story
Here’s what happened in one of the “big win” seats, Don Valley. (More detail in this post.)
2010: Nationally, the Labour vote drops by 6%, about half of which goes to UKIP or the BNP; here, the drop is double that, and it all seems to go to anti-political candidates of the Right and far Right: UKIP, the BNP and the English Democrats.
2015: Nationally, a big drop in the Lib Dem vote and a rise in UKIP’s. Here, again, it’s the same but much more so: the Lib Dem vote doesn’t drop so much as collapse; Labour picks up some Lib Dem votes, but UKIP takes a big chunk from the Lib Dems as well as absorbing the far Right and taking some from the Tories, finishing in a close third place. (It’s possible, as suggested above, that there were two big vote flows rather than one – Lib Dem to Tory, Tory to UKIP – but in this seat the numbers involved make it doubtful: more than half of the people who voted Tory in 2010 would have had to switch to UKIP and be replaced by former Lib Dem voters.) The division of the Right vote leaves the Tories looking substantially weaker; if all you were focusing on was the comparison between the winning Labour vote and the Tory runner-up, it would look like a big improvement.
2017: Nationally, a big shift to two-party polarisation, with a substantial UKIP vote collapsing and Labour’s vote going up about twice as much as the Tories’. Here, the UKIP vote in 2015 was only a couple of points shy of the Tories’, and it disappears completely in 2017 – no candidate on the ballot paper. Labour’s vote goes up, but not twice as much as the Tories’ – it’s the other way around. Labour’s vote increases by nearly 7%, but its majority falls by nearly 10%.
2019: Nationally, Labour’s vote drops by 8% with compensatory gains by multiple different parties; here, the drop is closer to 18% and the main recipient of Labour votes is the Brexit Party – although it’s the Tories who take the seat.
A few conclusions, quoted from that earlier post.
- There’s a substantial anti-political, “none of the above” vote in this seat: 10-20% of the vote at every election since 2001
- The Lib Dems profited from this, until they didn’t: joining the government was the kiss of death, and the Lib Dems have effectively been irrelevant (at least in this seat) since 2010
- Parties of the Right and extreme Right are legitimate in this seat as a repository for anti-political votes; the strength of the BNP and ED vote in 2010, and the extent to which UKIP built on this, is not to be underestimated
- UKIP/BXP is strategically ambivalent, operating as a pure protest vote (2010, 2019), as a more respectable alternative to the far Right (2015) and as an ante-chamber to voting Conservative (2015, 2019?)
- Without Corbyn, the 2017 result would have been much worse for Labour: Labour’s acceptance of Brexit and Corbyn’s image as an anti-system outsider both prevented the 2015 UKIP vote transferring to the Tories en masse
- However, 2017 looked worse than 2015 for the sitting MP (Caroline Flint), as – with higher Labour and Tory votes – Labour won by a much narrower margin; this supported the narrative that a decline in Labour’s vote had continued or even accelerated under Corbyn (whereas in reality it had begun to be reversed)
- What lost the seat in 2019 was the strength of the anti-system vote up to 2015, the Tories’ success in converting 2015’s UKIP protest vote to 2017 Conservatives, and Labour’s inability in 2019 to pre-empt the appeal of the BXP protest vote, due to the ambiguity of Labour’s Brexit positioning and the tarnishing of Corbyn’s image.
Wakefield? Here you go:
The angles are shallower but the shape of the line’s the same: Wakefield’s a DUD. It all looks very much the same, albeit on a smaller scale – Wakefield wasn’t a big loss in 2019, and it hadn’t ever really been weigh-the-vote territory before then (it was a bona fide marginal back in the 1980s, albeit one that the Tories never actually took). Still, the key indicators are all there. The drop in the Labour vote in 2010; the far Right presence (the BNP saved their deposit in 2010); the Lib Dem vote collapsing in 2015, with UKIP the main beneficiary (directly or indirectly); the uneven redistribution of the UKIP vote in 2017, leading to a larger absolute Labour vote share with a smaller majority; and then the hoovering up of disaffected Labour votes by the Lib Dems and the Brexit Party, to the benefit of the Tories.
2010: disaffection with Labour, legitimation of anti-system parties. 2015: disaffection with the Lib Dems, rise of UKIP. 2017: collapse of UKIP, diversion of some anti-system votes to Labour under Corbyn, conversion of most anti-system votes to the Tories. Then in 2019 we’re back to disaffection with Labour and legitimation of anti-system parties, this time with some diversion of anti-system votes to the Tories under Boris Johnson (portrayed – absurdly but effectively – as an anti-system figure). It was blowing much harder in the Don Valley, but the same wind was blowing in Wakefield.
Some of these trends were operating nationally to a greater or lesser extent, admittedly. Labour’s vote share fell across the country between 2005 and 2010, having also fallen between 2001 and 2005; indeed, considering that turnout was lower across the board, Labour’s vote in 2010 was two million down on its level in 2001 – and that was nearly three million down on 1997. (And that epoch-making result was itself substantially below the 55%+ that Labour had been polling when the election was called. 55%!)
The key difference in this sub-group of seats was what happened in 2017, where the national trend was for Labour’s vote to increase substantially more than the Tories’ relative to 2015. The seats where the Tories picked up more votes than Labour in 2017 seem also to show some relative rather than absolute differences from the national trend: a bigger fall in the Labour vote between 2005 and 2010; a bigger rise in the far Right vote in 2010; a bigger fall (a more complete collapse) of the Lib Dem vote in 2015 and a bigger rise in the UKIP vote.
4. What’s Really Happening?
We’ve reached the point where we can identify the three key driving forces of the last election result for Labour. The first – which is arguably one of the key driving forces of the last twenty years of politics – looks back to those five million voters Labour lost between the two victories of 1997 and 2005, and all the people who thought like them. That’s a big bloc of actual and potential Labour voters shaken loose from the party. Over time many, perhaps most, will have got the habit of staying at home at election time, but many will have taken the opportunity in 2005, 10, 15 to cast an anti-system, “sod the lot of ’em” vote – for the Lib Dems, for the BNP, for UKIP. (The British public also had an opportunity to cast an anti-system vote in 2016, of course. How’s that working out?) Roll that bloc of voters forward to 2017 and 2019, and the only question is how many of those people will be open to Corbyn’s anti-establishment appeal, and how many will prefer their anti-system politics with a definite Right-wing stamp.
Which brings us to the “Red Wall” question – why did some Labour seats go particularly heavily for the Tories? – and here we see the second key factor at work in 2019: the greater strength, and legitimacy, of the far Right in some areas than others. This is a factor that feeds through from 2005 on, first in higher votes for far Right parties, then in substantially higher votes for UKIP – and then, in 2017, in blunting the edge of Corbyn’s “anti-establishment” appeal. Local factors will be involved here; I can’t put the blame on Tony Blair, as much as I’d like to. What I will say, though, is that the BNP got more than 8% of the vote in eight constituencies in 2005 and twelve in 2010, including five in both – and they were all Labour seats. If I ran the Labour Party, having a fascist party even retain its deposit in a Labour constituency would be grounds for deselection; it doesn’t exactly suggest an assertive local party. To be fair, Labour didn’t go on to lose all of those seats – just a little over half, eight out of the fifteen – so it’s not an exact science. Anyway, we couldn’t go around holding the threat of deselection over good socialists like Yvette Cooper, Hilary Benn, Margaret Hodge and checks notes Ian Austin, could we? That would never do.
Have I looked at the data? You bet I’ve looked at the data. Here’s “total far Right vote in 2010” vs “rise in UKIP vote between 2010 and 2015”, Labour constituencies only.
(I tried it without the zeroes; very little difference.) And here’s that rise in the UKIP vote, 2010-15, vs the total Labour vote in 2019.
I don’t bandy R-squared values around a lot, but those look pretty chunky to me.
(Interesting point for anyone else who wants to run some numbers: the relationship between the changes in Lib Dem and UKIP votes between 2010 and 2015 did not look like these charts. It’s actually a (very weak) direct relationship: a big drop in the Lib Dem vote makes a big rise in the UKIP vote slightly less likely – or rather, knowing that the Lib Dems did really badly in a seat makes it slightly less likely to be a seat where UKIP did really well. Three-way traffic, I guess, or something more complicated than that.)
Are those factors really relevant to 2019, though? 2019 wasn’t 2017, and Jeremy Corbyn’s anti-establishment appeal was wearing pretty thin by election night. Actually that just makes those two factors more effective: if you didn’t want your anti-system radicalism with a Glastonbury ribbon on top, you certainly weren’t going to want it wrapped in allegations of antisemitism. Which is to say, those two factors applied along with a third: an anti-Labour campaign of a scale, viciousness and shameless mendacity that I’ve never seen before in British politics (and I remember Bermondsey), much of it emanating from people Labour should have been able to count as supporters or allies. (Or, y’know, as members, officers, elected representatives – people like that.)
The significance of this last point, in the present context, is what it implies for Wakefield, and for all the other seats Labour lost which didn’t tick all of the ‘Red Wall’ criteria I’ve set out. Past voting figures suggest that Wakefield, like many other constituencies, didn’t have that big a deracinated ex-Labour bloc from the mid-2000s on, and nor did it have that big a far Right bloc later. So, while those first two factors operated to depress the scale of Labour’s vote gain in 2017 – and depress Labour’s majority – the effect wasn’t that big (Labour’s majority fell by 1.4%). What really made the difference was that third factor – which hopefully won’t be a factor at the next election. Even without the Tories conveniently falling apart as they did at the recent by-election, a relatively normal election campaign for Labour – without the relentless negative campaigning of 2019, other than from the Tories – should be enough to keep Wakefield in the fold, and win back quite a few of 2019’s other losses.
5. The Fourth Reason
So that’s nice. Whether that will be enough to win a general election, though, is another question, and it’s a question to which the answer is No. This brings us back to the fourth reason why it’s hard to draw lessons for Labour from past elections, as promised earlier: it’s a fact that everyone knows but that tends to be quietly ignored. The fact is that Labour, by and large, doesn’t win general elections. Eleven Labour governments have been formed after an election, the first being Ramsay MacDonald’s first government in 1923. Eleven in a century doesn’t sound that bad, even if five of the eleven only added up to seven years between them. But consider: three of the eleven were minority governments (1923, 1929, 1974/Feb), and another three had majorities of five of less (1950, 1964, 1974/Oct). Of the remaining five, three were second- or third-term governments (1966, 2001, 2005) – which is to say that Labour went into the election with all the advantages of an incumbent government. That only leaves two – and one of those, the 1945 landslide, saw several of the leading figures of the wartime coalition remaining in office (Churchill’s Deputy PM Clement Attlee not least), meaning that it too was effectively a second-term government.
So that’s one Labour government, with a workable majority, formed after an election won by the Labour Party in opposition: one out of eleven; one in a century; one, not to put too fine a point on it, ever. Mr Tony Blair, come on down! 1997 wasn’t just unusual for Labour, it was – literally – unique. It follows that when people talk about Labour winning the next election, they’re not talking about competence and putting the grownups in charge and getting back to business as usual – at least, they certainly shouldn’t be, because that’s not going to do it. And when they talk about 2017 as if the only salient fact about that election was that Labour didn’t form the next government – as if a solid majority over the Tories was there for the taking by any half-decent Labour leadership, so that any campaign that failed to win outright deserved only contempt – they’re talking “absolute tripe”, as Michael Foot and my father used (separately) to say. (Born the same year, those two. Inter-war slang, I guess.)
New Labour’s victory in 1997 had many parents. The property crash of 1992 dislodged many natural Conservative voters from their party; standing at 45% in mid-1992, the Tories’ polling average had dropped below 40% by the end of the year and below 30% by the end of 1993, only beginning to recover in the months before the election was called in 1997. This fed through into both vote-switching and abstention: the Tory vote was down 4.5 million in 1997 relative to 1992, while Labour’s vote only increased by two million. (Yes, ‘only’. We put on 3.5 million between 2015 and 2017.)
Labour was polling 15-20 points ahead of the Tories when John Smith died in 1994. A new-broom leadership was meat and drink to political commentators, who seemed equally dazzled by Tony Blair’s personal style and his bite-sized political philosophy. Taking no chances, Blair and Brown were in any case assiduous in getting the media (as well as the City) on side. Meanwhile the “cash for questions” scandal was undermining the credibility of the Conservative government even further, enabling political commentators to pick up and amplify popular disaffection from the Tory Party under the general heading of “sleaze”.
As for the other parties, the Lib Dems were tacitly operating as New Labour’s small-L liberal outriders, in the (forlorn) hope that their contribution to victory would be rewarded. (Don’t get fooled again, eh?) Scotland was in the bag before campaigning began: the SNP finished as the third largest party in Scotland in 1997, just as they had in 1992. On the Right, the Referendum Party stood 547 candidates and took 2.6% of the vote; UKIP (est. 1993) stood 193 candidates and got 0.3%. (The only UKIP candidate not to lose their deposit was a Mr N. Farage; wonder what happened to him.)
You get the idea. 1997 really was a perfect storm: a confluence of multiple wildly different factors, some of them not of Labour’s doing, some of them downright impossible to recreate now. (O my 20% Lib Dems and my 20% (in Scotland) SNP of yesteryear!) And 1997 is the only time Labour’s won a working majority from opposition. Not the biggest or the most enduring or the most elegantly arranged – the only one.
So when people talk about Labour winning the next election, they really aren’t talking about getting back to business as usual; what they’re talking about is replicating a single, bizarre and unparallelled combination of circumstances. Or else they’re thinking – or they need to start thinking – in terms of striking out in a new direction, and pulling together some other combination that might prove equally effective. (We could try populist rhetoric attached to radical policies, perhaps, with a likeable figurehead and a credible pitch to young people. Worth a shot.)
I get the impression that the current leadership team believes that a competent campaign and a halfway friendly press will win Wakefield and seats like it, but that they need to reposition Labour – as patriotic, fiscally responsible, tough on crime and so on – in order to get the true “Red Wall” seats back. They also seem to believe that once the Red Wall comes back, the country will come with it. I agree that Wakefield’s probably staying Labour next time out, and some – perhaps as many as half – of the “Red Wall” seats will probably come back. But Don Valley and seats like it are almost certainly gone; that damage has been done, a long time ago. And – more importantly – where the “Red Wall” goes, goes… nowhere else in particular. Neither of those two groups of seats are big enough, or typical enough, to mean that winning them would be enough to win the election.
What will? One thing we know is that we aren’t going to have another 1997 – the preconditions for it aren’t there. That in itself means that something fairly dramatic is going to be needed – something a great deal more dramatic than talking about tax cuts and waving the flag. Unfortunately, unless there’s a big change in the leadership or (less probably) in leadership style, Labour just isn’t going to have the kind of inspirational appeal that enables the party to make new inroads into Tory territory. (I mean, it’s not as if people were oh say for example spontaneously chanting Keir Starmer’s name, is it.)
Wakefield or no Wakefield, on a national level it all adds up to another loss. I’ve got a nasty feeling that the leadership’s already working on that basis, and that the next election will see Starmer, well, doing a Kinnock – putting a brave face on a few gains and arguing that the gains the party didn’t make just show that his work is not yet done.
If it does come to that, I just hope that some of the people who backed Starmer on the basis that Corbynism was a vote-loser will draw the right conclusions.
Workers’ Playtime 