The Hartlepool by-election result has renewed discussion about how Labour re-connects with working-class voters. This article looks at two components of this. Firstly, it examines the idea that root of Labour’s problems come from having ‘triangulated’ away from its core vote in the 1990s. And secondly, it explores the phenomenon of ‘deprived bellwethers’ (as I call them) and other poorer seats that Labour doesn’t currently represent. It asks what the party can learn from areas which — as those on the left sometimes put it — vote ‘against their interests’.
‘Nowhere to go’
A contention from the far left, so common it’s now seldom contested, is that Labour in 1997 ‘abandoned its base’. The argument is that, under Tony Blair, the party ‘took for granted’ the electors it was created to represent. New Labour assumed that the core vote in its heartlands had ‘nowhere to go’, and instead appealed to affluent seats in middle England. This is said to have denied working-class communities a voice — creating voter apathy, allowing in the far right and catalysing the collapse of the Red Wall. Just last month, Ian Lavery wrote a piece for Tribune which included this charge.
I was always puzzled by this account of things. I remember working for the Labour candidate in South Thanet in 2015, for example. The seat is the 113th most deprived constituency of 533 seats in England. Having been gained by Labour in 1997 and held in 2001 and 2005, it was lost in 2010. By the time I was working there it was effectively beyond reach. The Tories won it by more than 10,000 votes in 2019.
The analysis below looks at data for all General Elections from 1983 to 2019, gauging how well Labour did with its base in each one.
There are different definitions of the Labour base, of course. So I’ve used seven metrics, each of which could potentially describe those the party sees itself as representing. These are:
· Deprivation score, as Labour tends to regard itself as the champion of society’s poorest;
· Proportion in social grades C2DE, working-class voters being Labour’s traditional social constituency;
· Proportion that are social/ private renters, as the left sees itself as a protector of those without property and those living precarious lives;
· Average weekly salary, as progressive politics aims to equalise economic outcomes and pay;
· Average house price, as we hope to represent those with less social status and fewer economic assets;
· Unemployment rate, as Labour aims to champion those denied opportunities;
· Percentage in mining/ manufacturing jobs, as the party speaks, historically, for communities in industrial and unionised sectors.
There is considerable overlap between seats scoring highly for these things, but they are not, automatically, the same places. Only 76 seats are not in the more deprived half of the country for a single one.
The charts below look only at seats in England. They are based on an attempt to trace constituencies’ back to 1983. This is an inexact science, as boundaries have changed significantly in some cases. But there are 496 seats that have existed, in some recognisable form, across all ten elections during the period.
Chart 1 looks at constituencies which are in the top half of English seats for each of our seven metrics — i.e. the 260 or so that are above the median average for deprivation, proportion C2DE, number of renters, etc. (The dates in brackets signify the year the data was taken from). The chart shows how many of these seats Labour won over the course of the ten elections that took place — excluding abolished or newly created constituencies.
Each metric follows the same rough pattern. Labour starts from a low base in 1983, wins steadily more of the seats that might constitute its base in the 1987 and 1992, and does extremely well in 1997 and 2001, with a little momentum lost in 2005. The party falls back dramatically in 2010 and 2015, has a minor uptick in 2017, and then does very poorly indeed in 2019.
The exceptions to this are ‘proportion unemployed’ and ‘proportion renting’. With these two measures the above pattern also exists, but Labour has not, by 2019, fallen to the same levels as in the 1980s. The reason for this is largely the party’s solidification of a number of inner-city seats. Westminster North and Streatham, for example, have large numbers both of social and private renters. Both are safely Labour now but were Tory in 1983.
Chart 2 shows the same thing as Chart 1, except that it focuses only on constituencies which are in the top 50 seats for each metric — rather than the top half. As we can see, the patterns are much the same, although the differences between the years in government and those in opposition are less pronounced. Labour’s solidification of seats with high numbers of renters is even more evident.
Of course, it may be that what critics of ‘triangulation’ are talking about is the overall makeup of the Labour vote — rather than the number of deprived seats won. Compass’s Neal Lawson (who I had a very enjoyable published email exchange with in March) argued this a few years ago. In the New Labour years, he wrote, “the tent was too big and [Tony Blair] spent the next 10 years trying to keep the wrong people in it.”
In other words, it could be that New Labour’s abandonment of its base came about by winning richer seats — and thus ‘diluting’ its appeal — not by losing poorer ones. This proposition is, in my view, pretty shaky, raising tough hypothetical questions. Should Labour even field candidates in affluent constituencies, for example, if winning them represents a capitulation? Nevertheless, the ‘dilution’ premise is worth investigating.
Chart 3 shows the composition of the seats won by Labour in all ten elections (removing, once again, those which cannot be traced all the way through the period). It illustrates the breakdown in terms of seats won in the top half for deprivation versus those in the bottom half. As we can see, Labour won an extra 67 less-deprived seats in 1997, as well as an extra 63 more-deprived seats.
Charts 4 and 5 show the same thing for class and pay, respectively. Labour in 1997 won an extra 63 of the more middle-class seats, for example, in addition to 67 of the more working-class ones. Likewise, the party won 80 extra better-paid seats, in addition to 50 worse-paid ones.
This does, indeed, mean that the average seat held by Labour was more affluent in 1997 than before — particularly on the pay metric. But it’s hard to see why this is an issue. If Labour had been losing working-class seats at the same time as it gained middle-class ones then there might at least be an argument to be had. But, given that the party was gaining both at a roughly equal pace, it’s hard to see that the ‘dilution’ critique carries much water.
One thing to point out, meanwhile, is the result in 2005. In this election, with Blair at his least popular, the seats lost were mainly those furthest from Labour’s base. The party lost 21 less-deprived seats compared with 15 more-deprived ones; 22 of its more middle-class seats compared to 14 of its more working-class seats; and 26 of its better-paid seats compared to 10 of its worse-paid ones. Essentially, the non-traditional, more affluent parts of the 1997 coalition started deserting the party — not the working-classes.
Also of note is the makeup of the seats won by Labour in 2017, at the high noon of Corbynism. Of 19 net gains, for example, 16 were at the more middle-class end of the spectrum. Unlike in 1997, Labour was winning middle-class seats at a much much faster rate than it was winning working-class ones — a ratio of around 5:1, in fact. (Indeed, when we look at actual numbers instead of net change, we can see that six more-deprived seats were lost in this election — such as Stoke-on-Trent South and Walsall North — compensated for by gains in better off places, like Leeds North East and Sheffield Hallam).
Chart 6 looks in more detail at seats lost. It takes the number of seats won in 1997 as the high water mark, and analyses the composition of those lost in subsequent elections — including by Tony Blair himself. It uses five of the seven metrics (removing those that are difficult to compare on a single chart).
Of the three tranches of seats lost, those that were Labour in 1997 but had been lost by 2005 were, on average, those furthest from Labour’s base. And those that were Labour in 2015 but not by 2019 were closest to the party’s historic core vote. The 43 seats lost by Blair between 1997 and 2005 had an average deprivation score of 20.6, for instance. The 41 lost between 2015 and 2019 had an average score of 28.8.
The only exception to this is the percentage renting, where the proportion is lowest among the Blair losses. This reflects, once again, Labour’s steadily strengthening support, during the 2010s, among young, inner-city private renters. This group are locked out of the housing market, especially in London, and may have spent much of the decade struggling to find skilled work. Termed ‘educated left behinds’ by Tim Bale, this was, in a certain sense, the core Corbynite vote.
The chart below shows the figures for the two metrics not shown on Chart 6. Again, we can see the same pattern in place, with the seats lost by Corbyn tending to have lower property prices and worse pay.
The Financial Times columnist Janan Ganesh wrote a few years ago that “New Labour was always misread as a middle-class takeover of a working-class movement. It was something close to the opposite.” Charts 1–6 appear, in different ways, to corroborate this. It’s hard to see much basis for the idea that the seeds for Labour’s present woes were sown during the party’s time in office.
Of course, there are factors beyond the simple test of the number of seats won. Enthusiasm for New Labour was Luke warm in the party’s traditional heartlands, it’s sometimes argued — reflected by the low turnout at the 2001 election.
I tend to think the meaning of turnout is in the eye of the beholder, here. 2001’s low turnout could be the result of a radical electorate denied a ‘real choice’. But it could also indicate a contest with very little jeopardy, and a voter base who were fairly happy with the direction of travel. So I don’t think turnout alone can be used as a measure.
However, the number of votes going to Labour in deprived seats does matter, as does the party’s vote share. It could be that disillusioned communities in working-class seats voted Labour out of habit during the 1990s and 2000s — without enthusiasm and with increasing numbers opting to stay at home. In this light it’s worth examining the size of the Labour vote in the heartland seats we’ve been looking at, as well as who won.
One major caveat here is that we need to also chart the performance of the Conservatives. If we look at seats which are in the more deprived half for at least four of our seven metrics — of which 241 can be traced throughout the period — we can see why. Labour got an average of 19,533 votes per seat across these constituencies in 2001, whereas in 2017 the average figure was 24,241. However, the Conservatives also got many less votes in these seats in 2001 than in 2017 — an average of 11,094 in the former election and 18,351 in the latter. Presented with a ‘real choice’ in 2017, an average of 4,708 more voters per seat chose Labour than in 2001, but an additional 7,257 per seat were motivated to vote Tory. If the ‘real choice’ you’re proposing is driving people in deprived seats to vote for the other main party in greater numbers than for your own then you have a problem.
It seems like the most sensible thing to look at here is Labour’s average majority — or, rather, at the party’s lead over the Tories (this being where the primary battle lies). Chart 7 shows this, for the 241 seats in the top half for four or more of the seven metrics. Seats that cannot be traced through the duration of the period are again removed. And the red line indicates the mean average.
What this shows, firstly, is that the patterns observed in Charts 1–6 essentially hold out when you look at vote share — albeit in a flatter way. Secondly, the range of vote shares in these deprived constituencies is much narrower during the Blair elections and much broader during the Corbyn years. Across the 241 seats, Labour’s biggest leads over the Tories in 1997 were around 30,000. Its smallest were around minus 15,000. By 2017 Labour was beating the Tories by about 40,000 in its safest deprived seats, but losing to them by as much as 25,000 in others.
Chart 8 examines this using a narrower definition — the 82 seats that are in the 50 poorest for at least two of our metrics. Here, Labour’s 2017 average performance is slightly better, but the sawn-off shotgun effect post-Blair is even more pronounced. Whereas Labour won all 82 of these deprived seats in 1997, 2001 and 2005, by 2019 there was no discernible pattern. 25 had bigger Labour leads over the Tories than in 1997. But many more had seen their majorities fall substantially. 21 are now Conservative-held — rising to 22 as of yesterday.
The deprived seats where Labour saw significantly increased majorities post-2015 are interesting to look at. These are poor and working-class areas where there appears to have been a genuine appetite for Corbynism. In some cases these constituencies had a significant, left-leaning Lib Dem vote, which overwhelmingly converted to Labour.
If we return to our list of 241 deprived seats (examined in Chart 7), we find that there are 48 Labour-held seats where a) 2017 Corbyn got a bigger majority than 1997 Blair, AND b) 2019 Corbyn a bigger majority than 2005 Blair. These radical working-class areas fall into several sub-groupings:
· Seats in Merseyside (e.g. Knowsley);
· Seats with large Pakistani or Bangladeshi heritage populations (e.g. Blackburn);
· Seats with student populations (e.g. Portsmouth South);
· Seats in parts of inner London with a radical history (e.g. Tottenham).
Although it’s worth noting that the party was also doing poorly in these areas pre-Blair, these 48 seats are the closest thing I can see to a genuine example of working-class areas clamouring for Corbyn-style socialism. I wouldn’t necessarily assume that the constituencies in question were tub-thumping fans of the then Labour leader — any more than the poorer swing-seats won by Blair were vocal advocates of the Third Way — but they certainly became more tribally Labour in the 2010s. They were not alarmed by the perceived radicalism of the party post-2015, or else were so hostile to Boris Johnson’s Tories that it didn’t register.
These 48 seats do not represent working-class communities per se, any more than Red Wall constituencies, historically Tory-voting areas with rural poverty, or anywhere else does. But they are an interesting sub-set to be aware of, and should not be dismissed.
The problem for Labour was that, for every seat like this there were several other deprived constituencies that went in the opposite direction. Chart 9 returns to seats won as the primary metric of success, and shows the electoral consequences of this, in our first-past-the-post-system. It reveals the overall pattern among the 241 more deprived seats looked at in Chart 7 (including only those that can be traced all the way through). Whereas just 30 were won by a party other than Labour in 1997, 121 were in 2019.
All of the above is relatively crude, and big demographic shifts in the past 40 years mean there are limits to how much you can infer the further you go back. Also, the absence of Scotland and Wales in the charts has probably not been kind some of the former Labour leaders on there, such as Michael Foot and Gordon Brown.
Moreover, it’s important to be aware that we have started at a low point for Labour, 1983 being an especially bad election for Labour’s working-class support. The Blair years are flattered a little by this in the charts.
So, to be clear, I didn’t write the above because I think Keir Starmer’s Labour needs to revert to the New Labour template, or that Tony Blair would somehow magic back the Red Wall were he to become leader again. I’m just trying to point out that the narrative preferred by the far left, whereby ‘centrists’ robbed the working-classes of their only political mouthpiece, is some way off the mark. Also, more importantly, I want to underscore that the idea of poorer and more working-class voters as a politically homogenous group is dangerous for the left — particularly if our approach to that group is to romanticise or stereotype it.
To the extent that the modern Labour Party can learn from the 1990s, it’s in the desire to listen to these communities rather than align them with our own assumptions. Projects like ‘Southern Discomfort’ — or the much-derided use of focus groups, for that matter — sought to think about the types of seat we’ve looked at above in a more nuanced way, rather than mythologizing them. Tony Blair’s use of ‘aspiration’ was an attempt to speak to poorer voters who wanted socio-economic divisions to be removed — rather than turned into the battle lines for a ‘class war’.
Instead of the automatic belief that working-class non-Labour seats are voting ‘against their interests’ progressives need to ponder what those ‘interests’ actually are, in the hugely varied range of places which might plausibly constitute the party’s base. The second section of this article assesses this a little more.
Once you include the seats which cannot be traced back for the entirety of the period we’re looking at (which adds 11 constituencies), there are 252 seats in England which are above the median average for at least four of our seven metrics. 122 were Labour-held prior to the Hartlepool result.
Chart 10 shows the cumulative swing in these places, between the 2010 election and the 2019 one. Neither of these were good results for Labour, but the cumulative swing data shows how much worse things got between the former at the latter, under the surface. Of our 252 ‘more deprived’ places, 169 swung to the Tories and 83 to Labour. When you look at very significant swings — over 10 points one way or the other — the disparity is even greater. 68 ‘more deprived’ seats swung Conservative by more than 10 points (e.g. Bassetlaw, Amber Valley, Cannock Chase and Waveney). Just 15 had a similarly pronounced shift in the other direction (e.g. Manchester Gorton, Birmingham Hall Green and Bradford East).
This shows the extent to which Labour rowed away from its base — whatever definition of the term we’re using — during the 2010s.
If we look at the same thing for our list of very deprived places (i.e. those looked at in Chart 8, which were in the top 50 for at least two of our seven metrics) the pattern is a less pronounced. Of these 85 seats 45 swung to the Conservatives (24 by more than 10 points) and 40 swung to Labour (14 by more than 10 points). However, given that these are places with acute poverty — the likes of Middlesbrough, Oldham and Blackpool South — this more mixed picture is hardly cause for celebration.
The collapse of Labour’s Red Wall has attracted a great deal of debate and speculation and, important though this is, there’s a risk that it masks the wider haemorrhaging of a range of deprived seats, over a much longer period. The swings taking place even since the 2010 electoral defeat are profoundly worrying.
Hence, rather than regarding of the collapse of the Red Wall as a distinct phenomenon, it’s important to think of it as a sequel to the loss of deprived constituencies in 2010, 2015 and 2017. In this timeframe 48 ‘more deprived’ seats went to other parties.
Likewise, we need to see it as a potential prelude — if we’re not careful — to further backsliding. There are 21 places which are currently Labour-held but which a) have leads of less than 5,000 over the Tories AND b) have seen cumulative swings to the Conservatives of at least 5% since 2010. Examples include Wentworth and Dearne and Hull East. Hartlepool was not among these, its cumulative swing to the Tories being 3% during this period, yet was lost last night.
Chart 11 roughly categorises the political allegiances and voting histories of all 252 seats that are in the top half for at least four metrics. It divides them into seven sub-groups:
· Strong Labour, socialist leaning/ reclaimed bellwether. These are ‘Always Labour’ or ‘Mostly Labour’ seats, but ones where — as was mentioned in the previous section — there appeared to be a genuine enthusiasm for Corbyn’s leadership. (Examples: Liverpool Riverside, Blackburn, Brent Central). Also included here are a handful of ‘reclaimed bellwethers’, which were relatively marginal historically but were won in 2017 and held in 2019. (Example: Brighton Kemptown, Bedford).
· Almost always Labour. These are solid Labour seats, which are currently held by the party and which elected Labour MPs for at least nine of the ten elections we’ve looked at. The profile of Labour’s support varies from one to the next, but they were not so hostile to Labour’s direction after 2015 as to join the Red Wall — nor so enthusiastic as to be classed as ‘socialist leaning’. (Examples: Jarrow, Coventry North East, Bristol South).
· Mostly Labour. These seats have voted Labour for between six and eight of the ten elections and are currently held by the party. They are somewhat miscellaneous as a group, but have not followed the same clear pattern as the Red Wall. (Examples: Southampton Test, Rochdale, Ipswich).
· Red Wall. These seats have got a lot of attention in the past eighteen months. They were strongly Labour until 2017 or 2019, but were lost for the first time under Corbyn. I used Steve Rayson’s Red Wall definition here, with one or two tweaks. Hartlepool is sadly now among them. (Examples: Newcastle-Under-Lyme, Wakefield, Mansfield).
· Deprived Bellwethers (broad definition). This is a category of seat we’ll look at in more detail. They are places which returned Labour MPs for between two and five of the ten elections covered in this analysis, including at least two of the three Blair elections. (Examples: The Wreckin, Pendle, Nuneaton).
· Deprived Bellwethers (narrow definition). This set of constituencies are very similar to the group directly above, but they fulfil a purer version of the definition. They are places which Labour won in all three of 1997, 2001 and 2005, but not before or since. (Examples: Great Yarmouth, Tamworth, Dover).
· Almost never Labour. These are places held continuously by the Tories (often in rural parts of England) or by the Lib Dems (in the case of the South West). Some were won once by Labour, anomalously, during the ten election period looked at here. (Examples: Yeovil, Louth and Horncastle, North East Cambridgeshire).
There has been a bit of ‘rule of thumb’ involved in categorising some places. But the groupings, while inexact, let us see a more granular breakdown of England’s deprived seats.
As Chart 11 shows, around half of the deprived seats in question are ‘Strong Labour/socialist leaning’, ‘Almost always Labour’ or ‘Mostly Labour’. Of the 122 deprived seats currently held by Labour, 82 are still what might be termed ‘safe’, with a 2019 majority of at least 5,000 (although post-Hartlepool I’d be cautious of classing anywhere as safe).
Just under a sixth are in the Red Wall, meanwhile, and a similar proportion are ‘Never Labour’.
This leaves 52 seats that I’ve classified as one or the other type of ‘deprived bellwethers’. These are listed below. Also shown, in the third column below, are bellwethers that are a little less deprived — sitting in the top half for two or three of our seven metrics, but not for four or more. Of the 52 seats listed in the first two columns, 45 have swung towards the Tories during the past decade.
Within this ‘deprived bellwether’ group there are several interesting types of place, which repeatedly crop up.
Firstly, a large number of seaside towns are deprived bellwethers. This includes Blackpool North and Cleveleys, Cleethorpes, Morecambe, Sittingbourne and Sheppey and Waveney. Within this group are some of the poorest communities in the UK — Sheerness on the Isle of Sheppey, for example. Many have distinct types of social issue, such as significant pensioner poverty and high numbers of drug deaths.
These resorts are geographically less connected. They often have older, non-diverse populations, with a strong sense of cultural loss, rooted in genuine economic decline. The predominance of informal job sectors and service-based economies means workforces are less likely to be unionised or organised.
Secondly, there are a number of ports or maritime centres, including Dover, Chatham and Aylesford and Portsmouth North. If we include slightly less deprived places then other seats, such as Brigg and Goole, feature as well.
The nature of work may be a determining factor in these places not being safer Labour. Historic jobs in ports, while unionised, will often have been less skilled, or else have been solitary occupations — such as operating cranes or forklifts — with less sense of collective endeavour. In some of these seats, furthermore, immigration is very much a ‘live issue’.
Thirdly, there are numbers of new towns within the ‘deprived bellwether’ grouping. This includes Harlow, Corby, Telford and Redditch. Again, if we broaden the definition to add slightly less deprived bellwethers (those in the top half for 2–3 of the seven metrics) then Crawley and Milton Keynes can be included.
New Towns often have large amounts of social housing and significant populations of older residents, many of whom grew up in bigger cities but migrated out in the post-war years. More recently, low property prices and good transport links have made New Towns a destination for younger and less well-off people looking to get on the property ladder. Fewer community assets and shared institutions exist — like pubs or football clubs — and, historically, the ‘newness’ of these seats may mean they’re beholden to neither party, even when deprivation is acute.
Fourthly, we come across a clutch of market towns and small cities: Gloucester, Worcester, Lincoln, Shrewsbury, Stourbridge. These areas would not automatically conjure images of deprivation; but they tend to be in the more deprived half for a number of our metrics, without being in the top 50 for any.
A lot of the market towns and small towns in question have universities, but they’re not strong enough academic centres to be hubs for left-liberalism — in the ways that Cambridge or York would be. Meanwhile, although industrial jobs exist, such seats are not so reliant on manufacturing as to be died-in-the-wool Labour.
Lastly, there are a group of towns which I’ve called, for want of a better label, ex-industrial multi-towns. This includes Keighley, Camborne and Redruth, Cannock Chase and Pendle. The most notable factor with these places is that a mining or manufacturing history exists, and that there are two or three smaller towns stung together, in an otherwise non-urban part of the country.
Levels of deprivation here vary. The towns in Pendle, for example, are very deprived, whereas in Calder Valley they’re only a little poorer than the median average. But what ‘ex-industrial multi-towns’ seats have in common is that they’re less cosmopolitan or urban. We might infer that rural and industrial traditions vie for political primacy here, creating a blue-red tug-of-war at the ballot box.
In addition to the above, we find some broader characteristics that occur across our deprived bellwethers. A number, for example, are characterised by lighter or more modern forms of industry. This includes brewing and shoemaking in Kettering and Burton, and car-making in Swindon. Often, industrial decline may be a more recent or gradual factor, with the strife of the 1980s not scorched on the memory in the same way as in Yorkshire or Wales.
Moreover, although the ‘deprived bellwethers’ are scattered across the UK, there is a distinct southern, eastern and Midlands slant. This may reflect areas which are more culturally Anglo-Saxon rather than Celtic, with stronger cultural attachments to English identity. It could also be due to wider differences, again relating back to experiences during the 1980s.
This leads on to the question of diversity. The ‘deprived bellwethers’ tend to have smaller non-white populations than other parts of the country, without the collective memory of migration that you might find in urban centres. However, this is changing in many such seats. According to a Webber Phillips report last year, 26 of the 52 ‘deprived bellwethers’ are Newly Diversifying (i.e. experiencing rises in migration that are faster than the national average, from a start-point which is lower). This compares with 105 of 533 seats across England. It may help to explain the salience of migration in these places.
A final interesting element of these ‘deprived bellwethers’ is specific geographical position. Almost none are in the centre of large conurbations, but several are situated in the edge of major cities: Bolton West, Dudley South, Keighley, Thurrock, Harlow, etc. One or two others, meanwhile, are what you might, perhaps pejoratively, call ‘second seats’ — the poorer and less prestigious constituency in a two or three seat city.
This tells us a lot. One of the defining traits of our ‘deprived bellwethers’ is that they are marginalised in national debates, and do not quite ‘fit’ the preferred groupings of left or right. Many are places without a great deal of glamour and status attached to them, which exist in the shadow — metaphorically and sometimes physically — of our city hubs.
Chart 12 illustrates this. It shows the number of clusters which our ‘deprived bellwethers’ fall into the top half for (bearing in mind that being in the top half for at least four is a prerequisite for the category). It reveals that 49 of our 52 ‘deprived bellwethers’ are below the median average for house prices. This is particularly striking, given that a number are in the south east, where property prices are higher.
This makes me think a focus on ‘place’ is really important for understanding ‘deprived bellwethers’. If I was Labour I’d focus, in a nuanced and granular way, on some of the ‘deprived bellwethers’ mentioned above — seaside towns, ports, New Towns, etc — to understand what they’re looking for from the party. Working groups could be set up to engage, within each type of place, with former Labour representatives, unsuccessful candidates in past elections, Council group leaders, local activists, etc.
Speaking for myself, I only know a small number of the ‘deprived bellwethers’ well. But my hunch would be that appealing to ‘deprived bellwethers’ does not require progressives to jettison all they hold dear. Perhaps there would be specific policy demands, on immigration or crime, that might make the average Labour member squeamish. But more likely, I suspect, the answers lie in changes to our language and culture, and in a focus on policies which are practical and constructive.
It’s often assumed that debates over ‘the flag’, for instance, are about demonstrating Labour’s patriotism. Yet the reality is that these discussions are, to a much greater extent, about showing that Labour is professional and no nonsense. Being comfortable singing the national anthem or bowing to the Queen is the political equivalent of wearing a suit to a job interview. You’re not expected to love wearing suits in your spare time, but if you have trouble doing it when it matters then you’re clearly not a serious candidate.
As things stand, Keir Starmer sitting next to a Union Jack provokes an extended period of soul searching and hand-wringing among progressives. As a result Labour is regarded as unserious — obsessing over gestures and stances rather than focusing on the things that matter.
In late 2019 I released a book, Warring Fictions. Subsequently republished as The Dark Knight and the Puppet Master, it criticised several myths inherent within left populism, including what I called the Dark Knight myth (the tendency to categorise politics into good and bad) and the Golden Era (the tendency to idealise the past).
A lot of the discussions I subsequently had with Labour members about the book focused on its moral and intellectual implications. Was I defending the architects of austerity as decent and well-intentioned people? Was I trying to say that ‘neoliberalism’ hadn’t happened?
Yet the wider aims of The Dark Knight and the Puppet Master were just as much about electoral cul-de-sacs as ethical ones. It critiqued the self-defeating habit of romanticising our own ‘side’ and caricaturing opponents. The progressive tendency, after all, is often to idealise Labour’s electoral base, imagining it as some mix of the Durham Miner’s Gala, Notting Hill Carnival and I, Daniel Blake. Conversely, we assume that those who vote for the Conservatives are stock-brokers and property developers, who only leave their gated communities to attend Ascot and Henley Regatta. (I myself am probably guilty of caricature here, but only slightly!)
The analysis of Labour seats over time reveals how baseless many of our assumptions here are. It shows that working-class is not an automatic synonym for left-wing, and that tacking away from conventional left-wing policies is not necessarily the same thing as abandoning the base. The misty-eyed recollection of Labour heartlands betrayed by Tony Blair does not, for the most part, stack up.
The focus on ‘deprived bellwethers’ meanwhile, points to a whole set of places which defy simple categorisations. There are southern seats with ‘northern’ problems, home-owners who are financially struggling, deprived communities voting ‘against their interests’ and working-class towns with a strong sense of upward mobility. By deploying the morally-charged, ‘Dark Knight’ language of villains and victims, Labour is unlikely to appeal to people in these places, and will probably end up putting noses out of joint.
Looking at these ‘deprived bellwethers’ — rather than at the Red Wall — may be a strange thing to do on the morning after an electoral defeat in Hartlepool. Yet if we look at Labour’s lead over the Conservatives in the seat prior to the 2021 by-election we can see that it follows very much the same pattern as many of our ‘deprived bellwethers’, as chart 13 shows. The collapse of the Red Wall is emblematic, but is a symptom of something far deeper.
I don’t know what the answer for Labour is here. But we need to start by analysing accurately what has actually happened and is happening. The aim, as much as possible, must be to think beyond archetypes and stereotypes. This will help us to define Labour’s purpose as a political party, to think through what our ‘base’ might possibly look like in 2020s Britain, and to understand how we can re-connect with that base.
 This is because the electoral context in the other three nations is different, and also because comparable Welsh, Scottish and Northern Irish data does not exist for several of the metrics (the Index if Multiple Deprivation, for example).
 The data on deprivation, social grade etc used is mostly from the 2010s. Accurate constituency data is harder to find the further back you go. And the truth is that, while the gaps between places have narrowed and widened at certain points — and there have been changes affecting all areas — there are not that many cases of dramatic shifts in the ordering (e.g. of an area going from the 100 highest seats for deprivation to the 100 lowest). With than said, there is a degree of weakness in the method, in that data for our seven metrics during the earliest elections covered is likely to be a bit less accurate than that for recent polls. With the two metrics based around housing this is most true.
 This could be seen as a weakness of my analysis, insofar as Corbyn was competing in a largely two-party system, unlike any other leader in our ten elections; of course his majorities were larger — they were not split by smaller parties. However, engaging with different electoral contexts is simply the nature of the beast, I think. Blair did not have the complex trade-offs of Brexit allegiances to content with, for example. Brown did not have the additional threat of Ukip which his predecessor Ed Miliband faced. A political leader can only beat what is in front of them.
 There are about 40 more seats which fulfil this criteria once we include places in the top half for just two or three criteria. These are overwhelmingly university cities like Exeter and inner London seats like Hackney South or Vauxhall.
 It’s worth pointing out here that The Conservatives were not the second party in 2005 and 2015, the Lib Dems and Ukip running Labour closest.