Marginal notes

We can be fairly sure that a general election is coming soon. (I’ve been saying that since last December, admittedly, but surely it can’t be much longer now.) With that in mind I’ve been thinking about marginals: the Tory seats that Labour needs to gain in order to form the next government, the Labour seats the party needs to hold in order not to cancel out its gains. Can we identify any patterns, or is Labour just going to be keeping multiple plates spinning – attracting the centrists while holding the loyalists, attracting Remainers while holding Leavers, and so on?

As a starting point, I looked at the size of Labour’s majority over the Conservatives – or vice versa – in the most marginal Labour/Conservative battleground seats, in general elections over the last twenty-odd years, i.e. going back to 1997 and New Labour. For the following chart I’ve used the Election Polling list of Conservative targets, and selected the first twenty constituencies where (a) Labour currently hold the seat and (b) the seat has existed at least since 1997. All figures are % shares of the vote; figures are rounded to the nearest whole % except for figures below 0.5, which are rounded up to 1. The ‘average’ marked with an X is the average Labour lead over the Tories across all these elections.

I expected to see three different patterns, split more or less evenly: safe Labour seats gradually going marginal due to changing demographics or incumbent complacency; vulnerable Tory seats going marginal and being narrowly taken by Labour; and permanent marginals, switching back and forth between the main parties. Here’s what I actually found (click to embiggen):

Reading from left to Right (with a couple of adjustments), we have:

• one long-term Tory seat (Canterbury), which was marginal in 1997 and 2001 and safe in the next three elections; in 2017, Labour overturned a majority of 18.3%
• 13 seats with two distinctive characteristics:
  • they were held by the Conservatives in 2015
  • they had a healthy Labour majority in 1997, 2001 or both; Labour majorities in this group range from Stroud (4.7% in 1997, 9.1% in 2001) to Crewe and Nantwich (31.2% in 1997, 23.8% in 2001). (Bear in mind that these are the majorities, i.e. the difference between the Labour and Tory vote shares. Labour’s actual vote in Crewe and Nantwich, in 1997, was 58.2%.) After 2001, in each of these seats, the Labour majority dropped and went on dropping; 11 of the 13 went Tory in 2010, and eight of those had a larger Tory majority in 2015. (Of the other two, Peterborough went to the Tories in 2005 and Derby North in 2015.)
• six seats which had never gone to the Tories, but where
  • Labour had similar or even larger majorities in 1997 (Labour’s majority in Bishop Auckland was 45.7% – higher than Labour’s vote share in 2010 or 2015)
  • Labour’s majority had dropped and carried on dropping at every subsequent election, including 2017; by 1% or so in Barrow & Furness and Newcastle-under-Lyme, but by over 10% in Dudley North and Ashfield

This was unexpected. Apparently Labour’s battleground, at least when it comes to defending home turf, consists almost entirely of former safe seats where a substantial majority was allowed to trickle away over successive elections – between 2001 and 2005, 2005 and 2010, 2010 and 2015. And, while it’s true that some of these seats saw another drop in the Labour majority in 2017, the large majority of them – 13 vs 6 – were actually taken back from the Tories in 2017, representing a substantial improvement in Labour’s majority (from negative to positive, apart from anything else).

I repeated the exercise using Election Polling‘s list of Labour targets, again selecting the first twenty constituencies where (a) the Tories hold the seat and (b) the seat has existed at least since 1997; again, X marks the average Labour lead over the Tories across all these elections. Again, I expected to a pretty even split between safe Tory seats gone marginal, former Labour seats where the Tories had squeaked in and permanent marginals. And here’s what I found:

Déjà vu, anyone?

What we seem to have here – again, reading roughly from left to right – is

• two marginals (1997-2001), turned solid Tory seats (2005-15), turned marginal again in 2017
• 12 seats with two distinctive characteristics
  • they had a healthy Labour majority in 1997, 2001 or both; Labour majorities in this group range from Finchley & Golders Green (6.4% in 1997, 8.5% in 2001) to Southampton Itchen (26.4% in 1997, 27.1% in 2001). After 2001, in each of these seats, the Labour majority dropped and went on dropping; 10 of the 12 went Tory in 2010, and seven of those had a larger Tory majority in 2015. (Of the other two, Preseli Pembrokeshire went to the Tories in 2005 and Southampton Itchen in 2015.)
  • they were marginal in 2017 but not in 2015, i.e. the Tory majority over Labour was substantially reduced
• six seats where
  • Labour had similar or even larger majorities in 1997 and 2001 (Labour’s 1997 majority in Mansfield was 43.3%)
  • between 2015 and 2017, Labour’s majority (or lack of one) had stayed the same (Bolton West, Telford, Thurrock) or dropped even further (Middlesbrough and Cleveland East, Stoke-on-Trent South, Mansfield)

Extraordinarily, Labour’s offensive battleground seems to be very much the same terrain as the area it needs to defend. In both cases, we’re looking at former safe Labour seats where a substantial majority was allowed to trickle away over successive elections – between 2001 and 2005, 2005 and 2010, 2010 and 2015; and in both cases, in 2017 two-thirds of these seats saw either a Labour gain or a substantial cut in the Tory majority in 2017.

Three conclusions. Firstly, the battleground seems to be the legacy of years of post-New Labour complacency: a decade and a half when some Labour MPs allowed themselves to think they had a job (and a fan base) for life, and didn’t see their support wearing away – or how insecurely it was founded – until it was too late. Secondly, something happened between 2015 and 2017 which – in the great majority of cases – stopped this process dead and reversed it. Look at Battersea, Ipswich, Colne Valley or Stockton South; look at Pendle, Preseli, Southampton Itchen. The Labour majority goes down, and down again; goes negative, and goes down again; and then there’s 2017. Town and country, north and south, it’s the same pattern. (Of course, any current Labour member could have told you precisely what’s happened in the last four years – how the mood’s changed among the membership and, apart from anything else, how much more campaigning is getting done these days – but it’s nice to see it’s had some effect.) Thirdly, there are places that this process hasn’t reached, or at least hadn’t reached by 2017 – places where the long erosion of Labour majorities continued in 2017, even to the point of tipping a couple of seats to the Tories – but they are the minority. Not that you’d know about it from the way that they’re covered in the press. Stoke and Newcastle-under-Lyme, Mansfield and Ashfield, these places have a story to tell – but it’s not the story of Labour heartlands turning against Corbynite radicalism (unless the radical rot had set in by 2010). Just as importantly, it’s not typical – it’s not the story of Labour’s battlefield seats more generally.

To demonstrate the similarities between the two groups of ‘battleground’ seats and highlight the two trends I’m talking about – the long slump from 1997 to 2015, the fightback in 2017 – here are all 40 together. Remember, these are Labour’s and the Conservatives’ most marginal seats, excluding only (a) seats which haven’t existed throughout the period since 1997 and (b) seats where a third party is or has been the main contender for the seat. (Which means that Scotland doesn’t get a look-in in this post; sorry about that, but it really is a different story.) For clarity I’ve stripped out the 2001, 2005 and 2010 results, to emphasise the contrasts between 1997 and 2015, and between 2015 and 2017. They’re arranged in a different order here: the X measures the difference between Labour’s 2015 and 2017 majorities over the Conservatives.

Now there are three groups:

• seven seats in which Labour’s majority was halved or worse between 1997 and 2015, then fell substantially in 2017
  • Tory gains: Mansfield, Stoke-on-Trent South, Middlesbrough South and Cleveland East
  • Labour marginals: Ashfield, Dudley North, Bishop Auckland, Keighley
• five seats where a large fall in Labour’s majority between 1997 and 2017 worsened only slightly in 2017 or was unchanged
• 28 seats – all but two held by Labour in 1997 – where the long fall in Labour’s relative vote share was reversed in 2017

That’s an across-the-board trend (a steady falls in Labour’s relative vote share from 2001 to 2015) and a partial but very strong countervailing trend (a reversal of that fall in 2017, in 28 seats out of 40). Before looking at the numbers, I had no idea that either of these existed (although I could have guessed at the second one).

What this suggests is that we need a lot more reporting from places like Ipswich and Colne, High Peak and Lincoln, and a lot less focus on now rather over-exposed places like Mansfield and Ashfield – and, when we are thinking about places where the Labour vote failed to recover in 2017, a less sympathetic focus on the MP who had, in many cases, presided over the decline in local party support for years before Corbyn was elected leader. (And that goes for Chris Williamson (Derby North) as much as for John Woodcock (Barrow & Furness).) It also suggests that when the election comes, in a lot of places we’ll be pushing, if not at an open door, at a door that we’ve already given a good shove in 2017. The next election campaign may be more winnable than we’ve been allowing ourselves to think.

Advertisement