There is no doubt that Labour’s performance in the Copeland by-election last week was historically bad: a government last won a by-election seat off an opposition in 1982. But Jeremy Corbyn’s most die-hard supporters have tried to explain this poor result away. Do their arguments hold any weight?
You have read the headlines: this is one of the worst by-election results for a main opposition party ever. The last time a governing party won a seat of an opposition party at a by-election was Mitcham & Morden in 1982: at the very slump of Labour’s early 1980s’ despair, with the Tories’ benefiting from both from a war-bounce from the on-going Falklands conflict and Labour’s vote being split with the newly formed SDP (the resignation of the sitting Labour MP to stand for the SDP caused that election). Copeland and its predecessor constituency, Whitehaven, had been in Labour hands since 1935.
The shifts in votes were hardly seismic, as shown in the graph below comparing the 2015 general election outcome in the constituency (left-hand columns) and the by-election result (right-hand columns), but the outcome certainly was.
Until the by-election, constituency was the geographically largest Labour-held seat in England, covering a large swathe of the Lake District and Cumbrian coast. It is home to Sellafield and England’s highest mountain, Scafell Pike (together with its second highest and part of its third). Population-wise of course, few people live in those areas and the majority of the population live in the west Cumbrian town of Whitehaven and its neighbours – Cleator Moor, Distington and Egremont. The other significant towns in the constituency are Keswick to the east and Millom to the south.
Of all the excuses trotted out by the Corbyn faithful, the easiest to dismiss is that Storm Doris on the day meant that Labour voters stayed at home. It is certainly true that turnout in the by-election was lower than the last general election in the constituency: 51.3% compared to 63.8% in 2015. But by-election turnouts are invariably lower than those at general elections. In fact, turnout was pretty good for a by-election: of the 31 held since the 2010 general election, only four (Eastleigh, Mid Ulster, Newark and Richmond Park) have been higher. The highest, Mid Ulster, was 55.7% (not much more) whilst the lowest (Manchester Central) was a measily 18.2%.
Perhaps a better measure than absolute turnout is to look at how much turnout dropped since the preceding general election. The drop of 12.5% in turnout at Copeland is remarkably small: only two of the same 31 by-elections had smaller drops: the remarkable (if under-reported in Great Britain) Mid Ulster by-election and the Stoke Central by-election also held last Thursday. The graph below shows those 31 drops – whilst it is perhaps unfair to compare to Batley & Spen (when Labour was given a free run by the other parties following the murder of its former MP Jo Cox), the average of the other 30 is a drop of 21.1%.
(The colours represent the winning party – Labour (red), Sinn Fein (green), Respect (dark red), Lib Dems (gold), Conservatives (blue) and UKIP (purple)).
It certainly seems as if Cumbrian voters are made of stern enough stuff that a storm did not put them off voting. The reality is probably that any voters that the storm may have deterred just went along to the polling station after it had passed.
It is no longer a natural Labour seat – part 1
It is certainly true that the vast rural swathes of the constituency do not vote Labour at either a local or national level. The map below, of the last council elections, does at first glance look very blue.
But look a little closer: the blue bits are the very sparsely populated areas, often with a single councillor. More densely populated wards can have two or three and are more likely to be redder in shade. In fact, of the 57 councillors within the constituency, 30 of them are Labour and 21 are Conservative.
This pattern of votes also shows the fallacy of the Storm Doris story where Labour voters stayed at home whilst car-driving Conservatives dd not: in fact, Conservative voters in rural areas are likely to have found it harder to vote during the storm than town-dwelling Labour ones.
As the graphs below show, there has always been a reasonable Conservative showing in the seat, albeit never enough previously to overtake Labour.
Labour’s support has been in long-term decline in the constituency
There is certainly some truth in this, as the graph below illustrates. This shows the percentage vote by party in every general election, together with the by-election, since the constituency name of Copeland has been used.
Labour’s support climbed from 44% in 1983 to 58% in 1997, and has been sliding slowly back down ever since. The balance between the two leading parties was altered slightly in 2010, when the area around Keswick was added to the seat, helping the Conservatives a little.
But what appears in percentage terms to have happened last week is that 2015 UKIP voters moved to the Conservatives whilst Labour voters switched to the Lib Dems. It is possible therefore that there is some post-Brexit voting behaviours on display here.
Taking turnout into account, this may or may not be the case. It is certainly true that the Lib Dems gained votes between the elections: about 600. The number of Conservative votes stayed fairly constant, whilst Labour’s and UKIP’s fell considerably. It may well be therefore that some previous Labour voters switched to the Lib Dems over Brexit whilst others stayed at home – more likely through of their concern about Labour than the weather though.
The graph below shows the changes in the number of votes cast between elections. Note that both the Labour and Conservative tallies rise in 2010 due to the boundary changes.
What the graph also shows is that the margin between the two parties has often been slight, outside of the Blair era.
Boundary changes mean than the comparison to 1935 is not accurate
Some left-wing commentators have commented that comparing the results for the old Whitehaven constituency is not fair.
In fact, the old Whitehaven constituency hardly changed at all between 1935 and its abolition in 1983. When Copeland was created in that year, it had precisely the same boundaries as the previous Whitehaven the constituency – the name change was just a cosmetic one to give the seat the same name as the local authority district that it was co-terminous with.
The sole significant boundary change to the constituency at all was in 2010, when four wards from the neighbouring Allerdale district were added to the seat: Crummock, Dalton, Derwent Valley and Keswick. Whilst one of Keswick’s councillors is Labour, the other wards are solidly Conservative.
In 2010, the BBC estimated that this change cost Labour 3.8% of the vote from the 2005 result whilst the Conservatives gained 1.8%, so a net swing of 2.8% or a combined effect of closing the gap by 5.6%.
Playing those numbers back into the pre-2010 results does show an impact: that change in votes would have in fact meant that the Conservatives would have narrowly won in 1983, 1987 and (just) in 1992. Labour would also have lost the 1935 election in Whitehaven with that shift, but held it in every subsequent election until the 1983 name change.
So, there might be some validity in the argument to say that it is not quite accurate to say that an identical seat was held by Labour since 1935. That said, the result on Thursday was bad enough that even without the 2010 boundary change Labour would still have lost the by-election.
It is no longer a natural Labour seat – part 2
It is worth asking the question whether Copeland is a natural Labour seat given what the party is now seen as representing. Copeland is reasonably unusual for a Labour seat in having a nuclear power station as such a significant employer (not quite unique – the same might be said for Ynys Mon in Wales). Clearly, voters’ concerns about Corbyn’s perceived views on nuclear energy may have had a particular role to play in this constituency.
Another factor may be demographics. Whilst Copeland is obviously a lot more white than Jeremy Corbyn’s Islington constituency, it is also a lot older and older voters are (1) most likely to vote and (2) most likely to disapprove of Corbyn: Theresa May has a 68% lead among over 65s on who they think is a better leader.
In Islington, that differential among older voters does not matter much: over-65 year-olds make up less than 9% of the population. But in Copeland, they are over 21% of the population: you are 2.4 times as likely to be 65 or over in Copeland than Islington.
That statistic should make Labour worry about its old industrial heartlands. At the next election, it may hold on to its urban and university seats with their younger than average demographics but those old heartland seats in the north and elsewhere tend to have populations that are older than the average.
Labour did not lose Copeland because of the weather. Whilst Labour’s vote has declined over time in the seat, it declined rapidly more last Thursday and that is an insufficient answer. So too are the 2010 boundary changes, although it is clear that they mean that the narrative about Labour holding the seat for 82 years is not quite true.
The reality is that a party threatening to close the constituency’s local A&E department still won this constituency off Labour when no sitting government has done the same since at least 1982 and arguably has not done as well in a by-election since the Conservatives won the Worcester by-election in 1878.
No-one blames the candidate – indeed there seems to have been universal praise for local councillor Gillian Troughton. Whilst part of the problem is that Labour’s message has not resonated for some time in seats like Copeland it is also clear that the message has weakened much further under its current leader.
Part of that may have been the nuclear debate, but if the by-election is not to herald a swathe of defeats in the next general election it needs to work out how to appeal again to older voters.