May 2021 election benchmarks

May 2021 election benchmarks

From David Cowling – former head BBC political research

Because the 2020 local elections were postponed due to the pandemic, this May will witness elections for incumbent candidates who were, overwhelmingly, last elected in either 2016 or 2017. Both national opinion polls and projections of national vote shares at the time suggest two very different benchmarks against which to measure the 2021 results.

Those elections taking place with a 2016 benchmark are: the Scottish Parliament, Welsh Senned, London Mayor, Greater London Assembly, the city mayors of Bristol, Liverpool and Salford, Police & Crime Commissioners, and councillors in over one hundred English local authorities.

Those with a 2017 benchmark are: 6 Metro Mayors – Cambridgeshire & Peterborough, Greater Manchester, Liverpool City Region, Tees Valley, West Midlands and West of England, two city mayors in Doncaster and North Tyneside, and councillors in 21 English Shire Counties, one Metropolitan District (Doncaster) and six Unitary Authorities (Cornwall, County Durham, Isle of Wight, Northumberland, Shropshire and Wiltshire).

Westminster voting intention polls are not predictors of outcomes in local or devolved elections: the latter are different political institutions with different functions and powers and voter turnout for them is usually significantly lower than in Westminster elections. However, they do offer a broad indication of the relative popularity of political parties at various times.

In the run-up to the 5 May 2016 local elections, Westminster voting intention polls suggested fairly even levels of support for Conservative and Labour, with the Lib Dems pushed into fourth place by UKIP. The table below gives the average poll figures for March and April 2016 (the full list of polls can be found as an appendix below).

Monthly averageConLabLDUKIPGreenOths.
April 2016353271647
March 2016363371446

The polls published in the run-up to the 4 May 2017 local elections suggested a very different story: the Conservatives were 20 points ahead of Labour, the Lib Dems were narrowly back in third place and UKIP’s support had halved compared with 2016. The table below gives the average poll figures for April 2017 (the full list of polls can be found as an appendix below)

Monthly averageConLabLDUKIPGreenOths.
April 2017472710736

As United Kingdom governments have never invested in a national system for the immediate collection and tabulation of local authority votes, two estimates have emerged that regularly feature in discussion following the close of voting on polling day. Based on extensive samples of results collected in the immediate aftermath of each election, both aim to represent what the party vote share would have been if the whole country had participated rather than just the councils with elections each year.

One estimate is the BBC’s Projected National Share (PNS) which is usually broadcast during the broadcaster’s overnight election results programme. The other is the National Equivalent Vote Share (NEV) compiled by Colin Rallings and Michael Thrasher and usually published in the first Sunday Times edition following polling day.

I have set out the run of both estimates from 2010-19. They mirror the significant difference in party support between the 2016 and 2017 local elections as evidenced in the Westminster voting intention polls: both show a one point difference between Conservative and Labour in 2016, compared with an eleven point difference in 2017.

Projected National Share (PNS) of the vote (BBC) 2010-19

ConLabL. DemUKIP
2017382718  5

National Equivalent Vote (NEV) Share (Rallings & Thrasher) 2010-19

ConLabL. DemOthersUKIP
2011383716  9
2016323314  912
201739281811  4
201931311717  4

The other obvious measure of relative performance is the total of seat gains and losses by the various parties in each set of elections. Once again, there is no official collection of these results that allows a national picture to be created. Over many years the Press Association (PA) has published their own totals of council seats gained and lost by the parties, and Colin Rallings and Michael Thrasher publish their own totals. It is rare that such totals result in the same figures: the PA measures gains and losses from the status of each seat at dissolution rather than when it was contested four years earlier (if someone was elected as UKIP but subsequently joined the Conservative group on her council, then the PA would  designate the seat being defended as Conservative); Colin and Michael would not as they calculate a seat by the party that won it at the appropriate previous election (also, Colin and Michael produce notional results for re-warded seats which the PA does not). I have used the PA figures for local election results 2010-19 for the purposes of this note. Clearly, one significant factor in outcome is often the total number of seats up for election. We can see from the table below that there were almost twice as many seats contested in 2017 than in 2016.

Net gains and losses at council elections 2010-19

YearConLabLib DemOthersTotal seats
2010  -93+391-208  -504,223
2012-328+534-190  -522,412
2015+539-201-419  +559,352
2016  -48  -18  +45  +212,782
2017+563-382  -42-1395,211
2018  -33  +77  +75-1194,404
2019  -1,269  -63+676+4348,410

However, the difference in total numbers of seats does not explain the different outcomes in the two elections. The neck-and-neck race between Labour and the Conservatives in Westminster polls prior to the 2016 local elections appears to be reflected in the small net gains/losses recorded at that election. By contrast, the significant Conservative lead over Labour in polls in the run-up to the 2017 election appears to play at least some part in explaining the substantial Conservative gains as well as Labour losses which occurred then. Government parties are usually vulnerable in local elections. 

The past would suggest that, in 2021, the Conservatives are particularly vulnerable in the seats they are defending from 2017.

Whilst the benchmarks for this set of elections are either four or five years ago, we should not forget the remarkable features of the most recent set of local elections in 2019. The Conservatives sustained their biggest local election losses since the mid-1990s and Labour’s contribution to this massacre was to register a net loss of seats as well. The beneficiaries were the Lib Dems who, according to Colin Rallings and Michael Thrasher, made “their biggest single jump in councillor numbers since 1987”; and Independent/Other candidates who took 14.1% of all the seats – their highest total in at least 20 years, having only reached double figures on three previous occasions during that period. 

In the introduction to their 2019 Local Election Handbook, Colin Rawlings   and Michael Thrasher observed:

Local voters are demonstrating frustration with the behaviour of the two main parties. Together the parties haemorrhaged more than 20 percentage points from the 85% of the vote they took in combination in 2017 as English local government once again took on a more fluid and multiparty look”. It will be a matter of keen interest for many as to whether the voter frustration, so manifest in May 2019, persists in May 2021.

Westminster voting intention polls in the run up to the 5 May, 2016 local elections

April 2016%%%%%%
March 2016
ComRes18-20.31,00237357  947

Westminster voting intention polls in the run up to the 4 May, 2017 local elections

Opinium*2-3.52,0054630  9  725
YouGov*2-3.52,066482910  526
Panelbase*28.4-2.51,034473010  526
Kantar*27.4-2.51,205482411  746
ICM*28.4-2.51,9704728  8  844
YouGov*27-28.41,612443111  626
ICM*26-28.42,0124728  9  844
Opinium*25-28.42,0074730  8  736
ORB*26-27.42,093423110  8n/a9
YouGov*25-26.41,590452910  736
MORI21-25.41,004492613  417
ICM*21-24.42,024482710  735
ComRes20-24.41,001482911  327
Kantar*20-24.41,196462411  847
Panelbase*20-24.41,026492710  536
YouGov*20-21.41,590482512  536
ICM*19-21.42,027482610  835
ComRes*19-20.42,074502511  734
ORB*19-20.41,8604429  810n/a  10
Opinium*19-20.43,003452611  934
YouGov*18-19.41,727482412  727
ICM*18.41,000462511  846

David Cowling

Comments are closed.