How close is the first election for West Midlands Mayor?

Since the inaugural elections for metro mayors in three city regions were announced — in Greater Manchester, the Liverpool City Region and the West Midlands — the latter has been talked up as the one to watch in 2017. Unlike the first two, the West Midlands has been said to be one of the closest. The Conservative candidate has consistently been mentioned by the prime minister during conference speeches and in parliament. Since November Andy Street has been the favourite with Ladbrokes; his odds are currently 4/9 against 15/8 for Sion Simon, the Labour candidate. But how likely is such an upset on 4 May 2017?

While it’s a new role, we can use publicly available election data to initially look at previous results in the area. Next we can consider the effects of the voting system, before looking at how turnout and additional factors may influence the results.

Electoral history

The seven boroughs that vote for the mayor this year — Birmingham, Coventry, Dudley, Sandwell, Solihull, Walsall, and Wolverhampton — had a total electorate of 1.93 million in May 2016. We can look at election results over the last decade in these boroughs to judge each party’s performance.

Firstly we can see that Labour has outpolled the Conservatives in every local election since 2010. From a high of 49.5% in 2012, the Labour vote declined to 41.2% in 2015, before reviving to 46.3% in 2016. However, the Conservatives entertained narrow leads over Labour in 2006 and 2007 and a larger lead of 9.3% in 2008 during the financial crisis and amid relatively high results for the Liberal Democrats and the BNP. On the face of it, it looks like it’s a question of whether Labour is performing as poorly as 2008, or akin to recent years, including in 2016 despite the party’s national problems.

Total vote shares in West Midlands local elections since 2006 (author’s calculations and data from

Likewise, looking at general election results for the 28 contingent parliamentary constituencies, Labour narrowly outperformed the Conservatives in 2010 and 2015 despite general election defeats. Labour polled 37.6% to 33.5% in 2010 and 42.8% to 37.6% in 2015.

This suggests that Labour should be in the lead, but a sufficient swing to the Conservatives may be enough for them to achieve first place, particularly if conditions are similar to 2008. Based on local election results, the Conservatives would require an 8.6% swing compared to 2016 (or 5.0% compared to 2015) to take the lead.

Redistribution effects

However, the analysis so far has ignored the actual system in use. Rather than first-past-the-post (FPTP), where the candidate with the most votes wins, the supplementary vote (SV) system enables voters to put a first and second preference. If no candidate receives more than 50% of votes in the first round, all second preferences are allocated to the top two candidates. The winner is the remaining candidate with the most votes.

This means that how votes are redistributed can determine the final winner if no party achieves more than 50% in the first round. Election results for the West Midlands Police and Crime Commissioner (PCC) in 2012 and 2016 can indicate how they tend to redistribute.

In 2016, Labour polled 49.9% of valid votes cast with turnout of 30.0%. While unhelpful for those at the count, it enabled us to see how the 134,329 votes cast for UKIP and an independent candidate were allocated to Labour and the Conservatives. It’s worth noting only 65,177 votes were actually redistributed to either remaining candidate. The rest were spoilt (17,621) or had no valid second preference. In the end Labour picked up 47.6% of valid second preferences compared to 52.4% for the Conservative candidate.

In 2012, three independent candidates plus UKIP and the Liberal Democrats led to Labour polling 42.0% in the first round against 18.5% for the Conservatives on a 12.3% turnout. 52.1% of redistributed votes went to eliminated candidates; only 30.6% went to the Labour or Conservative candidate. In the end Labour picked up 60% of redistributed votes compared to 40% for the Conservatives.

What does this indicate?

  • A candidate with a sufficient lead in the first round is hard to beat.
  • In a Labour-Conservative runoff, redistributions either favour Labour (as in 2012) or are broadly equal (as in 2016).
  • Many second preferences are wasted if not given to Labour or the Conservatives (presuming they make the final two). This is exacerbated in a crowded field — 30.6% votes were successfully redistributed in 2012 compared to 48.5% in 2016.

Turnout factors

Might turnout on the day affect the result? It has been suggested that because concurrent local elections in the seven-borough West Midlands are not taking place in 2017, that could reduce turnout and alter the election result.

The PCC elections indicate this may not have a large effect. The initial election in 2012 plus a by-election in 2014 saw turnout as low as 12.3% and 10.4% respectively. In the first round Labour’s lead over the Conservatives was 23.5% and 24.0% in each case. In the 2016 election, despite turnout of 30.0% (when it was run alongside council elections) the lead was still 24.0%. The second-round lead for Labour was 19.9% in 2012 and 26.8% in 2016.

Even looking at local election results, there is limited correlation with turnout. The 9.3% Conservative lead in 2008 occurred with 32.5% turnout while Labour’s decade-high lead of 21.8% in 2012 coincided with 28.2% turnout. Plotting the Labour lead against turnout does not appear to show much difference when looking at local elections outside general election years (in blue, below). On the other hand, when looking exclusively at results since 2010, Labour’s lead appears to diminish a little at slightly higher turnout levels.

First-round PCC results (in orange) tend to have a comparatively large lead whatever the turnout. Local elections during general election years (in purple) do not appear to differ much, though Labour’s leads are a little lower compared to other post-2010 results.

Labour lead over Conservative against turnout in West Midlands elections since 2006. Orange for PCC elections, purple for local elections in general election years (data from Rallings and Thrasher and

Lower turnout may have a small effect in narrowing a party’s lead, but other factors appear to contribute more to the final result.

Additional factors

Just as when PCCs were implemented in 2012, it has been argued that a ‘protest vote’ by voters who disagree with the existence of the role may be a factor. While 12 of 41 PCCs initially elected across England and Wales were independents, in 2016 this declined to just 3. In the West Midlands area, three independent candidates together polled 25.7% in 2012 and none managed to make the second round. In 2016 the lone independent polled 7.3% of the vote. In fact, a ‘protest vote’ may more likely be exhibited through abstention, as may have been the case in 2012 or the 2014 by-election, though notably not in 2016. It’s not clear whether a ‘protest vote’ helped or hindered any particular candidate.

Could it be party over person? In 2016, Labour’s 49.9% for PCC was not dissimilar to its 46.3% in local elections on the same day. The difference could be explained by the lack of a Liberal Democrat or Green candidate for PCC. Turnout for both elections was effectively the same. This suggests that once a voter is at the polling booth, they are prone to revert to party allegiance.

Findings and concluding thoughts

  • Since 2010, Labour has led in local elections in the seven-borough West Midlands. But this is tempered by a Conservative edge before 2010, particularly in bad years for Labour nationally. A swing of 8.6% compared to 2016 (or 5.0% compared to 2015) would give the Conservatives a lead.
  • Redistributed votes have tended to favour the first-placed candidate due to poorly allocated second preferences. Recently this has favoured Labour.
  • A crowded field can make it harder for second preferences to affect the result due to a high level of discarded votes. So far six parties have declared candidates.
  • Turnout in itself does not appear to have affected previous election results, though slightly higher levels may narrow a party’s lead.
  • It is not clear whether a ‘protest vote’ has much effect on the overall result.

It would appear that victory for the Conservatives requires them to lead in the first round and hold onto that lead in the second round. If the Conservatives cannot achieve a first-round lead, their path to victory depends on them gaining sufficient second preferences from other parties’ voters, or for enough Labour-inclined votes to be discarded due to poorly allocated preferences in the second round.

The election for West Midlands Mayor will certainly be closer than its counterparts in the North of England. But taking into account previous elections, the voting system, turnout and further factors, probably not as close as people seem to think.