“It’s time to come home,” the Prime Minister recently implored Ukip voters, and he has been joined by Boris Johnson. Tory strategists hope this ‘vote Ukip, get Miliband’ message will scare enough people that it transforms purples into blues and hands Cameron an election victory on Thursday.
But I’m unconvinced. The alchemy doesn’t work. Only if Ukip almost disappears would a combined Tory-Lib Dem coalition win, assuming Clegg is willing to renew his marriage vows.
Tories presume Ukip voters are natural Tories whose ranks are plentiful enough to tip the balance of seats Cameron’s way. Well we can test this by seeing what might happen if Ukip collapsed in the UK’s marginal seats, as measured by the 130 whose pulse has been taken by Lord Ashcroft’s constituency polls.
If Ukip fade, who benefits?
I took a look at this for May2015 a few months ago. That research showed how Ukippers were most likely to have voted for Cameron in 2010, but many also voted for Blair in 1997. The Tory plea for Ukippers to ‘come home’ paints too simplified a picture.
Let’s look again at how Ukippers voted in 2010. We have two useful sources for this: Ashcroft’s nationwide polls, which provide a breakdown of respondents’ 2010 recalled vote, and the British Election Study (BES).
A rough guess of Ukippers’ second preferences: three ex-Tories for each ex-Labour voter.
By using Ashcroft’s national polls we can gather together 1,500 people intending to vote Ukip from the 19 surveys he conducted between September 2014 and late April 2015. Among these 1,500, we find that of those who voted for the three main parties, 45 per cent are former Tories, 29 per cent ex-Lib Dem and 26 per cent ex-Labour.
Next I used data from the BES, which was collected in three ‘waves’ throughout 2014. Its advantage is sample size: it surveys 24,000 people per wave. That’s mean we have around 4,000 Ukip voters per wave – 12,000 in all.
Two findings stand out. First, there’s very little difference across the three waves when it comes to the 2010 vote of Ukippers. Second, they overwhelmingly backed the Tories in 2010: 63 per cent for Cameron, 15 per cent for Brown and 21 per cent for Clegg.
The 4:1 Tory-Labour slant is twice as big as in Ashcroft’s numbers. This is a surprisingly large gap, but one which mainly reflects the different methodology and sample size of the two surveys. Averaging BES and Ashcroft gives a rough guess of Ukippers’ second preferences: three ex-Tories for each ex-Labour voter. We also account for Ukippers who didn’t vote or voted for minor parties.
What happens if Ukip fall to 7 per cent?
Now consider the most daring of plausible ‘Ukip demise’ scenarios: its 14 percent of the vote halves to 7 percent. As Caitlin Milazzo and Matthew Goodwin note, this could be the case if the half of Ukip voters who say they are not ‘highly certain’ to vote for the party defect to their second preference.
If we use the vote split data from Ashcroft’s national polls – a 2:1 Tory:Labour split among Ukippers – a halving of the Ukip vote gives the Tories five more marginal seats than suggested by Ashcroft’s seat polls: Thurrock, Torbay, Rossendale and Darwen, South Ribble and Norwich North. (It also puts them further ahead in Pudsey, where a recent Ashcroft poll put the Tories ahead by 1.)
Halving the Ukip vote isn’t a significant enough impact for Cameron.
If we use BES data – and assume a 4:1 Tory advantage – the Tories another: Halesowen & Rowley Regis (and shore up their slim lead in High Peak). So the Tories’ net gain on Labour is 5-6 seats.
May2015’s current election prediction is 273 Tory, 268 Labour. Because May2015 moves national polls every day in-line with changes in national polls, two of the seats the Tories would gain have already moved into the Tory column.
If we add the other four – Thurrock (from Ukip), Torbay (from the Lib Dems), Norwich North and Halesowen & Rowley Regis (both from Labour), the Tories rise to 275, with Labour falling to 267.
This isn’t a significant enough impact for Cameron. To reach the mid-280 mark he needs, he will either need a greater national poll win than May2015 currently predicts, or a greater first-time incumbency effect than Ashcroft has found.
One reason for the limited impact of a Ukip demise is that a significant minority of Ukip’s current voters did not vote for a main party in 2010 and, according to BES, do not list a main party as a second preference. So even if they leave Ukip, they do not affect the race.
Only if UKIP were wiped to zero does the Tory bloc win.
Only if UKIP were wiped to zero, which delivers a Tory gain of 16, Labour loss of 11 and Lib Dem/Ukip loss of 5 relative to Ashcroft’s polls, does the Tory bloc win, and then by a slim 325-319 margin.
The demise of the Lib Dems means the Tories have exchanged a seat-rich coalition partner for a seat-poor one. This could produce a Canadian-style divided right in which conservatism is out of power for a generation. (Or perhaps Boris will become leader, win 40 per cent and have a majority.)
The danger for the system is that centre-right and right-wing voters, if we include the centrist Lib Dems, may make up 55-60 per cent of the electorate but hold a minority of seats. The only cure for that is proportional representation – an unlikely prospect.
Eric Kaufmann is Professor of Politics at Birkbeck College, University of London. His latest publication is a Demos report, freely available, entitled Changing Places: the White British response to ethnic change.