Last summer, I looked at how the Labour vote in Conservative/Lib Dem marginal had fallen to far lower levels than occurred to the Lib Dem vote in Labour/Tory marginals or to the Conservative vote in Lib Dem/Labour marginals:
I concluded that if this was as a result of tactical voting for Lib Dem candidates by Labour supporters, as recommended by Billy Bragg, the Conservatives could look forward to substantial numbers of seat gains at the next election if Labour voters decided not to continue with their tactical voting.
The critical question, therefore, is to establish just how much tactical voting of this type was taking place, and then we might be able to assess how much unravelling we could expect. Easier said than done. PB superstar Andrea suggested to me that I might look at the overall vote tallies in whole counties at successive elections and this is what I have done.
Andrea suggested that I looked at Devon. At the 2010 election, there were 12 constituencies in Devon. 601,300 voters made their way to the polling booths and 85,556 of these cast their vote for a Labour candidate: 14.2% of the vote.
No two general elections are exactly comparable, but 1992 is probably the easiest election to compare with 2010. The Conservatives weren't too far ahead of Labour and, importantly for our purposes, the Lib Dems were new on the scene so a culture of tactical voting would not be as mature as it is now. How have things changed?
There were 11 seats in Devon in 1992. 638,104 voters cast their vote and 122,854 of these voted Labour: 19.3% of the vote. Evidently, there has been a decline in the Devon Labour vote in that time – but of course, Labour polled better nationally in 1992 than in 2010, so that doesn't prove much. Labour polled 29.0% nationally in 2010 and 34.4% nationally in 1992. Making the appropriate adjustments, we could have expected the Devonwide Labour vote to have been 16.2% in 2010 if there had been no additional tactical voting. Since it was in fact 14.2%, that suggests that there might have been some additional tactical voting, but it's hardly overwhelming evidence.
What if we strip out those seats where Labour was out of contention? Labour's vote in Devon is very efficiently distributed. On its puny vote share in 2010, Labour managed two MPs out of 12 (by comparison, the Conservatives tallied 16.7% of the vote in Scotland and managed one MP out of 59). Labour was a reasonably close second in a third seat. Nicely done.
In 1992, Labour had one MP and finished a reasonably close second in three other seats. Not quite so efficient, but then it had to overtake the Lib Dems in all four of those seats. A pretty good effort really.
The numerate will have realised that means that Labour's performance elsewhere in Devon was at joke party levels. In the remaining nine seats in 2010, 463,633 voters cast their votes, but only 36,280 of these were Labour votes: 7.8% of the vote. Labour was pushed into fourth in two of these constituencies by UKIP. This sounds strongly indicative of tactical voting. Was it?
In 1992, there were seven Devon seats in which Labour finished third. It gathered 55,221 votes out of 431,028 cast in those seats): 12.8% of the vote. So it should be noted that Labour was already performing dismally in rural Devon, even then. If we make the same adjustment to the vote share to allow for the differential performance of Labour nationally, we would have expected it to have got 10.8% of the votes in these constituencies. This suggests that an additional 3% of Labour's vote has been lost to tactical voting – not as much as I originally might have expected to see.
Of course, Devon is just one county, so I thought it worthwhile to look at Dorset and Somerset. In Dorset, Labour didn't take a single seat in either 1992 or 2010. In 1992, Labour took 54,613 out of 407,472 votes (13.4% of the votes), while in 2010 it took 47,594 out of 391,111 votes (12.2% of the votes). Allowing for its relative performance at these two elections, Labour actually did better in Dorset in 2010 than it did in 1992, and this remains true even if you strip out the 2010 result in South Dorset (where Labour previously held the seat in 2005). Billy Bragg's appeal for more Labour tactical votes appears to have fallen on deaf ears: a prophet without honour in his homeland.
Somerset tells a different story. Labour was neither first nor second in any seat in Somerset in either 1992 or 2010. So you might expect to see more evidence of tactical voting here. And you do. In 1992, 38,561 Labour voters dutifully wasted their vote out of 297,857 voters: 12.9% of the vote. In 2010, this shrivelled to 22,163 out of 286,279: 7.7% of the vote. You would have expected to see 10.9% of the voters going red in 2010 if national trends had been followed. The Lib Dems now hold four seats out of five in Somerset, where in 1992 they held one. It seems reasonable to conclude that tactical voting has increased.
What can we conclude? First, that Labour have polled appallingly in these three counties (outside Plymouth and Exeter) for a generation. There comes a point where a Labour tactical voter is indistinguishable from a lifelong Lib Dem voter. Secondly, that there has been increased tactical voting, at least in Devon and Somerset, but that it has been on nothing like the scale that some, including me, have assumed. Maybe 10 Lib Dem seats nationally are at risk to the Tories from an unwinding of the increased tactical voting, but to lose more, the Lib Dems would have to lose long term Lib Dem voters as well as reds in yellow clothing.
Which leaves one further question: just who are all these Labour supporters who claim to have voted tactically for the Lib Dems?