Since the local elections last week discussion in the media has focused mainly on the impact of UKIP success on the Conservatives and how they will react. But compared with last year's local elections Labour suffered at least as much.
By comparison with the last county council elections in 2009 it does look like UKIP gains were at the expense of the Conservatives. Over 80% of UKIP seats came from the Tories, and the Tories lost a greater proportion of their seats to UKIP than the Liberal Democrats, while Labour lost barely any to them. Also there were many seats and councils where the UKIP advance was enough to help another party topple the Tories.
It also makes sense that UKIP should be taking votes from the Conservatives more than others. UKIP are a party of the right. They typically take policy positions that are currently considered to be right wing, such as being against EU membership, immigration, gay marriage and climate change mitigation. So it is unsurprising that UKIP support in the polls comes much more from former Conservative voters than from Labour or the Liberal Democrats.
But there are a number of problems with these observations as a way of looking at the results of this year's local elections. First, UKIP seat gains were very modest because of the even distribution of the vote and effects of the First Past The Post electoral system. Their vote share of 25% where they stood, not the seat tally, is the main mark of their success. Second, the Conservatives did unusually well and Labour especially badly in 2009. The Tories were bound to suffer loses and Labour bound to make some gains, as they did in 2011 and 2012. Third, attitudes on Europe, immigration and moral issues are not highly correlated with economic left-right positions. So there are plenty of eurosceptic, anti-immigrant and socially conservative voters among those with relatively socialist values who might be persuaded to leave Labour for UKIP. Fourth, while the opinion polls tell us about general election vote intention, local elections are a different matter. UKIP's share of general election vote intention, at 14% last week, is much more muted than the 23% they achieved in the Projected National Share of the vote (PNS) from the local elections.
For these and other reasons, it is more instructive to consider party performance this year by comparison with last year's local elections, rather than those in 2009. Between 2009 and 2012 we saw a change of government, a recovery for Labour and a collapse of the Liberal Democrats, a decline for the Conservatives, and, significantly, a drop in UKIP support (they were running at over 15% in the polls in the run up to the 2009 European Parliament elections but at just 8% this time last year). There are important political questions about how the trends for the Westminster parties up to last year's local elections have been disrupted by the rise of UKIP since this time last year.
If we look at the opinion polls, while UKIP did seem to benefit from Tory losses in the month after the 2012 'omnishambles' budget, there has been relatively little change between the start of May this year and the same time last year: Con -1, Lab -2, LD no change and UKIP +4. Given the usual +/- 3% confidence interval, we can only really be sure there has been a rise of UKIP. If anything it looks like it has come from Labour more than the other two parties.
By contrast, there were dramatic differences in party performance between this year and last. The PNS shows that while all three main parties are down since last year, Labour has dropped more than the other two (Con -6, Lab -9, LD -2). Labour were the biggest losers and this alone suggests that UKIP hurt Labour substantially if not more than the Tories.
Certainly we can't account for the UKIP rise since last year by reference to Conservative loses alone. Either there was a massive shift of support from Labour to the Tories and also from the Tories to UKIP or considerable direct loses from Labour to UKIP.
The pattern of the results across divisions suggests that relative to last year new UKIP votes have come as much from Labour as from the Tories. As at the general election, the change in the UKIP share of the vote since 2009 is more strongly negatively correlated with the change in the Conservative share than it is with any other party. This is as you would expect if most of the direct switching to UKIP from the Conservatives. However, if we look at the places that had local elections on the same boundaries last year and this year (combining district wards to make county divisions), the UKIP change since 2012 is equally strongly correlated with both the Conservative and Labour change, suggesting that relative to last year both parties suffered equally from UKIP progress. (See footnote.)
The socio-demographic profile of the places where UKIP did well also suggests that it took more new votes from people who would otherwise have voted Labour this year. We are used to seeing the party doing better in places with more routine manual workers who are traditionally part of Labour's base. But the correlation between the UKIP share and the proportion of working class people is stronger this year than it was in 2009 or 2012. Correspondingly, the change in the Labour share since last year is negatively correlated with the percentage of routine manual workers. So it seems that Labour is losing more of its base to UKIP.
Some may argue that Labour can afford to lose some support in its heartlands so long as it does well where it needs to win seats at the next general election. UKIP advanced more in the South and East of England and least in the North West, while there reverse is true for Labour. This is the opposite of what Labour need to do.
Moreover, comparison with 2012 is not unfair to Labour. It was not a particularly good set of local elections for them from which they might be expected to fall. Their lead over the Tories in the PNS then, and more so now, is considerably lower than at any of the local elections while Tony Blair was leader of the opposition.
One saving grace for Labour is that in divisions where they started second to, but within 20 points of, the Conservatives (i.e. Con-Lab marginals) their vote share was up by 10.5 points since 2009, compared with just 8 points elsewhere.
Otherwise, the distribution of the new UKIP voters this year bodes at least as badly for Labour as for the Conservatives. But as Nigel Farage is fond of pointing out, the main effect of UKIP is probably more psychological than psephological. At the moment it seems that the Conservatives are the most panicked while Labour are rather nonchalant. Whether either reaction is helpful remains to be seen.
Thanks to John Curtice and Rob Ford for comments and to them, Jon Mellon and the BBC for help with the data.
Footnote: Admittedly there are only 69 such divisions where the three main parties stood each time and UKIP at least one time and they fought either all or none of the constituent wards in a division in 2012. They are places where Labour are relatively strong, but they are similar to other divisions in their pattern of 2009-13 change and the switch in correlations can be seen within these cases. It is also a phenomenon we have seen before at the 2009 European Parliament elections when the change in the UKIP share of the vote was much more strongly correlated with the Labour change than with that for the Conservatives. Note also that the correlations between Liberal Democrat and UKIP change are just as strong as for the other two parties, but the Lib Dems are reaching the point where it is hard for them to fall much further than they did in 2012.