Why it doesn't really matter who wins

For the first time ever in Britain, on 5 May 2011, voters will decide how the House of Commons is elected. The referendum vote would seem a welcome extension to Britain’s democracy. After all, it hardly appears healthy that our democratic representatives should decide for themselves how membership of their club is determined.

Moreover, the referendum apparently offers voters a clear choice. The two official campaigns have marshaled arguments about the merits and demerits of the Alternative Vote (AV) that appear to be diametrically opposed to each other.

According to the ‘Yes’ camp, MPs would have to work harder under AV. There would be fewer safe seats and victory could only be achieved by winning a majority of the vote. Consequently, our elected representatives would be more accountable to voters.

The ‘No’ camp, in contrast, says that under AV many a candidate who succeeds in coming first would find their ‘victory ‘overturned unfairly by the lower preferences of those who have initially backed fringe candidates. At the same time, there would be an endless series of hung parliaments in which the Liberal Democrats would be the perpetual kingmakers.

So the two camps agree that switching to AV would make a considerable difference. They simply disagree about whether the change would be desirable. But are they right in their shared assumption about AV?

AV could make a difference in two ways. First, voters might not give their first preference AV vote to the candidate against whose name they would have marked an ‘X’ under FPTP. Some parties might profit and others lose out as a result.

Second, even if voters’ first preference votes and their ‘X’ votes were to be the same, the second and subsequent preferences they express under AV might alter the outcome, turning many a ‘victor’ into a ‘loser’ and significantly altering the partisan composition of the House of Commons.

In practice, neither effect seems likely to be as marked as either the proponents or the critics of AV suggest.

In theory, as the ‘Yes’ camp are keen to point out, AV allows voters to support a small party that has little prospect of winning without running the risk that in doing so they might make it more likely for the party they least like to win. If their first preference candidate does badly their vote will be transferred to whichever of the front runners they would prefer to win.

However, theory and practice are not always the same. Voters are already supporting smaller parties in record numbers despite the disincentives to doing so under FPTP. Last May, no less than one in ten votes in Great Britain were cast for parties other than Conservative, Labour or the Liberal Democrats.

This suggests switching to AV might not make much difference to how voters vote after all. Shortly after last year’s election, the academic British Election Study found that, when asked to complete a mock AV ballot paper, no less than 87 per cent backed with their first preference the party for whom they had voted in the election under FPTP.  Moreover, the proportion of people in England supporting one of the smaller parties was, at nine per cent, unchanged.

Much the same result was obtained by YouGov when last July they asked people both their current vote intention under FPTP and their first preference party under AV. Once again, overall support for the smaller parties was, at nine per cent, identical under the two systems.

Still, even if voters’ first preference votes and ‘X’ votes are likely to look much the same, the second and subsequent preferences they express under AV could still change the outcome. If one party is persistently the second preference of those who back other parties it might well be able to win more seats than under FPTP.

At every election since 1983 voters have been asked not only how they were voting but also which party would be their second preference. This evidence indicates that the Liberal Democrats are a popular second preference. At every recent election roughly some 60-70 per cent of Labour supporters have said the Liberal Democrats were their second preference. Much the same also used to be true of Conservatives, though in their case the proportion has dropped somewhat at more recent elections to around 50 per cent or less, not least because for them UKIP has increasingly become an alternative second attraction.

Yet the impact these Liberal Democrat second preferences would have had on the outcome of previous elections is limited. Remember that, by definition, it is only the second and subsequent preferences of candidates that have come third or worse in the election that count. That means that typically we are talking about a relatively limited pile of votes that might be transferred.

Moreover, relatively popular as they may have been as a second choice, the Liberal Democrats have still been far from unanimously preferred. Consequently the net benefit the Liberal Democrats would have derived from transfers would rarely have been worth as much as half of the transferred vote in a constituency, and not always even as much as that. Therefore transfers are only likely to have made a difference where the Liberal Democrats were already a close second under FPTP — and the number of seats where that has been the case has typically been quite limited.

Thus, when we use these survey estimates of what people’s second preferences would have been at previous elections in order to estimate the likely outcome if AV had been in place at previous elections, we discover the boost the Liberal Democrats would have received is typically quite modest.  

As table 1 shows,  at most elections it would have been no more than some 20 seats or so. Only in 1997, when (relatively numerous) Labour voters were particularly keen to gang up on the Conservatives and when the Conservatives were vulnerable to a Liberal Democrat challenge in more constituencies than usual, would the Liberal Democrats have derived a major boost from the system.

Yet even then, and indeed at every other election between 1983 and 2005, the boost to Liberal Democrat representation would apparently not have been sufficient to produce a hung parliament. Indeed, in four of those seven elections the government would have had an even bigger majority.

The explanation is simple. The second preferences of Liberal Democrat voters have tended to follow the general tide. Up to and including 1992 they were more likely to prefer the Conservatives to Labour. So, when transferred, their second preferences would have helped the Conservatives claim some marginal seats from Labour and thereby more or less compensate the Tories for losses to the Liberal Democrats. Since 1997, in contrast, Liberal Democrat supporters have preferred Labour, albeit only narrowly so in 2010, thereby enabling Labour to win more seats.

So the historical record suggests a very different picture from what we might have expected.  Although AV can be expected to produce rather more Liberal Democrat MPs than under FPTP, the boost the party would receive from AV will not necessarily be sufficient to produce a hung parliament. Indeed, AV can sometimes produce even bigger single party majorities than does FPTP.

It follows, too, that the outcome in most individual constituencies would also be the same under FPTP as under AV. The survey based estimate for 2010, for example, suggests that the winner would have been different in just 30 seats. An estimate based on the British Election Study mock ballot is, at 43 seats, a little higher, but still constitutes no more than seven per cent of all parliamentary seats in Great Britain.

So claims by the ‘No’ side that AV means that parliament would contain many an MP who had only been elected thanks to the second and subsequent preferences of those who initially backed fringe candidates seem wide of the mark. But so equally does the claim by the ‘Yes’ camp that many an MP’s seat would be less safe, thereby forcing them to reach out to a wider section of the electorate.

Moreover, transfers may have even less impact than the estimates in our table suggest.  Asked in a survey what their second preference would be, around four in five people give an answer. Much the same is true if they are invited to complete a mock ballot.  But when voters in Scotland have voted under AV in local by-elections, as they have done since 2007, only between a half and two-thirds have expressed a second preference. Less than two in five put down a third.

If that were to happen in Westminster elections, the outcome under AV would be even more similar to that under FPTP than our table suggests.  There would be fewer transfers capable of upsetting the FPTP applecart. It would mean too there would often be too few transfers to enable any candidate to pass the 50 per cent hurdle. Even under AV the winner can still end up with only a minority of the vote — as has been the case in more than one in three Scottish by-elections.

In reality, the similarities between AV and FPTP are more striking than their differences. Both are majoritarian systems in which the reward a party secures in seats depends on where it wins votes as well as how many. Voters are, in truth, not getting much of a choice after all.

John Curtice is Professor of Politics, Strathclyde University.