The proposed constituency boundary changes will hurt the Liberal Democrats and not help the Tories much either
With the conclusion of the AV referendum last month, focus now turns to one of the few certain electoral reforms that this parliament will contain; the redrawing of constituency boundaries, and the reduction of the number of seats by 50 to 600. Presenting recent research by Democratic Audit, Lewis Baston finds that the Liberal Democrats will suffer the most by far, and Labour and the Conservatives will suffer very similar seat reductions.
The two most striking findings of the recently released Democratic Audit model of the boundary changes were the damage it inflicted on the Liberal Democrats, and the relatively even pattern of losses between Labour and Conservative. These findings caused a certain amount of surprise in politics and the media, but they are fairly predictable from the point of view of political science, leaving aside the detail of the projection.
Overall impact of boundary changes by party
The Liberal Democrats lose the most…
The Liberal Democrats will suffer severely in boundary changes. The model suggests 14 out of 57 seats will go. This harsh result stems from two factors common to most of their seats. They tend to have smaller majorities than Conservative or Labour MPs; the mean Liberal Democrat majority is 12.5 percentage points, about two-thirds the size of the other parties. This makes it less easy for them to withstand adverse boundary changes. The other is that they tend to represent yellow islands in a red or blue sea, rather than clump together. This means that in exchanging territory with neighbouring seats, Liberal Democrat seats will tend to acquire areas where the Liberal Democrat vote in 2010 was low. The model takes out several such island seats, such as Burnley, East Dunbartonshire, Mid Dorset & North Poole and Lewes.
However, Liberal Democrat incumbents have been able to survive radical and unhelpful boundary changes in the past – David Alton in 1983, Malcolm Bruce in 1997 and Sarah Teather in 2010 all managed to engineer huge swings from the ‘notional’ result in their altered seats. Local activism has succeeded in changing the way that voters in newly arriving areas see the contest and persuaded them that they can now vote Liberal Democrat with a good chance of winning (and conversely that by voting as they did previously they might ‘let in’ the main party they dislike more). However, the context may be different now. In past changes, only a tiny number of voters would be completely unwilling to ever vote Liberal Democrat, and therefore there were a lot of persuadable voters. There will now be far more people who will never consider voting Liberal Democrat. The party has succeeded in running up the down escalator in several boundary changes in the past, but it is harder now. A regime of permanent revolution in parliamentary boundaries such as that created by the 2011 Act is a difficult environment for Liberal Democrats, even beyond the particularly tricky election of 2015.
… but the Labour and Conservative losses are even
Net losses in seats for Labour and the Conservatives came out more or less even in the model, with Labour down 18 and the Conservatives down 15 overall (it was 17 and 16 respectively before a late tweak to the model in Warrington).
This finding, too, should not be very surprising for solid reasons of political science. Variation in constituency size is a very minor contributor to the pro-Labour bias in the electoral system. It should be expected that a neutrally-implemented measure altering the size of constituencies should therefore not do very much to alter the balance between Labour and Conservative. Constituency size was worth something like six seats in Labour’s favour in 2010, and seven against the Conservatives according to Rallings, Thrasher, Borisyuk and Johnston)
Even before looking further at any local detail, a net effect of 13 seats (11-12 when applied proportionally to a smaller House) is small beer. A lot of the existing bias stems from the over-representation of Wales – but cutting down Welsh seats then will probably have a disproportionate effect on the Welsh footholds the Conservatives currently enjoy, as they are mostly either marginal or obvious candidates for merger. It should also be borne in mind that proportionately, Labour loses 7.0 per cent of its seats, while the Conservatives are down only 4.9 per cent. The scale of change in the Democratic Audit model is within the bands one might expect from first principles, allowing only for Welsh local factors.
Looking forward to 2015
The boundaries model is static – it assumes that everyone voted as they would have done in 2010. While the result and pattern of support in 2015 may resemble 2010 more than most people expect, there will be significant changes in electoral behaviour. It is impossible to predict what might happen in terms of the swing between Labour and Conservative. But it is likely that the Liberal Democrat vote will fall, and that the SNP vote will rise, and that the regions of England will continue to polarise between a Labour north and a Conservative south. How the new boundaries translate the national swing and its local variations into seats is more important than what would have happened in 2010, but much harder to estimate.
The Democratic Audit model may not have changed the 2010 election result in any important way had the boundary changes been in place then, but it might look different in 2015. This is one of the reasons why the predictable chorus of ignorant people claiming it is an unduly pro-Labour map is wrong. It is not in Labour’s interests to maximise its notional number of seats in 2010, but to get the best result possible in seats when the national shares of the vote of Labour and Conservative are fairly close. This was the underpinning of Labour’s successful boundary review strategy in the 1990s. Labour would be best off creating Tory marginal seats – ideally ones with sitting Labour MPs to benefit from the incumbency effect!
It is difficult to be precise, but the Democratic Audit model appears to depopulate a vital area for Labour – seats with narrow Conservative majorities that would be susceptible to a small to medium sized pro-Labour swing. Of the 50 top target Tory seats for Labour on current boundaries (winnable with swings of up to 4 per cent), 15 would see Labour’s prospects improved, 17 would see the Conservatives strengthened instead, 14 would be unchanged or not significantly politically altered, and 4 would be abolished – a fairly even spread.
Of the seats improved for Labour, 8 of them would be flipped to the party on the notional 2010 results, and therefore not available for gaining in 2015. Eight of the seats are also so much improved for the Tories that they would no longer count as marginals. This makes 20 seats taken out of Labour’s easiest target list. In exchange, four new targets result from Labour seats being flipped into being Tory marginals by boundary changes, and probably four previously safer Tory seats now made marginal. The number of Tory seats vulnerable to a pro-Labour swing of 4 per cent or so falls from 50 to 38. The hill Labour has to climb to get an overall majority is therefore steeper than it would be under existing boundaries, while it is probably a little easier for the Conservatives.
This was probably the intention of the more sophisticated Conservative supporters of this proposal – to put a finger on the scales a little by systematically adding bits of rural territory to marginal seats in towns outside the big metropolitan areas (e.g. Great Yarmouth, Harlow, Redditch, Stevenage, Lincoln, Stafford, Tamworth, Brighton, Dover – and indeed Bath and Southport…). They may have been less aware that there will be locations where increased size of seats might lead the Boundary Commission to create a core urban (Labour-inclined) seat rather than two increasingly Tory marginals (as for instance in Thanet and Norwich in the Democratic Audit model).
The Democratic Audit model illustrates the simple principle that altering something (constituency size) which is not much to do with the problem you are seeking to address (electoral bias) is unlikely to achieve your aim. It also illustrates some of the complexities in measuring the effects of boundary changes. It may superficially look a better outcome for Labour than expected, but there is a story below the headline numbers.