Monday, 17 October 2011

Everything is Arbitrary

Well, okay, not quite. But this post was prompted by a Twitter discussion with Kevin McNamara (@WoolyMindedLib) in relation to the changes to constituency boundaries overseen by the Boundary Commission. The aim was firstly to reduce the number of seats in the House of Commons to 600 from 650 or so and secondly to equalise constituencies (that is, to make the number of eligible voters in each single-member constituency as consistent as possible).

The main bone of contention Kevin seemed to have was that whilst he agreed with equalising constituencies the 600 figure was "arbitrary". He agrees that there's scope to reduce the overall number of MPs, but he argued that it was better to equalise the constituencies first.

There are a few issues with this. First of all, 600 is no more or less "arbitrary" than any other number of constituencies. They could have picked 100 if they really wanted to, although most would argue that single-member constituencies that large would not be appropriate for a body wielding so much power. It may be that you think the number of seats is greater than is necessary to represent adequately whilst maintaining efficient decision-making But anyway, the main point is that 600 is no more "arbitrary" than 601, 623, 646 or 19.

The second issue is that you should equalise the constituencies before reducing the number of MPs. The problem with this is that you sort of, well, can't! By reducing the number of constituencies, you definitionally have to divide up existing constituencies into new ones or, well, disenfranchise all of those in the abolished constituencies! You can only change the number of MPs by changing the boundaries so that they represent a larger or smaller proportion of the electorate.

If you start a review with a view to equalising constituencies, you need a number in mind for roughly how many constituents you want in each constituency. The thing is, this is directly related to the number of MPs you're going to end up with. If A is the total eligible voting population, B is the target standard number of eligible voters in each constituency, and C is the number of seats available, then C=A/B. We have to assume that A is relatively constant, because we can't just invent new people! So the size of a constituency is inversely proportional to the total number of seats.

It is therefore no more arbitrary to decide the size of equalised constituencies than it is to decide the total number of seats. You are setting exactly the same thing. If you equalised constituencies first, then reduced the number of MPs later, you'd just have to completely redraw the boundaries again. It would be a redundant exercise.

Another arm of Kevin's argument is that we should be able to make constituencies more equal by setting the number of constituencies afterwards. Except you can equalise constituencies without changing the total number of seats at all! All you do is don't eliminate any constituencies, but shift the boundaries about according to population distribution. If anything, making special cases for smaller "communities" like the Highlands, as many try to argue in favour of, harms not the arbitrariness of the number of seats, but the very premise of equal constituencies itself.

And that sort of leads on to my more general point about this boundary review. Making substantial special cases of communities in representation fundamentally misunderstands the notion of an electoral system where individuals vote for electors. Members of Parliament represent constituents, not communities per se; their community representation exists only insofar as a community happens to be partly or wholly contained within their constituency. In truth, constituencies themselves are relatively arbitrary. Attempts are made not to create absurd situations where bits of a constituency are landlocked from the rest, or to prevent the drawing of boundaries so as to create an inherent advantage for one party or group. Gerrymandering can still happen even when the size of constituencies is relatively equal. For example, you could ringfence seats with predictable voting patterns so that they become uncompetitive even if the overall numbers are roughly the same.

What we observe about the current system is that the old boundaries created an inherent bias towards Labour, and that not all of that was itself down to unequal sized constituencies. This explains why Labour won a handsome majority (66 seats) off 35% of the vote in 2005 whilst the Tories fell considerably short (by 20 seats) with 36% in 2010. Equalising constituencies does not, in and of itself, eliminate this gerrymandering, although the actual act of redrawing the seats has the practical effect of reducing its effects. Whether that has served simply to gerrymander in reverse rather than undo that which already exists is a legitimate point of discussion. I think there is a case, from a Lib Dem perspective, to argue that this is the case, although I do not find this terribly convincing. The new boundaries can only really be judged by some actual elections to see how they operate in practice, and even then there are other variables at play there. I'm rather more convinced that the problems people are pointing to are rather more fundamental, and pertain to the necessary effects of single-member constituencies.

It would be a lot easier to end gerrymandering and to equalise voter influence if we adopted some form of proportional representation. Larger multi-member constituencies, chosen by preferential method, are much more flexible to deal with large variations in population density. If a natural region does not comfortably correlate with a fixed "number" in mind, you have more options to maintain a constant voter value. Not only can you increase or decrease boundary sizes, but you can also increase or decrease the number of representatives without inherently prejudicing any particular party.

Distribution and number of seats can absolutely affect to a greater extent the outcome of an election than a relatively modest change in the system by which our representatives are elected from FPTP to AV. But it is procedural in character and a question of administration of the existing system rather than an overhaul of the way we elect our members of parliament. The idea that we need a referendum every time the boundaries and number of seats are tweaked by the Electoral Commission is, I think both impractical and misguided. It is an administrative and not a substantive concern.


  1. I find your arguments convincing. I am relieved that we at least agree that large multi-member proportional constituencies should be the end result. We were likely to disagree on what should be the interim since this bastardised review is not really what either of us would have called for. I am still not pleased with what the boundary commission is proposing as it carves up my own constituency in a completely nonsensical way, while also paradoxically increasing its representation by one mp.

    p.s. i thought your blog was going to be a lot more nihilistic judging by the title =P

  2. I think in the context of the existing system the best we can do of a bad job is to create equal constituencies on a basis similar to what's on offer, if nothing else to make it more difficult for people to fudge issues that are distinct sources of inequality within the system if/when electoral reform ever re-emerges.

    We're ultimately left with two choices in the existing set-up: constituencies which are equal but which butcher communities or constituencies made vulnerable to vested interests. Neither is that attractive but the former is the lesser of two evils. It's important to remember that whilst a constituency might not reflect natural community boundaries, there are many more sparsely populated constituencies where MPs have to represent a number of diverse communities anyway. A Highland MP, for example, has the small city of Inverness, lots of countryside and farming hamlets, and coastal towns. If MPs can't cope with serving multiple communities they're not doing their job properly.

  3. I too have written adovcating large multimember constituencies: