There has been some recent political drama regarding proposed constituency boundary changes. It’s rather an unseemly mess and this post is not about the politics of that. Instead I’m writing here about why constituency boundaries are deemed important and, rather more significantly, why boundaries should not really matter.
Why do boundaries attract such attention?
The significance of boundaries lies in the fact our MPs are elected in single-member constituencies using First Past The Post (FPTP) voting. The results of such elections are very sensitive to boundaries. In fact, it can be demonstrated that the process of modifying boundaries can have a huge effect on the results: new boundaries change constituency composition, change turnout and (particularly under FPTP) can change the manner in which voters vote tactically. The fact that boundaries are so significant is exactly why people tend to argue about them.
Disproportionality and bias
When FPTP is used to elect MPs in single-member constituencies in a national election, there are two major faults with the results:
- Disproportionality, and
Disproportionality is what makes it possible for a party to gain significantly more (or fewer) seats than their share of the vote would merit. Disproportionality is a built-in ‘feature’ of FPTP and cannot be avoided when using that voting system. In contrast, bias explains why disproportionality can affect different parties to a greater or lesser extent.
For example, in 2005 Labour won 55% of seats with 35% of the vote: that’s disproportional. In 2010 the Conservatives won 47% of the seats on 36% of the vote: that’s also disproportional. The fact that Labour’s 2005 results were significantly more disproportional than the Conservatives in 2010 is a demonstration of bias.
When viewed nationally, FPTP is disproportional by definition. One could use the term unfair instead of disproportional. The fact that there is bias in the system means that it could be considered not to be fairly unfair! It is easy to see why the Conservatives are in favour of a change where some of this bias would be removed (in their favour). Even with bias reduced or eliminated, the system will remain disproportional as long as FPTP is used and there will still be a strong chance of majority governments formed by parties gaining no more than 40% of votes.
Would AV have helped here?
No. Any voting system which elects its parliament in single-member constituencies will always exhibit disproportionality and, almost always, bias too. AV will indeed improve matters locally to eliminate tactical voting, but when the wider national picture is inspected, it will possess similar disproportionality and bias as FPTP.
How bias occurs
To explain how bias occurs requires some complex analysis and is a result of a combination of factors. It is affected by the geographical distribution of votes, differences in constituency size and in turnout.
As an example, to see how constituency size might have an effect imagine a single constituency with a single-party majority. Split that seat in two: now the winning party has two seats instead of one, without anyone’s vote changing. Similar demonstrations show the effect of turnout and geographical distribution of votes. Broadly speaking, the reason for the bias seen in recent UK elections is that Labour tend to be popular in constituencies with lower populations and/or lower turnout. The Conservatives tend to be popular in larger seats with higher turnout.
Can bias be eliminated?
One proposal to eliminate bias is to “make all constituencies the same size”. That sounds incredibly sensible at face value (and is the primary political point made by the Conservatives) since it sounds fair. However, there’s much more to it than that. What does “all the same size” actually mean in practice? Same population? Same size electorate? Population or electorate as of what date? It’s not necessarily obvious or straightforward. The ratio of population to electorate varies and so does the stability of that population. Generalising hugely, city populations tend to be younger and more likely to migrate, while rural populations tend to be older and more likely to stay put.
Of course, even assuming one really could eliminate bias in the current system, the disproportionality would remain. Elimination of bias would simply allow us to move from unfairly unfair to fairly unfair!
A related issue to the choice of boundaries is that of safe seats. As a result of a coincidence of boundaries, many seats are safe seats; this ‘safeness’ only changes if the boundaries are modified which then as a result makes a previously non-safe seat into a new safe seat. Safe seats are bad for voters, because a single party may control such a constituency for decades, often with the support of under 50% of the electorate.
Can we eliminate disproportionality?
Can we eliminate disproportionality by changing the boundaries? With single-member constituencies, No. However, the answer is Yes, if we change from single-member constituencies to multi-member constituencies.
How would multi-member constituencies help?
Electing MPs in multi-member constituencies using a voting system such as STV (Single Transferable Vote) results in little disproportionality and little bias: the levels of each are largely determined by the number of MPs in each constituency. For example, 100 constituencies which each elect six MPs would be less disproportional and less biased than 200 constituencies which each elect three MPs.
In such a system, a party polling x% of the national vote will gain approximately x% of the seats in parliament, thus being broadly proportional. The level of proportionality improves as the number of MPs per constituency goes up.
Of course these larger constituencies will still need boundaries. However, unlike single-member constituencies, the sensitivity of the election results to the actual boundaries is severely reduced. Review of boundaries need not take place so often: for instance, if the population changes significantly, the number of MPs returned could be changed rather than the boundaries. This would allow the larger, multi-member constituencies to be more ‘natural’ compared to current constituencies, following existing local authority/county boundaries.
There’s plenty of precedent for multi-member constituencies around the UK in various tiers of government. It’s used, together with STV, in the Scottish local government elections and for the Northern Ireland assembly. There’s no reason why they can’t be used for Westminster elections too.
I’ve written this up because every time I see discussion of small detail of a proposed boundary change, I’m frustrated because fiddling with the boundaries of a single-member constituency system is Solving The Wrong Problem, in my opinion. Attempting to make the system more fairly unfair seems wrong, when the opportunity exists to make it more fairly fair!
The boundaries should not matter and, with multi-member constituencies, they will not matter.
Multi-member constituencies is the right solution for eliminating unfairness, together with the STV voting system. This is the approach proposed by the Electoral Reform Society for Westminster elections. There are advantages and disadvantages of multi-member constituencies, of course, but that’s not really a subject for this post. Perhaps I’ll just say that the precious “constituency link” held as important by MPs may not seem so important to voters when it links an MP who was voted-for by less than 50% of their constituents with an area having fairly arbitrary, artificial boundaries.
Sadly both Labour and the Conservatives gain from the current system, because they benefit from the effects of disproportionality. They would both have fewer seats under a proposed multi-member system, so it’s going to be extremely hard to bring about such a change. They will continue bickering about small changes in the boundaries of the single-member constituencies, intentionally missing the point of further reform. It is worth pointing out, however, that there are advocates of proper reform in all parties, including Labour and the Conservatives.