(Ln(x))³

The everyday blog of Richard Bartle.

3:13pm on Sunday, 5th June, 2005:

Electoral Reform

Miscellaneous

I mentioned after the recent general election that I had some ideas for reforming the electoral system in the UK, and threatened that one day I might blog them. This is that day...

Necessary criteria:

MPs should be tied to constituencies (for local representation).
Representation should be proportional (but not so much so that tiny single-issue parties can call the shots).
Voters should be able to dump unpopular MPs (so no closed party lists, then).
The system must be simple (because politicians think we're all morons).

Desirable criterion:

The need for tactical voting should be reduced.

OK, here's the central thrust of the system I propose. All constituencies are parcelled into groups by adjacency. The groups would have 3 or (for small-population constituencies) possibly 4 component constituencies. People stand for parliament in only one constituency, and if they win then they're elected, first-past-the-post style. Thus, there will be 3 "by right" MPs in the group. There is the possibility, however, of additional top-up MPs for the whole constituency group. The votes of losing candidates are added up on a party-by-party basis, and if the total number of votes for one party exceeds the total won by all of the by-right MPs individually, that party gets a top-up MP. The top-up MP will be the losing candidate fromt he constituency group who polled the most votes.

Here's an example of how this works:

Let's suppose the constituencies of Colchester, Essex North and Harwich formed a constituency group. The results of the three main parties in the 5/5/05 general election were:

Colchester (44,899)
LD 21,145
C  14,868
L  8,886

Essex North (47,959)
C  22,811
L  11,908
LS 9,831

Harwich (50,408)
C  21,235
L  20,315
LD 5,913

The total number of votes cast in these three seats in 2005 was 143,266, of which 58,914 went to the Conservatives, 41,109 went to Labour and 36,889 went to the Liberal Democrats (the rest went to smaller parties, such as UKIP). So that means the Conservatives won 41% of the votes, Labour won 29% and the Liberal Democrats won 26%. The actual number of seats won was 2 to the Conservatives and 1 to the Liberal democrats. Labour get no seats, despite their strong showing.

Under my proposal, we add up the votes of the losing candidates. For the Conservatives, that's just the Colchester total; for the Liberal Democrats, it's Essex North plus Harwich; for Labour, it's all three. This gives:

L  41,109
C  14,868
LD 15,744

The total needed to get a top-up MP is (more than) 22,811, which is the largest number of votes won by any single candidate in the group. Only one party has this, Labour, so they would get a top-up MP. That top-up MP would be their candidate for Harwich, who obtained 20,315 votes — the highest of the three losing Labour candidates for this constituency group. The result is that of the 4 MPs returned by this constituency group, the Conservatives would get 2, the Liberal Democrats would get 1 and Labour would get 1. Given the percentage breakdown of the votes (41/26/29), this representation (50/25/25) isn't too bad.

Let's see how this would work on a larger scale. I applied this technique to the whole of Yorkshire and Humberside, which is 56 constituencies. I grouped these into 18 groups of 3, passing the remaining two constituencies (Scunthorpe and Colne Valley) to imagined constituency groups in Lincolnshire and Lancashire respectively. Overall, 2,110,738 votes were cast in the remaining 54 constituencies.

Actual results for the three main parties:
Party  Votes   Seats %Vote %Seats
L      922,927  42     44    81
C      609,953   9     29    17
LD     437,237   3     21     6

So, under the present FPTP arrangement, Labour won 81% of the seats on 44% of the votes. What difference would adding top-up votes make here? Well, under the constituency groupings I used (which I tried to make fair), Labour get 3 top-up MPs, the Liberal Democrats get 7 and the Conservatives get 9. This means the region would return 73 MPs rather than 54, with a percentage breakdown of:

L  62
C  25
LD 14

62/25/14 lies between the 44/29/21 that "true" proportional representation would deliver and the 81/17/6 that FPTP gave us.

You may have noticed that in my original explanation, I said that the combined party votes total for a constituency group should be more than any by-right MP wins. The reason for this is that it's possible for some MPs to win with fewer votes than the second-placed candidate in another constituency. The winning candidate in Great Grimsby, for example, polled fewer votes than the second-placed candidates in both the other constituencies that make up its group (Brigg and Goole and Cleethorpes). These near-miss candidates would both have a case to be awarded top-up seats, given that they both claimed more votes than someone who was directly elected in their group. If the criterion for winning a top-up seat is that the party total must exceed the total for any individual MP, that removes the case for having two top-up MPs from the same party in the same group (or at least it makes it much less persuasive).

Nevertheless, what if we were to relax this rule, such that to qualify for a (single) top-up MP you only had to beat the lowest-polling by-right MP of your constituency group? Well, what would happen is that Labour would still pick up only 3 top-up MPs, but the Conservatives would get 12 and the Liberal Democrats would get 14. There would be 81 MPs in total for the region, the percentage breakdown for the seats being:

L  56
C  26
LD 19

This is closer to the "ideal" of 44/29/21, but whether it's close enough to warrant the lack of legitimacy argument it brings with it is another matter.

The system is still not very fair for the much smaller parties. In Yorkshire and Humberside overall, the next three largest parties by votes were:

Greens 30,930
UKIP   25,021
BNP    59,660

Surely those results should be worth one extra seat each for the region as a whole? The BNP could perhaps argue for two, given that they have more than UKIP and the Greens put together?

I haven't done an analysis for the whole country. A while back, I did do it for maybe two thirds of the country using 1997 election data, and the results followed the same lines as the Yorkshire and Humberside figures above: a much fairer result, but not one that would have robbed Labour of a majority. It would be interesting to see how it would play out country-wide for 2005, but I'm too spreadsheet-inept to want to try that right now.

This system fits all my initial criteria: by-right MPs are tied to their constituency, and top-up MPs are tied to a group of 3 constituencies; representation is proportional, but not "too much" so; unpopular individuals can't sneak in through the back door; the system is simple (you just make an X as with the current FPTP system). I believe it would reduce the need to make tactical votes, but of course the figures I'm using to trial the effects thenselves include tactical votes, so it's hard to gauge what would happen overall.

Some disadvantages:

There would be too many MPs if we used the current constituency boundaries. Thus, we would have to redraw the map so as to make for bigger constituencies such that overall we got roughly the same number of MPs in total as before. In so doing, we could make the constituencies group together in threes according to "natural" boundaries (which isn't actually all that hard with the current constituency boundaries, except sometimes you get awkward singletons).
We don't know how many MPs there will be in the House of Commons until all the votes are counted. I don't think this is much of a problems: it just makes predictions about where the finishing line might be add to the fun on election night TV...
A massive vote for one party (eg. its leader) in a single constituency could mean there were no top-up MPs despite a good showing. Perhaps, then the rule to win a top-up seat should be that you beat the middle-ranked by-right MP, rather than the top-ranked or (as I discussed earlier) bottom-ranked?
MPs have to declare for a party if they want to try get in on a top-up vote. Independents therefore wouldn't be able to get top-up seats, but then if they're independent it's hard to see how proportional representation could make much of a difference anyway.

In summary, I feel that this modest alteration to our existing electoral system could lead to much fairer results while retaining the basic principles of representation and accountability that a straight first-past-the-post arrangement embodies.

It's just a damned shame I'm a nobody, so have no voice in the debate whatsoever.

Referenced by In Proportion.

Referenced by Opposition.

Latest entries.

Archived entries.

About this blog.

(Ln(x))3

Electoral Reform

(Ln(x))³