Parliament would have two types of MPs who would be equal in every way apart from the manner in which they are elected. There would be the same total number of MPs as we have currently.
One type would be Constituency MPs (CMPs) who are elected exactly as MPs are elected under the first-past-the-post system now, just for slightly larger constituencies. Then, for each region of Britain (South West, London etc.), there will be ‘top-up’ Regional MPs (RMPs) who are allocated in proportion to the votes each party received in that region.
For each region, all of the unsuccessful candidates are put on a list in order of the percentage of the vote they received in their constituency. This means that the most locally popular candidates will be at the top of the list.
The Regional MP seats are allocated by running the d’Hondt formula over all votes cast in each region. Candidates are elected from each party in the order they appear on the newly created party list.
It’s thus very similar to the Additional Member System used to elected the Scottish Parliament and the Welsh and London Assemblies. The differences are:
- AMS requires voters to vote twice: once of the constituency, and again for the region. RTU only requires one vote. A problem with AMS is that a party could “cheat” by standing on one name in the constituency seats and another name for the regional vote, and by doing so the constituency seats they won would be disregarded when allocating top-up seats (Labour considered this in the 2007 Scottish election, but didn’t eventually do it). It is not possible to game the system in this way with RTU.
- With AMS, the top-up seats are allocated to a party according to that party’s own ordering of candidates, so those candidates favoured by the party are most likely to get elected. With RTU, on the other hand, the ordering of a party’s top-up candidates is based on the share of the vote that they get, and a candidate is therefore more likely to do well if they are popular with the voters.
There are in fact a whole load of different voting systems that could be constructed on these principles. The parameters by which they vary are:
How constituency members are chosen. AMS and RTU both use First Past The Post (FPTP). Roy Jenkins’ AV+ system was the same as AMS but used AV for constituency seats. It has been proposed that the Scottish Parliament move to being elected by STV, with a small number of Scotland-wide top up seats; in this case the whole country would be treated as one big region.
Whether there is a separate vote to decide the number of top-up candidates. Systems that use this, such as AMS, are prone to parties gaming the system by standing under different banners in the constituency and top-up sections. Furthermore, it makes it easier for the electorate if they only have to vote once.
What proportion of those elected are top-up members. To ensure rough proportionality, this would need to be about 20% if the constituency seats were FPTP or AV. But if the constituency seats were STV, it wouldn’t need to be as high, because STV is already a roughly proportional system. Of course, one could deliberately design a system where there aren’t enough top-up seats to ensure full proportionality, in order that a party might win an overall majority with less than half the vote, to ensure “strong government”. At the moment, a party can win a majority in the Commons on a third of the vote (Labour got 55% of the seats on 35% of the vote in 2005), which I find unacceptable, though I’d have a lot less problem with a party winning a majority of 45% of the vote.
How big the regions are. The larger the top-up regions, the lower share of the vote a party needs to get elected. If the number of top-up members is enough to ensure proportionality, then a party would win a seat on 1/(n+1) of the vote, where n is the total number of people elected (constituency and top-up members). So in a Scottish region with 16 members, a party would need 5.88% of the vote to be guaranteed a seat. Some people suggest that a PR system should have an effective threshold of about 5% to prevent nutters from getting elected. However, at present we have a government that thinks women who take turns looking after each others children are criminals, so if more MPs from minor parties were elected, it’s unlikely there would be more nutters in the House of Commons than at present.
What mechanism chooses who becomes a top-up member. AMS uses a party list. RTU goes by a candidate’s vote share. Another possibility would be to use the number of votes that a candidate gets; this means that a constituency with a bigger electorate or where more people turned out to vote would be more likely to get top-up members elected to it.
If I were designing a PR voting system for Westminster, I’d use constituencies of variable size, elected by AV (if they are single member) or STV if they are multiple member. Having variable size constituencies means that they can more easily represent natural units, such as a town or city. So if Edinburgh has 5 parliamentary constituencies, it would be more naturally represented by 1 5-member STV constituency, and 5 individual ones, because people think of themselves as living in Edinburgh not in “Edinburgh South” or wherever. Because constituencies aren’t all forced to be the same size, their boundaries can more easily follow local government boundaries, and wouldn’t have to be changed every few years as populations change; indeed all the boundaries commission would have to do every ten years is change the number of seats a few multi-member constituencies have, merge the odd small constituency, and split the odd large single-member constituency into two, or make it a 2-member one.
In terms of regions, I’d either use the same ones used in Euro constituencies, or modify that scheme slightly. These regions mostly correspond fairly closely to where people think of themselves as living — for example London, east Anglia, Wales, Scotland, Northern Ireland. I’d perhaps merge East Midlands with West Midlands, and North East England with Yorkshire And Humberside, and it’d be nice to re-use the names of the historic kingdoms of England (East Anglia, Sussex, Wessex, Mercia, Northumbria). 12 regions and 646 seats means an average of 54 seats per region, meaning that a party would need a threshold of 2% to get elected, which I think is about right.
Regarding the mechanism to choose top-up members, I don’t have a strong preference among them, but would on balance choose top-up members in the order of 1st preference votes they achieve. This would favour candidates in large constituencies, of course, which some voters might see as problematic but I think that most wouldn’t. For example, if a Glasgow voter votes Labour than that vote is used to elect a top-up Labour MP in Edinburgh, the voter probably doesn’t mind, because he’s getting an MP of the party designation he wants. If he does mind, and enough people think like that, they can always set up a Glasgow Party, which runs only in Glasgow.