Mixed-member system changes?

Wilf Day sent me a link on a proposal to change the MMP system in Wales.

Alan Renwick, in a feed I get on electoral reform, says there are plans to change Hungary’s system as well. Apparently to remove the partial compensation linkage between the tiers, change to FPTP for the nominal tier, and reduce the size of the (overly large) assembly. (Link in Hungarian, in case someone can read or translate it).

27 thoughts on “Mixed-member system changes?

  1. Thanks to Matthew and Wilf for this – I had not seen it, but am not surprised this has come up. As for Labour’s allegation of ‘rigging’ the system, pots and kettles come to mind. It was Labour (led by Peter Hain in this instance) that banned dual candidacy for Welsh Assembly elections, and action condemned by all the non-Labour people who commented on it. This outrageous ban, which unashamedly reduces competition, making seats safer for Labour, needs to be removed in any future changes to the electoral system. The ban does not exist for Scottish Parliament or London Assembly elections (let alone for most other mixed-member cases around the world). Indeed, I am doing research into why so many Labour candidates in the last Scottish Parliament election failed to practise dual candidacy (the Labour Party in Scotland does not like it, but does not entirely prohibit the practice here) – the result was that many experienced Labour members lost their seats, with highly inexperienced Labour candidates winning regional list seats. This compounds the electoral disaster for Labour. Louis Massicotte’s research makes a strong case for keeping dual candidacy, and I wrote about the British ‘issues’ with it in my book if anyone is interested. Hain and other Labour people invoked the notion that there is widespread public dissatisfaction with dual candidacy, but there is no evidence of this (the only study I am aware of was a Labour-commissioned focus group in Wales that gave a mixed response). The fact is that most people in Wales (and Scotland) do not really understand MMP and do not distinguish between constituency and regional members.

    Tom Lundberg, University of Glasgow

    • Good points, Tom. And please give the publication details on your book!

      One point: It is true, as far as I know, that no MMP systems ban dual candidacy, aside from Wales. However, there are some cases of MMM that do so. Mexico, for one. Possibly Thailand?

      I agree that it should not be banned under either type.

  2. The Japanese Upper House is another MMM system that bans dual candidacy. District tier is SNTV, although most districts have M=1 or M=2, so not much different from FPTP.

  3. Could elections be held in two rounds, the first round for the district seats, and the second round for a similar number of MPs elected by party list proportional representation? But the votes each party gained in the first round would be deducted from their list in the second round (there would have to be some provision against dummy or shadow parties). Candidates who lost in the first round could then run again in the second round and maybe get elected to a list seat, with the voters already knowing that they were defeated on their first attempt.

  4. @Ed
    That is one way to deal with the “problem”, but a more elegant solution, at leas in cases like Wales, Scotland, and London is to simply open up the regional lists. Clearly this is not practical in all MMP systems, certainly not in New Zealand for example, but district magnitudes in Wales, Scotland, and London are low enough to avoid bedsheet ballots. If I may be permitted a slight digression, I really don’t see why the Brits don’t open up the lists in Euro elections too.

  5. On the banning of dual candidacy, I looked around the world of MM systems in 2005-6 when I was writing my book and could only find Ukraine (which no longer has an MM system), Thailand, and the Palestinian Authority as cases that had actual bans in place (the Thai ban was still in place in July’s election, with the government hoping that list-elected MPs would become ministers, while constituency-elected MPs would focus on serving constituents, but they did make other changes to the system, such as more list seats, a national (not regional) list system, and all single-member districts for the nominal tier). On Mexico, there is not a ban on dual candidacy, but rather a limit on how many dual candidacies that can be placed per party (this is what Louis Massicotte found and reported in a 2004 publication, and I also found the Mexican government publication (from the Instituto Federal Electoral, 2006) that stated – available in both English and Spanish to those who are interested – that the limit is 60 ‘dual’ candidates per party). If this has changed in Mexico since 2006, please let me know, but at that time, Mexico did not actually ban dual candidacy per se. I did not know about the Japanese upper house (thanks for that information above), but was limiting myself to lower or only houses.

    If anyone is interested, the issue of dual candidacy is part of Chapter 6 of my book Proportional Representation and the Constituency Role in Britain (Palgrave Macmillan, 2007), and was my favourite chapter in the book. Labour MP Peter Hain features prominently – he was the Welsh Secretary at the time and is still making statements about how much he hates proportional representation! He’s now pushing for the Welsh Assembly to move to a two-member district block vote (sorry, Matthew – multiple nontransferable vote) system, claiming that there is a consensus on making such a change. This is complete rubbish! Such a change would only help the Labour Party and will not find any support outside of it.

    On the opening of British European Parliament party lists, this was considered, but rejected by the previous Labour government. Speculation was that they only grudgingly introduced PR in the first place and wanted it to be as unattractive as possible. I personally think a flexible list system would be an improvement, though suspect that it wouldn’t change very much (the highest profile people, who are now generally at the top of the lists, would probably obtain the most personal votes). Nevertheless, such a change would respond to much of the criticism of PR in Britain.


  6. Tom Lundberg says “I personally think a flexible list system would be an improvement, though suspect that it wouldn’t change very much (the highest profile people, who are now generally at the top of the lists, would probably obtain the most personal votes).”

    This crucial design detail may become more important soon. New Zealand will, if it keeps MMP, review possible improvements. Arbuthnott made similar recommendations for Scotland. Quebec has a governing party which needs MMP and, if they make up their minds, might introduce it on short notice; and they have a report from their Director-General of Elections discussing the respective advantages of closed-list and open-list, and then gently implying that flexible list is the best of both worlds. The Law Commission of Canada recommended flexible list MMP in 2004, which is still the only real electoral reform proposal on the table for Canada.

    But what will the threshold be for personal votes to be sufficient to move a candidate up the list? What would Tom recommend?

    When there are 12 list seats (the typical region size in Sweden’s pure-regional-list system), and it takes 8% of the party vote to move up the list, then if everyone uses the individual vote option roughly half the 12 candidates will move up the list, and the ranking will be determined entirely by the individual votes unless a party wins more than 50% of the vote in that region. But being flexible list, not everyone uses the individual vote option.

    If Canada had MMP regions averaging 14 MPs with five list seats as Henry Milner suggested, suppose parties nominate ten candidates. Suppose 60% of voters use the individual vote option. (In Brazil it’s about 90%). Now suppose we set the individual threshold at 5% rather than copying Sweden’s 8%. Since the average candidate gets 6% of the party vote, the majority will likely pass the threshold, so they will be ranked by the individual vote, not by the list ranking. Or we could use 4%. The Law Commission left this open.

    On a related point: how big should MMP regions be? Bavaria has six regions with manageable size: 13, 12, 10, 9, 9 and 9 local, plus the same number of list deputies. So the typical region elects 9 list deputies. Scotland elects 16 including seven list. Wales 12 including four list. These are all manageable without bed-sheet ballots. But Bavaria has a seventh region that includes the capital of Munich, which elects 29 local deputies and 29 list deputies, and has a hideous bed-sheet ballot.

  7. Thanks to Wilf for the interesting question and all the information on electoral systems. I will spoil this by saying that I do not see any point to open/flexible lists with MMP – the whole point of MMP is to put a ‘human face’ on PR with the constituency vote (the Germans call MMP ‘personalised proportional representation’). If you want open or flexible lists, then go with such a system without mixing members – I just don’t understand why people would want to complicate MMP any more than it already is. The Arbuthnott recommendation to open up the lists went nowhere in Scotland. If people are hostile to the closed list element of MMP, then they should probably focus on STV or some kind of open or flexible list system. Denmark, Sweden, and Finland seem to make this work well enough.


  8. Why not just use STV with a top up best loser nation wide tier? 90% of the legislature is elected by STV, and 10% is the at large best loser candidates by party.

    Also why not just use MMP with a best loser model.

  9. There will be a review of MMP in New Zealand if the system is retained in the November referendum. It is actually required by the legislation.

    The required areas to review are thresholds (5% of party vote, as well as the one-electorate loophole that allows a bypass of the 5% minimum), proportionality (the impact of population shifts as well as surpuls/overhang mandates), dual candidacy, and ranking of candidates by the parties. Other areas may come up for review, but the Maori seats and size of Parliament are forbidden topics.

    The guarantee of a review might help the ‘yes’ side, though these issues were reviewed about ten years ago and nothing was changed.


  10. Yes, but the two reviews will share a lot of interested parties. As a taxpayer, combined hearings/deliberations but separate reports sounds useful.

  11. Why not just have all list candidates run in all districts? If a candidate wins in more than one candidate, that candidate will represent the district where it got the most votes or highest percentage of the vote.

    The ideal plan would be to have a transferable vote system if this were to be implemented.

  12. “Why not just have all list candidates run in all districts? If a candidate wins in more than one candidate, that candidate will represent the district where it got the most votes or highest percentage of the vote”

    I’m not sure if this works with small parties. Say you have a party that gets 6% of the vote nationwide, entitled to representation under most list systems. But there is very little deviation in the vote, so the most it gets in a particular district is 9%. The vote of the FDP in Germany follows this profile. Do you really want a district “representated” by someone who only got 9% of the vote there?

    One thing I could see is to experiment more with the concept of voting members and non-voting members. The assembly is elected from single member districts, and if a candidate wins a plurality in his or his district, he or she becomes a member of the assembly. But the member doesn’t necessarily get a vote -each party in the assembly gets a number of votes in the assembly in proportion to the total vote they received in the election, to distribute to amongst their members (and they can rotate this). Non voting members can still make speeches, table questions, introduce bills, serve on committees, etc. This would mean that a small party would have to win a district, somewhere, to get representation at all so it wouldn’t be a proportional as a pure list system. But it would be quite proportional among the larger parties and effectively end in parliamentary systems the practice of a party forming a government alone without having come close to winning a majority of the popular vote.

  13. I once filled out all 257 squares on a ballot paper the size of a tablecloth for the NSW legislative council. Having all list candidates run in all districts would result in ballot papers the size of football fields. There can be transaction costs to electoral proposals.

  14. Ed and Alan,

    Thanks a lot for your points! However, I should’ve said that all list candidates within an electoral region (be that Lander, state, province, etc.) should run for the district races and that a ranking voting system would be necessary, in the case that candidates win in more than 1 district.

    This proposal would be interesting to see in countries without electoral regions, like New Zealand.

  15. For example, if the UK had used Jenkins MMP/AV+ nowadays, the electoral regions would’ve had, on average, between 8 and 11 districts.

    Under this scenario, a UK voter living in an electoral region would’ve had a ballot with at least 8 candidates from all parties running in the district races.

  16. No, assuming the 3 main parties each nominate a candidate for each available seat, you are looking at 24 candidates without any independent or minor party candidates. There were 4150 candidates for the House of Commons in 2010. Averaging across 650 seats and multiplying that by 8 to 11 gives a fairly large ballot paper, if not quite football field sized.

  17. OK, I get that.

    How about this idea?

    First, each electoral region would be the size of the smallest Lander/state/province. If a province has 60 seats and the smallest one has 5, the smallest province would be 1 electoral region, while the large province with 60 seats would have 12 electoral regions of 5.

    Second, a party can have a list of candidates equal to the size of the electoral region. In this case, using the example from above, each party would nominate up to 5 candidates.

    This would mean that a party can have 5 candidates per district. Obviously, independents and minor parties are also included.

    Finally, a party creates a list of 5 names and these candidates would need to run in every district within the electoral district. These are the same 5 candidates, not more than that.

  18. What are the benefits of forcing all candidates to run in all districts, whether of an electoral region or some other area?

  19. I would support dual candidacy if it meant that a candidate must run in at least 2 or 3 districts.

  20. How about one vote MMP with optional preferential voting? The first preference is a vote for a candidate as well as a party vote. At least with this version of MMP, one would avoid two ballots, and the possibility of decoy lists.

  21. The only fair and rational way to combine preferential voting with MMP is to elect the list MPs with exhausted votes (those votes that cannot be credited to a district MP). You would have long ballot problems and the list tier would probably need to be divided into smaller regions to make the ballot manageable. Really it is hard to see the advantage of preferential MMP when a more rational plan would be simply to use optional preferential voting in multimember districts.

  22. Open-list MMP, whether fully open or “flexible” (which the British call “semi-open”), requires middle-sized regions to prevent “bed-sheet ballots” and make the regional MPs more personally accountable.

    But in the most recent election for the Scottish Parliament, using 8-region MMP, the Scottish National Party won a majority of seats on a minority of votes. Was this because of using eight regions?

    Not really. Using eight regions gave the SNP one overhang seat. The bigger problem was that Scotland uses the “highest average” calculation, which Germany abandoned decades ago; it favours large parties. The other methods are “highest remainder” which has the virtue of simplicity, and “Sainte-Laguë” which is the best mathematically. A new study has just shown that, if Scotland changed to Sainte-Laguë but kept its eight regions, that would generate the fairest result, and the SNP would not have won a manufactured majority.

  23. OK, so after many years, I’ve continued reviewing different electoral systems. Still I’m sold on the idea of a 2-tier system. My idea would be to use a ranked ballot system, with unused votes and losing candidates from the district tier, going to the regional/county level.

    If you have 10 seats to be allocated in a region and there are 6 district seats and 4 constituency seats, with 4 parties, I’d add a special tweak: all districts will have 2 district candidates while the remaining parties qualify for the at-large ballot. It keeps the district tier being an Instant Runoff and the at-large tier being proportional. This would have districts see races between 2 major parties or 2 third parties. It would be at random.

    D1 DR/GL
    D2 GL/DR
    D3 DG/RL
    D4 RL/DG
    D5 DL/RG
    D6 RG/DL

    In this example, with 4 parties, each party has 3 district candidates and within each district, the at-large tier has a general ticket for each party (2 candidates). So, each party would nominate 9 candidates in total. For example, if you live in the 1st district, you would have to choose between a Democrat and a Republican for district reprsentative and choose between the Greens and Libertarians for the at-large representatives. A funny tweak would be for the voters to rank the candidates from both tiers and mix up the preferences. But it would be complicated and voters would have to rank at least 5 candidates (4 at-large and 1 district) to have a valid ballot, unless you want to declare those votes “exhausted” in the district tier.

    Voters would cast a first preference for a candidate in the district race and must rank the at-large candidates in the at-large race (obviously the candidates that were nominated in that district, not all the at-large candidates). Simple. You’ve introduced a moderate ranked ballot system for both tiers (something Jenkins’ AV+ clearly lacked with its mix of open list PR and IRV). The question would be if the electorate wouldn’t mind having to choose between 2 candidates in the district tier, even if none of them represent their favorite party. But, if voters are already used to a 2-party system mindset, why can’t a system like this offer a district race only between a Green and a Libertarian?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s