MMP and dual candidacy

Should candidates be allowed to run simultaneously in the single-seat districts and on the party lists of mixed-member proportional (MMP) systems? Such dual candidacy is sometimes seen as a problem with MMP, in that a candidate can “lose” in the single-seat district (SSD), yet still win a seat based on a high ranking on the party list. Make My Vote Count has a thoughtful post on this question.

In Japan, where the electoral system is mixed member, though not proportional (because the SSD and party-list seats are allocated in “parallel” rather than to ensure overall proportionality, as with MMP), they have a term in the political lexicon for such losers-yet-winners: ZOMBIES. That such a term is used suggests the potential political problem with dual candidacies.

At least for MMP, supporters of the system and of dual candidacy can argue that the very concept of calling candidates who fail to obtain the plurality of votes in the SSD in which they were nominated “losers” is misleading. After all, MMP is a PR system, and in PR systems, the priority is on proportionality and parties, not on the SSDs or the candidates. (Parallel systems are different: There really is a premium on parties’ having strong SSD candidates, so if they lose, one could argue that they are indeed ‘zombies’ who ought to be considered “losers.”)

Nonetheless, I think we MMP advocates have a PR problem–that is, public relations–if we argue in favor of dual candidacy, particularly where MMP is being considered for adoption in a political culture grown up around the idea of SSDs and thus clear winners and losers.

Fortunately, the problem is easily fixed, and does not require a ban on dual nominations. (Even if one sees no problem with such bans, or limits, for obvious reasons, politicians–such as those on New Zealand’s 2001 MMP review committee mentioned in the Make My Vote Count post–are an obstacle to implementing them.)

The solution is to combine MMP with open lists. There is no reason why MMP has to be in the paradigm of SSD and party vote, where the party vote is for a closed list (i.e. one in which parties themselves determine the order in which candidates not elected in an SSD will be elected off the PR list). All MMP systems in use currently do have closed lists, except for a couple of German states (the details of whose rules I do not know), but it need not be that way.

I would recommend the following ballot design, in which voters give two candidate votes. These votes would be clearly indicated as one for the district representative, and one for a party representative. The latter vote would be counted for the party, just as the party vote is in Germany or New Zealand. The difference is that these votes for party representatives would determine not only the overall partisan balance, but also the order in which candidates not elected in an SSD are elected off the list.

I would further advocate that the party-representative portion of the ballot include the names of candidates for the party througought the region in which proportionality is applied, except the candidate who is also nominated in the voter’s SSD. In this way, candidates who are “hedging” on a possible loss in their SSD would have an incentive to make themselves known outside their SSD, because only from party voters on the outside could they obtain votes that would assist their ranking on the party list.

Obviously, this proposal would work well only where the magnitude of PR allocation is relatively small–unless one is willing to tolerate extremely long lists of candidates for party representative. It would not be practical where the PR lists are as long as in New Zealand (50 plus) or the larger German states. However, there are ways to have national proportionality while having smaller regional lists that would make campaigning feasible. (In fact, even with its closed lists, Germany does this, having both state party lists and nationwide proportionality.)

The critical part of the proposal is the ban on a candidate for party representative being listed in the district in which he or she is also a candidate for district representative. Well known party leaders would be able to run both in a district and outside it, and could win in either, because they are well known. But candidates who are lesser known would be encouraged to accept only one kind of nomination, in order to focus their efforts. Thereby the public relations problem of zombies would be eliminated, because SSD losers would either win based on being known outside the SSD where they lost, or would also lose the party-representative race–or would not have run for party representative in the first place.

38 thoughts on “MMP and dual candidacy

  1. I’m curious about your take on the Bolivian use of MMP (which uses the German system). Among the problems there, is that the SSDs are determined by simple plurality. On average, the winners have won (in the 1997 & 2002 elections) by about 20% of the vote (these are multiparty contests, after all). So these are minority winners in winner-take-all districts (often, the margin beetween winner & runner up is slim).

    But since the arrangment of the proportional half of the legislature is compensatory, who wins the SSDs has a dramatic influence on how many seats are won by parties. In districts w/ small magnitudes, this has lead to very disproportional outcomes.

    I’ve thought a solution might be some form of “vote pooling” method to ensure that the SSD winners are also Condorcet winners. In my idea, for simplicity, the losing candidate’s votes would be handed over (en bloque) to some other candidate after the candidate made a public statement (say, three days before the election) about what candidate to transfer his/her votes to. W/ the added benefit that such a process might encourage moderated inter-party barganing strategies, rather than polarization (as is the current case).


  2. I agree an open list would be better. I’d mention the danger of the main party leaders getting a huge share of the votes and the priority of candidates further down the list becoming almost random because of the giant votes locked up by the leadership.

    I won’t say that because I’d end up saying the solution to wasted votes is STv and I’d have to mention excessive STV partisanship at my next confession. In Wales, the Richard Commission looked in some detail at dual candidacy and recommended STV.


  3. I think Matthew’s idea about open lists in MMP systems is a good one well put. In the situation in Scotland and Wales, it would be manageable because of the low size of their district magnitudes. The idea of excluding the local candidate from the list in that area is an interesting one too, in that it would reduce the force of the ‘elected by the back door’ critique of list members. I think the evidence from open list and preferential systems is that people do vote for the leader, but not in sufficient proportion to make the lower placings or orderings meaningless. Still, I like the design of the two-column New Zealand ballot which is perfectly clear about the purpose of both votes that are cast.

    Some countries, like Germany, seem to have few problems of this nature. I recall in 1990 that Helmut Kohl won his own constituency for the first time in ages, and thinking how odd that was! But people in Germany don’t, and I don’t think Kohl lost any authority or legitimacy in 1982-90 because he was a list member.

    The design wrinkles in MMP that bother me much more than dual candidacy (which I believe may be necessary, in either the Welsh or Shugart variant, to ensure that small parties can participate on reasonable terms) are party-jumping and the front-party fiddle. By the latter I mean the possibility of nominally separate (but in fact allied)parties running a slate of candidates and a list under separate names, and converting it in effect into a MMM system. The fact that people seem not to have done it suggests that there is a disincentive to cheat the system because some other authority (a higher tier of government, constitutional court or referendum) could retaliate. But in a Westminster-style system, it’s a risk.


  4. I should have added that the 1998 Jenkins Commission in the UK designed a system called ‘AV+’ for the House of Commons. The constituency seats would be allocated by preferential voting, and there would be a small (15-20%) compensatory top-up in relatively small list regions, only electing one or two list members at a time. Jenkins proposed that these be elected from open (short) lists.

    I see the appeal of AV+, although I have reservations. It uses two different modes of voting (preferential and categoric) in the same ballot paper, which is complicated and less than ideal, and it’s not very proportional.


  5. Just to clarify, and in response to Lewis’s two previous comments: In open lists, strictly defined, only preference votes affect the order. There is no list order established by the party, as there is in what are often called ‘flexible lists’ (where the list order prevails, except for any candidates who obtain some stipulated quota of preference votes to ensure their election regardless of pre-election list rank). With those definitions in mind, it is my understanding that the Jenkins Commission actually was understood to be proposing (even if it was not explicitly stated) a flexible, rather than strictly open, list.

    So, in flexible list systems (Belgium, the Czech Republic, and the Netherlands, for example) it is indeed the case that most voters vote for the party’s national or local leader and the preference votes for other candidates make little difference. But in open lists (Finland, pre-1994 Italy, Brazil), even extremely small differences in votes won by candidates can be crucial to making the difference between being elected or not. This can be a drawback, as implied by Alan’s comment (above) about the essentially “random” treatment of down-list candidates.

    In fact, as a result of the incentive for candidates to seek preference votes when these alone determine who wins the party’s seats, such votes in open-list systems tend to be far less concentrated on the more prominent candidates than is the case with flexible lists. But some are indeed winning with very small shares of the party’s votes, especially if the district magnitude is large and even more so if voters are not required to select a candidate (rather than just give a party vote). (These issues are matters I am looking into in research now underway.)


  6. AV (IRV) is not such a bad thing for MMP systems that have low compensation magnitudes, given that many seats may be won by a party’s candidate who is well short of a majority. When that happens, proportionality can be sharply reduced because there won’t be enough PR seats to make up the difference. (Miguel, above, noted this problem with respect to Bolivia, where MMP with varying–and some quite small–PR districts is used in a context of a very fragmented party system.)

    The Jenkins proposal was rather ridiculously complex, including both the alternative vote in the single-seat districts and flexible lists. I wrote about this odd combination (and the overly small PR magnitudes) in a special issue of Representation devoted to the proposal.


  7. This whole topic has been much debated in Canada. By far the best material, including original research on Germany not previously done even in Germany, is in Louis Massicotte’s Report [PDF]. See especially Chapter 10 “Double Candidacy, The Key to Compensatory Systems.”

    As for the Zombie fallacy, Massicotte kills it. In Germany, list deputies compete with local deputies. In Wales, Labour AMs, and in Scotland, Labour MSPs fear and loathe this competition, and suggest it bothers the voters. It does not. Get used to it, says Massicotte.

    The question of open or flexible lists has also been debated at length. The Law Commission of Canada’s Report favouring MMP recommends flexible lists:

    However, the excellent Report of the New Brunswick Commission on Legislative Democracy (PDF) recommends closed lists, not only to help parties promote women, but also to help parties maintain geographic balance and prevent the urban centre of a district preferring its candidates and short-changing the outlying areas.

    And yet, most people expect that Canadian voters will insist that all candidates face the voters. Hence, flexible lists, and small natural regions linked German-style for proportionality, are the other possible answer.

    Quebec’s model, on which public hearings are about to resume, is a “large-party” model with five-seat districts (three local deputies, two regional), the highest threshold in the world, with closed lists and a single ballot: a model made in heaven for the governing Liberals, and in fact made in their caucus. Will it survive public scrutiny?

    The Bavarian open list model, considered and rejected by New Zealand when reviewing their MMP model, has few fans here, for the very reason mentioned by Alan above: the main party leaders get a huge share of the votes, and the priority of candidates further down the list becomes almost random.

    As for the suggestion of regional candidates being ranked only by those outside their local riding, imagine a nine-seat district: five local seats, four regional deputies. Three of the local seats are in a larger urban centre, while the other two are both from smaller communities. Regional candidates from the urban centre will still be ranked by two-thirds of its voters, while candidates from the smaller centres lose their only advantage. No, thanks.


  8. Excellent information, Wilfred, especially the “zombie” fallacy in Germany. The NB Commission’s recognition that closed lists are actually better for women than open lists is impressive. The conclusion is counter-intuitive yet correct, according to research by one of my (now former) PhD students.

    The Quebec proposal seems rather similar to the Jenkins Commission report in the UK. I have not looked at Jenkins in a while, but wouldn’t its effective threshold be even higher than Quebec’s? I wrote a piece about Jenkins in Representation in which I declared it a “worst of both worlds” form of MMP.

    I am doing research on open and flexible lists, and I agree that a problem with open lists is the “randomness” of who gets elected farther down the list. In that sense, it is at least as bad as SNTV (and, of course, open lists are indeed SNTV inside the party).

    I prefer flexible lists, but many of them seem to have the flaw of making the effective threshold for a candidate’s upsetting the list order too high. However, it is always possible that the reason most preference votes go to candidates with high enough party-provided rank to have won anyway is that parties are particularly attentive to voter preferences in such systems, thus anticipating what the voters want. But it is also possible that any such effect is an illusion, and the systems are really more or less closed lists. I am hoping to get at this very question in my own research (though I have not yet figured out just how!).


  9. There seem to be as many open list mechanisms as countries.

    In Iceland, you can rank all the candidates. A first rank counts as one point, a second rank as a 1/2, a third as 1/3, and so on. You still have only one vote for which party you support, but within the party’s list, the candidate with the most points is first in line for a seat.

    In Belgium, if you just vote the slate, that gets only half weight in ranking list candidates, giving more ranking weight to voters who rank the list. They introduced this weighting recently, to create more voter interest and choice, because it was too hard to break the slate. The result, by recent election results, is that the best position for a candidate hoping to break the slate is at the very bottom of the list, where ornery party supporters give their preference. It seems that the presumed desire of Belgian voters to choose one person from a provincial list is exaggerated, perhaps because the province is too large a region, although Belgium does not link the lists of the different provinces (understandably in a bi-national country,) making it problematic to use smaller regions: the threshold would be too high. The new Swedish optional personal vote may work better, in Sweden’s more numerous regions with a national compensation feature. Women feared this would hurt them, but it didn’t. As Lenita Freidenvall has noted:

    The preferential voting (PV) system was introduced for the first time in Sweden in the general election of 1998. In this election 12 MPs, 5 women and 7 men, were elected who would not have been elected if not for the PV system. In the Riksdag, this reform meant that one more woman was elected than with the previous electoral system. The PV system has thus, to the contrary of expectations, been positive for women. In the 1998 election women’s representation increased from 40 to 43 percent. In the 2002 election 10 MPs, 6 women and 4 men, were elected because of the PV system. 2 women were elected who otherwise would not have made it being too far down on the party list. Changes were however quite small. In the 2002 election womens representation rose from 43 to 45 percent.

    The real problem with dual candidacy was clear in the UK Government’s White Paper on Wales (PDF):

    1.28 The electoral system in the UK has historically attached great importance to the relationship between Members – whether of Parliament, local authorities or the devolved bodies – and their constituents. The present operation of the regional lists in Wales is damaging that relationship.

    1.29 There is considerable dissatisfaction that candidates who are rejected by a particular constituency can still become Assembly Members through the regional list and so be able to claim to represent the constituency that rejected them. It has happened in Wales that candidates defeated in a constituency have become Assembly members via the regional list for that same constituency. This can have done little to enhance the credibility of the electoral process or the voters’ level of engagement with it. It also causes considerable confusion among the electorate that list members can set up constituency offices and seek to deal with constituency case work as a “local member” in competition with the constituency member.

    1.30 The Government therefore proposes to legislate to provide that candidates should be able to stand either in a constituency or on the party list, but not both. This would ensure that candidates who had been rejected by the voters in a particular constituency could not then become members of the Assembly via the list, and claim to represent that constituency.

    The “considerable dissatisfaction” is of course strongest among the Labour Assembly Members. They won 29 of the 40 local seats. Two other ex-Labour independents also won: Labour has a defacto majority, no coalition partner.

    By contrast, the 20 regional AMs are 10 Conservatives, 7 Plaid Cymru (Welsh Nationalists) and 3 Liberal Democrats. Imagine, they have the gall to “set up constituency offices and seek to deal with constituency case work as a “local member” in competition with the constituency member.” Just as bad as Germany! The historical privileges — er, responsibilities — of the local Member must be preserved, mustn’t they. 😉


  10. Reiterating a point at comment no. 5 to this post, as I (and most in the electoral-systems lit) use the term, “open list” means that preference votes alone determine order of election. In other words, there is no party-provided rank. By that standard, Belgium’s lists are not open, but flexible. Of course, this is really a continuum, and as Wilfred notes, the Belgian case has become more “open” recently. But by the usual dichotomous classification, Belgian lists are still “flexible.” Really, the category of “flexible” is not all that useful, given that it really is one continuum all the way from closed (only party pre-election rank matters) to open (only preference votes given at the election matter). That leaves an awful lot of room for “flexible”!

    I am most grateful to Wilfred for the links and case descriptions he has provided, as they are really going to be helpful to me in thinking these issues through. Another big “plus” for blogging: It puts me in touch with people like Wilfred! (And I very much share his irony regarding dual candidacy.)


  11. Blog readers may be interested in the simple explanations of the mixed proportional representation system provided by the Quebec government.

    For example, here they stress how each vote will count, and speak of “division MNAs (Member of the National Assembly)” and “district MNAs” [PDF].

    Here they address the “two types of MNAs” (in each little region they will have three local division MNAs and two district MNAs) and note “Citizens could deal with either the MNA for their division or the one for their district, whichever one they feel would better act on their behalf or better reflect their convictions. This is one direct benefit of mixed proportional representation: it diversifies political representation on a regional basis.”

    Since they are proposing a closed list model, they don’t stress the personalized aspect as much as they otherwise might. Canadian voters may not respond well to terms like “nominal tier” or “personal tier” contrasted with “party tier” or implying “impersonal tier.” With an open list model, I can easily see both levels being stressed as respecting personal choice: “you have a vote for both your local MP and your regional MP.”


  12. This is an appeal for help on procedures used for nominating women in Germany’s SPD.

    In the 2002 Bundestag election, 97% of SPD constituency candidates were also on the state lists. I have seen no figures for 2005, but I expect they are similar. The SPD has a quota system: since 1998, at least 40% of list candidates must be women. According to the 40% rule, it is required that the lists should be zipped, with the option of allocating every fifth place to someone of either sex.

    But if 97% of them are also local candidates in single-seat districts, how is this arranged? If the local “gatekeepers” nominated too many men, women would have to be added to the state list who were not also local candidates. Clearly this is rarely necessary. I suppose some constituencies are allowed to nominate in advance of the state convention, while others are told to wait until later, encouraging them to nominate a local candidate who has already won a place on the state list. But this is a pure guess. I can find no description of actual SPD practice. Is one to be found?


  13. Dual candidacy has received a strong endorsement from the Arbuthnott Commission on the Scottish voting systems.

    The report said: “Dual candidacy only seems problematic to some people because of the legacy of constituency representation within British political culture and the hegemony this creates for some parties.

    “Candidates coming in second or third place who are then elected through the regional list are only ‘losers’ in the context of a first past the post ‘winner takes all’ electoral system.

    “This logic does not sit well within a proportional system and introducing it devalues and undermines the concept of proportionality. The criticism and the pejorative terms in which it is sometimes put, does little to enhance the legitimacy of regional MSPs.”

    The full Report runs to 100 pages [PDF]:

    “In order to give voters more choice over the election of regional members, the closed list system should be replaced by open lists. The Electoral Commission should investigate how best to devise such a system while minimising complexity for voters.”

    “We suggest that the structure for open lists for Scottish Parliament elections would be one where, in addition to the option of expressing a preference for an individual, the voter is able to vote for a party without expressing any preferences for candidates.”


  14. I think that this proposal, for either open or flexible lists for the PR-tier in MM systems, coupled with a ban on having one’s name on both sides of the ballot *within a single district*, is one of these proposals that may appeal to political scientists but to few others.

    It would, I think, not be popular, because it would add a significant tier of intra-party compatition to the existing two tiers of inter-party competition (national and district levels). I’m aware that this happens in low-DM STV systems such as Ireland, and in those situations this system may be manageable. But as I understand it, most countries with MMM or MMP have PR tier districts with DMs substantially larger than 3-5, which characterize Irish districting schemes. As you note, large DM PR tiers would make for highly unwieldy and multi-dimensional competition between large groups of co-partisans. Given the commonness of that situation, this proposal would be of only limited applicability.

    I think further voter confusion would result from the ban on running as a district candidate and on the PR ballot in those districts. Outside of the national leadership, the local MP / candidate is likely the politicians that local citizens like / know best. Some of them may want to support **their** candidate on both parts of the ballot, and would, I think, feel aggrieved when they find out that they are not allowed to fully support a candidate that they really like.

    The question then becomes whether this kind of public relations problem – where people want to support a legitimate candidate but aren’t allowed to – is worse that the zombie problem.


  15. Pingback: Fruits and Votes

  16. In Response to MMP and Dual Candidacy from the 13 Dec 2005, instead of voting for individuals in a party list, why not simply make the party vote go to the candidates from that party who got the most votes, but were not elected? This way, there is no party list, and the voter doesn’t need to choose from a long list of party candidates. Everyone elected will be chosen based on local support, its just that the party candidates need less of it.

    The only disadvantage I see is that you would then have some districts with multiple represesentatives from that district.


  17. “Best loser” provisions exist in some MM systems. While having no lists eliminates “dual candidacy,” it doesn’t address the main problem troubling opponents of dual candidacy: Candidates can “lose” (not get the plurality in a district), yet “win” (be elected).

    My effort (above) at a solution was to make sure that district “losers” become winners based on support obtained outside the district whose voters (or, more precisely, a plurality of whose voters) they have already failed to sway.

    (By the way, thanks, Jeff, for keeping this thread alive! I think it easily qualifies as the most popular thread ever at F&V.)


  18. The debate in the UK about competition between MPs under the Mixed Member Proportional system continues.

    Labour Assembly Members in Wales, and Labour Members of the Scottish Parliament, have been griping about one of the best characteristics of the MMP system: if a party wins lots of local seats, it gets few “top-up” MPs, so the list is no guarantee of election. Regional list MPs hope to run locally next time, as well as regionally. So they compete with local MPs to serve constituents, and have constituency offices. “Bad for the politicians, good for the voters.” There are no second-class MPs. But in the UK, locally-elected Labour MSPs and AMs are not yet used to the competition, and resent it.

    This has been well settled in Germany for decades. “List MPs” maintain constituency offices in the local riding where they were defeated last time, and in which they plan to run again next time. Even an MP elected from the small Left Party from the Bavaria list alone, without running locally — a rarity — and with no hope of winning locally next time either, does her duty by adopting a riding near Munich won by the conservatives, where no local candidate won a list seat: Kornelia Möller stepped into the vacuum to give local citizens a competing “shadow MP.”

    In Scotland, the Arbuthnot Commission held hearings last year on revising the Scottish system. Their recommendations this January were well-founded and ring true to a Canadian electoral reformer. In particular, on dual candidacy the Commission said:

    9. Candidates for election to the Scottish Parliament should not be prohibited from standing in a constituency and on the regional list at the same election.

    However, in Wales the Labour AMs’ campaign against competition has been more successful.

    They persuaded Tony Blair’s government to support them. A bill to that effect passed the House of Commons but was defeated in the Lords for obvious reasons. There was some hope that the government would be embarrassed to override the Lords on such a blatantly partisan fiddle with the voting system. However, this past Tuesday the government did so: “The government reversed a Lords defeat on moves to ban so-called “dual candidacy” in Welsh assembly elections.”


  19. What about a one vote mixed system? There would be a preferential ballot where voters could rank candidates. But it would have two tiers: the SMD and PR. The SMDs would use AV and the PR seats would be allocated to parties reaching a certain threshold of the vote. But a winning candidate’s party in a district would transfer the 1st votes to voters 2nd choices and so would parties not reaching the threshold. Then voters would understand that their vote can actually elect somebody!!


  20. Pingback: Fruits and Votes

  21. Matthew – banning dual candidacy, or implementing open lists under MMP would risk opening significant intra-party problems.

    The New Zealand National Party has already faced significant problems from having electorate candidates who are not also list candidates (although the 2003 NZNP reorganisation prevents this*). Introducing another aspect of intra-party conflict could easily de-stabilise the party and force a split into a liberal and conservative party (British meanings not US meanings).

    However, I do think it will end many of the problems of ‘Zombie’ MPs and also give list MPs a personal mandate, which many feel they lack.

    *The 2003 reorganisation is the topic of my MA thesis


  22. I think the legitimacy problem for MMP has tended to be underrated. A bigger problem is that the list members for MMP are by necessity “overhang”. In theory, a party can perform well in the election and elect no list members at all, because it has elected more than enough constituency members to be represented proportionally in the legislature. Its common for small parties to have their entire delegation composed of list members. Even without the problem of members losing a constituency election and bouncing back via the list, this creates the question of what the MMP list members actually represent.

    Note that in a system where all members are elected proportionally, from party lists, you don’t have this issue. The more votes the party list receives, the more members it elects so party list members can be said to be “elected”. With MMP there is no direct translation between votes and the number of list members returned, since the relationship is between votes and list plus constituency members.

    I think STV more effectively combines an element of constituency representation and an element of proportionality, though they are both somewhat diluted compared to single member distrit and more pure PR systems. Alternatively, elect one chamber using single member districts and one using PR, as is done in Australia.

    If MMP has to be used, I would suggest that ALL list members be candidates who have lost in the constituency elections. The catch would be that the order on the list would be determined by the number of votes the defeated candidates received in the constituency elections. This would give all the candidates running in constituency races, even in safe seats for the other party, and incentive to try to run close races and get their voters to the polls. In single member districts, you sometimes get situations where defeated candidates in close races wind up with more votes than many elected members. This would give some districts effectively double reprsentation, but these would be districts with high voter turnout and which are not fiefs of one party or another, which would also create positive incentives for voters.


  23. Thanks for the thoughts, Ed. I hope others will propagate.

    Just one terminological point: I understand “overhang” to mean, by definition, a number of SSD’s greater than the party’s proportional allocation.

    And a point of clarification: of course, we can’t expect a party’s list seats to be proportional to list votes if it won any SSDs, because MMP is concerned with overall proportionality (across the two tiers) and not with proportionality of the list seats, per se.

    The ‘legitimacy’ questions are, of course, normative–what do we value? But let’s start with clear terms and consideration of the mechanics of systems before getting into their legitimacy.


  24. I’ve always been a little bothered by “best loser” MMP. Two reasons:

    1. As you mentioned, there’s the disparity between districts when some have more representatives than other. “Swing districts” already have a lot of power since their MPs have to face real competition. They’re likely to also have most of the best-losers; do we really want to give these districts even more advantage?

    2. In a best-loser system, it’s impossible to vote *against* someone. There are so many incumbency advantages in elections (and political power generally) that I want to make sure voters can express displeasure in their MP. This is difficult in list systems in general, but at least each vote for an alternative list lessens the chance of the disliked-MPs election. Under best-loser, however, a vote against a candidate (compared to abstention) may not lessen their chance of election.


  25. Why not use the “best loser” method but insure that every district gets exactly one additional representative? Have as many compensatory seats as there are single seat districts and use a seat-at-a-time method for distributing the extra seats. Each seat that gets awarded to a party immediately goes to the party’s candidate who has the most votes but has not yet won a seat, and who is running in a district that has not yet received a compensatory seat. Once a district has received its second seat, no more candidates from that district are eligible to win seats. This way we would have proportionality, and both representatives from each district would be dependent on the support of local voters to win and retain their seats.


  26. I probably don’t understand Chris’ suggestion. In a two party system, it would result in every district having a representative from each party, and the legislature always split evenly as a result of an election.

    I don’t have a big problem with areas with high voter turnout getting a second representive. It strengthens the relationship between representives and voters, and voters in other areas can “fix” their underrepresentation by voting in higher numbers the next time, and not always voting as a bloc for one party.

    One other thing I would suggest for MMP would be to ban re-election via party list. You could lose or not win a consitutency race and still get elected via list the first time you run. If you want to stay in the legislature, you better find a constituency. Pace Vasi’s objection to “best loser”, voters could now kick representives they don’t like out of Parliament since defeated incumbents won’t bounce back in via the list. In fact, this would go some way to fixing the problem of politicians rejected in the constituencies appearing on the lists.

    I realize I used “overhang” incorrectly, or there is a more common use of the turn. I was referring to the fact that under MMP there isn’t a direct relationship between votes and list seats, looking at the list in isolation. The party’s total representation will reflect the number of votes it receives. But for list candidates only, their election depends not only on the total votes gained by their party, but also on the fairly random factor of whether their party wins a large number of constituencies given its overall vote total. Should the list candidates in MMP be steering large parties to greater geographic concentration, since they benefit from the party piling up votes in its strongholds that can’t be use to gain more seats? Should list candidates for smaller parties push their parties towards greater dispersal of the vote? I don’t like the fact that you can’t directly vote for a MMP party list except maybe by voting for the list and making sure you vote for another party in the consitutency race.


  27. Dropping this link with comment on NZ’s version of MMP plus electoral reform here as the most recent NZ posts are closed to comment.


  28. [Re NZ, I tried to post this on the thread to that topic but they won’t accept any new comments – MSS, please transplant at will…]

    Recent news item on proposals to give NZ’s North Island and South Island official names (probably in Maori). Apparently someone did some research and discovered the two islands have no official names that are enshrined in legislation.

    Which mildly surprises me because I read, two decades ago, that NZ’s then (pre-MMP, SM-FPTP) electoral law fixed the size of the House by laying down that the South Island was guaranteed 25 seats – no more, no fewer – than the North Island got 1 seat for every quotient equal to 1/25 of the South Island’s population.

    Wasn’t something similar used, or proposed, for the Canadian House of Commons – permanently allocating 75 MPs to Quebec and then using the average as the quotient for the Rest Of Canada?


  29. New Zealand should be guaranteed its own parliament, legislative competence with respect to Maori and Pacific issues, its own foreign minister, 24 seats (twice the usual state delegation) in the common senate and proportionate representation in the common assembly. Ideally the principle of separate indigenous representation would be carried into the common senate. I’ll run and hide now.


  30. Island naming was subject to a bit of a beat-up in some media, what will very likely happen is that the over-looked formal gazetting of NI + SI will occur, with official Maori alternatives being listed as well.

    Under MMP, the SthIs gets 16 seats (this presumably covers various small-population other islands, some of which are in a different timezone), this then sets the target size of NthIs and Maori seats as well.

    The SthIs seat count is driven by the physical size of the district that one MP has to cover. Part of the post-election agreement between the National and Maori Parties was that the larger districts (all but one of the Maori seats and IIRC 2-3 of the general ones) would get funding for an additional office to serve constituents.


  31. If the costs of opening an office and staff were the same in the cities and the countryside, I can see the rationale for extra funding for large, rural districts. But in fact the office and staff will be more expensive in the smaller, more densely populated districts. Wouldn’t the extra distance and higher costs cancel out?


  32. Please, folks. This thread is about dual candidacy in MMP.

    As for the blog shutting down comments, I have no idea why this happens. I am not happy about it, as the efforts some of you are making here to revive or add to a previous topic are very much one of the things that makes this blog work. But so is keeping things classified by topic as much as possible.

    The closing of comments on older posts is not a setting I ever changed. It did it itself, strangely enough.

    I can manually re-open some (one by one), and will do so for one or two NZ plantings.


  33. I see that Lundberg’s famous 2005 study of Scotland, Wales, Hesse and Brandenburg is online.

    It is somewhat useful analysis of competition between regional deputies and single-member-district deputies. But his anecdotal evidence and quotes are far more useful than his statistics.

    I also find it annoying that he didn’t use Bavaria, with its open lists, as one of his German states. The results would have been much more interesting. Presumably Bavarian regional deputies behave even more like local deputies.

    But his stats fail to distinguish between genuine “dual candidates” and token ones. In Germany, CDU and SPD deputies regularly swap roles from list member to local member, as the parties’ fortunes ebb and flow. List members are always acting as shadow local members, with constituency offices, sometimes listed on the legislature website as the member for that district “elected on the list” along with the other member “directly elected.”

    But the Green and FDP members have no hope of winning the local seat, and being fewer, they must serve up to ten local districts each. Naturally their behaviour is different from local deputies. They follow the norms of the political culture by having constituency offices, like the New Zealand Green Party MPs and the Scottish Green Party MPs who didn’t even bother to run locally, but it’s a more token effort than those whose political life is likely to depend on it.

    Lundberg explains all this. But first he purports to analyze his useless stats which lump all “additional members” of large and small parties together. Why bother?


  34. Pingback: Fruits and Votes

  35. Pingback: Thailand political reforms (yes, again) | Fruits and Votes

  36. Pingback: Not ‘zombie’ but what? | Fruits and Votes

  37. Pingback: Open lists in MMP: An option for BC and the experience in Bavaria | Fruits and Votes

Leave a Reply

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

You are commenting using your 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 )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.