MMP in NZ: An example of “best of both worlds” in action

In Shugart and Wattenberg (2001) we ask if mixed-member systems offer a “best of both worlds.” That is, do they allow simultaneously for the benefits of local representation and individual-member accountability that are the (supposed) advantages of single-seat plurality (FPTP) and the representation of smaller national parties that might struggle to win districts but would be represented under proportional representation (PR).

There was a question mark in the book’s subtitle. Over time, I have come to believe that indeed the proportional type (MMP) does have a strong tendency to offer the best of both worlds. The reason is that members elected in districts have incentives to behave as local representatives at the time that there is close approximation between party vote and seat shares (assuming compensation is carried out nationwide or in large regions). The majoritarian type (MMM, as in Japan and Taiwan) probably does not; it is much closer in its overall incentive structure to FPTP, even though it does indeed permit smaller national parties to win seats.

For MMP, the “best of both worlds” argument assumes that parties nominate dually–meaning many elected members will have run in a district and had a (realistically electable) list position simultaneously. If they do, then even the list-elected members will have a local base, and should have incentives to act as the local “face” of the party, including possibly by offering constituent services. Both prior anecdotes I have shared from New Zealand (e.g., “shadow MPs” who win from the list and maintain a local office) and my forthcoming coauthored book, Party Personnel, offer further evidence that MMP does indeed work in this way.

Now comes a terrific anecdote from New Zealand’s 2020 election. In this election, Labour won a majority of seats (64/120) with 49.1% of the nationwide party list vote. In the nominal tier of single-seat districts (electorates) it won 43 of the 72 available seats. Its win included some districts that are normally strongholds of the center-right National Party (which won 35 seats overall and just 26 districts).

Commenting on some of the Labour wins in mostly rural districts, Federated Farmers president Andrew Hoggard said:

in some “flipped” electorates Labour list MPs had worked hard to raise their profile and get involved with the community and this had paid off when they campaigned for the electorate.

This is an ideal description of how the “best of both worlds” argument works: list-elected members have incentives to attend to local needs of the district in which they ran for the nominal seat (but “lost”) in hopes of capturing the local plurality in the next election.

Of course, there were other factors at work as well. I will offer another planting about one of those factors separately. There is also some uncertainty at this stage just exactly the degree to which rural voters flipped, as the wins may have come in significant part from very large swings in the town areas within districts that also include large rural areas. Regardless, MMP offers the key advantage of giving most elected members, if dually nominated, a tie to a local constituency while ensuring close approximation of overall seat totals to party-list votes.

35 thoughts on “MMP in NZ: An example of “best of both worlds” in action

  1. Pingback: NZ 2020: Strategic coalition voting? | Fruits and Votes

  2. I wanted to test how far an empirical 43-26-1-1 result in single-member plurality seats diverges from what the cube rule would predict… but the total number of local electorate votes cast for each party nationwide seems to be a state secret, bro’. No shortage of information on how many nationwide list votes each party polled, and I suppose I could go through and add up the results for all the National, Labour, etc candidates in 72 districts, but surely someone else on the internet must have done this directly? I understand that the parties’ totals of Erstestimmer and Zweitestimmer can diverge by significant percentages, so I don’t want to apply the cube rule just using the Sikkind Billits on their own.

    • MSS, your thesis is persuasive. A large part of my rationale for ranking MMP a very distant second behind STV is a feeling that PR supporters should give no quarter whatsoever to the anti-PR crowd’s claim that voters are deprived proper local representation if you don’t carve the polity up into fairly arbitrary districts and assign exactly 1.0 representative to each.
      (I say “arbitrary” because (a) they have to be fairly close in population, unlike STV which allows different magniitudes to fit different historic regions and (b) even with uniform magnitudes, STV districts are 5 or 7 times larger so you are less likely to end up in a different electorate without moving house because the boundaries were adjusted).
      What MMP actually provides is not “single-seat electorates” but more like “1.5-seat electorates” – you get “your” local member, whom you are 45% of so likely to have voted for, but another 40% or so have access to a shadow local member from the list, whom they can still approach. True, you can’t really vote either in or out of parliament https://www.canberratimes.com.au/story/6970524/this-act-election-is-one-to-watch-closely/ – except in rare cases, voters determine only the overall number of seats in parliament – but your first (local) vote can affect which of the Labour and National nominees gets to claim pride of place at local ribbon-cuttings and giant-cheque-handings-over, which I suppose is something, and avoids the grave danger to democracy of voters getting to decide which individual candidates enter the Beehive, as the Royal Commission warned against in 1987.

      • So your reason for not ranking is MMP higher than STV is… because “PR supporters should give no quarter whatsoever to [an] anti-PR crowd’s claim”?!

        Since we’re all likely to be pro-PR here, shouldn’t we be ranking the various systems based on their own merits, rather than on which spurious arguments against reform those with vested interests might use?

        Here in the UK, STV would mean constituencies in the Highlands of more than 20,000 square kilometres, which sounds tiny in the Australian outback but would feel very remote by historical British standards. And STV units with maximum proportionality (6 seats) would involve constituencies with populations of over half a million people, which would be much less easily based on natural communities than we are accustomed to, unless you want a 3,000-member House of Commons.

        With MMP you could have, for instance, 450 single-member districts only slightly larger than the current ones, plus 200 compensatory seats to keep the Commons the same size it is at the moment.

      • Thanks, Manuel, yes I did miss it, I was trying Google, Wikipedia and a couple of NZ Govt sites.
        So I see at http://electionresources.org/nz/house.php?election=2020 that Labour’s candidate votes were 2% lower and National’s was 9% higher than their list votes nationwide, which probably explains why the result at district level was closer.
        With the cube rule, 47.5% to 35.8% “should” distribute 69 single-members seat in ratio of 107172:45,882, or 48 to 21. so the actual 43 to 26 is only about 20% off. Do the boundaries consistently favour National, ie rural areas? NZ used to have a very explicit vote-weighting, 1881-1945 https://en.wikipedia.org/wiki/Country_quota but this was abolished and the current law permits only 5% deviation from the average, even though MMP would ensure proportionality with wider population divergences.

      • The National Party didn’t contest any of the seven Māori electorates, and won just 3.5% of the party vote in them. However, it contested all 65 general electorates, where it lost the electorate (FPTP) vote to Labour, 47.1% to 38%, and 37-26 seat-wise, with the Green Party and ACT New Zealand winning one seat each.

  3. What about decoy lists where all the parties conspire to mutate a MMP system to a MMM system? Then MMP is not the best of both worlds in that scenario. What about a MMP system where the SMD is extremely fragmented and only the top two parties surpass 5%? Then there is a party winning 3.5% of the vote and no SMD yet parties with fewer list votes get seats because 1 party won a SMD, the thresholds are set at 5% or 1 seat. Are there any MMP systems with no thresholds? Is it better to have a threshold or not?

    • I believe it’s possible to design an MMP system to protect reasonably well against decoy lists. Use a single ranked-choice ballot.

      I also disfavour an arbitrary threshold; I’ve always thought that “one seat’s worth of votes” is the only truly defensible one. I don’t see how someone can support PR, but not also want every voter to be able to say “my vote helped elect this representative”.

      Sometimes it’s said thresholds are to keep the fascists out. I too would like to keep the fascists out, but I also think any sufficiently high threshold is also going to preclude almost every other minor party; the percentage of “uncomfortably far right” people in the modern west is above 10%. That’s a lot of “collateral damage”.

      • I think the key stated advantage of a threshold (which is not necessarily an arbitrary figure: PR systems that use districts have ‘effective’ thresholds that make “one seat’s worth of votes” insufficient to win a seat) is not necessarily about keeping fascists out, but about reducing fragmentation full stop, whether from the left, right or centre.

      • An explicit threshold also makes the de facto cutoff more a matter of explicit choice as the size of the legislature expands (or sometimes contracts) with population. With any more than 18 seats, the cutoff will be below 5% (and almost no national or even state-level assemblies have under 19 seats). It may be seen as more consistent that the question of whether the Democratic Liberal Party with 1.9% qualifies for representation is decoupled from the question of whether the National Legislative Assembly should be expanded from 190 to 210 seats because the population has risen.
        (The Australian House of Reps tends to expand in unusually large leaps – from 75 to 120 in 1948, then to 144+ in 1984 – but this is because of the 2:1 nexus ratio; with six states and two classes of Senators, no increase under 12 MHRs is constitutionally valid and no increase under 24 MHRs in politically feasible, because no one wants 6 Senators per State at one election and 7 and the other: nothing personal, Brazil).
        Too-small legislative chambers are a nuisance that dilutes democracy. At present, attempts to bring them closer to the cube-root ideal run into two objections: (a) that it will cost money to pay more legislators (true, but trivial, since the costs of an MP are a tiny drop in the budgetary bucket) and (b) that it means “bigger government” in the sense of more intrusive and pointless regulations (absolutely unsupported empirically – note that the small-government Lockean purists of New Hampshire have a 400-seat assembly that is huge in both absolute and per-capita terms – but surprisingly widely-believed. By this logic, a senate of only 35 Williams Proxmire would be fourfold efficient at regulation-busting, but.. oh, why bother).
        Adding in “a larger assembly means a party that is currently considered ‘ratbag’ becomes ‘respectable'” adds extra lead to the saddlebags.

      • The way to prevent decoy lists is not to have lists at all, as in Baden-Württemberg.

  4. Clarification: “With any more than 18 seats, the quota (Hare, Droop, or Imperiali) will be below 5%.” Whereas with fewer seats, a 5% cutoff may well be below the quota, although it can still “bite” since parties with a sub-quota might still have qualified as one of the largest remainders or highest averages.
    Many European polities apply a threshold within a sub-national region – 5% in Belgium, IIRC, and also in France during its 1986 PR revival phase – and 3% in a province for Spain. In these cases, the average number of seats per regional constituency is usually under 20, so the threshold may not “bite” in many cases. It can be seen as ensuring some consistency across regions, given the advantages of using fixed boundaries based on historic regions with seats reallocated according to population shifts, and thus district magnitudes vary widely.
    So, if Madrid elects 36 deputies, the superimposed 3% threshold supersedes the “natural” quota of around 2.7% (possibly lower with multiple competing lists, although not too much lower since Spain uses D’Hondt). This reduces – slightly- the variation in thresholds between Madrid and the least populous provinces (with 2 deputies each), where the quota of around 30% will supersede the formal legal threshold of 3%. A more typical 4% or 5% threshold would reduce the variation still further, of course at the cost of summarily throwing more ballot-papers in the bin.
    Whereas in Switzerland, with a similar constituency arrangement but no explicit threshold, the effective threshold in the cantonal constituency of Zurich, with 35 deputies, will be just south of the D’Hondt quota [*] of 2.77%. In theory, if any Swiss Canton ever qualified for 50 or more National Councillors, its threshold could drop below 2%. Meanwhile, with an average of 7.69 National Councillors per Canton, the average quota and threshold across Switzerland would be as high as 11.11 to 12.5%.
    One of my first ever encounters with the work of MSS was reading his 1990 book arguing against legal thresholds as an added layer of arbitrariness, but on this point I would respectfully disagree with our host (if one is going to use party lists).

    • The STV electoral system doesn’t use any arbitrarily set electoral thresholds depending on the small district magnitudes to weed out the extremes. I think that 9 seats would be the maximum district magnitude for STV, as it is naturally set at 10%, maybe with the exception of the last candidate who is elected because everyone else has been eliminated.

      Has anyone devise a STV system with thresholds and how would that work?

      • Rob: STV system with thresholds = “a candidate who is currently the highest surviving preference on less than [five per cent] of the ballots is eliminated before a candidate who is currently the highest surviving preference on [five per cent] or more of the ballots, even if the former has more vote-points [] standing to her credit than the latter.”
        Eg, in an STV count for 25 seats and quota 3.846%, the 8th candidate of the Purple Party, which polled 30%, has 3.07% of the votes (ie, her team’s remainder after deducting seven quotas) but nonetheless defeats the sole candidate of the Transnational Socialists Party with 4.0% of the vote. Because Purple #8 is currently highest on 30% of the ballots, well over the 5% threshold (even though their remaining transfer value is only 3.07%), as against Karl Drago of the TSP who has only 4.0% of the ballots (even though each is still at its full, first-preference value of 100,000 vote-points. [
        ]
        This example assumes that voters closely follow (or at least stay within) the party ticket, but the formal rules do not assume parties, lists, teams. or tickets and would operate the same with candidates standing as individuals.
        I do not say I endorse or promote such a system – just saying that would be the way to do it if the constitution, or the political Overton Window, allowed only voting for individual candidates only (eg, something like that one of the entrenched Hare-Clark principles in ACT law]
        [* “vote-points” is a convenient accounting term to allow for transfer values. If calculations are taken to the fifth decimal place, then each voter has 100,000 vote points, which outflanks those whiners and trolls who complain that a “single” transferable vote denies them their right to cast FIVE!!! votes when filling five seats. We probably need a snappier term than “vote-point”… may I suggest “schlechton”]

      • Every election to the Australian senate produces 1 or 2 candidates whoa rte elated off very small first preference votes. Thisleads to a flurry of political, journalistic and social media comment about how unfair it is that Senator X received so few first preferences but was stilleclted went they got a quota of votes. It should be noted that Senator X never belongs to a government party. The government reassures an anxious nation by referring the matter to the joint standing committee on electoral matters.

        The committee holds public hearings. A very long parade of the great and good assures the committee that thresholds are unknown in STV systems, would mean voters casting votes of unequal value, and would make no difference to the number of parties in the senate. The committee then writes Anti-Threshold Report #287, which bears a marked resemblance to Anti-Threshold Report #286 and all its predecessors before it, and the nation happily resumes its accustomed calm.

      • During the 2016 Australian Senate electoral reform process, a number of parties proposed a threshold based on group totals (before you begin the normal STV process, exclude any candidates who are part of groups with less than, say, 4% of the vote). This would have ‘worked’, in the sense that it would have weeded out the most egregiously unrepresentative results of the Group Ticket Voting system without substantially changing the way people vote, but ultimately didn’t get anywhere coz it wouldn’t have been guaranteed to solve the problem (winning a seat with 4% under DM=6 thanks to group tickets would still be a little unrepresentative) and would have been questionably constitutional since Senate elections need to be “direct” and it’s not clear whether that’s compatible with treating candidates as part of parties during the counting process.

      • I believe South Australia once tried STV with a threshold and it produced a total mess that was swiftly abandoned.

        The Australian High Court’s interpretation of its country’s Constitutional articles related to Senate elections is flawed. It fails to make the distinction between what the Germans call ‘unmittelbare Wahlen’ (i.e. direct as opposed to indirect elections) and ‘Direktmandate’ (i.e. each elected member being able to claim some kind of personal mandate). The Constitution explicitly calls for the former, but the Court interpreted this as a requirement for the latter – hence the use of a preferential system. If only senior judges Down Under spoke German, they might accept that list PR would be a perfectly valid system for the Aussie upper house!

      • Hi Olivier,
        No, South Australia (1973-1985) used preferential voting among closed lists, not STV. Any list under 0.5 a Droop quota (ie, 4.1667% with 11 seats statewide) was summarily eliminated at the first count, its votes were transferred, and then seats were allocated by Droop quota and largest remainders, without any further transfer of preferences. (As I’ve noted before, this is a bad system because if you vote for a party under 0.5 quota your votes may get transferred, if you vote for a party over 1.0 quota it is sure to win at least one seat, but if you vote for a party between 0.5 and 1.0 quota there is a risk – growing less as you get closer to 1.0 quota, but still – that your vote will simply get binned).
        SA replaced lists with “closed-ticket” STV in 1985. The ACT, similarly, used preferential voting among open lists in 1989 and 1992, with a first-count summary exclusion threshold equal to the Droop quota (5.55% with 17 seats Territorywide). Again, List-PR was replaced by STV after a handful of elections. You may be sensing a theme here.

    • I think ‘the advantages of using fixed boundaries based on historic regions’ might be overstated here a little, particularly in Spain. The problem with the Spanish electoral system really is that small parties are less likely to win seats if they win votes in rural areas than in urban areas. Thresholds add a tiny bit of consistency here, but it’s fundamentally only a slight change: if you win 6% of the vote in Madrid, you win seats, but if you win 6% of the vote across six or so 3-5 member seats, you may not even come particularly close. If you want to address that problem of consistency, you’ve really got to go after the DM.

      • The provinces, which function as electoral districts in Spain have no particular claim to be historic regions. Spanish provinces only date from 1833 (although some, for example Granada, have an older history) and were grouped into larger entities called historic regions until the current political structure was adopted.

      • Provinces only 187 years old sound sufficiently established and tinker-proof to serve as electoral districts (at least provisionally… ideally Spain should consider an Alaska-type rule for automatically merging the smaller districts with neighbouring ones, for electoral purposes, until each one remaining is a decent district magnitude: I believe Ireland used to do this with counties) but that may be my Australian bias coming through, as something raised in a country that, as Tory MP Matthew Parris observed, “has no history”

      • Catalonia has existed as a historic region, in the form of the county (later principality) of Barcelona, since the early 800s. Spanish provinces have a way to go. Spanish provinces are also not natural regions in the sense used by Lakeman because they have extremely limited autonomy in relation to both the central and autonomic governments.

    • I don’t have any principled objection to legal thresholds. What I do prefer is that electoral systems have either nationwide PR (whether one district or two-tier) with a nationwide threshold, or districts but no threshold and especially no national threshold.

      That is, stick to one method for limiting fragmentation, rather than employ both simultaneously.

      In Votes from Seats (2017), Rein and I offer a predictive model based on a legal threshold (assumed nationwide) only, as a “competitor” to the Seat Product Model which is based on assembly size times district magnitude. The threshold model is elegant theoretically but much less successful empirically than the SPM.

      If a system has both districting and a threshold, I do not know which model to use, and I would be less than surprised if neither works all that well for such systems.

      • Your preference may work in unitary countries, but it is unthinkable in Canada that Alberta votes would be part of the calculation electing Quebec MPs, or vice versa. Similarly, any legal threshold would have to be applied at the provincial level. However, with 121 MPs Ontario will have a number of regions (6 to 10; with open-list likely 10). Should a Northern Ontario Party get a seat in Northern Ontario with 0.1% of the provincial vote? Canadians expect a PR model to prevent such excessive fragmentation, especially when 19% of our MPs are elected from smaller provinces with median DMs or 10 or 11, and we may not want two-speed proportionality. So we might well want a provincial threshold (4% or 5%) as well as the effective threshold of the regions.

      • Wilf, I think that’s more of a problem in theory than it would actually be in practice. It’s true that you could have a Northern Ontario Party, a Montreal Party, a Vancouver Island Party or so on, each winning about 10% of their vote in their regions and thus one or two seats in the Commons. However, it’s also the case that this could be a problem under FPTP, and seemingly a more serious one. The fact is that Canada institutionally could be Papua New Guinea under FPTP, but isn’t because the national/provincial effects of the party system are so strong. This would continue to be the case under a regional-based version of PR.

      • Explicit thresholds seem to be less needed in STV systems for the other purpose of limiting the divergence in threshold across different districts. Probably because actually existing STV districts usually vary in magnitude by at most 2 seats (Ireland), if not uniform (Aust Senate, Victoria) – in fact the trend over time is for STV systems to move from divergent to uniform district magnitudes (Malta, formerly 5-6, not 5-all; WA upper house, formerly 5-7, now 6-all; ACT, formerly also the “haiku” 5-7-5 model, now 5-all) or at least to reduce the degree of divergence (Ireland was 3-9 in the 1920s but has been 3-5 since, I think, the 1950s). Proposed models like the UK Electoral Reform Society, the NZ Royal Commission (second choice) and (barring a few single-seaters for the Orkneys etc) the 1982 UK Liberal/ Social Democrat Constitutional Commission) usually propose at most a 3-7 range, and the PR Soc of Australia supports (rather optimistically) a 5-9 range. So even with non-uniform magnitudes, STV quotas will at most vary between 10% and 25%, or more likely (assuming STV is ever adopted) between 12.5% 16.667% and 25%.
        Contrast the much wider range in PR-list systems that use sub-national electoral districts for the final determination: 7-36 for the Finnish Eduskunta (1-36 if we include Aaland, but see above re the Orkneys), 2-36 for the Spanish Chamber, 1-35 for the Swiss National Council.
        The push, such as it is, for supra-quota thresholds in STV systems is less to cap divergence in entry points across constituencies than to screen out minor parties even with uniform quotas and district magnitudes. (To be pedantic, I should say “as well as the quota” rather than “higher than the quota”, since a 4% cutoff in a 6-seat Senate race is not actually higher than the quota, but could exclude candidates who might otherwise reach the quota).

      • In Canada, the ten top-up regions could be the 10 Provinces and ‘The North’ without breaking the Constitution.

        In the UK the top-up regions could be the four nations, although there would probably be a strong temptation to carve up England further, as at European elections.

  5. [* Yes, D’Hondt does have a quota. It’s the largest number that will allocate exactly the desired number of seats if you divide each party’s total by that number and discard any remainders. Likewise the Sainte-Lague quota is the largest number that will allocate exactly the desired number of seats if you divide each party’s total by that number and discard any remainder less than 0.5. The Danish method, etc etc except read “quotas under 0.333” and so forth]

  6. Well, you could always use first-past-the-post voting to deal with the problem of fascist tendencies in the electorate. That was what happened in Britain. In the 1970s, the National Front was expected to poll in the double digits, but the Mrs Thatcher took over the Conservative Party, gave the country Strong Leadership, and the NF’s vote collapsed to low single digits.
    Historians disagree over whether the UK’s use FPTP caused these far-right voters to change their beliefs, like Kilgrave, or simply dissolved them into piles of ash, like Thanos, but however it worked, it definitely saved Great Britain from the risk that governments would adopt far-right-wing policies in the hope of appealing to far-right-wing voters.
    This is also why the USA uses first-past-the-post voting, which is a simple, easy-to-explain system that always guarantees that the candidate with the most votes will win, along with wise checks and balances like the Electoral College, which may not be either simple or easy-to-explain, and may not always guarantee that the candidate with the most votes will win, but at least it keeps out populists who appeal to “us vs them” tribalism, just as Andrew Jackson designed it to.

    • In FPTP, there is no guarantee that the party with the most votes will win a majority. NZ moved to MMP partly because of the 1978 and 1981 elections where the National Party won a majority despite being the 2nd most voted party across the country. Then when Labour won in 1984, they established a Royal Commission to study electoral reform. Both parties were very unhappy with the conclusions and both outmanuver each other on a referendum, Labour made a promise by accident, then didn’t honor it, and National temped fate with a referendum expecting it to fail in 1993. We all know what happen to that. NZ democracy improve dramatically with MMP despite growing pains in beginning.

      The same is with the electoral college, despite being a plurality at large system, we all know what happened last time. Will it happen again? If that is the case, is it worth keeping? What will the new system be?

      • The Electoral College should be retained for the reasons Tom Round states.

        It should abandon winner-take-all for the reasons Rob states.

  7. Pingback: “Effective Seat Product” for two-tier PR (including MMP) and MMM | Fruits and Votes

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 )

Google photo

You are commenting using your Google 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.