Piggyback MPs, part 2

With apologies to New Zealanders’ somewhat complicated memory of Robert Muldoon, I am sticking to my “piggyback MPs” as a preferred term for members elected under MMP via an alternative threshold to the one based on party-list votes.

Here I want to address briefly the question of whether allowing an alternative threshold, by which a party qualifies for list seats through the winning of one (or more) district seats, is itself a problem in electoral-system design. I have been wanting to address this issue for some time, and some of my thoughts are anticipated by a comment left by Rob at the previous thread.

Up front, let me state that I see no problem with the principle of an alternative threshold. If mixed-member proportional systems are to have a chance of delivering on the “best of both worlds” promise, then one really should allow both worlds to coexist simultaneously. One of those worlds is one in which local concentrations of support for particular parties or candidates are able to attain representation. The other world is one in which only nationwide levels of support for particular parties are worthy of representation. Any one of us might prefer one conception of representation over the other, but MMP is explicitly designed to promote both.

Now, one might respond that one need not have the alternative threshold in oder to obtain both of these worlds. Parties could still exist to target one or a few district seats, and earn their representation that way, without being entitled to any list seats.* I concede that this is a perfectly valid argument, and it seems to be the position taken by the New Zealand Electoral Commission in its MMP Review. That is fine; they have thought much more about these issues, and the needs of New Zealand society, than I ever can do.

However, I think it is a perfectly valid “best of both worlds” provision to say that we want to give incentives to smaller parties to attract support outside their district-based strongholds, while still being able to win representation based on their regional concentration. A very small party may have supporters around the country, but be concentrated in one area. Voters outside the areas of strength have little reason to vote for the list of such a party if it won’t win seats; by the same token, voters in a single district where the party has local strength may have little reason to vote for the party if it lacks any chance to win further seats via list votes obtained elsewhere. (If one seat is expected to affect the balance of power, the second consideration vanishes, of course.)

It seems to me that the decision whether to abolish the alternative threshold should be made not on the basis of disliking particular parties that take advantage of it. (Search on “Key cup of tea” if you are unfamiliar with the debate.) Rather, it should be taken after considering what minimal size of party is considered optimal in a given country’s proportional system. (One can never squeeze out all one-seat parties, as they are at least a latent possibility in any system that has single-seat districts, including mixed-member PR, but one can eliminated the opportunity for such parties to exist to seek additional seats via the list.)

What is the optimal minimum size for parties that win more than just a given district (or two or more), but also win list seats?

I do not know! I will say that there is something fundamentally strange about the current New Zealand thresholds. If you are a party that can’t win a district, you need 5%, which implies 6-7 seats as your minimum. However, if you can win one district, you need only about 1.4% of the list vote to get two seats. The problems of these widely disparate thresholds became apparent most starkly in 2008, when New Zealand First won no seats despite 4.1% of the party vote, while ACT New Zealand won five seats on only 3.7% of the party vote. The difference was, of course, that ACT won an electorate (Epsom).

I would recommend bringing the two thresholds closer together. Any specific proposal would be arbitrary, but then so is the very idea of a nationwide threshold. I might suggest something like 3% (instead of the current 5%) party-vote threshold.** In a 120-member parliament, this would be equivalent to about four seats. Then I’d suggest an alternative threshold, the surpassing of which would entitle a party to list seats, of either one seat and 2% of the list vote, or else require two seats.*** (De-facto, the latter would mean a party needing to clear around 2.5% in order to get more than its two seats; so I would then see no need to specify any party-vote threshold.) Whatever might be chosen, I would suggest the goal should be to bring the two types of threshold closer to one another, based on a consideration of what size is deemed appropriate for a party that is more than just the sum of however many districts it can win.

Basically, I just do not see anything wrong with the concept of having an alternative threshold under MMP. I can understand why some folks would disagree, and I don’t think abolishing the alternative threshold undermines MMP. My point is that having such a threshold is perfectly consistent with MMP, and if one disparages small parties piggybacking on the support of one electorate MP (who might get elected due to explicit or tacit deals with a larger one) then maybe one really should prefer a non-mixed form of PR. That is, maybe that is an argument for being more anchored in one “world” of electoral systems, rather than trying to have the best of both. And there is nothing wrong with thinking only one of these worlds is “good” at delivering representation! Here we really are in the realm of values, not strictly objective criteria.

* This is what the Maori and Mana parties have done, with the former having won multiple districts in recent elections. This is feasible only because of a further “twist” in the New Zealand electoral system, which predates MMP: There are two layers of single-seat districts, one for the general electorate and one set aside for voters who claim Maori ancestry and choose to vote for candidates in these districts instead of the general electorates. Note that the principle of some parties winning based on the ability to win one or a few districts, but having no realistic chance to earn party-list seats, is generalizable beyond the parties representing Maori voters.

** The Electoral Commission said that 3% would not cause any problems, but came down in favor of lowering it only to 4% (see its MMP Review, Final Report, p. 16, paragraph 1.28). Interestingly, the Commission cites Taagepera and Shugart’s (1989) stated preference for reforms to be incremental as justification for not recommending lowering the threshold below 4%.

*** Germany’s thresholds are, of course, either 5% of the list vote or three district wins. Note that 3 out of 200-325 total districts (depending on the time period) is very roughly “equivalent” to New Zealand’s 1 out of 60-70 districts. However, it is probably more difficult for a very small party to have as many as three candidates capable of winning district pluralities than it is to have one, even holding constant the aggregate levels of support needed. Unless, that is, the party is exploiting a specifically regional cleavage.

27 thoughts on “Piggyback MPs, part 2

  1. Surely the rational alternate threshold is the number of votes required to win a district seat. Anything less is going to advantage regionally concentrated parties over regionally dispersed parties.

    Small quibble, both the tangata whenua and New Zealand electoral law have quite a precise definition of Māori which goes beyond claiming Māori ancestry, and indeed whakapapa (a much more complex idea than ancestry in English) is in some ways central to the culture.


  2. Wouldn’t the number of votes needed to win a district seat be 0.41% of the vote (just over one half of maximum potential vote in a district with One-120th of the national population); less really considering that there is less than 100% turnout and a candidate doesn’t need a majority to win a seat.

    Might as well not have a threshold at all. FWIW, I do support an actual one seat threshold rule. Any party that wins a district seat or has enough votes to earn a list seat if it doesn’t win a district gets all of the seats it is entitled to. To limit the influence of tiny parties, I would provide that a government can only be brought down by a constructive vote of no confidence.


    • Absolutely. The principle that the value of your vote should not depend on where you live has been accepted since the 1960s. Thresholds mean that the value of your vote depends on where you live in the political space rather than the geographical space.


  3. Well, as I tried to say in the post, there are different competing principles of representation. If one uses MMP instead of nationwide PR, one apparently wants some of at least two of those principles to apply. I don’t particularly object to geography having a role in determining representation (though I very much object to its playing the dominant role). But if one does object to it, then eliminating the alternative threshold, as several New Zealand parties and the Electoral Commission propose, makes MMP worse, not better. A party like ACT can win its one seat in Epsom, or wherever, but ACT voters elsewhere in the country can’t get representation on the party’s usual national level of support (unless the national party-vote threshold is made much lower).

    I have thought about trying to stipulate a threshold based on what percentage of the national vote it takes to win a district, but such a figure can only be approximated, depending as it does on how many candidates compete in the district in question, variations in district turnout (and population), etc. In any case, such a threshold would be too low, in my assessment. Others may differ, of course, as this is a very normative question.

    On Maori voters, I will defer to Alan. I was using language similar to what the Electoral Commission itself uses in a brief overview on its website, not claiming to describe the more detailed provisions of the law (because that’s way off topic for what I wanted to convey).


    • There is a pretty sad meme in both Australia and New Zealand. although less so in New Zealand, that alleges indigenous people are not really indigenous (usually based on blood quantum claims) and in any case should not have sic ‘special rights’ like separate representation or native title. Writing of ‘those who claim Maori ancestry’ rather than ‘Māori ‘ echoes that meme to a certain extent although of course I know that was not your intention.


      • Alan, I do not think this is precisely the same page that I remembered at the time I drafted my evidently insensitive sentence, but the following quote is similar to what I recall.

        The following factors may influence the number of Māori electorates:

        population change since the last census
        the number of people included in the census
        how people answer the census Māori descent question
        the level of enrolment by people who have indicated they are of Māori descent on the General roll or Māori roll
        the proportion of electors of Māori descent who are on the Māori roll at the end of the Option period.

        [My emphasis]

        I guess I must be really insensitive, because I still do not get your objection, Alan. It is not as if the Electoral Commission requires a DNA test or a genealogical report to prove who is Maori (sorry, I do not know the codes for the diacritic). It is based on how someone answers a census questionnaire.

        I don’t have anything invested in whether my phrasing was correct or not. You are right, if there is something wrong in my phrasing, it was not my intent. But I just don’t see the problem–at least thus far.


  4. Say the one-electorate threshold is abolished in NZ but overhang is not, and in an election one party does win an electorate seat but fails to cross the (only, national) threshold. Now the electorate seat satisfies the definition of an overhang seat, and the size of Parliament should increase by one. Notwithstanding the MMP review report recommendation, would there be merit in not considering the seat an overhang seat, but essentially won by an independent, and keeping the size of Parliament unchanged?


  5. The terminology surrounding overhangs always confuses me, as I think different authors use the term subtly differently. I understand the overhang as any seat in the nominal tier that puts the party winning it above its proportional entitlement based on its party-list vote. So, by that definition, one can’t “abolish” overhangs and still have a mixed-member system, which means one in which there is a nominal tier of seats awarded solely based on votes for candidates. Of course, one can have, or not have, a rule that increases the size of the list tier whenever an overhang occurs.

    From memory (always risky!) the MMP Review specifically recommends both abolishing the alternative threshold and the expansion of the list tier to cope with overhangs. So, I think that does indeed mean that such a seat is essentially treated as if it were won by an independent, given that for purposes of proportional entitlement, the party in question might as well not exist.


  6. “If one uses MMP instead of nationwide PR, one apparently wants some of at least two of those principles to apply.”
    Yes, but not necessarily in both intra- and interparty dimensions. One might argue for MMP’s nominal tier only in the interest of local representation and constituency services, but not in terms of overall party representation.


  7. JD, that is absolutely correct!

    However, I don’t think this element has figured into the debate in New Zealand on these question. If it has, I’d really like to see it. (OK, actually, my submission to the review was largely about the intra-party dimension, but I mean other than me.)


    • Looking back to the report of the Royal Commission on the electoral system, I could find no explanation for the 1-seat threshold, which is mentioned rather casually in the conclusion. But besides that, the district tier is mostly cited in terms of ‘effective representation of constituents’, being said to continue the local representation advantages of FPP, while the PR aspect is the one delivering ‘effective [party] representation’. Even if the element has not featured in this specific debate, I’m still convinced it is the logic that underlies MMP in New Zealand.


  8. Those who have endured this thread for so long would surely find interest in Rob Salmond’s blog entry on the potential for “unfair advantage” from electorate deals even if there is no piggybacking (“coat-tailing”).


  9. I continue to find it odd that Germany and NZ (and other party-list-using societies) (1) won’t allow joinder of lists into alliance before an election, but then (2) complain after an election that some smaller party went against public expectations in its choice of coalition partner (eg the FDP in 1972 and 1982, Winston Peters in 1996).
    From Australian experience, the surest way to avoid morning-after surprises following polling day is for the electoral system to encourage smaller parties to ally themselves in advance with larger parties to maximise their joint share of seats. Preferential systems do this with the active support of the voters. Alliances in list-PR systems do this with at least the passive and informed acquiescence of the voters (if you really don’t think the Social Christian Democrats should be allying with the Nationalist People’s Party, you can vote accordingly). MMP, as actually practised, allows this to be done only indirectly, in an underhand and essentially dishonest way, that gives the voters in just one of several dozen constituencies the power to tip a number of seats in a closely-balanced parliament one way or the other by perhaps just a few hundred votes.
    This is not inherent in MMP: as noted, it could be avoided if the system allowed pre-election alliances and vote-pooling.
    As for the problem of dummy candidates winning districts with no list affiliation, this could be solved by allowing the runner-up candidates in the district 48 hours after the poll to nominate which party list the local winner’s seat should be “billed to”. If they disagreed, whichever list is nominated by the losing candidate or combination of candidates with the most votes, is selected.


    • Tom, the Mana and Internet cooperation is just that: a pre-electoral alliance (which requires a joint list). National and Act are obviously not a formal alliance, but are a de-facto one (with separate lists), and this is certainly well known.

      The cases of a party tacitly or overtly assisting a smaller party to win a seat and thus pass the alternative threshold are not the same cases in which a party made a coalition with–or even bargained with–a party other than the one it had a pre-electoral arrangement with. So I don’t quite follow your diagnosis.

      (The only case I can think of in New Zealand of a party breaking a pre-election statement about what party it would work with was New Zealand First in 1996, and the election results had made a Labour-led government impossible unless it included either National or Act. That is actually the only such case, although NZF and United Future bargained with both parties following the very close 2005 election.)


    • Just realized I was wrong about the FDP in 1982, or at least stretching the meaning of “after the election”. But these three cases are often cited by anti-MMP (and anti-PR in general) writers as proof of broken promises. In NZ the Peters fishing trip seemed (from what I could tell at the time: Internet in the mid-Nineties was primitive) to be seized upon with particular schadenfreude by many of the SMP supporters who’d lost the 1992 and 1993 referenda: one big argument for ditching SMP had been that the 1984-90 Lange/ Douglas Labour govt had used its absolute majorities to push through deregulatory economic reforms that it hadn’t made clear to the voters before they voted. And now here, the critics argued, was MMP enabling a minor-party politician to commit the same sin.
      By the by, NZ has held 13 elections in the past four decades at each of which either all, or roughly half, the seats were filled by SMP. At how many of those elections did the party with a plurality of votes win an absolute majority of district seats? Answer: at nine – that is, just over 69%. In 1978 and 1981 the second-largest party won an absolute majority of districts, while in 1996 and 2005 no party won half the district seats. (All else being equal, if minor parties can win nine districts out of 65 or seven districts out of 69, they would presumably do at least as well if there were still 99 districts, or a fortiori if there were 120-odd). In other words, over nearly one-third of this author’s lifetime, SMP in NZ has not performed as advertised.


  10. Aha, sorry I was imprecise in my terminology. By “joint list” I didn’t mean several parties combining to present a single merged list of candidates. Maybe “alliance of lists” would be more precise – several parties, formally allied before the election, each presenting a separate list, with remainders being pooled among the parties. Poland, I believe (as one example), has allowed such alliances, with the threshold being 5% for a single party and 8% for an alliance.
    My point is/ was that given how much (say) Angela Merkel wants and needs the FDP – or John Key wants and needs ACT-NZ – to be in Parliament supporting their coalition, rather than being excluded by a 5% threshold, I continue to find it surprising that these parties don’t support such a widely-used-elsewhere mechanism whereby separate party lists can pool their votes.
    While pre-election vote pooling doesn’t of course prevent one party leaving the alliance and crossing the floor in the subsequent parliament, it does signal to voters which major party the minors are going to support.
    Allowing for the usual bias of the Australian media when reporting on (a) NZ and (b) PR elections, the reaction in 1996 to Winston Peters’ actions in 1996 were widely reported on this side of the ditch (newspapers, radio, water-cooler conversation) as a betrayal that had caused NZ voters to turn en masse against the MMP system they had voted for only 3-4 years earlier. I’m glad it was an isolated incident. I think small parties help to discredit PR when they pull such stunts.


    • Separate lists that can pool votes also exist in Israel. Each list must separately pass the threshold, but the threshold remains the same for individual lists and collections of lists. These arrangements are often consequential in that they get the combine an additional seat, which in such a fragmented system can matter.

      Note that both types of electoral cooperation can exist simultaneously, and do, in Israel. Some individual lists are blended between two or more parties (Likud-Beiteinu is one, Bennet’s Jewish Home is another), and these lists can also enter into pooling agreements with another list.

      The Internet-Mana list would be a blended list (a term I just made up, actually). And it is remarkable in having a clause that says their agreement expires after the election. That strikes me as far more opportunistic and manipulative than the piggybacking (which Internet-Mana, or at least the Internet component thereof, is also counting on). But in either case, it is transparent.

      In the Polish case that you mention, these are open lists, unlike the New Zealand and Israel examples. This means that parties within the list can’t prearrange the “blending” of what their caucuses will look like, for a given number of seats won by the list. Rather, they have to think about the distribution of votes across candidates of their respective parties. I am not sure that separate lists in Poland can enter into cartels (or whatever you want to call these pooling arrangements). Maybe they can, or at one time could, do so.


      • In fact I believe (sources: Lakeman and Lambert 1973, McLaren Carstairs 1984, and the Britannica) the reason the Scandinavian countries changed from D’Hondt to either St-Laguë (Sweden) or an even more small-party-generous formula (the Danish “0.33” method, compared to “0.5” for St-Laguë and “1.0” for D’Hondt) to avoid the need for joint “cartels” of lists, which had come to be viewed with suspicion.
        St-Laguë does do the job in ensuring that any given bloc of voters representing, say, 25% of the electorate will end up with somewhere around 24-26% of the seats, regardless of whether they concentrate their votes on a single party or divide them among four or five. Fragmenting might lose them a seat under some circumstances, but gain a seat under others, so it randomizes out. By contrast, under D’Hondt, a grouping of five parties (without joinder) will never win more, and almost certainly win fewer, seats than a unified list of equal aggregate size.
        However, St-Laguë isn’t much help solving the other problem of non-preferential vote-splitting under PR-List, which is where a party loses not one seat here or there due to rounding error but misses out on up to 31 Bundestag seats (or as many as six NZ House of Representatives seats) because its nationwide support falls just short of the threshold.
        There’s much to be said for requiring individual member parties in a combined-lists alliance to each get, say, half a quota to win a seat (as well as the alliance en bloc reaching the 3% or 4% or 5% threshold). Otherwise one might see a Czech or Polish Ricky Muir winning a seat http://tinyurl.com/nuvuhb4 on 0.5% of the vote, just by heading (or topping the poll for) the plurality list in a 20- or 30-list alliance that just cracks the threshold.


  11. How would a one vote MMP system would have look? Would there had been any rorts? I know I may have mention this before, it seems as if this would be the best MMP system, a one vote MMP system using the Australian preferential voting for electorates, and with first preferences as the list vote? This system would avoid the problems of decoy lists or maybe possible lessen them.


    • MMP really should have two votes. Ticket-splitting is not a bad thing. Exception: if a country has a problem with the so-called decoy lists. I mentioned Lesotho up-thread, or perhaps in the other one. But this is clearly not a problem in Germany or New Zealand. And I don’t see how one vote would eliminate the incentive to piggyback, where permitted. These are separate issues.


  12. Has anyone tried an MMP system using the SNTV, and a List component or would that lead to problems? Would it lead to patronage and clientism? I know that some countries might use SNTV, and List hybrid as a parallel system.


    • Rob, that is exactly what was proposed by the Dutch Balkenende II government about ten years ago, but that proposal got nowhere. It would be interesting to see what would happen, but I still wouldn’t recommend it, due to the negative effects you mention that feature in SNTV’s poor record.


  13. Pingback: New Zealand to have referendum questions on 2020 ballot, potentially including “tweaks” to MMP | 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 )

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.