Labour+Greens or +NZF?

There is an interesting item about a recent poll in the New Zealand Herald. The upshot of it, as I read it, is that the New Zealand Labour party may be so concerned about being tied too closely to the Greens that they’d at least like to signal a preference for forming a coalition with Winston Peters and his New Zealand First Party. Now that’s desperate!

The story notes that Labour+NZF would be unlikely to be a majority, but they may hope they can just say to the Greens that they can either agree at least to abstain and allow a minority government to function, or be blamed for a fresh election.

Then again, maybe Labour and Greens will yet end up forging a joint program before the election, and this is all posturing. As also noted in the news item, the Greens have made no secret of wanting the Finance ministry, and Labour certainly has an interest in signaling that this would be non-negotiable.

12 thoughts on “Labour+Greens or +NZF?

  1. > “like to signal a preference for forming a coalition with Winston Peters and his New Zealand First Party”

    MMP, unlike closed lists simpliciter, gives some voters some means of signalling to the political elites what sort of coalition partners they’d be happy with. As Lakeman noted, in Germany many FDP supporters – realising that an FDP vote was “wasted” at district level – would give their Firstvote to the CDU/ CSU candidate, if they leaned to the centre-Right, or to the SPD candidate, if they leaned to the centre-Left. Thus, the proportion of ballots with a Secondvote for the FDP list but a Firstvote for one of the Big Two gave German pundits information roughly similar to the two-party preferred vote figure in Australia.

    (I think I recall reading elsewhere that in a couple of elections either SPD or CDU/ CSU actually won more than 50% of the district-level votes, although of course less than 50% of the list votes and hence less than 50% of the Bundestag seats in every election bar one).

    Unfortunately this isn’t much help to major-party supporters wanting to signal which minor party they would prefer a coalition with. It would be hard to persuade National or Labour supporters to tick a minor-party candidate at district level (except in highly unusual circumstances such as the Hide the Key manouvres in Epsom) since a vote for NZ First really would be wasted. Also, since they are several minor parties the votes may be split, and may be only a small (and thus statistically meaningless) percentage of the National or Labour vote anyway. Finally, since the district MPs are supposed to provide personalised local representation, it seems contrary to the purpose of MMP to use them as pawns in the game of coalitions among parties.

    This is yet another reason why I strongly opposed butchering STV/ AV by introducing ticket-voting, nearly as much as FPTP. The flow of preferences among competing political parties is a vitally important data set in a democracy. If a Tory candidate’s 45% plurality really represents somewhere between 30% and 40% who actually support the Conservative Party, and somewhere between 5% and 15% who would prefer UKIP or the LibDems but want to keep Labour out, this information is merely guesswork. And if the official Senate returns show that 93% of ALP voters preferred Family First or the DLP over a Green, this really tells us little more than that half a dozen bright former student politicians with bad haircuts on the ALP Executive decided in one or another Chinese restaurant that this would be a brilliant piece of tactical voting that would help elect three ALP Senators instead of two. It provides zero useful feedback as to how millions of living, breathing ALP voters actually think.


  2. Tom, isn’t there a bit of a trade-off here? If we reject ticket voting, medium-to-large district magnitudes yield huge “tablecloth ballots”. In turn, this gives us either many exhausted ballots, or donkey voting, neither of which is great.

    We’ve talked before of various compromise forms of ticket voting. Some of the ones I recall:

    – “Party ticket voting”, where the voter can make inter-party choices without being forced into intra-party choices. IE: “I prefer ALP, then Green, then National, then Liberal–but within each party, I accept their list order.”

    – “Multiple ticket voting”, where each voter casts multiple STV ballots.

    – I don’t know if anyone came up with a catchy name, but it was mentioned that each party could propose several different tickets. So you could vote “Green: leans left” or “Green: leans right” or “Green: leans populist”.

    So do you consider all these forms of ticket voting to be “butchering STV”? Or is it just the Australian Senate’s method that you object to?


  3. Hi Vasi, I’m not sure I agree with you about cause and effect here. Large Australian Senate ballot-papers existed before ticket-voting but were smaller. Eg, the 1974 double dissolution, there 73 candidates for 10 seats in NSW (supposedly a businessman with Liberal Party ties had signed up a number of them in the hope of crowding the ballot field and hurting Labor by driving up the informal vote) That was a record high at the time, but it’s absolutely standard these days – eg NSW had (by my manual count) 84 in 2010 and 79 in 2007.

    Note also:
    (a) how many of these groups have exactly two candidates – no more, no fewer; and
    (b) unlike 1974, when 73 candidates meant a paper with 73 squares, the 2007 ballot also had 25 squares above the line for the 25 groups, making a hundred squares in total.

    I discussed the reason in more detail here: Basically, ticket-voting encourages more cand diates to nominate, for two reasons:

    A reform that was originally supposed to help manage the problem of large numbers of candidates has actually exacerbated that problem. Your party only gets a square above the line, and its own column below the line, if it stands two or more candidates (although sitting Senators can have one if they run as Independents). It is much easier for a small party or even an independent to get ballot visibility by adding a second or third candidate. Not only does a lone candidate (party-endorsed or not) get shoved in he “ungrouped” column – which by law always goes at the end, on the far right of the ballot-paper – but s/he can only be voted for by citizens who are willing to number dozens of candidates carefully; s/he also loses the ability to direct his/her preferences via a voting ticket. So, naturally, small parties and independents who before 1984 might have stood one candidate (incurring a small disadvantage in electoral visibility to save a deposit) now find that a deposit is worth paying, and even forfeiting, to avoid a massive loss of electoral visibility that is incurred when you run only one candidate. (Even Independents sometimes put up a running mate for this reason.)

    Candidates of very small parties (NDP in NSW 1987, FF – that’s Family First, I rush to clarify) do win Senate seats with a primary vote of only 1-2% when they have their own column and ticket. But no ungrouped candidate has ever come close to winning a Senate seat since 1984.

    [comment posted at “Electoral reform to increase female representation,” The Irish Politics Forum, May 2011, ).


  4. Tom, I agree that ticket voting caused ballots to expand. But I think it’s clear that a large-ballot-with-ticket-voting is a very different animal than a large ballot without it. With ticket voting, you don’t have the exhausted preferences problem, you don’t have donkey voting, and you don’t make voters stand around numbering fifty boxes.

    For what it’s worth, I don’t think the Aus. Senate has a high enough DM that ticket voting is a must, as long as other methods are used to keep voting easy. Higher nomination fees and signature requirements could keep the number of candidates low, and optional preferences would make a big difference. I favour higher DM than six, though, and eventually some sort of tickets become truly necessary.


  5. I ran two student elections where ticket voting was allowed but:

    1. Tickets and candidates appeared on the same area of the ballot-paper in random order.

    2. You could vote any combination of candidates and tickets you pleased.

    3. If you voted for the same candidate more than once only the first time counted.

    You could, for example, vote:

    1. Maverick

    2. Shugart

    3. Greens

    4. Labor

    5. Round

    The vote counts for Marge while she is a continuing candidate, then to Shugart (if Shugart is included on the Greens ticket at say Number 2 he gets the vote before the Number 1 Greens candidates), then to Labor and so on.


  6. I may be the only person who actually likes Above the Line Voting. If I had to redesign the system or write for another country, I would make the following changes to keep the benefits, at least as I see them, of being able to vote for parties or people.

    1. Instead of a tablecloth ballot for everyone, I would actually have two “ballots.” A voter could ask for the “Long Ballot,” which lists every candidate in random order without reference to party status. A voter could also ask, instead, for the “Short Ballot.” The Short Ballot would essentially be the Above the Line Vote, but on its own concise page. Every party would have one box on the Short Ballot as would every independent. If several independents run as a joint ticket, they would effectively become a party for the purposes of the ballot.

    2. On either ballot, voters rank one or more candidates. A Long Ballot vote is counted as it stands. A Short Ballot vote is counted so that a choice for a party is a choice for that party’s published list of candidates and those candidates only. So a vote of “Labor, John Smith, Crazy Fringe Party, Greens…” is read as a vote for all Labor candidates, John Smith, all Crazy Fringe Party candidates, etc.


  7. I’ve been trying to come up with a system that could be called “STV for stupid people” and it would probably involve something like above the line voting. Essentially try to keep most of the effects of STV but avoid asking the voters to rank candidates, which seems to be what some people have trouble with.

    One thing I’ve come up is a smallish (5 to 7 members) multimember district, where each voter gets as many votes as there are seats in the district. Voters are not allowed only one vote per candidate, but can still bullet vote (using one of the votes and waste the others), scatter the votes among candidates from different parties, use them all on candidates from the same party, or vote for a party list in which case the votes are scattered equally amongst all the candidates in the list (the list can’t be longer than there are seats to fill).

    The STV similar portion would be in the count, which would work like an STV count except with the assumption that every voter each stuck to candidates from one party. So if a candidate more votes than the quota, the excess gets redistributed to the candidate with the next most votes on the same list (or the next ranked candidate from the party in the case of ties); when a candidate is eliminated the votes get redistributed to the candidate with the most votes on the same list and so on.

    If the count gets to the point where the only remaining viable candidates are all from different parties/ list, and there are still seats to fill, then the greatest remainder method is used to select which of the remaining group would be elected.

    This could also be termed “extreme open list”. It would sacrifice the ability by voters to rank candidates, but keep the rough proportionality of STV and the ability to vote for individual candidates.


  8. In STV districts as small as Ed proposes I really cannot see a usability problem.

    The informal votes in the ACT (M=5-7) and Tasmania (M=5) are not very high. The most extreme case of STV without a whole lot of awareness in the electorate would be the Eve Online Council of Stellar Management (M=15) where STV was dropped on a population of hundreds of thousands of seriously ill-informed game geeks who voted preferentially without much complaint about the system and without any formally organised parties.

    There is a slight tendency to reinvent the wheel when what we have already rotates quite well.


  9. Like Alan, based on Tasmanian and ACT evidence, I would say that optional or semi-optional (no fewer preferences than seats) preferential voting is sufficiently simple in 5- to 7-seat districts (which is what most Australian STV advocates support… although the dominant faction in the NSW branch of the PRSA has been flirting with the idea of Statewide at large elections, or 10-to 15-seat districts). Ticket-voting is a non-solution to a non-problem in those cases – or at least not a solution to the problem that ticket-voting is publicly claimed by the politicians to be fixing.

    For districts with more than seven seats, I’d advocate what I’ll call group-voting as an alternative to ticket-voting. Basically:

    (1) Voters can number groups as well as, or instead of, numbering individual candidates.

    (2) The group-votes for a party’s team of candidates are not all cumulated on the highest-ranked remaining candidate but, instead, are deemed allocated in the way that will equalise (as far as possible) the totals for that team’s surviving candidates.

    Thus, if the Purple Party has three candidates remaining in the count, with 5,000, 2,000 and 1,000 personal votes respectively, and there are 4,000 group-votes for Purple Party team as a whole, then these candidates are credited with 0, 1,500 and 2,500 votes respectively. This leaves their respective totals as 5,000, 3,500 and 3,500 votes, which increases their hope of winning more seats by having more candidates left in the count. (If the highest candidate had not exceeded the average then a completely equal division – 4,000 votes each – would be attainable).

    Once all of a team’s candidates have been elected or excluded, its group-votes go to the next preferences (if any) expressed by those voters, with transfer value reduced where appropriate.

    The purpose of this is to make it as easy for the candidate-indifferent voter to cast a ballot in an STV election for, say, 10 seats as in an AV election for one single seat. He or she can simply number the parties irrespective of whether there is one candidate in each team, or ten.

    I have no problem with an electoral system accommodating the voter whose preference order is “Every Purple Party candidate over every non-Purple candidate, but other than that I don’t know or care about differentiating among individual Purple candidates.” What I do object to is when ticket-voting interprets a ballot cast on this basis as if the voter had solemnly decided to rank the highest candidate on the Purple ticket above the other candidates, and to express 34th and 35th preferences among non-Purple candidates whom the voter has never so much as thought about.

    Some Australian commentators (Antony Green is probably the most prominent) dislike optional preferences in STV elections because it means the later candidates elected often get in on a sub-quota, whereas the earlier candidates have to make the full quota. I can see why this might look undesirable from a distance, but what’s the alternative?

    STV with compulsory full preferences means candidates may often be elected with only 2-3% of the first preference votes. True, they did have to reach 14.28% or 16.67% on the final count, but it’s the 2-3% figure that sticks in voters’ minds (and throats) when they think of Senators Wood, Chamarette, Fielding or Madigan.

    By contrast, an STV race with optional preferences might see candidates take the fifth or sixth seats with only 10-11% on the final count. (Vide NSW upper house elections where semi-optional preferences mean the early seats require a quota of 4.5454% while the last few seats out of 21 may be won on 2.5 to 3%). But all else being equal, they will need more like 7-8% of first preferences to have a chance.

    It’s analogous to the position of optional preferential AV being midway between FPTP (winning candidate may be liked by 44% but detested by as many as 56% of the voters) and compulsory full preferential AV (winning candidate may be liked by only 38% but may be tolerable to, say, 53%). OPV AV means the winning candidate will be liked by, say, 41% and tolerable to, say, 48%. It’s a fair trade-off between “Which candidate do the most voters like?”and “Which candidate do the fewest voters detest?”


  10. Just a note (on the most recent NZ planting) that the initial results of the 2013 NZ Census are out. This impacts the number (as well as the boundaries) of electorates (districts). As the number of MPs and South Island electorates are fixed, the northward population trend means an additional North Island seat.
    The census was two years later than scheduled, as the forms were just going out in 2011 when the most severe of the Christchurch earthquakes struck.


  11. Thanks, Errol. So there will again be a reduction in the size of the compensatory tier, I take it.

    The electoral commission noted that this is a long-term problem. Some future government will need to address it, given the current one decided to do nothing on any of the “improve MMP” recommendations.


  12. My preferred longer-term solution is the dropping of the Maori Seats. If you leave the South Island at 16 (rather than adding the Maori Seat which is mainly in the South Island to the SI fixed number), then you get 7 or so seats more in the compensatory tier.
    I looks like we might not get more Maori Seats for a while, so that particular source of loss of compensatory tier is on hold.


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.