Compensatory STV?

On a theme that we discuss here from time to time, the third party in Malta (Democratic Alternative) is proposing a compensatory element to the country’s single transferable vote system.

we are proposing a double threshold, with a district quota of 16.6% that would allow an individual to be elected on her/his own steam, and a national quota with a threshold of 2 quotas for a party to be represented in parliament.

(Thanks to Rob R. for the tip.)

15 thoughts on “Compensatory STV?

  1. This is probably where some of the lengthy discussion on Malta ought to have gone.

    The AD proposal of 3 years ago was interesting, but I would argue that Malta more or less has “compensatory STV” already in the form of its majority/plurality bonus provisions.

    This was also originally planted around the same time as Buhagiar and Lauri’s article about a modified STV was released (they essentially proposed that in each district, after 4 candidates have been elected, the remaining candidates go into a national proportional process to ensure the majority winner gets the majority of seats).

  2. Malta has elections this Saturday 9 March), as the government lost a budget vote to a defection (the peril of a single-seat majority). The most recent polling I’ve seen shows Labour ahead 49-47, with AD-The Greens on 4%. Their leader’s goal is to get above 2000 votes in one district. Quota is generally 3500-4000, but 2000 would let them survive long enough to pick up some transfers from Labour and PN, and may be enough to be declared elected without a quota. The leaders of both major parties have expressed their willingness to form a government with AD; the fact that a coalition is even on the radar is a huge change for Malta.

    On the subject of compensatory STV, the article below says that a constitutional amendment was passed in 2007 which made the bonus seats proportional rather than just a 1-seat majority, though in 2008 a one-seat majority was the closest to proportional. I wasn’t aware this had passed. I’m not sure whether it adjust the total if the largest party wins a majority of the 65 district seats.

    http://www.timesofmalta.com/articles/view/20080316/opinion/fourth-perverse-result-since-1981.200496

  3. After the 2010 election in Australia there was considerable play with the two-party preferred vote. Initially the Coalition led the 2PP and insisted that it gave them a right to govern. They reverted claiming that right on the basis on the count primary votes after Labor drew ahead on the 2PP count.

    I have been reading a proposal for reiterative STV called the Wright system, where after each exclusion the count is reset as id the excluded candidate had never stood. It strikes me that Malta needs to calculate the corrective bonus (in an area littered with confusing terms let us keep ‘compensatory’ for MMP) by resetting the count by the Wright rules.

    That is a much simpler formulation than the current Malta rules and avoids the distortion of using first preferences as if they were final preferences.

  4. I’ve at times thought perhaps Australian voters should be handed a third ballot-paper – as well as House and Senate, you’d have a non-binding plebiscite asking “You you want to see the Prime Minister replaced by the Opposition Leader?” or like effect.

    This would give a more accurate measure of 2PP than even an exhaustive pairwise comparison of candidates in each district. (The 2010 2PP differs according to whether one takes the “Coalition” candidate in O’Connor, WA as Wilson Tuckey or Ken Crook. Antony Green cites an anomalous case that – rarely under Australian IRO-AV – allows Condorcet-style comparisons: http://tinyurl.com/bxgjm6t. Note that AG, true to form, speed-reads my comments then misconstrues them. At least Alan pores over my comments in detail before he misconstrues them. (-;)

    (1) When one party or coalition wins a majority of seats, such a plebiscite would give an objective basis for deciding which party “should” have won. Note the problems in South Australia – where the electoral commissioners are required by law to arrange electoral boundaries in a way that ensures a majority of seats for a majority of votes – when third-party candidates and independents win seats, as they often seem to do under SA’s complicated system of single-member electorates. In some cases the commissioners have resorted to interviewing cross-bench MPs for their views on policy issues…

    (2) When no party or coalition wins a majority of seats (in which case the 2PP is going to be haywire anyway), such a plebiscite would help wavering balance-of-power MPs decide whom to support. Of course, if they lean ideologically towards one side (Katter or Crook towards the coalition, Bandt or Martyn Evans towards Labor), they can disregard the result. But other balance of power MPs (eg, Liz Cunningham in Qld, 1996) have cited the 2PP as one factor swaying their decision.

  5. During the 2010 government formation in Australia, Labor claimed successively that (1) the 2PP was an academic irrelevance (2) only the number of seats matters (3) the crossbench should be guided by the 2PP.

    The Coalition claimed successively that (1) the crossbench should be guided by the 2PP (2) only the number of seats matters (3( the 2PP is an academic irrelevance.

    These positions of high principle taken after the deepest refection and analysis may or may not have been related to the 2PP initially favouring the Coalition and then shifting to Labor.

    Adding in a third measure (whose function would be deeply unclear to the electorate) would just give the major parties something else to misrepresent while considerably raising the cost and obscurity of the election process.

    Anomalous cases like Crook could be resolved by the simple expedient of looking at who nominated them. Crook claimed to be a WA Nationals MP but was actually nominated by the federal branch of The Nationals. You get similar anomalies under public funding at most elections. I would be surprised, for example, if Crook had declined any funding by the The Nationals federal branch.

  6. I’m a big fan of the Wright system–it weights every ballot paper equally, rather than giving disproportionate weight to those papers which transfer later in the count, making it conceptually similar to the use of Weighted Inclusive Gregory. My only major criticism is that it lowers the quota after each exclusion (treating exhausted papers as if they were informal)–this means that some candidates who would never receive enough transfers to attain a quota end up elected because the quota gets smaller (think Pauline Hanson at the 2011 NSWLC election).

    It is more complicated than most other STV methods (possibly invoking the Tideman trade-off), but it’s not so complicated one needs a computer to calculate the algorithm; a calculator or else a sheet of paper and a good head for fractions would suffice.

    I’m just confused as to how it would apply to the Maltese bonus. Are you suggesting using final count totals with the Wright system for determining proportionality?

    I think the simplest change would be to based the bonus on a 2PP count if there are only two parliamentary parties. If there are more than two parliamentary parties, a party would need a majority of first preferences to win bonus seats. However, this wouldn’t establish proportionality in a 3-party parliament if no party won a majority of first preferences; if a minor party had won one seat at the last election, Labour would have had 33-34 seats and formed government despite the two opposition parties polling a majority of votes.

  7. Tom, your third ballot paper idea sounds awfully similar to the Israeli direct PM elections, though with the caveat that Australia’s Parliament is nowhere near as fractious as Israel’s.

    I think a non-binding PM plebiscite is a recipe for disaster. 2PP numbers have enough ambiguity to give crossbenchers some cover in a hung parliament (as well as the cover of the fact that first-preference pluralities and 2PP numbers are constitutionally irrelevant, as Gov. Underwood reminded us all). A plebiscite would lack this ambiguity; there would be enormous backlash, in Bush-Gore fashion, if the electorate supported one candidate and yet the other formed the government.

    If the plebiscite came with bonus seats for the winner (if necessary), that would be a very different story. Australia could even hold an AV election of the Prime Minister and give corrective seats to the winner, to allow more than two candidates to compete.

    I think Malta would also be better off simply moving to a second ballot to determine the national preferred-PM rather than trying to ascertain it from district-level votes. It also eliminates the issue in STV of how to consider a vote that has given preferences to multiple parties–I think the Maltese know that their first preference is what’s counted, but a voter who goes back and forth between parties obviously isn’t expressing a very strong preference for any of them.

  8. @Chris

    I agree about lowering the quota. Much better to continue the count under the initial quota until all vacancies are filled. I’m unclear why the author of the Wright rules included a quota recalculation at each exclusion, it does not seem to be necessary to his system.

    Naturally I also agree with your extraordinarily cogent critique of the third ballot paper idea.

  9. I apologise for my ignorance, but how does Wright actually differ from other STV counting-mechanisms?

  10. It essentially is PR-STV with Weighted Inclusive Gregory surplus transfers, but with two major caveats:

    1. If after all surpluses have been counted, there are still vacancies to full, the lowest places candidate is excluded. The count then is reset, with the excluded candidate treated as they’d never stoof. This means that the ‘first count’ total at each round of counting the highest remaining preference on each paper, not just first preferences; and

    2. The quota is recalculated after the first count of each round (that is, after each exclusion). Papers on which there are no remaining preferences (those on which every candidate had been excluded) are treated as if they were informal, and thus the quota lowers at every count.

    Counting continues until m candidates (where m=magnitude) have attained a quota in the same round of the count, or until all but m candidates have been excluded.

  11. Erggg…that should read ‘if after all surpluses have been transferred, there are still vacancies tI fill, the lowest places candidate is excluded. 

  12. The first part looks completely normal to me – isn’t that just like any STV? If counting the surpluses from quotas that were reached and exceeded isn’t enough to fill all seats, the lowest candidate is eliminated until another candidate reaches a quota.

  13. The distinction in what happens when a candidate is excluded. The election is reset and any preferences for the excluded candidate at any stage of the count are disregarded. In most flavours of STV, an excluded candidate’s preferences are distributed among the continuing candidates. The advantage is that Wright treats all ballots equally, where other flavours give some slight weighting to different ballots.

    The Gregory inclusive transfer value still applies to surpluses from elected candidates. The Wikipedia pageis very comprehensive.

    Like Chris I disagree with the idea of recalculating quotas when a candidate is excluded.

  14. For instance, take two groups of voters.

    One group voted 1 A, 2 B, and 3 C.
    The second group voted 1 Y, 2 B, and 3 Z.

    In traditional STV systems, if A is the first candidate excluded, all of group 1’s preferences transfer to B at full value.

    Say B is then elected. In fractional systems, the votes transfer to C at a small value. In sampling systems, two things could happen: if A’s votes were the parcel that brought B over a quota, some of those papers, picked at random, transfer to C at full value, and the rest do not transfer. If A’s preferences don’t bring B over a quota, and some other votes do, none of A’s preferences transfer at all.

    If Y is excluded after B is elected, all those papers transfer to Z at full value. This is despite the fact that, like the first group, they supported an excluded candidate first and had the same second preference.

    By restarting the count after each exclusion, both A’s and Y’s preferences would be treated as first preferences for B at all further counts.

    Basically, it does for excluded papers what Weighted Inclusive Gregory did for surpluses. Previous surplus systems either only transferred papers from the ‘last parcel’ received, giving later exclusions and surplus transfers more weight, meaning these papers could transfer at full value everal times, or gave every ballot [i] paper[/i] the same value regardless of the votes it represented, weighting transferred papers more than first preferences (for instance, for a Greens candidate with 100 first prefs and 300 Labor transfers at a 0.33 transfer value, representing 100 votes, the Labor papets make up 75% of the transfers, as they are 300/400 papers, rather than 50% as 100/200 [i]votes[/i].

    Basically, Wright takes STV from ‘one vote, similar value’ to as close as possible to ‘one vote, one value’ as is possible without computer calculation.

Leave a Reply

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

WordPress.com Logo

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

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s