Elections New Zealand has released the Proposals Paper that it has prepared in response to the submissions to the review of the country’s Mixed-Member Proportional electoral system.
Rather than summarize it here, I will encourage interested readers to go read it themselves, and ask them to come back here and discuss.
For earlier discussion of the referendum and review, please click the words, MMP Review, in the “planted in” line above. A couple of the prior entries include portions of a submission I sent.
My submission is cited (under the last subheading on the page on dual candidacy); I am honored!
Propagation: Seeds & scions (57)
RSS feed for comments on this post.
TrackBacks
To graft a scion to this planting, please use the following URL:
http://fruitsandvotes.com/blog/wp-trackback.php?p=6403
(Non-MT bloggers click here to send pings.)
Grafted scions that are not compatible with this planting's stock will die or be pruned out by the Orchardist.
About the comment form
Please note that the name you enter below and the first several words of your comment will appear on the right sidebar of the blog's front page, under "Propagation." New propagators might want to look at the comment policy.
Please do not enter long URLs into the seedbed. Either mark them up using html hyperlinks or convert them to a "tiny URL." Thank you!
The soil is ready for planting:
`
Just for ease of cross-referencing, follow this link for the previous discussion of dual candidacy.
Seed planted by MSS — 13 August 2012 @ 01:41
I found Jack Vowles analysis (PDF with a lamentable number of typos) really interesting, particularly his observation that turnover is significantly higher amongst list MPs, at least in New Zealand. It would be interesting to repeat this analysis in other MMP jurisdictions. (A summary is available.)
The proposal paper seems reasonable to me, based on the generally applicable theory that “if it ain’t broke, don’t fix it.” However, the graphs could easily have been extracted from any of a number of anti-Powerpoint rants. (The linked graph is “Simple analysis of submissions on whether the one electorate seat threshold should be kept”.) I have no idea why they would choose a graphical presentation style that encourages readers to think that “percentage of submissions saying X” is even remotely correlated with social preference. It’s clear that they did not view their work as simply choosing the most popular option, although in fact I think they mostly did so. (That was all inspired by a reply to Wilf’s last sentence, which I decided in the end not to make.)
Seed planted by rici — 13 August 2012 @ 03:09
I strongly commend United Future supporting the best loser approach. This approach lends itself well to MMP, and would be very easy to implement compared to a any open list system in terms of ballot format.
I noiced that they implied that the Baden Wurtenburg system “always leads to overhangs”. I don’t believe that is true. What was the context of the comment?
Overall, it looks like they invalidated the best loser system in the context of open lists, though the two are very different.
Seed planted by Nick Anand — 13 August 2012 @ 12:45
Another question, how would the abolition of the 1% threshold lead to more overhangs (as suggested throughout the review). Wouldn’t it increase the national allocation of seats to other parties, thereby preventing overhangs?
Lower the threshold to 4% would likely increase overhangs as suggested.
I think the bigger issue with eliminating the one-seat threshold is that it more or less targets the ACT. So though I support it in principle, as I think there should only be one threshold where possible, it should not be done when it can so easily be tied to one party. That will cause many to believe there is a sense of targeting.
Seed planted by Nick Anand — 13 August 2012 @ 12:50
Nick, the reason the one-seat threshold might create more overhangs (unless overhangs are prohibited, as the proposal suggests) is that the one seat (or more, in the case of the Maori party) is awarded anyway. If the party fails to qualify for list allocation, then its list allocation does not offset the direct seats, and so they are all overhangs (five in the case of the Maori party in 2008).
I don’t see why this change would be seen as specifically counter-ACT. It would have made no difference to ACT in the 2011 election. As the table in the proposal’s section on thresholds shows, there have been five occasions in which the one-seat threshold led to an allocation of additional list seats; of these, ACT was the beneficiary only twice.
With respect to “best loser”, they seem to agree with the Royal Commission that the “best loser” metric’s correspondence with voter preference is “illusory”, but what they say in the end is: “This would constitute a fundamental change to our system of MMP. In particular, parties would not have the mechanism of a party list through which they
could promote the representation of women, M?ori and minorities thereby having a major impact upon the diversity of Parliament. It would also confuse the purpose of the two votes, one of which is to elect a local representative, the other to represent a party of choice.”
Seed planted by rici — 13 August 2012 @ 15:04
Removing the provision of the threshold by which winning one nominal-tier (electorate) seat qualifies the party for list-tier compensation seats would be targeting Act more than any other current party. While Progressive and United Future have benefited from it in the past, both these parties were always mainly vehicles for one safe-seat MP (Jim Anderton, now retired and taking his “party” with him, and Peter Dunne, respectively). These are not parties with staying power, but Act is; different MPs have carried it to the one-seat threshold at different elections–indeed, in different electorates.
While it is true that the threshold provision did not matter to Act or the overall result in 2011, the abolition of that threshold provision would remove one of the main incentives for voting for Act–both in its one winnable electorate (recently Epsom) and nationwide.
For this reason, I doubt such a change will pass, because National may want to keep Act around. But others who are closer to current NZ political developments may have other assessments of this situation.
Seed planted by MSS — 13 August 2012 @ 15:25
As for “best loser” provisions for the list tier, I can’t say more than what the Proposals Paper says (in particular, the passage quoted by rici), or what I said in the prior entry on dual candidacy (see link in comment #1).
(I know it may seem odd to call it a “list tier” when technically there is no list under a best-loser provision. But all the other mechanics of the mixed-member system are the same, so I think of it as a list tier where the list is all dual-candidacy, and ordered by some provision concerning relative performance in the nominal-tier districts.)
Seed planted by MSS — 13 August 2012 @ 15:28
Also, with respect to Baden-Württemberg, I tried the review’s “search” functionality, which led me to the submission of Philip Temple. It’s a word document, so I’m not pasting the link here, but it’s easy enough to find.
It’s true that Baden-Württemberg’s system does not always result in overhangs. There was no overhang in 1972, for example. But there have been overhangs in every election since then. Note that the B-W system also has balance seats, so it’s not correct to say that there are currently 18 overhang seats. There are, I believe, nine overhangs seats and another nine balanced seats.
See the complete state election results (PDF in German; Baden-Württemberg is the first page.)
Seed planted by rici — 13 August 2012 @ 15:34
MMS, does this look like a party with staying power, or a party dying a slow death? Compare with New Zealand First which benefited from the one-seat threshold once, but would never have needed it had the threshold been four percent instead of five.
Also, looking at Epsom, it doesn’t seem like Epsom voters view themselves as having much incentive to vote for ACT as a party.
(But I admit that I know nothing about NZ politics on the ground; I’m just looking at numbers.)
Seed planted by rici — 13 August 2012 @ 15:56
MSS, sorry about the typo. By the way, how common do you suppose it is that the one of the two significant candidates in an election is the biographer of the other?
Seed planted by rici — 13 August 2012 @ 16:22
I would say that Act could be dying a slow death over the last couple of years. My point is that, unlike UF and Progressive, it has not been led by the same politician since its founding (which is one indicator of staying power, or more political-scientifically, institutionalization). Nor, unlike the other two, has it relied on the same district (electorate) each time it has entered parliament without clearing the 5% threshold.
Further, the indicator of a small party’s strength in an electorate that is the only one it wins is not to be found in the party vote. In all the cases I am referring to, voters have split tickets at a high rate, voting for the small-party candidate in the electorate, and the preferred large-party coalition partner with the party-list vote. This is entirely rational strategic behavior by these voters. There truly is no incentive for such voters to give their list vote to the party.
For voters outside the electorate that the party can win, the only rational vote for that party is a party-list vote, but then only if the party is expected to win its one electorate elsewhere. Act did not have that advantage in 2011, as it was perceived to be a close race in Epsom.
Seed planted by MSS — 13 August 2012 @ 19:06
Okay, I see that the overhang for the Maoris and United Future would in fact have been 5 if the one seat threshold was abolished.
Kinda makes then to keep it as is.
However, it might then make sesnse more for parties not crossing the threshold to have their SMD seats removed from the count. So in the last election, the above threshold parties would then divide up 114 seats.
New Zealand already does this for independents. apparently. In theory, the Maoris could have increased their “relative” presence in parliament by calling themselves independents last time at the expense of another party.
Though, this sacrifices proportionality a tiny bit, the impact would be minimimal, and would resolve this overhang problem.
Seed planted by Nick Anand — 13 August 2012 @ 19:08
The point about Baden-Württemberg in the Proposal Paper struck me as odd, too.
It seemed to be claiming that the type of list used there (best loser) was the cause of the parliament being expanded so much due to overhangs. But surely the greatest cause is that B-W uses “balance” seats. (Let me cross-reference the related thread on German overhangs here.)
Seed planted by MSS — 13 August 2012 @ 19:08
In my view, and I am free to be questioned on this, my thoughts are that the BW system would likely lead to some type of underhang (moreso than an overhang). This might occur due to small parties receiving significant second votes; however they may not have fielded significant amount of candidates to avoid splitting the vote. As a result, they might not have enough SMD candidates to fill their list position.
Seed planted by Nick Anand — 13 August 2012 @ 19:33
Nick, the actual B-W system doesn’t have second votes, which is what makes it different. That’s the logic of “best loser”, isn’t it?
But I would expect a fused vote to result in fewer overhangs, not more, since I’d expect CDU’s vote to be higher on the nominal tier than a hypothetical list tier.
Seed planted by rici — 13 August 2012 @ 20:05
Logically, “best loser” does not require a single vote, because it is a provision regarding the intra-party dimension (who gets the seats a party has won, other than its nominal-tier district wins).
Japan–which is MMM (parallel)–has a variant of best-loser provision, and also has separate nominal and list votes.
(Despite the established nomenclature in Germany, “first” and “second” votes are unclear conceptually.)
Seed planted by MSS — 13 August 2012 @ 20:12
If the Maori Party MPs were Independents, they would have got less Parlimentary funding I think, and certainly no extra salary for the Party Leader.
Seed planted by Errolwi — 13 August 2012 @ 21:24
Party funding! Now that’s a perverse incentive if I’ve ever heard one.
Seed planted by Nick Anand — 13 August 2012 @ 21:43
Parlimentary funding is for support services etc. The only state-funded election funding is for TV advertising – not sure if this is available to Independents. No-one can do TV advertising with their own money (!!!)
Seed planted by Errolwi — 13 August 2012 @ 22:00
MSS @6: It may or may not be true that removing the 1-electorate threshold targets ACT more than any other party. But it undeniably also targets the practice of what you call “vehicles for one safe-seat MP”.It thus perhaps also prepares the stage for the return of independent politicians, who in the status quo have a strong incentive to create and run as part of a small ‘party’ in order to gain more influence. Under the commission’s proposals, this incentive is removed as such politicians can’t win list seats on their coattails just because they’ve won an electorate seat.
Nick @12: If I’ve understood correctly, the removal of sub-threshold SMDs from the count is precisely what the commission proposes. When they say overhang should be abolished for sub-threshold parties, they are talking about overhang as ‘seats over the regular 120 MPs’ rather than ‘seats beyond what the party deserves based on party-votes’.
Seed planted by JD — 14 August 2012 @ 04:21
JD,
From my understanding. All they were saying is parties which don’t cross the 5% threshold would not get the 1-seat exception to qualify for list seats. I did not see (possibly by my error) the logical extension that they would be treated as “independents” vis-a-vis the overhang rules.
However, I guess that’s a logical extension.
One thing that struck me about my own thought @ point 12. If the rule was that only parties crossing the proposed 4% threshold having overhangs augment the house, this could create a negative vote weight for a party with significant overhangs near the threshold.
Let’s say a party is @ 3.99%, but has elected 6 SMDs. In theory, their entitlement is 5 MPs. An additional 1 basis point of votes would likely push them above resulting in no additional seats for that party. However, it could trigger an additional seat for other parties (through the augmentation of the house for overhangs). This would be an instance of a negative vote weight (though the party itself would not lose seats).
Seed planted by Nick Anand — 14 August 2012 @ 10:53
This is completely off topic, but I think its relevant to an academic electoral reform blog and probably deserves its own planting. Unfortunately, the only threads on U.S. state politics I could find on this site were about California and so this doesn’t quite fit anywhere. But check it out:
Yes, the article is referring to the general election, not to primary challenges as I had thought. It also doesn’t mention if there are any fringe party candidates on the ballot in any of the races the Democrats or the Republicans are not contesting.
[not only completely off topic, but too long a URL for the template. Please, let's keep this thread to NZ and the MMP Review. There actually is a thread on an aspect of New York party politics.--MSS]
Seed planted by Ed — 14 August 2012 @ 10:55
Nick, quoting from the proposals paper:
“The provision for overhang seats should be abolished for parties that do not cross the party vote threshold.”
“The simplest method of abolishing overhang seats generated by parties that do not cross the party vote threshold and,
therefore, the Commission’s preference, would be to reduce the number of quotients to be allocated by the number of electorate seats won by the party.”
Seed planted by JD — 14 August 2012 @ 11:10
Thanks JD.
Seed planted by Nick Anand — 14 August 2012 @ 11:29
Regarding the observation of Jack Vowles, mentioned by rici in comment #2, that turnover is significantly higher amongst list MPs, at least in New Zealand, I can confirm that this is also so in Germany.
Or at least it was in the two elections for which I have consecutive-term data. I can look at two pairs of elections, 1983-1987 and 2005-2009.
There are 1,104 incumbent MPs at the time of the 1987 and 2009 elections (and for whom the needed data are entered), of whom 777 (70.4%) continue as MPs in the next term. 47.2% of them were elected in the nominal tier. Of these, 393 (75.3%) continue in the next term. 52.7% were elected from the list. Of these 384 (66.0%) continue next term. This difference is statistically significant (p=.0004), so list MPs are less likely to return.
873 MPs, or 79.5%, ran for the next term in 1987 or 2009. (The rest would be “retirements”, whether voluntary or the party denying renomination.) Of these, 96 (11.0%) were defeated. The defeat rate for those in the nominal tier is 34/427 (7.96%). The defeat rate for those in the list tier is 62/446 (13.9%) The difference is statistically significant (p=.0025).
I did not check to see whether defeats by list-tier MPs were due to swings against the party (meaning it elected fewer off its list) or incumbents being moved down the list. Nor did I check to see if there is variation by party. But the pattern is the same in these two German elections as Vowles found for New Zealand: more turnover in the list than in the nominal tier.
I wonder if this is true in all mixed-member systems. I think I just found myself a paper to write based on the data my project members have collected (which includes several other MM systems)!
Seed planted by MSS — 14 August 2012 @ 16:18
I’m reluctant to provide any potential ammunition for the white-shoe brigade who funded NZ’s anti-MMP campaign, but I should point out that a high rate of list MPs being dropped doesn’t necessarily answer their main point.
The real core of the complaint is not that list MPs serve for a long time but that they serve (whether for a long or a short time) at the pleasure of the party HQ. If they were dropped (from the list completely, or to a position below the electoral high tide mark), then – to a List hater* – this may well have been because the MP was trying to act independently (standing up for his or her local area or region, uncovering some matter of corruption or maladministration, whatever) and was punished for it by the faceless men.
Note how SMD supporters like to point to examples of district MPs in various systems who bucked the party machine, got expelled or disendorsed, yet still managed to get re-elected by their local constituents. In this case, the defeat of a sitting MP would be seen as a bad thing, if it’s brought about by the diktat of party office (by taking away the party’s brand label and nudging the voters to reject the disendorsed ex-nominee). I take it as read that SMD-only advocates would see the defeat of a sitting MP as an even worse thing if it was done via list-re-ordering, ie, not just by nudging the voters to switch to the new official candidate but by leaving them no chance (at the Zweitestimme level, anyway) to re-elect the now-expelled MP.
* I don’t hate party lists, I just think preferences are better. Pooling of like-minded candidates’ votes is better than compulsory vote-splitting, but voluntary pooling is even better than compulsory pooling. Not all STV shills feels this way, though.
Seed planted by Tom Round — 14 August 2012 @ 22:25
The most spectacular example of the list problem is South Africa, where the ANC has developed an unlovely practice known as redeployment. As in, Thabo Mbeki was redeployed to other duties.
Redeployment of MPs between the National Assembly and provincial legislatures or even municipal councils is not uncommon. Admittedly South Africa has a very tight floor-crossing rule, but it is rarely invoked because generally redeployed representatives just wear it.
Seed planted by Alan — 14 August 2012 @ 22:45
Well, yes, what Alan said. South Africa empowers parties to expel parliamentary representatives in mid-term and thereby unseat them. (I have some idea Indonesia has a similar provision, or at least had it during the Suharto era). This does not assuage the fears of the party-list PR opponents. If anything, it exacerbates it.
Interestingly, some advocates of MMP for the UK (eg Vernon Bogdanor in his 1981 and 1984 books, Dawn Oliver) stress the German Constitution’s guarantee of the personal independence of Bundestag MPs as a valuable safeguard and a partial corrective to this danger. In other words, they calculate that British voters will be more likely to warm to MMP if guaranteed that MPs will be harder to remove.
Seed planted by Tom Round — 15 August 2012 @ 00:35
Tom: I should point out that a high rate of list MPs being dropped doesn’t necessarily answer their main point.
I disagree. I worry much more about the case Vowles addresses, where list MPs are disliked by voters, but can’t be thrown out as long as they retain support by the party muckety mucks. If one of these MPs is near the top of the party’s list, it would take a massive amount of voter coordination to turf them.
In the other case, of list MPs who are popular but shunned by their party, the newly-independent MPs have options, so I’m less worried. There may be other parties eager to latch onto their popularity, or other MPs who will join them on a new list, or perhaps they could win a constituency on their own.
This is why I’m puzzled that most open-list systems make it so much easier to push a given candidate higher up the list than to lower their position.
Seed planted by Vasi — 15 August 2012 @ 00:58
> “list MPs who are popular but shunned by their party, the newly-independent MPs have options”
Hmm, good point, although those options may be so limited – if the MP represents a minority who don’t come close to a plurality in any local district, and if there’s a high threshold for new party lists to win seats – that bowing to the whips may stand out as the easy option by far.
Having said that, loss of party machine favour is probably a career-killer under most electoral systems, with partial exceptions for Ireland and the USA. In Australia, expelled Independents tend to get re-elected once or perhaps twice (Katter stands out as an exception) and then eventually defeated by their former party (rather than retiring voluntarily).
Seed planted by Tom Round — 15 August 2012 @ 01:45
Yes, the next step would be to sort out how it is that the turnover in the lists occurs. But Vowles does show that the so-called zombie (a term I hate), who loses as an incumbent nominal-tier MP, but remains in parliament via the list, is very unlikely to be renominated in the next term. That suggests parties pay attention to voters’ verdicts on individuals even when they do not have to.
As for other systems, South Africa has an unusually draconian party-leadership dominance, in both law and party practice.
Indonesia is now open list.
I suspect that open-list systems* make it very hard for voters to coordinate and reject an incumbent. The exception would be when M is low; otherwise, as long as the incumbent retains a core following, he is likely to be reelected unless he is very marginal to start with. And presumably the very marginal ones are not the ones that are a problem, from the standpoint of holding individual MPs accountable.
If the pattern Vowles identifies is common under mixed-member systems (or at least MMP), then maybe this is a better system than most for individual accountability after all. This is an open question. And I hope to be among those answering it in the coming years!
____
* I mean actual open lists, not “flexible” lists where there is a party-given list position as well as preference voting.
Seed planted by MSS — 15 August 2012 @ 14:24
MSS, with respect to ACT and the one-seat threshold, you said (at #6):
“the abolition of that threshold provision would remove one of the main incentives for voting for Act–both in its one winnable electorate (recently Epsom) and nationwide.”
And at #11:
“the indicator of a small party’s strength in an electorate that is the only one it wins is not to be found in the party vote… There truly is no incentive for such voters to give their list vote to the party.”
These aren’t contradictory, but I fail to see the model which includes both of them. Removing the threshold does not remove the ability of the party to win the single constituency seat, so it should not affect the payoff of voting for the likely-to-win candidate in the single constituency. If there is “truly no incentive” for voters in that constituency to give their list vote to the party as well, that must imply that they have no payoff from additional list-seats won by the party. And that can only mean that they are not really party supporters.
In all three elections (2005, 2008, and 2011), the “ACT” voters in Epsom who chose to give their list vote to National (that is to say, the vast majority of “ACT” voters) could have won an additional seat for ACT instead had they voted for it at the list level.
In 2011, it wouldn’t even have been very difficult; it would have required only about 25% of Bank’s voters to have voted ACT on the list.
I get the bit about list votes in other districts being contingent on the likelihood of success at winning the one-seat and hence the threshold, but I don’t see why that is different for list votes in the Epsom constituency itself. Obviously, it is better if the list vote is useful, but the probability of success is reasonably high, assuming that the seat is won, and the probability of harm is relatively low.
So someone with a mild preference to ACT might prefer to play it safe, but anyone with any sense of party allegiance should have some incentive to give ACT their list vote.
If there were a chance that ACT would end up with an overhang, that would change the arithmetic, but even in 2011 polls were predicting one to two percent, which would be well over the effective threshold.
In 2005 and 2008, it would have required a lot more voters shifting their list votes from National to ACT — around 75% would have been necessary — but it would still have been quite possible, if the Hide voters actually wanted ACT representatives. (Epsom could also have bumped ACT over a four percent threshold, but not the actual five percent threshold, which is perhaps also an interesting fact.)
Seed planted by rici — 17 August 2012 @ 00:12
I may in the wrong thread, but there was at least one interesting idea about floor crossing in the electoral reform proposal for South Africa from EISA.
They propose 200 list MPs and 200 district MPs. The floor-crossing rule would remain for list MPs and be abolished for district MPs.
Seed planted by Alan — 17 August 2012 @ 04:33
rici said, “Removing the threshold does not remove the ability of the party to win the single constituency seat, so it should not affect the payoff of voting for the likely-to-win candidate in the single constituency.”
I am assuming that in most situations, just one seat is not all that useful for a smaller party. Yes, it can have influence, as it has now, so I am not claiming one seat is useless. But I am assuming most of the interest in a place like Epsom in giving the nominal (electorate) vote to the little party is that it might also collect enough list votes around the country to win at least one more seat, and need the electorate in order to clear the threshold.
Now, you may be right that Act voters in Epsom itself could have gotten their party another seat by giving it their list vote, too. That’s a very interesting observation! However, it seems like an extra level of strategic sophistication being required. Besides, 2 or 4 seats–or for that matter 7, as it won several elections back–is not all that useful if National is not the largest party overall. (Unlike the Greens, Act is not a party likely to find some key issues on which to cooperate from the opposition benches with the bigger party that is farther away from it.) And so it seems to me that the way voters have split in districts such as Epsom is strategically rational.
It would be very interesting to game out the conditions when this would, or would not, be the “correct” strategy. I am not the one to do that, but it would be great if someone would.
Seed planted by MSS — 17 August 2012 @ 16:45
Alan, is the MMP proposal for South Africa similar to the one already used in that country’s municipal elections?
Seed planted by MSS — 17 August 2012 @ 16:48
MSS: “I am assuming that in most situations, just one seat is not all that useful for a smaller party.”
Yes, I agree.
“most of the interest in a place like Epsom in giving the nominal (electorate) vote to the little party is that it might also collect enough list votes around the country to win at least one more seat, and need the electorate in order to clear the threshold.”
But here you’ve lost me. Why would the Epsom voter want to provide the opportunity for a non-Epsom vote to cast a vote that the Epsom voter is not interested in casting? Or in other words, if the Epsom voter believes that more list votes would be useful, why don’t they cast one? Their list vote is worth no more nor less than anyone else’s.
That doesn’t seem to me to be particularly complex or sophisticated reasoning. In fact, it is no more than a “sincere vote”, if they are sincere about ACT. Would you consider it likely for a Green voter to decide not to give their list vote to the Green party because in their constituency the Green candidate was likely to win? (This is different from the question of whether the Green candidate in such a constituency might attract nominal votes from supporters of other parties, which is quite likely.)
So I interpret the massive failure to vote ACT in Epsom an indication that the voters in that constituency are not sincere about ACT, and equally with the other examples of single-seat parties. And if eliminating the one-seat threshold is a measure against this phenomenon, it makes perfect sense to me.
Seed planted by rici — 17 August 2012 @ 18:50
From Gschwend and Van der Kolk, Split Ticket Voting in Mixed Member
Proportional Systems: The Hypothetical Case of
The Netherlands, Acta Politica (2006) 41, 163–179. doi:10.1057/palgrave.ap.5500150:
(pp. 168-169, emphasis added. Document available here)
Seed planted by rici — 17 August 2012 @ 19:29
Rici, dead URL. Alan, if the 200-200 split were a parallel system, there would be a nice sort of symmetry in the 200 party-list MPs being removable by their parties while the 200 district MPs are recallable by voter petition. Not sure how to apply it to MMP, though.
List-PR has one advantage over STV in making it much easier to identify which “constituents” should have the legal right to remove one MP of several (for a discussion of whether and how to apply this to STV, see http://tinyurl.com/8lul2f8). But it also creates the problem that an MP elected by tens of thousands of voters (even with closed lists, there might be a general feeling that his or her name brought votes to the list) who is then expelled or demoted mid-term by a small clique on the party executive.
One possible solution might be to stipulate that
(a) the party has to hold a mid-term ballot of its members if it wants to re-arrange its list (so we are talking hundreds of voters – not tens of thousands, true, but at least not a dozen)
(b) the candidates victorious in the ballot go to the top of the list version 2.0, above any names from the list version 1.0 as of the previous general election (unseating sitting MPs and seating new ones as required)
(c) however, the number of new MPs who may be seated as a result of the recall ballot is in some way capped by the number who actually vote in it. Eg, one per 500 votes cast or part thereof. So if the party holds 10 seats in that district and 1,817 members vote in the recall ballot, then only the 3 highest candidates in the ballot go to the top of the list, so at most only 3 sitting MPs (as per the list order last shown to the general electorate) will be unseated, assuming they’re not also among the top 3.
Seed planted by Tom Round — 17 August 2012 @ 20:01
Tom, thanks. Appears that WordPress doesn’t like urls with apostrophes in them; if you view source, you’ll see that it turned the apostrophe into a quote, thereby truncating the url. (You’ll have to take my word for the fact that it was a quote when I pasted it, though.)
corrected url (I hope).
Seed planted by rici — 17 August 2012 @ 20:14
Same deal, the list MPs remain under the same rules they are now.
Seed planted by Alan — 17 August 2012 @ 21:42
rici (relating to comments # 36, 34, 32), let me try again.
If a voter wants a smaller party as a coalition partner for their preferred big party–tilting the latter in a more extreme direction or pushing it to take up some set of distinct issues–then the rational vote under MMP is:
1. If the small party has no chance in your district, vote for it on the list. (And vote for the bigger party you most like in the district, even though this is unlikely to affect the partisan balance in parliament.)
2. If the small party has a chance in your district, vote for it; give the list vote to your preferred major party. This way both of your two votes are potentially effective. You have voted for your preferred coalition.
I assume the incentive for scenario #2 is no weaker and likely stronger if you also expect that there are enough voters for the small party nationwide that it will win more than just your district–provided it also wins your district. Looking at it from the other end, of course, the incentives for voters outside the district to give this party their list vote are only strong if they expect the party to win its one best-shot district–assuming its reaching the 5% threshold is iffy at best.
As for your question: “Would you consider it likely for a Green voter to decide not to give their list vote to the Green party because in their constituency the Green candidate was likely to win?”
See Coromandel 1999. This was the Green’s “Epsom”; the pattern is the same.
By the way, I agree that the small-party vote in a district like this–including Epsom now–is not a “sincere” vote for the small party. It is sincere for a relatively more extreme or issue-based coalition, however.
Final point: we should never assume that only the list vote can be considered the sincere one.
Seed planted by MSS — 19 August 2012 @ 15:26
@MSS 35
I think so yes. The proposal at EISA keeps the National Assembly at 400 with half each from districts and lists. Apparently this was discussed briefly during the constitutional negotiations but omitted during the finalisation of the permanent constitution between 1994 and 1996.
Seed planted by Alan — 19 August 2012 @ 18:41
I don’t really get this whole “best loser is illusory”. In any list based MMP system, the candidates awarded list seats are not the list seat winners. In fact, they will typically be far down the list in terms of votes received (in an open list) as they are coming from lower parties. The whole crux of MMP is to award “losing parties” seats.
An additional point not mentioned is the use best loser at least provides some guard against “decoy lists” in that the list portion of the decoy twins are required to at least field candidates ressulting in a possible vote split. This of course can be overcome by co-ordination, but that is at least another hurdle.
And again, I just want to make sure we’re clear so that this does not somehow stay on the record. The use of a best loser system does not lead to more overhangs.
In the case of BW, I believe the overhang issue results from overhangs being awarded at the regional level within the state.
Seed planted by Nick Anand — 23 August 2012 @ 13:10
Nick, the full paragraph referred to is 2.196 in this scanned document (so don’t bother trying to search).
What they describe as “illusory” is the idea that best-loser in any way measures the value placed on a given candidate by voters. That’s because the voters evaluating candidate A are not the same voters as the ones evaluating candidate B, and the size of the two sets is a measure of the party popularity in the respective constituencies, which probably has a lot more to do with how boundaries were drawn than voters’ perceptions of the candidates. In particular, voters for a party (even a major party) who have the misfortune to live in a constituency where there are few other supporters of that party are doubly unblessed: first, they have no chance of electing their candidate as the constituency candidate, and second they have little likelihood of having their candidate selected from the best-loser list.
If, in addition, best-loser is combined with a fused vote (as I think you are recommending, and as was recommended in the Hansard MMP proposal referred to in paragraph 2.196), then it is likely to lead to a decrease in vote for minor parties, as a result of increased influence of strategic voting for the major parties.
Seed planted by rici — 23 August 2012 @ 16:40
Don’t mean to be dismissive. The report in 2.194 uses a faulty premise that “best losers” requires one-vote MMP. Read it yourself in 1.94. It simply is not true. Two-vote MMP can always be used with “best losers”. I believe we covered this above.
Though, I see some point that a member in a safe seat for another party has less chance of being elected in a best loser situation. I’m not sure how closed lists are preferable.
It simply assumes that party HQ knows what’s best in terms of nominating candidates. In fact, closed lists really enforce the idea the second class nature of list MPs, as they can be shown to be nothing but drones.
So to point to a flaw in best losers without at least acknowledging the exact same issue is prevalent with the status quo (except much worse) makes me think the report writers had “skin in the game”., especially since they’ve used a logical trick in 2.194 and 195.
Furthermore, best losers typically yields something important. The lsit members will often be from an area where support for that party is relatively high. Closed lists do no such thing. So if Labour ran a close second in a marginal riding, those votes are still likely to be represented by a local member.
Furthermore, it also ensures some geographic distribution of list seats.
This comes at the expense of a “worthy” candidate who finished a distant second. But is that the end of the world? If anything, all that really requires then is some “parachuting” of candidates.
Now open lists are probably the best choice in terms of selection of individuals (if that’s preferred over list members coming from where party support is highest). However, they require a wholescale change especially in systems like New Zealand which have national allocations. Or to put it more bluntly, they require a huge ballot which may lead to more confusion among voters. See point 2.199.
It’s a debate worth having, and I can see arguments either way.
But this report is basically dismissive without any logical backup.
Seed planted by Nick Anand — 23 August 2012 @ 18:34
Nick,
Perhaps the difference in our views comes from our histories. I don’t know yours, but I grew up in a place where the party I most identified with was never going to get elected (although things change, including the fact that I no longer live there [1].)
So what you see as a good thing (the marginal district with two representatives) seems awful to me. The labour supporters of districts where labour is running around 15% have zip: there local representative has a different, possibly even hostile ideology, and their party representative is busy dealing with their lucky cousins in Marginal District. Whereas those lucky cousins have two representatives they can choose between. How lovely for them.
I continue to think that if you want to compare the vote for A with the vote for B, then the voters should be the same. Otherwise, it’s apples and oranges.
Most of the various German systems use regional allocation of list seats (so does Mexico, at least for the lower house, and I’m sure there are many other examples) which forces some kind of geographical diversity on list, if that’s really needed. It suffers from the same problem as I complained of above, though normally not as strongly: representation becomes regionally-biased at the list level as well.
I think there’s some argument for counter-balancing nominal representation bias with list representation biased in the opposite direction, or at least unbiased; the latter could be done the same way that (closed) lists can be used to provide gender and minority balance. Suppose that a party believed in that, and puts on the top of its closed list a minority representative from a part of the country where the party gets very little majority support. Is that bad? Or is it good? Or, more generally, is it something voters can decide?
Because, at the end of the day, the composition of closed lists are known to voters in advance of the vote (at least, in most jurisdictions, and I would strongly oppose any system in which that wasn’t the case). And what I’ve seen — even here, where list order is basically irrelevant — is that parties are very well-attuned to voters, and deliberately put candidates they think will be popular at the head of their lists. And they’re usually right. Furthermore, when they’re wrong, voters react. In the case of MMP, it’s very easy for a voter to vote against the party list because they don’t like whose on it, since they can still vote for their local party candidate. Why do you think there is a problem?
fn1. I’m actually a fan of open lists, and moreover I live in a country with an open list PR system, so that’s all good.
Seed planted by rici — 23 August 2012 @ 19:27
Rici,
I am going to table the open list v. best loser debate for a second, as you seem to have defended closed lists.
I am going to try and answer a couple of themes which you’ve articulately put forth in the last post.
Promotion of minorities: Promotion of minorities (or women) is typically only done in closed lists where electoral law so requires. Any election system (closed, open, best loser) could incorporate some form of minority/gender promotion into it. Though I do not support the idea on principle, I believe it could be easily completed, and the idea of best loser are not mutually exclusive.
The “15% labour seat”: As you say the person in a heavily National riding (assume 85% for simplicty) who is a labour voter is very unlikely to have his “worthy” member elected. My response is “Not every voter is going to get a perfect mix of ideology and localness”. Now I know that sounds harsh. But think about the reverse, the “45% labour seat losing candidate”.
In a closed list which promotes the 15% labour candidate over the 45% based on Labour HQ’s beliefs, there are going to be a lot more people who have no voice (the 45% of Labour voters will have no local representative) rather than the 15%.
A bigger group (the 45%) will need to deal with the “local representative has a different, possibly even hostile ideology”. Not sure I understand why that’s better than 15% having to deal with that.
“Deliberately put candidates they think will be popular at the head of their lists.” I don’t think that’s entirely true in MMP.
Logically, large parties would usually put their best candidates in constituencies (when there’s a likelihood of overhangs).
Whereas, small parties put their best candidates on list as that’s the most likely means of electing them.
Furthermore, I find it highly doubtful that any voter would target a list vote in a compensatory list (like MMP or Scandinavian levelling seats) based on members given the amount of other items that come into play, except for very small parties.
“In the case of MMP, it’s very easy for a voter to vote against the party list because they don’t like whose on it, since they can still vote for their local party candidate.”
That assumes that your local candidate has a viable chance at winning locally. Otherwise, if you don’t like who’s on your party list, aren’t you basically screwed?
Whereas at least with best losers, instead of a party needing to target their list choices to voters who choose to approve it or not approve, that choice is made by the voters themselves.
“even here, where list order is basically irrelevant ”
I’m not sure where you live, but your positive response to the transparency of a list system suggest you live in some type of flexible list PR country. Just a guess.
By the way, I am from Canada, and am generally supportive of some type of MMP (preferably one-vote though).
Seed planted by Nick Anand — 23 August 2012 @ 20:50
Nick, for a Canadian perspective on the issue:
I live in Montreal, and some people I know vote Conservative. I may not agree with them, but there’s a hundred thousand such voters across the city, much more than the 20 000 or so votes that it usually takes to win a single riding. Despite this, they win no representation in Parliament, because they’re so spread out—in fact, only one Tory candidate even achieves second place.
In a best-losers system, these voters would continue to have no representative from their chosen party who’s even vaguely local. Obviously, as you state, it’s better if a 45% candidate wins over a 15% candidate, all else being equal. But regions are important, so not all else is equal.
However, under a closed- or flexible-list system, the Conservative party could place a candidate from Montreal in a winnable position on their list. This would let them cultivate voters here, and perhaps help prevent some of the alienation that’s common in Canadian politics.
Seed planted by Vasi — 24 August 2012 @ 01:26
Vasi,
One would argue that, under a hypothetical MMP system, the Conservatives in Montreal do get their voices represented in parliament through the significant overhang the Conservatives achieve in the rest of the country. My modelling based on 2011 shows the Conservatives qualify for a compensatory at 74-26 SMDs vs. list (nationally) and tehn it’s only one seat.
However, to your point, they don’t get local Conservatives from Montreal.
Again, I am not sure why you assume the Conservatives, under any closed list system, would feel obliged to put their first candidate as a Quebecois, more than some other region such as downtown Toronto.
The argument only works based on the goodwill of the party.
Again, open lists may be a better option, however they have a separate
Any system could work with some type of regional allocation mechanism above it, which first allocates nationally calculated seats to regions and then chooses members based on best loser criterion (or open list or closed list for that matter).
This could work something like the current German model (except eliminating the negative vote weight), and would give a real test to determine if Montreal Conservatives were underweighted in the party after accounting for overhangs, based on quants.
I’ve actually tried one out that more or less works except under extreme fragmentation.
Seed planted by Nick Anand — 24 August 2012 @ 16:50
This is a fascinating normative discussion about the merits of best-loser vs. closed lists. There really is no way to resolve it, because it depends on what your values are. I guess that’s exactly what makes it fascinating!
Regarding a couple of points raised in #47…
“Promotion of minorities (or women) is typically only done in closed lists where electoral law so requires. ”
Actually, I think the literature is pretty clear. Except for less developed countries, women (and often other under-represented groups) have tended to be more likely to be elected under closed lists than under other electoral systems, holding constant the presence of legal mandates (quotas).
“Logically, large parties would usually put their best candidates in constituencies (when there’s a likelihood of overhangs).”
In MMP, it seems logical to me that large parties would put their best candidates both in a nominal-tier constituency that was relatively safe for the party (hence the overhang issue is not relevant), and in a top list position. And they would advertise that the candidate was on the list so as to attract votes from the wider list-tier district. I am fairly certain this is what such parties do, although I do not know systematically, and of course the analysis would depend on how one defined “best” candidates.
Seed planted by MSS — 24 August 2012 @ 17:18
Can’t a Best Loser MMP system be combined with an open list system?
I always go back to this, what about a one vote MMP model using FPTP with a best loser system, or a one vote MMP model using Australian style preferential voting with an open party list allocation with the first count of votes for list seats, and the elimination of candidates until one wins 50% plus one vote for the district seat.
Let’s say we have the NZ MMP using closed lists and the two vote method, however, It might be better to use a threshold method in the district seats, such as all candidates have the right to be on the party list if they surpass 5% of the vote in the district.
At least minor parties who run in a district are force to campaign in a small district to win votes to be on the list. The party can set up the list as it is closed, but if any member of their party wins less than 5% in a district, then they are off the list. Then this may discourage some members of that small party not to run in a district if they are fearful of polling less than 5% in a district, as they would be off the list.
This wouldn’t be a best loser method, but a minimal threshold method. Some may want to make this more elaborate and more complex, combining this with a an open list system would be interesting, but might lead to perverse results where the candidate wins lots of votes nationwide, but is rejected in their home district. I guess as they say all politics are local.
This system would not hurt large parties at all, but make small parties candidates achieve at least over 5% in the district to be on the list.
Seed planted by Suaprazzodi — 24 August 2012 @ 20:16
I think MSS is right that there is no real answer to this discussion — I originally wrote “resolution” but obviously there is a resolution, and that will be what New Zealand decides to do, independent of the opinions of non-New-Zealanders. [1]
So let me just clarify a couple of things, and then I’ll stop.
First, Nick asks why I think that it’s better for 45% of a constituency to be unrepresented than 15% of a different constituency. The reason is that the National MP (to keep with the example) with 85% of the vote has no particular need to be concerned with the minority who vote for other parties, while the Nat with a bare majority needs to watch out. (In other words, I would expect that constituency representatives in MMP would conform to the median voter model just as well — or not — as elected FP(T)P representatives.) That’s at least a testable hypothesis although I don’t actually know of any tests. I’ll look, though.
Second, I think Vasi hits the nail on the head with his example of Montreal Conservatives. I don’t actually have any insight into the thinking of this rare breed, but if they are not otherwise different from other Montrealers I know, they would not feel themselves much represented by someone from Bloor Street, even a Bloor Street Tory.
Furthermore, I don’t think you need to rely on the goodwill of the Conservative Party to put a Montrealer high on the list of a hypothetical supplementary closed list. They’d be slavering at the bit to do it — or at least, other political parties I’m more familiar with would have been.
Why? Because political parties actually want to get elected. This motivates them to present attractive candidates. They have essentially no motivation to present unattractive candidates, which is different from presenting attractive candidates who lack a (geographical) constituency.
How can you (supposing that you’re a political party) make inroads into a province if you don’t have anyone from that province in your caucus? Your only hope is to convince someone electable in that province to run in some riding in which they might just barely have a chance to win on their own merits. Unfortunately, it’s not easy to convince really good candidates to take you up on this offer, since the only ones eager to do it will be people with an inflated view of their own attractiveness coupled with an inability to work with their former colleagues in whatever party they came from. So it’s not a strategy without risks. (It is a strategy the NDP tried for years without success until it finally succeeded in spades, but I’m far too far away to guess what it means for the future.)
If, on the other hand, you can pretty well guarantee to someone that they will be elected (by putting them high up on the closed list), then you can attract a much better range of candidates, and they will likely serve you well in your future attempts to be credible in that region.
So, yes, regions matter.
Finally, I don’t see how best-losers without a fused vote can save you from the delinking problem. Given that every (large) party has some number of really secure districts, they only need to run “non”-party candidates in those districts. Even with a fused vote, this strategy can provide some advantage, but not nearly as much.
I’ll agree that best-loser with a fused vote does help, but then I’ll argue that strategic voting for the fused vote (further) distorts the meaning of the “best” loser’s vote, to the point where its correspondence with voter preference becomes, to coin a phrase, illusory
And with that, enough. Back to the New Zealanders, if any.
fn1. I’m also Canadian, although I now live in Perú, which has a simple open-list system.
Seed planted by rici — 26 August 2012 @ 02:17
MSS, I guess I’m too dumb to figure out how the comment editor is supposed to work, so if you can fix that italic error (it was supposed to just be “wants to get elected”), I will be (even more) in your debt.
Seed planted by rici — 26 August 2012 @ 02:22
@Suaprazzodi: I think your idea of using a best losers method with a 5% district threshold in order to be elected on the party list actually makes alot of sense!
Seed planted by Derek — 26 August 2012 @ 05:10
Rici,
A couple of points and some further analysis.
“First, Nick asks why I think that it’s better for 45% of a constituency to be unrepresented than 15% of a different constituency. The reason is that the National MP (to keep with the example) with 85% of the vote has no particular need to be concerned with the minority who vote for other parties, while the Nat with a bare majority needs to watch out. (In other words, I would expect that constituency representatives in MMP would conform to the median voter model just as well — or not — as elected FP(T)P representatives.) That’s at least a testable hypothesis although I don’t actually know of any tests. I’ll look, though.”
Your method is overvaluing the wish to pander to marginal consistuencies under MMP. Because MMP is proportional (for the most part) parties are less concerned with pandering to a riding than under FPTP. The bigger thing is to simply get more votes.
You are right that the local member from the Nationals in the marginal constituency may be very concerned with appealing to his community he represents (assuming he’s not a dual candidate).
However, how will he do this in all reality? The closed list system does lead to a significantly more powerful party whip. I think we’ve alluded to this above.
The only means for a compensating member to be able to go against his party (when standing up for his local constituents) is to have the moral authority to do so. If the prospect of a bad list position is dangled over his head, he is much more likely to not have that fight.
Moreso, in a marginal seat you’re creating a much harsher punishment for losing by a few votes.
Best loser creates more of a continuum in lieu of threshold in terms of chances of winning a seat. If I lose my SMD by a few votes, it doesn’t necessarily mean I’m screwed, because I may still have a good chance at a compensating seat.
“Second, I think Vasi hits the nail on the head with his example of Montreal Conservatives. I don’t actually have any insight into the thinking of this rare breed, but if they are not otherwise different from other Montrealers I know, they would not feel themselves much represented by someone from Bloor Street, even a Bloor Street Tory.”
I agree with this. But I think it’s better served by a regional allocation of compensatory seats which makes a decision based on verifiable quantitative facts. This works on the basis that compensatory seats can be allocated to regions which each have some specified amount of compensatory seats (accross all parties).
These are then allocated based on wasted votes (total votes less seats received*party’s hare quota).
I’ve worked on a model which I ran for the 2011 election. It gave the Conservatives two compensatory seats in Quebec (out of 10 nationally for the Conservatives and out of 32 in all of Quebec). This was based on 70% of the seats coming from SMDs nationally which produced a Gallagher Index of 1 (mainly coming from small parties and independents).
This model could be run to fine tune intra-province allocations (Montreal v. Quebec v. ROQ). I have not had the time to set that up because it requires some data input of riding results. But would function the same way as the national to province allocations. On the face it appears likely that it would give one of the compensatory seats to Montreal (I think it would be Larry Smith).
The point really is that a regional allocation system would be better because it would leave the choice of where list seats are given to the voters.
The question would still remain which candidate within a sub-provincial district should be given the compensating seat that’s been allocated to it. This again could be done by any system (closed list, open list, or best loser)
And I still think that’s better decided by the voters then the party HQ. But hey I beleive in democracy. It allows these decisions to be made through quantitative voting patterns. Basically, it approximates open lists, without logistically having to put it on the ballot.
It also better legitimizes the compensating seat winner as representing a group of people and not simply party policy.
In my ideal structure, the allocation would be detailed enough to go so that each set of 2/3 SMDs would have a compensating member, while still maintaining national allocations.
“Furthermore, I don’t think you need to rely on the goodwill of the Conservative Party to put a Montrealer high on the list of a hypothetical supplementary closed list. They’d be slavering at the bit to do it — or at least, other political parties I’m more familiar with would have been.”
You’re right in that they may be wishing to put a Monreal member high on their list. But this come at the cost of Conservatives in another area who may be equally underrepresented. The best means to judge that fact, and thereby make decisions, is through a quantitative mechanism based on actial votes, not based on what a party deems important.
“If, on the other hand, you can pretty well guarantee to someone that they will be elected (by putting them high up on the closed list), then you can attract a much better range of candidates, and they will likely serve you well in your future attempts to be credible in that region.”
You mean you can guarantee someone a seat without him getting the confidence of voters. That sounds awesome. I’m sure that will seem very credible to voters. If it gets abused a few times, people will scream bloody murder.
“Finally, I don’t see how best-losers without a fused vote can save you from the delinking problem. Given that every (large) party has some number of really secure districts, they only need to run “non”-party candidates in those districts. Even with a fused vote, this strategy can provide some advantage, but not nearly as much.”
How does the use of best losers versus any other intraparty allocation system affect the strategic de-linking issues. It is a common issue to all MMP systems. Germany has the issue with regards to CDU/CSU right now and it’s not closed list.
This argument is like debating first-past-the post versus alternative vote and saying, “The problem with AV is that the minority isn’t represented. That’s why FPTP is better in this circumstance.”
“I’ll agree that best-loser with a fused vote does help, but then I’ll argue that strategic voting for the fused vote (further) distorts the meaning of the “best” loser’s vote, to the point where its correspondence with voter preference becomes, to coin a phrase, illusory ‘
Let’s be clear, you don’t need a fused vote for best losers to function any better/worse. You could easily have the members chosen based on their proportion of district level votes (or first votes), after intially determining national allocations to regions based on party votes (or “second votes”).
The only issue would be that people may choose to vote for one the top two candidates in the district election, while saving their party vote for their “true” choice. This could slightly distort small party choices (parties who have no chance in those ridings) under a best loser system.
However, in the context of some type of regional allocation system (on wasted votes), this would be somewhat “self-regulating” in that parties would be allocated seats where they were marginal losers.
Seed planted by Nick Anand — 27 August 2012 @ 15:07
Suaprazzodi,
This type of threshold would pre-suppose that all candidates would have to be riding candidates. Please confirm.
An issue I could see is that you may well force small party voters to “waste” their nominal tier vote, making sure their candidate gets on the list.
This would have the benefit of discouraging “vote doubling” like we see in German elections and more importantly decoy lists. However, many (especially on this site) seem to worry that this leads to a drawback that smaller parties’ voters are much less likely to have an influence on the local race.
An issue is that small parties around the threshold could easily not have enough qualified candidates to fill the positions they’ve won in the list tier. Basically, there would be an “underhang”. I’m not sure that’s a great thing, as this would dispropotionately fall on small parties and not big parties.
And this would really not solve the issue of big party “yes men” as the threshold is still quite low for a big party candidate.
The best part versus other list-based systems is that it requires candidates to run in SMDs. This at least ties list winners to an SMD, and legitimizes then somewhat.
However, you don’t really need the 5% threshold to achieve that.
Seed planted by Nick Anand — 27 August 2012 @ 15:32
Yes, my proposal of a 5% threshold to be on the list will require small parties to be in the SMD. It may be better to make this a one vote MMP system, then at least small party candidate has to win more than 5% to be on the list.
Seed planted by Suaprazzodi — 27 August 2012 @ 23:04