This is one section in a planned series on “improving MMP”, as the New Zealand Electoral Commission sets up its process of reviewing the electoral system that New Zealanders voted to retain at the referendum in November, 2011.
Campaigners for “Keep MMP” in downtown Wellington,
the day before the vote.
The terms of reference for the mandatory review now underway include the question of whether to continue allowing dual candidacy–the right of a candidate to stand simultaneously in a district (electorate) and on a party list. In this essay, I will suggest that banning dual candidacy outright would be mistaken, but I will consider some possible provisions that would ameliorate the potential “legitimacy” problem that can arise from losing electorate candidates gaining a seat via the party list. I will caution against a “best loser” provision, which would allow candidates to win PR seats only if they have run close losing races in their electorate. Instead I will suggest an “incumbent defeat assurance”, whereby incumbent MPs who earn fewer votes that their party’s list in the district, and lose the district, would not be able to retain a seat via the list. I will also suggest an alternative form of “best loser” rule that rewards those electorate losers who outpoll their parties in their districts.
One of the most criticized aspects of MMP in New Zealand is the possibility that a candidate can lose an electorate contest, yet remain in parliament via the party list. This seems to be especially subject to criticism in the case of incumbent electorate MPs, who are perceived as having been defeated by their voters, yet reelected thanks to the closed party list.
One can argue that this is not really a problem at all. The idea of a mixed-member system is that there are two ways of winning election: being a local constituency MP, or being a party MP. It is sometimes alleged by critics that losing a district race but winning on the list is entering parliament “by the back door”. But that is not accurate, because MMP is a system that might be said to have two “front doors”, neither of which is less legitimate than the other. It is not as if the lists are secret. If voters do not like the list of candidates of a party, there are other parties to choose from. This is the advantage of proportionality–more choice of parties–and in a competitive inter-party context, parties that get a reputation for nominating bad candidates to their list are not likely to prosper. (Work by Jack Vowles has shown that, in any case, incumbent MPs who lose the district but remain MPs due to their list rank tend to last only one more term. That is, their parties do not renominate them, or do not give them a viable list rank, next time around.)
However, if one concedes that the presence of MPs who are district losers but list winners is a problem–whether of actual legislative performance and accountability, or simply bad PR (in the sense of public relations)–there are various ways to address it. One is to open up the lists so that list MPs also need to win votes. This is a separate issue, and one that I will address separately in a later section. There are also ways to address the issue within the context of closed lists.
Ban dual candidacy?
One way is simply to ban dual candidacy. Candidates, and their parties, would have to decide between electorate candidacy and list candidacy. This is the simplest approach. It is also the worst. It would mean parties needing to nominate close to the full 120 candidates rather than 70 to 75, as is currently typical.
The only sign I saw that was for both a party and (more subtly) also for a position on the referendum. Seen in central Christchurch.
A ban on dual candidacy would be especially disadvantageous to smaller parties that depend entirely or almost entirely on winning list seats. These parties would not be able to nominate their leaders in districts, because if they did, they’d have their leadership outside parliament. Thus banning dual candidacy would weaken the links of smaller parties to geographic constituencies, which would seem to go against the deeper preferences of those advocating changes to dual candidacy rules. That is, the criticism of MMP from these reform advocates is that list MPs are not “accountable”, but if a ban on dual candidacy results in smaller parties not contesting electorates with their best personnel, then the resulting elected members (from the list) are less accountable because they’d have little reason to pay any attention to geographic constituencies. And, yes, leaders and other candidates of parties that rely on the list vote to win all their seats do in fact campaign in the districts. I witnessed such campaigning by Greens, including their national co-leader Russel Norman, while I was in New Zealand prior to the 2011 election.
Choosing the compensation members from among the “best losers”?
A possible solution is to adopt a “best loser” provision, as advocated in the New Zealand Herald.
A party’s list seats could, for example, go to its candidates who came closest to winning an electorate.
This has some appeal, in that it more closely ties every MP and serious candidate to a geographic constituency. Thus on the face of it, it seems like a good “accountability” corrective. However, as I already noted, many list candidates already campaign in electorates (single-seat districts) even if they are unlikely to win that way. If New Zealand currently had a clear division between plausible electorate winners, who campaigned personally in their districts, and likely list-only winners who avoided getting out and meeting voters, then a “best loser” provision might be a corrective. But it seems like a solution to a problem New Zealand does not actually have.
Moreover, the unintended consequences of a best-loser provision could be worse than the status quo. A similar provision exists in Japan, and the experience there should give advocates of “best loser” MPs pause. By rewarding a candidate for driving up his or her vote even in a district where the candidate has little realistic chance of winning, best-loser allocation exacerbates some of the worst features of FPTP. Whereas under current MMP rules, the two big parties have a strong incentive to seek to position themselves near the center of the nationwide electorate to maximize their party vote, under the best-loser rule, they would be back to the old days of targeting districts and seeking to appeal to voters located in relatively closely contested districts. They would want to do so because, even if they are not likely to win such electorates, those are the electorates where their highest quality candidates have the best shot at entering parliament despite losing the district race.
In seats that are “safe” for one party, the other major party likely would run a very weak candidate, as the party’s candidate there would have less hope of being among the several “best losers” who would enter parliament. Under current MMP rules, there is no strong disincentive to running a quality candidate in the electorate (who is also on the list). I can’t say how often it happens that a “quality” second-party candidate contests a seat that is safe for another party. It would depend on how we define “quality”, but the point is that parties now have no reason not to run a good candidate even if they aren’t likely to win the electorate. With a best-loser provision, they would want to avoid “wasting” such candidates on districts that are relatively weak for the party, who would be unable to enter parliament unless they ran close to the electorate winner.
The best-loser concept also has a clear “top two” bias in it. The provision might make sense for candidates of the big parties, where the provision might let the candidate who loses, say, 48-47, enter parliament, but keep out one who loses on the order of 52-35. Yet for the smaller parties, they may not have any electorates in which they are close to winning. Thus their “best losers” might be those who lost “only” 50-35-15 instead of 50-40-10. What is gained by ensuring that a candidate who loses by 35 points is elected ahead of one who loses by 40 points?
A proposal: Incumbent defeat assurance
Here is a proposal that could address the (perceived) problem of the so-called back-door MP: Incumbent electorate MPs who fail to win the plurality of votes in their reelection bids are skipped on the party list if their own electorate votes fall behind their party’s vote in the electorate. In this way, if the MPs in question lost because their party became less popular (which will often be a nationwide swing, and not a specifically local effect) they can retain their seat if they are sufficiently valued by their party as to have a list rank that permits their election. However, if they lose because they are less popular than their party, they are done–the list can’t “save” them from rejection by their electorate.
As noted by Kate Chapman in a Fairfax NZ article
In November, Labour MP Clayton Cosgrove lost the Waimakariri seat and National’s Chris Auchinvole lost West Coast-Tasman, but both returned to Parliament.
Sign for Chris Auchinvole near Greymouth.
Under my proposed incumbent-defeat assurance (IDA) provision, Cosgrove would have remained a list MP, but Auchinvole would not have been returned to parliament. In Waimakariri, Cosgrove won 16,145 votes, while Labour won 8,431 party-list votes. In other words, even as a loser in the district, Cosgrove was almost twice as popular with his constituents as was his party! Why should he be considered unqualified to continue when he is both popular in his district, relative to his party, and sufficiently highly regarded by the party to have earned a high list rank?
The National winner in Waimakariri, Kate Wilkinson, was less popular than her party: she won 16,787 votes, this beating Cosgrove by just over 600 votes, but her party won 20,489 party-list votes in the electorate. She won fair and square, by getting more votes than Cosgrove. But, if the IDA provision were already in effect, she would now be on notice: she would have to improve her personal standing vis-a-vis her party’s voters in the district, or she could be out next time.
In West-Coast Tasman, on the other hand, IDA would have seen Auchinvole tossed out, despite his high list rank. He ran behind his own party in the district, by getting 13,214 votes in a district in which there were 15,462 National party-list voters. As for the district race itself, he lost it by 2,539 votes (7.5%) to Damien O’Connor, one of the few Labour candidates not to be dual-nominated. That Auchinvole ran so far behind his party looks like repudiation by the voters who know him best–including those who voted for his own party–and yet he continues via the list.1
Sign for Damien O’Connor along the main highway on the West Coast.
So the most recent election gives us examples of both an incumbent against whom IDA would bite, and one who would still be allowed to remain via the party list. Both examples come from large parties. What about small parties? There is one key case of considerable importance, Jeanette FItzsimons, then Green co-leader, in 2002. She still almost doubled the Green Party vote in Coromandel that year, when she lost the electorate seat she had won in 1999. So she would have remained in parliament; more generally, a small party that has a leader who is popular enough in her district base is not necessarily doomed to be tossed out if she can’t retain the seat in the face of tough competition from the bigger parties.2
Before concluding that the incumbent defeat assurance provision is viable, it would be useful to know just how common it is for an electorate winner to trail his or her list in the district. If it is relatively common for candidates to underperform their party’s vote, then it might be the case that IDA could prove overly punitive (even though we have already seen that it would have cost only one MP in 2011).
I have readily available the necessary electoral data in spreadsheet form only for 2002 and 2005 (for now). These are actually good years to investigate, due to the very different patterns of competition between the two big parties. In 2002, the election was not close nationally, as Labour easily secured the leading position. In 2005, by contrast, there was a very close contest between Labour and National.
In looking at these two elections, I find 16 legislators who won fewer votes in their electorate races than their party won in the electorate, yet who won the electorate. An asterisk indicates an incumbent.
year, name, party, district
2002, *Tizard Judith, Labour, Auckland Central
2002, Pillay Lynne, Labour, Waitakere
2002, *Mahuta Nanaia, Labour, Tainui (Maori electorate)
2002, Fairbrother Russell, Labour, Napier
2005, *Tizard Judith, Labour, Auckland Central
2005, *Hodgson Pete, Labour, Dunedin North
2005, *Benson-Pope D, Labour, Dunedin South
2005, *McCully Murray, National, East Coast Bays
2005, *Field Phillip, Labour, Mangere
2005, *Hawkins George, Labour, Manurewa
2005, *Williamson M, National, Pakuranga
2005, Clarkson Bob, National, Tauranga
2005, Hayes John, National, Wairarapa
2002, *Mahuta Nanaia, Labour, Tainui (Maori electorate)
2005, *Horomia Parekura, Labour, Ikaroa Rawhiti (Maori electorate)
2005, *Okeroa Mahara, Labour, Te Tai Tonga (Maori electorate)
These sixteen candidates, twelve of whom were incumbents, were all less popular than their own parties, yet they are obviously legitimate winners due to a basic provision of how MMP works: they were more popular than any other candidate in their district.
These sixteen represent 20.3% of all candidates of the two big parties in 2002 and 2005 who were “list-trailers”; there were an additional 23 National and 40 Labour list-trailers who lost their electorates.
On the other side of the coin, the two big parties in these two elections had 60 candidates whose votes surpassed those of their list, yet did not win the electorate. (Eighteen, or 30%, of these won anyway, due to the list.) Electorate-losers are 53.5% of all list-surpassing candidates. So losing the electorate race when you run ahead of your party is far more common than winning the electorate despite trailing your party. This indicates that the IDA provision would hit relatively few MPs, but those few are precisely the ones that are most a lightning-rod for critics of dual candidacy.
Taken together, this partial analysis of the patterns of electorate-level votes for candidates and their associated lists suggests that there is no noteworthy tendency of candidates to be “rejected” by voters in their electorate yet be “saved” by the list–the proverbial “back door” MP. Further, the analysis suggests that a simple rule would solve whatever legitimacy problem there might be with incumbents who trail their list vote, yet are ranked high enough to win anyway: the incumbent defeat assurance provision would have such incumbents skipped on the list.
Another possible provision, and conclusions
Finally, I want to propose one other provision that the above analysis suggests could be beneficial. One could offer an alternate form of the “best loser” provision that defines a good loser not by reference to his or her margin of defeat at the hands of another party’s candidate in the electorate, but rather by the margin by which the candidate surpassed his or her party’s list vote in the electorate. Such a provision would reward those who are more popular with their own district’s voters than their own party. It would thus be the mirror image of the incumbent defeat assurance provision, which punishes those incumbents who lose the electorate race by trailing their list. Under this concept, when filling the list seats for a party, the first winners would be list-surpassing electorate losers with the greatest margins over their own party list. If a party still has more list seats earned after all list-surpassers have been elected, the remaining seats would be taken from the party list by rank order. Or, if preferred, one could continue to fill out the list with candidates who trailed their list by the narrowest margins–still skipping any incumbents among these list-trailers.
The bottom line is that if New Zealanders conclude that there is a problem with candidates losing one way (in the electorate) but winning another way (via the list), there are several ways to address it without banning dual candidacy. An outright ban on dual candidacy undermines one of the main beneficial features of MMP: the incentive of parties to have most of their candidates campaign among local voters even if they are unlikely to win the electorate seat. A ban on dual candidacies would be especially harmful to smaller parties, who necessarily have smaller pools of quality candidates to draw from in the first place, and would be deterred from running their top candidates in electorates at all. Those wanting to improve MMP should drop consideration of banning dual candidacy and instead consider the proposals outlined above: incumbent defeat assurance, and list-surpassing provisions for filling list seats.
All photos included by this post are by Matthew Shugart. Some rights reserved. Many more at the Flickr set.
- Another, and potentially more plausible interpretation is not that Auchinvole fell victim to outright repudiation, but that he was expected to win as a list MP anyway, so voters felt free to vote for the Labour electorate candidate–who was not on his party’s list–thus ensuring the district would have two representatives, one in each major-party caucus. It might be thought that this is a type of strategic behavior that should be discouraged, and IDA would indeed discourage it. (As an aside, it should be noted that this electorate actually has three current MPs, as the Green candidate, Kevin Hague, was elected from his party’s list. Green list votes were more than twice as high as Hague’s electorate vote. The difference must be overwhelmingly Green voters who tactically voted for O’Connor; without them he would have lost narrowly to Auchinvole. Yet his margin was much greater than just Green tactical votes; he evidently won some significant share of the National list voters who did not vote for Auchinvole.) [↩]
- In 1999 in Coromandel, Fitzsimons won 13,682 votes (just under 40%, and a margin of 250 over the second-place National candidate). The party votes were Green 2,640, Labour 12,390, National 10,747. In 2002, her own vote fell to 7,724, losing to a National candidate with 14,706, but her votes were still more than double the 3,232 Green votes in the district. There can be little doubt that Fitzsimons owed her win in 1999 to Labour voters tactically voting for her to help ensure the Green Party would have seats in parliament–the party ended up barely over the nationwide threshold, so it was almost dependent on her win to have any seats. Yet even in defeat in 2002, she remained personally far more popular than her party. [↩]