One reason people worry about coalition governments falling quickly is the issue of confidence votes, where the government in power falls if it does not pass issues labelled as "issues of confidence".
I don't understand the distinction between votes that are issues of confidence and those that are not very well, but it seems that in Canada many votes in legislature are issues of confidence. Such issues are the easiest way for minority government participants to "pull the trigger" on the government.
My understanding was that votes of confidence are a feature of Westminister systems (i.e. most systems built from Britain's heritage). I was not overly concerned about the issue because I knew that New Zealand also comes from a Westminister tradition, so presumably they dealt with the issue. Then I misinterpreted
this post by localgrit as saying that New Zealand does not have motions of no-confidence. In fact, the claim is that New Zealand handles no-confidence motions differently, which is correct. The difference is illustrated in a paper by Aucoin and Turnbull (Aucoin Turnbull 2004). In Canada, the convention is that no-confidence motions result in immediate elections. Upon losing a no-confidence motion, the premier or prime minister asks the Crown (Governor General or Lieutenant General) to dissolve parliament and force an election. In Canada, the convention is that the Crown always agrees to this request. In New Zealand, the convention is that the Crown has the right to determine whether another government could be formed by the opposition and some coalition partners. If no such government is possible, then another election is scheduled. The big difference is that in New Zealand no-confidence motions do not immediately trigger elections, whereas in Canada they do.
New Zealand does have confidence and no-confidence votes. In fact, some votes are explicitly labelled "No Confidence Motions" (see motion 90 from April 9 2004 in (Gillon 2005), a press release published by the Progressive Party of New Zealand).
Despite the existence of these confidence motions, New Zealand does not fall apart every year and a half, as the critics of MMP would have you believe. What is going on?
The key point is that none of the no-confidence motions pass in New Zealand. The country has had four elections under MMP: 1996, 1999, 2002 and 2005. As far as I can tell, none of these governments fell on motions of non-confidence. Prime Minister called an early election in 2002, but as far as I know she did not engineer the fall of her government. The 1996 government was close to falling apart by 1999: National's coalition partner New Zealand First had left the coalition, and National had to cobble together support from a bunch of independents who had left the New Zealand First party. However, even that government (which was crazy in many ways) hung together long enough for the election to take place at its regular time.
This should astonish the MMP critics, especially given New Zealand's dirty secret: three of the four governments in New Zealand (all except the 1996 fiasco) have actually been minority coalitions, in which the formal coalition partners do not hold a majority of seats between them. Surely this is exactly the kind of situation that gives MMP critics and FPTP supporters night sweats -- and yet New Zealand plods along. How could this be?
It happens because there are actually two levels of agreement the biggest coalition partner (thus far, the Labour party) has made. The first is a formal coalition where the biggest partner chooses parties to participate in the government directly, and the parties rule more-or-less together. The other agreement is one of "confidence and supply", where the big coalition partner makes agreements with certain parties to support them in confidence votes. This is a weaker partnership than being in coalition. The confidence and supply parties can oppose the main party for almost everything except confidence votes, while coalition partners are expected to support all (or almost all) legislation together.
This is one reason it sometimes takes a while for governments in New Zealand (and other countries that use PR systems) to sort themselves out. Sometimes the transition is smooth: the coalition partners agree to partner with each other during the election. Sometimes forming the government is harder, so the deal-making takes longer.
What this means is that the different ways that New Zealand and Canada is not the definitive factor why New Zealand's coalitions don't fall apart, because the no-confidence rules don't come into play. That does not mean Ontario is safe, however, because New Zealand is doing a lot of other things that make its coalitions work.
The first difference is that New Zealand Prime Minister Helen Clark has been effective at cobbling together coalitions and confidence and supply partners that work. Less skillful leaders might make poor choices in coalition partners, or might be stubborn and attempt to rule without support from other parties. In order to be successful under MMP, the big parties will have to develop skills in coalition making, but this will likely take time, and we might expect that the first few governments under MMP will be crazy.
Another difference is that New Zealand has a convention of three year terms between elections. In a few cases -- the National government of 1996 that almost fell apart, and the Labour government that called an election in 2002 a few months early -- the length of this term could have been significant. The McGuinty government legislated four year election terms, which makes it more likely that impatient parties might try to break government early. Again, parties will eventually learn that calling elections too frequently in Ontario is political suicide, but it might take an early election or two for parties to get the lesson.
That does not mean that we should ignore the no-confidence issue. It does mean that the issue itself is not the primary barrier to coalition stability, and thus is not a very strong argument for sticking with FPTP.
References
(Aucoin Turnbull 2004) Peter Aucoin and Lori Turnbull. "Removing the Virtual Right of First Ministers to Demand Dissolution". Canadian Parliamentary Review, Summer 2004, p. 16-19. Available at
http://www.parl.gc.ca/Infoparl/english/issue.htm?param=160&art=285 (Gillon 2005) Grant Gillon. "80% of laws gain broad support; MMP is working."
http://www.progressive.org.nz/modules.php?name=News&file=article&sid=1484 Posted 04 February 2005. Accessed July 5 2007.