New Zealand has a mixed bag when it comes to minor parties surviving a full Parliamentary term if they've entered into a coalition or confidence and supply agreement. Since the first of these was signed in 1996, if you exclude agreements with single MP parties, five have failed and five have succeeded. Though the United Future split in 2005 nearly made it to the election, falling about a month short.
Progressives 2002 - 2005
NZ First 2005 - 2008
Māori Party 2011-2014
Māori Party 2014-2017
NZ First 1996 - 1998
Alliance 1999 - 2002
United Future 2002 - 2005
United Future 2005 - 2007
Māori Party 2008 - 2011
That makes for a 50% chance that either NZ First or the Green Party will experience a schism during this Parliamentary term. That being said, We haven't had a minor party combust in Parliament since the 2011 election. This might suggest parties are learning to manage the pressures of these arrangements better, but then again the Māori Party had three MPs in the 50th Parliament and two in the 51st Parliament, which likely lends itself to better stability.
Here's the length that each of the five failed coalition or confidence and supply agreements have lasted.
Given that the Labour and United Future confidence and supply agreement did nearly last the term, if you exclude this from the results, it drops both averages for agreements with the Labour Party and overall agreements to 711 days.
If there is a split, and there's a roughly 50% chance* one of the two parties will splinter, when is it likely to happen? Using the averages in the above table we're looking at a period anytime from 7 October 2019 through to 12 February 2020. Excluding the United Future 2005 split, leaves us squarely on 7 October 2019.
When you think about it, this makes sense as to when a split might occur. It's roughly 12 months out from the next election and both the major and minor parties in the agreement, but especially the minor parties, are beginning to flex their muscles to differentiate themselves from their partner and demonstrate some independence to get attention and show voters why they still matter.
If you want to look at a broader time period on the above numbers, the earliest a split might occur is 24 May 2019, and the latest (excluding the 2002-2005 United Future split which did run for three full years if not the full Parliamentary term) would be 19 February 2020.
It's also likely that by this point, most, if not all, of the undertakings made in the coalition or confidence and supply agreement have been, or are being delivered, and the two parties are having to negotiate on a policy-by-policy basis.
Usually it's taken a specific policy decision or external event to cause underlying tensions to erupt into a schism. In 1998 it was the sale of Wellington airport that triggered NZ First's breakup, for the Alliance in 2002 it was the build up of tension following a string of poor poll numbers and an internal party perception of subservience to Labour, and for the Māori Party in 2011 it was Hone Harawira's objection to National Party policies.
As the issues over the Kermadec Ocean Sanctuary this weekend past illustrated, there are differences in policy and ideological approaches between NZ First and the Green Party that Labour is going to be stuck in the middle of trying to bridge. While these will be easy to manage in the early days as each of the minor partners is this agreement focus on getting there policy wins on the board, as we close in on the 2020 election, the pressure on the two minor parties will grow, especially if Labour remains, as I expect they will, very popular under Prime Minister Jacinda Ardern.
While National will struggle to put pressure on the Greens, they will be able to squeeze NZ First's support by targeting their voters in rural areas and provincial cities. If National is successful in doing that, and they could very well be, then with the added combination of Winston Peters' advancing years NZ First may well be the ones to give out first.
Which is why Winston Peters is pushing so hard for his Waka Jumping Bill. He can see the dangers that lie ahead for his party, and he's trying to nullify them before it's too late.
*In terms of the 50% chance of a split, I've calculated this off the five failed and five successful agreements featuring parties of more than one MP. If you drop the United Future confidence and supply agreement of 2002 - 2005 from this list, as it very did nearly run the full Parliamentary term, you could also argue that the Māori Party agreements from 2011 to 2014 and 2014 to 2017 should be considered as one and the same, largely because there was significant continuity between them.