In 1979 a law professor named Geoffrey Palmer wrote that New Zealand had the fastest lawmakers in the West. In 2024, he updated the diagnosis: "We've gone from being the fastest lawmakers in the west to the fastest repealers in the west."

Palmer would know. He went on to be Prime Minister, and the machine he described in Unbridled Power? is the one we still run: one chamber of Parliament, no upper house, no written constitution, and almost nothing a government with a one-vote majority can't do before lunch. Most democracies have speed bumps. We bought the racing package.

Part 1 of this series counted the bill: well over a billion dollars in directly sunk and cancellation costs from a single change of government, plus billions more spent redoing and replacing. This part is about why it keeps happening, because the answer isn't that we elect unusually flighty people. The answer is the company structure.

A company with no board

Stay with the picture from Part 1: New Zealand as a company whose entire exec team faces replacement every three years. Now look at the governance around that team.

Most companies of consequence have a board that long-horizon decisions must clear, a constitution that takes more than a staff vote to change, and shareholders who can force a special resolution. NZ Inc has none of that. Parliament is one chamber, its decisions need 61 votes out of 120, and the few protected rules we do have (six electoral provisions, entrenched since 1956) don't cover a single spending commitment, infrastructure plan, or health strategy. A hospital programme and a tweak to dog registration fees enjoy exactly the same legal durability: none.

The three-year term turns that freedom into a metronome. Section 17 of the Constitution Act gives a government roughly 36 months, and anyone who's run a delivery team can sketch the rhythm from there. Year one, ship something visible. Year two, panic about the things that aren't visible yet. Year three, campaign. A 2024 study in Public Choice tracked 22 democracies over two decades and found exactly this shape: significant reforms cluster right after elections and evaporate as the next one approaches. Every democracy has the cycle. Ours just spins faster, with fewer brakes, and the 100-day plan has hardened it into ritual: each new exec team now publishes a list of the previous team's projects it will kill, and gets graded on killing them quickly.

A third of the year in urgency

Speed shows up in the record as something with a technical name: urgency. It's the parliamentary setting that compresses or skips the normal stages of making law, including select committee, which is the one stage where the public, experts, and officials get to find the bugs.

Every engineer reading this knows what shipping without review does to quality, and you don't have to be an engineer to follow the numbers. In 2025 the House sat 644 hours, the most in decades, and spent 32% of those hours under urgency, also a record (the long-run average sitting load is around 508 hours, per analysis by David Farrar). By Newsroom's count this April, more than half of the government's legislation has had at least one stage under urgency, and around 30 bills have gone through with no select committee scrutiny at all. There's a community-built tracker keeping score in real time: as I write this, it shows 21 separate urgency occasions this term, touching 57% of all bills before the House — 125 of 219.

I want to be fair about what those numbers mean. Urgency is sometimes legitimate, and every government uses it. But a third of the legislative year in skip-review mode isn't an exception regime. It's the workflow. And law made at that speed is exactly the law most likely to need repealing, which feeds the cycle we counted in Part 1.

The fix that ran out of time

Here's my favourite piece of evidence that the problem is structural, because you couldn't script it better.

The obvious patch is a longer term: four years instead of three, more runway between campaigns. It's one of the rare ideas with support across the aisle, and this Parliament actually had a bill for it, the Term of Parliament (Enabling 4-year Term) Legislation Amendment Bill. Voters had said no twice before, emphatically (68.1% against in 1967, 69% in 1990), so the plan was to do it properly: select committee, a binding referendum, the works.

The select committee stripped out the bill's novel design. The referendum got shelved on time grounds. And in February this year the bill formally stalled, unfinished, as the parliamentary clock ran down toward the election.

The fix for the three-year term ran out of time in a three-year term.

Even the things that survive get taxed

You might reasonably ask: can't anything be made to stick? Two case studies, and they're the two edges of the same blade.

The Zero Carbon Act passed in 2019 with 119 of 120 votes, about as close to corporate consensus as this country produces. It survived, technically. But in October 2025 the government cut the 2050 methane target so far that the top of the new range sits below the bottom of the old one, despite the Climate Change Commission advising movement in the opposite direction. Climate policy consultant Dr Christina Hood put it plainly: she was not aware of any government, anywhere, that had weakened an existing legislated domestic climate target. Near-unanimity bought the framework six years of life before the dial got turned anyway.

The other case is quieter and costs more. The NZ Super Fund exists to pre-pay the retirement of people who are currently in school: contributions go in now, compound for decades, and pay out when today's eight-year-olds retire. It's the single most generational financial commitment on the country's books. In 2009 a new government suspended contributions, and they stayed suspended for eight years. The fund survived; the suspension still happened. The Guardians' own estimate of what it cost, in missed contributions, forgone returns and forgone tax, is around $25.5 billion.

That's the number I'd put on a billboard, because it makes the stakes concrete in a way potholes never will. One pause, made inside one three-year window under fiscal pressure, quietly compounding into a twenty-five-billion-dollar hole in a fund for people who couldn't vote on it. The light rail fiasco cost $228 million and got years of headlines. The Super Fund pause cost a hundred times more and most people couldn't date it within five years. The most expensive churn is the kind nobody films.

The honest counterargument

Before anyone fits this series for a constitutional reform t-shirt, the other side of the ledger deserves its say, because it's not weak.

In a country with one chamber and no written constitution, the three-year election is close to the only real check we have. Shorten the leash because governments misbehave, runs the argument, and don't be surprised when a bad government uses a four-year leash to entrench worse. Hipkins supports a four-year term; so does Seymour, eventually; the voters, twice asked, have preferred to keep their grip.

Locking things in has its own pathologies. Supermajority requirements hand a veto to whoever can muster 31% of the House, which is how minority blocking works everywhere it exists. And when Parliament did once try to entrench actual policy (an anti-privatisation clause in the Three Waters legislation, 2022, passed under urgency at 60%), constitutional lawyers across the spectrum called it a dangerous precedent and it was unwound within weeks, with the then Prime Minister conceding the mistake. Entrenchment, it turns out, is a weapon both teams can fire.

All of which leaves a genuinely uncomfortable shape: the accountability mechanism and the churn engine are the same machine. You can't simply slow it down without asking who benefits from the slack.

So if term length alone won't fix it, and entrenchment cuts both ways, what's left? My answer is unglamorous: you start by measuring it. No one in New Zealand currently publishes a churn metric — how much of each Parliament's lawmaking is spent unpicking the last Parliament's work, which sectors get whiplashed hardest, how long a law actually lives. You can't manage what you don't measure, and right now nobody's even measuring.

So I measured it. I downloaded the entire statute book, all 19,000 files of it, and parsed every amendment and repeal back through 2008. That's Part 3, and the numbers are better than I hoped and worse than you'd think.