Why does New Zealand spend more than almost anyone, and build so badly?
- Grant McLachlan
- 18 hours ago
- 10 min read
The familiar excuses — too little money, too few people, an awkward geography — do not explain why New Zealand’s infrastructure is late, incremental and over budget. The real cause is a system built to be defensible rather than right, and nothing this government has planned will change it.
New Zealand has spent the last twenty years as one of the biggest infrastructure investors in the developed world. Over that period the country put an average of 5.8 per cent of GDP into infrastructure each year, near the top of the OECD, and between 2010 and 2019 it invested more public money in infrastructure, as a share of the economy, than any other OECD country. And yet, by the Infrastructure Commission’s own benchmarking, New Zealand sits in the bottom 10 per cent of high-income countries for the value it gets back. We pay like a leader and build like a laggard.
The comfortable explanations do not survive contact with that gap.
It is not, in the main, that the country spends too little. It spends more than most.
It is not simply geography, either — Switzerland and Chile carry harder terrain and get more for their money.
The harder question — the one the politics keeps stepping around — is why a country that spends so much builds so late, so incrementally, so far over budget, and, above all, decides so poorly what to build at all. The answer is not a shortage of money. It is a system quietly optimised to protect everyone involved in a decision except the public that pays for it.
Contents
The numbers are not in dispute
Start with what no one seriously contests.
New Zealand’s infrastructure problem is not a spending problem; it is an efficiency problem.
The Commission’s research is blunt: our investment levels run higher than Australia and the median OECD country, yet we rank near the bottom 10 per cent of high-income countries for the efficiency of that spend.
And the gap is widening, not closing.
Infrastructure prices have risen about a third faster than prices elsewhere in the economy, while infrastructure construction productivity has grown at roughly a third the rate of the wider economy.
We are paying more each year to build less well.
Layer the demographics on top and the squeeze turns severe.
On the Commission’s numbers, simply renewing and replacing what we already own will consume around 60 cents in every dollar of future investment — before a single new project is funded — even as net Crown debt is forecast to climb from 42.7 per cent of GDP toward 200 per cent by 2065 on current settings.
The country is about to spend a great deal more just to stand still. Which makes the efficiency question not academic but existential: if we cannot get more from each dollar, we cannot afford the future we have already promised ourselves.
The problem was never the money
If the money is not the problem, where does the cost go?
Not, mostly, into concrete and steel.
The evidence points the other way: cost in New Zealand is driven by process, not construction.
Marcos Pukenis de Castro Neves, a chartered professional engineer, sets it out plainly in an analysis appended to this feature:
multiple sequential reviews and approvals at every project phase,
programme durations long enough to inflate overheads,
risk priced conservatively into every contract and professional fee, and
assurance duplicated across organisational boundaries.
What the country reliably buys, and largely receives, is certainty and traceability. What it does not automatically receive is correctness, quality or innovation.
The Ministry of Business, Innovation and Employment has said the quiet part aloud, linking “excessive process, slow delivery, and regulatory complexity” directly to weak productivity outcomes across the infrastructure sector.
That is not an outside critic but the ministry responsible for the growth programme, conceding in its own Infrastructure for Growth documentation that the system it administers has turned against its stated purpose.
The New Zealand Infrastructure Commission’s National Infrastructure Plan, published in February 2026, reaches the same conclusion and calls for system-wide reform, clearer decision ownership, and reduced rework.
New Zealand invests over $20 billion a year in infrastructure and achieves less output per dollar than international peers.
The Commission’s own benchmarking says so, and the gap keeps widening.
This is no marginal effect.
Central government alone spends about $52.5 billion a year procuring infrastructure, goods and services, and the Auditor-General’s multi-year review of that spending located the weaknesses not in the building but in the deciding — in how work is planned, sourced and managed.
A system that piles review upon review does not grow safer in proportion to the cost it adds. Past a point it simply grows slower and dearer, and learns to call the delay prudence.
A system built to be blameless
Why would a sector load itself with process it cannot afford?
Because the incentives reward exactly that.
New Zealand’s liability settings mean an engineering or project decision must be defensible years after it is made; courts routinely test whether an organisation exercised “reasonable professional care.”
Under those conditions the rational move is always toward the documented, approved decision and away from the undocumented but technically superior one.
The result, as the structural diagnosis appended here puts it — drawing together the OECD, the Auditor-General, Deloitte and the Commission itself — is a system optimised for three properties: defensibility, documentation and traceability. It is not optimised for speed, learning or technical excellence. Those are not incidental gaps; they are the direct price of the properties the system has chosen to buy. A decision becomes defensible rather than optimal, and measured performance slips quietly behind measured compliance.
Deloitte and the Productivity Commission have both identified procurement and governance settings as a “handbrake” on innovation. The structural diagnosis that informs this feature is blunt about why: novel approaches introduce perceived risk but offer limited reward, so the system defaults to incrementalism and precedent.
That is not passivity or poor intent on anyone’s part. It is the rational response of engineers, project managers, and clients facing a system in which unapproved right decisions are more dangerous than approved wrong ones.
The behaviour the system rewards is following process and avoiding exposure. The behaviour it discourages is challenging assumptions, late learning, and trying something new.
Innovation requires variance; New Zealand’s governance structures are built to detect and suppress variance.
The paradox of the approved mistake
Here is the part that should trouble anyone who assumes more checking means fewer errors.
Highly risk-averse systems are demonstrably poor at correcting the mistakes they make.
Review and assurance in New Zealand mainly test two things — compliance with process and the defensibility of the decision. They do not reliably test whether the assumptions were sound, whether the parts integrate, or whether the thing will work in the field.
So approved mistakes survive.
Worse, once a decision has been signed off, the system actively discourages anyone from fixing it, because correcting a known error late exposes whoever approved it in the first place.
As Pukenis de Castro Neves puts it, an approved wrong decision becomes safer, within the system’s own logic, than an unapproved right one.
That is the engine of cost overrun and poor outcome in a single sentence: the people best placed to stop a bad project are the very people most exposed if they admit it is bad.
An approved wrong decision becomes safer, within the system’s own logic, than an unapproved right one.
There is a further wrinkle that analyses of this system rarely name directly: many of the roles the system has created exist because the system is complex.
When complexity is reduced, those roles are at risk. The resistance to reform that observers keep encountering is therefore not obstruction for its own sake.
It is rational self-preservation — streamlining perceived as loss of work, loss of relevance, and risk of redundancy.
The structural diagnosis is direct on this point: reform efforts encounter passive resistance, produce superficial change, and find that complexity is quietly re-created in new forms.
If jobs depend on inefficiency, inefficiency will be protected.
This is the human engine behind the institutional one, and it is why removing process steps without changing the underlying incentive structure reliably fails.
Everyone is accountable, so no one is
If approval is the shield, diffusion is the armour.
The Auditor-General’s recurring finding is that responsibility for outcomes is not consistently aligned with the authority to decide.
Risk is socialised across committees, panels, consultants and agencies; when a project underperforms, accountability is diffuse, and the after-the-fact justification rests on whether the process was followed, not whether the decision was good.
This protects institutions and individuals from blame, and it is precisely why nothing self-corrects.
A structure in which everyone signs and no one owns cannot learn from a failure, because there is no single point at which the failure lands.
The country has built committees to spread risk so thinly that, when the bill finally arrives, there is no one left to send it to.
The comfort of the familiar
The same instinct shapes who gets the work.
Review after review finds the system leaning on trusted individuals, established consultancies, and familiar delivery models. Not out of corruption. Out of an ingrained preference for the safe and the known.
A familiar bidder is defensible. A new one is a risk. So the familiar bidder wins.
Competition softens.
The solution space narrows.
A better proposal loses if it introduces novelty or discomfort.
The structural diagnosis names the winners: consultancies with entrenched government relationships, intermediaries who specialise in process, gatekeepers who control access rather than outcomes.
It names the losers too: homeowners, ratepayers, future taxpayers, and engineers with world-class capability whose judgement the system declines to use.
The structural diagnosis puts it plainly: in this environment, engineering quality has become a hygiene factor.
You need enough of it to pass a check. Beyond that, it barely matters.
Advanced technical skill and international experience do not translate into more authority.
What counts is who you know and whether you are safe to appoint. The engineers with the most to offer are often those with the least influence over what gets built.
The plan that prescribes the disease
Set that diagnosis beside the government’s flagship response, the National Infrastructure Plan, and the mismatch is glaring.
The plan’s central finding is correct and overdue: the country plans far more than it can pay for.
The seventeen Roads of National Significance, a new Waitemata Harbour Crossing, and major rapid transit together far exceed the revenue likely to be available.
The minister himself expanded the roads programme to an estimated $56 billion. On any honest budget, much of it cannot be built.
The plan’s answer is to defer the big builds, stage them, reach for low-cost fixes first, and time major roads to demonstrated demand.
As budgeting, that is disciplined.
As planning, it is the very stop-start the plan was written to end.
The Commission has already costed the pausing and cancelling of projects at $11.8 billion over 25 years.
Staged, deferred, demand-gated investment is incrementalism with a tidier name. The cure is a neater version of the disease.
The road north of Auckland shows exactly where this leads.
The Puhoi to Warkworth motorway opened in 2023 at about $880 million. Its second stage, Warkworth to Te Hana, was split off, costed separately, and left to stall. Proposed as a Road of National Significance in 2009, dropped in 2017, it is still only in procurement at an estimated $3.5 to $4 billion.
Splitting one corridor in two created the off-ramp that stalled the harder half for seventeen years.
The new plan reproduces the error, breaking the same route into segments and declaring each unlikely to reach capacity before 2055.
A method that assesses fragments will always find a reason to wait.
The minister who overrides the plan
Even on its own terms, the plan is not being followed.
The commission the government commissioned told it, in plain language, that its flagship roads are unaffordable and should be deferred and sequenced.
The government welcomed the plan as confirmation of work already underway and pressed on regardless.
It committed a further $1.2 billion to advance the roads and a tolling regime to “build the roads New Zealand needs sooner.”
The report says one thing. The minister does another.
The country is left with the worst of both worlds.
A plan that declares the roads unaffordable, and a programme that keeps committing to them anyway.
The discipline the plan preaches is the first casualty of the three-year electoral cycle. Which is the strongest evidence yet that the problem was never the plan. It is the incentives the plan does not touch.
Why none of it will fix it
Every remedy on the table operates at the level of process and sequencing: defer, stage, toll, reprioritise, shift assurance from one agency to another.
None of it changes the incentive that produces poor decisions: that following the process is safer than being right.
Strip out process steps without rebalancing who owns the risk, and Pukenis de Castro Neves warns you get fear, resistance, and the quiet rebuild of the very complexity you just removed.
If jobs depend on inefficiency, inefficiency will be protected.
The government’s instinct has run the other way.
It disestablished the Productivity Commission, the very body whose job was to diagnose this kind of systemic drag, and stood up a regulation ministry to hunt red tape.
As though the trouble were the volume of rules rather than the incentives beneath them.
The IMF’s 2025 assessment found the same pattern across the wider economy: competition, innovation, and technology diffusion all low.
The underlying rule holds everywhere.
Weak engineering capability drives more process. Strong capability allows less.
New Zealand keeps adding checking exactly where it should be trusting judgement, then wonders why the checking costs so much and gets so little right.
The Infrastructure Commission’s plan notes that New Zealand is now looking to the United Kingdom as a model.
What the UK figured out is that stripping back process only works if you replace it with something real: accountability, held by real people.
The Royal Academy of Engineering and the National Audit Office have documented the approach. It concentrates technical authority in fewer hands. It produces engineers who can make decisions in the field rather than waiting for a committee to grant permission.
New Zealand says it wants that model. It has not built any of the conditions that make it function.
Real reform means doing the unglamorous, threatening thing the plan avoids.
Concentrate decision authority in capable people.
Hold them accountable for outcomes, not for process.
Permit bounded failure so the system can learn.
That is a cultural and institutional change, not a sequencing exercise.
Nothing the government has proposed comes near it.
New Zealand has not built a system for building things. It has built a system for never being blamed.
Superb at producing a decision that is defensible. Incapable of producing one that is good.
Until someone is allowed to decide, and made to answer for it, the country will keep paying the most to build the least: on time in only one thing, explaining afterwards why no one was at fault.
