A New Zealand Act in force from 2024-11-26 added four years to every earthquake-prone building strengthening deadline that had not already passed — and as of 2025-09-29 about 5,800 buildings were still awaiting strengthening or demolition

What is happening?

Since 2017-07-01 New Zealand has run a national scheme that puts a legal deadline on individual buildings. A territorial authority assesses a building, issues an earthquake-prone building notice, records it on a national register, and from that point the owner has a fixed number of years to strengthen the building or take it down. The period depends on the seismic risk area and on whether the building is designated a priority building: roughly 7.5 to 15 years in high seismic risk areas, 12.5 to 25 years in medium, and 35 years in low. Breach carries a penalty ceiling of NZD 200,000, with up to NZD 20,000 for each day a breach continues.

Two changes have moved that deadline, and a third would move who the deadline applies to.

The Building (Earthquake-prone Building Deadlines and Other Matters) Amendment Act, in force from 2024-11-26, added four years to every deadline that had not already passed as at 2024-04-02, and created a power to add up to two further years by Order in Council. The responsible minister gave the cost of strengthening as the reason, citing a figure of about a million dollars per apartment.

On 2025-09-29 the government announced a redesign of the scheme itself. The percentage of the New Building Standard that currently decides whether a building is in scope would be removed as the test, and low seismic risk areas — Auckland, Northland and the Chatham Islands — would leave the scheme entirely. At that date more than 8,000 buildings had been classified as earthquake-prone at some point and about 5,800 were still waiting for strengthening or demolition. On the figures released with the announcement, the redesign would take about 2,900 of those out of the scheme altogether, leave 1,440 needing only lower-cost work, leave 840 needing nothing, and leave about 80 buildings in the whole country still required to be strengthened in full.

The government presented the redesign as a move to a risk-based system, in which the New Building Standard percentage test gives way to targeting of the buildings said to carry the most risk, and put the saving in demolition, strengthening and compliance cost at NZD 800 million in one opened report of the announcement and NZD 8.2 billion in another. An industry association representing commercial property owners submitted to the select committee in support of the bill, including of the provision taking Auckland out of scope.

The bill giving effect to this was introduced on 2025-12-09, passed its first reading, and went to the Transport and Infrastructure Committee, where submissions closed on 2026-02-16 and oral hearings were held on 2026-03-12. It passed its second reading on 2026-06-30 and sits at the Committee of the whole House stage. As of 2026-08-17 it has not been enacted, and the register figures above are therefore projections of what the bill would do, not a record of what has happened.

Whose problem is this?

No source opened here counts the people who occupy or pass beside these buildings, and no step in the legislative process requires that count to exist. The register is a list of buildings. Everything the scheme measures, defers and now proposes to narrow is measured in buildings, so the group whose exposure the scheme was created to reduce is the one group that never appears as a number.

Where does this problem end?

The boundary matters because the disagreement here is not about whether these buildings are dangerous. A public authority has already assessed each one and put it on a register, which is a judgment the scheme itself made. What is in question is how long the obligation lasts and how many of the buildings it continues to apply to.

What is the state now, and what should it be?

Now

What it should be

The target state is written into the scheme itself and needs no outside standard to state it. Every building carrying an earthquake-prone building notice is to be strengthened to the required level or demolished before the date on its own notice, and that date is what the 2024 Act moved and what the bill would in many cases remove along with the notice. The scheme names no other goal. No opened source gives a target for how many buildings should be off the register by a given year, and none gives a target for how far exposure is meant to fall, so the only benchmark this file can use is the one the statute already sets for each individual building.

How big is it?

About 5,800 buildings, on the figures released on 2025-09-29, were still waiting to be strengthened or demolished, out of more than 8,000 ever classified. On the same figures, if the bill comes into force about 80 buildings in the entire country would still be required to be strengthened in full. The difference between those two numbers is the size of what the redesign would change.

The published breakdown and the total it is set beside do not cover the same set of buildings. About 2,900 leaving the scheme, 1,440 needing reduced work, 840 needing none and 80 needing full strengthening come to about 5,260, which is roughly 540 short of 5,800. The reported share of 55 percent for the 2,900 matches 5,260 rather than 5,800, which reads as though the four categories were drawn from the narrower total. No opened source says what the 5,800 figure includes that the breakdown does not, so every figure here is reproduced as published rather than adjusted to fit the others.

The number of people is not derivable. Every opened source counts buildings. None gives occupancy, employment or footfall for buildings on the register, and none breaks the register down by building use, so there is no second term to multiply the building count by. A car park building and a three-storey apartment building both count as one building on the register and carry entirely different exposure. The population file records this as not-derivable with the missing terms named, rather than filling it with an assumed number of occupants per building.

Under what conditions does it arise?

1. The obligation lives inside a date. The scheme sets a date for each building and the date is the whole of the obligation, so moving the date moves the obligation by the same amount without repealing a single duty. The 2024 Act did exactly that: four years added to every deadline not yet passed, in a single instrument, with a power to add two more by Order in Council.

2. Scope is set by a technical threshold, and a threshold can be redefined. Whether a building is in scope currently turns on its rating against the New Building Standard. The bill would remove that test. Changing the definition of who is captured achieves the same result as exempting them, but it reads as a technical amendment rather than as an exemption.

3. Risk is classified by region, so a region can be removed at once. Because the scheme grades seismic risk by area, taking low seismic risk areas out of scope removes a large block of the register in one step rather than building by building.

4. The cost is dated and the benefit is not. An owner can be quoted a strengthening price this year. The earthquake that the strengthening is for has no date, so one side of the comparison arrives as an invoice and the other as a probability. The stated reason for the 2024 extension was cost.

5. There is a penalty ceiling but no published record of enforcement. The scheme can fine up to NZD 200,000 and up to NZD 20,000 a day for a continuing breach. No opened source gives a national count of penalties or prosecutions actually issued since 2017, so how much the deadline was worth before it moved cannot be read from any source here.

6. A deferred deadline produces no event. Nothing is demolished, nothing is announced building by building, and nothing changes on any street on the day a deadline moves. The register stays in place with the same buildings on it and a later date beside each one.

What has been tried?

Two of these went in one direction and the rest went in the other. The scheme was built to capture existing buildings that no new-build standard would ever reach, and every step since 2024 has reduced either how long that capture lasts or how many buildings it holds.

What was found?

Why is it still unsolved?

Institutional exemption — the obligation was created, it captured the buildings it was meant to capture, and it is now being narrowed from inside by ordinary legislation rather than being enforced or repealed.

An obligation that lives inside a date can be softened without repealing a single duty, because the duty stays on the page while the year it points at moves. That is what the 2024 Act did to every deadline that had not yet passed. Nothing was withdrawn, no building left the register, and no owner was released. Four years were added to a number, and the entire scheme continued to describe itself in exactly the same words as before.

The second part is that scope can be redefined without anything being repealed. The bill does not say that buildings in low seismic risk areas are exempt from strengthening. It removes the test that put them in scope and redraws the seismic risk map. The projected result is that most of the register leaves it, but the mechanism is a definitional change to a technical threshold, which reaches the public as an amendment to a standard rather than as a decision about several thousand identified buildings.

The third part is that the scheme produces no event when it retreats. A completed strengthening is visible on a street. A deferred deadline is visible nowhere. The register keeps the same entries, the notices stay on the buildings, and the only thing that changes is a date held in a database. So the two changes that have reduced the reach of this scheme since 2024 have each arrived without anything happening that a person walking past one of these buildings could see.

The fourth part is that no opened source records what the deadline was worth in practice. The scheme carries a penalty ceiling of NZD 200,000 and a daily addition for continuing breach, and no opened source records how often either has been used since 2017. An obligation whose enforcement record cannot be read from any source here also cannot be shown to have been weakened, because there is no baseline to compare against.

What observation would mean it is solved?

Candidates — (a) the count of buildings awaiting strengthening or demolition falls toward zero because the work was completed, with completions published (b) statutory deadlines stop moving, so that a deadline set in one year is still the deadline in the next (c) a published national record of enforcement under the scheme exists, showing how many notices were followed, how many were breached and what happened to the breaches.

(a) alone can be produced without any strengthening. The register count can fall because buildings were fixed, because buildings were demolished, or because the definition of what belongs on the register changed. Those three are indistinguishable in the total. A falling count is only evidence of a solved problem if the reason for each departure is published alongside it, and no opened source publishes departures by reason.

(b) alone measures stability and not safety. A deadline can hold still and still be far enough away that nothing happens before it. The 2024 extension moved deadlines that in the low seismic risk band already ran to 35 years from the date of the notice. A stable deadline in the 2050s is stable and is also, for anyone standing in the building this decade, indistinguishable from no deadline.

(c) alone counts enforcement and not exposure. A published enforcement record would settle whether the penalty ceiling is real, which is currently unknown. It would still say nothing about how many people occupy the buildings, because the register does not hold that and no source opened here supplies it.

The three have to be read together, and all three would still leave the same hole underneath. Until some source counts the people rather than the buildings, every observation available here measures the administration of the scheme and not the exposure the scheme exists to reduce.

What is it connected to?

Fills with researchnatural hazard adaptation and relocation policy, the insurance and lending treatment of buildings carrying an earthquake-prone building notice, seismic retrofit obligations for existing buildings in other jurisdictions, and heritage building protection where strengthening and preservation obligations meet. Relation type and evidence grade were not established in this round.

What these sources do not say

• How many people are exposed. Every opened source counts buildings. None gives occupancy, employment or footfall for buildings on the register, and none splits the register by building use, so the exposure the scheme exists to reduce has no published number at any point in this file.
• What the 5,800 figure covers that the breakdown does not. The four projected categories total about 5,260 against a stated 5,800, and the reported share of 55 percent for the 2,900 matches the smaller total rather than the larger one. No opened source says what the remaining 540 or so buildings are, or whether the breakdown was drawn from a different set.
• Which saving figure is correct. One opened source reports about NZD 800 million and another about NZD 8.2 billion for the same redesign. Neither states its basis and neither acknowledges the other.
• Whether anyone outside government checked the projections. No opened source records an independent review of the 2,900 figure or the saving estimate by an audit office, a parliamentary analysis unit or any other body outside the department that produced them. That is an absence in the sources opened here, not a finding that no such review exists.
• How many of the 5,800 are already past their deadline. The sources record buildings as awaiting work and do not divide them into those still inside their statutory period and those beyond it.
• How often the scheme has been enforced. The penalty ceiling appears in the sources as a feature of the law. No opened source gives a national count of penalties issued or prosecutions brought since 2017.
• What individual councils hold. Territorial authority registers were not among the sources that could be read here; the council pages approached returned no usable response, so only national figures are carried.
• The government position in its own words. The official release announcing the redesign is not among the sources opened here, and neither is the government department page carrying the register definitions. The stated reasons — cost, and a shift to a risk-based system — reach this file only through news reports and professional commentary that describe them.
• What happens to a building that leaves the register. No opened source says whether a building removed from scope retains any assessment record, any disclosure obligation to tenants or purchasers, or any route back into the scheme if the seismic risk classification of its area changes again.

See the evidence

No primary legal or government document was read in full. The Building Act provisions, the 2024 amending Act, the bill text and the register itself are described here only through professional and news sources that cite them, and the three rows above with no URL are the primary and official pages that could not be retrieved. That matters most for two things. First, the register figures of more than 8,000 and about 5,800 come from reporting of a government announcement rather than from the register, so no figure here has been checked against the register it describes. Second, the projections of 2,900, 1,440, 840 and 80 are what the government said the bill would do, and they are reproduced as projections because the bill was not law at 2026-08-17. Where sources overlap they agree: the extension of four years and its commencement date appear in two opened sources, the figure of about 5,800 buildings awaiting work appears in two, and the breakdown of 1,440 and 840 and about 80 appears in two. Where they disagree the disagreement is left visible rather than resolved — the reported saving is about NZD 800 million in one opened source and about NZD 8.2 billion in another, and the projected departures from the register are given both as a count of 2,900 and as a share of 55 percent, which do not describe the same quantity out of 5,800. The arithmetic gaps noted in this file are between figures as published and are not adjustments made here. This is a Path A output, research-based definition, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range Not derivable

The reason and what is missing are listed under “What is missing” below