Pharmac may not exceed the NZD 1.806 billion the New Zealand government set for medicines in 2026/27 — and in a consultation that closed on 2025-11-17 Pharmac proposed declining the lowest-ranked funding applications that had waited more than two years

What is happening?

New Zealand buys publicly funded medicines through one agency, Pharmac, and the amount that agency may spend is fixed before it decides anything. The Combined Pharmaceutical Budget is a separate appropriation set by government, and Pharmac carries a duty to ensure that spending does not exceed the amount set. The appropriation has grown, from NZD 795 million in 2013/14 to NZD 1.806 billion in 2026/27, but each annual figure is a ceiling rather than a forecast.

Measured against comparable countries the ceiling is low. New Zealand spent 4.9 percent of public health expenditure on medicines in 2025 while the average across 13 comparator OECD countries was 13.3 percent, in an analysis produced by an economic research institute for a pharmaceutical industry body and released on 2025-10-23.

The visible result is a queue of medicines that Pharmac has already assessed and already ranked as worth buying. The Options for Investment list held 123 medicines as of November 2025, and the average time a medicine had already spent on it was 5.9 years. Nothing on that list is waiting for evidence, for a clinical opinion or for a safety decision. Each item has been through the assessment and stopped at the budget line.

The gap this produces is measurable against Australia. Since 2011 Australia has publicly funded 215 new medicines and New Zealand has funded 86, leaving 142 medicines funded across the Tasman and not here, of which about 38 percent are cancer medicines. Average time from application to listing was 556 days in Australia against 1,050 days in New Zealand, in a report prepared by a health data analytics firm for the same industry body and published on 2025-10-03.

In October 2025 Pharmac proposed to shorten the queue rather than fund it. A consultation opened on 2025-10-20 and closed on 2025-11-17 on declining applications that sit in the bottom 20 percent of the ranking, or the bottom 10 percent where fewer than 100 applications are in play, and that have stayed there for more than two years.

Whose problem is this?

The agency that ranks the medicines is not the body that sets the amount available to buy them, and the body that sets the amount does not publish a target for how much of the ranked queue it intends to reach. Between those two facts sits the entire problem.

Where does this problem end?

Four things are outside this frame for a specific reason rather than by omission. Medicine safety and market authorisation sit with Medsafe and are not the bottleneck here, since most of the 142 medicines identified in the Australian comparison were already approved by a regulator somewhere. Medical devices used in hospitals fall outside the Combined Pharmaceutical Budget and are funded separately. Pharmacy dispensing and consultation fees are a separate Health New Zealand budget line. The Exceptional Circumstances pathway, which allows individual applications outside the funded schedule, exists but was not sized in this round and does not substitute for a funding decision that applies to everyone with the condition.

The boundary here is not whether these medicines work and not whether they were assessed, but whether an assessment that says yes can reach a budget line that is already full.

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

Now

Needs a new measurementthe target state: no source opened for this document gives an official target, from government or from Pharmac, for the medicines share of health spending, for the length of the queue, or for how long a ranked medicine should wait. The only stated destination comes from one party in the governing coalition, which in July 2026 published a policy to raise the medicines share by one percentage point a year until it reaches 12 percent of the health budget in 2033. At the time it was announced that policy had not been formally put to the coalition partners, and in the 2026 Budget Pharmac received NZD 13.5 million out of a NZD 1.5 billion health increase. A party policy that has not been proposed to the government it belongs to is not a target, and no opened source supplies one.

How big is it?

The queue is counted in medicines rather than in people. As of November 2025 the Options for Investment list held 123 medicines with an average wait of 5.9 years, and by July 2026 the volume of applications had reached 190, covering roughly 100 medicines, with the average ranking assessment taking 43.8 months against a five-year average of 31.1 months. Measured against Australia the shortfall since 2011 is 142 medicines, of which about 38 percent treat cancer. New Zealand has a population of about 5.33 million.

Two published patient figures exist and both answer a different question. When the 2022 Budget added NZD 191 million over two years, the resulting 20 new listings and 22 widened listings were estimated to reach about 365,000 people. When the 2024 Budget added NZD 604 million over four years for cancer and other medicines, the resulting 54 new or widened listings were estimated to reach about 175,000 people. In the third quarter of the 2024/25 financial year Pharmac recorded 64 investment decisions estimated to benefit about 79,131 patients.

The countable quantity here is medicines rather than people, and the two are not interchangeable because one medicine can serve a handful of patients or tens of thousands.

Needs a new measurementthe number of people waiting. Every patient figure above counts people newly reached when the ceiling moved up, and none counts people standing behind it. The affected population is recorded as not derivable for that reason, and the arithmetic that would appear to close the gap is worse than leaving it open: dividing 175,000 by 54 medicines and multiplying by 123 assumes that the medicines Pharmac chose to fund and the medicines it could not fund serve populations of the same size, when the ranking is built to fund the larger health gain first. The same person can also be eligible for more than one waiting medicine, so the product would double count by an unknown amount.

Under what conditions does it arise?

1. The budget is a ceiling and not a consequence. Pharmac assesses applications and ranks them, and the amount available to act on that ranking is set separately and in advance. An assessment that concludes a medicine is worth buying therefore changes the order of the queue and not the length of it. A budget ceiling set before the assessments are read cannot respond to what the assessments find.

2. The queue has no expiry and no clock. A ranked application can remain unfunded indefinitely without any decision being taken about it. Average time on the list stood at 5.9 years in November 2025, and the average time merely to complete a ranking assessment rose from a five-year average of 31.1 months to 43.8 months by July 2026. Nothing in the process converts elapsed time into pressure.

3. Growth in the appropriation is absorbed before it reaches the queue. The budget rose from NZD 795 million to NZD 1.806 billion across thirteen years, and both large one-off increases in that period were tied to particular listings rather than to the standing ceiling. The 2026 Budget directed NZD 13.5 million of a NZD 1.5 billion health increase to Pharmac.

4. The cost of not funding is invisible in the accounts. Money not spent on a medicine appears as a saving in the year it is not spent. The illness that follows appears later, in a different budget line, attributed to something else, and no source opened here attempts that calculation for New Zealand.

5. Shortening the queue is administratively easier than funding it. In October 2025 Pharmac consulted on declining the lowest-ranked applications that had sat in place for more than two years. A separate process running since 2019 had already consulted on about 600 inactive applications, of which more than 500 were declined. Removing an application changes the visible size of the problem without changing anyone who has the condition.

What has been tried?

Three directions have been tried and only one of them touches the mechanism. Two large injections of money bought listings without changing the rule that produces the queue, one structural review produced recommendations that were partly implemented and partly set aside, and the most recent operational change reduces the recorded length of the queue rather than the number of people in it. The proposal that would change the mechanism is a party policy that has not yet been put to the government that would have to adopt it.

What was found?

Why is it still unsolved?

Cost structure — the failure is not in the assessment and not in the payment, it is in the shape of the appropriation that the assessment feeds into.

A demand-driven appropriation and a capped one behave identically while the cap is above demand and diverge permanently once it is below. New Zealand has been on the lower side of that divergence long enough for the divergence to become the normal state. Pharmac does what it is designed to do, which is to buy the most health available for a fixed sum, and it does it well enough that its ranking is treated as authoritative. The consequence is that a medicine can be publicly certified as good value by the agency that would buy it and remain unbought for years, without anyone having made a decision to refuse it. Refusal would be visible and reviewable. A ceiling is neither.

The second part is that the queue absorbs the pressure that would otherwise reach the budget. As long as an application can remain on the Options for Investment list, no one has to say no to it. The list functions as a place where an unfunded decision can be stored indefinitely at no political cost, and the longer the average wait becomes the more clearly it is doing that. When the list itself became embarrassing in 2025, the proposal that followed was to remove entries from it rather than to fund them, which is the logical move for an organisation that controls the list and does not control the budget.

The third part is that every partial remedy has been structured to avoid the mechanism. Two budget rounds added money attached to named packages of medicines, which produces a headline count of listings and a headline count of people reached, and leaves the recurring ceiling to resume its ordinary behaviour the following year. That is a rational way to spend money if the goal is a visible result. It is not a change to the rule that produces the queue, and after each of those rounds the queue was still there.

A queue that grows without a deadline looks the same from outside whether it is being worked through or simply held.

What observation would mean it is solved?

Candidates — (a) the medicines share of public health expenditure rises toward the comparator average and stays there across more than one budget cycle (b) the Options for Investment list shrinks through funding decisions rather than through declines (c) the average time from application to listing falls toward the Australian figure of 556 days.

(a) alone is a ratio and ratios move for two reasons. The share can rise because the medicines budget grew or because total health expenditure fell, and the second of those is not an improvement. This document has already had to record that two opened sources give the share as 4.9 percent and about 6.5 percent without either stating its denominator, which is a warning about how easily this particular number moves without anything happening.

(b) alone can be produced by the proposal already on the table. Declining the bottom fifth of the ranking would shorten the list immediately, and by exactly the mechanism that this document identifies as the failure. The observation only means anything if the reduction is decomposed into medicines funded and medicines declined, and no source opened here publishes that split.

(c) alone measures the medicines that made it. Average time to listing is computed over applications that reached a listing, so it is silent about the ones that never did. A queue can post an improving average while the items with the longest waits are removed from it. The 5.9 year average on the current list and the 1,050 day average to listing are not measuring the same population, and no opened source reconciles them.

The three have to be read together, and each of them has to be read against whether the number of people actually receiving a previously unfunded medicine went up.

What is it connected to?

Fills with researchthe funding mechanism for medical devices and diagnostics outside the Combined Pharmaceutical Budget, the workforce and hospital capacity constraints that determine whether a funded medicine is actually administered, the health equity gradient in access for Māori and Pacific populations, the Exceptional Circumstances pathway as an individual route around an unfunded schedule, and the same capped-appropriation structure wherever else it is used to buy medicines. Relation type and evidence grade were not confirmed in this round.

What these sources do not say

• How many people are waiting. No source opened here answers how many people in New Zealand would be clinically eligible for a medicine that Pharmac has assessed as cost-effective and has not funded. The figures that exist count people reached by a funding decision, never people standing behind one. This reads less like a gap in the reporting than a gap in the instrumentation, because a national indicator of unmet eligibility does not appear to be published by anyone.
• Which medicines are on the list. No opened source names the 123 entries on the Options for Investment list or the 142 medicines funded in Australia and not here, beyond the statement that about 38 percent of the second group treat cancer. Without the names, the conditions affected cannot be listed and the people affected cannot be counted even in principle.
• What the 4.9 percent is a percentage of. One 2025 analysis gives 4.9 percent and one 2024 account gives about 6.5 percent. Neither states whether the denominator is total public health expenditure or a particular appropriation, and neither acknowledges the other, so this document reports the more recent figure and records the disagreement rather than resolving it.
• Whether the two waiting-time figures describe the same thing. The 5.9 year average is measured across medicines still on the list, and the 1,050 day average is measured across medicines that reached a listing. One counts the unfinished and the other counts the finished. No opened source states this, and read carelessly the two look like a contradiction rather than two different measurements.
• What happened to the October 2025 consultation. It closed on 2025-11-17. Whether the proposal was adopted, modified or dropped is absent from every source opened here, which means this document cannot say whether the lowest-ranked applications have since been declined.
• What the primary documents say in their own words. The commissioned economic research report file and a Treasury document on the Combined Pharmaceutical Budget both returned successfully to the research tools and neither could have its text extracted, so every figure attributed to them here is taken from a page that cites them.
• How the Australian scheme actually works. This round confirmed the outcome difference between the two countries and did not open an Australian primary source on the mechanism behind it. The claim that the Australian scheme is demand-driven and uncapped is therefore recorded as unverified, and no number in this document rests on it. The financial sustainability debate around that scheme was also not examined.
• What the Exceptional Circumstances pathway handles. Application volumes and approval rates for the individual-approval route were not searched in this round. That is an absence of investigation rather than an absence of data.

See the evidence

No primary source was read in full. Four Pharmac pages were opened directly and supply the statutory position, the annual budget figures, the consultation terms and the status of the 2022 review recommendations. Everything else comes from commissioned reports read through their publishing pages, from news and trade outlets, and from academic commentary. Two files that would have been primary evidence, the commissioned economic research report and a Treasury document on the medicines appropriation, returned successfully and could not have their text extracted, so the figures attributed to the first are taken from the release that commissioned it and nothing is attributed to the second. Where sources overlap they agree: the 2026 Budget split of NZD 13.5 million against NZD 1.5 billion appears in two opened outlets, and the Australia comparison figures appear in the report page and in coverage of it. Where they disagree the disagreement is left visible rather than resolved, in the medicines share of health spending given as 4.9 percent by one source and about 6.5 percent by another with neither stating its denominator. Both reports that supply the comparison figures were commissioned by an industry body whose members sell the medicines in question, which is stated here rather than corrected for, since no independent restatement of either figure was found in this round. 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