Annotation: Stats NZ OIA (Death statistics, 2015–2025) — what was answered, what was refused, what is still pending

Stats NZ has now issued its substantive response to the parts it retained, and earlier advised that parts 3, 4, and 7 were transferred to Health NZ – Te Whatu Ora because those parts were more closely connected to Health NZ’s functions. 

What Stats NZ did provide (practically)

For several items (e.g., annual deaths, age-standardised mortality rates, weekly deaths, and age-banded mortality counts), Stats NZ largely pointed to existing public sources (notably Infoshare and their births/deaths releases) and described how weekly registered deaths data is published (provisional, quarterly updates). 

This is a common approach: agencies will often treat “already publicly available” material as “provided” by directing requesters to the authoritative source, rather than exporting custom spreadsheets.

What Stats NZ refused (and why it matters)

Stats NZ refused the “excess mortality / expected deaths / baseline methodology” components under OIA s18(g) on the basis that the requested information is not held by Stats NZ (they state they do not measure excess deaths or maintain baseline methodologies/technical documentation for that purpose). 

This creates a clear “analytical gap” for the public:
• NZ has robust mortality counts and rates available,
• but official “expected vs observed / excess mortality” analysis is treated as outside Stats NZ’s holdings, and therefore must be sought from whichever agency (or cross-agency process) is responsible for that modelling—if it exists.

Is this OIA “complete” yet?

Not fully. The request is still live in substance because Health NZ has not yet delivered the transferred parts (3, 4, and 7). Stats NZ’s portion is complete; the overall request is not. 

What to watch next (without assuming bad faith)

When Health NZ replies, the key questions are:
1. Do they hold cause-of-death / ICD-coded outputs at the level requested (and what lags/limitations apply)?
2. Do they hold any internal analysis that compares mortality patterns pre-2020 vs post-2020, or is that also “not held”?
3. If they refuse as “not held”, do they identify who does hold the relevant modelling (or confirm it does not exist in NZ government)?

Comparable FYI requests (pattern-spotting, not allegations)

There are numerous FYI requests seeking mortality analytics, “excess deaths,” or “expected deaths” models. Results commonly include:
• directing to public data portals,
• “not held” refusals for modelling/baselines,
• partial transfers across agencies where holdings are split.

(If you want, I can maintain a rolling “related FYI requests” list under this annotation as new ones surface.)

International comparison (governance models)

In multiple jurisdictions, “excess deaths” is treated as a defined statistical product supported by published methods:
• US CDC / NCHS publishes excess-death estimates using modelling approaches (baseline/expected deaths with uncertainty), described in peer-reviewed/official methodology write-ups. 
• UK ONS routinely publishes weekly deaths and excess-death style comparisons with transparent statistical framing and documentation in its bulletins/method notes. 

The key governance distinction is: some systems treat excess-mortality as an owned, documented, routinely-published product (with defined methods and revision policy), while NZ’s response here suggests that, at least for Stats NZ, it is not treated as an internal “held” output in the way requesters often assume. 

Status: Stats NZ response received; Health NZ response pending (transferred parts).