Stats NZ is consulting on an infringement scheme that would let it issue $400 fines to individuals who refuse to respond to mandatory data requests. The agency frames this as a proportionate middle ground between doing nothing and criminal prosecution, which can already result in penalties of up to $2,000 for individuals and $12,000 for businesses under the Data and Statistics Act 2022.
General Manager Policy Tennille Maxey says expanding the enforcement toolkit will “strengthen data collection and support high-quality official statistics.” General Manager Data Collections Lorna Curran adds that fines would only apply to those who “actively refuse to provide information,” not people unable to participate due to illness or bereavement.
That is the official line. The more revealing question is why the agency needs this tool now, and the answer has nothing to do with a few stubborn survey holdouts.
Trust collapsed before participation did
Newsroom’s analysis of NZ data credibility identifies the core dynamic: “trustworthy and accurate data relies on people’s cooperation with the government employees in charge of these numbers”, and that cooperation is draining away in a society where trust in government is dropping and younger generations avoid answering both the door and their phones.
This is not abstract. Stats NZ’s employment surveys cover roughly 30,000 people every four months, each surveyed for eight consecutive quarters. When participation falls, the data skews. Professor Thomas Lumley of the University of Auckland warns that self-selecting respondents are “very unrepresentative” and flags an emerging threat: “In the future, there’s the potential for LLM-based fake people to take surveys for the money”.
Lumley makes a particularly sharp observation about how errors get caught: “The reason we found out the report was inaccurate was precisely that other evidence didn’t point the same way.” The implication is chilling. Errors that happen to align with expectations may never be detected at all.
A smaller agency making bigger mistakes
Stats NZ has shed around 300 staff following a seven percent budget cut. Government Statistician Colin Lynch insists quality has not suffered, but the record disagrees. Multiple data failures have occurred in quick succession, including bungled food price index figures and problems with fuel shipment data at MBIE.
Former Government Statistician Len Cook has criticised the agency’s website as “almost impenetrable for an ordinary person to use”. Lynch’s response, that the site is being rebuilt to accommodate AI-driven data consumption as fewer people visit directly, reads less like a strategy and more like a rationalisation of declining engagement.
Now layer on the structural shift. The Government has announced plans to abandon the five-yearly census in favour of administrative data and annual surveys. The cost argument is real: the 2023 census cost $325 million, up from $104 million in 2013, more than double the per-capita cost. But researcher Paul Spoonley argues the decision was made “without any apparent consideration of how the census is used” and what will be lost.
What will be lost is longitudinal depth. As Spoonley notes, “one of the primary functions of a census is to allow comparison with previous censuses over time”. Administrative data can tell you what is happening now. It cannot easily tell you how things have changed across decades.
Every business decision sits downstream of this data
For business owners, the $400 fine is almost beside the point. Employment figures, inflation indices, demographic projections, regional population data: all of these flow from Stats NZ’s surveys and census. Businesses use them for workforce planning, market sizing, pricing models, and investment cases. Lenders use them to assess risk. Government uses them to set policy that directly affects operating costs.
The US offers a preview of where this leads. Falling labour survey participation created a weeks-long crisis over employment data reliability that rattled financial markets and complicated monetary policy. New Zealand is a smaller, less diversified economy with fewer redundant data sources. The margin for error is thinner.
When that data degrades, the cost is invisible right up until it is catastrophic. A workforce projection that understates regional population growth leads to under-investment. An inflation measure that misses price movements leads to mispriced contracts. A demographic forecast built on administrative data with no census benchmark leads to infrastructure planned for a population that does not exist in the way the numbers suggest.
Stats NZ is reaching for the stick because the carrot stopped working. But fining people into compliance does not fix an agency that has been cut, has made visible errors, and is now losing the census that anchored its entire data architecture. The infringement scheme is a symptom. The disease is an institution being hollowed out while every business in the country still depends on its output.
Sources
- Stats NZ opens public consultation on infringement scheme
- Our approach to survey compliance
- What experts say about NZ’s stats credibility amid US jobs data meltdown
- Dangers of opt-in surveys
- Collecting statistics: When the numbers don’t add up
- Govt’s census shake-up advances despite ‘unreliable’ data concerns
- Data gaps and demographic change: the end of the NZ census will create big blind spots