The census is an official count of people and dwellings in New Zealand. It provides information on how households and families are structured, changes in cultural diversity, how people live and travel, and the qualifications that they hold.

The Data and Statistics Act requires the Government Statistician to take a census of the population and dwellings in every fifth year. The last census was taken in 2023. The current full field enumeration (survey-first) model requires everyone to complete a census survey.

Work to modernise the census has been under way for more than a decade. Statistics New Zealand has been exploring how to provide the census more effectively by relying more on administrative data—that is, information already collected by government agencies. Cabinet first considered this direction in 2012, and in 2015 directed further exploration and development of an admin-data-first approach.

The new approach is considered necessary to respond to a decline in data quality, affordability, efficiency, and timeliness under the traditional model. Over the last three censuses, Stats NZ has faced difficulties with the survey-first model, exacerbated by interruptions due to natural disasters. In May 2025, Cabinet endorsed the Government Statistician’s preferred approach to modernise the census by using an admin-data-first model.

Under the new approach, future censuses will be taken using admin data supported by survey data. Data collection will be ongoing and will involve:

• admin data on key census variables using more of the information already collected by government agencies

• a new annual census survey beginning in 2027 building over time to achieve a sample size of 60,000 randomly selected households

• tailored solutions to meet the needs of Māori and harder-to-measure communities.

From 2030, statistics produced under the new census model would be published and available annually, rather than five-yearly.

The Data and Statistics (Census) Amendment Bill is an omnibus bill as it amends more than one Act. Its amendments can be regarded as implementing a single broad policy, namely that there will be an annual census from 2030.

Although the Government Statistician has discretion to decide how they will take the census, changes to the Data and Statistics Act are necessary to support an orderly move to an admin-data-first census model. The bill would amend the Act to:

• require the Statistician to publish, in each financial year, statistics from the census of population and dwellings based on a specified reference date in the previous financial year so that census data is published annually

• move the next census from 2028 to 2030 to allow time to prepare for the change in census approach

• update requirements regarding Māori engagement, public consultation, and post-census review to align with an annual census

• remove the census-specific enforcement and infringement provisions that support compulsory participation in the five-yearly census survey because these will not be required under an admin-data-first census.

The bill would also amend legislation that relies on census or population data. It would update references to census concepts and population data so the following legislation would continue to operate under an annual census model:

• Local Electoral Act 2001

• Waste Minimisation Act 2008

• Bay of Plenty Regional Council (Maori Constituency Empowering) Act 2001

• Wellington Regional Water Board Act 1972.

We considered the bill alongside the Electoral (District Boundaries) Amendment Bill (on which we are reporting separately). Census data is used by the Statistician to determine the number of general and Māori electorates (including the electoral population in each district), and by the Representation Commission to review and adjust electoral boundaries. At present, each five-yearly census triggers a boundary division. Without legislative change, shifting to an annual census would trigger an annual boundary division process, which would be impractical and unworkable. The Electoral (District Boundaries) Amendment Bill would amend the Electoral Act 1993 to decouple future boundary divisions from the census. That bill provides that electoral boundary divisions would instead be triggered every second electoral cycle.

Local electoral boundaries are reviewed every six years. Some councils will next review their boundaries in 2027 and others will in 2030. For historical reasons relating to the availability of census data, the reviews are often conducted based on the Statistician’s population estimates rather than census data. The Local Electoral Act enables either population estimates or census data to be used.

Clause 19 would amend the definition of “assessment date” in section 5 of the Local Electoral Act. As introduced, it provides that an assessment date could only be used that was after both:

• the date on which the Statistician provided the most recent electoral population figures to the Representation Commission for national electoral purposes; and

• the reference date of the most recently released census.

We understand that this definition is intended to ensure that local electoral boundary reviews are based on up-to-date population data. However, we consider that the date the electoral population figures were provided to the Representation Commission for national electoral purposes adds nothing, and the definition could be simplified by removing it. We recommend amending clause 19 to this effect.

We note that population estimates are often assessed as of 30 June in any given year. The bill would also set 30 June as the default reference date for the census. If the assessment date was defined as being after the reference date, it would effectively prevent the use of population estimates for local electoral boundary reviews. Instead, only census counts could be used. This would effectively remove the discretion to use either population estimates or census data, which was not intended. We therefore recommend amending clause 19 to provide that the assessment date would be the date on or after the reference date of the most recently published census.

Submitters generally framed the request for a review as a need for an independent assessment to test whether the preconditions for high-quality statistical outputs are in place. “Independence” was variously described as being independent from Stats NZ or the Government Statistician, or politically independent. Submitters considered that a review would test whether New Zealand was ready to transition, and would check the main assumptions underpinning the proposed model. It could also provide more transparency and assurance to data users and the public about methodologies and data quality. Suggested criteria for the review included Māori representation and expertise on Māori data governance and sovereignty, and consideration of alternative census approaches. An example of an alternative was adopting a 10-yearly field census with continuous population data from administrative sources.

Other views expressed about a review included:

• concerns about the process leading up to the Statistician recommending a shift to the new model

• disappointment about the amount and quality of Stats NZ’s engagement with the public, Māori, and data users and suppliers, which should be remedied through a review

• a request for the Justice Committee to commission an independent study of the feasibility and consequences of undertaking an admin-data-first census.

Some submitters considered that independent oversight could help maintain public confidence and enhance transparency. Their suggestions included establishing an independent board to oversee administrative data use, and requiring regular public reporting on a range of data-related matters. One submitter proposed that the bill should be amended to require regular independent reporting to Parliament about the quality, completeness, and fitness for purpose of the administrative data tabled annually.

We acknowledge submitters’ requests for an independent review of the new approach and independent oversight of the use of administrative data. We are concerned about some submitters’ lack of trust in the shift to an annual census based on admin data. We discussed how to build trust in the new approach, as well as provide oversight and assurance. We considered possible amendments including requiring a statutory review in the legislation or an external review committee.

We were advised that Stats NZ was considering options for a potential independent external review, and we sought more information as to what this would entail.

Stats NZ gave examples of several established assurance processes, which assess data quality and census operation after the census has been delivered. The processes for the 2018 and 2023 censuses included an independent advisory panel of external experts who reviewed the output data quality of selected variables after the census. Section 37 of the Act also currently requires the Statistician to conduct a review of the operation of each census and present a report on that review to the Minister of Statistics.

In addition to the existing design and assurance processes, the Government Statistician intends to establish an external technical assurance body. This External Methodology Advisory Panel will undertake an external, formal assessment of:

• the quality of integrated admin data

• the readiness of the data to meet the population and dwelling measurement requirements of the census

• the elements of the methodology that underpin the new census model.

The panel’s first report will be provided within six months of establishment, which will allow enough time to inform the planning and design of the new census. The panel will add a forward-looking focus, and aims to provide timely, practical input to improve the development of the next census.

Under the Data and Statistics Act, the Statistician is responsible for determining how official statistics are produced, including the choice of census methods. This responsibility is intended to support trust in statistics by keeping decisions independent from political and other undue influence. The advisory panel will therefore not revisit the Statistician’s decision to move to a new census model.

Stats NZ intends the panel to continue beyond the initial formal assessment, advising the Statistician as needed after its first report. It will be independent from Stats NZ and provide advice to the Statistician. The panel will include local and international experts and be transparent, with membership, terms of reference, and findings all made public. Stats NZ has tested the concept with an international statistical expert who concluded that the panel is fit for purpose. We believe that it is absolutely essential that this panel comprise people independent of Stats NZ who can provide expert advice and oversight on this process to ensure that this change is delivered with confidence and within the projected time frames and cost forecast.

Stats NZ also highlighted a range of existing processes and channels that incorporate diverse perspectives and provide additional assurance and transparency. They include data advisory groups, a community design forum, gateway reviews, privacy impact assessments, and publication of methods papers.

We note that the bill would maintain the requirement for the Statistician to periodically review the operation of the census and to present a report about the review to the Minister of Statistics. The bill would initially provide for more frequent reviews, with reviews required after the 2030 and 2033 Censuses, and then at least every five years. Stats NZ also remains accountable to Parliament through annual reviews, Estimates examinations, and corporate reporting, including annual reports and statements of strategic intentions.

We acknowledge Stats NZ’s existing assurance processes and accountabilities to Parliament. However, to help address submitters’ accountability concerns, we urge the External Methodology Advisory Panel to report directly to the select committee that is responsible for census-related matters in the next Parliament. We would expect this reporting to occur quarterly. For the avoidance of doubt, we note that this would not confer decision-making powers about the census to the committee.

• Loss of small area (meshblock) statistics

• Loss of family, household and housing statistics

• Loss of information about occupation, and workplace connection with residence

• Major discontinuity between information from past censuses, and those of administrative data-first model so trends are difficult to ascertain.

This has required the Government Statistician to construct unvalidated new metrics for the Electoral Act. The Electoral Commission has not received any independent statistical advice on measures which could be subject to a challenge by judicial review.

“Engagement” with Māori every three years is not true partnership, as defined in the principles of te Tiriti o Waitangi. Additionally, the bill does nothing to afford data sovereignty to Māori, or give Māori tino rangatiratanga over their data, which is a taonga predominantly held by the Crown.

Te Mana Raraunga (The Māori Data Sovereignty Network) brings together over 800 Māori researchers, practitioners, and entrepreneurs. Their submission acknowledges potential benefits but they are not ready to throw away the in-person census “We have serious concerns about whether this approach will be effective, and there are still large gaps in the census plan. In our view this Bill is premature – much more work and assurances are needed – and could create long lasting harm to our data and statistics system, electoral system, and to Māori.”

Retail NZ’s submission states “We are concerned that administrative data will be insufficient to produce quality iwi affiliation data following the delay in release of the 2018 Census iwi data. This data is vital for understanding New Zealand’s unique demographic landscape.”

We also agree with the submissions made by academics across the country that the rich data the physical census gives would be lost in an abstracted administrative census. We also note that countries that have abandoned a physical census such as Israel, Sweden, and Netherlands, have compulsory registration systems that provide the administrative records underpinning population statistics.

Finally, it remains to be seen as to whether democratic systems can rely on agencies to produce reliable data, given the failure of MBIE’s biometric border security project. The data-driven census requires coordination in combining data from NZTA, MBIE, Ministry of Education, Ministry of Health and several more into the Integrated Data Infrastructure. New Zealanders need to be confident of sufficient protection of personal data, particularly in light of recent data privacy breaches.

The Green Party recommends that at least one more 2028 physical census be undertaken to ground-truth data provided by different government agencies.

Submitters consistently told the committee that the public was given too little opportunity to engage with a reform of this scale. Mr Len Cook, a former Government Statistician, noted that recent census reviews were conducted largely within the public sector, and that this select committee process has been one of the few real opportunities for experts and the wider public to test the proposals. Research Association New Zealand made a similar point on behalf of the research and insights sector. Labour shares the view that the consultation undertaken by departmental officials, constrained by the timeframes they were given, has not been adequate for a decision of this importance. The same concern applies to the Electoral (District Boundaries) Amendment Bill, where the consequences for the timing and work of the Representation Commission have not been sufficiently explained or consulted on. We are sympathetic to the proposal of an independent review of the proposed changes, and that review being made available for further public scrutiny.

The central question is whether the enactment of an administrative data-first model as proposed by this bill can produce statistics that are accurate and reliable enough to replace the census. Submitters were not confident that it can. Numerous submitters cautioned that administrative records collected by different agencies are not held to consistent standards and can be out of date. They warned that the quality of information for small areas and for smaller population groups is likely to suffer, and that the continuity of statistics New Zealand has relied on for generations could be broken.

We are particularly concerned about the effect this could have on Māori, Pasifika, rainbow communities, disabled people and hard-to-reach rural communities. We concur with the submission from Te Kāhui Raraunga that Statistics New Zealand’s proposed tailored solutions to address these issues rely on trust, and that they are not a sufficiently certain basis for a decision of this importance. We also note that officials have themselves acknowledged that accuracy and coverage would reduce in the early years of the new model. Until there is far greater certainty on these points, we do not accept that the case for ending the census has been made.

A recurring theme across submissions was the disproportionate impact these changes may have on Māori. The census is a key source of the data that supports Māori development and the fair allocation of resources, and it underpins the calculation of the Māori electoral population. Submitters, including Māori organisations, were concerned that a model that is less accurate for communities that are less easily reached risks undercounting Māori in a way that is harder to see and therefore harder to challenge. They also emphasised that a change of this nature should be developed in genuine partnership with Māori, consistent with Te Tiriti o Waitangi and the principles of Māori data sovereignty, rather than through limited consultation. Labour considers these concerns have not been adequately addressed.