Statistics New Zealand’s approach to the 2030 Census
Statistics New Zealand (StatsNZ) is consulting on a major change to the census: moving from a five-year full count to an administrative data-based model supported by an annual survey.
StatsNZ announced in June 2025 that it will move from a five-year full enumeration census to an administrative data-based census complemented by an annual sample survey. This decision was made on the basis that, on balance, the lower costs and more timely data availability outweigh the risk of lower quality and less granular data from the census.
The great value of a full enumeration population census is in the breadth and depth of information that such a census gathers for all individuals and households in New Zealand at a common point in time. This allows for highly detailed granular analysis of population groups, regardless of their size.
Public consultation on the new census approach closes 19 December
As part of StatsNZ’s process for moving from a five-yearly full enumeration survey to an administrative data-based census complemented by an annual sample survey, it is currently consulting on its proposed data collection approach and the content for the annual sample survey.
Those wanting to participate have until the 19 December 2025 to have their say. More information on this can be found here
StatsNZ plans to replace the full census with admin data and annual surveys
The new census annual sample survey aims to survey 60,000 households. Its purpose is to provide reliable national and regional data from a single year of data, complemented with administrative sources.
To provide reliable data for areas the size of local authorities, for example, Wellington City or Selwyn District, StatsNZ will need to pool three years of survey data. This is partly why the annual census survey will start in 2027, and results will be available from 2030, allowing StatsNZ to gather three years of data.
Smaller communities risk losing detailed data under the new model
For smaller communities of 5,000 to 9,000 individuals, five years of data will need to be pooled together. However, StatsNZ is not contemplating the pooling of six or more years of data together. Therefore, it is likely that data from the annual survey will be unable to be published for population groups or geographic areas under this limit of 5,000 people.
This means that if data cannot be gathered via administrative data for these groups, information will not be collected or published for these smaller populations.
Will cost savings outweigh the loss of granular insights?
StatsNZ’s discussion document also clarifies the limitations of the annual census sample survey in providing the granular information that is not available from administrative data.
Of the information collected from the new census, 23 data variables will be collected via the annual census survey only, while another 33 data variables will be collected via both the census survey and administrative data sources. Only 13 data variables are expected to be collected solely via administrative datasets.
This means that the potential information loss, for population groups under 5,000 individuals, is around 40 percent of the information expected to be collected through the annual census sample survey. With a large number of data variables still requiring collection through the annual survey, and with this limited to a sample of the population, some large-scale analysis of individuals’ responses may be at risk under the new approach to the census.
Can government agencies supplying administrative data fill the gap?
The discussion also outlines the additional data that StatsNZ is seeking to gather from other government agencies. This raises the issue of whether or not other government agencies can gather or supply this additional information, given the current requirement that government agencies do not gather any more personal information from the public than they require for their purposes.
At present, most of the 23 data variables being gathered solely from the census survey are unable to be gathered via administrative sources. This is because these variables concern subjective personal information such as sexual identity, smoking and vaping, religion, languages spoken, transgender status, disability indicators, and unpaid activities.
Improvements over time can be made regarding the 33 data variables gathered via both the survey and administrative sources. These improvements in data collection could reduce the need to gather data via the survey.
Do the efficiency and frequency benefits outweigh the potential loss of information?
In light of the clarifications and details from the discussion document, we need to ask whether the benefits of lower costs and more timely data availability outweigh the risks in terms of the loss in the quality and granularity of data from the census, and thus the potential loss of our ability to use this data to the benefit of the country.