Subnational Population Estimates (2023-base)

Subnational population estimates The estimated resident population for subnational areas by age and sex at 30 June is produced annually. First release of provisional subnational population estimates at 30 June of a reference year generally becomes available in October of the same year. These estimates will remain provisional for 2 years. The release of subnational population estimates for a reference year includes: provisional estimates at 30 June of the reference year, revised provisional estimates at 30 June of the previous year, and final estimates at 30 June for two years previous. Base population The estimated resident population at 30 June 2023 forms the base population for deriving current post-censal population estimates. The estimated resident population of each area at 30 June 2023 is based on the census usually resident population count from the 2023 Census (held 7 March 2023), with adjustments for: people missed or counted more than once by the census (net census undercount), residents temporarily overseas on census night, births, deaths, and net migration between census night and 30 June 2023, reconciliation with demographic estimates at ages 0–14 years. 2023-base territorial authority and Auckland local board (TALB) estimates TALB estimates are derived using a component method. They are based on the latest census data and updated to account for births, deaths, and net migration since the census. The formula used for estimating the resident population in each TALB is: Population (current year) = population (previous year) + births - deaths + net internal migration + net international migration. All components of the formula (on the right-hand side of the equation) are estimates of the number of events that occurred during the current year ended 30 June. The subnational estimates are constrained to agree with the corresponding national estimated resident population at the time of publication. Births/Deaths Birth and death registrations are used to estimate the number of births and deaths that occurred during each June year. There is often a delay between when a birth or death occurs and when it is registered, and this delay is accounted for when estimating birth and death occurrences. Net migration For subnational areas, net migration includes net gains from international migration (flows into and out of New Zealand) and net gains from internal migration (flows between areas in New Zealand) during the June year. Annual international and internal migration flows are estimated as two different processes. Multiple administrative data sources are traditionally used to derive estimates of subnational migration because: there is no direct measure of internal migration in New Zealand other than the periodic Census of Population and Dwellings, the direct measure of international migration (via the outcomes-based migration measure) will use integrated address information from administrative data to inform the geographic distributions of the international migration estimates when estimates are final. International/external migration From October 2019, international migration estimates have been derived using the outcomes-based migration measure (see Outcomes versus intentions: Measuring migration based on travel histories). The revised subnational estimates at 30 June 2013–18 also incorporated the outcomes-based migration measure. International migration data (by age and sex) are used to estimate international migration for all ages. Travellers arriving in or departing from New Zealand are classified as long-term migrants according to the 12/16-month rule: A migrant arrival is an overseas resident who arrives in New Zealand and cumulatively spends 12 out of the next 16 months in New Zealand. A migrant departure is a New Zealand resident who departs New Zealand and cumulatively spends 12 out of the next 16 months out of New Zealand. We use a statistical model to produce provisional migration estimates. Statistics produced using these provisional migration estimates have uncertainty for 16 months; after this time, we can finalise the classification of all border crossings (according to the 12/16-month rule). Once migrant status has been finalised according to the 12/16-month rule, these records can be linked to Stats NZ’s Integrated Data Infrastructure (IDI) to obtain migrant address information. See International travel and migration – DataInfo+ for more details. For provisional subnational population estimates, we estimate subnational international migration by imputing TALB from observed records of the reference year, using country of citizenship, Māori ethnic group indicator, age, sex, and direction as predictors. Historical geographic distributions of final international migration estimates may be used to inform the provisional geographic distributions of international migration. Internal migration The process for estimating annual internal migration to and from TALBs uses primarily linked address notifications data from the IDI combined with an estimation model. Additional data sources are used for evaluating and/or informing the distributions of small domain estimates. These sources include: 2018 and 2023 Census data (specifically, transition data derived from the current usual residence address and address one year ago questions), and primary health organisation (PHO) enrolment data. See Subnational migration estimates for territorial authority and Auckland local board areas at 30 June: Data sources and methods for more details. Internal migration is the most difficult component of subnational population change to measure, because there is no single authoritative data source with full coverage of movements between areas. Some changes in data selection and model specifications may be necessary to improve estimates following a thorough analysis of the latest data available. Summaries of these changes follow. Changes in 2025 The model and software were updated for the Subnational population estimates: At 30 June 2025 (2023-base) released in October 2025. The main timeseries used in the estimation of internal migration remains the address notifications from various administrative data sources in the IDI. PHO data, census data, and historical internal migration estimates were used to inform the estimation model and subsequent adjustments. Impact of 2023 Census integration on address transition volumes Address notifications data in the IDI include address information from administrative data sources and censuses. Following the integration of census addresses into the address notifications data, the internal migration transition data derived for census years, showed a decrease in volume compared to the same data derived prior to integration, especially for 2023. The net effects (differences between inflows and outflows) for most areas, however, were mostly consistent between the data derived in 2024 and 2025. Therefore, we decided to use the latest administrative counts as they are, without adjustments for the drop in volume. This approach prioritises consistency and transparency, while acknowledging the limitations in the data. Summary of changes: The estimation models were fit to a timeseries of annual transitions from 2018 – 2025, and internal migration estimates for 2024 were revised. The 2023-base ERP provided the new population base that was rolled forward, and therefore, the back series for internal migration estimates prior to and including 2023 were not revised. The PHO data series was not directly used in the estimation model, however, analysis of historical differences between PHO (and other data series) and address notifications data informed adjustments made to a few areas with implausible age distributions during the derivation of the estimates. The address notifications data was adjusted for possible under and overcount using a measurement error model based on average historical errors before inclusion in the estimation models. Address notification data is updated with a lag, and while the quality of the data is improving, the data has its challenges as the main source for estimating internal migration. An average of age and sex ratio distributions from 2018 and 2023 Census transition data were used to further disaggregate the TALB internal migration estimates by age group to single year of age and sex. Manual adjustments were made to some areas and ages. Changes in 2024 The method used for estimating internal migration in the Subnational population estimates: At 30 June 2024 (2018-base) released in October 2024 changed, with revisions made back to 2019. Different data selection methods and model specifications were tested to estimate internal migration, while the availability of the 2023 Census subnational population distributions provided a benchmark with which to compare. Main findings of the testing included: The quality of linked administrative data has improved over the year and can be used as the main source to measure internal migration. Some areas still required specific modelling processes to be applied, e.g., Dunedin. Lags in administrative data updates remained the main challenge to produce timely and quality internal migration estimates. Areas with smaller population and sparse data on people movements resulted in larger uncertainties of the estimates. Major interruption was seen if administrative data sources changed their data collection method. For example, in 2019 Primary Health Data (PHO) changed the way they collect address information. This distorted the changes in the derived movement data. Internal migration fluctuates from year to year, and as a result, while some trends can be informed by historical time series, it doesn’t always represent the latest movements that happened. As a result, the method used to estimate internal migration was refined to produce population estimates with subnational distributions similar to the 2023 Census subnational distributions. This methodology was then applied to subnational population estimates back to 2019. Summary of changes: The method of determining a transition happening between TALBs changed. Two data inputs into the models were dropped: 2018 Census usual residence one-year-ago information, and derived net internal migration data from the series of revised subnational population estimates between 2014 – 2018. Shortened time series admin data as data input into models. Reduced adjustments to compensate for implausible estimates in certain areas or at specific ages due to quality limitations in the existing data sources. Refined data model which works best using the adjusted list of input data sources. The changes to the method used for internal migration estimates are still interim due to following reasons: Final evaluation to be done after the 2023-base estimated resident population is available Updates to the R packages used to model the estimates is still to come. These are mainly to improve efficiency of the modelling processes. However, some impact to the modelled estimates are expected. SA2 estimates The SA2 population estimates by age and sex are prepared using a combination of administrative data sources: primary health organisation (PHO) enrolment data, electoral enrolment (EEC) data, and building consents. SA2 estimates are initially a combined estimate using administrative population counts of PHO and EEC data. In some areas, building consents data is used to adjust the estimates. The initial SA2 estimates are adjusted for coverage at the broad age group level based on previous year’s SA2 estimates. The initial set of SA2 estimates by age and sex represent an update of the previous year’s estimates based on the current year’s PHO and EEC coverage adjusted population data including an update for natural increase (births minus deaths). The initial estimates are adjusted to meet the TALB constraint at the age by sex level. SA2 estimates are not produced using a component method due to limitations of estimating the migration flows at the more detailed SA2 level. Rounding All published figures are calculated using data of greater precision. Individual figures may not sum to totals due to rounding. The rounding of estimates of the five-year age groups, broad age groups, and total all ages is determined by the total population size of the geographic area. Five-year age groups For areas with a population less than 2,000, estimates are rounded to the nearest 5. For areas with a population of 2,000 and more, estimates are rounded to the nearest 10. Broad age groups and total all ages For areas with a population less than 10,000, estimates are rounded to the nearest 10. For areas with a population between 10,000 and 20,000, estimates are rounded to the nearest 50. For areas with a population of 20,000 or more, estimates are rounded to the nearest 100. The rounding of estimates of natural increase, net migration, and population change is determined by the size of components. For components have an absolute value less than 1,000, estimates are rounded to the nearest 10. For components have an absolute value of 1,000 or more, estimates are rounded to the nearest 100. en-NZ