Stats NZ reported 3,440 new dwelling units consented in August 2026, in its release dated 1 October. The observed monthly total comprised 1,704 stand-alone houses, 1,455 townhouses, flats and units, 153 retirement-village units, and 128 apartments. These are units approved through consents, not completed homes, construction starts, or the number of consent documents. One consent can cover a multi-unit project. Stats NZ says monthly totals can move sharply when large townhouse or apartment projects are timed into a particular month. This composition describes approvals by dwelling type, not the eventual mix of homes delivered.

A separate seasonally adjusted measure put August at 3,525 dwellings, 5.6% above July after a 4.5% decline in July. It differs from the 3,440 actual count because adjustment removes estimated seasonal and calendar effects to make adjacent months more comparable. It is not another count of permits issued. The adjusted series is re-estimated when observations arrive and can be revised, especially at the recent end. Stats NZ also reported that the seasonally adjusted number of stand-alone houses rose 2.8% after falling 0.1% in July. That is a separate type-specific series, not the movement in all dwelling units.

In the year ended August, the actual total was 41,268 new dwellings, 21% above the year ended August 2025. Regional rolling-year counts were Auckland 17,264 (+19%), Canterbury 8,933 (+35%), rest of North Island 5,106 (+15%), rest of South Island 4,390 (+17%), Waikato 3,219 (+14%), and Wellington 2,356 (+18%). These are not August-only counts. Auckland had the largest count among these published regional groupings; Canterbury recorded the fastest percentage growth. Month and year-ended measures answer different questions: the former records one month, while the latter smooths the comparison window across twelve months.

The annual total was 7.7 consented dwellings per 1,000 residents, against 6.4 in the year ended August 2025. This rate uses estimated resident population as its denominator, so population updates can change it. It is neither a measure of completed construction nor a count of available housing stock. The annual regional groupings reported in the release are also different in granularity from the full regional table in the workbook. Reading counts alongside population and time-window definitions helps avoid interpreting higher approvals as an immediate increase in finished homes.

Consent statistics record plans approved for construction and serve as an early indicator of building activity; they do not prove that work proceeds. The series covers construction requiring consent and excludes predominantly demolition work and consents valued below NZ$5,000. Land value is excluded. Consent values include GST and are not inflation-adjusted. The annual value of non-residential building consents reached NZ$8.9 billion, up 2.2%; Stats NZ notes that prices can influence this series. The count of dwellings, floor area, and dollar value are separate measures and should not be treated as interchangeable evidence of construction volume.

Stats NZ compiles monthly counts from consents issued by territorial authorities with complete coverage of the target population, not a sample. Non-sampling errors can still arise through missing or incorrect form information, transfer, interpretation, or classification. Seasonal adjustment uses X-13 ARIMA-SEATS and includes trading-day and Easter effects when statistically significant; adjusted and trend series are updated monthly. Large projects may be consented in stages, and small-area type figures can contain errors or omissions that are not material nationally. These methods and limits matter when reading a single-month change or comparing a small locality.