Estimating completions of new dwellings in New Zealand

Almost everyone we talk to in the construction industry at the moment wants to know how the residential build rate, or the number of completions, compares to new dwelling consent numbers. The heightened interest is understandable: no one believes that we’ve been building new homes at the same rate as we’ve been consenting them. Although the consent total peaked at 51,015pa in May last year, most industry players believe the build rate will only have got up to somewhere between 35,000 and 42,000pa, because the industry doesn’t have the capacity to deliver a greater volume of new homes.

We’ve undertaken modelling work, using data from Stats NZ and Auckland Council, to estimate historical completion numbers,1 as well as providing forecasts of completions that are consistent with our consent forecasts. We plan to include updated predictions of completions alongside our consent numbers in future Building Forecasts.

Completions yet to peak

This section contains the results of our modelling. An overview of our methodology is contained later in the article.

Stats NZ’s experimental figures are only available up until early 2022, and they showed that national completions got up to almost 34,700 in the year to March 2022. At that time, Auckland Council’s completions were running at 13,130pa, and they have since climbed to 16,769 in the year to September 2023.

Applying the recent trends in Auckland Council’s data to activity around the rest of the country suggests that the nationwide completion rate has now climbed to about 40,700pa (see Chart 1). It is also likely that the completion rate has yet to peak, something we expect to happen in the March 2024 quarter at 42,245pa – almost two years after the peak in consents.

From early 2024 onwards, we expect completion numbers to start declining. However, based on our current forecasts, completions are predicted to hold above consents through until mid-2027, reflecting the continued downward trend in consents as well as the pipeline of projects awaiting completion.

Our estimates of completions in Auckland and around the rest of New Zealand are shown in Chart 2 and Chart 3.

The most substantial caveat around our estimates is the assumption that completion rates and the lags between consent and completion for building in Auckland are mirrored around the rest of the country. This assumption means that the effects of Auckland’s Covid lockdown in the second half of 2021, creating greater delays to building projects, could be distorting numbers for the rest of the country somewhat. There is also a risk that differences in building lags between Auckland and the rest of the country are not picked up when demand conditions are strong in one area but not in the other. However, this latter risk is likely to be minor currently, given that consent numbers and demand have been highly elevated across the entire country.

Some risks to consider around our estimates

Apart from the simplifying assumption that trends in Auckland are mirrored around the rest of the country, we see two main risks around our estimates of completions over the coming year. Firstly, as Chart 1 shows, the peak in completions is significantly lower than the peak in consents, so there is a chance that completion numbers could push higher as the industry continues to work its way through the backlog of consents. In our view, though, significant further increases in completions are unlikely given the capacity limitations in the construction industry and its supply chains.

Secondly, the downturn in completions could take longer to come through than our March 2024 peak would imply. Nevertheless, we are relatively comfortable with this estimated timing, occurring almost two years after the peak in consents. Our discussions with firms supplying materials into the residential construction sector suggest that all businesses are grappling with an emerging downturn in demand for their products.

In our view, the wide gap between consents and completions at the moment also adequately reflects a likely higher cancellation rate for projects from 2022 given falling house prices, much higher building costs, and weak economic conditions. Stats NZ’s historical estimates show a fall in the completion rate from an average of 90.8% for consents in the year to June 2007 to 84.5% for consents in the year to September 2008. Significantly weaker economic conditions from mid-2008 onwards, due to the Global Financial Crisis, saw a much higher proportion of consented projects fail to be completed.

Completions v work put in place data

Chart 4 compares year-end growth rates for the number of consents, the number of completions, and the volume of new residential work put in place (WPIP). One might expect changes in WPIP to lie somewhere between consents (the very start of the build process) and completions (the very end of the process). However, the graph shows that it’s all a bit of a mess. Sometimes WPIP seems to move in line with consents, and at other times it seems to move more in line with completions.

WPIP remains our favoured measure for the volume of work being completed, partly because it is official data published by Stats NZ and so provides a definite benchmark to measure our forecasts against.

Since 2020, Stats NZ has been utilising CCC data internally to help refine its WPIP modelling, given the disruptions that Covid caused to construction activity. Serendipitously, these refinements have also helped refine the WPIP modelling to cope with the capacity limitations experienced by the industry over the last couple of years.

Nevertheless, we continue to express some caution about the residential WPIP estimates, because of the high proportion of consents that are now modelled rather than surveyed. In September last year, Stats NZ increased the threshold for residential projects to be surveyed to $2.4m or over, with all consents under this value being modelled based on historical relationships. This cut-off means that work on almost all standalone houses is now likely to be modelled rather than surveyed. As a result, Stats NZ has a limited ability to quickly pick up changes in how house consents are being translated into WPIP.

How we calculated our estimates

The remainder of this article provides more detail on our data sources and the modelling work we have done to estimate and forecast completion numbers.

The foundation provided by historical estimates

The information we have on completions, as measured by code compliance certificates (CCCs), and lags between consent and completion, is as follows.

  • Stats NZ published quarterly completion rates for standalone houses and attached dwellings for 2006-2019 as part of its experimental building indicators in 2022. The completion rates were dated by the quarter of consent issuance and calculated as a proportion of consents issued in that quarter.
  • The same Stats NZ data release also included quarterly numbers of completions for standalone houses and attached dwellings for 2009-2022, These completions were dated by the quarter of CCC issuance.
  • Auckland Council publishes monthly data from 2013 onwards on the total number of completions on the same basis as the Stats NZ data, but for total new dwellings only.
  • Auckland Council also provides a split of CCCs issued each month into how long it is since the building consent was issued: 0-2 years, 2-4 years, or over four years. As with the other CCC data, these figures are dated by the month of CCC issuance.

Filling in the missing pieces

The missing pieces of the puzzle are the number of completions by the date of the building consent issuance and, by extension, a matching of consents and CCCs to better understand the lags between consent and completion.

Stats NZ’s data provides information on the number of completions by date of consent issuance, thanks to the nationwide completion rates published up to March 2019. Some caution needs to be taken about the later estimates given that they were published with a three-year lag, and a small proportion of projects is still likely to have been working towards completion at that stage (as evidenced by Auckland Council’s data showing some CCCs being issued after four years or longer). Most critically, the Stats NZ data provides no information about more recent completion rates, which is where Auckland Council’s data comes in.

When were consents granted for dwellings getting their CCC?

The shortcoming with Auckland Council’s figures is that they only give an imprecise two-year window for when a building consent was issued relating to a new CCC. We have attempted to “reverse engineer” completions by the quarter of building consent issuance, with the resulting completion rates shown in Chart 5. They are compared with our estimates of Auckland’s completion rates during this time using nationwide Stats NZ data, adjusted for the different mix of standalone houses and attached dwellings in Auckland.

The results for any individual quarter are far from perfect – a completion rate of over 100% is impossible! However, the average figures over the period are reasonable and provide us a guide for future completion rates across the various windows of 0-2 years, 2-4 years, and over four years. We have used this modelling to inform our estimates of future completions, as shown in Chart 6.

With Auckland Council CCC data now available up to the September 2023 quarter, we estimate that just 57% of consents issued in the September 2021 quarter have received their CCC within the two-year timeframe. This figure is well below our estimated longer-term average of 70% of consents since 2015 receiving their CCC within two years. At the same time, we estimate that the proportion of consents taking 2-4 years or more than four years to receive their CCC is higher than the historic average. These proportions will remain higher throughout the next couple of years due to the lagged effects of delays caused by Covid lockdowns, supply chain disruptions, and capacity limitations in the construction industry.

Modelling future completion rates

The mix of completions obtained from the above modelling is supplemented with an overall target completion rate for future quarters, which is modelled on economic growth, house price inflation, and Tobin’s Q (the ratio of house price inflation to building cost inflation) over the two years following the consent’s issuance. Stronger economic growth typically leads to a higher proportion of completions, as does faster house price growth (either on its own, or relative to building cost inflation, as measured by Tobin’s Q). This modelling allows for significantly lower completion rates for consents issued during 2022 and 2023 due to falling house prices and weaker economic growth, which have reduced the viability of proceeding with new developments.

We have then applied Auckland’s profile of completions, both in terms of lags between consent and completion, and overall completion rate, to the rest of New Zealand. For example, our estimate suggests that 13% of Auckland dwelling consents issued between July 2021 and September 2022 received their CCC in the September 2023 quarter. We apply this figure to consents around the rest of New Zealand for the same period to obtain our estimate of completions for the 0-2-year timeframe outside Auckland in September 2023.2 A similar process is used for the timeframes of 2-4 years and over four years to obtain an overall completion figure for the rest of New Zealand for the September 2023 quarter.

In summary

We have been through several iterations of our modelling work trying to estimate completions. The complexity of the work is heightened by the combination of two different data sources for New Zealand and Auckland, the availability of completions data on dwelling types for New Zealand but not for Auckland, the lack of availability of completion rates for Auckland at all or for New Zealand since 2019, and the limited information on what consenting period Auckland’s CCCs pertain to. In essence, the modelling exercise is like a maths problem from high school with simultaneous equations – but one where there are more unknowns than equations, so it is impossible to fully solve!

Most importantly, there’s the requirement for the results to be realistic given what we know about the construction industry, which is particularly important at the moment given the extremes of the recent construction cycle. Is the peak in completions appropriately constrained given the limitations on capacity in the industry? And is the drop-off in completions sufficiently delayed and moderated given the backlog of consented dwellings waiting to be built?

Sometimes it is only when additional input data becomes available that shortcomings in the model become apparent and further refinements need to be made. We intend to provide forecasts of completions alongside our regular residential consent and WPIP forecasts in coming quarters. As part of that process, we will monitor the model and adjust and refine our assumptions and calculations as necessary.

1 We take the issuance of a code compliance certificate as a proxy for a dwelling being completed.

2 The percentage applied to the rest of New Zealand will differ slightly from the Auckland figure due to variations in the mix of dwellings (standalone houses v attached dwellings) and the timing of consents within the period we estimate that the CCCs pertain to.

Related Articles