What is the current status of official national completion data?

Currently, there is no official data release, although we are expecting Stats NZ to commence a regular release in 2024. However, Stats NZ did publish an experimental data set from Q1 2011 to Q1 2022, which remains available, and Auckland Council does publish its monthly CCC data. This Stats NZ experimental data set covers less than 2/3 of total new dwellings consented in New Zealand. So alongside national consents data, the primary predictor of completions, these sources do combine to provide a credible opportunity to model a more complete analysis to assist the industry in the absence of official data.

Why is it important to revise the model?

Better understanding of the relationship between consents issued and eventual completed builds is crucial for the industry to best predict workloads. We know that not all consents are converted to builds, and therefore, demand for materials and labour needs to be understood more from a lens of what is going to be built than what has been consented. It is also important to understand completions as we need to shift the narrative from building intentions to completions. We have a tendency to compare dwelling consents to population growth and assume that everything that we consent is going to be built. The Stats NZ experimental data set showed that completion levels may be significantly below what is being consented though, particularly as consent numbers reached record levels.

What are the key inputs and assumptions that contribute to the model?

Firstly, from experimental CCC data by Stats NZ we can assume detached homes typically take 9 months to complete from consent issuance, and attached dwellings 12 months. So this allows us to incorporate this lag effect and model our expected completions taking into account dwelling types. The second input is the typical completion percentage rate of consents issued. With the 11 years of Stats NZ experimental CCC data, we can see the rate is not consistent, in fact quite varied.

So today, how does the model suggest we’re tracking for annual completions?

Using the new model, we now estimate 9,950 residential dwellings were completed nationally in the second quarter of 2023 contributing to a rolling 12-month completion number of 39,930. We think our 12-month rolling prediction model is accurate in identifying most of the variance, but we cannot easily include factors such as COVID-19 lockdowns or the subsequent catch-up in CCCs in the modelling.