Getting around the limitations of building work put in place data

Weather has significantly affected New Zealand throughout the first half of 2023. The effects of the extreme weather events on Auckland, Gisborne and Hawke’s Bay, and other parts of the upper North Island have been well documented. Aside from these events, the persistent wet weather has also made conditions difficult for the construction industry. Building work put in place data for the March quarter was surprisingly strong, but we have some concerns about the accuracy of this data. This article looks at whether ready-mixed concrete data provides any different information on construction activity.

Modelling, rather than surveying, most residential work

One of our ongoing concerns with building work put in place (WPIP) data has been the increasing use of modelling by Stats NZ to estimate activity on projects with consent values below a certain threshold. When it was introduced in 2015, this threshold sat at $800,000 for residential consents and $1.8m for non-residential consents. It has steadily risen in subsequent years, partly due to building cost inflation, but also due to increasing consent volumes, particularly for residential consents. These growing volumes have meant the number of surveyed projects has lifted substantially, making the survey increasingly and unsustainably expensive. Due to these factors, the cut-offs now sit at $2.4m and $2.9m for residential and non-residential consents respectively. According to Stats NZ, these thresholds mean that approximately 70% of residential WPIP is modelled, and about 25% of non-residential WPIP.

This approach means that, for residential building, construction work on almost all standalone houses is now being modelled. Any changes in the way consents end up being translated into WPIP are unlikely to be captured by Stats NZ. For example, prior to COVID-19, these changes were most likely to be cyclical, with longer lags between consent and construction at the peak of the cycle due to capacity constraints in the industry.

Since early 2020, Stats NZ has conducted an additional survey of low-value projects and benchmarked its expected model outcomes against the results, due to the changing effects that the pandemic has had on the relationship between consents and WPIP. There were downward adjustments to modelled activity for both the 2020 lockdown and the Delta lockdown in 2021, with downward adjustments continuing throughout the first half of 2022 due to labour, capacity, and supply constraints. Activity has been adjusted upward since mid-2022 as the industry has caught up on previously delayed projects.

Stats NZ expects to continue adjusting and revising its estimates of activity for the three most recent quarters on a rolling basis going forward, as more data and information becomes available. However, it’s unclear how long these adjustments will continue, given the fading effects of the COVID-19 pandemic.

Let’s concentrate on pouring concrete

Given the shortcomings of WPIP data, what other indicators of construction activity are available? Ready-mixed concrete (RMC) data is published by Stats NZ on a quarterly basis across 12 “regions” about six weeks after the end of the quarter. It provides a timelier and possibly more accurate indicator of building activity that’s taken place, even if it only represents one stage of the construction process and doesn’t offer any breakdown of activity by building type. The survey covers all RMC producers across the country. For all these reasons, we will start to report to clients on the RMC data releases going forward, to add to the insights we provide on the building sector.

For the six months to March 2023, the total volume of building WPIP was up 9.2% from a year earlier. In contrast, the volume of RMC was down 10.6% over the same period. As Chart 1 shows, these two series can sometimes diverge, but it’s very unusual for them to move in opposite directions.

The divergence in the December 2022 quarter might have been a catch-up in building activity as materials such as plasterboard became more readily available again – supply chain disruptions that had not delayed or prevented concrete being poured in previous quarters. However, we believe the March 2023 RMC result is likely to provide a more accurate picture of construction activity in the quarter. Extreme weather events and an unusually wet summer in the upper North Island have made it much harder to get work done – especially pouring concrete!

The effect of the weather is borne out by the biggest regional underperformers in terms of RMC volumes in the March quarter: Northland, Gisborne/Hawke’s Bay, and the Auckland metropolitan area. Auckland and Northland had been among the strongest regional performers in previous quarters, as activity recovered from the extended lockdown restrictions in place across the upper North Island in the second half of 2021. The swing of these regions to be among the worst performers in March 2023 is a sharp contrast. Excluding lockdowns, Northland recorded its largest annual fall in RMC volumes since 2013, and the Auckland metro decline was the biggest since 2009. The decline in RMC production in the “rural” Auckland area (encompassing the Rodney and Franklin wards) was also the largest since 2011.

Looking backwards, looking forwards

We use building consent data at a city and district level to break down WPIP data from the six “super-regions” published by Stats NZ. We also use these WPIP estimates to break down RMC data from the 12 “regions” published by Stats NZ to a city and district level.

Chart 2 shows the areas with the largest estimated growth in RMC usage over the last 30 years. These districts are generally ones that have experienced strong population growth, driving a surge in housing and associated infrastructure, although Ōtorohanga provides an example of a specific large project (Waikeria Prison) affecting recent RMC estimates. The increased unaffordability of housing and land constraints in the largest urban areas is reflected in the population spread into many of the areas shown in Chart 2.

We have also taken this opportunity to re-estimate our forecasting model for RMC volumes. Our model had previously not included infrastructure activity as an explanatory variable, primarily because of the limited availability of infrastructure data at a regional level. However, we’ve now incorporated infrastructure data in our nationwide model, as well as testing the usefulness of regional infrastructure estimates in our 12 regional models of RMC volumes.

At a regional level, infrastructure data has generally proved to be of limited value because of the lack of information on quarter-to-quarter movements in infrastructure – the variation in our regional estimates is based on June year data from our Infrastructure Pipeline Profile. However, the quarterly nationwide data has driven an adjustment in our national RMC model, reducing the importance of non-residential building in our modelling. Chart 3 compares our updated model with the previously modelled figures published in our July update of our Regional Construction Outlook.

We will use our updated model in our next set of RMC forecasts, which will be published in October. We will also monitor future RMC data releases to get a better understanding of construction activity trends, with June quarter data set to be published by Stats NZ on 10 August.

Related Articles