By Rodney Dickens*
Annually since 2013 MBIE has released the National Construction Pipeline report. The reports include forecasts for national and regional residential building, non-residential building and infrastructure spending that are supposed to help with “improved planning by all participants in the sector” and with “scheduled investment in skills and capital”.
Instead, the forecast for new dwelling consents in the 2018 report are likely to provide extremely misleading insights for firms planning for the future including scheduling capital spending.
The problem is that the forecasting approach used by BRANZ – the organisation that provides the residential building forecasts for the reports – is fundamentally flawed.
The 2018 forecasts for new dwelling consents effectively just add KiwiBuild on top of the current high level of building. The current high level of building is a result of low interest rates coinciding with high net migration. Only three times in the last 45 years have low interest rates coincided with high net migration. The super-high base level BRANZ is effectively adding KiwiBuild on top of is a rare event. It is highly likely over the six year forecast period used in the report that these two primary drivers of cycles in residential building deteriorate with net migration already in the process of falling.
Ironically, the more that KiwiBuild boosts the level of building the more interest rates are likely to ultimately increase and undermine the base level of building.
Should you take any notice of BRANZ’s forecasts for new dwelling consents?
The extract below is from page 12 of the 2018 National Construction Pipeline report and shows BRANZ’s forecasts for annual new dwelling consents out to 2023.
If I was managing a firm operating in the residential building industry my greatest concern based on the forecasts above would be whether my firm had the capacity to deal with a 39% increase in new dwelling building activity over the next six years. I would be looking to significantly wind up capital spending and employment that would involve taking on more debt (i.e. leave my firm exposed if building activity levels ended up significantly lower than predicted by BRANZ and if interest costs increased significantly).
Appendix C of the report spells out the various assumptions behind BRANZ’s forecasts although the guts-of-the-matter is spelt out on page 9:
The dwelling unit forecasts are based on Statistics New Zealand’s December 2017 household formation data, which provides estimates of the number of new dwellings required derived from population estimates. This information provides estimates of the number of new dwellings required to meet both expected population growth and to remedy already existing housing shortages.
There is no allowance for the fact that the Stats NZ projections don’t take into account cycles in net migration that are a key part of building cycles while the BRANZ forecasting approach takes no account of cycles in interest rates that are also an integral part of building cycles. Interest rates have been the most powerful driver of cycles in new dwelling consents but there was no allowance for this or the likelihood of a cycle in interest rates in BRANZ’s new dwelling consent forecasts.
Effectively BRANZ has started with the current high level of new dwelling consents (i.e. around 32,000 per annum) and added the boost required to “remedy already existing housing shortages” (i.e. the boost from KiwiBuild). By contrast, a key insight provided in our monthly building reports is warning clients not to add the boost from KiwiBuild on top of the current super-high level of building.
Understanding what drives cycles is critical to residential building forecasts
KiwiBuild doesn’t sort out the most important factor that has contributed to the level of residential building running below the level justified by population growth over the last decade: a massive increase in section prices. But as covered in our monthly building reports, there are several reasons why it will make significant progress; especially the large injection of taxpayers’ money the government has generously contributed.
However, the starting point for KiwiBuild is a high level of building by historical standards that is the result of low interest rates coinciding with high net migration as shown in the two charts below. Interest rates take approximately 12 months to impact on new dwelling consents and reflecting this, the red line in the first chart has been advanced or shifted to the right by 12 months with the same true for net migration in the second chart.
The high level of new dwelling consents in recent years and the mid-2000s was because low interest rates coincided with high net migration. By contrast, high net migration in the mid-1990s didn’t result in such high new dwelling consents because interest rates were much higher. You have to go back to the early-1970s for the previous major boom.
Net migration has already started to fall and is generally expected to experience quite a significant cyclical fall over the next few years, implying a significant slowing in population growth. Even the Reserve Bank predicts a significant deterioration in net migration as shown in the chart below from the May Monetary Policy Statement and for once we agree with the RB for the reasons outlined in our monthly building reports.
Source: https://www.rbnz.govt.nz/- /media/ReserveBank/Files/Publications/Monetary% 20policy%20statements/2018/mpsmay18.pdf
The BRANZ’s dwelling consent forecasts assume away the likelihood that population growth and therefore a key source of demand for new housing is likely to fall significantly over the next several years. But just as importantly, BRANZ’s forecasts take absolutely no account of the impact of interest rates despite them historically being the most powerful cyclical driver of new dwelling consents.
The experiences in 1999 and 2000 provide a good example of interest rates being a much more powerful cyclical driver of new dwelling consents than net migration. Net migration deteriorated and remained low over this period but despite this there was a major upturn in new dwelling consents in 1999 followed by an equally large fall in 2000, as shown in the second chart. As shown in the first chart, the spike in new dwelling consents in 1999 followed the large fall in interest rates in 1998 with the normal 12 month lag while the tumble in new dwelling consents in 2000 followed the significant increase in interest rates in 1999 with the normal lag although low net migration most likely added to the scale of the fall of consents in 2000.
As covered in our building and economic reports, the Reserve Bank is again setting things up for another significant increase in interest rates.