Dwelling unit statistics

Yesterday's look at the housing finance statistics prompted me to dig out some old code I wrote a few years ago and look at dwelling unit approvals, commencements and completions. The headline statistics suggest we are passing the peak of the current dwelling unit building boom. The first chart is the rolling annual total statistics from the original data. The second chart is the quarterly trend estimate.

In the next set of charts, we can see that the recent decline in approvals appears to be more pronounced in the apartment sector, rather than the house sector.

Looking at approvals by state, we can see the decline is fairly widespread. The decline is not only obvious in the mining states (WA, Qld and NT), but also in NSW and ACT. Approvals in South Australia and Tasmania are lacklustre, but not in decline. Victoria is the exceptional state, with approvals continuing at historically high levels.

The quarterly building commencements data is only available to the September quarter 2016. The December quarter will be released on 12 April this year. As can be seen in the first two charts above, the commencements pipeline appears to be trending downwards. Commencements tend to follow approvals closely, so I expect this number to continue to decline over the next two quarterly prints.

Completions can be compared with the ten year rolling mean for each state.

Notwithstanding the recent growth in approvals and commencements, the reality is that completions were not keeping up with population growth in most states between 2006 and 2013. The situation has softened since 2013 (as supply caught up with pent-up demand), and now there appears to be an  over-supply of new stock coming to market in the ACT, NT, SA, Tas and WA.

We can view these as ratios. In both the 2006 and 2011 census, there was on average 2.55 persons per dwelling across the nation. The additional stock and additional persons at the margin currently appears to be below this average (although it should be noted that these statistics are for gross increases in the housing stock, not the net increase after knock-downs and accounting for housing stock that becomes uninhabitable).

Notwithstanding this difference (between gross and net stock increases), the statistics suggest in most jurisdictions we have had a recent period where new stock is more than keeping up with population growth. It is not implausible, therefore that the demand for new stock has been exhausted in most jurisdictions, and that the usual building boom-bust cycle is moving towards the bust end of the cycle (in all jurisdictions but Victoria).