How the 2011 census shows Australia's housing market is undersupplied: BIS Shrapnel's Angie Zigomanis

There have been several reports on the number of households as reported by the Australian Bureau of Statistics 2011 census in relation to estimating the number of households nationwide. The most commonly reported number of households reported is 7.8 million.

Researchers and analysts have been using this number as a justification or otherwise of claims that the housing market is in undersuply are legitimate. However, a pure count of households or occupied dwellings in itself doesn't say much at all. A count of households and occupied dwellings just tells you that one equals the other. Therefore it doesn’t measure whether there is any underlying undersupply. However, an oversupply can be measured if there are a rising number of unoccupied dwellings.

Some analysts compare their count of the number of households with the National Housing Supply Council estimate of the underlying level of households to provide evidence that the market is in oversupply. The 7.8 million households at the census normally bandied about by commentators compares with the National Housing Supply Council underlying figure of 8.8 million households to imply the NHSC overstated households at 2011 by 1 million, and therefore its estimate of undersupply was actually an excess.

However, a closer look at the 7.8 million households quoted by most industry commentators shows that this figure is just plain wrong. This is the result of these commentators plucking a headline figure from the census data without taking a closer look at the numbers behind the numbers.

When looking at household numbers to determine demand, the number of households will equate to the number of occupied dwellings – i.e. when a household is created it needs to occupy a dwelling. However, the figure that most commentators refer to for the number of households falls well short of the number of occupied dwellings on census night. To find this figure, we need to examine the census data based on place of enumeration.

The place of enumeration count is the physical count of all households in the dwellings where they were counted on census night. This shows that there were 7.8 million households nationally. However, this actually only refers to the households for which they have obtained household structure information. There are two other categories of households occupying dwellings that have not been accounted for. These are "not stated" households and "visitor only" households.

“Not stated” households are dwellings where the census collector has confirmed that someone was present, but was not able to get a completed census form to confirm the household structure. This happens across all dwelling types but is most prevalent in high-rise apartment buildings, where it is difficult to access the dwellings to drop off and pick up the forms (which is why the internet was introduced as a means of filling out the forms). While the type of household has not been confirmed, there is nevertheless a household in the dwelling. There were 279,000 “not stated” households on census night across Australia.

“Visitor only” households are households where all the occupants are not at their usual place of dwelling. This could be overseas visitor households (but not necessarily short- term holiday households), dwellings where the occupant might be away for work, households that were at their holiday house on census night, etc. While not permanently living at the dwelling, they nevertheless reflect demand for dwellings as there will always be a component of the population occupying and requiring a dwelling on this basis. There were 143,000 “visitor only” households on census night.

When comparing households over time, a further adjustment can be made in 2011 for persons who were overseas on holidays. The high Australian dollar has meant that a greater proportion of Australians were overseas on census night. At the 2001 and 2006 census, around 1.7% of Australians were recorded as being temporarily overseas on census night. However, on the 2011 census night, around 2.1% of residents were temporarily overseas. Based on the 2011 population, this reflects an extra 100,000 or so persons, or around 40,000 additional households that weren't counted. Remember that the household count reflects the physical count of occupied dwellings on census night, so the dwelling would have been recorded as empty. Conversely, subtracting these households from the unoccupied dwelling count reduces the unoccupied dwellings from equivalent to 11.4% of households to 10.8% of households, which is below the 11% of households at the 2006 census – another indicator of a tightening market over the last five years.

Adding all of these components – 7.76 million households counted, 279,000 “not stated” households, 143,000 “visitor only” households, and an allowance for an extra 40,000 households being on an overseas holiday on census night than there would normally be – provides a total houshold count, and therefore occupied dwelling count, of 8.22 million occupied dwellings.

This is still well below the NHSC underlying household count of 8.8 million households in its most recent State of Supply report (although closer to BIS Shrapnel’s pre-census figure of 8.3 million that was used to calculate the underlying stock deficiency in the June 2012 editions of our Residential Property Prospects and Building in Australia reports). However, it should be remembered that the NHSC figure (and BIS Shrapnel’s) is not an estimate of actual households, but an estimate of the underlying requirement for households, which is the number of households that would normally be expected in an unconstrained environment – i.e. under normal dwelling supply and affordability circumstances.

BIS Shrapnel’s forecasts of households used in calculating underlying demand are based on trends over time in household formation rates across different age groups, which accounts for changes in demographic trends as well as the likely impact of housing affordability constraints on household formation. In many states it is assumed that the reduction in household formation between 2006 and 2011 due to a deterioration of affordability (both owning and renting) has now become permanent and is unlikely to recover in the foreseeable future. This then places most of the states at, or close to balance, in most states.

Based on the 2011 census data, BIS Shrapnel’s estimate of the deficiency at June 2011 is 37,000 dwellings, concentrated in New South Wales, where the number of households appears lower than it otherwise would be (than explained by prevailing trends in household formation rates) due to people being forced to squeeze into the existing dwelling stock.

A national deficiency of 37,000 dwellings at 2011 was insufficient to drive any upturn in dwelling construction in 2011-12. However, declining new dwelling completions, even at the lower level of new household formation rates derived from the 2011 census, have fallen below underlying demand in 2011-12, and a rising deficiency is now emerging. As the deficiency continues to rise, it will first be seen in vacancy rates (already evident in Brisbane and Perth), before flowing through to rental growth, construction, and some level of price growth.

 Angie Zigomanis is senior manager of BIS Shrapnel.