Senate Standing Committee on Economics - question taken on notice on 29 May 2012
Senator LUDLAM asked:
Senator LUDLAM: We are pretty short of time so I might seek a briefing from you when we are not in this context and get a sense of where we are up to. My other question is much more specific. It is about how we calculate the CPI. I know there are a lot of different ways of chopping it up and a lot of different indices, but the CPI is most commonly quoted and is understood to be the rate that we follow. How are housing and, particularly, rental expenditure incorporated into measures of the CPI? Tell me if my maths are wrong, but I understand with housing that you weight house purchases, rents, other maintenance, utilities and so on. With respect, you say average weekly expenditure on rent is estimated at $92, which contributes 6.71 per cent to the CPI. Median rents, of course, in places like Perth and Sydney are up around $400 per week. Can you tell us exactly how that works? Is the CPI intended to accurately represent the cost of housing or not?
Mr Hockman: It is intended to represent the purchase cost on average across the capital cities of housing. It includes, therefore, in the survey of household expenditures where the weights are drawn from parts of the population that, of course, do not rent their house and are using imputed rents. We also have an imputed rent calculation within the CPI. We use that in the national accounts. So it is not meant to be a representative rental for any representative house in the Australian community.
Senator LUDLAM: So what is it doing in there? This is obviously no fault of yours. The CPI is understood to be the cost of living-consumer prices, what we pay to live here. Has the way in which
housing is represented within the CPI changed over, say, even the last 10 years or since 2004 to represent the much larger share of household expenditure that is going into housing? Or has it stayed flat?
Mr Hockman: It has certainly changed with the incorporation of the new weights drawn from the household expenditure survey, which is conducted every six years to represent the housing proportion in total household expenditure. The CPI is actually measured on a household expenditure basis across the whole of the population in the capital cities. That is where the weight you referred to was drawn from. It was reviewed in our 16th series review and that acquisitions basis was determined to be the best way of measuring the CPI for the purposes of managing inflation.
Senator LUDLAM: Has the weighting taken up within the different factors that constitute the CPI and given to housing risen in proportion to its actual weighting in real life?
Mr Hockman: To be honest, I would have to check that.
Senator LUDLAM: If you could.
The table below presents the expenditure weights of the Housing and Rents components of the Consumer Price Index (CPI) over the period 1998 to 2011. In 2004, the 14th series CPI was
constructed using weights derived from the 1998-1999 Household Expenditure Survey (HES). 15th series CPI was constructed using weights derived from the 2004-2005 HES, while the most
recent 16th series CPI was constructed using weights derived from the 2009-2010 HES. The expenditure on Rents relative to total household expenditure has remained relatively stable over the
last 13 years, increasing from 5.80% in 1998 to 6.71% in 2011. The expenditure on total Housing relative to total household expenditure increased from 19.35% to 22.30% over the same period,
while the weight of Rents relative to total Housing expenditure has increased from 30.0% to 30.1% over the same period.
Table 1. Expenditure weights of Rents, Total Housing, and Ratio of 'Rents' to 'Total Housing'
(Sources: ABS cat. no. 6431.0 - Consumer Price Index, Historical Weighting Patterns, 1948-2011
16th series -
15th series -
14th series -
13th series -
Rents weight in
6.71 5.22 5.60 5.80
weight in the
22.30 19.53 19.75 19.35
30.1 26.7 28.4 30.0
Correction to information given at the Senate Estimates Hearing
It should be noted that the earlier statement from Mr Hockman in paragraph 2 relating to imputed rents in the CPI was incorrect. Rent in the CPI is directly priced from real estate agents in each capital city under a matched sample approach.Mr Hockman's comment regarding imputed rents in the National Accounts is still valid.