Monday, February 8, 2010

Fail: CPI, housing and the cost of living

This short article about inflation, and the ABS review of the Consumer Price Index (CPI), prompted me to write again on the subject.  In fact, some friends and family who recently returned from a few years abroad have also been bugging me about why Australian prices have shot up so much, yet we hear almost nothing about rapid inflation.  Housing prices in Brisbane have typically increased 200% in the past decade - around 7% per year.  If housing was just 10% of the CPI basket, it alone would contribute 0.7%pa growth to the index.

First, let's be clear about the purpose of the CPI.  In principle I believe that measuring the price level is impossible as the type and quality of goods and services in the economy changes constantly. However, some indicator about the changing cost of living is still important for determining changes to welfare payments and calculating real growth in incomes. 

Unfortunately, not many people know that the CPI does not measure the change in the cost of living (see p6 of the ABS CPI overview report here).  The ABS method, including quality adjustments and the weighting of household consumption goods, means that the CPI fails to be a useful measure for determining real incomes and purchasing power. 

To do that you would use what is know as an Outlays Method, rather than the Acquisitions Method currently adopted by the ABS.  What the ABS tries to do is the impossible.  It tries to measure the rate of change of prices of goods and services, or the change in the price level, using the acquisitions approach.  

The main difference between the two methods discussed above applies to durable goods – those goods that are that consumed over a long period, such as housing.  Because the acquisitions approach ignores the interest cost of debt (as this is not a cost in a given time period, but a cost bringing forward spending form future periods), the cost of servicing home loans is completely ignored.  Any changes to interest rates do no impact the CPI until prices themselves adjust to new interest rate conditions.

Quality adjustments are also killing the CPI.  For housing in Brisbane for example, the CPI attributes a mere 80% growth in prices from 1999 to Dec 2009, yet my observations suggest that homes have typically increased 200% (meaning it is now 3 times the price).  Are our homes really more than ‘twice the quality’ they were in 1999? The ABS thinks so. 

With the quality adjustments to rental price (not ‘cost’ remember) we get this absurd result:

…that for the period December 1994 to December 2005 the average rent paid by private tenants included in the ABS Survey of Income and Housing increased by 41.1%, while the increase in the CPI rental component was 22.9%

And guess what the share of household budget the CPI uses for rent.  Go on.  You wouldn’t believe it.  5.22%. 

And while I’m at it, what about the weighting for house purchase? 11.21% - slightly less than the weight given to motor vehicle costs.

While I admit that the weightings actually reflect the combination of household types, and those who own their own home outright will face no house purchase costs or rents, it still seems a little on the low side to me.

Another massive shortfall of the CPI is a bias towards higher income households.  For example, low-income households spend more money on what we would commonly call ‘the basics’, such as food, fuel, and rent, but less on luxuries like computers (whose quality adjusted price declined by 50% in the past 10 years according to the ABS), dining out, holidays and airline flights, and hotel accommodation.  So when the basics increase in price at a greater rate than the luxuries (which they have been in the past decade), the cost of living for the poorer households increases by more than the published figure, while the cost of living for wealthier households increases by less than the official number.  

Finally, all the weightings of the CPI are here (on page 113), and the index for each component over time is here.  If anyone thinks they reflect their reality please let me know.

The simple way around all nonsense is for the ABS to publish a CPI using the acquisitions approach, and a ‘cost of living’ index using the outlays approach.  Discrepancies between the two would have simple explanations, and the relative change between the two indices would be of great assistance to policy makers.  

I suspect however, that admitting the cost of living for many Australians may be outpacing wages is not going to be politically palatable to any party in power.  Maintain the illusion of a low CPI is also one reason why hedonic indexes (which adjust for quality) are becoming more politically popular.

I will be interested in the outcomes of the current CPI review to see whether they improve or degrade this key national economic measure.

The ABS produces an Analytical Living Cost Index (ALCI) here, using the appropriate outlays method.  Unfortunately, while interest rate are factored in, the price of home purchase is removed
the ALCIs, the changes in the amount of interest paid on mortgages (measured as part of financial and insurance services) and other costs (e.g. maintenance costs and council rates) are included for owner-occupied housing. In addition, changes in rental are measured for that part of the reference population that resides in rented dwellings. The ALCIs do not include the net purchase of housing and the increase in volume of housing due to renovations or extensions.
Furthermore, most other components of this measure are simply taken from the CPI in their quality adjusted form.  Which gives the ABS an ALCI which mirrors CPI almost exactly.  


  1. I couldn't agree more - what a load of useless overmanipulated statistical bullsh*t!!

    The name Consumer Price Index - on the face of it you would expect an index that measures changes in general consumer prices.. However as you have shown it is so over manipulated that any sign of significant movement will be "adjusted" out of the index and we're left with something that is kept within the ABS, of sorry- RBA target range.

    Anyhow, thanks for putting this into a well structured logical argument.

    By the way, it is not only CPI but many other national statistics which could use an honest review, this is true not only for Australia but around the globe.

  2. I am currently creating a 'Make your own CPI' spreadsheet that will use some ABS data inputs and allow you to put your own weightings on the basket of goods. I have found some CPI weightings from around the world to use as comparisons - in Germany rents are weighted 20% (with another 10% in the housing basket for maintenance and building services).

    I will try and have it up this week (but don't hold me to that).

    And some more official stats may get the once over in the future.

  3. Awesome work Cameron - you should also make a submission to RBA. You have a lot of people reading this blog by the way...bloody legend!

  4. Sorry, submit to the ABS I meant$File/64680_dec%202009.pdf

  5. You say;
    "And while I’m at it, what about the weighting for house purchase? 11.21% - slightly less than the weight given to motor vehicle costs."

    Please consider that house purchase only count "cost of new build". Land NOT included. Existing houses NOT included.

    It can be seen that any increase in productivity in the building sector, even if that is using cheaper materials or moving to pre-fab structures (inherently cheaper) will reduce the CPI for housing purchase!!! Also consider that cost of building not changed for like twenty years, so housing purchase really little part of CPI.

    CPI only good for someone whose basket is same as CPI. Very useless for others.

    Also consider US of A and China where governments not like food price inflation so they just remove from basket.