ABS: CPI overstated owing to smokes tax

Via the ABS:

The Consumer Price Index (CPI) measures price change for goods and services purchased by Australian households in the eight capital cities. It is a measure of inflation that informs monetary and fiscal policy. It is also used widely by economists and the general community to assess the health of the Australian economy.

An area of interest is whether prices are increasing at the same rate for goods and services that could be considered essential (non-discretionary), compared to goods and services that are more discretionary in nature. To inform this, the ABS has classified CPI goods and services into two categories: ‘Non-discretionary’ and ‘Discretionary’.

Analysis indicates that prices of non-discretionary goods and services increased slightly faster than for discretionary goods and services. Over the period 2012 to 2019, cumulative non-discretionary inflation was 14.8 per cent, whereas discretionary inflation was 12.9 per cent. Excluding the impact of tobacco (which more than doubled in price over the period) resulted in lower discretionary inflation of 6.4 per cent.

Defining non-discretionary and discretionary

There are no international standards or definitions for non-discretionary and discretionary household spending. Deciding whether a good or service meets a basic need (non-discretionary) is somewhat subjective and will differ across households. In light of this, to categorise the 87 CPI components as either ‘non-discretionary’ or discretionary,’ the ABS has developed the following definitions:


Goods or services which are purchased because they meet a basic need (food, shelter, healthcare), are required to maintain current living arrangements (car maintenance, school fees), or are a legal obligation (compulsory insurance, stamp duty).

Spending on these goods or services may be less responsive when there are changes in household wealth and incomes, or changes in the relative prices of goods or services.


Goods or services which could be considered ‘optional’ purchases, such as take away meals, alcohol and holidays.

Spending on these goods and services may be more responsive to changes in household wealth or relative prices.


Goods and services were classified in line with the above definitions. Table 1 shows the allocation of the expenditure weight for each CPI group to the non-discretionary and discretionary categories. A list of each of the 87 CPI components classified as either non-discretionary or discretionary is provided in the appendix.

Table 1: Weight contribution by CPI group (a)
CPI Non-discretionary Discretionary
Weight % Weight % Weight %
All groups 100 58.5 41.7
Food and non-alcoholic beverages 16.2 7.6 8.6
Alcohol and tobacco 8.0 0.0 8.0
Clothing and footwear 3.2 0.0 3.2
Housing 22.7 22.7 0.0
Furnishings, household equipment and services 8.9 3.8 5.1
Health 5.8 5.8 0.0
Transport 10.2 6.9 3.3
Communication 2.4 2.4 0.0
Recreation and culture 12.4 0.8 11.6
Education 4.4 2.8 1.6
Insurance and financial services 5.8 5.8 0.0

a.   Weights represent the March 2020 quarter. Values may not sum due to rounding

Key Findings

Between 2011 and the start of 2020 non-discretionary inflation exceeded overall CPI inflation. Price increases in housing, health and education costs were the main contributors to non-discretionary inflation. Figure 2 shows a fall and subsequent rise in non-discretionary inflation in 2020. This was the result of free child care being introduced in the June quarter and removed in the September quarter, as well as volatility in automotive fuel prices.

Figure 2 also shows that from 2011 discretionary inflation rose more slowly than overall CPI inflation. Price falls for goods such as clothing, furniture, household appliances and motor vehicles were the main reason, while price increases for discretionary food has also been subdued in recent years.

Tobacco was the biggest contributor to discretionary inflation, with prices more than doubling since 2011 following annual increases in the tobacco excise. Removing tobacco reduces cumulative discretionary inflation since 2011-12 from 16 per cent to 8 per cent. With ABS data showing that fewer than 15 per cent of Australians are daily smokers, removing the impact of tobacco price increases from discretionary inflation makes it more representative of the majority of the population.

The rate of non-discretionary inflation has been more than double that of discretionary inflation (excluding tobacco) over the period analysed. This analysis shows how prices for goods and services which could be considered ‘essential’ were increasing more rapidly than those for more ‘optional’ goods and services. This provides valuable insights into how Australian households have experienced inflation in recent times.

The ABS is seeking feedback on the usefulness of these measures of discretionary and non-discretionary inflation and the potential to continue to produce them in the future.

Obviously, taxes should not be included in the CPI. That will lead to all kinds of unintended consequences such as distorted wage outcomes and monetary policy.

That said, we should be including property prices in the CPI. AS we, and Gareth Aird at CBA have argued previously using a house price-adjusted measure:

There is a massive flaw in using the CPI as a proxy for changes in the cost of living.

The index ignores price changes in the single biggest purchase a person (or household) is likely to make in their lifetime – a dwelling.

For households that do not own a dwelling and aspire to purchase one, the CPI is a very poor measure of changes in the cost of living.

David Llewellyn-Smith


  1. DLS – So when it say’s housing at 22.7% non- discretionary, is that just rental costs and does it exclude mortgage repayments?

    • Nope. The RBA has ignored this in the last decade.

      The 22.7% on housing looks low to anyone living in Sydney or Melbourne, too.

      • You sure? This is what the RBA say’s about ‘Housing’ in the CPI


        The two largest housing components of the Consumer Price Index (CPI) basket are rents and new dwelling purchases by owner-occupiers, which together account for around one-sixth of the CPI basket.[1] Price inflation in these two components is influenced by conditions in local housing and construction markets and can vary considerably by city. Inflation in rents and new dwellings have each averaged a little over 3 per cent in year-ended terms since 2002. Over the past year these two components have been much weaker than average and have contributed notably less to CPI inflation; rent inflation was 0.4 per cent and new dwelling cost inflation was 1.2 per cent (Graph C1).

        • So in other words, rents staying flat, or going down, is bad for inflation and so therefore undesirable from the RBA’s perspective?

        • The one thing that has not been included is Land Value. Go check on the Web. Or go look at archives of John Adam’s interview by Martin North. If I remember well (I could be wrong in my recall), he says that up to 1998, RBI did not use CPI as the main measure of inflation. In 1998, Land value was removed from CPI and few months later RBA started using CPI as the main measure of inflation.

          What has increased the most in last 17 years or more? It is the land value. From what little I know, in many houses, the Actual House value (land value removed from it) is pittance, comparatively. As time goes by it further goes down due to wear and tear, unless it is a Heritage house.

          Remember RBA interest rates are set based on inflation measure. Now you know why despite quick rise in home values (huge upward movement in short 17-20 years time), the interest values are still low.

          Had the Land value not been removed from CPI (and RBI used it, as it does now, for measuring inflation) do you think House prices would have gone up this much. A big No. Because every time Land value increases it will have a telling effect on CPI and therefore interest rates would be increased big time by RBA. This will lead to massive mortgage payments and therefore less and less people will take loans from banks, under such conditions. This means that there would have been greater control over house price increase via interest rates leading to only very modest and healthy increase in house prices over longer periods of time and therefore higher housing affordability for the general populace.

          If I remember well, I have seen a graph posted here in MB a while ago on volume of housing sale transactions over several years, wherein it shows 2003/2004 to be having the highest volumes and in last 16-17 years this volume has only come down big time. Can we read this graph as an indicator of housing affordability too? Hmmm…. Don’t know.

          Anyhoo, what it is is what it is. Looks like, We as a Nation are in a Big time economic problem.

          I think, come March-July 2021 we will know the amount of the proverbial we are in. Knee Deep, Waist Deep or Neck Deep.
          Until that time have a nice Christmas, Happy new year and Happy Easter folks.

          • Jumping jack flash

            1 for many reasons.
            March/April will be the proving grounds for this latest can-kick. Will it have re-ignited the debt economy? Will the new debt created by then be able to create enough capacity to repay the last load of debt, its interest, create the capital gains, and leave enough to boost spending/wages to the point where even more debt can be created at the same price?

            Its a big ask. They havent succeeded for 10 years. It will also mean phenomenal inflation (conveniently hidden, or maybe not even) but absolutely not having any effect on interest rates at all!

  2. That said, we should be including property prices in the CPI.

    Isn’t rent included though. That’s a decent proxy for property prices.

    • Fabian AlderseyMEMBER

      Rent Is included, as is the cost of building a house. The price of the land is excluded though (and that’s the majority of the cost when buying a house), since land isn’t a consumption item.
      The ABS also publishes residential property price indexes. If an employer uses just the CPI to index their employee’s wages, that employee is going to go backwards in terms of being able to afford a house using their wages. Ideally employers would use a mix of CPI and RPPI, but I’d say that very rarely happens.

  3. Jumping jack flash

    CPI is not cost of living. CPI is not anything. It is a deliberately engineered and largely meaningless measure that is manipulated as required to produce the result they need, to enact the policy they think is required at the time.

    It has some semblence of reality of course. Im sure they do try to make it as believable as necessary.

    Fortunately they decoupled it from interest rates a while ago so if CPI does accidentally show “high” inflation for some reason then they dont necessarily need to do anything to interest rates.