6.1. The previous section described the new HCP price index which the ABS proposes to develop. In recognition of the increasing interest that policy makers have shown in measures of underlying inflation in recent years, this section considers alternative conceptual approaches to modifying the HCP index to obtain a measure of underlying household inflation.
6.2. As stated at the beginning of section 3, one of the ideal properties that inflation measures would possess is that, from an analytic viewpoint, the effects of changes in government charges and taxes would be capable of separate analysis and the effects of erratic price fluctuations would be identifiable.
6.3. That is, a market-related measure of inflation which abstracts from the effects of exogenous factors should be available, providing a measure of endogenous, or underlying, inflation.
6.4. A measure of underlying inflation should exclude three types of price change, namely:
- changes which reflect the impact of monetary and fiscal policy decisions;
- changes which reflect a regular seasonal pattern; and
6.5. A number of modifications would need to be made to the HCP price index in order to obtain a measure of underlying household inflation. There are three different methods that could be applied:
- changes which reflect inherent volatility rather than underlying price pressures.
6.6. The exclusion method can be considered in relation to the Treasury underlying rate and other measures presented in the CPI publication such as price indexes covering ‘All Groups excluding interest and volatile items’ and ‘Private sector goods and services’. Most attempts to measure underlying inflation in Australia have been based on the exclusion method, with the difference between measures reflecting differences in the definition of an exogenous price pressure.
6.7. The Treasury underlying rate is derived by defining certain components of the CPI basket as being significantly influenced by exogenous factors (monetary and fiscal policy, seasonality and volatility). These components are then excluded from the CPI basket to derive a measure of underlying inflation. The Treasury measure of underlying inflation excludes about half of the items (by weight) included in the CPI.
6.8. The specific adjustment method abstracts from the effects of fiscal policy decisions by separating transaction prices into market prices and taxes and subsidies, and it also removes the effects of seasonal factors and volatility.
6.9. While the decomposition of transaction prices may be difficult to do at the point of price observation, the separation can be made through a modelling process. Several countries have developed a form of net price index which represents an attempt to develop a measure of underlying inflation by excluding the effects of changes in fiscal policy. The net price index has been considered a particularly useful addition to the existing set of inflation indicators in those countries.
6.10. To abstract from seasonal factors and volatility, seasonal adjustment and smoothing techniques could be applied. Current ABS techniques for deriving trend estimates are discussed later in this section.
6.11. The third approach involves applying outlier-based methods which give a relatively smaller weight to any price which is rising or falling in an extreme way (outliers). The weighted median and trimmed mean measures of underlying inflation are based on this approach, under which the contribution of the different sources of price change would change from quarter to quarter.
6.12. These three methods are assessed in more detail in Appendix 2 where it is concluded that the specific adjustment method is conceptually superior to the other methods.
6.13. Therefore, the ABS proposes to pursue investigations into the measurement of underlying household inflation through the development of an experimental, underlying price index of HCP. The approach proposed is to apply the specific adjustment method to the price index of HCP, to exclude the effects on prices of changes in indirect taxes and subsidies, seasonality and volatility.
6.14. This approach would overcome several limitations associated with the current exclusion-based underlying indexes, that is:
- the specific adjustment method; and
- almost all items in the CPI basket would be represented (interest rates and non-market items would be excluded under the concept of the DFP model);
- specific adjustment to remove the effects of fiscal policy, seasonality and volatility would be preferable to the total exclusion of the affected items which results in the exclusion of many valid market determined price signals;
- under the specific adjustment method, non-underlying price changes which affect included items (e.g. broad increase in sales tax, increase in tax on motor vehicles) would be specifically adjusted for, while they would not be removed under the exclusion method; and
6.15. However, the specific adjustment method does have some practical disadvantages which are outlined in Appendix 2. These arise from the difficulty of accurately and objectively estimating the magnitude and timing of the effect on prices of changes in government policy, seasonality and volatility. In particular, it is difficult to specifically adjust for subsidies. These issues are discussed in more detail below in relation to net price indexes.
- prices charged by public corporations for telephone and electricity services are likely to be reflected, while they are excluded from Treasury’s measure on the basis that they are significantly influenced by government policy.
Net price indexes
6.16. Finland, Sweden, Denmark, Netherlands, United Kingdom and Canada have at some point published net price indexes which attempt to estimate indirect tax effects on the CPI, although some of these countries no longer produce a net price index.
6.17. The potential uses of a net price index are summarised below.
6.18. ‘The foremost use of a Net Price Index is for purposes of analysis, such as in the assessment of the importance of taxes and tax changes in the CPI, and in price formation studies. Another important use is as a tool relating to compensation for only part of general price change, where the intention is not to compensate for tax changes. Other uses include the indexation of pensions to maintain purchasing power, and the adjustment of income taxes for the effects of inflation. Another use could be the deflation of value series, to obtain constant dollar measures that are not affected by taxes. Finally, a very important use of a Net Price Index is to provide information that is of assistance in understanding price change for consumer commodities during a period of tax reform.’ (Statistics Canada 1987.)
6.19. Indirect taxes are generally defined as taxes on goods and services (rather than on incomes) while subsidies are treated as negative taxes. To estimate the first round effect of indirect taxes on the CPI it is usually assumed that indirect taxes are fully passed on to prices of household purchases and that this pass-through is instantaneous. While neither assumption is entirely realistic, any practical deviations are unlikely to significantly detract from the suitability of the proposed approach. A net price index provides a measure which helps to explain movements in measured inflation without the clouding effects of changes in taxation.
6.20. There are several alternative treatments that can be used to produce net price indexes. A net price index can be calculated as a tax free price index (by excluding indirect taxes from both base and current period prices) or as a constant tax price index (by excluding only the effect of changes in taxes since the base period). Further, the net price index may exclude only those indirect taxes levied directly on consumer goods and services purchased by households, or alternatively, indirect taxes levied on all goods and services (whether used for intermediate or final consumption) may be excluded. The former measure can be said to remove only the first round effects of changes in the indirect tax system by, for example, removing the effects of changes in tobacco excise duties and wholesale sales tax on the price of tobacco products, whereas the latter measure would also remove the second round effects of changes in petrol excise on the price of tobacco products.
6.21. As a result, there are four alternative methods for calculating net price indexes which are shown in diagram 7. It is necessary to decide not only whether a constant tax or a tax free index is preferred, but also whether adjustments should be made for only the first round effects of a taxation change or for all effects.
6.22. A tax free index may be calculated by excluding indirect taxes from both the current period and the base period. This has the effect of significantly altering the weighting system, with heavily taxed goods (e.g. tobacco) contributing a considerably lower weight to the net price index than to the HCP index. The interpretation of such an index, in the context of the potential applications outlined above, is not always straightforward. By contrast, a constant tax index can be directly compared to the HCP index to assess the impact of taxation changes. For example, when the HCP index shows an increase of 10%, a 5% increase in the constant tax index means that half of household inflation in that period was accounted for by changes in indirect taxation. The weighting pattern of a constant tax index is identical to that of the household index, because only changes in taxation since the base period are excluded. For these reasons, it is proposed that a constant tax net price index be developed.
6.23. The choice between whether to adjust only for first round effects or not hinges on more practical considerations. Adjusting only for taxes levied directly on consumer goods and services purchased by households is the most reliable method because data are readily available. To estimate the effects on consumer prices of taxes levied on intermediate goods and services, more information is required and the effect is most commonly estimated using Input Output analysis. Empirically, these latter round effects of a taxation change have been found to be relatively unimportant in comparison with the first round effects. On this basis, it is proposed to only exclude the first round effects of a taxation change on HCP.
6.24. As subsidies can be considered negative indirect taxes, they should be afforded symmetrical treatment in a net price index. However, unless the amount of the subsidy is directly relatable to either the quantity or the price of the product, it may not be possible to assess the impact of changes in subsidies for each product in each period. It is therefore proposed to directly exclude those residual goods and services (i.e. those not already excluded from the HCP index on the basis of having economically insignificant prices) which may be heavily affected by subsidies. Although this may be seen to be at odds with the treatment of indirect taxes, in the Australian context the effect on the index is assessed as minimal due to the expected small number of items involved.
6.25. The following quotation illustrates some of the problems recently experienced by the United Kingdom when they introduced a net price version of their Retail Price Index. This index, compiled monthly, is constructed as a tax free index, removing only the first round effects of taxation change.
‘Debatable issues related to
- subsidies for which, because of practical difficulty, no adjustments were made in spite of the fact that they have the opposite effect to indirect taxes and should therefore strictly be excluded,
- the prices of items such as television licenses, prescription charges and rail fares which are affected by Government policy but which are different types of taxes and are elements of a payment for a service,
- the weighting system of a net price index which is affected by the fact that some sections of goods subject to high rates of taxes and duties, will have a much reduced weight, and
6.26. The Australian proposal for a constant tax net price index (removing only first round effects) with provision for direct exclusion of residual items for which price change is largely affected by changes in subsidies or direct policy initiatives, would appear to avoid some of the problems experienced in the United Kingdom. A trend estimate of this constant tax index, adjusted for seasonal factors and volatility, would provide a measure of underlying household inflation.
- the lagged effects of duty changes and genuine price rises.’ (Central Statistical Office 1995, p. 8.)
Seasonal adjustment and smoothing
6.27. An underlying index should exclude seasonal and irregular price movements. There are a range of smoothing techniques which could potentially be used to specifically adjust for the effects of inherent volatility on prices. It is likely that standard ABS seasonal adjustment and trending techniques would be applied.
6.28. An analysis of the CPI over the period September 1980 to September 1990 showed that seasonal and irregular contributions to the CPI were small, each being about 1% of the original series (Zarb 1991). Further, both seasonal and irregular contributions to the CPI were found to be diminishing with time.