1350.0 - Australian Economic Indicators, Nov 1998
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 30/10/1998
Page tools:
 Print Page
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Feature Article - Modifications to Reserve Bank Of Australia's Commodity Price Index
This article reports on the changes. They were adopted in the October release of the RBA Commodity Price Index on Thursday, 1 October 1998. SIMPLIFYING THE CONSTRUCTION OF THE INDEX The two most commonly used formulae for constructing a price index are the Paasche and Laspeyres indexes. A Paasche index uses weights determined in the current period to establish the relative importance given to individual components - in this case, individual commodities - in order to compare current prices with those of a base period. A Laspeyres index uses base-period weights in making the same comparison. In tracking an index through time, the weights of a Paasche index will change as the composition of commodity exports changes, whereas for the Laspeyres index, there is no movement in the weights over time. There is no unambiguous theoretical rationale for preferring one of these two types of index over the other (Endnote 1). The current Commodity Price Index is a modified-Paasche index. (It is modified in the sense that the quantities used to determine the current period weights are moving averages of the physical amounts of each commodity exported in the preceding 12-month period, rather than just those in the current period.) The main reason for choosing a Paasche index was that this most closely corresponded with the way the export price deflator has been calculated in the National Accounts (Endnote 2). The use of a Laspeyres formula for the Commodity Price Index has two practical advantages. The first is that the data requirements are less onerous - since weights are determined in the base period only - and the Index is therefore less subject to revision. The second advantage is that it is simpler to interpret. Movements in the Index from one period to the next would simply reflect underlying price movements a 10% rise in the prices of all commodities in a given month, for example, would always be reflected in a 10% rise in the Index, regardless of quantity changes. In contrast, movements in a Paasche index reflect changes in both prices and quantities in the period. Reflecting these considerations, the Commodity Price Index is to be constructed as a Laspeyres index commencing in October 1998. A technical description of the Index appears in the Appendix. RE-BASING THE INDEX To maintain the relevance of the Commodity Price Index, the base period used in its calculation has been updated every five years. (Periodic re-basing is desirable regardless of the index formula used.) The last time this occurred was in 1993, using a base period of 1989-90, the same as for the National Accounts. The National Accounts are now to move to annual re-basing, but the Bank has decided to retain a five-year re-basing period for the Commodity Price Index. The Index will be re-based to 1994-95. The desirable frequency of re-basing reflects a trade-off between the short-run and longer-run properties of the Index, the nature of which, in turn, will depend partly on the volatility of the component series. Frequent re-basing has the advantage of ensuring that weights are up-to-date, which helps to ensure that short-run movements are accurately measured. The disadvantage is that erratic price and volume fluctuations, by being built into the base period, would cause the level of an index to drift from its initial level, such that the Index would not necessarily return to its starting point even if all individual prices did so. Longer-run comparisons can therefore be rendered less reliable. The potential for drift is most serious when the underlying volumes and prices are relatively volatile, as is the case for the Commodity Price Index. The decision to retain the five-year re-basing period reflects these considerations. (Endnote 3) CHANGES IN THE COVERAGE OF THE INDEX The Commodity Price Index has been designed and constructed with a view to providing information about an important source of fluctuation in export income. This has, in particular, motivated the use of export weights for its construction. (An index designed to provide information about movements in commodity producers incomes, by contrast, would ideally use production weights in its construction.) The further question arises as to whether the weights should reflect gross or net exports of each commodity. It has been decided to retain the current approach of using gross export weights, but to remove crude oil, since Australia is a net importer of crude oil. The other commodities in the Index for which there are significant imports are gold (where imports amounted to around 16% of exports in 1997) and aluminium (11%), but in both cases the adoption of net export weights would have a negligible effect on the Index. IMPACT OF THE CHANGES The new weights, which incorporate the three sets of changes outlined above, are shown in Table 1; the weights in the current modified-Paasche index are also shown. Table 1 highlights the growing relative importance of resource commodities. Under the old weighting scheme these commodities accounted for 59.2% of the Index. They will now account for 66.8% of the Index. This reflects both a relative decline in the price of rural commodities - the rural price component of the Commodity Price Index, measured in SDRs, has fallen by around 35% since the beginning of the decade, compared with a fall of around 18% in the non-rural component - and sizeable increases in resource commodity export volumes.The impact is most notable for wheat (the share of wheat has fallen by 4.4%), cotton (by 4.4%) and wool (by 4.3%); the commodities with the largest increase include gold (4.2%), sugar (2.7%), aluminium (2.5%) and beef and veal (2.3%). Graph 1 illustrates the impact of the changes on the history of the index (Endnote 4). It is apparent that these changes do not greatly affect the overall trends in the Index, although the Index is slightly less volatile. For example, the rise in commodity prices from the trough in late 1993 to the peak in March 1997 was 18.1% on the new Index, compared with 23.1% on the old basis; similarly, the fall in commodity prices from March 1997 to August 1998 has been 17.3% based on the new Index, compared with 20.2% based on the old Index. Most of the difference between the two series is attributable to the move to the Laspeyres index, rather than the re-basing of the Index or the exclusion of crude oil. The changes do not vary the interpretation of movements over the past 12 months. FURTHER INFORMATION To subscribe to the RBA monthly Bulletin or to order reprints of RBA articles, contact the RBA Information Office on telephone (02) 9551 9724 for subscriptions or (02) 9551 9722 for reprints. RBA articles can also be obtained from the Bank’s web site at http://www.rba.gov.au, under Reports, Statements and Bulletin Articles. Table 1: COMMODITY PRICE WEIGHTS ALL ITEMS INDEX; per cent
(b) Laspeyres index using 1994-95 as the base period. 1 RBA COMMODITY PRICE INDEX SDRs; 1994-95 = 100 ENDNOTES 1. For a discussion of the relevant merits of the Paasche and Laspeyres indexes, as well as a number of alternative indexes, see Johnson L. (1996), Choosing a Price Index Formula: A Survey of the Literature with an Application to Price Indexes for the Tradable and Non-tradable Sectors, Australian Bureau of Statistics Working Papers in Applied Econometrics and Applied Statistics No. 96/1. <back> 2. The Australian Bureau of Statistics (ABS) is soon to vary the way the export price deflator is derived. It will remain a Paasche index, but the volume estimates used to derive the export price deflator will be derived from annually chained volume data, rather than volume data chained every 5 years. For reasons discussed below, it has been decided not to apply that approach to the Commodity Price Index.<back> 3. Both the United Nations, in its 1993 System of National Accounts, and the ABS, in Information Paper: Introduction of Chain Volume Measures in Australian National Accounts (ABS cat. No. 5248.0) note that in series where prices can vary quite widely from period to period, it is preferable to re-base less frequently than annually.<back> 4. In order to provide a time series back to July 1982, the re-based Index is spliced to earlier Indexes at July 1989 (based on 1989-90 weights) and July 1984 (based on 1984-85 weights) using the ratio of the re-based Index to the earlier Index at that time as a splicing factor.<back> APPENDIX The RBA Commodity Price Index will be calculated as a fixed-weight Laspeyres index. The Index will be calculated as follows: where: The denominator measures the value of commodity exports in the base period (in the case of the latest base period, average export values in 1994-95). The numerator measures export volumes in the base period at current prices. Alternatively, Equation (1) can be expressed as: where: Thus the Index is a weighted arithmetic average of the individual commodity price series. The weights used correspond to the export value of each commodity as a share of total commodity exports. Document Selection These documents will be presented in a new window.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
