APPENDIX 5 ABS PROPOSAL FOR A MONTHLY CPI
1. The ABS has evaluated the viability of producing the CPI on a monthly basis and believes it can produce a monthly CPI measure of equivalent quality to the current quarterly measure. The monthly CPI would be constructed taking international best practice into consideration. This appendix outlines the current quarterly CPI processes, the procedures and costs for producing a monthly CPI. A comparison of selected international statistical agencies' monthly compilation procedures is also provided.
2. The current CPI is designed to measure average pure price movements over a period of time. Although it is not practical to measure all prices on this basis, the ABS adopts a number of strategies to approximate this concept across the index. Key to these strategies is the ABS practice of undertaking regular price collections throughout the quarter. Spreading the collection of prices from different outlets for particular products over the month is a practical and cost effective way of obtaining a set of prices representing the whole period. In addition, there are practical considerations which favour continuous price collection. Specifically the workload is more even, avoiding some of the operational problems associated with point-in-time collection. Pricing over most of the reference period also allows for smoothing techniques to be adopted. The general approach is to price each item as frequently as is necessary to ensure that reliable measures of quarterly price change can be calculated, resulting in a more robust CPI dataset.
3. Publishing the CPI quarterly provides sufficient time to resolve any quality change issues during the period in which they occur. Most prices are collected by personal visits to the selected outlets by trained ABS field officers, who observe actual market prices as well as discussing matters such as discounts, special offers and volume-selling items on the day with the retailers. The field officers record this information in handheld computers which facilitates interactive real time editing. Regular personal visits by field officers to the retail outlets also enable the field officers to actively monitor market developments such as market shares or possible quality changes. This information is used in maintaining the representativeness of the samples and making quality change assessments. This approach allows for two stages of price assessments. First, when field officers enter price data into their computers the figures are checked for validity. If there is a large divergence from the previous price, the field officer is asked to confirm the new price. Second, after the prices have been sent to head office they are also checked. Prices outside the upper or lower bounds for similar items are identified as outliers that require further investigation.
4. Collecting prices over the quarter is also consistent with other economic statistics, most particularly those feeding into the National Accounts.
ABS RECOMMENDATION FOR A MONTHLY CPI
5. The preferred option discussed below is expected to be achievable as it is derived from current ABS CPI procedures. The main features of the proposal are:
Definition: The ABS will maintain the current definition of the CPI (except for frequency). That is, the CPI will measure the average monthly price change of a 'basket' of goods and services which accounts for a high proportion of expenditure by the CPI population group.
Reference period: The CPI will have a reference period of one month.
Price collection period: Prices will be collected over four weeks of the month. Current ABS practices will be reviewed so the present quality assurance procedures are continued and the CPI is still able to be released in a timely manner. In particular, increasing the use of transactional and administrative by-product data is being investigated to assist in meeting the volume of quality assurance work in a tight time frame.
Release timeliness: The CPI will continue to be released within one calendar month of the end of the reference period as per the IMF SDDS.
Population coverage: All private households in the capital cities. If the target population of the CPI was to change it would have an additional impact on the costs of producing a monthly CPI.
Approximate price observations: 80,000 per month. Although this is a decline of approximately 20% compared with current observations of about 100,000 per quarter, it is in line with other international statistical agencies' practices and it is expected that this will not have an adverse affect on quality. To obtain a robust figure of average price change for the month, the requirements of the CPI subgroups are taken into consideration. The drop in price observations occurs because approximately 50% of the CPI, by expenditure weight, is already collected monthly and not all of these items are deemed volatile enough to be priced more frequently than monthly.
Benefits: The full economic benefits to the Australian economy are difficult to quantify at this early stage, however at least one benefit would be more timely adjustment of the cash rate target in response to a more timely indicator of inflationary trends. A monthly CPI will comply with IMF guidelines which will assist in international comparisons. Pricing over four weeks of the month is ideal to calculate a robust measure of price change for the month. It also allows enough time for processing techniques to be employed.
Challenges: Pricing over all four weeks does create some added processing risks for the ABS. It will put more pressure on data providers and staff to adhere to tight deadlines and investigate data issues. The ABS will have to adapt current procedures to process more prices in an efficient manner.
6. The presentation of the CPI publication will be reviewed to take advantage of publishing developments, but the content is expected to remain similar to the current publication. Data for the weighted average of the eight capital cities and for each capital city will continue to be released down to the expenditure class level within one month of the end of the reference period. The index numbers as well as percentage change from the previous month and corresponding month of the previous year will be released.
7. A number of users have expressed concern about the possible absence of a quarterly index. In addition to publishing monthly and annual CPI figures, standard quarterly indexes (i.e. in respect of the March, June, September and December quarters) will continue to be published. They will be calculated as an arithmetic average of the three monthly index numbers, similar to the UK approach. This is also the current methodology for obtaining an average quarterly figure for the indexes presently compiled monthly in the CPI.
8. Some users have also expressed concern that monthly seasonality may increase the 'noise' reported in the CPI. The ABS will continue to release the Special series in the CPI publication. The Special series show the headline CPI excluding a particular group of items. This shows the effect of large, one off movements such as annual increases in electricity or education costs. The Analytical series will also be continued in a monthly release. Trend and seasonally adjusted estimates of the CPI are discussed in Chapter 6.
9. Currently the CPI is published four times per year. Additional funding will be required to publish the CPI an additional eight times per year (i.e. twelve monthly releases). It is projected that the CPI cost will increase from approximately $10 million to $25 million per annum. The costs have been calculated based on the ABS preferred monthly option outlined above and are based on the assumption that the quality of the CPI is maintained and that there are no other major changes to the CPI and the current suite of ALCIs. Any additional changes such as increasing the population coverage, adopting more frequent weight updates or using scanner data will have an additional impact on the cost of the CPI.
10. These costs are in line with expectations as currently around 50% of the CPI, by expenditure weights, is collected at least monthly. The number of price observations per annum will increase from approximately 422,508 to 960,000. In the first three years of the project a one-off investment of $6 million will be required for aspects such as redesigning and implementing a new CPI processing system and increasing sample sizes.
11. A monthly CPI will be first released within two to three years of receiving funding. It is estimated that it will take one year to evaluate and improve CPI price samples, increase the number of staff and to implement computer systems. The ABS then requires at least a year to run the monthly series parallel to the current quarterly series to evaluate the quality and robustness of the index. This will also ensure that a ‘percentage change from the corresponding month of the previous year’ figure can be included in the first release. Assuming funding was received for 2011-12, the earliest a new monthly series would be published would be in July 2013.
12. Australia and New Zealand are the only Organisation for Economic Cooperation and Development (OECD) countries that do not publish a monthly CPI. Outlined below are some common variants on the monthly CPI used by international statistical agencies. It should be noted that practices in other countries are not necessarily considered best practice for Australia. If the ABS is to produce a monthly CPI, the quality of the index as it is currently produced, will be maintained.
Table 1 - CPI compilation information(a)
|Country ||US. ||UK. ||Netherlands. ||Canada. ||Australia. |
|Reference period ||Monthly. ||Monthly. ||Monthly. ||Monthly. ||Quarterly. |
|Price collection period ||4 weeks. ||1 day. ||First 3 weeks. ||First 3 weeks. ||First 11 weeks. |
|Definition ||Measures pure price change in a market basket of goods and services of constant quality. ||Average measure of change in prices of goods and services bought in the UK for the purpose of consumption by all UK households, foreign visitors and residents of institutional households. ||Indication of the price increase or decrease of a basket of goods and services purchased by an average Dutch household in one year. ||An indicator of the changes in consumer prices experienced by Canadians through time and obtained by comparing the cost of a fixed basket of commodities purchased in a particular year. ||Pure price change in a selected basket of goods and services typically purchased by Australian households. |
|Release timeliness ||Approximately 2 weeks (14-31 days) after the end of the reference period. ||2nd or 3rd Tuesday (6-20 days) following the end of the reference month. ||1st or 2nd Thursday (1-13 days) following the end of the reference month. ||16-22 working days after the end of the reference period and no later than a month. ||4th Wednesday (21-26 days) following the end of the reference period. |
|Population ||All urban consumer units. ||All private UK households, foreign visitors and residents of institutional households. ||Covers expenditures on the Dutch territory and abroad from all Dutch resident households. ||All Canadian families and individuals living in urban and rural private households. ||All private households in the capital cities. |
|Coverage of the population ||Prices are collected monthly for food and energy items in all cities and for all items in the three largest cities. Prices for the remaining items are collected every second month. ||All of the UK. ||All cities with more than 100,000 inhabitants and a sample of cities with between 10,000 and 100,000 inhabitants. Municipalities with less than 10,000 inhabitants are excluded. ||Prices are collected in 15 to 76 locations based on the price behaviour of the commodity. ||The eight capital cities |
|Approximate price observations ||78,500 prices from 25,500 outlets. ||120,000 prices from 20,000 outlets. ||65,000 prices from 9,000 outlets plus supermarket scanner data. ||60,000 prices from 7,000 outlets. ||100,000 prices from 10,300 outlets. |
|Other factors ||Use of hedonics for quality assurances. ||Point in time inflation. ||Use of scanner data. |
|Contracts out price collections. ||No regional CPI indices are published. |
|Benefits ||Coverage and time to adopt smoothing techniques. ||Time for quality assurance. ||Time for both quality assurance and smoothing techniques. ||Time for both quality assurance and smoothing techniques. ||Time for both quality assurance and smoothing techniques. |
|Drawbacks ||Reliance on modelling, limited time for quality assurance. ||Limited scope for smoothing techniques, logistically costly. ||Not collecting prices over the entire reference period. ||Not collecting prices over the entire reference period. ||Reference period is not compliant to IMF guidelines. |
|(a) Information obtained from the IMF SDDS website. (http://www.imf.org/external/data.htm) |