Feature Article 3: Forthcoming Updated National Accounts Annual Benchmarks
Each year the annual and quarterly national accounts are revised using updated annual benchmarks from annual supply and use tables. Annual supply and use tables provide a balanced measure of GDP from the production, income and expenditure approaches. Usually, only data for the previous four years are revised, however, for this annual cycle, the complete time series will be revised. The impact of the revisions will largely be seen in changes to the levels of GDP and component series rather than in revisions to the movements in these series. Further, because revisions to annual movements will be small the quarterly pattern of the time series will largely remain as current published.
The revisions come from four sources:
The purpose of this article is to inform users of the nature of the changes that are likely to occur when the revisions are first released in the annual national accounts publication Australian System of National Accounts (cat. no. 5204.0) on 2 November, 2005. Quarterly data consistent with these revised annual data will be released in the September quarter 2005 release of Australian National Accounts: National Income, Expenditure and Product (cat. no. 5206.0) on 7 December, 2005. Annual state accounts data consistent with these revised annual data will be released in the 2004-05 release of Australian National Accounts: State Accounts (cat. no. 5220.0) on 9 November, 2005.
- changes to annual source data due to changes in the methods used to collect and compile the annual economic activity data that form the basis of the annual supply and use tables
- the incorporation of updated benchmark information
- changes arising from work on the 2002-03 Information and Communication Technology (ICT) satellite account
- changes from improvements in methods used to estimate various components of GDP.
CHANGES TO ANNUAL SOURCE DATA
The primary source used to compile annual benchmark estimates of gross value added by industry and the compensation of employees and gross operating surplus components of income is the information that underlies the ABS publication Australian Industry (cat. no. 8155.0). The most recent edition of this publication was released in February 2005. Those estimates incorporate the impact of a number of significant changes to the way ABS collects and compiles business and economic statistics generally. These changes will have the largest impact on the annual national accounts series to be published in this annual cycle.
The most significant revisions to the data published in 8155.0 came from the introduction of changes associated with The New Tax System (TNTS) that had a number of significant implications for ABS business statistics. A description of the changes was initially presented in Information Paper: ABS Statistics And The New Tax System (cat. no. 1358.0) that was released on 26 April 2000. The paper foreshadowed changes in the statistical infrastructure used by the ABS to support compilation of ABS economic series. Information Paper: Improvements in ABS Economic Statistics [Arising from the New Tax System] (cat. no. 1372.0) released on 6 May 2002 provided further information about these changes and their treatment in statistical series.
Most of these changes directly impacted on the ABS Business Register (ABSBR), which contains a list of businesses from which samples are selected to collect data for ABS economic series, including the annual Economic Activity Survey (EAS). The series have been impacted by the changes in the following ways:
Annual benchmark information for the annual supply and use (SU) tables comes from a data set that is a combination of directly collected data from EAS and annual Business Income Tax (BIT) data. This combination creates the 'EASTax' data set. The changes to the ABSBR listed above impact on both elements of the EASTax data, and hence the aggregate impact of the changes have fed directly into the annual supply and use benchmarks leading to an overall increase in the level of output, intermediate consumption and GDP.
- the population of businesses has increased
- the statistical units model, i.e., the way in which business structures are represented on the ABSBR, is more closely aligned with taxation reporting requirements
- the industry classification of some business has changed
In addition, because the industry code being used for the majority of units is now sourced directly from the Australian Business Register, there have been shifts in the allocation of output and intermediate consumption, hence value added across industries. The main industries affected by this change are manufacturing, property and business services and retail trade.
In order to implement the revised level in the time series of national accounts, EASTax data were compiled on both the new and old basis for 2000-01. Data on the new basis was also available for 2001-02 and 2002-03. This enabled a good understanding of the changes in levels and the impact on movements from the changes to be assessed. The general strategy for backcasting prior to 2000-01 has been to preserve the existing movements by industry but to shift the level of the series to the new basis. Care has been taken to ensure that the revised time series are economically meaningful by using all available information.
The updated information from the annual EASTax data source will provide a comprehensive measure of the Australian economy. In the past the annual SU benchmarks were compiled using more industry specific information (for example, using annual manufacturing survey information for the manufacturing industry estimates). While these industry specific data were not inconsistent with the broader EASTax dataset there were some concerns about possible overlaps and gaps between the different surveys in terms of complete coverage of the economy. The application of the updated EASTax data to this round of annual SU tables has been completed in a far more comprehensive fashion and while there have been some shifts in the relative size of different industries the overall basis for the measurement of macroeconomic activity has been improved. Work is currently underway to further improve the integration of the range of economic surveys with a particular interest in better aligning product and industry level data and quarterly and annual data.
INCORPORATION OF UPDATED BENCHMARK INFORMATION
In addition to incorporation of updated benchmark information from the EASTax data source, a range of other updated benchmarks will be incorporated.
Capital formation on alterations and additions to dwellings
Estimates of capital formation on alterations and additions to dwellings are based on the compilation of benchmarks that use a variety of source data. A new benchmark for 2002-03 has been compiled through confrontation of data from the ABS Construction Industry Survey 2002-03, the ABS Household Expenditure Survey 2003-04 and the 2003 BIS Schrapnel survey on the Home Improvements Market in Australia. Interpolation between benchmarks and extrapolation from the latest benchmarks are based on quarterly building activity survey data for alterations and additions.
Output of financial intermediaries
Estimates of the output of financial intermediaries such as banks and credit unions use data on the balances of deposits and loans held by these institutions and the counterparties to those balances. This information is published quarterly in Australian National Accounts: Financial Accounts (cat. no. 5232.0). In 2003 the balances data in the financial accounts was revised to incorporate changes required to record the data on the conceptual basis presented in the international standards for national accounts, System of National Accounts 1993 (SNA93). This updated balances information has now been incorporated in the estimation system for measuring the output of financial intermediaries. There will be revisions to estimates of financial sector gross operating surplus and to consumption of financial services both by using industries and by the household and general government sectors.
Taxes on production
In the course of introducing the goods and services tax (GST) into the national accounts, data relating to collections of wholesale sales tax (WST) were removed from the accounts from the September quarter 2001 and later quarters. This was done to more accurately record the taxes on an accruals basis, as these transactions where considered to relate to production in previous periods. In the course of making these adjustments, the wine equalisation tax (WET) and the luxury car tax (LCT) were also inadvertently excluded. These ongoing sales taxes will now be included in the estimates resulting in an increase of the level of taxes on production for 2000-01 and later years.
Other annual source data
In addition to the changes outlined above updated data from the balance of payments, public finance and the Australian Prudential Regulatory Authority (APRA) have been incorporated.
CHANGES ARISING FROM ICT SATELLITE ACCOUNT WORK
The ABS has been developing a satellite account on the measurement of information and communication technologies (ICT) in respect of 2002-03. This work examines a range of ICT products and essentially involves confrontation of measures of these products through a supply and use framework, thereby improving the overall cohesiveness of our ICT-related estimates. This exercise has given rise to benchmark level and time series revisions for a number of national accounting aggregates, in particular to gross fixed capital formation on computer software and computer hardware and to certain components of household final consumption expenditure.
Benchmark estimates of gross fixed capital formation of computer software are produced for those years for which detailed information is available. Appropriate indicators are then used to interpolate between these benchmarks and extrapolate beyond the most recent benchmark year. Prior to the release of this publication, benchmarks for gross fixed capital formation of computer software (the most recent of which related to 1994-95 and 1995-96) had been moved forward solely on the basis of imports of computer software. Various data collected by the ABS have allowed us to now produce more recent benchmarks and, at the same time, the indicators used to interpolate between benchmarks and extrapolate beyond the most recent benchmark have been improved. Of particular note is that the number of computing professionals will now be used as the indicator for the series 'customised computer sooftware', since this is considered a more appropriate indicator for this type of product. Revisions to gross fixed capital formation on computer software extend back to 1984-85.
The 2002-03 estimate of gross fixed capital formation of computer hardware has been revised upwards based on data obtained from the 2002-03 ABS Government Technology Survey and the 2002-03 ABS Economic Activity Survey. Historical estimates for this item have been revised back to 1993-94.
Investigations associated with the 2002-03 ICT satellite account have also driven changes to the household final consumption expenditure categories of 'communications' and 'recreation and culture'. For communications, the 2003-04 Household Expenditure Survey (HES) has been used to create updated benchmark information on household spending on telephone carrier services. Data from the ABS Household Use of Information Technology Survey (HUIT) and the ABS Internet Activity Survey provide the basis for updated estimates of household expenditure on Internet provider services. The 2003-04 HES also provided the updated benchmarks for household final consumption expenditure on computer hardware and computer software. For these latter items, the indicators used to interpolate between benchmarks and extrapolate beyond the most recent benchmark have also been updated and are now based on information derived from HUIT and the ABS Census of Population and Housing.
Improved estimates of price change for computer software have also been developed. The current estimates for price change on the three types of computer software (prepackaged, customised and own-account) assume a constant price fall of 6% per annum. The new methodology provides separate price indexes for each of the three components. The US Bureau of Labour Statistics index for prepackaged software is used, adjusted for some exchange rate variation. The price index for own-account software was derived from labour price index data, adjusted for changes in labour productivity for the IT industry. The customised index is a weighted average of the prepackaged index and own-account software index. The revised indexes show an ongoing price decline for computer software but to a smaller degree than the current estimates. This will cause downward revisions to the volume measures of software investment.
CHANGES TO METHODS
Compensation of employees
The current method for estimating compensation of employees (COE) is complex and requires the use of a range of data sources to ensure complete coverage of all employees and the various types of remuneration. To date, the primary data sources for COE have been the Quarterly Business Indicators Survey (QBIS) and the quarterly Survey of Employment and Earnings - Public Sector (SEE). These two sources measure wages and salaries for the private and public sectors respectively. Other elements of COE such as superannuation contributions and payments in kind are sourced from the annual EASTax data set and some ATO data on fringe benefits. Some recent conceptual developments and data confrontation have led to a change in approach that will affect the level of the COE series but will not impact greatly on movements in COE.
It has been decided to incorporate fringe benefits tax (FBT) as part of COE. In the past FBT was treated as a production tax paid by business and, while included in the calculation of GDP it was not part of COE. The new treatment will leave GDP unchanged but will increase the income taxes paid by individuals in the household income account. Household saving will be unaffected. It has also been decided to change the treatment of the superannuation contributions tax. Currently, this tax is treated as income tax payable by corporations but this will be changed to be shown as income tax payable by individuals. There is no impact on GDP from this change but the level of household saving will be reduced. This will be offset by an increase in the savings of the corporation sectors.
A number of changes have been made to data sources. First, the annual benchmarks for wages and salaries for private sector units and public corporations will now be sourced from the annual EASTax data rather than by summing quarterly survey data. Data for general government units will continue to be sourced from SEE. This change means that around 80% of COE will come from a single data source. Second, a new benchmark for payments in kind has been introduced using data from the irregular Survey of Labour Costs 2002-03. Third, better measures of COE flows to and from non-residents have been introduced. Fourth, some small adjustments for salary sacrificed components have been introduced, based on ABS surveys that measured the extent of salary sacrificing. A comprehensive ABS wide information paper on this subject will be released later in 2005.
Agriculture industry value added
An improved derivation of intermediate use in the Agriculture industry will be introduced that will lead to changes in the estimate of value added. Previous estimates of intermediate use were derived as a residual, by deducting estimates of compensation of employees, gross operating surplus and other taxes on production from output. Intermediate use is estimated directly in the new method. This is possible because reliable external data sources and EASTax data is now available. Other methodological changes to measures of Agricultural industry value added include valuing agriculture output at gross value of agricultural production, rather than at local value of agricultural production, and using EASTax based wages and salaries and employer social contribution estimates.
The way in which data on agricultural income are published will also change. The separate Agricultural income tables that are published in the various national accounts releases will be maintained but there will be changes to some line items to better reflect the valuation concepts used in the supply and use tables.
Construction industry value added and associated outputs
Previously, the estimates of output and intermediate input recorded in the supply and use tables were on a net basis, that is transactions within the construction industry by sub-contractors and the like were not recorded. The improved methodology adopted for this update recognises these intra industry flows, i.e., the transactions are now recorded on a gross basis. This approach has been facilitated by data from the 2002-03 Construction Industry Survey, which will enable improved estimation and analysis for this industry into the future.
The outputs of the construction industry such as dwellings and other buildings are of particular analytical interest. Using data from the Construction Industry Survey 2002-03 and the new method outlined above, a data confrontation exercise was undertaken against the long standing ABS data sources for building and construction activity. It is difficult to reconcile the estimates from these different sources precisely for a variety of reasons. However, on the basis of this information and some additional demand side data available on the value of alterations and additions, improved estimates of the value of alterations and additions on residential buildings and non-residential buildings have been compiled. The overall effect will be an upward revision to the level estimates of capital formation, private and public, for alterations and additions to both residential and non-residential buildings. However, movement estimates will remain largely unaffected.
Retail industry value added
Annual estimates of value added for the Retail industry will be based on the annual EASTax data source described above. These were previously based on the 1998-99 Retail Industry Survey, moved forward using subannual indicators such as the Monthly Retail Trade series. The level of sales by retailers will rise as a result of this change, although movement estimates will be largely unaffected. This will in turn result in various upward adjustments to some categories of household final consumption expenditure.
Education industry value added and associated outputs
Estimates of value added for the education industry will now directly incorporate information from the EASTax data set. This information provides a measure of the activity of non-government schools and commercial education providers. The contribution of government schools and other government educational facilities are estimated separately, then consolidated with the above information. These estimates were previously compiled through a hybrid methodology that involved aggregation of various demand side estimates of education.
Adoption of this improved method for estimating the output of education industry highlighted a deficiency in the estimates of household final consumption on education. Specifically the current estimate of household final consumption of non-government schools output was found to be understated. This component will now be estimated as the sum of the recurrent expenses incurred by those schools in delivering their education service. The improved measures of output will lead to upward revisions in this component of household consumption on education.
Household expenditure on motor vehicles
Investigation has shown that, in recent years, household expenditure on motor vehicles has been significantly understated. The existing method estimates expenditure on motor vehicles by multiplying the number of new cars purchased by an average price. This average price is benchmarked infrequently and is moved forward using movements in the consumer price index for motor vehicles. However, the CPI series is adjusted for the increasing quality of motor vehicles thus leading to relative price falls in the CPI compared to changes in the nominal prices paid for cars by households. While this treatment is correct for CPI purposes it represents an understatement of the movements in actual prices paid by consumers.
A new national accounts methodology to estimate current price expenditure has been developed utilising a price index that is not adjusted for quality change. The volume of household consumption on motor vehicles will now be derived using the new current price series deflated by the quality adjusted CPI series. The methodology for estimating household expenditure on motor vehicles has also been altered to account for changes in the structure of the motor vehicle market, particularly the rising proportion of sports utility vehicles, and for improvements in data available on numbers of motor vehicles purchased by households.
Recording of general government transactions
Previously, data for the general government sector was compiled on a consolidated basis, i.e., by netting out transactions between general government units. This is also the basis used by the ABS to compile government finance statistics (GFS). In order to more completely measure activity recorded in the production account, adjustments have now been included to gross up the production account items affected by consolidation. These items are state payroll taxes (PRT) paid by government units and commonwealth fringe benefit taxes (FBT) paid by government units.
The effect of this change will be to increase the level of government output, thus value added for industries that include government units, and to increase the level of government final consumption expenditure. In the case of PRT the level of taxes on production will also increase. In the case of FBT, the level of compensation of employees will increase and taxes on income paid by individuals will increase, consistent with the changes to the treatment of FBT outlined above.
There is a substantial amount of change foreshadowed in this article. Nonetheless, employing new methodologies and data sources ensures that the best available data are being used in the compilation of accounts. Ongoing reviews of methods and data sources will continue although it is unlikely that revisions of a similar scope will occur in the near future.
Overall, while there will be a noticeable impact on the level of GDP as a result of the changes, our assessment is that there will only be a small impact on movements in GDP both in current price and in volume terms.
A more detailed statement of the revisions, including the magnitude of the impact of the individual changes listed above, will be provided in Australian System of National Accounts (cat. no. 5204.0) to be released on 2 November, 2005. Questions concerning the nature of the revisions should be directed to Ian Bobbin (email <firstname.lastname@example.org>, phone 02 6252 6908).