APPENDIX 5 METHODOLOGICAL ISSUES
The release of SNA93 with its recommendation that statistical agencies treat computer software spending by producers as gross fixed capital formation rather than intermediate consumption expenditure brought into focus the understanding and measurement of computer software. The intangible nature of software, the ways in which it is produced and copied under licence, and deficiencies in business accounting for software have provided significant conceptual and measurement challenges for official statisticians.
There has been considerable work undertaken recently towards developing an understanding of, and consensus on, how software production and distribution should be treated in the national accounts and how some of the practical data difficulties might be overcome. Key international forums advancing this work include the OECD's Task Force on Software Measurement in the National Accounts (hereafter referred to as the OECD software task force) and the Canberra II Group on the Measurement of Non-Financial Assets.
This appendix describes the concepts and methods used to generate estimates of software supply and use for the various types of software included in the satellite account. It describes the notion of a software 'original' within a broader discussion of the production of packaged software. Two areas of interest associated with packaged software are the treatment of computer games, and data collection difficulties caused by the widespread practice of selling computer hardware and software as a 'bundle'. A brief description of relevant issues is provided. Australian production of software relates mainly to customised software and own account software. Measuring the latter presents a considerable challenge to national accountants and detailed attention is provided in this appendix on the estimation methods used in the satellite account. Finally, this appendix describes how gross fixed capital formation on computer software has been derived within the supply and use balancing process, and how industry-based estimates of gross fixed capital formation on computer software were produced.
Broadly speaking, software can be can be divided into original software and reproduced software. Original software can be created for the purpose of reproduction (of so called packaged software), for on-sale, or it can be used for direct input into the creation of goods and services more generally, including other software. As mentioned, software can be readily reproduced at minimal cost and on-sold to others. This characteristic underlies much of the debate about its treatment in the national accounts.
Packaged software: the production process and accounting for licence payments
It is helpful to describe how packaged software is typically produced and how this production process is treated within the framework of the national accounts.
Packaged software generally reaches its buyer through a three stage process. The software production process commences with the creation of an 'original' piece of software which is clearly an asset and is treated as gross fixed capital formation as long as it is used in production for longer than one year. The second stage relates to reproduction of the software 'original'. The vast majority of packaged software purchased in Australia is copied from an overseas-produced original. The reproduced software may be imported in a 'boxed' format (i.e. disk(s), manuals and packaging) though increasingly, a copy of the 'original' is sent over the Internet or provided on disc and reproductions of this 'original' are then made in Australia. In the latter case, licence fees or royalties are paid to facilitate reproductions of the 'original' ('licence to reproduce'). In Australia, reproduction of the 'original' is typically organised by a wholesaler. The third stage in the software production process involves the software user securing use of the reproduced copy through 'licence to use' payments.
In the national accounts and the satellite account, payments relating to licences to reproduce and licences to use receive different treatments. SNA93 recommends that 'licence to reproduce' payments be treated as intermediate consumption by the licencee and as service income by the owner of the original (paragraph 6.146). They represent payments for capital services rendered by the software original and are not considered to be investment spending.
The more difficult issues surround the treatment of licences to use software. Software can be 'purchased' in many different ways but typically the owner of the original will retain certain rights over the use of the software. Notwithstanding these conditions, software copies can be purchased outright in the form of a once only payment of a licence fee, or of regular licence payments that may or may not include the provision of updates and other services such as help desk. Recent international discussions by an OECD software task force considering the update of SNA93 concluded that the use of software copies by the final user should be treated as investment if this user is a business or government unit who intends to use the software for longer than one year. It should be treated as final consumption expenditure if it is purchased by a household. In all cases, the software copy is regarded as separate to the original.
The value of packaged software to be recorded as gross fixed capital formation is straightforward where a once only payment for the use of the software is made, as the payment can be considered the market value of the asset. However, there are complications in situations where contracts require regular licence payments to be made. The OECD software task force report identified three different scenarios:
Bundling of packaged software
In many cases, vendors sell computer hardware and computer software together as a 'bundle' and the purchaser will not necessarily know how much of the purchase price relates to the computer software component as opposed to the computer hardware component. When reporting spending on computer hardware and computer software to ABS surveys, respondents are asked to estimate the relative proportions of a bundled purchase. However, in some cases respondents are unable to do this and the combined total is then simply reported against computer hardware. The practice of 'bundling' will lead to some understatement of reported computer software spending and corresponding overstatement of computer hardware spending.
- initial payment followed by smaller maintenance updates payments. In this case the up-front payment can be considered the initial acquisition cost and the series of smaller payments are for the rights to updates. Both payments are treated as gross fixed capital formation.
- regular annual payments. Where the contract requires regular payments over the expected lifetime of the asset it has the characteristics of a finance lease and the software is therefore considered to be the gross fixed capital formation of the lessee. Where the contract agreement is only for one year but annual payments are made to extend the licence, the situation is more akin to an operating lease where the lessee acquires a service from the lessor. However, the task force concluded that if the intent was for the lessee to use the asset for more than one year (i.e. renew the licence) then it should be regarded as gross fixed capital formation of the end user.
- licences of less than one year - the software is treated as consumption expenditure of the licence holder.
There are no international standards or guidelines on whether computer games should be included in the definition of 'computer software'. While there are arguments either way, in principle we would prefer to exclude them from the ICT satellite account, although in practice this can only be done in a limited way. One reason for the 'in principle' decision is that the same software will drive applications run on a range of different platforms, some of which are clearly not computer hardware according to internationally accepted product definitions. Also, while some games can be networked, many are not.
Australian production of software occurs in all industries, including government administration. While much of this is centred around the ICT specialist industries (where customised software is mainly produced), substantial amounts are also produced in other industries, especially own account software. It is this latter component that causes the most difficulties.
ICTIS is the source of information for software production by the computer services industry grouping, with estimates based on revenue from the provision of software services. It also provides information for the manufacturing industry which contributes to software production through copying and packaging of software.
Customised software is generally produced under contract to a particular client rather than being intended for reproduction. It is generally provided by businesses in the computer services industry and one practical issue is establishing the boundary between software production and computer services more generally. Typically, the following services are provided within the computer services industry:
Category 1 relates to customised software. Category 2 includes such things as functional analysis and detailed analysis, and is considered largely related to customised software. Much of software maintenance (Category 3) is in reality an improvement to existing software (since software does not wear out like tangible capital goods) and could also be considered a customised software product. The remaining categories making up 'computer services' do not relate to the creation of software.
1. Customised software services and solutions (including web site design and other Internet applications)
2. Computer systems analysis
3. Software maintenance
4. Computer hardware consultancy
5. Computer hardware maintenance and repair
6. Disaster recovery
7. Data processing services
8. Computer time sharing
9. Information storage and retrieval
Although ABS surveys ask for details of income from businesses in the computer services industry, in some cases they are unable to itemise their receipts to accurately identify the customised software service component. In these cases, it was necessary to estimate the proportion of total computer service income relating to customised software. For example, ICTIS provided information on income from web site design, other Internet applications and other customised software services. These are all considered to be related to customised software. ICTIS also captured income from 'other computer consultancy services'. Based on a general understanding of the composition of this item, 60% of this category was allocated to customised software production.
Own account software
Own account software describes that software developed by an entity's own employees for its own use. It is valued according to its estimated market value, or if this is not possible, the market value is approximated by summing the costs of production (including both labour and non-labour costs, and consumption of fixed capital). Own account software is regarded as being produced by the business or government entity and is therefore part of its output and gross value added. As it is a capital item produced for own use, it is also regarded as gross fixed capital formation by that business or government entity. Own account software ranges from the development of large scale systems in large organisations to small scale building of databases around packaged software. In that sense all industries contribute to the wider 'economy-wide ICT industry' and are included in the scope of the satellite account.
In principle, both own-account and customised software under development should be recorded as work in progress until finished and available for use. At that point, both become part of gross fixed capital formation. In practice, data constraints mean it is most likely that gross fixed capital formation will be recorded as work progresses. This is the approach used in the satellite account.
Deriving values for own-account software production of industries outside the computer services industry grouping is a vexed issue. ABS and international experience has been that it is not possible to collect satisfactory information directly from businesses on the costs of own-account software development. During the collection of EAS and GTS 2002-03, Australian accounting standards in force at that time allowed for the recognition of internally generated intangibles as assets. Nevertheless, accounting practice was mixed. ABS experience from EAS 2002-03 was that only very large, discrete software developments had a strong chance of being capitalised and therefore the reported EAS value was substantially understated. On the other hand, general government sector units were more likely to record internally generated software as assets. The latest International Accounting Standards generally prohibit the recognition of internally generated intangible assets and so the costs are now likely to be written off, generally indistinguishably from other business expenses. The adoption by the Australian Accounting Standards Board of the International Accounting Standards from 2005 means that internally generated software will now be recognised in business accounts as assets only under very restrictive circumstances.
In recognition of these practical data collection problems, the OECD software task force recommended that:
'...In practice, business reports on software capitalisation underestimate software capitalisation and may be affected by changes in tax regulations and business practices. As a consequence, member countries are recommended to implement an estimate of gross fixed capital formation in software independent from the estimate derived from business reports on capitalised software.'
The task force further recommends an indirect macro-approach to the estimation of own account software using data on the number of software professionals and other information. That is:
'...best practice for the estimation of own-account software investment at macro level is the following;
1. Estimate the labour cost of own-account software
= Labour cost of software professionals (number of software professionals * Average compensation).
Adjustment 1: Exclude labour cost linked to the production of software to be sold (however, do not exclude labour costs for originals for reproduction).
Adjustment 2: Exclude labour cost linked to other activities (maintenance, management, etc.).
2. Add non-labour costs of own-account software (intermediate consumption, consumption of fixed capital and net operating surplus).
Adjustment 3: in adding non-labour costs, avoid double counting costs that have already been recorded as purchases.
This approach is essentially that used in the ASNA and in the satellite account. The exception being that the value of output (and investment) estimates used here and in the ASNA do not include an imputation for consumption of fixed capital and net operating surplus. Insufficient data exist to reliably estimate these components and the valuation therefore falls somewhat short of a full market price equivalent. The estimates for investment in own account software included in the satellite account were introduced into the ASNA 2004-05 release in late 2005. Previous estimates in ASNA were based solely on labour costs.
The method used to derive estimates of own account software in the ASNA and in the satellite account involves many assumptions and approximations and the estimates should be regarded as 'indicative' only. The method is described in some detail below.
Operation of the own account software estimation model
The five step model described below was used to generate own account software estimates for all sectors. Estimates for general government - based on adjusted GTS data - were then deducted from this total to derive the estimate for private/public businesses. As mentioned earlier, this residual estimate is significantly higher than that directly collected from businesses in EAS 2002-03.
Step 1. Number of software professionals:
The starting point is estimating the number of software professionals employed. In Australia, LFS collected total number of 'computing professionals' (ASCO 2231) employed in 2002-03. The LFS figure includes contractors who are, technically, self employed.
Step 2. Proportion of time 'software professionals' spent writing software:
Persons classified as 'computing professionals' will not spend all their time working on software development. Also, not all stages in the development of a software product should be considered part of the valuation of the final product. The OECD software task force has described the following stages in building software:
According to the software task force, only those costs associated with stages 2 - 6 should be included in the value of the own account software asset. Activities such as LAN administration, help desk operation, IT security and so on are excluded. In addition, there are other non-software related activities undertaken by 'computing professionals' such as those associated with hardware architecture and maintenance.
1. Feasibility analysis
2. Functional analysis
3. Detailed analysis
An important input to the own account software estimation model is the proportion of time computing professionals spend writing this type of software. Case studies of several large organisations were used to estimate this proportion.
Step 3. Estimate (and remove) time spent producing software for market output:
Of those computer professionals assumed to be producing software, a proportion will already be contributing to the market output of businesses in ICT specialist industries and will therefore need to be deducted to avoid double counting with customised software. ICTIS publishes estimates of 'ICT employment numbers' for each ICT specialist industry and this source was used to make an appropriate deduction. An estimate of ICT employment associated with production of customised software was then required.
The ICTIS definition of ICT employment includes all those:
'...who spent the majority of their time engaged in ICT activity and includes help desk staff, information technology managers, electronics engineers, systems managers, administrators, analysts, designers and programmers, application programmers and computer and communications technicians. Data entry and call centre staff are excluded.'
This ICTIS measure is therefore broader than ASCO 2231, but since only ICT employment of the computer services industry was used in the calculation of customised software, the correspondence should be quite reasonable (the computer services industry will, for example, employ minimal numbers of telecommunication technicians).
Employment related to writing software for own use is therefore derived as employed 'computing professionals' (ASCO 2231) from LFS, less ICTIS-based 'ICT employment' in the computer services industry.
We can be certain that these persons are not exclusively engaged in writing software and, as stated, we estimated the proportion of time that computing professionals spent writing own account software.
Step 4. Apply earnings data to time spent producing own account software:
Using these assumptions, and using an average weekly wage cost derived from the Survey of Employment, Earnings and Hours for ASCO 2231, the salary cost directly associated with production of own-account software is derived.
Step 5. Estimate relevant non-labour costs:
In addition to labour costs, related non-labour costs need to be incorporated in order to produce estimates on a 'sum of costs' valuation. Again, there is limited information available to do this, and an analysis of several large own account software developments was undertaken to determine the ratio of wage costs to related non-wage costs for this type of work.
Imports and exports
The OECD software task force identified a number of ways in which software can be traded internationally:
These transactions may be recorded as trade in goods or services, including licence and royalty transactions. This can contribute to measurement difficulties and will to lead to understatement where customs information alone is taken into account. The ABS takes account of associated services flows through its Survey of International Trade in Services, though the capturing of these flows is not straightforward.
- Physical disc (including manuals) where valuation is based on the medium rather than how it is used
- Installed on equipment
- Physical or on-line copy where licences are paid to use one or multiple copies
- Physical or on-line copy for reproduction and on-selling (including bundled with hardware)
- Customised software in physical format
- Informal purchases on-line
- International transactions between affiliated businesses.
Supply and use of software
Although the ABS collects a wealth of ICT-related data compared with many countries, the nature of software means that more than the usual amount of judgement and approximation were needed to integrate the data within the supply and use framework. This problem has been recognised internationally, as many countries have observed substantial differences between data for the supply of software into the economy and data for purchases of software by businesses, government and households. Particular problem areas identified relate to the difficulty of measuring imports and the production of own account software on the supply side, and the tendency for business surveys (such as EAS) to understate business and government investment on the use side.
Broadly, the approach adopted for valuing software in the satellite account has been to examine and confront the data available from the various sources and, after investigation, make judgements about what data are likely to be of the highest quality. These data are given preference and are used as benchmarks to adjust data considered of lower quality or to derive some items as a residual. Generally, the estimates of supply into the economy have been used to benchmark the data for use of software and gross fixed capital formation has been derived as a residual.
Gross fixed capital formation, by industry
Estimates of gross fixed capital formation by industry have been derived for own account software and for purchased (packaged and customised) software. The industry estimates for own account software are based on LFS employment numbers for 'computing professionals', after deducting an estimate of numbers of workers engaged in customised software production. The industry estimates for gross fixed capital formation of purchased software are necessarily based on the directly collected (EAS and GTS) investment data--adjusted to the benchmark derived from the supply side on a pro rata basis.
'Computer services' as a product in the satellite account covers a range of activities, including maintenance and repair of ICT products, disaster recovery, data processing services, computer time sharing, information storage and retrieval and help desk functions. The prime data source for information on computer services (ICTIS) includes customised software solutions within its broader description of 'computer services'. In the satellite account, 'computer services' excludes production (including enhancement) of computer software.
Separating 'customised software' solutions from 'computer services' is a necessary step, since the national accounts treat customised software as a capital item and spending on computer services as intermediate and final consumption. They are separate ICT products. Issues relevant to this step are described earlier in this appendix.
In relative terms, imports are not overly significant to the supply of computer services, and the estimate of domestic production therefore represents the majority of supply of computer services. Supply side estimates have a much firmer basis than do the demand side numbers and they therefore underpin the estimates on the use of computer services. Households consume relatively small amounts of these services, with the largest proportion being consumed by business and government units as a current expense. Exports of computer services are thought to be relatively small. Since we do not have information on business and government spending on computer services, intermediate consumption has been derived as a residual.
The industry-based estimates of intermediate consumption of computer services are derived using a composite indicator of industry activity made up of industry data on net capital stock of computer hardware, net capital stock of computer software, and industry expenditure data (from EAS and GTS) on ICT contractors and consultants. All three factors are equally weighted to form an indicator that is applied to the Australia total.
Understatement adjustment applied to ICTIS data
Because ICTIS includes only employing businesses, it was decided to include an upwards adjustment to certain ICTIS results to estimate for the contribution of non-employers. The ABS undertook a study to establish the extent of non-employers in industries covered by ICTIS. Such a study is only indicative and drawing firm conclusions can be difficult. Nevertheless, it provided a very useful broad indication of the likely understatement, especially when combined with an informed view of business structures typically used within these industries. For example, it is clear that the provision of telecommunication provider services is dominated by large, employing firms and we are comfortable in concluding that no adjustment to ICTIS data for non-employers was required for this industry.
It is clear that significant numbers of non-employers operate in the computer services industry. Therefore, for the purpose of the satellite account, an 8 per cent upward adjustment based on the study mentioned above has been applied to the ICTIS figures for computer services income of the computer services industry. No adjustments for non-employing businesses have been made for any other industry.
Broadly defined, 'telecommunication services' is made up of 'phone carrier services' and 'Internet provider services'. The supply and use of each of these components has been independently derived and is described below.
Phone carrier services
Phone carrier services represent the largest single ICT item in Australia. Australian production primarily occurs in the telecommunication services industry, with minor contributions from ICT wholesalers and the computer services industry. Estimates of Australian production of phone carrier services are derived from the ICTIS collection which provides a breakdown of the various types of phone carrier services including, for example, data and text services, short messaging services (SMS), other mobile and paging services, as well as basic telephony services. Estimates of international trade in these services are generated within the BOP system, while HFCE of phone carrier services has been estimated using the 2003-04 HES.
Intermediate consumption of phone carrier services by business and government has been collected through the EAS and GTS collections. However, supply side and HFCE estimates are considered more soundly based for this product and intermediate consumption has therefore been derived as a balance within the ICT product supply and use framework. It is encouraging to observe that the figure derived as a residual for intermediate consumption is almost identical to the figures reported in EAS and GTS. EAS and GTS have been used to generate industry-based estimates of spending on phone carrier services.
Internet provider services
The value of total supply of Internet provider services is collected through ICTIS. A small number of ISPs are not collected by ICTIS, for example, ISPs run by community organisations; however the impact of these ISPs is thought to be minimal. Estimates on the use side are not directly available from ABS collections. An indirect method of estimating household spending on Internet provider services has produced results of defensible quality (see below). Intermediate consumption of Internet provider services is derived as a residual.
Household final consumption expenditure
The satellite account generates estimates of HFCE on Internet provider services through a method utilising data from HUIT and the IAS.
HUIT provides detailed information on numbers of households with 'Internet access'. This information has been collected for some time and, as might be expected, the resulting time series shows strong growth in numbers of households with Internet access. IAS supplies information on the proportion of households with dial up Internet access and with broadband access (including a breakdown across various ranges of broadband download speed). Together these data allow the construction of a profile of the numbers of households with Internet connections, by type of Internet connection. To this we add observed prices for monthly ISP costs (for dial up access, basic broadband and high speed broadband packages) to derive HFCE on Internet provider services.
The model rests on a number of assumptions, one of which is that household Internet connections are used only for household purposes and not for business-related purposes. On the other hand, we have used monthly prices for Internet Service Provider (ISP) services that do not include an estimate for Internet connection charges, or for ISP charges for 'excess' use. Internet connection charges are difficult to measure because HUIT data provide only a net increase in Internet connections by households (i.e. connections less disconnections), and because ISPs frequently offer free Internet connections. There are no data available on ISP charges for 'excess' use. In short, there are a number of factors that we are unable to reliably account for but the net overall impact of these assumptions will be at least partially off-setting. The resulting estimates of HFCE on Internet provider services are considered defensible.
MEASUREMENT OF TELECOMMUNICATION ASSETS
Supply of telecommunication assets is comprised of imports and domestic production, with own account production (construction) of telecommunication assets being a notable inclusion.
Estimates of imports of telecommunication equipment are drawn directly from the ABS International Trade System. Because these are goods imports and are therefore generally physically shipped into Australia, customs information are believed to provide good coverage. Mobile phones comprise over one third of the value of these imports.
Domestic production of telecommunication assets is made up of two broad components--the manufacture of telecommunication equipment in Australia and the construction of telecommunication assets on own account by telecommunication service providers. Data on the first component are collected by the ABS from relevant Australian manufacturers. Estimation for the second component is described below.
On the demand side, HES provides the basis for estimation of HFCE, while EAS and GTS collect gross fixed capital formation on telecommunication equipment. Telecommunication equipment is a narrower concept than telecommunication assets - the latter includes own account (construction) work on telecommunication structures. Intermediate consumption of businesses and government mostly relates to purchases of mobile phones and the equipment used in own account production of telecommunication assets. Directly collected data relating to gross fixed capital formation and margins complete the picture for supply and use of telecommunication assets.
Own account (construction) work by telecommunication providers
The 'standard' ABS ICT product definition (see Appendix 3) defines a range of products making up 'telecommunication equipment'. Essentially, this definition includes only telecommunication equipment. However, a substantial amount of telecommunication equipment is purchased to be integrated within telecommunication structures such as transmission towers and telecommunication networks more generally. In the satellite account we have chosen to treat spending on construction activities that are integral to the operation of telecommunication equipment as part of ICT output, value added and gross fixed capital formation. Appendix 2 contains a more complete discussion of this treatment.
Estimates of own account construction by telecommunication service providers were made by summing acquisitions of broadly defined 'telecommunication assets' of the major businesses involved. The value of these acquisitions was then partitioned into the following cost components: purchases of telecommunication equipment; labour costs; and cost of purchases of 'materials' consumed in building the assets. This was done using both publicly available information and returns submitted to the EAS collection. In practice, and consistent with the measurement of own account software, no allowance for consumption of fixed capital or net operating surplus has been included in the valuation of own account construction work by telecommunication service providers.
The total value of these telecommunication assets was treated as output of telecommunication service providers - own account production (construction) of telecommunication assets - and as gross fixed capital formation. The telecommunication equipment purchased by the telecommunication service providers for use in own account production was treated as intermediate consumption to avoid double counting these assets within gross fixed capital formation. The impact on 'ICT GDP' of this treatment is equivalent to the value of labour and material costs involved in the own account construction activity.
ICTIS does not provide sufficient data from which to estimate margins related to telecommunication equipment and calculations are complicated by the impact of mobile phones on the data. Businesses organising mobile phone 'plans' appear to be recording purchases of mobile phones as an explicit expense item, but revenue from the sale of these phones is subsumed within 'Income from provision of mobile and paging services'. Consequently, we have not allocated a margin to mobile phones sold as part of a 'plan'. It is instead included as part of telecommunication services.
A decision has been made to treat spending on mobile phones by business and government as intermediate consumption rather than as gross fixed capital formation in the satellite account. This is not a clear-cut matter and a valid case could also be made to treat this spending as gross fixed capital formation, since mobile phones are typically used repeatedly or continuously in production by producing units for more than one year. On the other hand, producing units are themselves generally treating these purchases as a current expense rather than as a capital cost. More importantly, the SNA93 (paragraph 6.158) 'small tools' rule allows for certain items that might otherwise be 'capital' in nature to be treated as intermediate consumption provided they are small, inexpensive and are used to perform relatively simple operations. Further, the expenditures should be made regularly and be small compared with expenditures on machinery and equipment. These guidelines provide solid support for the treatment adopted.
Household final consumption expenditure
The 2003-04 HES provided detail on household expenditure on telecommunication equipment. It asked householders to separately identify outright purchases of mobile phones and payments related to mobile phone 'plans'. Of the latter, we have estimated a certain proportion relates to the 'purchase' of the mobile handset itself. Overall, the estimates of HFCE telecommunication equipment are thought to be sound.
We might normally expect the vast majority of expenditure by producers on telecommunication assets to be treated as gross fixed capital formation. However, because of the treatment we have adopted for own account (construction) of telecommunication assets and because business and government spending on mobile phones have been treated as intermediate consumption, a significant amount of expenditure is allocated to intermediate consumption.
Gross fixed capital formation, by industry
Because the satellite account includes own account (construction) work on telecommunication assets in its definition of 'ICT output', this is also included in gross fixed capital formation of telecommunication assets. Derivation of this item has been described above, and the resulting estimate has been entirely allocated to the telecommunication services industry. The remaining gross fixed capital formation relates to business and government spending on telecommunication equipment (but not mobile phones). EAS and GTS data provide both national and industry estimates for this item.
As with all other ICT products in the satellite account, estimates for computer hardware were generated within a supply and use framework. Supply side estimates are considered superior to the demand side numbers and these provide the benchmark estimate.
Australian production of computer hardware is relatively small and is mainly undertaken by manufacturers of ICT products. ICT wholesalers and computer service providers both contribute smaller amounts, primarily related to computer hardware installation services. The majority of Australia's supply of computer hardware is imported from overseas.
Estimates of imports of computer hardware are drawn directly from the ABS International Trade System and are published in ICTIS. Because these are goods imports and therefore almost invariably physically shipped into Australia, customs information should provide good coverage. Even so, there are classification issues, particularly in relation to the various parts that may or may not relate (partially or wholly) to computer hardware. These issues were investigated in the course of constructing the satellite account and it was decided that no adjustment to ICTIS computer hardware import data was required.
Trade margins relating to computer hardware are based on ICTIS data. For ICT wholesalers, ICTIS provides estimates of value of purchases of computer hardware and sales of computer hardware. This provides the basis for estimating wholesale margins on computer hardware, though the practice of 'bundling' of computer hardware and packaged computer software is a factor affecting our interpretation of this data source (see entry on 'bundling' in this appendix).
Household spending on computer hardware is based on HES data. The HES collects household spending on 'computer equipment' separately from 'computer software'. Nevertheless, the HES figure for 'computer equipment' contains an element of bundled packaged software and this component has been estimated and removed based on an analysis of a typical computer purchase.
Within the ICT product supply and use framework, gross fixed capital formation on computer hardware was derived as a residual, reflecting the judgement that supply-side data are, overall, more robust than the various demand side data. Nevertheless, the ABS collects information on business and government spending on computer hardware through, respectively, its EAS and GTS collections. The gross fixed capital formation computer hardware value from these collections is somewhat lower than the figure generated in the satellite account. There are a number of possible reasons why this has occurred, including, especially, the effect of leased computer hardware. This apparent understatement was corrected for by benchmarking the estimates to supply side data.
There is evidence that significant amounts of ICT computer hardware (and telecommunication equipment) are leased by their users. There are potentially two ways that leasing computer hardware can hinder our attempts to measure these products.
First, if both the lessee and lessor are reporting the same ICT product as capital expenditure to ABS surveys then a double-count exists. In practice, survey respondents appear to be following finance lease accounting rules and double-counting does not appear to be an issue. If it were happening, we would expect to see exceptionally large figures for capital expenditure for computer hardware and telecommunication equipment in the Finance and Insurance industry (where most lessors are operating). However, this is not the case.
The other problem potentially arises in the EAS and GTS collections, where respondents are required to report spending on computer hardware and telecommunication equipment acquired under finance lease arrangements as 'capital expenditure'. However, in some cases, lessees have misreported this spending as 'rent, leasing and hiring' expenditure (an appropriate response only for assets used under an operating lease). While every effort has been made to 'clean' the survey data as far as possible, it is likely that the reported EAS survey data somewhat understate actual levels of ICT-related capital expenditure.
In practice, the satellite account derived gross fixed capital formation computer hardware as a residual, and EAS and GTS were used only to derive industry details for this item. Thus the issue of leasing has had minimal impact on the estimates produced.