APPENDIX 6 QUALITY OF THE ESTIMATES
Overall, the ABS considers that the estimates are of reasonable quality, especially at the most aggregated level, such as for ICT gross value added and ICT GDP. However, because the satellite account is the product of development work requiring a number of assumptions and synthetic estimates for some components, the estimates should be considered experimental. There is less confidence in the quality of the results at various levels of disaggregation. For example, the boundary between 'customised software' and 'computer services' is reasonably clear in principle, but in practice it can be quite difficult for survey respondents to separate these two components.
The estimates were prepared from a wide range of statistical sources. Most key data sources are compatible with the desired national accounting basis, however in some instances this is not the case and some data sources are less than completely satisfactory in terms of coverage, concepts and/or timing.
Estimates of own account computer software produced by non-general government units were generated using modelling techniques that required use of significant assumptions, as described in some detail in Appendix 5. Estimation techniques essentially follow those recommended by the OECD software task force and represent international best practice. Nevertheless, this remains a challenging area in estimation, and users are warned to exercise caution when using own account software estimates. The estimation of own account production of telecommunication assets by telecommunication service providers essentially follows the general principles for valuation of own account production of capital goods as outlined in the international standards for national accounts (SNA93), and as used by the major telecommunication service providers.
The ABS International Trade System is not purpose-designed to capture ICT product detail. Instead, fine level international trade data are examined and each product category is designated as either ICT or non-ICT in nature. Since some of the product items are in reality partially related to ICT products and partially to other products, this compilation technique represents a potential quality issue. This is especially the case for certain types of components which may find use in a wide range of final goods.
The measurement of international trade in computer software is known to be a difficult area for official statisticians. The OECD in its Information Technology Outlook, 2004 comments that recorded international trade in computer software is surprisingly low. There are a number of reasons why this is the case - reasons principally related to the nature of software production and the way in which software is traded across borders. Appendices 5 and 7 provide a fuller description of relevant issues and of related work recently undertaken by the ABS.
Within the national accounts, wholesale and retail margins are generally estimated indirectly rather than being based on purpose-designed surveys, so the estimates are an approximation. Nevertheless, ICTIS 2002-03 provides appropriate structural data that allow the generation of estimates of wholesale margins on various ICT products. Retail margins on ICT goods have been estimated at least partly through reference to the related wholesale margins. In dollar terms, it seems clear that the greater proportion of sales of ICT goods are through wholesalers rather than retailers.
Overall, data sources for the supply of ICT products are considered superior to the independently-derived demand-side estimates. Supply side estimates have generally been accepted and adjustments made to the use estimates considered weakest. In practice, this has generally meant that intermediate consumption or gross fixed capital formation were derived as residuals. Given this approach, the fact that resulting revisions to official estimates of gross fixed capital formation of computer hardware and computer software have been relatively modest, provides some confidence in the estimated levels (past and present) for these aggregates.
Estimates of the number of persons employed in ICT-related occupations have been derived from the LFS. As this is a household survey, it has some deficiencies when used to derive detailed industry estimates. In order to mitigate some of these potential quality problems, estimates of employment have been published at the ANZSIC division level.