Proposal for a simplified bridge tables between FM to SNA[1]
In the first meetings of the NESTI SNA-FM group and the Canberra II group the feasibility of using R&D expenditure data collected according to the Frascati Manual on Proposed Standard Practice for Surveys on Research and Experimental Development (FM) for national accounts purposes was examined. Since the latest version of the FM has not been fully harmonized with the SNA, the differences between the two systems were analyzed in a background paper and it was proposed to construct bridge tables linking between the two systems.
The bridge tables should ideally provide a complete link between data on gross expenditure on R&D (GERD), according to the FM, and estimates of R&D output based on SNA93’s concepts and definitions. The preliminary versions of such bridge tables together with a questionnaire were distributed to OECD countries in the beginning of 2003. The intention was that compilers of R&D statistics should cooperate with national accountants and respond to the questions in order to estimate the importance of the differences and the feasibility of filling out the bridge table.
Only a few of the countries – Netherlands, USA, Slovakia, Israel – bridged between the two systems, and the impression was that one of the obstacles to performing the bridging was the difficulty to establish a connection between the compilers of R&D statistics and national accountants, who mostly work in two different institutions. It was however, possible to analyze the importance of the differences in the few cases where bridging was performed, and in addition an examination of the surveys conducted by the various countries provided more information on the importance of some of the items.
Many of the differences included in the preliminary bridge tables are insignificant in practice. Other differences, such as consumption of fixed capital or operating surplus, may be estimated combining existing data from the R&D statistics and national accounts. A few of the important differences require additional information that could most efficiently be collected in R&D surveys, and the necessary changes to the FM in this respect are outlined in the second paper presented.
Since the experience has shown that it may be difficult to establish the cooperation between national accountants and compilers of R&D statistics in practice, it is proposed that the construction of bridge tables should be managed on a current basis by compilers of R&D statistics, and that they collect the relatively few but important additional data needed to complete the bridge tables from the national accountants. Future revisions of the FM should in our opinion preferably include recommendations for the preparation of such bridge tables.
On the basis of the experience gathered so far we also propose to eliminate some of the bridging items at this stage and construct a simplified bridge table.
Below the results so far are discussed and the simplified table is presented.
Insignificant differences
A number of differences were insignificant, and could be omitted from the bridge tables in the short term.
1. Adjustment for changes in inventories – It may be assumed that the inventories held for R&D are relatively small, since the share of purchases in total expenditure is not large (around a quarter to a third of GERD in most of the countries that report on this item), and such purchases are mostly not for an on-going production process. The few countries reporting either already collected data excluding changes in inventories, or could not supply data for this item.
2. The adjustments of capital expenditure do not affect the measurement of output directly, since the capital expenditure is subtracted when passing from GERD to R&D output in national accounts. But the data on gross fixed capital formation affects the output measure indirectly, since they are used for the estimation of consumption of fixed capital applying the perpetual inventory method. However, in practice the two adjustments (disposals of fixed assets and land) were insignificant, and could be omitted from a simplified bridge table:
Disposals of fixed assets – in the case of Israel such disposals amounted to two percents of the acquisition of fixed assets, and other countries did not report any numbers on this item. The small size of disposals could also be expected, since service lives of many of the fixed assets used in R&D are quite short, so that the disposal of them to other producers probably is not common.
At this stage it is probably not possible to obtain data for adjustment for land (only one country (Israel) reported to a questionnaire distributed among NESTI members a few years ago that it had data excluding land). Capital expenditure on buildings and land is a relatively small part of expenditure - around 2% of total R&D expenditure in OECD countries reporting separately on expenditure on buildings and land, so that any errors in the evaluation will not play a very important role.
3. Exports and imports by non-market producers were not reported by any of the countries, and are probably insignificant: Output of non-market producers is generally not sold, and even if sold, only in very rare cases sold to abroad. Consequently, it does not seem worthwhile to include these items in the bridge tables for non-market producers.
4. Similarly only small parts of output in non-market institutions are sold at prices including operating surplus, and to include an estimate for this small part will be in most cases be very difficult.
Differences estimated by using data from R&D surveys
Acquisitions of R&D to be used in R&D activities: Data on acquisitions of R&D by producers of R&D, which are omitted from estimates on expenditure prepared according to FM, were readily available from R&D surveys in a few countries, and the amount of acquisitions was found to be quite large and growing. The fact that international transactions between R&D producers have grown over the years is well known, but there also is an even more important amount of transactions between domestic producers. Data collections on these items should be mandatory in the future as outlined in the second paper presented by us at this meeting.
Classification by sector should also be changed by making a subdivision of the data for the Higher Education Sector. The first bridge tables did not contain such a subdivision, however, since data in R&D surveys are mostly collected for each institution, it seems feasible to adjust the data for most countries.
Differences estimated by combining R&D statistics with national accounts data
Some important differences were estimated by combining data from the national accounts with R&D statistics:
1. Operating surplus for market production of R&D. Netherlands and Israel were able to add operating surplus, using data from the national accounts, and a survey of national accounts data for other countries shows that most countries have separate data on output for the R&D industry in ISIC 73, which can used for estimating operating surplus as a percentage of costs of market production.
Saying that, it still is a question, whether the operating surplus in the R&D industry is a valid measure for all market production of R&D, especially since the proportion of operating surplus is usually quite volatile over the years, and probably also varies widely between various projects. Each R&D project is by definition different from other R&D projects, so that there should not be a reason to assume that operating surplus would be similar for all R&D. The production of R&D also typically takes more than one year, so that a large proportion of the activities of the producers is on work in progress.
To have a good measure of operating surplus, details on R&D projects should be collected and analyzed over a number of years. Details on R&D sales could be collected in the framework of R&D surveys (as recommended in our second paper).
2. Consumption of fixed capital (CFC): Data collected in the framework of R&D surveys on expenditure on fixed capital can conveniently serve as a basis for estimates of stocks of fixed capital using the perpetual inventory method (PIM). In practice the easiest way to obtain such estimates will be by using data on rates of depreciation, prices etc. already prepared for the relevant capital goods in the national accounts.
3. Other taxes on production less other subsidies on production: The SNA93 includes them as part of the production costs. The mentioned flows are recorded in the generation of income account, as use components of the gross value added, for output measured at costs or at basic prices.
The FM does not show the flow of other taxes on production explicitly, but the flows are included, at least partially, in the current expenditures, e.g. payroll taxes are part of the labour costs. However, the flow of other subsidies on production is not accounted for in intramural expenditures, but as a financing source of them.
Netherlands was able to adjust the expenditure for this item in an adequate way, and other countries will probably have similar data in their national accounts.
4. R&D on software: Two countries – Netherlands and Israel mentioned the growing prominence of R&D on software. As explained in the paper on R&D and software we recommend including R&D on software with other R&D, following the FM. USA did include R&D on software in their estimates on R&D output.
Differences where additional data collection is needed
As said above, the domestic and external R&D transactions of R&D performers are currently investigated in a few countries through R&D surveys, and such transactions should be investigated in additional countries. But producers that are not themselves performing any R&D may also have external sales and purchases of R&D. In countries where such transactions are of importance, they will have to be covered through other types of sources – for example in surveys of exports and imports, which are becoming more common in recent years.
Conclusion
The bridging of the remaining gap between the R&D statistics and SNA is a crucial prerequisite to the capitalization of R&D in the national accounts on the basis of the FM data.
The first experience with collection of data for bridging between the two systems has shown that it may be difficult to establish the cooperation between national accountants and compilers of R&D statistics in practice, since they often work in separate institutions. On the other hand, most of the data needed is obtainable from R&D surveys, so that it is proposed that the construction of bridge tables should be managed on a current basis by compilers of R&D statistics, but that they contact the national accountants to obtain the information needed to complete the bridge tables, similar to the way that data on PPP for GDP or total GDP are obtained.
Future revisions of the FM should preferably include recommendations for the preparation of such bridge tables on a current basis.
The appendix shows a proposed simplified bridge table for the total economy, each of the FM sectors and a disaggregation of de Higher Education sector in order to bridge this FM sector with the SNA sectors.
References
“Basic Science and Technology Statistics”, OECD, Paris, 2002
de Haan, Mark, and Myriam van Rooijen –Horsten: “The translation of R&D statistics from Frascati to National Accounts guidelines in the Netherlands”, Canberra II Group meeting, 15-17 April 2003, Statistics Netherlands, Voorburg, the Netherlands
Guellec, Dominique, “Improving Compatibility between Frascati and SNA: Proposals at the October 23-24 NESTI Meeting” DSTI/EAS/STP/NESTI(2001) 28, October 2001
Mandler, Pablo and Soli Peleg, “Background and Issues Paper for the R&D-SNA Task Force” NESTI SNA-R&D task force meeting, 14 April 2003, Statistics Netherlands, Voorburg, the Netherlands
System of National Accounts 1993, Commission of the European Communities, International Monetary Fund, Organisation for Economic Co-operation and Development, United Nations, World Bank, Series F, No. 2, Rev. 4, Brussels/Luxembourg, New York, Paris, Washington D.C., 1993.
“The Measurement of Scientific and Technological Activities. Frascati Manual 2002: Proposed Standard Practice for Surveys on Research and Experimental Development”, OECD, Paris, 2002.
Annex
1
[1]Prepared by Pablo Mandler and Soli Peleg, Central Bureau of Statistics, Israel