Depreciation of research and development satellite account expenditure

Warning View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

85. There are two cases in which assets need to be depreciated in the RDSA. One is for depreciable assets used to generate R&D, as part of the process of moving expenditure data to output. The depreciation of these assets is relatively straightforward since they involve tangible assets such as buildings and other equipment. The other is the depreciation of intangible R&D assets by an industry/sector as used in a production process.

86. R&D differs from physical capital (like machines and buildings) in that it does not wear out. Rather, it loses value due to obsolescence, or possibly as a result of the ending of patent protection or the leakage of information to competitors about a given product. In fact, an argument given against changing the treatment of R&D in the SNA was that its asset lives are not known. Little information is also available concerning the rate of depreciation (e.g., linear or geometric). However this is not that different from other produced assets in which accurate information of asset lives and rates of depreciation is not always available.

87. Much like other assets, various assumptions can be made about lives and whether the rate of depreciation is accelerating over time for R&D. Several methods have been used to determine the asset life of R&D. They include using patent lives common to an industry, using the asset lives associated with a given industry for other capital and using the asset lives in the research and development related industries (NAICS 5415 and 5417). A survey of R&D producers asking for the expected service life of R&D has also been employed.

88. It should be noted that the consumption of fixed capital (CFC) that occurs with the depreciation of R&D assets used in a production process, are a portion of total operating surplus. Therefore, when business capital is depreciated, the type of model used (e.g., linear or geometric) or the length of the asset lives chosen, does not affect total GDP. It only changes the composition between capital consumption and net operating surplus.

89. It is only in the cases of government and non-profit funded R&D that asset lives and the model used have an effect on GDP as well as saving. This is because for these sectors the amount of CFC attributable to R&D is added to both income and expenditure for these entities thereby increasing factor incomes, final expenditure and value added for this sector.

90. Government R&D (which includes university R&D) is characterized by basic research. Studies from the U.S. indicate that two-thirds of university research and a quarter of federal government R&D is basic research.25 This research tends to have longer and more stable asset lives than that of the fast moving technology-based R&D like computer software.

91. As a result, the asset lives for government and non-profit R&D were fixed at ten years for the whole study period. Various scenarios for asset lives were done to analyze the effect of asset lives on GDP. These results are presented in the results section at the end of the paper. Asset lives for business investment were set equal to the rate used for other capital in the same industry. A linear Perpetual Inventory Model (PIM) was used for all estimates, consistent with the estimation procedure used for other government assets in the Canadian SNA.26 The depreciation of business investment used the average asset lives of the industry in which the R&D was performed.


92. The deflation of research and development output encounters many of the same challenges as the deflation of other non-market products. In Canada, approximately 70% of R&D is undertaken on own-account and therefore is not directly sold on the market. That which is sold on the market is very heterogeneous and therefore using these prices as proxies for own-account R&D may not be accurate. As there is no market price in the conventional sense associated with the service provided, the cost of the inputs are used to measure price change. This method is suggested in the FM for the deflation of expenditures.

93. Input costs are broken down into two main types in the FM data–labour and other. The "other" component encompass various goods and services that enter into the production function such as rent, heat, electricity, office supplies and even raw materials used up in the research process. Since the FM surveys do not provide this detailed input information, corresponding data could be taken from the R&D related industries North American Industrial Classification System (NAICS 5415 and 5417) in the Input-Output (IO) system. Corresponding price indexes could be used to measure the change in prices of these items over time. These price indexes should take into consideration any changes in quality of the commodity in question, in order to measure pure price change.

94. This is also true of the labour component, however the separation of price and quality can be difficult to achieve. In this case, real estimates of non-market services usually rests on the assumption of constant productivity, with the volume estimates reflecting solely the change in the level of employment (hours worked). However, it is generally accepted that labour, in aggregate, has become more productive over time but the rate at which it has done so is not known. An R&D producer may also be using its other inputs more efficiently over time.

95. The U.S. RDSA used four different methods to deflate R&D estimates. The first used input prices. Labour costs were available to deflate labour inputs and other costs were deflated using prices from the R&D industry (NAICS 5417). The second method used prices from the five industries with the highest productivity in manufacturing; the third used the highest productivity services industries, while the fourth used prices from industries that were the most R&D intensive. The choice of method had a relatively small impact on GDP (a difference of less than 3% of R&D GDP).

96. The Canadian RDSA used one composite price index to deflate the detailed industry R&D expenditures. This price index was a weighted average of various price indexes for commodities thought to be representative of R&D inputs. The latter include price indexes for non-residential rent, computer-related goods and services, office supplies and telephone and other communications, as well as the All-items Consumer Price Index for the portion related to labour remuneration.27 The All-items Consumer Price Index was also used for operating surplus. The Canadian RDSA does not introduce productivity differentials through prices as was done in the U.S. account. Rather, the possible productivity gains are theoretically measured by the premiums found in the various scenarios for operating surplus.

Stock-flow measurement issues

97. As the result of the nature of R&D as an intangible asset, the measurement of stocks and flows in R&D, when traded is a difficult issue. In this study, it is assumed that all exports or other traded R&D are for sale only. That is, the seller does not keep any of the stock of this R&D after the sale. This is not a fully realistic assumption in the Canadian case since there are documented cases of exchanges in assets and not just services. For example, R&D assets, in the form of patents, have been traded abroad. Further investigation of this issue is required.

Research and Development Satellite Account

98. The objective of the RDSA is to examine the effects of changing the expenditures of selected components of research and development from current spending28 to capital formation in the Canadian System of National Accounts (CSNA) sequence of accounts – both from the industry account side and from the sector accounts dimension. The RDSA detail will meet the requirements of both CSNA architectures, and will make for a straightforward link to the SNA93 Rev1 sequence of accounts.

99. The RDSA has been suggested as a method to investigate the impact of this change as well as to explore various options with respect to R&D inclusion.29 This satellite account approach has been put forward by the Canberra Group. This approach would allow for work on the major balancing entries in the macro-economic system so that in the next few years, consistent estimates for measures such as GDP and savings, as well as measures for gross fixed capital formation and stocks of fixed capital can be implemented across countries.

100. In the Canadian satellite account all R&D is included, such that the impact on GDP is calculated with and without, for example, software as part of the R&D total. The RDSA then allows for the accounting of the total impact of R&D in the national economy as well as the additional impact or the change to GDP by broadening the coverage of R&D beyond the current treatment.

Structure of the Canadian Research and Development Satellite Account

101. The aim of this satellite account is to build a framework for incorporating measures of R&D capital into the core accounts. With this in mind, the satellite account will follow the sequence of accounts as presented in the SNA and show the impact on the key macro-economics variables.

Impact on the System of National Accounts component accounts

Production account and industry detail

102. The change in treatment of R&D will have varying impacts in the SNA depending on whether transactions involve the business, government or non-profit sector. A new R&D commodity will be created (similar to the commodity created for software capitalization) to record R&D. R&D associated with software will remain in its current commodity. R&D expenditures on mining will be included with R&D. However, exploration expenditures will continue to be presented separately.

Business sector

103. In the business sector, if an establishment was a producer but not a user of R&D, there is no impact to its industry. The R&D will remain as a sale and not be capitalized within that industry. Producers of R&D will continue to have some knowledge capital (they may or may not derive income) that is not capitalized. This knowledge capital is not dealt with in the RDSA.

104. The situation changes for non-producing buyers of R&D. In this case, the R&D, which is currently being treated as the purchase of an intermediate input, will now be capitalized and recorded in the capital account. Total current expenses will drop by the amount of R&D purchased resulting in an increase in operating surplus. Output remains unchanged as the decrease in expense will be offset by the increase in operating surplus. The increased surplus will be used to make the investment in R&D, thereby leaving net lending unchanged.

105. Own-account producers of R&D will see an increase in output by the amount of the R&D capitalized. Operating surplus will increase with depreciation of the R&D invested. Intermediate expenses will remain unchanged. In the case of the various rate of return scenarios, the extra income resulting from these premiums would also increase surplus. Any purchases of R&D by an R&D producer, whether from a domestic source or imported, are considered as capital. In this case, intermediate expenses will decrease while operating surplus increases by the equivalent amount, leaving output unchanged.

Other sectors

106. In the government and non-profit sectors, a purchaser (funder) of R&D will record a decrease in current expenses since the R&D is now capitalized. Since government and non-profit output is equal to total cost, including these intermediate expenses, output will decrease. However, some value will be added back since the CFC of the R&D capital will be included in the valuation of the output. Therefore output will decrease by the amount of R&D investment less the CFC for the R&D capital stock.

107. In the case of government and non-profit own-account R&D, intermediate expenses will remain the same in total. However the intermediate expenses for non-R&D output will go down, as in the case above, but these intermediate inputs will now be used to produce R&D output that is capitalized. Also, as the CFC from the new capital investment is included in the valuation of the government output, value added and output increase by the amount of CFC added.

Primary income - surplus

108. As a result of the change in treatment of R&D, income arising from production increases in the SNA. This occurs because R&D is no longer treated as an expense in the accounts. Since expenses decrease, surplus increases. This increase matches the gain in investment. In the CSNA, the increase in surplus is registered in both CFC and net operating surplus.

Final expenditure

109. Business and government investment in the SNA will increase due to the change in treatment of R&D. Government current expenditure is reduced by the amount of expenditure on R&D. This is partially offset by an increase in CFC on the depreciated R&D assets.

Accumulation account

110. As a result of the change in treatment of R&D, investment and saving will increase in the capital account. In the Canadian SNA, the increase will be registered in the machinery and equipment line of the accounts, similar to the impact of software capitalization.

111. There is no direct impact in the sector financial accounts of including R&D as capital assets. This is because the net addition to investment is equal to the addition to saving from the reduction of operating expenses. Thus, there is no impact on net lending/borrowing for the total economy. There are no subsequent changes to corresponding financial instruments and therefore no explicit entries in the financial account.

Balance sheet account

112. A broadening of the treatment of capital to include R&D assets has a direct impact on both national wealth and sector estimates of assets, liabilities and net worth. National wealth includes both produced and nonproduced assets. R&D is a produced intangible asset, which adds to the productive capacity of the economy. The asset values are sensitive to both the service lives assumed and the depreciation method chosen.

113. At the same time, sector account estimates of produced assets are more correctly measured by the inclusion of R&D capital and, as a result, estimates of sector net worth are improved. For corporations, the market estimate of future stream of earnings would implicitly reflect knowledge capital. Therefore, this increased coverage of assets allows for a better reconciliation of the market value of corporate equities and the net asset value estimate of net worth.30

114. The impact of changes in asset positions with respect to purchases/sales of R&D assets is an area that requires further research.

Other changes in asset accounts

115. The difference between the opening and closing values of R&D assets would reflect, in addition to the investment and depreciation of these assets: changes in the cost of reproducing these assets (price fluctuations), in the Revaluation account; and; write-ups/downs of R&D capital in the Other changes in the volume of assets account.

Alternative estimates of research and development capital

116. One issue that will be examined in the RDSA is the impact of various rates of return for R&D output in the business sector. Four scenarios, using different assumptions for rates of return, are presented in this study along with their impact on R&D GDP.

117. Another dimension to the RDSA is a tabulation of both total R&D capitalization as well as the additional impact on the economy of capitalizing additional R&D expenditures. For total capitalization, adjustments made for spending on software, trade of R&D, the CFC involved in the process of producing R&D and government and nonprofit capital are included. The additional capitalization measure, on the other hand, examines the change to the core accounts as a result of broadening in the treatment of R&D and therefore excludes the above adjustments.

118. In the future, this satellite account can also be broadened to include other intangible assets on the SNA. This could include a study on innovation and its role in the economy.


25 . Siddiqi and Salem, A Proposal for Treating Research and Development as Capital Expenditures in the Canadian SNA, Statistics Canada, page 17.

26 . The U.S. satellite account provided results for two asset life scenarios. The first scenario used a 15 percent geometric rate of depreciation, the second used a rate that varied between four and six years, getting shorter as the capital expenditure became more current. This was an attempt to model the rapid pace of recent technological change.

27 . Average earnings from the Survey of Employment Payroll and Hours (SEPH) undertaken by Statistics Canada could be another possible deflator for the labour data.

28 . Satellite accounts are based on the principles of the National Accounts but are developed as an extension to the core National Accounts System. This allows for a comparison between the satellite account (or area of interest) and the entire economy as measured by the SNA. With the RDSA, the impact on the level and growth rate of value added or GDP can be examined. A satellite account is a useful tool when developing a new variable for the accounts because: (a) it allows for the analysis of the impact of that variable on the account without disturbing the core of the account, (b) various definitions of the area of study can be examined and (c) several methods of implementation can be tried.

29 . See minutes of the National Accounts Advisory Committee, June 2005.

30 . This increased coverage of assets would, all other things being equal, reduce residual corporate net worth.