Measuring R&D in the National Economic Accounting System.

Crawford, Marissa J. ; Lee, Jennifer ; Jankowski, John E. 等

ECONOMISTS HAVE LONG recognized that research and development (R&D) products have the characteristics of fixed assets--their ownership rights are well defined, they are long lasting, they can be used repeatedly in the production of other goods and services, and their value depreciates over time. In the early 1990s, the Bureau of Economic Analysis (BEA) began researching how R&D expenditures should be treated in the core economic accounts. That research culminated when as part of the 2013 comprehensive revision of the national income and product accounts (NIPAs) and the 2014 comprehensive revision of the industry economic accounts (IEAs), BEA expanded the asset boundary to recognize business, academic, nonprofit, and government R&D expenditures as investment, not as a current expense. This new approach allows the accounts to better measure the effects of innovation and intangible assets on economic growth and productivity.

The enhanced treatment of R&D investment in the NIPAs and IEAs reflects the consensus view that R&D and other intellectual property products are critically important to the modern economy. Treating R&D as an investment allows economists to analyze its contribution to growth and productivity using the same framework as other capital goods. (1)

BEA's treatment is also consistent with the recommendations of international guidelines, notably the System of National Accounts 2008, which defines R&D as "creative work undertaken on a systematic basis to increase the stock of knowledge, and use of this stock of knowledge for the purpose of discovering or developing new products, including improved versions or qualities of existing products, or discovering or developing new or more efficient processes of production." (2)

This article describes the methodology underlying the measures of R&D investment and output in the NIPAs and the IEAs. The article discusses the following:

*Current-dollar R&D estimates and key issues, such as (1) the main source data, which are the National Science Foundations (NSF)'s R&D surveys, (2) how ownership of R&D is established, (3) how R&D developed for an entity's internal use is differentiated from R&D that is purchased, and (4) how BEA distinguishes between establishment and company R&D investment

*The development of depreciation rates and price indexes to derive R&D investment and output, including real measures

*The methodology for estimating R&D investment in the current quarterly estimates In addition, a brief history of R&D economic accounting at BEA is discussed in the box "A Brief History of BEA's Capitalization of R&D."

Methodology: Current-Dollar Estimates

Overview

Estimating R&D investment requires valuing R&D expenditures when market prices are not available and identifying the owners of R&D investment.

BEA's standard approach for measuring output is to value it at market prices if possible. When market prices are not available-mainly when valuing output for own-account and nonmarket uses-the standard approach is to use the costs of production as a proxy.

BEA defines the owner (or investor) of R&D as the funder of the R&D. This decision was made largely because funders typically reserve some, if not all, rights to the outcome of the R&D and receive economic benefits from the R&D.

BEA estimates of R&D investment include (1) R&D that is funded by one entity but produced by another entity, which is considered "purchased" R&D, and (2) R&D produced for an entity's own use, which is considered "own-account" R&D. Market prices are used to value purchased R&D, while the value of own-account R&D is estimated as the sum of production costs.

Source data and adjustments

In producing estimates of R&D investment, BEA relies primarily on NSF's R&D surveys, which provide information on expenditures associated with the production of R&D. These surveys include the business research and development and innovation survey, higher education research and development survey, survey of federal funds for research and development, and survey of state government research and development. All are conducted annually except the survey of state government research and development, which is conducted every other year. (3) In general, the NSF surveys provide data beginning in 1953.

For years not covered by the surveys, BEA uses a variety of sources to estimate R&D investment. For example, for years before 1953, business R&D is estimated primarily by using employment data for scientists and engineers. For the details on the primary source data for all of BEA's R&D investment and output estimates, see table 1.

BEA adjusts the NSF expenditure data for coverage, for scope, and for alignment with the NIPA framework and concepts. The adjustments are as follows:

*BEA converts NSF R&D expenditures data on a fiscal year basis to a calendar year basis to align with the NIPAs.

*NSF surveys of business R&D collect some information on exports of R&D (including R&D performed domestically that is funded by sources outside of the United States). However, these data do not fully represent all of the R&D foreign trade flows recorded in BEA's economic accounts. BEA must also separately record imports of R&D, which are not captured in NSF's surveys of business R&D. As a result, instead of relying on the NSF data, BEA integrates its international transactions accounts (ITAs) data on imports and exports of R&D services when deriving measures of total R&D output and investment. This approach ensures consistency with BEA's methods for estimating imports and exports and with existing estimates of R&D services in the ITAs. To align the trade data with the definition of capitalized R&D services, a portion of the ITA-based imports and exports is removed because they represent non-R&D testing services.

*BEA's estimates of R&D investment exclude software R&D, which is already captured in BEA's estimates of software investment. To prevent double-counting, BEA removes the portion of software development R&D already embedded in its software estimates. Table 2 shows the amount of software R&D removed from the R&D estimates.

*BEA makes an adjustment to classify the enterprise data on an establishment basis in the IEAs. This approach assigns an establishment's activities to the industry most closely associated with that activity. In contrast, the NSF's industry classification assigns all of a multiunit company's R&D expenditures to one industry.

Tables 3-6 show in detail how BEA builds the R&D investment estimates from the NSF data. Table 3 shows total private business, table 4 shows the pharmaceutical and medicine manufacturing industry, table 5 shows the scientific R&D services industry, and table 6 shows federal government.

R&D ownership

BEA recognizes R&D investment by corporate and noncorporate business, nonprofit institutions serving households (NPISHs), academic institutions, federal government, and state and local governments. Because the accounting methods used by most businesses and governments do not capitalize R&D and because other data sources such as patent records do not provide comprehensive coverage of R&D assets, BEA must provide its own allocation of ownership to the corporate and noncorporate business sectors, NPISHs, and the general government sector. The allocation of ownership is important because it affects measures of net income such as corporate profits and personal income as well as measures of net saving because consumption of fixed capital is a charge against the income of the owner. (4)

NSF's surveys of R&D expenditures are conducted based on the performer of the R&D, but they also provide extensive information on the funder of R&D for each performer.

For the private sector, the allocation of R&D ownership to the funder is clear. However, the ownership of government-funded R&D may be less transparent. Government-funded R&D is supported through two primary mechanisms--purchases and grants. As an example, the federal government may directly purchase R&D from an academic institution, or the federal government may provide funding for R&D at an academic institution via a grant. The ownership of purchased R&D is straightforward, because the federal government normally retains ownership of the outcome of the purchased R&D activity. For grant-based R&D, however, the ultimate beneficiary is difficult to ascertain because both the federal government and the performer can benefit from the transaction. Historically, NSF surveys have not collected information on the allocation of purchased and grant-based R&D for the federal sector. (5) Because of this lack of direct information, federal purchases and grants of R&D are both treated as investment by the federal sector because the federal government is assumed to receive the primary economic benefit. The funders of R&D generally invest in R&D because they receive economic benefits; in the case of government funding, the economic benefits include the benefits that the government obtains on behalf of the general public.

For a perspective on how funding differs for each performer, see tables 7 and 8, which show the different funding sources for each R&D performer. The statistics are BEA-adjusted estimates based on NSF data.

Own-account versus purchased R&D

The NSF data used to derive R&D investment provide expenditures, or costs, for R&D performed by domestic entities regardless of whether it is produced for their own use (own-account) or produced for and funded by others.

To calculate total business own-account R&D, BEA starts with total NSF R&D costs plus a profit markup on the amount of R&D sold, (6) and subtracts R&D sales by businesses reported by the Census Bureau's economic surveys. (7) The Census Bureau data are based on receipts and already embed a profit markup. To calculate total business R&D investment, BEA subtracts from its estimate of NSF-based total business R&D expenditures, which includes the profit markup, the following: R&D sold to other sectors, such as governments and NPISHs and exports of R&D. The R&D sold to government and NPISHs is classified as R&D investment for those sectors. BEA then adds the following: business purchases of R&D from government and NPISHs and imports of R&D. Table 3 provides a walkthrough of the steps BEA uses to calculate business R&D investment from the NSF expenditure data. Investment of all funders of R&D is estimated using a similar methodology.

Establishment versus company

NSF's industry classification is company based, which assigns all of a multiunit company's R&D expenditures to one industry. For a given company, expenditures can be made in several different types of establishments, including manufacturing plants or services-producing facilities, company headquarters, a freestanding R&D laboratory that provides all its output to the rest of the company (auxiliary establishment), or an independent establishment that sells its R&D output. Since the NSF surveys do not link R&D expenditures to establishments, using the NSF industry classification would imply that all pharmaceutical and medicine manufacturing R&D activity was conducted in manufacturing plants when more realistically, it was conducted in R&D laboratories, company headquarters facilities, manufacturing plants, and other types of establishments.

BEA's industry input-output accounts are classified based on establishments. According to this establishment-based approach, for example, a multiunit company with one establishment whose main activity is manufacturing and another establishment whose main activity is R&D services will have the output of the manufacturing establishment assigned to the manufacturing sector (the primary industry) and the output of its R&D services establishment assigned to the services sector. To incorporate R&D as investment in BEA's input-output accounts, the NSF data are converted to an establishment basis. BEA applies a company-to-establishment adjustment based on a reconciliation of Census Bureau-based data from the scientific R&D services industry (establishment-based) with the NSF's company-based data. See the box "BEA's Company-to-Establishment Adjustment for the Scientific R&D Services Industry" for more information.

Quarterly estimates

BEA prepares quarterly estimates of current-dollar R&D investment back to 1947. However, the most complete and accurate source data are the NSF surveys of R&D expenditures, which provide only annual data. BEA thus relies on a number of methods and data sources to interpolate annual estimates of R&D investment into quarterly estimates. These methods and sources vary depending on the period.

For NPISHs, academic institutions, federal government, and state and local governments, little information on quarterly trends in R&D activity is available. Therefore, quarterly estimates are interpolated from annual estimates assuming no quarterly pattern. For business R&D, the quarterly interpolation methods are split into three periods, each with a different interpolation method: 1947-1990, 1991-2007, and 2008 forward.

For 1947-1990, the quarterly estimates are interpolated assuming no quarterly patterns.

For 1991-2007, current-dollar business R&D is interpolated using a seasonally adjusted composite series from the quarterly census of employment and wages (QCEW) from the Bureau of Labor Statistics (BLS). The composite quarterly indicator is constructed by weighting three-quarter moving averages of industry-specific QCEW wage data. The composite QCEW series is constructed using a weight of 60 percent for the scientific R&D services industry's wage growth. The remaining 40 percent weight of the series is allocated to the other private industries using the annual distribution of R&D investment.

For 2008 forward, quarterly business R&D is interpolated using seasonally adjusted R&D expenses reported on publicly traded companies' quarterly financial statements as an indicator. See table 9 for information on the specific quarterly interpolation methods for private business, nonprofits, and government sectors.

Methodology: R&D Depreciation

The depreciation rates used by BEA to estimate business R&D depreciation are based on the concept that unlike tangible assets, which depreciate because of obsolescence that causes physical decay or wear and tear, R&D capital depreciates because its contribution to a firm's profit declines over time. (8) BEA developed a forward-looking profit model and used its annual industry output and R&D investment for 1987-2007 to estimate industry-specific R&D depreciation rates. Table 10 shows the R&D depreciation rates for 10 R&D-intensive industries. As described in Li's working paper (2012), the results are consistent with expert opinions and conclusions from empirical analysis that R&D depreciation rates should be higher than the traditional assumption, 15 percent, and should vary across industries. Details of the model and estimation results are described in Li (2012).

Industries not specifically identified use the depreciation rate for the scientific R&D services industry of 16 percent. This rate is also used for NPISHs and academic institutions.

For general government R&D, a contribution to profits is not estimated, but like business R&D, depreciation reflects obsolescence over time. Based on this concept, BEA observed that R&D investments were associated with the production of other products that eventually became obsolete; for example, investments in stealth technology were associated with the development of particular military aircraft. As innovations give way to newer technologies, the original R&D becomes less valuable or obsolete, thus bringing an end to the effective service life of the R&D.

BEA derived service lives for four federal government functions: defense, health, space, and energy. To estimate service lives for these government R&D asset types, BEA selected a sample of representative projects. The service life for an R&D project spans the time from prototype design through the end of production of the physical tangible asset. An average for all sample projects for a particular R&D asset type, such as defense R&D, was used to calculate the overall life for the R&D asset (table 11).

For nondefense transportation and other R&D and for all state and local government R&D assets, BEA adopted the same rate (16 percent) used for the private scientific R&D services industry.

Methodology: Price Measures

Price deflators for R&D are used to develop quantity measures for R&D investment, stocks, and depreciation. Unfortunately, these prices are generally unob-servable, given that much R&D is produced for own-account use and that R&D projects are heterogeneous in nature (so that similar outputs are not available for observation). Consequently, it is generally accepted that measures of own-account R&D investment are derived as the sum of the costs of their inputs, such as compensation, materials and supplies, and capital services. In the absence of direct price measures for R&D output, this input cost approach is the preferred method for constructing R&D price indexes. (9) However, by using an input cost index as a proxy for the output price change, the impact of productivity change on the value of real R&D output or on investment is not captured. Considering the widely held view that R&D expenditures are an important source of increases in productivity, BEA uses the economy-wide measure of multifactor productivity produced by BLS as the estimate of unobserved R&D productivity. This productivity adjustment is made by subtracting the growth rate of private nonfarm business sector multi-factor productivity from the growth rate of the R&D input cost index.

In periods that productivity increases, the price of R&D is adjusted downward because of the underlying assumption that more R&D can be produced per unit of input; conversely, the price of R&D is adjusted upward when productivity declines because less R&D can be produced per unit of input.

Three R&D price indexes are used in the NIPAs and IEAs: a business R&D price index, an academic R&D price index, and a federal own-account R&D price index. All three price indexes are based on an input cost approach. The academic and federal price indexes are used for R&D performed by academic institutions and the federal government, respectively. (All of R&D performed by the federal government is own-account). The business price index is used for all other R&D performers; that is, private business, NPISHs (excluding academic institutions), and state and local government because they are assumed to have similar R&D cost structures.

Measures of real or inflation-adjusted R&D output and investment are derived by deflating each current-dollar R&D estimate with the corresponding R&D price index.

Business R&D price index

The annual business R&D price index is a weighted average of the cost components of the scientific R&D services industry (60 percent) and an "all private industries" aggregate (40 percent). Beginning with 1997, prices for compensation, material inputs, depreciation, and other costs are separately constructed for the scientific R&D services industry and the all private industry group. These prices are then aggregated using a Fisher formula to derive a single business R&D price index. For 1929-1996, prices are constructed for compensation and other costs for the scientific R&D services industry and the all private industries group. These prices are also aggregated using a Fisher formula to derive a single business R&D price index. The price index for each cost component is derived using source data that vary over time because of availability. Table 12 provides a full description of each cost component price.

Weights for each cost component are from NSF surveys of business R&D expenditures. Table 13 presents the weights for each cost component for selected years.

Annual prices are interpolated to derive quarterly prices using quarterly indicator data. For periods prior to 1975, quarterly indicator data are not used, and the annual prices are interpolated without a pattern.

Beginning with 1976, the business R&D price index is interpolated using a quarterly pattern series based on average wages derived from QCEW data. The indicator series for 1976-1990 is based on a three-quarter moving average of seasonally adjusted average wages for all private industries. For 1991 forward, the QCEW-based pattern series is constructed using seasonally adjusted wage and employment data to derive an average wage for the scientific R&D services industry and for all other industries. The growth rate in the average wages for each industry set are then weighted (60 percent for the R&D services industry and 40 percent for all other private industries), and a three-quarter moving average is applied to construct the final indicator series.

Academic R&D price index

Because the characteristics and cost structure of academic institutions differ greatly from other types of for-profit and nonprofit institutions, an academic-specific R&D deflator enhances the accuracy of the estimates of real R&D produced by academic institutions. The academic R&D price index is adapted from an approach outlined in Inflation Measures for Schools, Colleges, and Libraries: 2001 Update, which is a data volume that reports price trends for educational institutions. (10)

The academic R&D price index comprises the following input cost components: R&D personnel wages and salaries, overhead/materials and supplies/purchased services, and economic depreciation. The source of the price data for each cost component varies over time, depending on availability. Table 14 provides the composition of the academic R&D price index for each period.

Two primary sources are used to develop weights for the academic R&D price index: Inflation Measures and NSF's higher education research and development (HERD) surveys. (11) In addition to data from Inflation Measures and the HERD surveys, estimates of current-cost depreciation of equipment and structures used in the production of R&D were used as the weights for the measure of capital services for R&D production.

The weights from each source are used to create an annual time series of weights for each cost component; these are presented in table 15 for selected years. The price index weights and input costs are combined to create the Fisher-weighted academic R&D price index.

Federal own-account R&D price index

Federal own-account R&D prices are based on a weighted average of the prices of various components of federal nondefense consumption expenditures and gross investment, including compensation of general government employees, selected intermediate goods and services, structures, and equipment. Prices for federal own-account R&D are developed by first decomposing current-dollar estimates into the following cost components:

*Compensation of federal employees performing R&D

*Intermediate inputs and overhead

*Consumption of fixed capital (CFC) for structures used in the production of own-account R&D

*CFC for equipment used in the production of own-account R&D

These cost components are then deflated using a variety of BEA price indexes. Compensation of federal employees performing R&D is deflated using the price indexes for federal nondefense compensation of employees, which are based on an indicator derived from BLS employment data. Intermediate inputs and overhead are deflated using an aggregate price built up from prices used to deflate selected components of nondefense intermediate goods and services (excluding the consumption of fixed capital and petroleum), which are largely based on various BLS producer price indexes (PPIs), consumer price indexes, and employment cost indexes. Prices for the CFC for the structures and equipment used in the production of own-account R&D are based on the prices for federal nondefense structures and equipment investment, which are based on a variety of PPIs and BEA price indexes.

Federal nondefense prices are used to deflate both nondefense and defense own-account R&D because they better reflect the mix of input materials and employees used in the R&D process than prices including defense inputs and personnel.

Methodology: Current Quarterly Estimates

BEA's current quarterly estimates of R&D are less detailed than its annual estimates. BEA produces separate estimates for R&D investment of total private business, nonprofit institutions (including private academic institutions), federal government, and state and local government (including public academic institutions). The source data available, and thus the estimation methods used, vary for the advance, second, and third estimate vintages. Real measures are derived by deflating the current-dollar measures by the appropriate price index. Table 16 summarizes the source data used for each R&D investment category and each vintage of estimate.

Private business R&D investment

For the advance estimate vintage, current-dollar business R&D is extrapolated forward from the previous quarter with the growth rate of seasonally adjusted aggregate wages for all private industries derived from average earnings, hours, and employment data from the BLS Current Employment Statistics (CES). (12)

For the second and third vintages, BEA constructs an indicator based on reported R&D expenses from a matched sample of publicly traded companies' quarterly financial statements.

The business R&D price index for the current quarterly estimate is derived as the growth rate in a three quarter moving average of a composite average wage index based on data from the BLS CES in which average wages of the scientific R&D services industry receives a 60 percent weight, and the average wages of all private industries receives a 40 percent weight. (13)

Nonprofit and state and local government R&D investment

The methodology for measuring current-dollar R&D investment of private nonprofit institutions and state and local government is consistent for all three estimate vintages. BEA's estimates of private nonprofit R&D include private academic institutions' R&D, while its estimates of state and local government R&D include public academic institutions' R&D. Both nonprofits and state and local government use an indicator based on the growth rate in an aggregate wage series constructed using BLS CES data for the education and health services industry. The price index for R&D performed by academic institutions is calculated by deriving an average wage for the education and health services industry using CES data.

Federal R&D investment

During the preparation of current quarterly estimates, estimates of federal R&D investment are derived by extrapolating NSF survey data using growth rates based on figures from The Budget of the United State Government. The price index for federal own-account R&D is calculated by using the same source data as the annual price index.

By Marissa J. Crawford, Jennifer Lee, John E. Jankowski, and Francisco A. Moris

Marissa J. Crawford and Jennifer Lee are economists at the Bureau of Economic Analysis (BEA). John E. Jankowski is the Director of the R&D Statistics Program at the National Science Foundation, where Franciso A. Moris is a Senior Analyst.

The authors gratefully acknowledge significant contributions from the following BEA economists: Peter G. Beall, Jennifer A. Bennett, Kyle A. Cisco, Wendy Li, Benjamin A. Mandel, Nicole M. Mayerhauser, Kyle D. Mulgrew, Erich H. Strassner, and Dave B. Wasshausen.

(1.) For further analysis on the impact of R&D capital on multifactor productivity and value-added growth, see Steven Rosenthal, Matthew Russell, Jon D. Samuels, Erich H. Strassner, and Lisa Usher, "Integrated Industry-Level Production Account," Survey of Current Business 94 (August, 2014).

(2.) European Commission, International Monetary Fund, Organisation for Economic Co-operation and Development, United Nations, and World Bank, System of National Accounts 2008 (New York: 2009): paragraph 6.207, 119.

(3.) The NSF surveys collect information on the sources of funds for R&D expenditures and for some sectors, the types and fields of research. For more information on the scope and design of each NSF survey, please see the information on the NSF Web site.

(4.) In the context of the integrated macroeconomic accounts, the assignment of ownership affects the balance sheets of the owning sectors.

(5.) Since 2010, the NSF business and academic surveys have information on grant-based and purchased R&D for federally funded R&D. However, without historical data, BEA cannot properly account for grants and purchased R&D prior to 2010.

(6.) Because the NSF data are based on production costs, the market price of the purchased R&D is not captured. To estimate the market value of R&D, BEA applies a profit markup. BEA calculates this profit markup based on the net operating surplus of the miscellaneous professional, scientific, and technical services industry. This is an industry category from BEA's GDP-by-industry data that contains scientific R&D services within it as a subindustry. Net operating surplus is estimated as gross operating surplus less consumption of fixed capital, and thus reflects a residual profit-like measure for the industry.

(7.) The Census Bureau economic surveys used to estimate R&D sales include the economic census, the service annual survey, and the annual survey of manufactures.

(8.) For more details on R&D business depreciation rates, see Wendy C.Y. Li, "Depreciation of Business R&D Capital," Bureau of Economic Analysis/National Science Foundation Paper (Washington, DC: BEA, 2012).

(9.) In preparing the satellite account, BEA experimented with several R&D price indexes. The featured price index was an aggregate output price index, which was a weighted average of the output prices of the products produced by R&D-intensive industries. This price index exhibited average annual decreases. The business price index used for implementation, based on an input cost approach, into the core accounts increases at about 2 percent per year.

(10.) Research Associates of Washington, Inflation Measures for Schools, Colleges and Libraries: 2001 Update (Washington, DC: Research Associates of Washington, 2001).

(11.) Beginning in 2010, the NSF HERD surveys collect information on R&D expenditures by universities and colleges. The HERD survey is a newly redesigned higher education survey, which collects additional information, compared with its predecessor, the survey of research and development expenditures at universities and colleges. The HERD survey collects information on R&D expenditures by type of cost, including compensation costs, other direct costs, and indirect costs.

(12.) Aggregate wages are the product of the number of employees, weekly hours, and hourly earnings.

(13.) Average wages are the product of weekly hours and hourly earnings divided by the number of employees.

A Brief History of BEA's Capitalization of R&D

The Bureau of Economic Analysis' (BEA) effort to measure R&D investment and to examine the effects of R&D expenditures on the U.S. economy began with a set of satellite accounts. These accounts provided a means of exploring the effect of R&D activity on the economy and a framework through which various methodological and conceptual issues, such as how to measure changes in R&D prices, could be examined. In 1994, BEA first published an R&D satellite account. In 2006, BEA introduced a revised satellite account and updated that account in 2007 and in 2010. This satellite account formed the general framework for how BEA would implement R&D into its core economic accounts.

However, the R&D estimates incorporated into BEAs core economic accounts differ from the satellite account estimates in two ways:

*Source data differences. To incorporate R&D as investment into the national income and product accounts (NIPAs) and into the industry economic accounts (IEAs), BEA uses additional data sources, such as the Census Bureau's economic census and information from publicly-traded companies.

*Classification differences. In the core accounts, R&D software remains in BEA's estimates of software investment. In the satellite account, BEA had removed R&D software from the software investment estimate, retaining it in R&D investment because the R&D satellite account was designed to focus on R&D as a capitalized asset. The development of the R&D satellite account was funded by the National Center for Science and Engineering Statistics (NCSES) at the National Science Foundation (NSF). NSF data on R&D expenditures continue to serve as the primary data source for BEA's estimates. NSF and BEA are working together to collaborate on survey methodologies appropriate for BEA's R&D estimates. (1)

(1.) NCSES R&D statistics are collected from an expenditure, or current-cost perspective, following international guidelines in Frascati Manual: Proposed Standard Practice for Surveys on Research and Experimental Development (Paris: Organisation for Economic Cooperation and Development, 2002). See also "R&D Recognized as Investment in U.S. Gross Domestic Product Statistics: GDP Increase Lowers R&D-to-GDP Ratio," NSF InfoBrief (forthcoming on the NSF's Web site).

BEA's Company-to-Establishment Adjustment for the Scientific R&D Services Industry

The National Science Foundation (NSF) publishes statistics for the costs incurred in the production of research and development (R&D) performed by the scientific R&D services industry. These statistics are collected on a company basis rather than on an establishment basis. As a result, they exclude production costs for R&D establishments that are part of a multiunit company but that are classified in another industry, such as the pharmaceutical and medicine manufacturing industry.

For the Bureau of Economic Analysis (BEA) to produce R&D investment estimates, it must convert the NSF company basis data on the scientific R&D services industry to an establishment basis in keeping with the BEA industry accounts. The conversion yields an estimate of R&D costs that includes expenditures on a company basis within the scientific R&D services industry plus expenditures for scientific R&D performed by establishments that are part of multiunit companies outside of the industry. (1)

NSF cost data include (1) expenditures for R&D paid for by the company, which are treated as own-account R&D, and (2) expenditures for R&D paid for by others, which are treated as R&D sales. For most of the time series, NSF does not differentiate between these two categories. Beginning with the publication of the business research and development and innovation survey (BRDIS) in 2008, however, NSF started reporting data that can be used to approximate costs for own-account R&D, which was approximately 50 percent of total expenditures in 2008 and 2009. Based on the first couple of years of BRDIS results, BEA makes the assumption that half of the NSF scientific R&D services industry expenditures are associated with the production of own-account R&D. The other half of the expenditures are associated with R&D produced for sale.

To estimate R&D costs incurred by establishments outside of companies classified in the scientific R&D services industry, BEA uses data from the Census Bureau's economic census and service annual survey (SAS). These data are already on an establishment basis and are based on receipts, or sales, of R&D (2). Economic censuses, conducted every 5 years, provide detailed receipts of R&D services produced by the scientific R&D services industry. For noneconomic census years, SAS data are used to interpolate between the economic census years. Because the economic census data and the SAS data are based on receipts, they do not cover own-account R&D expenses. BEA assumes that NSF scientific R&D services company-based sales, estimated as half of published NSF R&D expenditures, are included in the Census Bureau-based data.

So from the Census Bureau-based data, an estimate of total establishment-based R&D sales is calculated. And from the NSF data, an estimate of sales by establishments that are part of a scientific R&D services company is calculated. Subtracting the NSF sales from the Census Bureau-based sales produces an estimate of R&D sales for an R&D establishment that is part of a multiunit company. This amount is moved from the primary industry (for example, the pharmaceutical or computer manufacturing industries) as reported by the NSF company-based data to the scientific R&D services industry.

(1.) BEA assumes that all establishments that are part of a scientific R&D services company are also scientific R&D services establishments.

(2.) The Census Bureau provides a special tabulation during economic census years for auxiliary establishments. The primary function of an auxiliary establishment is to manage, administer, service, or support the activities of the other establishments of the company. There are no market transactions because all of the output is provided to the rest of the company and expenses instead of receipts are collected for these establishments.

Table 1. Primary Source Data for BEA's Estimates of R&D Investment
                              and Output

                    1929-1941  1942-1946  1947-1952  1953-1954  1955

Private business     A,B,C      A,B,C,D    A,B,C         E       F
Academic
institutions                               A,N
State and local
government                         R
Nonprofit
institutions                      A,N                  A,C
Federal
government                        A,N

                    1956  1597-1959  1960  1961  1962-1963  1964  1965

Private business     G        H           I                  J
Academic
institutions                            O
State and local
government                              R
Nonprofit
institutions                  AD      AE          F           AF    F
Federal
government                            AM,AN

                     1966  1967  1968-1969  1970  1971  1972  1973

Private business                            K
Academic
institutions                                O
State and local
government                T       U         F             V
Nonprofit
institutions          AG          F         AH      F          AI
Federal
government                                               AM,AN

                     1974-1976  1977  1978-1980  1981-1986  1987-1988

Private business                      K                       L
Academic
institutions                                  P
State and local
government            F          W            F                X
Nonprofit
institutions                                  F
Federal
government                                AM,AN

                      1989-1994  1995  1996-1997  1998-2001  2002

Private business                             L
Academic
institutions                                 P
State and local
government              F          Y         F
Nonprofit
institutions                 F              AJ    AK          AL
Federal
government                                 AM,AN

                      2003-2005  2006  2007  2008  2009  2010-2012

Private business                       L              M
Academic
institutions                           P                        Q
State and local
government            F                Z        F    AA        AB
Nonprofit
institutions                  AK       AL            AK
Federal
government                                AM,AN

A Nestor Terleckyj, "Research and Development: Its Growth and
   Composition," Studies in Business Economics 82 (1963).
B George Perazich and Philip M. Field, Industrial Research and Changing
   Technology (University of California: Work Projects Administration,
   1940).
C Science and Engineering in American Industry: Final Report on a
   1953-54 Survey (Washington, DC: NSF).
D An adjustment for Manhattan Project R&D expenditures based on data
   from Richard Hewlett and Oscar Anderson, Jr., The New World,
   1939/1946:
A History of the United States Atomic Energy Commission, vol. I.
   (University Park, PA : Pennsylvania State University Press, 1962).
E Science and Engineering in American Industry (Washington, DC: NSF).
F Linking two NSF survey years using straight line interpolation, data
   from NSF surveys, or BEA NIPA data.
G Science and Engineering in American Industry: Report on a 1956 Survey
   (Washington, DC: NSF).
H Funds for Research and Development in Industry (Washington, DC: NSF).
I Research and Development in Industry (Washington, DC: NSF).
J Basic Research, Applied Research, and Development in Industry
   (Washington, DC: NSF).
K Research and Development in Industry (Washington, DC: NSF).
L Survey of Industrial Research & Development (Washington, DC: NSF).
M Business Research and Development and Innovation Survey (Washington,
   DC: NSF).
N John Kendrick, The Formation and Stocks of Total Capital (New York:
   National Bureau of Economic Research, 1976).
O "R&D Expenditures of Universities and Colleges by Source of Funds:
   Fiscal Years 1953-90" in Selected Data on Academic Science and
   Engineering R&D Expenditures, FY 1990 (Washington, DC: NSF).
P Survey of Research and Development Expenditures at Universities and
   Colleges (Washington, DC: NSF).
Q Higher Education Research and Development Survey (Washington, DC:
   NSF).
R BEA data on gross investment for state and local governments.
S R&D Activities in State Government Agencies Fiscal Years 1964 and
   1965 (Washington, DC: NSF).
T R&D Activities of Local Governments: Fiscal Years 1966 and 1967
   (Washington, DC: NSF).
U Research and Development in Local Governments: Fiscal Years 1968
   and 1969 (Washington, DC: NSF).
V Research and Development in State Government Agencies: Fiscal
   Years 1972 and 1973 (Washington, DC: NSF).
W Research and Development in State and Local Governments Fiscal
   Year 1977 (Washington, DC: NSF).
X Research and Development Expenditures of State Government Agencies:
   Fiscal Years 1987 and 1988 (Washington, DC: NSF).
Y Survey of State Research and Development Expenditures: Fiscal Year
   1995 (Washington, DC: NSF).
Z State Agency Research and Development Expenditures (Washington,
   DC: NSF).
AA State Government Research and Development (Washington, DC: NSF).
AB Survey of State Government Research and Development (Washington,
  DC: NSF).
AC Research Expenditures of Foundations and Other Nonprofit
   Institutions, 1953-54 (Washington, DC: NSF).
AD Scientific Research and Development of Nonprofit Organizations: 1956
   Expenditures and 1957 Manpower (Washington, DC: NSF).
AE Research and Other Activities of Private Foundations, 1960
   (Washington, DC: NSF).
AF Scientific Activities of Nonprofit Institutions: 1964 Expenditures
   and January 1965 Manpower (Washington, DC: NSF).
AG Scientific Activities of Nonprofit Institutions: 1966 Expenditures
   and January 1967 Manpower (Washington, DC: NSF).
AH Scientific Activities of Independent Nonprofit Institutions, 1970
   (Washington, DC: NSF).
AI R&D Activities of Independent Nonprofit Institutions, 1973
   (Washington, DC: NSF).
AJ Research and Development Funding and Performance by Nonprofit
   Organizations: Fiscal Years 1996 and 1997 (Washington, DC: NSF).
AK Census Bureau data on tax-exempt establishments in the scientific
   R&D services industry (NAICS 5417), data from NSF R&D surveys, and
   BEA data.
AL Census Bureau data from the Economic Census, 2002 and 2007
AM Survey of Federal Funds for Research and Development (Washington,
   DC: NSF).
AN The Budget of the United States Government for all fiscal years.

Table 2. Total Business R&D Including Software R&D
                        [Billions of dollars]

                             1990  1991  1992  1993  1994  1995  1996

Software R&D in BEA's
software investment             2     5     7     7     8     9    11
Business R&D investment
excluding software R&D         83    90    92    92    94   105    116
Business R&D investment
including software R&D         86    95    99   100   102   114    127

                             1997  1998  1999  2000  2001

Software R&D in BEA's
software investment            14    17    22    29    31
Business R&D investment
excluding software R&D        126   134   146   160   161
Business R&D investment
including software R&D        140   151   168   189   193

                             2002  2003  2004  2005  2006  2007  2008

Software R&D in BEA's
software investment            31    35    35    38    43    47    50
Business R&D investment
excluding software R&D        155   156   162   176   192   211   225
Business R&D investment
including software R&D        186   191   197   214   235   258   275

                             2009  2010  2011  2012

Software R&D in BEA's
software investment            51    51    61    62
Business R&D investment
excluding software R&D        214   224   231   241
Business R&D investment
including software R&D        265   275   293   303

Note. Estimates may not sum to totals because of rounding.

Table 3. BEA's R&D Investment Reconciled to NSF R&D Expenditures for
                        Total Private Business
                       [Millions of dollars]

Line                                     Description

1     R&D investment by    Calculated as the sum of own-account R&D
      private business     (line 2) and purchased R&D (line 8) and
                            equals line 7 in NIPA table 5.6.5.
2     Own-account R&D
      investment
3     NSF R&D              Funds for industrial R&D performed in the
      expenditures for     United States, excludes FFRDCs. For 2007,
      all industries       values come from NSF's SIRD. For 2008 and
                           2009, values come from NSF's BRDIS.
4     Plus: SIRD to BRDIS  An adjustment based on the NSF's survey of
      time series          federal funds for research and development
      adjustment           to bridge survey differences between the SIRD
                           and BRDIS for R&D funded by the federal
                           government.
5     Plus: Adjustment to  A capital consumption adjustment to convert
      account for the      business depreciation to economic
      full cost of         depreciation.
      production
6     Less: Software R&D   The removal of software R&D that is in BEA's
                           software estimates to avoid double-counting.
7     Less: Census Bureau  R&D sales, which include expor ts and R&D
      sales of R&D         paid for by the government, NPISHs, and other
      converted to         businesses, from BEA's annual input-output
      expenses             (I-O) R&D services estimates, converted to a
                           cost-basis. BEA's I-O estimates are based on
                           the Census Bureau's economic census, SAS, and
                           annual survey of manufactures data. FFRDCs
                           are removed from the Census Bureau data to be
                           consistent with the NSF data.
8     Total purchased R&D
9     Impor ts of R&D      Imports are based on BEA International
                           Transaction Accounts data for research,
                           development, and testing services.
10    Plus: R&D purchased  Industry-funded R&D performed by academic
      from academic        institutions from unpublished detail from
      institutions         BEA's academic R&D estimates that are based
                           on data from NSF's survey of research and
                           development expenditures at universities and
                           colleges.
11    Plus: R&D purchased  Industry-funded R&D performed by nonprofit
      from NPISHs          organizations from unpublished detail from
                           BEA's NPISH R&D estimates that are based on
                           Census Bureau data for tax-exempt scientific
                           R&D establishments.
12    Plus: R&D purchased  Industry-funded R&D performed by state and
      from state and       local government from unpublished detail from
      local governments    BEA's state and local government R&D
                           estimates that are based on data from NSF's
                           state government research and development
                           survey.
13    Plus: R&D purchased  Industry-funded R&D performed by FFRDCs from
      from FFRDCs          unpublished detail from BEA's estimates of
                           FFRDC R&D that are based on data from NSF's
                           R&D expenditures at federally funded R&D
                           centers survey.
14   Plus: Purchases of    R&D in the social sciences and humanities
     R&D in the social     from BEA's annual I-O estimates that are
     sciences and          based on the Census Bureau's economic census
     humanities            and SAS data.
     from for-profit R&D
     services
     establishments
     (NAICS 54172)
15   Plus: Cost of R&D     R&D transferred from company headquarters
     performed in company  and R&D services establishments to the
     headquarters          primary industry of the company, which shows
     (NAICS 551) and R&D   up as purchased R&D rather than as
     services subsidiary   own-account R&D. The estimates are from a
     and auxiliary         reconciliation of NSF SIRD data (for 2007)
     establishments        and BRDIS data (for 2008-2009) with BEA's
     (NAICS 5417           annual I-O R&D estimates that are based on
                           the Census Bureau's economic census and
                           SAS data and a special tabulation of
                           auxiliary establishments.
16   Plus: R&D purchased   Purchases from BEA's annual I-O R&D estimates
     from other            after subtracting sales to other sectors,
     businesses,           exports, and the cost of R&D that is
     specifically R&D      transferred to other establishments within
     services              the company. BEA's annual I-O estimates
     establishments        are based on the Census Bureau's economic
                           census and SAS data.

Line                         2007            2008           2009

1     R&D investment by
      private business      210,801         224,522        213,951
2     Own-account R&D
      investment            115,615         114,427         97,130
3     NSF R&D
      expenditures for
      all industries        269,267         290,680        282,393
4     Plus: SIRD to BRDIS
      time series
      adjustment              9,167               0         -3,816
5     Plus: Adjustment to
      account for the
      full cost of
      production               -101              -77          -418
6     Less: Software R&D     47,655           51,036        52,618
7     Less: Census Bureau
      sales of R&D
      converted to
      expenses              115,063          125,141       128,411
8     Total purchased R&D    95,186          110,096       116,821
9     Imports of R&D         10,843           13,580        13,845
10    Plus: R&D purchased
      from academic
      institutions            2,924            3,148         3,195
11    Plus: R&D purchased
      from NPISHs             5,605            7,122         9,789
12    Plus: R&D purchased
      from state and
      local governments           2                3             3
13    Plus: R&D purchased        86               94            94
      from FFRDCs
14   Plus: Purchases of
     R&D in the social
     sciences and
     humanities
     from for-profit R&D
     services
     establishments
     (NAICS 54172)            1,618            1,735         1,754
15   Plus: Cost of R&D
     performed in company
     headquarters
     (NAICS 551) and R&D
     services subsidiary
     and auxiliary
     establishments
     (NAICS 5417             64,562            71,803        76,323
16   Plus: R&D purchased
     from other
     businesses,
     specifically R&D
     services
     establishments           9,546             12,611        11,817

BRDIS      Business research and development and innovation survey
FFRDCs     Federally funded research and development centers
NAICS      Nor th American Industry Classification System
NPISHs     Nonprofit institutions serving households
NSF        National Science Foundation
R&D        Research and development
SAS        Service annual survey
SIRD       Survey of industrial research and development

NOTE. Estimates may not sum to totals because of rounding.

Table 4. BEA's R&D Investment Reconciled to NSF R&D Expenditures for
 the Pharmaceutical and Medicine Manufacturing Industry (NAICS 3254)
                       [Millions of dollars]
Line                                        Description

1     R&D investment by    Calculated as the sum of own-account R&D
      the pharmaceutical   (line2) and purchased R&D (line9) and equals
      and medicine          line 9 in NIPA table 5.6.5.
      manufacturing
      industry
2     Own-account R&D
      investment
3     NSF expenditures     Funds for industrial R&D performed in the
      for R&D for the      United States, excludes FFRDCs and funds
      pharmaceutical and   from the federal government. For 2007, values
      medicine             come from NSF's SIRD. For 2008 and 2009,
      manufacturing        values come from NSF's BRDIS.
      industry (NAICS
      3254)
4     Plus: SIRD to BRDIS  An adjustment, based on R&D expenses
      time series          reported on publicly-traded companies' annual
      adjustment           financial statements, is made to bridge
                           survey differences between the SIRD and
                           BRDIS.
5     Plus: Adjustment to  A capital consumption adjustment is made to
      account for the      convert business depreciation to economic
      full cost of         depreciation.
      production
6     Less: Software R&D   The removal of software R&D that is counted
                           in BEA's software estimates to avoid
                           double-counting.
7     Less: The cost of    Based on BEA's company to establishment
      R&D performed in     adjustment, which is from a reconciliation
      company              of NSF SIRD (for 2007) and BRDIS (for
      headquarters (NAICS  2008-2009) data and BEA's annual input-output
      551) and R&D         (I-O) R&D estimates. BEA's I-O estimates are
      services subsidiary  based on the Census Bureau's economic
      and auxiliary        census and SAS data and a special tabulation
      establishments       of auxiliary establishments.
      (NAICS 5417
8     Less: Exports of     The cost of exports is based on BEA
      R&D                  International Transaction Accounts data for
                           research, development, and testing services.
9    Total purchased R&D
10   Impor ts of R&D       Imports are based on BEA International
                           Transaction Accounts data for research,
                           development, and testing services.
11   Plus: R&D purchased   Total for all industry-funded R&D performed
     from academic         by academic institutions from unpublished
     institutions          detail from BEA's academic R&D estimates that
                           are based on data from NSF's survey of
                           research and development expenditures at
                           universities and colleges. Information on
                           industry shares comes from SIRD data (for
                           2007) and BRDIS data (for 2008-2009) on
                           company-funded R&D performed by universities
                           and colleges.
12   Plus: R&D purchased   Total for all industry-funded R&D performed
     from NPISHs           by nonprofit organizations from unpublished
                           detail from BEA's NPISH R&D estimates. BEA's
                           NPISH estimates are based on Census Bureau
                           data for tax-exempt scientific R&D
                           establishments. Information on industry
                           shares comes from SIRD data on company-funded
                           R&D performed by nonprofit organizations
                           (other than universities and colleges).
13   Plus: R&D purchased   Total for all industry-funded R&D performed
     from state and local  by state and local government from
     governments           unpublished detail from BEA's state and local
                           government R&D estimates that are based on
                           data from NSF's state government research and
                           development survey. Industry share is proxied
                           with SIRD data on company-funded R&D
                           performed by academic institutions and
                           nonprofit organizations.
14   Plus: R&D purchased   Total for all industry-funded R&D performed
     from FFRDCs           by FFRDCs comes from unpublished detail from
                           BEA's estimates of FFRDC R&D that are based
                           on data from NSF's R&D expenditures at
                           federally funded R&D centers survey. Industry
                           share is proxied with SIRD data on
                           company-funded R&D performed by academic
                           institutions and nonprofit organizations.
15   Plus: The cost of     R&D costs transferred from company
     R&D performed in      headquarters and R&D services establishments
     company               to the pharmaceutical and medicine
     headquarters (NAICS   manufacturing industry, described in line 7.
     551) and R&D
     services subsidiary
     and auxiliary
     establishments.
16   Plus: R&D purchased   Total for all industries' purchases from
     from other            BEA's annual I-O R&D estimates after
     businesses,           subtracting sales to other sectors, exports,
     specifically R&D      and R&D that is transferred to other
     services              establishments within the company. The
     establishments        residual is assumed to equal R&D sold to
                           other businesses. BEA's annual I-O estimates
                           are based on the Census Bureau's economic
                           census and SAS data. Industry shares are
                           estimated based on SIRD data (for 2007) and
                           BRDIS data (2008-2009) on business purchases
                           of R&D from other domestic for-profit
                           companies.

Line                         2007            2008           2009

1     R&D investment by
      the pharmaceutical
      and medicine
      manufacturing
      industry              56,756          61,235         26,903
2     Own-account R&D
      investment            27,780          27,363         21,888
3     NSF expenditures
      for R&D for the
      pharmaceutical and
      medicine
      manufacturing
      industry (NAICS
      3254)                 47,624          47,994         44,823
4     Plus: SIRD to BRDIS
      time series
      adjustment                 0             519              0
5     Plus: Adjustment to
      account for the
      full cost of
      production              -653            -382           -642
6     Less: Software R&D
7     Less: The cost of
      R&D performed in
      company
      headquarters (NAICS
      551) and R&D
      services subsidiary
      and auxiliary
      establishments
      (NAICS 5417                D               D              D
8     Less: Expor ts of
      R&D                   16,273          18,273         18,751
9    Total purchased R&D    28,977          33,872         35,872
10   Imports of R&D              D               D              D
11   Plus: R&D purchased
     from academic
     institutions            1,824           2,012          1,572
12   Plus: R&D purchased
     from NPISHs             2,701           2,847          3,913
13   Plus: R&D purchased
     from state and local
     governments                  1              2              2
14   Plus: R&D purchased
     from FFRDCs                  D              D              D
15   Plus: The cost of
     R&D performed in
     company
     headquarters (NAICS
     551) and R&D
     services subsidiary
     and auxiliary
     establishments.         16,273         18,037         18,751
16   Plus: R&D purchased
     from other
     businesses,
     specifically R&D
     services
     establishments           4,760          7,148          6,555

BRDIS     Business research and development and innovation survey
D         Data suppressed to avoid the disclosure of data from
          individual companies.
FFRDCs    Federally funded research and development centers
NAICS     North American Industry Classification System
NPISH     Nonprofit institutions serving households
NSF       National Science Foundation
R&D       Research and development
SAS       Service annual survey
SIRD      Survey of industrial research and development

NOTE. Estimates may not sum to totals because of rounding.

Table 5. BEA's R&D Investment Reconciled to NSF R&D Expenditures for
          the Scientific R&D Services Industry (NAICS 5417)
                       [Millions of dollars]
Line                                       Description
1     R&D investment by    Calculated as the sum of own-account R&D
      the scientific R&D   (line 2) and purchased R&D (line 8) and
      services industry    equals line 17 in NIPA table 5.6.5.
2     Own-account R&D
      investment
3     NSF expenditures     Funds for industrial R&D performed in the
      for R&D for the      United States, excludes FFRDCs and funds from
      scientific research  the federal government. For 2007, values
      and development      come from NSF's SIRD. For 2008 and 2009,
      services industry    values come from NSF's BRDIS.
      (NAICS 5417)
4     Plus: Adjustment to  A capital consumption adjustment to convert
      account for the      business depreciation to economic
      full cost of         depreciation.
      production
5     Less: Software R&D   The removal of software R&D that is already
                           counted in BEA's software estimates to avoid
                           double-counting.
6     Equals: NSF R&D
      expenditures with
      BEA adjustments
7     Divide by 2          Assuming that R&D expenditures for
                           own-account use are 50 percent of R&D
                           expenditures, then R&D that is sold is the
                           remaining 50 percent.
8     Total purchased R&D
9     Imports of R&D       Imports are based on BEA International
                           Transaction Accounts data for research,
                           development, and testing services.
10    Plus: R&D purchased  Total for all industry-funded R&D performed
      from academic        by academic institutions from unpublished
      institutions         detail from BEA's academic R&D estimates
                           that are based on data from NSF's survey of
                           research and development expenditures at
                           universities and colleges. Information on
                           industry shares comes from SIRD data
                           (for 2007) and BRDIS data (for 2008-2009)
                           data on company-funded R&D performed by
                           universities and colleges.
11    Plus: R&D purchased  Total for all industry-funded R&D performed
      from NPISHs          by NPISHs from unpublished detail from BEA's
                           estimates based on Census Bureau data for
                           tax-exempt scientific R&D establishments.
                           Information on industry shares from SIRD data
                           on company-funded R&D performed by nonprofit
                           organizations (other than universities and
                           colleges).
12    Plus: R&D purchased  Total for all industry-funded R&D performed
      from state and       by state and local government from
      local governments    unpublished detail from BEA's state and local
                           government R&D estimates that are based on
                           data from NSF's state government research and
                           development survey. Industry share is proxied
                           with SIRD data on company-funded R&D
                           performed by academic institutions and
                           nonprofit organizations.
13    Plus: R&D purchased  Total for all industry-funded R&D performed
      from FFRDCs          by FFRDCs from unpublished detail from BEA's
                           estimates of FFRDC R&D that are based on data
                           from NSF's R&D expenditures at federally
                           funded R&D centers survey. Industry share is
                           proxied with SIRD data on company-funded R&D
                           performed by academic institutions and
                           nonprofit organizations.
14    Plus: Cost of R&D    The costs incurred in NAICS 551
      performed in         establishments are transferred as the sum of
      company              production costs.
      headquarters (NAICS
      551) for use in
      scientific R&D
      services
      establishments
15    Plus: R&D purchased  Total for all industries' purchases from
      from other           BEA's annual input-output (I-O) R&D estimates
      businesses,          establishments within the company. The
      specifically R&D     residual is assumed to equal R&D sold to
      services             other businesses. BEA's annual I-O estimates
      establishments       are based on the Census Bureau's economic
                           census and SAS data. Industry shares are
                           estimated based on SIRD data (for 2007) and
                           BRDIS data (for 2008-2009) on business
                           purchases of R&D from other domestic
                           for-profit companies.

Line                           2007          2008            2009

1     R&D investment by
      the scientific R&D
      services industry       7,256          9,200          9,068
2     Own-account R&D
      investment              5,663          6,936          6,326
3     NSF expenditures
      for R&D for the
      scientific research
      and development
      services industry
      (NAICS 5417)           12,017         14,798         13,613
4     Plus: Adjustment to
      account for the
      full cost of
      production                 45             22            -15
5     Less: Software R&D        795            948            945
6     Equals: NSF R&D
      expenditures with
      BEA adjustments        11,267         13,872         12,653
7     Divide by 2             5,633          6,936          6,326
8     Total purchased R&D
9     Imports of R&D          1,622         2, 264          2,742
10    Plus: R&D purchased
      from academic
      institutions                D              D              D
11    Plus: R&D purchased
      from NPISHs                 7            187            455
12    Plus: R&D purchased
      from state and
      local governments         159            441            607
13    Plus: R&D purchased
      from FFRDCs                 0              0              0
14    Plus: Cost of R&D
      performed in
      company
      headquarters (NAICS
      551) for use in
      scientific R&D
      services
      establishments              1              1             1
15    Plus: R&D purchased
      from other
      businesses,
      specifically R&D
      services
      establishments            807             774           718


BRDIS  Business research and development and innovation survey
D      Data suppressed to avoid the disclosure of data from individual
        companies.
FFRDCs Federally funded research and development centers
NAICS  North American Industry Classification System
NPISH  Nonprofit institutions serving households
NSF    National Science Foundation
R&D    Research and development
SAS    Service annual survey
SIRD   Survey of industrial research and development

NOTE. Estimates may not sum to totals because of rounding.

Table 6. BEA's R&D Investment Reconciled to NSF R&D Expenditures for
                       Federal Government
                     [Millions of dollars]
Line                                         Description

1     R&D investment by    Calculated as the sum of own-account R&D
      the federal          (line 2) and purchased R&D (line 12) and
      government           equals line 16 in NIPA table 3.9.5.
2     Own-account R&D      Calculated as the BEA-adjusted NSF data
      investment           (line 9) less small equipment costs (line 10)
                           plus adjustment to account for the full cost
                           of production (line 11).
3     NSF R&D total
      obligations          Survey of federal funds for R&D.
4     NSF R&D intramural
      obligations          Survey of federal funds for R&D.
5     Convert NSF R&D      Uses a 3-year weighted moving average in
      obligations to R&D   which the most recent year (time period t)
      outlays              receives 50 percent of the weight, t-1
                           receives 30 percent of the weight, and t-2
                           receives 20 percent of the weight.
5a    Adjust the NSF R&D   Uses the calculation in line 5 to convert the
      total obligations    NSF R&D total obligations data (line 3) to
                           outlays data.
5b    Adjust the NSF R&D   Uses the calculation in line 5 to convert the
      intramural           NSF R&D intramural obligations data (line 4)
      obligations          to outlays data.
6     Create own-account
      ratio                Equals line 5b divided by line 5a.
7     NSF R&D total        BEA uses NSF obligations data by performer
      outlays              from NSF's Survey of federal funds to
                           allocate the federal outlays data into
                           performers because NSF outlays data is not
                           split between performer.
8     Calculate            Applies the own-account obligations ratio
      own-account outlays  (line 6) to the total R&D outlays data
                           (line 7) to derive the own-account outlays.
9     Convert fiscal year  CY(t) = 0.75*FY(t) + 0.25*FY(t-1).
      (FY) data to
      calendar year (CY)
      data
10    Less: Small          Removes small equipment costs included in
      equipment costs      NSF's R&D expenditure data in the Survey
                           of federal funds.
11    Plus: Adjustment to  A capital consumption adjustment is made to
      account for the      convert business depreciation to economic
      full cost of         depreciation.
      production
12    Total purchased R&D
13    R&D purchased from   Total for all federally-funded R&D performed
      industry             by private business from unpublished detail
                           from BEA's private business R&D estimates
                           that are based on data from NSF's SIRD (2007)
                           and BRDIS (2008 and 2009).
14    Plus: R&D purchased  Total for all federally-funded R&D performed
      from academic        by academic institutions from unpublished
      institutions         detail from BEA's academic R&D estimates
                           that are based on data from NSF's Survey of
                           research and development expenditures at
                           universities and colleges.
15    Plus: R&D purchased  Total for all federally-funded R&D performed
      from FFRDCs          by FFRDCs from unpublished detail from BEA's
                           estimates of FFRDC R&D that are based on data
                           from NSF's R&D expenditures at federally
                           funded R&D centers survey.
16    Plus: R&D purchased  Total for all federally-funded R&D performed
      from NPISHs          by nonprofit organizations from unpublished
                           detail from BEA's NPISHs R&D estimates that
                           are based on Census Bureau data for
                           tax-exempt scientific R&D establishments.
                           Federal funding of nonprofit R&D performance
                           is proxied using NSF's Survey of federal
                           funds data.
17    Plus: R&D purchased  Total for all federally-funded R&D performed
     from state and local  by state and local government from
     governments           unpublished detail from BEA's state and local
                           government R&D estimates that are based on
                           data from NSF's state government research and
                           development survey.

Line                          2007          2008           2009

1     R&D investment by
      the federal
      government            122,531       127,210        129,287
2     Own-account R&D
      investment             28,062        28,923         29,655
3     NSF R&D total
      obligations           127,263       127,106        141,090
4     NSF R&D intramural
      obligations
5     Convert NSF R&D
      obligations to R&D
      outlays                29,932        29,637        31,0543
5a    Adjust the NSF R&D
      total obligations     123,920       126,078        134,129
5b    Adjust the NSF R&D
      intramural
      obligations            28,443        29,333         30,649
6     Create own-account
      ratio                       0.23          0.23           0.23
7     NSF R&D total
      outlays               121,875       123,059
8     Calculate
      own-account outlays    28,085        28,495         29,920
9     Convert fiscal year
      (FY) data to
      calendar year (CY)
      data                   27,577        28,393         29,092
10    Less: Small
      equipment costs         1,136         1,209          1,245
11    Plus: Adjustment to
      account for the
      full cost of
      production              1,624         1,740          1,809
12    Total purchased R&D    94,469        98,287         99,635
13    R&D purchased from
      industry               44,234        46,424         44,350
14    Plus: R&D purchased
      from academic
      institutions           28,704        29,570         31,878
15    Plus: R&D purchased
      from FFRDCs            13,789         7,578          8,155
16    Plus: R&D purchased
      from NPISHs


              Table 7. Funding of R&D by Performer, 2007
                      [Millions of dollars]

Funder                   NIPA investment          Performer
                          sector            Federal     State and local
                                             government    government


Federal government    Federal government       28,230           160
State and local       State and local
government            government                                273
Private business (1)  Private fixed
                      investment                                  2
Nonprofit             Private fixed
institutions (2)      investment                                 18
Public academic       State and
institutions (3)      local government
Private academic      Private fixed
institutions (4)      investment
Total output
/performance                                   28,230           454

Funder                                   Performer
                      Private       Nonprofit       Public academic
                      business(1)  institutions(2)  institutions(3)

Federal government     48,930       10,795           19,487
State and local
government                251        1,445            3,097
Private business (1)  202,273        5,605            1,982

Nonprofit
institutions (2)          677        5,717            2,577
Public academic
institutions (3)                       123            8,546
Private academic
institutions (4)                       123
Total output
/performance          252,131       23,809           35,688

Funder                Private academic           Total investment/
                      institutions (4)            funding

Federal government     14,929                       122,531
State and local
government                368                         5,434
Private business (1)      942                       210,804

Nonprofit
institutions (2)        1,445                        10,434
Public academic
institutions (3)                                      8,669
Private academic
institutions (4)       1,960                          2,083
Total output
/performance          19,645                        359,955

NAICS North American Industry Classification System
NIPAs National income and product accounts
R&D Research and development
(1.) Includes industry-administered federally funded research and
   development centers.
(2.) Includes nonprofit-administered federally funded research and
   development centers.
(3.) Includes public academic-administered federally funded research
   and development centers.
(4.) Includes private academic-administered federally funded research
   and development centers.

NOTE. Estimates may not sum to totals because of rounding.

            Table 8. Funding of R&D by Performer, 2012
                   [Millions of dollars]

Funder                   NIPA investment          Performer
                          sector            Federal     State and local
                                             government    government


Federal government    Federal government       35,566           279
State and local       State and local
government            government                                294
Private business (1)  Private fixed
                      investment                                  4
Nonprofit             Private fixed
institutions (2)      investment                                 28
Public academic       State and
institutions (3)      local government
Private academic      Private fixed
institutions (4)      investment
Total output
/performance                                   35,566           605

Funder                                   Performer
                      Private       Nonprofit       Public academic
                      business(1)  institutions(2)  institutions(3)

Federal government     45,624       13,959           23,020
State and local
government                317        2,944            3,240
Private business (1)  227,047       10,899            2,073

Nonprofit
institutions (2)        1,162       10,422            3,192
Public academic
institutions (3)                       232           10,164
Private academic
institutions (4)                       225
Total output
/performance          274,149       38,681           41,689

Funder                Private academic           Total investment/
                      institutions (4)            funding

Federal government     18,130                       136,579
State and local
government                399                         7,193
Private business (1)    1,165                       241,188

Nonprofit
institutions (2)        1,822                        16,627
Public academic
institutions (3)                                     10,396
Private academic
institutions (4)       3,081                          3,306
Total output
/performance          24,598                        415,289

NAICS North American Industry Classification System
NIPAs National income and product accounts
R&D Research and development
(1.) Includes industry-administered federally funded research and
   development centers.
(2.) Includes nonprofit-administered federally funded research and
   development centers.
(3.) Includes public academic-administered federally funded research
   and development centers.
(4.) Includes private academic-administered federally funded research
   and development centers.

NOTE. Estimates may not sum to totals because of rounding.

Table 9. Indicators for Quarterly Interpolation of R&D Investment

                   1947-1990           1991-2007       2008 forward
Business        No quarterly      QCEW-based      Seasonally adjusted
                    pattern       composite wage       R&D expenses
                                  indicator        reported on quarterly
                                                   financial statements
Academic
 institutions               No quarterly pattern
NPISHs                      No quarterly pattern
State and local
 government                 No quarterly pattern
Federal
 government                 No quarterly pattern

NPISHs Nonprofit institutions serving households
QCEW Quarterly census of employment and wages
R&D Research and development

Table 10. Business R&D Depreciation Rates

NAICS                 Industry                            Depreciation
code                                                     rate (percent)

3254       Pharmaceutical and medicine manufacturing          10
3341       Computers and peripheral equipment                 40
            manufacturing
3342       Communications equipment manufacturing             27
3344       Semiconductor and other electronic component       25
            manufacturing
3345       Navigational, measuring, electromedical,           29
            and control instruments manufacturing
3361-3363  Motor vehicles, bodies and trailers,               31
            and parts manufacturing
3364       Aerospace products and parts manufacturing         22
5112       Software publishers                                22
5415       Computer system design and related services        36
5417       Scientific research and development services       16

NAICS North American Industry Classification System
R&D Research and development

Table 11. Government R&D Depreciation Rates

                                      Depreciation
                                      rate (percent)
    Federal government
Defense
  Purchased R&D                          20
  Own-account R&D                        16
Nondefense
  Aerospace R&D                           7
  Health R&D                              9
  Energy R&D                              9
  Other R&D                              16
    State and local government
Own-account R&D                          16

R&D Research and development

Table 12. Business R&D Price Index, Cost Components, and Source Data

Cost component    1929-1946     1947-1960     1961-1974     1975

Scientific R&D
services
(NAICS 5417)
Wages and
salaries         Wage and salary accruals     Average wages for
                 per FTE for all private      chemists and engineers
                 industries
All other
costs (1)        Wage and       Average PPI for            Average PPI
                 salary         intermediate materials,    for
                 accruals per   supplies, and              intermediate
                 FTE for all    components                 materials
                 private                                   and
                 industries                                supplies,
                                                           excluding
                                                           food and
                                                           energy
Materials and
supplies (1)                        Not applicable
Depreciation (1)                    Not applicable
Other costs (1)                     Not applicable
All other
private
industries
Wages and
salaries         Wage and salary accruals   Wage and        QCEW
                 per FTE for all private    salary          average
                 industries                 accruals        wages for
                                            per FTE         all private
                                            for             industries
                                            manufacturing
All other
costs (1)        Wage and       Average PPI for            Average PPI
                 salary         intermediate materials,    for
                 accruals per   supplies, and              intermediate
                 FTE for all    components                 materials
                 private                                   and
                 industries                                supplies,
                                                           excluding
                                                           food and
                                                           energy
Materials and
supplies (1)                        Not applicable
Depreciation (1)                    Not applicable
Other costs (1)                     Not applicable
Cost component    1976-1977     1978-1987     1988-1996     1997 forward
Scientific R&D
services
(NAICS 5417)
Wages and
salaries         ECI for        QCEW average  QCEW average  QCEW average
                 wages and      wages         for wages     wages for
                 salaries of    R&D           for           scientific
                 professional,  laboratories  commercial    R&D services
                 specialty,     (SIC 7391)    physical      (NAICS 5417)
                 and technical                research
                 occupations                  (SIC 8731)
All other
costs (1)        Average PPI for intermediate materials    Not
                 and supplies, excluding food and energy   applicable
Materials and
supplies (1)                        Not applicable         KLEMS price
                                                           for
                                                           materials
                                                           inputs by
                                                           industry
                                                           for
                                                           miscellaneous
                                                           professional,
                                                           scientific,
                                                           and technical
                                                           services
Depreciation (1)                    Not applicable         CFC price for
                                                           business
                                                           services
Other costs (1)                     Not applicable       Fisher-weighted
                                                           price for
                                                           energy and
                                                           services
                                                           inputs by
                                                           industry for
                                                           miscellaneous
                                                           professional,
                                                           scientific,
                                                           and technical
                                                           services.
                                                           Weights for
                                                           the Fisher
                                                           aggregation
                                                           from KLEMS.
All other
private
industries
Wages and
salaries                            QCEW average wages
                                    for all private
                                    industries
All other
costs (1)        Average PPI for intermediate materials    Not
                 and supplies excluding food and energy    applicable
Materials and
supplies (1)                        Not applicable         KLEMS price
                                                           for materials
                                                           inputs by
                                                           industry for
                                                           total private
Depreciation (1)                    Not applicable         CFC price for
                                                           business
                                                           services
Other costs (1)                     Not applicable       Fisher-weighted
                                                           price for
                                                           energy and
                                                           services
                                                           inputs by
                                                           industry for
                                                          total private;
                                                           weights for
                                                           Fisher
                                                           aggregation
                                                           from KLEMS.

CFC    Consumption of fixed capital
ECI    Employment cost index
FTE    Full-time employee
KLEMS  K-capital, L-labor, E-energy, M-materials, and S-purchased
       services
NAICS  North American Industry Classification System
PPI    Producer price index
QCEW   Quarterly census of employment and wages
R&D Research and development
SIC Standard Industrial Classification
(1.) Before 1997, all other costs was an aggregate measure. Beginning in
   1997, it was disaggregated into materials and supplies, depreciation,
   and other costs.

Table 13. Business R&D Price Index, Weights for Cost Components
                          [Percent]

Cost component           1929  1963  1975  1987  1997  2002  2007  2012

Scientific R&D services
Wages and salaries         58    58    60    52    58    54    58    61
All other costs (1)        43    42    41    48
Materials and
supplies (1)                                       13    12     11    9
Depreciation (1)                                    1     3      4    4
Other costs (1)                                    27    31     27   26
All other private
industries
Wages and salaries         58    58    60    52    59    56    59    62
All other costs (1)        43    42    41    48
Materials and
supplies (1)                                       13    12     11    9
Depreciation (1)                                    2     4      4    4
Other costs (1)                                    27    30     25   25

R&D Research and development
1. Before 1997, all other costs was an aggregate measure. Beginning in
1997, it was disaggregated into materials and supplies, depreciation,
and other costs.

NOTE. Estimates may not sum to totals because of rounding.

Table 14. Academic R&D Price Index, Cost Components, and Source Data

Cost component    1929-1946    1947-1959    1960-1974    1975-1977

Compensation
R&D university
faculty           Wage and salary accruals   AAUP average salary for
                  per FTE for educational    all faculty
                  services from the NIPAs
Research
associates                Wage and salary accruals per    QCEW
                          FTE for educational services    average wages
                          from the NIPAs                  for colleges,
                                                          universities,
                                                          and
                                                          professional
                                                          schools
                                                          (SIC 822)
Fringe benefits   Wage and salary accruals   AAUP average benefits for
                  per FTE for educational    all faculty
                  services from the NIPAs
Overhead
/purchased
services
/materials
and supplies      Gross domestic     PPI for SOP intermediate supplies,
                  purchases price    materials, and components
                  index
Depreciation
Private
equipment                      IPD (chain) for investment
                               in equipment by education
                               services industry
Private
structures                     IPD (chain) for investment
                               in equipment by education
                               services industry
Public
equipment                      IPD (chain) for investment
                               in equipment by education
                               services industry
Public
structures                     State and local government
                               price index for educational
                               structures
Cost component    1978-1989    1990-1996    1997-1998    1999 forward
Compensation
R&D university
faculty                        AAUP average salary for
                               all faculty
Research
associates      QCEW average   QCEW average wages for    OES data for
                wages for      colleges, universities,   life, physical,
                colleges,      and professional schools  and social
                universities,  (NAICS 6113)              scientists in
                and                                      colleges,
                professional                             universities,
                schools                                  and
                (SIC 8221)                               professional
                                                         schools
                                                         (NAICS 6113)
Fringe
benefits                       AAUP average benefits for
                               all faculty
Overhead
/purchased
services
/materials
and supplies    PPI for SOP intermediate    Fisher-weighted price for
                supplies, materials, and    energy, materials, and
                components                  services by industry from
                                            KLEMS for educational
                                            services (NAICS 61)
Depreciation
Private
equipment                      IPD (chain) for investment
                               in equipment by education
                               services industry
Private
structures                     IPD (chain) for investment
                               in structures by education
                               services industry
Public
equipment                      IPD (chain) for investment
                               in equipment by education
                               services industry
Public
structures                     State and local government
                               price index for educational
                               structures

AAUP   American Association of University Professors
FTE    Full-time employee
IPD    Implicit price deflator
KLEMS  K-capital, L-labor, E-energy, M-materials, and S-purchased
       services
NAICS  Nor th American Industry Classification System
NIPAs  National income and product accounts
OES    Occupation employment statistics
PPI    Producer price index
R&D    Research and development
SIC    Standard Industrial Classification
SOP    Stage of processing

Table 15. Academic R&D Price Index, Weights for Cost Components
                          [Percent]

                          1929  1962  1974  1986  1997  2002  2007  2012

Compensation                54    56    55    55    51    50    49    52
 R&D university faculty     23    24    23    23    22    21    21    22
 Research associates        20    21    20    20    19    19    18    19
 Fringe benefits            11    11    11    11    10    10    10    10
Overhead/purchased
services/materials
and supplies                36    37    36    38    42    44    45    42
Depreciation                10     7    10     8     7     6     7     6
 Private equipment           3     2     3     2     2     1     1     1
 Private structures          1     1     1     1     1     1     1     1
 Public equipment            4     3     4     4     3     3     3     3
 Public structures           2     1     2     1     1     1     2     2

R&D Research and development
NOTE. Estimates may not sum to totals because of rounding.


Table 16. Indicators for Current Quarterly Estimates of R&D Investment
and Prices

Component          Advance estimate          Second and third estimates

Business R&D       Aggregate wages from      R&D expenses as repor ted
investment,        BLS CES for all private   on publicly-traded
current dollars    industries.               companies' quarterly
                                             financial statements.

NPISH R&D                Aggregate wages from BLS CES for
investment,               education and health services.
current dollars

State and local          Aggregate wages from BLS CES for
R&D investment,           education and health services.
current dollars

Federal R&D              The Budget of the United States
investment,
current dollars

Business, NPISH,         Composite average wages from BLS
and state and           CES using 60 percent scientific R&D
local government        services and 40 percent all private.
performer price
index

Academic                 Average wage from BLS CES for education
performerprice                   and health services.
index

Federal                  Various BEA price indexes
own-account
price index

BLS   Bureau of Labor Statistics
CES   Current Employment Statistics
NAICS North American Industry Classification System
NPISH Nonprofit institutions serving households
R&D   Research and development