5
Intangibles and Government Measurement

An overarching theme to the workshop was that better information about intangibles is needed for firms to manage these assets efficiently and for policy makers to value them appropriately. It is also clear that the conceptual approach used in the accounting of intangible assets is likely to have a significant impact on measures of economic performance and growth. Many of the workshop’s participants suggested that improvements in the accounting of and information about intangible assets could be effectively initiated and orchestrated only by government.

With this as the backdrop, a session was organized to sort through the priorities of the statistical agencies for collecting higher quality data on private investments in intangibles, as well as the size and composition of public investments, and incorporating them into broader measures of economic performance. Beyond the measurement issues, narrowly defined, presenters asked: What should the government do to encourage company creation of intangibles? What should be the government’s role in creating or supporting more robust markets in intangibles? And, what are other governments doing in these respects?

Jonathan Haskel, who moderated the session, noted that inherent in this group of presentations were a lot of issues—about the redesign of research and development (R&D) and innovation surveys and better measurement of intangibles—that extend to European and other international contexts. The two speakers for the first government session were Steven Landefeld and John Jankowski, who spoke about developments at the Bureau of Economic Analysis (BEA) and the National Science Foundation (NSF), respectively. Earlier in the day, Brent Moulton provided details about BEA’s R&D satellite account.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 60
5 Intangibles and Government Measurement A n overarching theme to the workshop was that better information about intangibles is needed for firms to manage these assets efficiently and for policy makers to value them appropriately. It is also clear that the conceptual approach used in the accounting of intangible assets is likely to have a significant impact on measures of economic performance and growth. Many of the workshop’s participants suggested that improvements in the accounting of and information about intangible assets could be effectively initiated and orchestrated only by government. With this as the backdrop, a session was organized to sort through the priori - ties of the statistical agencies for collecting higher quality data on private invest - ments in intangibles, as well as the size and composition of public investments, and incorporating them into broader measures of economic performance. Beyond the measurement issues, narrowly defined, presenters asked: What should the government do to encourage company creation of intangibles? What should be the government’s role in creating or supporting more robust markets in intan- gibles? And, what are other governments doing in these respects? Jonathan Haskel, who moderated the session, noted that inherent in this group of presentations were a lot of issues—about the redesign of research and development (R&D) and innovation surveys and better measurement of intan- gibles—that extend to European and other international contexts. The two speak- ers for the first government session were Steven Landefeld and John Jankowski, who spoke about developments at the Bureau of Economic Analysis (BEA) and the National Science Foundation (NSF), respectively. Earlier in the day, Brent Moulton provided details about BEA’s R&D satellite account. 0

OCR for page 60
 INTANGIBLES AND GOVERNMENT MEASUREMENT 5.1. THE ROLE OF GOVERNMENT STATISTICS Presenters throughout the day characterized the role and value of intangible assets, both to firms and industries and in terms of their macroeconomic implica - tions. Similarly, the way that intangible assets are measured, or not, potentially has a significant impact on the statistics calculated and used for monitoring the performance of the economy. Rigorous measurement of intangibles can lead to improved accuracy of estimates of gross domestic product (GDP) and the ability to identify sources of economic growth. Brent Moulton, in explaining the rationale for classifying expenditures on intangibles as capital formation, noted that economic theory strongly suggests that investment in such assets is similar to tangible investment in terms of its effect on reducing current consumption and increasing future output. Consistency in national accounting requires that assets with these shared qualities be treated analogously. As noted above, certain types of intangibles, such as computer soft - ware, are already capitalized in the national accounts. Although rents and royal- ties and the services output associated with knowledge assets are included in the economic accounts, the production of all knowledge assets is not. The major thrust of Steven Landefeld’s presentation dealt with the measure- ment challenges facing BEA and the priorities for confronting them to improve data on intangibles. In order to provide a sense of the magnitude of the issue, Landefeld first summarized preliminary results that have emerged from the NSF/BEA R&D satellite account (discussed in greater detail in the next section). Of particular note is the observation that the impact on the broader economy of R&D appears to be expanding: Between 1959 and 2004, R&D accounted for 5 percent of growth in real GDP, whereas, for the period 1995-2004, its contribu- tion rose to 7 percent. If spillovers (the residual unexplained portion of growth) from R&D are, as research suggests, at least as large as the direct returns, this means that R&D may account for one-sixth of total factor productivity growth.1 Landefeld cautioned that, while substantial progress in measurement has been made and preliminary results indicate an influential role for R&D in the economy, a number of questions are unanswered about these statistics and much remains unmeasured. Next, Landefeld reviewed some of the data (see Table 5-1) from the research by Corrado, Hulten, and Sichel (CHS) that has specific implications for BEA’s immediate programs. A portion of the capital spending itemized in the table is already captured in the national accounts (e.g., the above-mentioned software), and methods for handling additional components are being advanced in the satel - lite R&D accounts. Landefeld stated that the pieces that BEA is particularly inter- ested in pushing further are nonscientific R&D (2b) and firm-specific investment (parts of 3b), in particular, human capital. For practical reasons, BEA does not plan to put significant effort into measurement of the brand equity arising from 1This figure is based on estimates from Jorgenson et al. (2005:38-39).

OCR for page 60
 INTANGIBLE ASSETS TABLE 5-1 A Broader Measure of Business Intangibles, 1998-2000 (billions of dollars, annual average) Capital Spending Total Comments on Evidence as (included in Type Spending Capital Spending NIPAs) (1) Computerized 154 Firms capitalize only a fraction 154 information of purchased software in (151) financial accounts. Relatively little is known about the service life of software assets. (2) Innovative property (a) Scientific R&D 201 Research suggests that scientific 201 R&D yields relatively long (16) lasting returns and is capital spending. (b) Nonscientific 233 Little is known about 233 R&D nonscientific research R&D, (40) but a portion of new product development expenditures in the entertainment industry apparently have relatively short-lived effects. (3) Economic competencies (a) Brand equity 235 Research shows that the effects 140 of some advertising dissipate (0) within one year, but that more than half has effects that last more than one year. (b) Firm-specific 407 Research suggests that firm- 365 resources specific training is investment. (0) Spending for organizational change also likely has long- lived effects, but a portion of management fees probably is not capital spending. Total 1,220 1,085 (205) Percent of Existing GDP 11.7 Ratio to Tangible Capital Spending 1.2 SOURCE: Corrado, Hulten, and Sichel (2006b).

OCR for page 60
 INTANGIBLES AND GOVERNMENT MEASUREMENT advertising (3a). BEA will leave it to researchers to figure out how this should be measured, and perhaps, at some point down the road, it can be addressed in the statistical agency context. It would be controversial for BEA, as a statistical agency, to simply add an extra 2 to 3 percent to GDP based on these findings. That is a significant magni- tude, and Landefeld said that the source data and methodologies for making such an adjustment will have to be examined under a microscope before BEA could even think about adding them to official statistics. One of the biggest jobs for BEA analysts and leadership is to arrive at a consensus methodology on which the experts and users (including policy makers such as the Federal Reserve) agree. International Dimensions to the Measurement of Intangibles Intangibles sold and transferred by multinational corporations to their over- seas units are particularly difficult to track. Yet, as Landefeld noted, these assets are a key part of the scenario—R&D by multinationals accounts for over 80 percent of U.S. R&D—so it is essential that BEA work toward solutions to the measurement challenge. Multinationals are, by nature, oriented toward maximiz - ing their global profits net of taxes, and their stock price will reflect those profits whether they are generated in the United States or elsewhere. Because companies are not typically required to repatriate profits, they may attempt to minimize their operation’s taxable income and assets in high-tax countries; one way to accom - plish this is to transfer intangible assets to operations in lower tax locales, such as Ireland or Switzerland. The incentive is to transfer the right to collect revenues from assets yielding royalties overseas. According to press reports, multinationals such as Microsoft have been able to dramatically lower their effective corporate tax rate worldwide in this way. As Table 5-2 indicates, these transfers of intangible assets have been grow- ing in importance over time. The percentage of receipts recorded by U.S. parent multinational corporations in the “tax haven” countries has risen from 13 to 38 percent during the 1977-2005 period. Similarly, U.S. direct investment in the same set of countries has gone from 19 percent in 1977 to 34 percent of the total in 2005. When one examines the rates of return to U.S. companies’ investments overseas versus what foreign companies earn on their direct investments domes - tically, there is a considerable divergence, with the former earning much more overseas than what the latter earn in the United States. Landefeld offered the view that the rates of return on investments, risk adjusted, made by U.S. companies do not look too far out of line relative to what U.S. companies earn domestically. The more notable effect is what is happening to foreign earnings in the United States, which seem very low. Many researchers have suggested that a significant transfer pricing issue is probably at work. This example raises the issue of how these assets should be treated given that they can often be used both overseas and in the United States where, in many

OCR for page 60
 INTANGIBLE ASSETS TABLE 5-2 U.S. Parent Companies’ Receipts for Royalties and License Fees from Foreign Affiliates in Lower Tax Countries (millions of dollars) 1977 1982 1989 2005 Belgium 104 149 326 580 Ireland 10 39 255 4,285 Luxembourg 2 1 5 91 Netherlands 107 166 633 1,589 Switzerland 45 83 255 4,160 Bermuda 2 10 4 (D) UK Islands, 0 0 0 (D) Caribbean Hong Kong 3 14 94 393 Singapore 10 24 151 2,278 Tax haven subtotal 283 486 1,723 13,995 Worldwide total 2,173 3,585 10,082 37,771 Tax haven share 13.0 13.6 17.1 37.1 (percent) NOTE: The table values are net of withholding taxes. (D) = suppressed to avoid disclosure of data of individual companies. SOURCE: The list of low-tax-haven destinations for foreign direct investment is from Sullivan (2004); receipts data are from BEA. cases, they were originally produced. This is as yet an unresolved measurement problem for BEA. The agency is working on getting the research and the empiri- cal results right, but a reasonable consensus also needs to be reached among users with respect to the BEA methodology and to what needs to be measured. Given the complexity of some of these tasks, Landefeld cautioned against integrating results into formal statistics too quickly. People have suggested to BEA that, since it has produced R&D estimates, why not simply integrate them into the accounts and move on to the next set of intangible assets. There is still much work to be done—not just on the international aspect, but also on regional ones. When dealing with the idea of adding figures of this magnitude to GDP, BEA must be extremely careful that everything is correct and the methodology is correct, defensible to users, and can be consistently reproduced. BEA Plans Next steps for advancing BEA’s agenda, both in the area of intangibles and elsewhere, will involve continued source data development and conceptual work. BEA will continue to move toward integrating the new version of the Interna- tional Accounting Standards which, in accordance with the 2008 Revision of the System of National Accounts (SNA), the United States and all other countries are recommended to follow. The SNA revision also recommends that R&D be capi - talized, an important first step if international comparability of macro data and

OCR for page 60
 INTANGIBLES AND GOVERNMENT MEASUREMENT common accounting standards are to be achieved. Also in adherence with SNA recommendations, work will continue on integrating the productivity accounts produced by the Bureau of Labor Statistics (BLS) to create a crosswalk across data sets to identify sources of economic growth. Noting that this has always been something of a cottage industry, with outsiders doing much of the work, Landefeld expressed the view that it was time that the federal government begin to develop fully integrated sets of accounts. Another major set of items on BEA’s agenda is to begin implementing recommendations from the Report on the Advisory Committee on Measuring Innovation in the st Century, which was summarized by Cynthia Glassman at the workshop (see section 5.4.). That committee offered practical guidance on expanding data sharing, or data synchronization, between the statistical agencies (mainly BLS, the Census Bureau, and BEA).2 In order to promote further understanding of technological change and inno - vation, discrepancies that exist within the U.S. statistical system need to be reconciled. For example, the growth rates in productivity for computers over time appear very different depending on whether the analyst is looking at Cen - sus Bureau data or BLS data. This is true for many of the most rapidly growing industries, many in the service sectors, which are major users of technology. BEA, which must integrate data from both BLS and the Census Bureau, is, at times, essentially dividing apples into oranges. The output series is produced by the Census Bureau, and the input series is produced by BLS. Some of the discrepancies have to do with classification issues because aggregate output data often look virtually the same. In a related initiative, also tied to recommendations by the innovation com - mittee, BEA hopes to soon finish developing an integrated macro model of GDP and productivity; this will involve the joint publication by BEA and BLS of real GDP and multifactor productivity data. There have been ongoing efforts over the years with BLS to reduce the differences between various data series. Some of the problems are attributable to methodological differences across programs, which Landefeld is confident can be eliminated. Landefeld expressed optimism that the relevant data sets could be synchro- nized and improved in other ways, since the key discrepancies exist for only about a half dozen industries. Fixler and Landefeld identified these in a paper prepared for an earlier workshop on data sharing (see National Research Council, 2006). Taking a hard look at the classifications systems reveals much of the problem. For example, because of their global approach to manufacturing, major U.S. computer manufacturers are sometimes classified as wholesalers rather than as manufacturers. This inconsistent classification of the activities of the computer 2 See also National Research Council (2006), a summary report from a workshop held by the Com - mittee on National Statistics on data sharing, for a detailed discussion of the goals of the statistical agencies on the topic.

OCR for page 60
 INTANGIBLE ASSETS industry creates problems for estimating this very important contributor to eco - nomic growth and productivity. Another top priority for BEA, identified by the innovation committee, is to continue improving definitions and measures of output in the service sectors. Landefeld congratulated BLS on the superb job that agency has done to expand its coverage of the service industries. Only a decade ago, the producer price index covered only about 5 percent of services; it now covers them up in the 70 percent range. This is a huge improvement, but there is still a significant amount of work to be done. It will also be important to continue to improve annual measures of overall services activity in the United States. Currently, such an aggregate is produced only once every five years in conjunction with the quinquennial cen - sus. The Census Bureau is making good progress through their expansion of the quarterly and annual surveys of services and, with a little funding, the agencies may be able to take the final step to provide full coverage of the service sector in an ongoing fashion. Another recommendation by the innovation committee is to link data from the existing NSF R&D survey to BEA and Census data on R&D conducted by U.S. and foreign multinational firms. The progress attained thus far has been made possible by NSF funding; Landefeld reported that there is a memorandum of understanding for an expanded project to combine data sources on R&D that will take advantage of each agency’s relative specialization (BEA for large firms and data on R&D, and the Census Bureau on overall coverage of small- and medium-sized firms) to create a more complete and consistent set of estimates. Landefeld also reflected on various ideas for expanding the measurement of intangibles, identifying candidates for an innovation account. For spending on scientific and engineering R&D, these involve further study of product and pro- cess innovation, developing more timely data and more frequent indicators, and collecting data on royalty and licensing fee receipts (as well as expenditures) and on associated capital investment expenditures. Landefeld also expressed the hope that someday better data and methodology would be available for the valuation of intellectual property. He agreed with Baruch Lev that developing an NSF survey geared toward such a measure would be a very interesting project. If standards can be developed that can be shared by industry and governments, it would go a long way toward getting the process started. Continuing the discussion of potential expansions in the measures of intan - gibles, Landefeld identified expenditures on social science R&D related to new products and processes. For example, explicit subcategories could be created for product and process design and development for both industrial and artistic and entertainment areas. Additional areas for expansion, identified in earlier presenta- tions as well, include employer spending on employee training and development, computer software, and investment in business models, such as inventory and distribution control systems.

OCR for page 60
 INTANGIBLES AND GOVERNMENT MEASUREMENT During the discussion of business processes, the example of Walmart was raised, with Landefeld noting that he was not sure of the extent to which things like inventory control systems were being captured in the current data. Later dur- ing the workshop, a participant pointed out that there had been much discussion about human capital and knowledge capital, but less about organizational capital, yet the latter appears to be one of the more influential types of intangible assets. This involves not only the business processes of Walmart, but also the organiza- tion of work, talent, and training. Organizational capital theory suggests that much more is at work than simply information technology (IT)—such as how work and workforces are structured—in determining which industries and which companies succeed. Ideally, these factors would be measured at either the firm level or nationally or both. Kossovsky commented that some intangible process and systems factors are captured in net income measures because a company’s cost of delivering its services is ultimately reflected therein. If a company is orga- nized efficiently, if the workers are well trained and are working well together, if the IT systems link them efficiently together and the employees are pleased with their environment, and if there is less turnover, the cost of delivering the product will be lower and so the net income will be higher. Landefeld concluded by reiterating the importance of collaboration between businesses and the national accountants. If the data being collected are not those that businesses have in their records, the project is, in his view, largely doomed to failure. BEA can make indirect estimates and undertake modeling to cover some of the transactions that are internal to the firm, but ultimately the goal has to be to take advantage of coincidental interests and reliance on business records for accu- racy and consistency and minimization of respondent burden. Both government and businesses have an interest in consistent valuation of intangibles in terms of understanding growth for the firm as well as for the U.S. economy. Landefeld expressed optimism that all of the talking would ultimately lead to valuable data products, but he cautioned that it will be a long road. 5.2. THE U.S. RESEARCH AND DEVELOPMENT SATELLITE ACCOUNT Brent Moulton presented an overview and results from BEA’s work on an R&D satellite account, an effort that has been under way for about four years, with sponsorship from NSF. He began by identifying several conceptual chal- lenges to improving measurement of R&D—challenges that are substantial but probably still simpler than those that will arise for some other categories of intangibles. Among these challenges are defining the unit of economic output, estimating R&D output price indexes, measuring depreciation and obsolescence, and dealing with the issue of the public good aspects of some R&D.

OCR for page 60
 INTANGIBLE ASSETS Price Indexes Most R&D is done within a firm for its own use, which means that observ- able market prices for R&D output are not frequently available. This is, of course, problematic for price index construction. Even for cases in which R&D specialist firms sell custom products to other firms, a problem still exists because rarely is it possible to monitor the same unit across periods. Unlike a gallon of milk, which remains unchanged, R&D projects typically are unique, and so a standardized unit of measurement is not available. Even when prices can be observed, it is still difficult to construct a price index using traditional methods. Moulton reported that BEA has considered several approaches to price indexing associated with intangibles. One option that is often used for nonmarket output is to simply look at the prices of inputs. This has the well-known problem that, when the value of the output is equated with the value of the input used to produce it, productivity change is defined away. If one of the main purposes of the R&D account is to generate insights into productivity, this is not a very use - ful approach. BEA has pursued other approaches. In some cases, the price of the output is largely driven by the innovation that it embodies. For example, the value of a semiconductor chip originates from the engineering that has gone into it and very little from the physical inputs. This has led BEA to examine output price indexes for the goods that are being produced through the innovative activity. Another approach is to try to establish a measure of productivity for the relevant industry and then to augment the price of inputs based on a broad-based productivity fac - tor. A fair amount of work has been done on the depreciation issue—by BEA and others—and, in Moulton’s view, the results, while not perfect, have produced a good range of estimates. During his presentation, Landefeld provided an indication of how price indexing results from the input- and output-based methodologies diverge, using the R&D example. Figure 5-1 shows two price indexes. The dashed line reflects input costs, which is probably not the right one to use; as noted earlier, it pre - sumes zero productivity growth. The solid line shows an aggregate output price index constructed for 13 R&D-intensive industries and reflects falling prices of key inputs to R&D, most notably computers. This is what was used to deflate the figures in the 2007 satellite account update. BEA’s theoretical research suggests that this was appropriate, although more empirical work is needed to examine the relationship between R&D’s input costs and output prices. Landefeld cited the example of the semiconductor industry (which has been studied by Ana Aizcorbe at BEA) to illustrate the complexities that the agency confronts in this research. The declining costs per unit of speed and, in turn, the declining price index for semiconductors are well understood. Complications arise when looking at the output price for semiconductors themselves because it is difficult to place the timing of when the R&D that led to improvements occurred. Relevant activities include not only those related to research to develop semi -

OCR for page 60
 INTANGIBLES AND GOVERNMENT MEASUREMENT 140.0 120.0 100.0 Index, 2000 = 100 80.0 60.0 Aggregate output price index 40.0 Input price index 20.0 0.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 FIGURE 5-1 R&D price indexes. SOURCE: Workshop presentation by Brent Moulton. Reprinted with permission. Fig5-1.eps conductors, but also the R&D of semiconductor manufacturing, which involves another set of people working to improve the equipment for that stage of the process. Over time, there has been a miniaturization of the equipment that makes the semiconductors, and development of faster and cheaper semiconductors is very much a product of R&D in both areas. Several feasible assumptions could be made about how these processes interact. Perhaps there is little market power on the part of, say, Intel, and, as a result, the market price that is being charged by the manufacturers of semiconductor equipment is fully captured. This is uncertain, however, and Landefeld articulated the need for more research in order to avoid potentially large misallocations of R&D across industries. The public goods issue is particularly a problem for BEA as it looks at mul - tinational corporations and also domestically for the regional accounts—GDP by state and so forth. As discussed above, a large multinational company that con - ducts R&D in the United States may transfer ownership of that R&D asset or that knowledge asset to a low-tax country. Microsoft, for example, has a large affiliate in Ireland, a low-tax country, that collects royalties from most of Europe. In this case, should the ownership be recorded where it was produced or in the country where it is legally owned? Or, since these are large multinational companies with subsidiaries throughout the world, should ownership be allocated among all of the countries where the asset is used? Similar issues arise in allocating R&D output

OCR for page 60
0 INTANGIBLE ASSETS in industries that have establishments in many states. Each option—other than the legal ownership—is likely to involve some imputations. Another option might be to allocate all of the ownership to one country and then to impute rentals. Some countries do charge their subsidiaries for R&D, but BEA researchers have found that, in most cases, those charges are not for use of existing R&D knowledge assets, but rather for ongoing R&D work—an overhead expense charge—which is less relevant. So internal company charges, which are not uniformly applied by companies at any rate, still probably are not conceptually what would be most appropriate to use when the goal is to measure a royalty for use of R&D within a company. These are a few of the issues that need to be resolved for this public good aspect of R&D assets, even within a single company. Progress to Date on the R&D Account Moulton provided some details of what progress BEA has made in the devel- opment of its satellite account. The starting point in the evolution of this work has been the NSF surveys, a valuable resource that provides over 50 years of consis - tent industry-level time series data on R&D expenditures. The focus of the data, which include such items as costs for employees, materials, and depreciation, is on physical and life sciences and engineering—scientific R&D. Construction of the R&D accounts entailed the following steps: current dollar investment is estimated as the sum of input costs, which is the best dollar value available for now. Current dollar investment figures are then deflated (using an output price index constructed for 13 R&D intensive industries, described above) to produce measures of real investment. Using a perpetual inventory model, capital stock measures are created, calculated as cumulative R&D invest - ment less R&D depreciation.3 For government and nonprofit institutions, R&D includes a net return measure. On that basis, the experimental measure of GDP gets recalculated. Moulton next presented summary results from the 2007 R&D account. As described above, scientific R&D has historically accounted for around 5 percent of growth in real GDP, although that figure has increased during the 10-year period of coverage beginning in 1995. In comparison, business gross fixed capital formation in commercial and all other types of buildings accounted for just over 2 percent of real GDP growth. The magnitude of the R&D numbers, as a portion of total real GDP growth, is comparable to those associated with major categories of tangible investment. In the satellite account, the contribution of R&D to GDP growth is almost as large as the measured contribution of computers. Figure 5-2 shows (in billions of current dollars) the addition to GDP attributable to changing 3 Depreciation rates are based on average estimates from the literature ranging from 11 percent for chemical manufacturing to 18 percent for transportation equipment and manufacturing.

OCR for page 60
 INTANGIBLE ASSETS for the collection, interpretation, and analysis of data on scientific and engineer- ing resources, and to provide a source of information for policy formulation by other agencies of the Federal Government.” With such a broad mandate, SRS is involved in a range of activities to enhance the comparability, scope, and availability of R&D and related data. Among these are • redesigning the industry R&D survey; • redesigning the academic R&D survey; • improving two surveys of federal government funding of R&D (a panel of the Committee on National Statistics was convened to explore issues and new approaches for survey improvement); • continuing a new state government agency R&D survey, for which data collections were conducted for 2006 and 2007 and will be conducted periodically in the future; • expanding the research facilities survey of academic and biomedical facilities to include, among other things, expanded data collection on cyberinfrastructure; • developing a nonprofit R&D survey, which is in the very early planning stage; • exploring innovation data collection possibilities from very small firms (1-4 employees); • continuing work on the R&D satellite account—the joint work with BEA discussed by Landefeld; • linking NSF’s business R&D data with BEA data on foreign direct invest- ment; this will include U.S. firms’ international R&D activities and for- eign firms’ R&D activity in the United States by state and industry; and • planning to add R&D and innovation-related questions to other surveys, such as the Kauffman Firm Survey (third follow-up) and the Census Company Organization Survey. Jankowski focused his comments on initiatives that the agency has taken in terms of collecting information on R&D, particularly the redesign of its industry and academic R&D surveys. The rationale for why the surveys need to be rede- signed can be established by observing the changes that have taken place in the business R&D context over time. During the 1950s, government was the largest source of R&D expenditures, and it was domestically focused; business is now the largest spender, and the context is global. A half-century ago, business was the largest performer of basic research; now the largest performer is academia. And, as emphasized throughout the day, the past 50 years have seen the transforma - tion from a manufacturing economy populated with large companies dominating R&D to an economy driven by a service sector in which large companies are not as dominant in the R&D picture.

OCR for page 60
 INTANGIBLES AND GOVERNMENT MEASUREMENT Although perhaps not quite as pronounced as these trends, a number of changes have taken place among the academic enterprises conducting R&D. The federal government has always provided the bulk of funding to universities for R&D, although the relative share has declined, and the amount of cost sharing that the universities provide has grown substantially. Collaboration has grown with universities along with the extent of multidisciplinary and interdisciplinary research. In addition, interest has grown in the commercialization of academic R&D,4 at times controversially, as when pharmaceutical companies farm out drug development to academic researchers who also have a financial stake in the outcome. The Survey Redesign Process Jankowski stepped through what NSF does when it embarks on a survey redesign. Perhaps most importantly, the agency seeks extensive input from the data user community that establishes a prioritization of activities. Expert panels— which typically include such people as vice presidents or heads of research for major universities or corporations—are convened and data user workshops are held. The process may involve surveys of record-keeping activities, cognitive interviews, and a number of other activities to understand what the available data mean and what gaps need to be filled. SRS also collaborates with other govern - ment agencies, such as BEA, to identify what can be done to provide data that will be helpful in the execution of their missions. Once data needs have been identified, SRS then focuses intense efforts on identifying data sources and estab- lishing data availability. If data are not available, work then begins to establish what might be reasonable proxies for some of the variables of interest. The first survey described by Jankowski is the industry R&D survey, an annual collection of industrial (manufacturing and services) R&D expenditure data that has been conducted since 1953. Data collection and tabulation have been carried out by the Census Bureau since 1957. The survey, conducted with a pledge of confidentiality, includes all for-profit R&D-performing companies with five or more employees, which are surveyed at the company level (as opposed to the establishment level). The annual sample includes approximately 32,000 firms. The overall unit response rate (in 2006) was 77.5 percent; the top 500 R&D performers responded at an 89.2 percent rate. Two different forms are used—a standard length typically used for the known R&D performers and a short ver- sion for smaller companies with unknown R&D status. The survey includes five 4 Legislative actions have encouraged some of this. The Bayh-Dole Act of 1980, for example, enabled universities and small businesses to patent discoveries created by research sponsored by government funding (mainly the National Institutes of Health) and then to grant exclusive licenses to drug companies. Prior to this legislation, although individual agencies still had a variety of agreements with universities, taxpayer-financed discoveries were in the public domain and therefore available to anyone who wanted to use them.

OCR for page 60
 INTANGIBLE ASSETS mandatory response fields: total and federal R&D, sales, employment, and R&D by state. All items are mandatory in economic census years. Data are collected through extensive company contacts who provide informa- tion about what data companies have and how they get the data. Record-keeping and environmental scanning interviews are conducted to find out what informa - tion companies track in their records. NSF has conducted 5 rounds of cognitive interviews with more than 100 individual companies. Interviews are done with accountants for financial sections of the questionnaire, human resources repre - sentatives for the employment section, R&D managers for technical aspects of the activities, and legal experts for intellectual property and technology transfer information. Jankowski cited several lessons that have been learned from interacting with companies during the R&D surveys. Perhaps most obviously, not all of the data that policy makers and researchers may want are knowable. And, for what does exist, different types of data are stored in different parts of the company, and no single person typically has direct access to all of the data. This means that getting the survey to the correct, most informed respondents in the company for specific topics is crucial to the task of obtaining the right data. In addition, questionnaire development requires contact with a variety of companies and the input from a variety of subject matter experts. The structure of the business R&D survey reflects this learning process. Because the survey content covers a range of topical areas and requires data from multiple parts of the company, NSF/SRS has structured it into the following separate sections: • financial measures of R&D activities (R&D expense in accounting terms), • financial measures of R&D funded by others (not classified as R&D in accounting definitions but is R&D performance), • nature and technical aspects of R&D, • R&D employment data, and • intellectual property and technology transfer. Content of the New Business R&D Survey Jankowski stepped through a description of the content of the new version of the business R&D survey, which will be sent out to about 40,000 businesses. A new set of initial check-off questions will be included that are geared toward gaining a sense of the role of innovative activities at the firm. These will ask about any introduction of new or significantly improved products (good and services) or of new or significantly improved processes over the past couple of years and about any patent and intellectual property licensing activity. Firms not engaged in R&D or funding R&D will be able to set aside the rest of the questionnaire;

OCR for page 60
 INTANGIBLES AND GOVERNMENT MEASUREMENT this in itself will generate some useful information about the patterns and extent of innovative activities. The heart of the questionnaire will collect information on the financial measures of the R&D activities. Many of the components of the upcoming survey, which are listed in Box 5-1, are new or expanded. One question that arose several times during the workshop is the extent to which the statistical system and the R&D survey specifically would pick up expenditures on design innovation, such as Apple’s investment in the iPod. Jankowski reported that NSF’s expert panel was asked about this. Although there are some exceptions—perhaps financial services—the consensus was that design expenditures largely should be in the R&D totals reported on the survey. That said, he emphasized that the question needs detailed investigation and that, for a specific example such as the iPod, there would need to be much more granular- ity about what exactly is being included in various categories. He also noted that some firms may report R&D and design activities separately in their filings to the Federal Communications Commission, but that is not necessarily the way it will be reported on their R&D survey. In some cases, there seems to be considerable expansion of what is being counted as R&D in the annual reports beyond the more narrow definitions of R&D as defined by the Financial Accounting Standards Board (FASB) and definitions used in the NSF/Census Bureau survey. Participants commented about categories of innovation that could be can- didates for further investigation. The point was made that most R&D is in fact design and development. If one were to ask engineers and the people in corpora - tions who manage them what they do, they would be likely to respond that they conceptualize and design first at very general levels and then at increasingly detailed levels. The real question about something like the iPod is where does design become product styling, as it used to be called. Whatever the industrial design or the styling is, it will typically only account for a fraction of the expen - diture involved in bringing a product to market. One participant pointed out that, for the iPod example, another element of innovation is the business model itself. The entire retail distribution industry of music—and now, more broadly, video and other types of entertainment—has changed. In some cases, it is not just a product innovation that may be important to track and measure, but also a much broader series of innovations that change the flow of economic revenues through the course of a number of different industries. Jankowski and Landefeld agreed that, because these are significant compo - nents of innovation, it is important that NSF continue to enlist expert panels from industry to move forward with the collection of data. In fielding these questions, Jankowski reminded participants that R&D funding is not the same thing as inno- vation funding and that business model innovation is, by design, not part of the current survey. This is not to say, he continued, that it would not be valuable at some point, but it is not part of the current effort. Similarly, the SRS team has not talked to companies about looking at marketing or advertising. In fact, in order to be consistent with what FASB includes, Jankowski reported that they would be

OCR for page 60
 INTANGIBLE ASSETS BOX 5-1 Business R&D Survey Content Financial Measures of R&D Activity • Detail on domestic U.S. R&D and on worldwide R&D activity (NEW) • Company R&D expense • Includes social science R&D (NEW) • “Business segment” (i.e., below the company level) (NEW) • U.S. state location and country location (NEW) • Type of expense (wages, materials, etc.) (EXPANDED) • Outsourced R&D by sector (universities, other companies, etc.) • Detail on domestic U.S. and worldwide sales and revenue (NEW) • Capital expenditures for R&D (buildings, software, equipment) (NEW) • Projected R&D expense Measures of Company R&D Activity Funded by Others • Funds for global R&D activity as well as domestic U.S. activity (NEW) • R&D funded by others • “Business segment” (i.e., below the company level) (NEW) • U.S. state location (NEW) • Type of expense (wages, materials, etc.) (EXPANDED) • Associated with single largest R&D project (NEW) • &D performed for others under grants, contracts, or other agreements R (NEW) • ype of organization (other companies, federal government, state and local T governments, others) • Foreign vs. domestic organization • Cinical trials and the production and testing of prototypes careful to encourage companies to ensure that they specifically exclude money spent on market research. Business R&D Survey Timeline At the time of the workshop, NSF/SRS had fully developed the survey ques- tionnaire using the interactive process described above and submitted the package to the Office of Management and Budget (OMB) for approval. Final cognitive testing took place in summer 2008. The plan was to launch the full-scale pilot of the redesigned survey in January 2009 and to send it to 40,000 companies, to collect data for 2008. Following the guidance of their industry panel and that of OMB, respondents receive a guarantee of confidentiality, and the survey is man - datory. During calendar year 2009, NSF/SRS will evaluate survey operations and analyze the pilot data that are returned so that, by January 2010, the questionnaire

OCR for page 60
 INTANGIBLES AND GOVERNMENT MEASUREMENT Measures Related to R&D Management and Strategy • Share of R&D • Devoted to new business areas for the company (NEW) • Involving science or technology new to the company (NEW) • Science or technology that is new to the market (NEW) • Spent on research versus development • evoted to specific application areas (health, defense, energy, etc.) D (NEW) • Devoted to specific technology areas (EXPANDED) • Counts of R&D projects (NEW) • Number active and number started • Number moved from R&D into production or marketing • R&D partnerships (EXPANDED) • Sector (universities, companies, government) • Type of organization (customer, vendor, competitor) Measures Related to R&D Employment • U.S. R&D headcount and worldwide R&D headcount (NEW) • Occupation (scientists, engineers, technicians, support) (NEW) • Gender and level of education (NEW) • U.S. R&D employees working under a visa (H-1B, L-1, etc.) (NEW) • R&D full-time-equivalent counts Measures Related to Intellectual Property and Technology Transfer (NEW) • Patent data (counts, external sources, foreign filings) (NEW) • Licensing to outside parties (NEW) • Importance of types of intellectual property protection (NEW) • Participation in specific technology transfer activities (NEW) • Importance of types of intellectual property protection (NEW) can be revised in advance of the official survey to collect 2009 data. By December 2010, the Census Bureau plans to deliver the 2009 survey data to NSF, in time for production of the Science and Engineering Indicators: 0. Jankowski reported that future plans call for possibly adding new/rotating modules to the survey—for example, for other innovation categories and for key industries and industry segments, such as financial services and pharmaceuticals. In addition, they may develop a pilot survey of firms with 1-4 employees to iden - tify innovative activities. And the survey content, methodology, and processing will be continuously reviewed and updated. NSF Academic R&D Survey Jankowski provided a brief overview of the agency’s academic survey. The survey is a census of all universities and colleges in the United States that con-

OCR for page 60
0 INTANGIBLE ASSETS duct at least $150,000 of R&D annually; since this is a small threshold level, it includes essentially all R&D-performing universities (about 680 institutions). The survey has been conducted annually since 1972; the web-based version is now used by 99 percent of respondents. It is a voluntary survey, but response rates are regularly in the 95-98 percent range. The survey requests information on expenditures for all separately budgeted R&D performed at the institution during the previous fiscal year. Data are published at the micro (institution) level, which allows for peer comparisons (something that cannot now be done with the industry survey). R&D expenditures are recorded in the following categories: • source of funds (federal, state/local, industry, institution, other); • character of work (what percentage is basic research?); • field of science and engineering (S&E); • federal agency sponsor and S&E field; • amount expended on research equipment, by S&E field; • amount passed through to subrecipients or received as a subrecipient; and • non-S&E field. Recently, NSF brought together data users for a survey redesign workshop. Expert panels were convened to discuss what should be collected and what is possible to collect. The top data needs identified were to expand coverage to academic technology transfer activities, academic and industry collaboration, and interdisciplinary or multidisciplinary research. The panels recommended the following: • including non-S&E R&D in the totals for institution rankings, • collecting data separately for medical schools, • capturing all sources of funding by field, and • collecting data on interdisciplinary R&D and emerging fields. Furthermore, the panels recommended collecting (if feasible) data on foreign sources of funding, R&D collaboration, proposals and awards, technology trans - fer activities, and R&D personnel. As with the business survey, NSF has efforts under way to improve the academic survey by way of redesign. Based on findings from 15 institutional visits, SRS has found a demand to expand the number of funding categories in the questionnaire; for example, nonprofits as a source would be added. Clinical trials in R&D, which are currently excluded, will be added. And information on all sources of funding (including foreign)—not just total and federal—will be requested for all fields. A major objective is to be able to determine how much industry funding is for specific fields, such as engineering or biomedical research.

OCR for page 60
 INTANGIBLES AND GOVERNMENT MEASUREMENT Jankowski also noted that fields of coverage need to be updated and a ques- tion on interdisciplinary research added. He was not overly optimistic about that happening soon. The survey also requests some minimal information about R&D faculty and other personnel, as well as R&D proposal submissions. Find - ings from the institutional visits also revealed a demand to add a small module on technology transfer activities, and information on total R&D expenditures by cost categories (salaries, indirect costs, equipment, supplies, etc.) to address some BEA data needs. At the time of the workshop, NSF was in the fairly early stages of putting the new questionnaire together. The goal is to go through the stages of development, similar to those for the business survey, to pilot the survey with 40 institutions in the fall of fiscal year 2009, and to be able to launch the redesigned survey by fall 2010 to collect fiscal year 2010 data. 5.4. ADVISORY COMMITTEE ON MEASURING INNOVATION IN THE 21ST CENTURY Cynthia Glassman spoke about a focused government effort directly relevant to the workshop topic: a major initiative to develop a comprehensive set of mea - sures of innovation in the economy by the secretary of commerce’s Advisory Committee on Measuring Innovation in the 21st Century.5 The advisory committee was established in September 2006 in response to Secretary Carlos Gutierrez’s concern that measures of innovation for the economy were inadequate. Consisting of 10 chief executive officers (CEOs) and busi- ness representatives and 5 academics, “it was meant to be practical rather than theoretical.” Carl Schramm from the Kauffman Foundation served as chair. 6 The advisory committee’s charter stated that it would advise the secretary on new or improved metrics to improve understanding of how innovation occurs in different sectors of the economy, how it is diffused across the economy, and how it impacts economic growth and productivity. Glassman noted that the initiative was never intended to establish a magic number—for example, that “innovation was X this year, and Y the next year.” It would be great if that were possible, she added, and maybe some day it will be, “but we are not there yet.” First, Glassman described the committee process, which influenced the manner in which its recommendations were developed: Prior to the first meet - ing, Glassman and her staff met with committee members individually to 5The final report of that committee can be read at http://www.innovationmetrics.gov/. 6 Business members were Steve Ballmer, Microsoft Corporation; David L. Bernd, Sentara Health - care; James Blanchard, Synovus Financial Corp.; George Buckley, 3M; Art Collins, Medtronic; Michael Eskew, UPS; Luther Hodges, Jr., Phoenix Associates, Inc.; John Menzer, Walmart; and Samuel J. Palmisano, IBM Corporation. Academic members were Ashish Arora, Carnegie Mellon University; Rajesh Chandy, University of Minnesota; Kathleen B. Cooper, Tower Center for Political Studies, Southern Methodist University; Dale W. Jorgenson, Harvard University; and Donald Siegel, University of California at Riverside.

OCR for page 60
 INTANGIBLE ASSETS understand how, in their view, innovation occurs and their ideas for measuring it. CEOs reported that the culture within organizations and their willingness to accept failure and to take risk were very important to how innovation happens in their companies. Drivers of innovation, as well as regulatory impediments to it, were discussed. Glassman noted that the interviews did not produce much information about how the companies measure innovation. There was no consistent measurement methodology across firms, although some consistent themes emerged. One theme was that companies tended to assess their innovation, at least to some extent, based on some concept of market share. If they were growing, and if the industry whose market they were in was growing, that was viewed as an indicator that efforts to innovate had been successful; again, there was no spe - cific measurement of that, only a basic concept. Another theme that was voiced sounded something like a “vitality index.” It was called different things by dif- ferent respondents but, basically, it is the percentage of growth, revenues, sales, or whatever was relevant to the firm, attributable to new innovative activities over the past three to five years. The committee’s definition of innovation referred not only to products, but also to services, organizational structure, marketing, and processes—whatever it was that companies did that was new and different and that resulted in an increase in performance, however defined. The fact that no con- crete or systematic measure of innovation was uncovered made the committee’s task all the more challenging. Glassman reported that the committee’s first meeting, held in February 2007, involved discussions of data gaps and potential recommendations, building on what had been gleaned from the pre-meeting conversations. The committee then requested public comments on its charge and more generally on ideas germane to the measurement of innovation, which resulted in submissions by 34 individuals and organizations (see the above cited website). The first meeting was followed by another round of phone calls with individual committee members to discuss what had been learned during the meeting and from the public comments and to identify potential key issues that the recommendations might ultimately address. A second public meeting was held in September 2007. The final report was drafted and circulated and then made public on January 18, 2008. In laying out key themes and guiding principles for measuring innovation, the committee recommended that measures be practical and relevant and that input required for their construction should impose a minimal burden on institu - tions and businesses. It also became evident that not all the measures could be quantitative and that there were some qualitative issues that justified some atten - tion and were relevant. Three sets of recommendations emerged—one for government, one for the private sector, and one for researchers. The recommendations for the government occupied the largest portion of the report. The first was to create a formal frame- work for identifying and measuring innovation in the national economy. This

OCR for page 60
 INTANGIBLES AND GOVERNMENT MEASUREMENT included developing annual industry-level measures of total factor productivity by restructuring the national income and product accounts. In addition, it was recommended that a supplemental innovation account be created for the national income and product accounts to expand the categories of innovation-related inputs. The purpose here was to broaden some of what BEA was already doing—specifically, the agency’s satellite R&D account—to include other aspects of innovation. Other recommendations specified the need to improve service-sector data, increase survey coverage, and improve measure- ment of intangibles, particularly intellectual property, by building on the work that has been led by NSF. Other recommendations for the government were to better leverage existing data and linkages, including synchronization among the statistical agencies. Glassman noted that most data synchronization efforts will require legislation beyond that specified in the Confidential Information Protec - tion and Statistical Efficiency Act of 2002.7 The committee also recommended increasing access to data to facilitate more robust cutting-edge research, including fostering the work of data tagging. Data tagging, something that the Securities and Exchange Commission (SEC) has been promoting, involves the use of extensible business reporting language (XBRL), a standards-based system that allows software vendors, programmers, and others developers to enhance the creation, exchange, and comparison of busi- ness reporting information. XBRL improves the ability of the SEC and others to . use data that companies file in a way that enables better analysis and research across companies and over time than can be done with the current flat files that the reports now come in. The SEC is pushing this idea, and some companies, including Microsoft, are already reporting in XBRL (as recommended in National Research Council, 2007). The committee also suggested that work continue toward development of a national innovation index, although the group thought that it was premature to do this now. The committee also supported funding to implement its recommen- dations. Responding to a question during discussion, Glassman noted that the committee examined existing innovation indicators, such as those that have been used by the Organisation for Economic Co-operation and Development, Eurostat, and Canada, in trying to work toward international comparability. In fact, the committee explicitly recommended that work continue to develop consistent, if not identical, measures as other countries. Moving beyond government, the committee issued guidance on how busi - nesses could help with innovation measurement. The guidance involved essen - tially two recommendations: one is to create firm and industry-level measures of innovation and develop best practices for innovation management and accounting. An important contribution of the committee was to raise the consciousness of the 7 SeeNational Research Council (2006, 2007) for detailed discussions of the Confidential Informa - tion Protection and Statistical Efficiency Act of 2002, and of ways in which it could be expanded to improve business statistics through data sharing and greater agency program coordination.

OCR for page 60
 INTANGIBLE ASSETS community of stakeholders about the importance of measuring innovation. Much work has been done on a variety of aspects of measuring intellectual property, valuing patents, rethinking the financial reports, and a whole range of issues, to which the committee called attention, raising awareness of the importance of innovation itself and the need to measure it. The other recommendation for busi - nesses is that they participate in research activities themselves and make innova- tion information available to outside researchers. On the research front, the committee recommended further work aimed at identifying and assessing innovation outcome measures. Thus far, measures of innovation have emphasized R&D, patents, intellectual property, and other inputs—all very important—but the committee recommended going further to focus on innovation output and how that affects the economy. The committee also recommended that researchers work to identify gaps in innovation data and how they might be filled, analyzing relationships between innovation activities and collaboration, innovation performance, and firm performance. One of the concepts that was raised by both the academic and business members of the committee, as well as in the public comments, is the importance of collaboration—within the firm, between innovators and the producers, among firms, and between companies and the government and companies and research - ers globally. Collaboration, though very difficult to measure, seems to be impor- tant, in the view of many. Glassman reported that, upon publication of the final report, the secretary of commerce asked BEA to work with BLS to provide a comprehensive accounting of the effect of high-tech goods and services on growth and productivity. BEA plans to unveil a design for a supplemental innovation account this year and is working with NSF to expand and collect R&D information on innovation-related inputs.