Chapter 2:

Accounting for Defense R&D: Gap Between Performer- and Source-Reported Expenditures

In many OECD countries, including the United States, total government R&D support figures reported by government agencies differ substantially from those reported by performers of R&D work. Consistent with international guidance and standards (OECD 1994), however, most countries' national R&D expenditure totals and time series are based primarily on data reported by performers. This convention is preferred because performers are in the best position to indicate how much they spent in the actual conduct of R&D in a given year and to identify the source of their funds. Although there are many reasons to expect funding and performing series to differ-such as different bases used for reporting government obligations (fiscal year) and performance expenditures (calendar year)-the gap between the two R&D series has widened during the past several years. Additionally, the divergence in the series is most pronounced in countries with relatively large defense R&D expenditures.

For the United States, the reporting gap has become particularly acute over the past several years. In the mid-1980s, performer-reported Federal R&D exceeded Federal reports by $3 to $4 billion annually-5 to 10 percent of the government total. This pattern reversed itself toward the end of the decade; in 1989 government-reported R&D total exceeded performer reports by $1 billion. The gap has since grown to about $5 billion. In other words, about 7 percent of the government total in the late 1990s is unaccounted for in performer surveys. (See figure 2-35.)

To investigate causal factors for the reporting gap, NSF-working with DOD contract-specific data-conducted on-site interviews with carefully selected companies that perform Federal R&D for DOD. Companies were asked about their R&D activities, data collection and reporting methods, and subcontracting practices. They also were asked to volunteer information about other factors that might influence the growing reporting difference. On the basis of these interviews and supplemental data analyses, the following factors appear to contribute most to the observed data gap.

Shifts in the composition of R&D, test, and evaluation (RDT&E) contracts during the past 10 years-since the end of the Cold War-introduced numerous changes in DOD's budgeting choices. Between 1991 (the last year that Federal funding and performing totals were relatively close) and 1998, DOD procurement spending (in inflation-adjusted terms) fell by 50 percent, whereas RDT&E spending declined by a relatively modest 7 percent. Concurrently, the proportion of DOD's RDT&E funding of traditional R&D program activities such as missile and space systems, tanks, ships, and other weapons systems has decreased; funding of more generalized technical, analytical and professional service contracts has increased. This trend has been accompanied by the emergence of new, nontraditional contractors (including large communication carriers and small high-technology firms) and firms specializing in program support activities within the DOD-funded R&D-performing industrial sector. Consequently, an increasing share of what DOD now funds, and therefore reports as R&D, is not necessarily perceived as R&D by industry performers. Industry representatives also mentioned significant changes in DOD's overall budget environment whereby RDT&E funds are now used to update military equipment under an emerging procurement management concept called "repeated R&D," whereby new technology is being incorporated on an ongoing basis into military systems. The effect is that RDT&E appropriations are now funding activities that could have been considered production 10 years ago. In short, there has been a change in what constitutes the R&D activity that is not similarly captured from Federal and industry respondents.

As a result of major changes in DOD's efforts to streamline its procurement environment and practices, the use of large, flexible, multiyear, multi-agency, indefinite order-type contract vehicles has become increasingly common. These contracts, which can be used by nearly every Federal agency, significantly reduce administrative and procurement actions needed to acquire services and technical support from previously selected contractors. They also have very high funding "ceilings" that allow government agencies to order tasks as needed. These contract vehicle characteristics tend to hide the ultimate funding source for particular activities and confuse the original "color of money" (i.e., the nature of the originating appropriation accounts). The effects of these procurement reforms were widespread in 1992 and 1993.

Finally, the consolidation of the defense and aerospace R&D business (see figure 4-10 in NSB 1998), as well as other corporate mergers and acquisitions, has considerably complicated industries' tracking of defense-related R&D. Few firms (especially extremely large, diversified companies) maintain award-specific data on R&D contracts for their many subsidiaries. Consequently, R&D-intensive activities of acquired firms may not be visible at corporate headquarters responding to national R&D surveys. This reporting problem is magnified with recent growth in R&D outsourcing. In such circumstances, the subcontracted ("routine technical service") activity often is performed by companies with only scant knowledge of the original funding source and perhaps even less knowledge on the overall DOD R&D objective to which their work is contributing.

The relative importance of these considerations in quantifying these data differences is unknown. Clearly, however, a variety of factors affect the collection of consistently reported R&D data from performers and funders. A similar mismatching of Federal R&D to academia as reported by universities and Federal agencies is now appearing in the data series. In this instance, however, totals reported by universities exceed those reported by Federal respondents. Indeed, other countries also have difficulty tracking and matching performer and source data (see NSB 1998)-indicative of the transitional changes affecting the S&E enterprise globally.