Repositioning the Advanced Technology Program
ATP should be refocused to better fill a major gap in the nation’s innovation system.
The Commerce Department’s Advanced Technology Program (ATP) has been the lighting rod of U.S. civilian technology policy during the past decade. Consider the program’s short but volatile budget history. ATP was initially funded in 1990 at $10 million and reached $68 million by the end of the first Bush administration. In 1993, the new Clinton administration promptly tripled ATP’s budget, with plans to boost funding to $1.5 billion, a level comparable to that of the Defense Advanced Research Projects Agency. Indeed, ATP was slated to be the civilian equivalent of that acclaimed agency.
Then beginning in 1995, the new Republican-controlled Congress targeted the program for outright elimination. For the rest of the Clinton administration, ATP was enmeshed in an executive-legislative impasse and limped along with appropriations of about $200 million a year. Now the Bush administration has proposed suspending ATP’s funding.
These budgetary convolutions reflect a deep political divide. Whereas advocates say ATP plays an instrumental role in filling a critical gap in the national innovation system, critics call the program the ultimate high-tech form of corporate welfare. There is some truth in each of these arguments. More important, they also share some common ground, which could provide the foundation for a renewed ATP. Indeed, there are three ways in which the program could be optimized to enhance the U.S. research and development (R&D) enterprise.
The valley of death
Of the standard R&D categories–basic research, applied research, and development–the middle category is the most problematic for the United States. A 1998 House Science Committee report eloquently described the “widening and deepening” of a “valley of death” between federally funded long-range basic research on the one hand and industry-financed near-term product development on the other. What is deficient is mid-range applied research that builds on fundamental scientific discoveries to address real-world needs short of actual product development.
National Science Foundation (NSF) data show that although total U.S. R&D investment has increased an average of 5 percent per year since 1993, applied research expenditures have not kept pace. The federal government’s applied research effort in particular has declined in constant dollars during the past 8 years. And contrary to conventional wisdom, neither corporate R&D nor venture capital has taken up the slack.
An Issues article by Richard Rosenbloom and William Spencer (“The Transformation of Industrial Research,” Spring 1996) drew attention to an alarming decline in corporate research. But last year, Charles Larson (“The Boom in Industry Research,” Summer 2000) pointed to a resurgence in corporate R&D. Although total industry R&D has indeed steadily risen since the mid-1990s, latest NSF figures show that gains in corporate applied research trail those in the basic research and product development categories. Moreover, the data show a deceleration through the 1990s in the corporate applied research effort. Although that effort enjoyed an 8.75 percent average annual increase for 1993-97, that rate has dropped to 4.66 percent for 1997-2000. As a percentage of total industry R&D, applied research is now lower than at any time since 1993.
Recent news accounts warn of a softening in overall corporate R&D intensity. The Ziff-Davis News Network reports that R&D-to-sales ratios are down for companies as diverse as Ariba, i2 Technologies, IBM, JDS Uniphase, Microsoft, Monsanto, Siebel Systems, Sun Microsystems, and Xerox. For 2001, the Battelle Institute, Industrial Research Institute, and R&D Magazine project the smallest gains in industry R&D growth since the mid-1990s. The slowdown is expected to continue into 2002. Although these latest reports do not distinguish between basic research, applied research, and product development, the overall slowdown would hardly be consistent with significant increases in either of the first two categories.
An analysis of venture capital (VC) funding also undercuts reports of the early demise of the valley of death. As reported by PriceWaterhouseCoopers and VentureOne, venture funding rose from $6 billion in 1995 to nearly $69 billion in 2000. Little of this financing, however, has gone to applied let alone basic research.
Particularly striking is the paltry level of venture capital support for start-up/seed financing, which goes to “proof of concept” and is thus akin to applied research. Even during the VC boom years in the late 1990s, only 4 percent of all funds went to start-up/seed financing. The balance went toward follow-up funding of business, not technology, development. Venture funds flowed almost exclusively into Internet-related investments that promised returns in as little as three to six months. Applied research efforts with three- to six-year development cycles were left with residual financing, if any.
The bursting of the e-commerce bubble in 2000 has only worsened matters as gun-shy venture capitalists have refocused their portfolios on second-, third-, and fourth-round financing of earlier investments. The latest data for 2001 show barely one-half of one percent of VC funding going to earliest stage investments. The valley of death is alive and not doing well for national innovation system.
The ATP approach
Helping to bridge the valley of death provides a core rationale for ATP. Managed by Commerce’s National Institute of Standards and Technology (NIST), ATP’s mission differs from most of the rest of the federal R&D portfolio. Rather than either basic or agency mission-oriented research, ATP supports industrially relevant technologies that promise significant commercial payoffs but are too high-risk and too far removed from the marketplace to adequately attract private investment. This is valley of death territory. During the past 10 years, ATP and more than 1,000 awardees have invested more than $3 billion in R&D.
ATP opponents, including conservative think tanks such as the American Enterprise Institute and the Cato Institute, regard the program as one of the federal government’s top examples of corporate welfare. Yet the most frequently cited liabilities of social welfare programs largely do not apply to ATP. For example, critics say such programs often have minimal or questionable eligibility criteria. Yet ATP employs rigorous merit-based selection criteria. Of 4,435 industry proposals submitted since its establishment, only 522, or fewer than 15 percent, have won ATP awards. Critics also see government assistance programs undermining the work ethic and individual responsibility. But bidders for ATP funding are required to submit detailed technology and business plans outlining credible commercialization strategies, including rough timetables, for bringing new technology to market. Plus ATP projects are at least 50 percent cost-shared by the award recipients. And ATP avoids the criticism that government programs inevitably turn into permanent entitlements. ATP awards are explicitly limited to three years for single-company recipients and five years for multifirm joint ventures.
The corporate welfare charge that does carry some weight, however, is the criticism that ATP may be supplanting and crowding out market-based mechanisms for technology development. Grounds for this criticism can be found in how ATP has positioned itself within the valley of death.
The valley of death is a wide expanse that cannot be filled by any single program or organization. A program such as ATP can be positioned in only a part of the applied research chasm: for example, either nearer to the side building off of basic research or nearer to the side leading to product development. ATP’s enabling legislation allows some flexibility in program implementation, and NIST has in fact steered the program to a particular position in the valley of death. Examining ATP’s general mission statements is not sufficient to specify this positioning. Instead, it’s revealing to look at the metrics by which the program assesses its own progress, impact, and achievements.
ATP commissions external reviews as well as conducting self-assessments. The metrics used in these assessments are numerous and vary widely in character. But a majority of them reveal a strong orientation toward a commercially penultimate form of applied research. For instance, ATP awards have been found to contribute to accelerated R&D progress; reduced R&D cycle times; accelerated commercial application, identification, and commercialization; reduced production costs; performance and productivity improvements; faster production prototyping; expansion of production capacities; efforts leading into early stages of production; faster speed-to-market performance; early revenue flows; and patents.
According to one assessment, 62 percent of ATP awardees have accelerated their commercialization plans by two years or more, and 77 percent of projects have completed product/process definition for at least one application. Forty percent of ATP awardees report that they are now able to make a new or improved product, and 71 percent of projects have conducted market-analysis activities, including sales forecasting and product development or testing. Twenty-five percent of projects have resulted in expanded production capacity, 41 percent have completed production prototypes for at least one application, 35 percent have completed pilot production or a commercial demonstration, and 12 percent have moved into actual production. Ten percent of companies have earned early revenues, and ATP awardees have filed 105 patent applications and seven copyrights and been issued 11 patents.
These figures derive from a set of completed ATP projects, and many of the reported results occurred after the end of direct government funding. Still, these types of results are used to demonstrate the program’s achievements. But these same metrics reveal a questionable aspect of ATP’s operational focus. By accelerating commercialization plans, product definition, market analyses, production prototyping, and pilot-line and actual production, and by contributing to new or improved products, expanded production capacity, early revenues, and new patents and copyrights, ATP comes as close as possible to corporate product development without actually doing it. Such close interface with activities properly relegated to the market is what makes ATP most vulnerable to the corporate welfare charge.
Although the Commerce Department denies that ATP crosses the line into product development, it can’t deny that a good part of the program does fall into the commercially penultimate category and that these efforts directly abut corporate commercialization activities. Moreover all parties must acknowledge that ultimately there is a gray line–if not a wide swath–where applied research blends into product development and where precommercial merges with commercial. The R&D community does not dispute that such a gray area exists between basic and applied research. That a similar fuzzy line exists between applied research and product development is also incontrovertible. Thus, claims that ATP is or is not subsidizing product development are essentially unresolvable.
Finding “higher” ground for ATP
ATP’s orientation toward commercially penultimate applied research is just one approach to helping bridge the valley of death. ATP could consider targeting other varieties of applied reserach that do not directly abut corporate product development. Moving to such “higher” ground would help protect it from charges of of corporate welfare.
A precedent exists for distinguishing among subcategories within the standard matrix of basic research, applied research, and product development. For instance, consider knowledge-driven basic research and targeted basic research. The former is captured by Vannevar Bush’s classic depiction of the “free play of intellects working on subjects of their own choice, in a manner dictated by their own curiosity.” The latter–most often undertaken by mission-oriented national laboratories and federal agencies–is largely basic in nature but pursued with a sense that some downstream use may exist.
The R&D literature also distinguishes among different varieties of applied research, including basic technological research and research on enabling and generic technologies. At issue for ATP is the identification of longer-range forms of applied research: in effect, “pre-precommercial” R&D. The key here is applied research that begins earlier in the innovation pipeline–research with longer payback periods and attendant higher technological and market risks.
Examples of higher-risk, longer-term applied research are plentiful. It took an 8-year development effort to produce the transistor and a 12-year effort to develop the microwave oven. And, of course, the Internet was under development by the government for 25 years before its commercial breakthrough in the early 1990s. Private-sector funding of such extended applied research is exactly what is lacking in the marketplace. ATP’s legitimacy can be strengthened by setting its sights not on basic research, but on this lowest point in the valley of death.
ATP can be repositioned in three areas.
Award selection criteria. ATP should invest in a broad portfolio of longer- as well as shorter-range applied technologies. Although no hard and fast boundaries exist between longer- and shorter-range research, ATP could aim to maximize relative differentials in the length of anticipated payback periods for its funded research. Contemplated here is a form of set-asides for higher-risk, longer-range applied research. For example, the Sematech semiconductor manufacturing technology consortium, when it was partially funded by the Defense Department, devoted 20 percent of its budget to targeted research at university centers of excellence, with the remainder focused on nearer-term R&D. Although it may not be advisable to set such a rigid ratio, ATP should also spread its bets when selecting research projects.
Networking with other R&D organizations. To reinforce the longer-range applied research focus, ATP should strengthen and build new relationships with research institutions further back in the innovation pipeline. In particular, ATP should make merit-based cost-shared awards to:
- Support research-intensive industrial organizations such as independent R&D enterprises and corporate research labs. Smaller research-intensive organizations such as BBN Corporation, SRI International, and RAND have made instrumental contributions to today’s Internet and human-computer interface technologies. Large corporate R&D facilities such as Bell Labs, Sarnoff, Xerox PARC, and the IBM Watson Center pioneered the transistor, liquid crystal technology, personal workstations, and disk drive technologies. The demise of this long-range industrial research infrastructure is the defining feature of the valley of death. Although some ATP awards are already made to such organizations, this aspect of the program’s portfolio could be enhanced.
Awards to corporate labs would still be vulnerable to corporate welfare charges, particularly in the case of big corporations with deep pockets. But because this funding would be directed to the research side of these companies, the force of such charges should be diminished. ATP licensing requirements could also help fend off corporate welfare criticisms. Indeed, part and parcel of a greater ATP focus on longer-range applied research should be requirements for research-oriented organizations to transfer sponsored technology to commercial-oriented organizations (including within the same company). Quite naturally, ATP would not want to fund applied research projects that goes nowhere, staying within the confines of research organizations. Licensing strategies to disseminate ATP project results could become an important criterion in the granting of awards to these organizations. Licensing would also diminish the prospects of any one company consolidating the benefits of an ATP project. In this way, licensing could transform corporate welfare into, at least, broader industry-wide welfare.
- Support industrially relevant applied research in universities and national laboratories. Current legislation limits ATP awards to industrial for-profit organizations. Universities and federal laboratories are restricted to secondary status in ATP joint ventures led by industry or as subcontractors to single-company ATP recipients.
ATP authorizing legislation should be amended to allow for direct cofunding of industrially relevant applied research in academia and in the national labs. Both of these sectors have, of course, developed keen interests in pursuing commercial applications of their more traditional basic research efforts. ATP should not be shut out from such promising applied research efforts and opportunities. At the same time, ATP should not by any means get into the basic research business. This extension into co-funding of universities and national labs should not extend to ATP support for basic research. That end of the innovation pipeline is already covered by NSF, National Institutes of Health, and other agencies.
- Support industry participation in applied research collaborations with universities and national laboratories. In addition to direct ATP cost-sharing with universities and federal labs, the program could serve as a supporting third party in university-industry and industry-national laboratory research collaborations. ATP support could be added to university and national lab-funded cooperative research and development agreements, technology transfer programs, and industrial parks.
A longstanding concern about university-industry and industry-national laboratory research collaborations is that commercial technology transfer is often stunted by proclivities toward pure academic research or mission agency R&D as well as more general bureaucratic impediments in universities and the national labs. ATP support for the industry side in these collaborations and its overall experience in providing incentives for commercialization efforts could help break down these technology transfer barriers. This recommendation not only helps to move ATP toward the center of the valley of death, it also moves the program into a central integrative role in networking the nation’s various applied research efforts. Strengthened ATP-university-industry-national lab links could help fill leaks in the innovation pipeline as technologies move from basic research toward product development.
Program Assessment. Reflecting a reorientation toward longer-range applied research, assessments and measures of the program’s achievements should focus on less near-term and less commercially oriented metrics such as publications; licensing; literature citations; joint ventures between award recipients and non-ATP firms; mobility of sponsored researchers into product development, manufacturing, or management positions; and entrepreneurial startups by sponsored researchers.
The thrust of this list (as compared with the previous list of commercially oriented metrics) is to get ATP-sponsored research out of the lab so that it is broadly available in advance of actual commercialization. The emphasis would be on the startup of new ventures, the establishment of joint ventures, and the mobility of personnel rather than their ensuing commercialization efforts. Technological, personnel, and organizational advances would be emphasized rather than market- or product-level achievements. These types of measures cannot be as easily associated with outright product development or commercialization.
These proposals for the reorientation of ATP toward longer-range applied research should help to insulate the program from the most critical charges of corporate welfare. At the same time, the repositioning of ATP along these lines would solidify its legitimacy in addressing fundamental innovation gaps in the valley of death.