Employment Growth from Public Support of Innovation in Small Firms

3 National Research Council Database

33 3 National Research Council Database BACkGRouND oN ThE DATABASE The Small Business Reauthorization Act of 2000 mandated that, among other things, the NRC conduct “an evaluation of the economic benefits achieved by the SBIR program” and make recommendations to Congress for improvements to the program. In its evaluation of the SBIR program, the NRC steering committee charged with the study took several approaches, including multiple surveys, interviews, and more than 100 case studies.1 The results of the NRC’s survey approach are described in this chapter. The NRC conducted an extensive and balanced survey in 2005 based on a population of 11,214 projects completed from Phase II awards made between 1992 and 2001 by five agencies: 1) the DoD, 2) NIH (within the HHS), 3) NASA, 4) DOE, and 5) NSF. It was assumed as part of the NRC’s sampling methodology that Phase II awards made in 2001 would be completed by 2005. The five agencies surveyed accounted for nearly 97 percent of the program’s expenditures in 2005, as was shown in Table 2.1 in Chapter 2. Table 3.1 shows the distribution of the population of 11,214 projects by funding agency, as well as the percentage of the 11,214 projects funded by each agency. The total number of projects surveyed from the 11,214 projects was 6,408. The number and percentage of respondents from these 6,408 surveyed projects is shown in Table 3.2.2 The total number of responding projects was 1,916, and the average response rate across all five agencies was 30 percent. Also shown in Table 3.2 is the total number of projects in the final random sample of completed Phase II projects, by agency. For ease of reference, definitions of the variables from the survey and descriptive statistics are in the Glossary of Variables at the end of the book.3 34 Link and Scott The NRC surveyed a number of nonrandomly selected projects because they were projects that had realized significant commercialization and the NRC wanted to be able to describe such interesting success stories. These nonrandomly selected projects are not considered herein. The data reduction process that was applied to each agency to arrive at the final random sample is in Table B.1 in the Technical Appendix, also found at the end of the book. VARIABLES IN ThE DATABASE Three employment variables in the NRC database are the focal variables in the chapters that follow.4 The variables, using 2005 data, include 1) number of employees, 2) number of employees hired as a result of the SBIR award, and 3) number of employees who, as a result of the award, were retained after the SBIR project was completed.5 Detailed information about types of employees is not available from the NRC survey, and this is not surprising. In small research firms, employees are technical workers, and they each wear many hats. As has been documented in previous case studies of Department of Defense SBIR projects (e.g., Wessner 2000), a given employee could be involved in research, marketing, and also management. Table 3.1 Population of SBIR Phase II Projects, 1992–2001 Agency Completed Phase II projects Percentage DoD 5,650 50.38 NIH 2,497 22.27 NASA 1,488 13.27 DOE 808 7.21 NSF 771 6.88 All agencies 11,214 100.00 NOTE: “Percentage” column does not sum to 100.00 because of rounding. SOURCE: Authors’ calculations. National Research Council Database 35 The size of the SBIR awardee firms, as measured by the number of employees in the firm in the year of the survey, 2005, varies across firms and across agencies from a mean low of 41 employees among the 141 NSF firms in the random sample to a high of 63 employees among the 155 NASA firms. See the descriptive statistics in the Glossary of Variables. However, the range of firm sizes in 2005 is from 1 employee to 4,001.6 ISSuES oF SAmPLE SELECTIoN BIAS The NRC database does not contain information on projects that were not funded by SBIR, and it does not contain information on projects for which a firm applied for SBIR support but was declined, or on comparable projects in firms that did not seek SBIR support. This lack of so-called matched pairs has been viewed by some (e.g., Wallsten 2000) as a...