Cover Image

HARDBACK
$73.75



View/Hide Left Panel

Statistical Analysis of the National Academy of Sciences Survey of Small Business Innovation Research Awardees: Analyzing the Influence of the Fast Track Program

David B. Audretsch,* Albert N. Link,** and John T. Scott*** Indiana University,* University of North Carolina at Greensboro,** and Dartmouth College***

EXECUTIVE SUMMARY

This paper summarizes the findings from a statistical analysis of the survey data collected by Peter Cahill for the National Academy of Sciences, under the sponsorship of the Department of Defense’ s Small Business Innovation Research program. Also, the findings from the statistical analysis are related to the findings from several regional case-based studies that compare, along a number of dimensions, Fast Track Phase II projects to non-Fast Track projects.

The primary conclusions from the statistical analysis of the survey data are:

  • Fast Track projects have greater expected sales (commercialization) than do non-Fast Track projects,

  • Fast Track projects experience a shorter funding gap between Phase I and Phase II awards than do non-Fast Track projects, and

  • Fast Track projects have greater employment growth than do non-Fast Track projects.

These findings are extremely robust. In addition, they complement the case-based analyses by other researchers.



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 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE Statistical Analysis of the National Academy of Sciences Survey of Small Business Innovation Research Awardees: Analyzing the Influence of the Fast Track Program David B. Audretsch,* Albert N. Link,** and John T. Scott*** Indiana University,* University of North Carolina at Greensboro,** and Dartmouth College*** EXECUTIVE SUMMARY This paper summarizes the findings from a statistical analysis of the survey data collected by Peter Cahill for the National Academy of Sciences, under the sponsorship of the Department of Defense’ s Small Business Innovation Research program. Also, the findings from the statistical analysis are related to the findings from several regional case-based studies that compare, along a number of dimensions, Fast Track Phase II projects to non-Fast Track projects. The primary conclusions from the statistical analysis of the survey data are: Fast Track projects have greater expected sales (commercialization) than do non-Fast Track projects, Fast Track projects experience a shorter funding gap between Phase I and Phase II awards than do non-Fast Track projects, and Fast Track projects have greater employment growth than do non-Fast Track projects. These findings are extremely robust. In addition, they complement the case-based analyses by other researchers.

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE INTRODUCTION Responding to a paucity of evidence about the impact of the Small Business Innovation Research (SBIR) program, recent studies document the fact that the SBIR does make a number of positive economic contributions. Most notably, research has found that the growth rates of SBIR firms exceed those of comparable small firms not receiving SBIR support (Lerner and Kegler, 1999), the social returns from SBIR-funded projects exceed the private returns (Link and Scott, 1999), and the SBIR program influences the entrepreneurial behavior of scientists by changing their career paths and inducing them to commercialize (Audretsch, Weigand, and Weigand, 1999). Although these studies conclude that the SBIR program makes a positive contribution to the commercialization of knowledge, they are somewhat limited in their scope of coverage. One of the more controversial aspects of the SBIR program was the introduction of the Fast Track Initiative in 1996. Under this initiative, firms winning Fast Track designation have priority for the funding of the Phase II award because additional outside funding is committed to the research. It is conjectured that this Fast Track option bestows at least three main advantages to firms. First, it provides a mechanism for avoiding, or at least reducing, the funding gap that often occurs between Phase I and Phase II research. The significance and impact of this funding gap is made clear in the case studies by Audretsch et al. (1999), Feldman (1999), Link (1999), and Scott (1999). Audretsch et al. (1999) report examples related to funding-gap problems in their Indiana-based case studies. For example, the cofounders of a new startup developing genetically based rats experienced a funding gap between Phase I and Phase II research. The turnover of key personnel that resulted from the lapse of funding forced the company to incur retraining because of key personnel turnover. One purpose of the Fast Track is to assist small firms in avoiding such redundant cost burdens. Second, the Fast Track program may reduce complications arising during the normal review process. For example, Audretsch et al. (1999) report in their case studies that Anthony Hubbard, founder of Endotech, Inc., noted from his non-Fast Track experience that [t]here was no continuity of reviewers between our Phase I and Phase II proposals. It was like the Phase II review board ignored our Phase I results and overlooked that we had met our Phase I goals. A third possible gain from the Fast Track program comes through certification. As Kegler and Lerner (1999) point out, there is a growing body of empirical research that suggests that new, technology-based firms are burdened with asymmetric information between them and external financing institutions. The certifi-

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE cation involved in the Fast Track process may encourage third-party financing by alleviating information asymmetries. In responding to the lack of evidence about the impact of the Fast Track Initiative, the U.S. Department of Defense (DoD) requested that the National Academy of Sciences (NAS) review its SBIR Fast Track program to determine, to the extent possible, if the Fast Track Initiative encourages more rapid commercialization of research results through the acquisition of private investment capital, and if Fast Track projects progress more rapidly than do the standard SBIR awards. To accomplish this, the NAS undertook a multifaceted research strategy that included both a broad-based mail survey to a representative sample of SBIR awardees and focused regional case studies of firms taken from that sample. An overview of the mail survey is presented by Cahill (1999). The focused regional case studies are described by Audretsch et al. (1999), Cramer (1999), Link (1999), and Scott (1999). We expand on the Cahill (1999) analysis in one very important aspect. We present a number of statistical analyses that examine the relationship between the Fast Track program and several different performance output variables. The analyses described herein control statistically for the relationship between other factors and performance output. The remainder of the paper is outlined as follows: In the second section, the rationale for public funding of small technology-based firms is explained. In the third section, selected characteristics of the Cahill survey sample are presented. In the fourth section, the statistical models that we considered are presented, and the relevant variables are defined. In the fifth section, we present our results and our interpretation of them. We conclude with a brief summary of our findings. THE FINANCE GAP CONFRONTING SMALL TECHNOLOGY-BASED FIRMS Are small technology-based firms merely replicas of their larger counterparts? This is an important question because an affirmative answer would indicate that there is no reason to expect them to be financed differently than large firms. A large literature in economics has addressed this question and provided a resolute answer. Small technology-based firms are, in fact, markedly different from large enterprises in a number of key dimensions that have important implications for public policy, in general, and the SBIR, in particular. The first important difference is that large firms have a proven track record based on success. New firms have no such proven track record, and many SBIR firms are relatively new firms. Having a track record is crucial because efforts to commercialize new technologies are characterized by a greater degree of uncer-

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE tainty, knowledge asymmetries, and transactions costs than other types of economic activity. The greater degree of uncertainty is the result of not knowing how the new technology-based product can be made, and if there is a viable commercial market for the product (Arrow, 1962). The high degree of uncertainty makes a track record of success dealing with such uncertainly very important to outside investors.1 An additional complication stems from asymmetries in knowledge about the project and its prospects for success between the firm and external financiers. The firm may have a better—or at least different—understanding of how the technology can be produced and commercialized than do external investors, who generally do not have the same technological background, experience with the technology, or degree of specialization in the technology. This gap in technological competence and understanding results in high costs of transacting information about the technology, possible products, and potential commercial applications. For external investors to understand and evaluate accurately the prospects of the project, they need to invest in technological competence and experience, which can be prohibitively costly. These knowledge asymmetries are compounded by the prohibitive cost of transacting the knowledge about the project to external parties. Not only do these asymmetries exist but, because of the specialized knowledge required for path-breaking technologies, it becomes prohibitively expensive for external financiers to learn enough to evaluate the project, or even hire experts who can. Thus, external financiers are confronted with an inability to evaluate accurately a proposed new technology-based product. This is true for larger, established firms as well as for new small firms. However, there is an important difference that tilts the decision to provide finance toward large enterprises. Larger enterprises have established a proven track record, whereas the new small firms have not. External financiers may be uncertain about the outcome of the proposed project and even unable to evaluate accurately its technological and commercial prospects, but they can be certain about the past performance of the established large firm. This often is not the case for new enterprises, with no proven track record. As one of the founders of Genetic Models, Joe Pesek, learned (Audretsch et al., 1999) after unsuccessfully trying to obtain funding from traditional financial institutions: “Nobody is going to finance a firm with no assets, no product, and no track record.” The founders of Genetic Models, like many of the other SBIR firms, have no commercial track record because they had been involved in scientific research at the university prior to starting their firm. SBIR provides the needed certification to reduce the uncertainty confronting external financiers. According to Paul J. Hall, President and CEO of Integrated Biotechnology: 1   As Hebert and Link (1989, p. 47) argue in their synthetic definition of entrepreneurship, “An entrepreneur is someone who specializes in taking responsibility for and making judgmental decisions that affect the location, form, and the use of goods, resources, or institutions. ”

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE Even if you only received a $1 award, the fact remains that you have people extensively reviewing your product with impeccable credentials. Receiving an award really boosts your reputation and your credibility. The greatest benefit from receiving an award is the sense of comfort it provides for investors when our small firm is trying to raise capital. It is well known that the rate of success of innovative activity is low. Many, if not most, innovative projects do not succeed. However, studies have documented that because a specific project did not result in a viable commercial product does not necessarily mean that the knowledge and experience generated by that project have no economic value. In fact, most successful new products are the result of previous attempts at either the same product or a related but different one that failed. For large firms, much of the knowledge and experience resulting from failed projects is then applied to successive projects, enabling the firm and its investors to capture the economic value from the learning process inherent in risky innovative activity. However, a failed project for a small firm typically means that the firm will go out of business. The economics literature has documented as a virtual “stylized fact” (Geroski, 1995, Caves, 1998) that the failure rate of small firms is systematically higher than that for larger enterprises. Thus, the scientists from these small firms going out of business will take their knowledge and experience to other firms, often other small firms. Although this experience and knowledge may result in a commercial success for a different firm, the investors in the original failed firm are unable to realize any of the financial gains from the original investments. However, although the high rate of failure by technology-based small firms may deter private investors, the public concern is that the science be commercialized. This is the direct result of the externality of knowledge and experience created in innovative efforts by small firms that ultimately fail. The gap between the valuable and useful knowledge with a potential commercial value created in small firms and the ability of private investors to earn a return on that knowledge results in an underinvestment in technology-based small firms. This underinvestment in technology-based small firms is particularly pronounced in regions where there is a deficiency of technology-based small firms. Several important studies (Link and Rees, 1990; Feldman, 1994a,b) have documented how external sources of scientific knowledge are much more critical to small firms than to their larger counterparts. This means that the success of new technology-based firms is highly conditional upon the existence of other small technology-based firms in the same geographic region. This second type of ex-ternality associated with small technology-based firms results in a high propensity for these firms to cluster within tightly concentrated geographic regions (Audretsch and Feldman, 1996, Audretsch and Stephan, 1996). An important and valuable function of the SBIR is to induce the start-up of a critical mass of new technology-based firms, which can trigger the start-up of subsequent new firms. In examining the decision to start a new biotechnology firm, both Feldman

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE (1999) and Audretsch and Stephan (1996) show the importance of accessibility to similar small biotechnology firms. Private investors will underinvest in small technology-based firms in regions where a cluster of technology-based firms is lacking. However, the externality generated by creating such a cluster of small technology-based firms, which will then make it profitable for private investors to finance subsequent start-ups, provides a clear mandate for public support to compensate for the finance gap existing between large and small firms. OVERVIEW OF THE FAST TRACK SURVEY As described in Cahill (1999), a mail survey was sent in early 1999 to a representative sample of companies that had received an SBIR Phase II award since 1992. This sample of 379 projects consisted of all 48 Fast Track projects funded since the inception of the program in 1996, all 127 BMDO co-investment projects funded between 1992 and 1996, a matched control group for Fast Track and BMDO projects, and an additional 29 projects for population adjustments. A total of 232 surveys were returned partially or totally completed. That responding sample was defined as the parent sample for this statistical study. The statistical models described in Section III, and the relevant findings discussed in Section IV, relate to a subset of the parent sample of 232 projects. That subset contains information related to 112 projects. A number of surveys were returned partially completed; when information relevant to the analyses of this paper was missing the project was deleted from consideration. In terms of Fast Track projects, which are the focus of this paper, 12.7 percent of the initial sample of 379 projects were Fast Track projects, 18.1 percent of the 232 returned surveys represented Fast Track projects, and 16.1 percent of the 112 projects considered herein are Fast Track projects. THE STATISTICAL MODEL The fundamental model considered in this study is described, in the most general terms, by Eq. (1). Performance output (OUTPUT) associated with an active research project is assumed to be functionally related to the experience of the research company (EXP), the associated strategy adopted by company (STRATEGY), the technical characteristics of the project (TECH), and whether the project was funded as a Fast Track (FT) project or not. We represent this behavioral model as OUTPUT = f (EXP, STRATEGY, TECH, FT) (1) Five alternative performance output measures are considered:

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE actual sales (in dollars) realized to date resulting from the technology developed during the Phase II project (ActSales); sales (in dollars) expected from the technology developed during the Phase II project between now and the end of 2001 (ExpSales); actual sales (in dollars) realized to date from the technology developed during the Phase II project plus sales expected between now (the time of the survey) and the end of 2001 (ActExpSales); duration of the funding gap (in months) between the completion of Phase I and the beginning of Phase II for the project (DurGap); and number of employees hired as a result of the technology developed during the Phase II project (Employ).2 The precedence for these five performance output measures comes from evidence, both quantitative and anecdotal, collected by Audretsch et al. (1999), Link (1999), and Scott (1999) while conducting case studies of SBIR awardees. The experience of each research company in the subset of 112 projects was characterized along five dimensions, each dimension hypothesized to have an independent influence on performance output. The variables that characterize these dimensions are age of the business (in years) defined in terms of its founding date (AgeBus); experience of the business founder(s) measured dichotomously in terms of whether the founder(s) most recently came from another private company, or not (ExpFounder)3; size of the business defined in terms of its total revenues during the previous fiscal year (Revenues)4; research experience of the company as measured by the number of previous Phase II awards that it has received (PhaseII); research stage of the company as measured dichotomously in terms of whether the Phase II award has been completed (Complete).5 Company strategy was characterized in only one dimension: marketing plans of the company as measured dichotomously if the company has a marketing (e.g., commercialization) plan under way or completed, or not (Market).6 2   A company could report a fractional unit of an employee’s time. 3   ExpFounder equals 1 if the founder’s most recent employment was in another private company and 0 otherwise (e.g., with a college or university or the government). 4   On the Cahill survey, respondents noted company revenue as the range of total revenue (less than $100,000; $100,000 to $499,999, $500,000 to $999,999, $1,000,000 to $4,999,999, $5,000,000 to $19,999,999, and more than $20,000,000). The variable Revenues is defined as the midpoint of each stated range, with the lower bound defined as $50,000 and the upper bound defined as $25,000,000. 5   Complete equals 1 if the Phase II research is completed, and 0 otherwise. 6   Market equals 1 if the company has under way or has completed a marketing plan, and 0 otherwise.

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE Two technical characteristics of each project were considered: length of time (in years) that the Phase II award has been active, measured as the time between the year that the award started and 1999 (Active)7; primary technology area of the research, defined in terms of the Small Business Administration’s (SBA) general technology areas.8 Finally, the process for funding each award was defined by one variable: funding process was measured dichotomously in terms of whether the award was a Fast Track (FT) Phase II, or not.9 Table 1 reports the mean value for each of the variables described above. 10 STATISTICAL FINDINGS Variations of Eq. (1), for each of the five alternative performance output variables defined above, were estimated using ordinary least-squares analysis. Each of the explanatory variables discussed earlier was included in each of the five estimated equations. The estimated regression results are in the second column of Table 2, Table 3, Table 4, Table 5 through Table 6. The findings in column 2 of Table 2, Table 3 through Table 4 can be discussed as a group because each has as a dependent variable a sales or expected sales measure. Thus, as a group, these three specifications relate to commercialization activity. An inspection of the regression results in column 2 of Table 2, Table 3 through Table 4 suggests that, ceteris paribus: The survey respondents associated with Fast Track projects have a greater expectation of future sales than those associated with non-Fast Track 7   All projects considered in the sample of 112 are active projects even if the Phase II research is complete. 8   On the basis of a reading of the technical abstracts of each award, SBA assigned several detailed technology codes to each project. Based on the assumption that the first code identified by SBA is the dominant technology, a more aggregate technology area was assigned to each project. SBA defines seven broad technology areas: Computer, Information Processing, Analysis (Computer); Electronics (Electronics); Materials (Materials); Mechanical Performance of Vehicles, Weapons, Facilities (Mechanical); Energy Conversion and Use (Energy); Environment and Natural Resources (Environment); Life Sciences (LifeScience). 9   FT equals 1 if the award was a Fast Track Phase II award, and 0 otherwise. 10   Two constructed variables are discussed. These variables are a probability of response to the survey variable (ProbResponse) and an associated hazard rate (HazardRate). These two related variables were constructed using all of the Cahill sample to estimate a model of the probability of response. The two variables were available for only 109 of the 112 observations. Observations were lost to either a perfect prediction in the probit model that generated the probability of response, or to missing observations that were used as controls in the probit model. The results from the underlying probit model are not reported herein, but are available upon request from the authors.

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE TABLE 1 Mean Values for Variables Used in the Statistical Analysis Variable Mean ActSales $175,021 ExpSales $6,299,554 ActExpSales $6,474,574 DurGap 7.58 months Employ 2.38 individuals AgeBus 11.13 years ExpFounder 0.7857 (78.57% from business) Revenues $5,547,321 PhaseII 6.01 awards Complete 0.3661 (36.61% completed) Market 0.5893 (58.93% with operational market plan) Active 2.82 years Electronics 0.5000 (50.00% in electronics technology) Computer 0.2143 (21.43% in computer technology) Materials 0.0714 (7.14% in materials) Mechanical 0.0496 (4.46% in mechanical performance of vehicles, weapons, and facilities) Energy 0.0982 (9.82% in energy conversion and use) Environment 0.0446 (4.46% in environment and natural resources) LifeScience 0.0268 (2.68% in life sciences) FT 0.1607 (16.07% Fast Track) ProbResponse 0.659 (65.9% probability of response) HazardRate 1.168 (conditional density) The survey respondents associated with Fast Track projects, in contrast, report lower current sales than those associated with non-Fast Track projects, although the estimated regression coefficient is only marginally significant in a statistical sense. This finding is not unexpected because Fast Track projects from awards in 1996 are just now being completed. Companies with a marketing strategy in place have realized, and expect to realize in the near future, greater sales than companies that do not have one. Older, more established companies, as measured by the age of the company, seem to have somewhat dampened expectations of future sales than do younger companies. This finding may reflect greater reasonableness in forecasting expectations. Regarding duration of the funding gap as a performance output measure, the results reported in column 2 of Table 5 suggest that, ceteris paribus: Fast Track projects are associated with a shorter funding gap, compared to non-Fast Track projects, as expected, given the focus of the Fast Track Initiative.

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE TABLE 2 Estimated Regression Results: Dependent Variable = ActSales   (2) Estimated Coefficient (3) Estimated Coefficient (4) Estimated Coefficient (1) Variable   t statistic Probability of Response t statistic Hazard rate for Response t statistic Intercept 41,094.9 (0.175) 553,777.5 (1.298) 253,384.2 (0.732) AgeBus −11,518.1 (−1.048) −11,879.3 (−1.071) −11,784.8 (−1.055) ExpFounder −47,862.1 (−0.350) −35,194.0 (−0.249) −30,692.7 (−0.215) Revenues 0.0236 (2.485) 0.0189 (1.882) 0.0207 (2.053) PhaseII −2,979.0 (−0.533) −1,973.2 (−0.346) −2700.2 (−0.473) Complete 263,649.4 (1.840) 201,098.6 (1.335) 232,003.6 (1.551) Market 231,439.1 (1.943) 258,994.1 (2.110) 250,487.2 (2.029) Active −42,668.4 (−0.667) −70,092.2 (−0.998) −56,503.3 (−0.802) Computer 327,922.9 (2.297) 298,169.2 (2.036) 313,263.7 (2.129) Materials 284,034.9 (1.251) 299,487.7 (1.262) 300,828.3 (1.257) Mechanical −14,2017.4 (−0.545) −92,018.0 (−0.347) −117,836.2 (−0.443) Energy 114,603.2 (0.611) 75,228.9 (0.394) 83,641.0 (0.434) Environment 290,793.3 (1.054) 256,384.1 (0.918) 265,770.8 (0.945) LifeScience −8,726.6 (−0.026) −48,978.1 (−0.145) −49,581.1 (−0.146) FT −210,619.4 (−1.276) −147,079.3 (−0.851) −166,812.4 (−0.954) ProbResponse __   −624,951.6 (−1.550) __   HazardRate __   __   −142,912.1 (−0.991) R2 0.224   0.244   0.233   F-level 2.00   2.00   1.88   Older, more established companies, as measured by the age of the company, seem to experience slightly longer funding gaps than do the younger companies, although the estimated regression coefficient is not significant in a statistical sense. One possible explanation for the positive coefficient is that there is less urgency among older companies to obtain gap funding because they have other sources of funds to rely upon; another possible explanation is that diseconomies of research and production scope associated with the relatively older companies decrease the ability of these companies to maintain an administrative schedule once new research begins. Finally, regarding employment growth during the Phase II projects, the results reported in Table 6 suggest that, ceteris paribus: Employment growth is greater in Fast Track projects than in non-Fast Track projects. Company founders who have a business background, compared to an academic or public-sector background, expand staffing slower, perhaps reflecting prior experience or lessons learned.

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE TABLE 3 Estimated Regression Results: Dependent Variable = ExpSales   (2) Estimated Coefficient (3) Estimated Coefficient (4) Estimated Coefficient (1) Variable   t statistic Probability of Response t statistic Hazard rate for Response t statistic Intercept 7,018,013.0 (1.193) 2.63E+07 (2.493) 2.08E+07 (2.46) AgesBus −575499.8 (−2.088) −594,588.8 (−2.169) −593407.7 (−2.17) ExpFounder −2,467,573.0 (−0.719) −2,531,579.0 (−0.723) −2,237,995.0 (−0.640) Revenues −0.063 (−0.264) −0.231 (−0.931) −0.2319 (−0.939) PhaseII 91,310.1 (0.651) 133,096.2 (0.944) 118,248.7 (0.846) Complete 241,319.1 (0.067) −1,875,286.0 (−0.503) −111,378.0 (−0.304) Market 6,712,359.0 (2.246) 7,451,051.0 (2.456) 7,375,226.0 (2.44) Active 248,885.0 (0.155) −873,956.6 (−0.503) −869,935.8 (−0.504) Computer 841,505.3 (0.235) −417,943.6 (−0.115) −277,023.8 (−0.077) Materials 302,986.1 (0.053) 664,196.8 (0.113) 457,868.9 (0.078) Mechanical −3,242,383.0 (−0.497) −1,475,863.0 (−0.225) −1,780,125.0 (−0.273) Energy 6,488,823.0 (1.380) 5,079,322.0 (1.077) 4,830,751.0 (1.024) Environment 1.54E+07 (2.229) 1.43E+07 (2.075) 1.42E+07 (2.067) LifeScience −7,648,164.0 (−0.915) −8,879,436.0 (−1.065) −9,498,163.0 (−1.140) FT 9,317,159.0 (2.252) 1.15E+07 (2.702) 1.18E+07 (2.764) ProbResponse __   −2.22E+07 (−2.225) __   HazardRate __   __   −8,204,561.0 (−2.325) R2 0.266   0.305   0.308   F-level 2.52   2.72   2.76   Companies with a marketing strategy in place report greater Phase II employment growth than those without a market strategy.11 Older, more established companies, as measured by the age of the company, seem to expand their staff slower than do the younger companies, although the estimated regression coefficient is only marginally significant in a statistical sense. As with the experience of the company founders, this may reflect prior experience or lessons learned. Estimation of our five basic models yields essentially the same results, as discussed earlier, whether or not we control for the possibility that the error in the estimating equation is correlated with the binary variable denoting Fast Track status, FT. There are two reasons that we might expect such a correlation, and if the correlation exists, our previously discussed estimates of the relationship between Fast Track status and performance output would be biased. First, if the probability of responding to the Cahill survey is associated with performance output, and if that response effect in the error term is correlated with 11   This finding should be interpreted with caution because the survey does not distinguish the category of employee growth being measured and hence it is possible that part of the reported growth included individuals in marketing.

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE TABLE 4 Estimated Regression Results: Dependent Variable = ActExpSales   (2) Estimated Coefficient (3) Estimated Coefficient (4) Estimated Coefficient (1) Variable   t statistic Probability of Response t statistic Hazard rate for Response t statatistic Intercept 7,059,108 (1.190) 2.68E+07 (2.529) 2.11E+07 (2.472) AgeBus −587,017.9 (−2.113) −606,468.1 (−2.197) −605,192.5 (−2.197) ExpFounder −2,515,435.0 (−0.727) −2,566,773.0 (−0.728) −2,268,688.0 (−0.644) Revenues −0.0391 (−0.163) −0.2123 (−0.849) −0.2112 (−0.849) PhaseII 88,331.1 (0.625) 131,123.0 (0.924) 115,548.5 (0.820) Complete 504,968.5 (0.139) −1,674,188.0 (−0.446) −881,776.4 (−0.239) Market 6,943,852.0 (2.305) 7,710,045.0 (2.524) 7,625,713.0 (2.505) Active 206,216.6 (0.127) −944,048.8 (−0.540) −926,439.1 (−0.533) Computer 1,169,428.0 (0.324) −119,774.4 (−0.033) 36,240.0 (0.010) Materials 587,021.1 (0.102) 963,684.6 (0.163) 758,697.2 (0.129) Mechanical −3,384,400.0 (−0.514) −1,567,881.0 (−0.238) −1,897,961.0 (−0.289) Energy 6,603,426.0 (1.393) 5,154,551.0 (1.085) 4,914,392.0 (1.034) Environment 1.57E+07 (2.253) 1.46E+07 (2.098) 1.45E+07 (2.09) LifeScience −7,656,890.0 (−0.908) −8,928,414.0 (−1.064) −9,547,744.0 (−1.137) FT 9,106,539.0 (2.184) 1.14E+07 (2.649) 1.17E+07 (2.705) ProbResponse __   −2.28E+07 (−2.272) __   HazardRate __   __   −8,347,473.0 (−2.348) R2 0.266   0.306   0.309   F-level 2.51   2.74   2.77   the variables included in our model, our estimates could be biased. To control for that possibility, we used the entire Cahill sample and estimated a probit model of response and constructed the probability of response and an associated hazard rate for each surveyed company. In our estimations, we controlled for the possibility that the probability of response affects our previously discussed results. We added in turn to our equations the probability density of response and the hazard rate for response. The specifications allowed us to see if there was something important in the error that was correlated with either the probability of response or the counterfactual conditional probability of response, counter-factually assuming nonresponse. In sum, we asked across the observations in the sample whether the variance in probability of response has an effect on the dependent variable, and whether controlling for that matters for the other estimates. Further, using the hazard rate, we ask across the observations whether the variance in the counterfactual hazard for response has an effect and whether the control affects the other estimates. Our results regarding the association of Fast Track projects with performance outputs are essentially unchanged when either the probability of response (results are shown in each table of regression results in column 3) or the hazard rate for response (results are shown in each table of regression results in column 4) is added to the specifications.

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE TABLE 5 Estimated Regression Results: Dependent Variable = DurGap   (2) Estimated Coefficient (3) Estimated Coefficient (4) Estimated Coefficient (1) Variable   t statistic Probability of Response t statistic Hazard rate for Response t statistic Intercept 6.8103 (2.918) 6.4712 (1.543) 7.7186 (2.288) AgeBus 0.1213 (1.109) 0.1165 (1.068) 0.116 (1.065) ExpFounder −0.6879 (−0.506) −0.7913 (−0.568) −0.7570 (−0.544) Revenues 1.26E−08 (0.133) 2.33E−08 (0.236) 7.52E−09 (0.077) PhaseII −0.0131 (−0.236) −0.0162 (−0.289) −0.0136 (−0.245) Complete 1.1487 (0.806) 0.6838 (0.462) 0.6059 (0.416) Market 0.5510 (0.465) 0.3477 (0.288) 0.4026 (0.335) Active −0.0154 (−0.024) 0.2084 (0.302) 0.0915 (0.133) Computer −2.7073 (−1.907) −2.5066 (−1.74) −2.6024 (−1.816) Materials 5.7556 (2.549) 6.9308 (2.97) 6.8676 (2.946) Mechanical −2.9899 (−1.154) −2.8411 (−1.091) −2.693 (−1.039) Energy 3.7107 (1.989) 3.7802 (2.014) 3.645 (1.941) Environment −1.4416 (−0.525) −1.1827 (−0.431) −1.285 (−0.469) LifeScience 0.3557 (0.107) 0.3762 (0.113) 0.2268 (0.068) FT −4.5941 (−2.799) −4.2573 (−2.506) −4.0109 (−2.353) ProbResponse __   −0.1179 (−0.030) __   HazardRate __   __   −0.810 (−0.577) R2 0.289   0.312   0.315   F-level 2.81   2.82   2.85   Second, the error in the estimating equations could be correlated with the binary variable denoting Fast Track status because Fast Track status can be modeled as an endogenous variable, with Fast Track status having as its ultimate cause better expected commercial performance. Thus, while we expect Fast Track status to help propel the commercial success of the so-funded SBIR project, it is also the case that projects with better prospects for commercial success will attract outside funding and thus will be placed on a Fast Track status. We recognize that both directions for causality—from Fast Track to expected performance and from expected performance to Fast Track—exist. That causality issue in itself is not necessarily a problem for estimating the relationship between Fast Track and performance output. However, we might expect the associated problem of the error in our estimating equations being correlated with the binary Fast Track variable. To control for that possible simultaneity bias, we estimated each of our five basic equations with instrumental variables (results not shown here), letting the binary Fast Track variable and the response variable be endogenous in addition to the left-hand-side variable for each equation.12 With both the survey 12   These results are also available upon request from the authors.

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE TABLE 6 Estimated Regression Results: Dependent Variable = Employ   (2) Estimated Coefficient (3) Estimated Coefficient (4) Estimated Coefficient (1) Variable   t statistic Probability of Response t statistic Hazard rate for Response t statistic Intercept 3.0047 (2.897) 6.1047 (3.276) 4.9009 (3.252) AgeBus −0.0737 (−1.517) −0.0762 (−1.572) −0.0758 (−1.559) ExpFounder −1.4127 (−2.336) −1.3659 (−2.208) −1.3238 (−2.130) Revenues 5.62E−08 (1.342) 2.87E−08 (0.655) 3.22E−08 (0.733) PhaseII −0.0025 (−0.102) 0.0037 (0.150) 0.0006 (0.025) Complete 0.4001 (0.632) 0.0238 (0.036) 0.1705 (0.262) Market 1.1675 (2.216) 1.310 (2.444) 1.2848 (2.39) Active −0.1308 (−0.462) −0.2964 (−0.965) −0.2693 (−0.878) Computer 0.0327 (0.052) −0.1464 (−0.229) −0.1009 (−0.158) Materials 0.5082 (0.506) 0.6195 (0.598) 0.5988 (0.575) Mechanical −0.8445 (−0.734) −0.5383 (−0.465) −0.6232 (−0.538) Energy 0.2280 (0.275) −0.0007 (−0.001) −0.0124 (−0.015) Environment 4.3522 (3.568) 4.165 (3.415) 4.174 (3.408) LifeScience −0.4954 (−0.336) −0.719 (−0.488) −0.7902 (−0.533) FT 2.0884 (2.863) 2.4804 (3.287) 2.4757 (3.251) ProbResponse __   −3.7334 (−2.119) __   HazardRate __   __   −1.2083 (−1.925) R2 0.357   0.387   0.382   F-level 3.85   3.92   3.84   data and the associated project-specific data collected by DoD, a number of instruments were identified—variables highly correlated with Fast Track status, yet arguably not affected by the error in expected performance—to use in the estimations. For example, experience with private investors prior to the SBIR award, the number of founders with a business background, and the agency making the SBIR award were among the instruments available. Our results regarding the relationship between Fast Track status and performance output are essentially the same qualitatively in the instrumental variable specifications. CONCLUSIONS Although the empirical results provide compelling evidence that the SBIR generally has promoted the commercialization of scientific knowledge, detailed case studies have revealed several impediments hindering the effectiveness of the program. Small technology-based firms often have trouble financing the gap between Phase I and Phase II awards. This funding gap is particularly troublesome because it can cause the loss of investments in scientific training for key personnel. Although the Fast Track Initiative was initiated to overcome this impediment, until now there has been no systematic evaluation of the program.

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE The empirical results presented in this paper clearly support the claim that the Fast Track Initiative has measurable impacts on the performance output of Phase II award recipients. The evidence is overwhelming, when other factors are controlled for, that respondents associated with Fast Track projects report greater expectations of sales associated with the Phase II research, shorter funding gaps between Phase I and Phase II awards, and more employment growth. Thus, the Fast Track program improves the ability of SBIR firms to transform science into commercially successful products. In addition, the empirical results presented here are consistent with a basic finding reported by Link and Scott (1999) based on a very different methodology. In the present paper, we have shown that Fast Track projects are associated with better commercial performance, where that performance is estimated directly by the SBIR respondents through survey responses. The paper by Link and Scott also finds that Fast Track projects are associated with better commercial performance, but in contrast to the present paper, a key measure of such performance is derived indirectly. In particular, instead of expected sales that the respondents have estimated on the Cahill survey, Link and Scott asked the respondents during extensive interviews a set of questions about investments and the duration of those investments, and then derived the expected profit stream from an economic model of investment behavior. Using their indirect method for estimating expected profits from the SBIR investments, Link and Scott find that the Fast Track projects are associated with greater rates of return on investment. Both our paper and the Link and Scott paper find that the Fast Track projects as a group have greater commercial potential than the non-Fast Track projects. Further, the basic finding of our paper—the association of Fast Track projects with better performance regarding expected sales and employment growth, and a more rapid movement from Phase I to Phase II development—remains when we control for the possibility that the failure to respond by some recipients of the survey could affect the results of our estimations. Additionally, the basic finding of the paper remains when we control for the possibility that there is a simultaneous-equations effect because Fast Track status itself is affected by expected performance. The basic result—the association of Fast Track with better performance—survives econometrically both the hazard-rate response bias scrutiny, and the instrumental-variables simultaneity bias scrutiny. REFERENCES Arrow, Kenneth J. 1962. “Economic welfare and the allocation of resources for invention,” pp. 609–625 in The Rate and Direction of Inventive Activity: Economic and Social Factors. Princeton, N.J.: Princeton University Press. Audretsch, David B., and Maryann P. Feldman. 1996. “R&D spillovers and the geography of innovation and production,” American Economic Review 86(3):630–640. Audretsch, David B., and Paula E. Stephan. 1996. “Company-scientist locational links: The case of biotechnology,” American Economic Review 86(3):641–642.

OCR for page 291
The Small Business Innovation Research Program: AN ASSESSMENT OF THE DEPARTMENT OF DEFENSE FAST TRACK INITIATIVE Audretsch, David B., Juergen Weigand, and Claudia Weigand. 1999. “Does the Small Business Innovation Research program foster entrepreneurial behavior? Evidence from Indiana,” this volume. Cahill, Peter. 1999. “Fast Track: Is It Speeding Commercialization of Department of Defense Small Business Innovation Research Projects?” this volume. Caves, Richard E. 1998. “Industrial organization and new findings on the turnover and mobility of firms,” Journal of Economic Literature 36(4):1947–1982. Cramer, Reid. 1999. “Patterns of firm participation in the Small Business Innovation Research Program in Southwestern and Mountain States,” this volume. Feldman, Maryann P. 1994a. “Knowledge complementarity and innovation,” Small Business Economics 6(5):363–372. Feldman, Maryann P. 1994b. The Geography of Knowledge. Boston: Kluwer Academic. Feldman, Maryann P. 1999. “Role of the Department of Defense in Building Biotech Expertise,” this volume. Geroski, Paul A. 1995. “What do we know about entry?” International Journal of Industrial Organization 13(4):421–440. Hebert, Robert F., and Albert N. Link. 1989. “In search of the meaning of entrepreneurship,” Small Business Economics 1(1):39–49. Lerner, Josh. 1999. “The government as venture capitalist: The long-run effects of the SBIR program,” Journal of Business. July, v. 72 3, pp. 285–297. Josh Lerner, and Kegler, Colin. 1999. “Evaluating the Small Business Innovation Research Program: A Literature Review,” this volume. Link, Albert N. 1999. “An Assessment of the Small Business Innovation Research Fast Track Program in Southeastern States,” this volume. Link, Albert N., and John Rees. 1990. “Firm size, university based research and the returns to R&D,” Small Business Economics, 2(1):25–32. Link, Albert N., and John T. Scott. 1999. “Estimates of the Social Returns to Small Business Innovation Research Projects,” this volume. Scott, John T. 1999. “An Assessment of the Small Business Innovation Research Program in New England: Fast-Track Compared with Non-Fast Track Projects,” this volume.