Employment growth from the Small Business Innovation Research program

The Administration’s view toward economic recovery is focused on job creation and employment growth, as reflected in the American Recovery and Reinvestment Act (ARRA) of 2009. While the ARRA was legislated in the aftermath of what is arguably the worst economic downturn in the past 80 years, there are indicators that an emphasis on innovation-based employment growth will become an integral part, if not a centerpiece, of the pending implementation of the America COMPETES Act of 2007 as the economy recovers.

Kendrick and Grossman (1980) reported annual total factor productivity growth in the private business sector from 1948 to 1966 to have been 2.9%, but only 1.4% from 1966 to 1976. See Link and Siegel (2003) for an overview of the productivity slowdown literature.

The University and Small Business Patent Procedures Act of 1980, also known as the Bayh–Dole Act, reformed federal patent policy by providing incentives for the diffusion of federally funded innovation results. Universities, in particular, were permitted to obtain titles to innovations developed with government funds. See Stevens (2004) for an historical account of the passage of the Bayh–Dole Act.

The Economic Recovery and Tax Act of 1981 (ERTA) included a tax credit for qualified R&E expenditures in excess of the average amount spent in previous years. See Atkinson (2007) and Tassey (2007) for an overview of the past effectiveness and current state of the R&E tax credit.

According to Birch (1979, p. 29): “On average about 60% of all jobs in the USA are generated by firms with 20 or fewer employees, and about 50% of all jobs are created by independent, small entrepreneurs. Large firms (those with over 500 employees) generate less than 15% of all net new jobs.”

In subsequent research, Birch (1987) found that small businesses with fewer than 20 employees accounted for 88% of all net new jobs over the 1981–1985 time period. Also, according to Birch (1981, p. 7), “Smaller businesses more than offset their higher failure rates with their capacity to start up and expand rapidly.”

Some disagree with Birch’s analyses and the findings of others who reached similar conclusions from analyses of firm and aggregated data. See, in particular, Davis et al. (1996).

According to Klein (1979), the slowdown in productivity growth in the 1970s can be viewed in terms of a decline in businessmen’s ability, or perhaps their desire, to deal with disequilibria.

Innovation and entrepreneurship are closely related concepts. For Schumpeter (1934), the entrepreneur was the persona causa of economic development. The process of “creative destruction” is the essence of economic development and growth, he wrote (1950); it is a process defined by the carrying out of new combinations in production. An underlying hypothesis in the examination of employment growth in this paper is that the Schumpeterian process of creative destruction is instigated in substantial part by small, entrepreneurial firms. The impact of the SBIR program on the process is explored.

Relatedly, the National Science Foundation and the U.S. Small Business Administration released the Gellman Report in 1976. It showed that small firms over the 1953–1973 period were more innovative per employee than larger firms.

“[I]t is the exchange of complementary knowledge across diverse firms and economic agents that yields an important return on new economic knowledge” (Thurik 2009, p. 10).

See Link and Tassey (1987) for documentation of this shift.

This section draws on Link and Scott (2000), Audretsch et al. (2002), National Research Council (2004), and Wessner (2000, 2008). For a taxonomy of public/private partnerships, see Link (2006).

The 1982 Act amended the Small Business Act of 1953 (P.L. 85-536) which established the Small Business Administration.

SBIR is a set-aside program; it redirects existing R&D funds for competitive awards to small businesses rather than appropriating new monies for R&D. The 1982 Act allowed for this percentage to increase over time.

As stated in the 1982 Act, to be eligible for an SBIR award, the small business must be: (1) independently owned and operated; (2) other than the dominant firm in the field in which it is proposing to carry out SBIR projects; (3) organized and operated for profit; (4) the employer of 500 or fewer employees, including employees of subsidiaries and affiliates; (5) the primary source of employment for the project’s principal investigator at the time of the award and during the period when the research is conducted; (6) at least 51% owned by U.S. citizens or lawfully admitted permanent resident aliens.

The $50,000 amount and the $500,000 amount below are not stated in the 1982 Act. They originally came from a Policy Directive to the Small Business Act of 1953, to which the 1982 Act was an amendment. Both amounts are referenced explicitly in U.S. Senate Report 110–447 (2008).

“The objective of Phase I is to determine the scientific or technical feasibility and commercial merit of the proposed research or R&D efforts and the quality of performance of the small business concern, prior to providing further Federal support in Phase II.” See: http://​grants.​nih.​gov/​grants/​funding/​SBIRContract/​PHS2008-1.​pdf, p. 1.

“The objective of Phase II is to continue the research or R&D efforts initiated in Phase I. Funding shall be based on the results of Phase I and the scientific and technical merit and commercial potential of the Phase II proposal.” See: http://​grants.​nih.​gov/​grants/​funding/​SBIRContract/​PHS2008-1.​pdf, p. 1.

“The objective of Phase III, where appropriate, is for the small business concern to pursue with non-SBIR funds the commercialization objectives resulting from the outcomes of the research or R&D funded in Phases I and II.” See: http://​grants.​nih.​gov/​grants/​funding/​SBIRContract/​PHS2008-1.​pdf, p. 1.

The percentage increased to 1.50 in 1993 and 1994, 2.00 in 1995, and 2.50 in 1997.

The reauthorization also stated that there should be an “adjustment of such amounts once every 5 years to reflect economic adjustments and economic considerations.”

It is not uncommon for Phase II awards to exceed the $750,000 threshold.

On October 7, 2009 the House and Senate Armed Services Committees recommended that the Department of Defense (DoD) SBIR program be reauthorized for 1 year, until September 30, 2010. This recommendation became part of the DoD 2010 Authorization.

See Congressional Research Service (2009, 2010) for a detailed overview of the reauthorization efforts.

We thank Dr. Charles Wessner of the NRC for making relevant data available to us for this study.

It was assumed as part of the sampling methodology that Phase II awards made in 2001 would be completed by 2005.

In the absence of a non-SBIR-funded sample of projects, we are unable to say from the data in Table 3 that SBIR-funded projects retain more or fewer employees than comparable non-SBIR-funded projects.

We are looking at the sample of projects for descriptive patterns in the data for the respondents and do not control for selection in the estimation of Eq. 2. Retained employees are essentially a commercialization decision. and elsewhere (Link and Scott 2009, 2010) we have shown that the commercialization decision is uncorrelated with selection for all but one of the five agencies’ samples of projects. For that one agency’s sample of projects, the error in the model of response is correlated with the error in the equation for the commercialization model. The effect of selection in the error of the commercialization equation is then of course important for the prediction of commercialization, conditional on the response or the population of that one agency. However, even in that one case, with or without control for selection, the estimates of the coefficients on the explanatory variables are essentially the same. It is those coefficients that are important in this paper; therefore, we use the simpler specification rather than the more complex simultaneous estimation of the negative binomial model for retainees and the probit model for selection into the sample by response.

To describe marginal effects, we use the incident rate ratio because it places the effect in the perspective of the outcome without the effect. From Eq. 2, holding constant all variables except x _i , given a change in x _i , the expected number of retainees relative to the expected number without that change is the base for the natural logarithms raised to the power of the product of the coefficient for x _i and the change in x _i . For a unit change in x _i , the expression is called the incident rate ratio and is simply the base for the natural logarithms raised to the power equal to the coefficient for variable x _i .

The probability of commercialization among the selected projects is actually somewhat higher for the NIH projects than for the DoD projects. See Link and Scott (2010, Table 11, p. 600).

Importantly, as we emphasize in the following discussion, generating retained employees is not a stated goal of the SBIR program, so an award that does not result in retained employees has not thereby failed to meet the SBIR program’s goals.

Such spillover effects are consistent with the findings from case studies of SBIR awards (Link and Scott 2000).