Despite the growing public and private investment in ... - SSRN papers

4 downloads 0 Views 234KB Size Report
two-stage net present value model to identify four effects from initial public investment on the private decision for follow-on investment. Our empirical analysis ...
How does initial public financing influence private incentives for follow-on investment in early-stage technologies?

Corresponding Author Andrew A Toole Department of Agriculture, Food, and Resource Economics Rutgers University 55 Dudley Road New Brunswick, NJ 08901 [email protected]

Calum Turvey Department of Applied Economics and Management Cornell University 356 Warren Hall Ithaca, NY 14853 [email protected]

ABSTRACT

One common rationale supporting public financing programs for small firms is that initial public investment creates incentives for follow-on private investment. However, there does not appear to be a unified statement in the literature describing how initial public investment creates incentives for follow-on private investment. Focusing on external private investors, we use a two-stage net present value model to identify four effects from initial public investment on the private decision for follow-on investment. Our empirical analysis uses a sample of non-venture backed firms entering the SBIR program to examine how reduced risk, the number of SBIR awards, and size of initial public investment influence the likelihood of follow-on venture capital investment. We find the probability of follow-on venture capital investment is more likely when firms reach Phase II of the program, is less likely as firms win multiple Phase I and Phase II awards, and is more likely as the size of initial public investment in Phase I increases.

2

1 Introduction One of the rationales supporting public financing programs for small firms is that initial public investment creates incentives for follow-on private investment. For example, as part of the congressional discussion leading to the enactment of the U.S. Small Business Innovation Research (SBIR) Program, policymakers clearly stated the desire to attract follow-on private investment from investors like venture capitalists for commercialization of early-stage technologies (Archibald and Finifter 2003). While this rationale has a long history in policy formulation, there does not appear to be a unified statement in the economics or finance literatures describing how initial public investment creates incentives for follow-on private investment. This is unfortunate since understanding these incentives is critical for gauging the impact of public programs on private investment behavior and, ultimately, for designing public policies that foster commercialization. In this paper we focus on how initial public investment affects the incentives for followon private investment by other external financing agents such as venture capitalists. 1 To shed light on how these private investors change their investment assessments, we construct a twostage net present value model for an early-stage technology. This model is based on the real options paradigm which argues that delays in financing, and hence the project go-ahead, can be tied to underlying technical and market uncertainties.

Our model suggests that initial

government investment serves to underwrite the risk of an early-stage technology thereby making follow-on financing by venture capitalists and others more attractive on a reduced risk and higher reward basis. In section 3, we highlight four effects from initial public investment that influence private incentives for follow-on financing. 1

David, Hall, and Toole (2000) review the literature examining whether publicly financed R&D complements or substitutes for a firm’s own private R&D investment, but they do not consider how external financing agents respond.

3

To complement the theoretical discussion, we also undertake an empirical analysis looking at the probability that firms receive follow-on venture capital investment after advancing their technologies in the SBIR program. 2 We find the probability of follow-on venture capital investment is more likely when firms reach Phase II of the program. This is consistent with venture capital investors responding to reduced risk following SBIR participation by early-stage technology firms. Given our data, however, we cannot tell whether this is due to real options investment behavior by venture capitalists or simply a traditional response to reduced risk. Consistent with Lerner’s (1999) discussion of SBIR certification, as the number of Phase I and Phase II awards to a firm increases, the likelihood of receiving follow-on venture capital investment falls. This reflects the diminishing marginal value of subsequent awards as signals to venture capitalists. We also find the probability of follow-on venture capital investment is more likely as the size of initial public investment in Phase I increases. This is consistent with the idea that initial public investment leverages follow-on private investment. While suggestive, these results should be viewed as exploratory rather than conclusive. Conclusive statistical tests require richer data to adjust for potential selection bias and to increase the number of controls for firm characteristics. The next section provides a short overview of the relevant literature on small firm financing of early-stage technologies. Section 3 presents the simple two-stage net present value model to shed light on how initial public investment influences the incentives for follow-on private investment. Section 4 describes the SBIR and venture capital data, hypotheses, and the empirical model. The final section contains concluding remarks.

2

The results from the theoretical model are not specific to venture capital investors, but apply to all external investors. See de Bettignies and Brander (2007) for an analysis of the entrepreneur’s choice between bank financing and venture capital financing.

4

2 External Financing for Early-stage Technologies Small firms seeking to commercialize new technologies often face the challenge of securing external financing to support product development. An important strand of the finance literature suggests these challenges stem from capital market imperfections such as asymmetric information and moral hazard (Hubbard 1998). Asymmetric information arises when the quality or riskiness of a project cannot be fully communicated or understood by a potential investor. Moral hazard refers to the incentive and monitoring problems once financing is received. In both cases, the cost of external funds will be higher or investors may simply refuse to finance the project. Hall (2006) discusses these problems within the context of financing research and development (R&D) investment. For small R&D performing firms, asymmetric information and moral hazard problems are more severe. When seeking R&D financing for early-stage technologies, firms face an even greater chance that external private investors will refuse to invest immediately since information asymmetries increase the option value of waiting. According to the real options theory of investment under uncertainty, investors delay irreversible investment as the degree of uncertainty increases (Dixit 1992, Pindyck 1991, Dixit and Pindyck 1994). While technical and market uncertainties are already high with early-stage technologies, imperfect information exacerbates these uncertainties.

This causes external investors to delay their investment until new

information arrives.

If firms are seeking external financing because internal funds are

insufficient for project go-ahead, this “financing gap” will hinder the commercialization of potentially valuable technologies. 3

3

Branscomb and Auerswald (2001) include a “financial gap” in their discussion of the challenges to crossing the “Valley of Death” between invention and innovation.

5

3 The Potential Role of Initial Public Investment In principle, public investment may serve a “bridging” function to facilitate external private investment in early-stage technologies. Lerner (1999), the key background paper for this analysis, suggests that public subsidy programs may be valuable if they “certify” small firms for follow-on private investment, particularly from venture capitalists.4 The real options paradigm discussed above helps to clarify one mechanism by which participation in a public subsidy program may create incentives for follow-on private investment.

In particular, if public

subsidies are used to support the research necessary to reduce technical and/or market uncertainties, private investors will find these opportunities more attractive since this initial investment reduces their option value of waiting. To help clarify the potential impact of a first stage public investment, consider the following net present value (NPV) model where investment occurs in two-stages. Start-up investment is K0 and follow-on funding at time T is given by KT. The investment NPV is risky with a probability of success θ(K0), which depends on K0, and assume θ’(K0)≥0 (increases in start-up funding increase the probability of success). If θ(K0) >0 then a second stage investment is made by external private investors and, if the investment is successful, it will return a present value of V(K0,KT). When discounted at the rate r, the NPV of this problem is given by

(1)

NPV = − K 0 + θ ( K 0 )[VT ( K 0 , K T ) − K T ]e − rT

Taking the derivative of the NPV with respect to investment yields

4

Feldman and Kelly (2003) find a similar effect in their analysis of the Advanced Technology Program. They call this the “halo effect.” Sine et al. (2003) show a “halo effect” from institutional prestige increases the licensing rate of university inventions.

6

(2)

⎡ ∂V () ∂KT ∂VT () ⎤ ⎤ − rT ∂NPV ⎡ ∂θ VT () + θ ⎢ T =⎢ + ⎥⎥ e −1 ≥ 0 K K K K ∂K 0 ∂ ∂ ∂ ∂ T 0 0 0 ⎣ ⎦⎦ ⎣

and

(3)

⎡ ∂V () ⎤ ∂NPV = θ ⎢ T − 1⎥ e − rT ≥ 0 ∂KT ⎣ ∂KT ⎦

where (2) represents the change in expected NPV given the initial funding and (3) represents the change in NPV with respect to follow-on funding. Of course, there is no requirement that initial and follow-on funding come from the same source. Suppose that K0 is public funding from a program targeted at small technologically innovative companies. The first term in (2) represents the change in the probability of success due to the public startup funds. If the initial phase is unsuccessful, perhaps due to the unanticipated failure of necessary technical requirements, this initial capital is lost. But as written the purpose of the initial investment is to increase the probability of success by reducing these uncertainties. The second term of (2) has two parts. The first part captures the cross-effects between public start-up and private follow-on funding, assuming the two are related.

If this cross-effect is positive, then public investment gets

“leveraged” by follow-on private investment. 5 Of course, if the two are not related, then there is no leveraging effect from public funding of research since follow-on funding is independent of initial funding.

The second term represents the increase in the value of the early-stage

technology as a result of public investment, a social return. This portion of the present value is 5

The possibility of the cross-effect being negative is not sensible in this model since initial public investment keeps the project alive. It would indicate that initial investment actually reduces the “effectiveness” of follow-on investment. In that case, no rational investor would provide follow-on financing.

7

split between the firm and follow-on private investors depending on bargaining power. The marginal private return is given in (3). Note that in (3) there is no transfer of wealth from the private investors back to the government. In this simple model, the term θ(K0) in (1) incorporates the influence of both technical and market uncertainties on the decision to invest.

For early-stage technologies, private

investors see θ(K0) as too low to justify private start-up investment today. Since

∂θ ≥ 0 , as ∂K 0

seen in (2), there is a potentially valuable role for public start-up investment. Initial public investment keeps the project alive by reducing the probability of failure. As uncertainty is reduced, follow-on private investment increases with an increased probability. This relationship is governed by θ

∂VT () ∂KT in equation (2) which simply states that the relationship between ∂KT ∂K 0

public and private investment is governed by the probability of success. The probability of success is in turn governed by public investment. Another related point examines the timing of investment in relation to its underlying risk. Reference has previously been made to the real options model which in some forms holds that increasing uncertainty about future probabilities can delay investment. From time to time it may not be in the public interest, for reasons of health or international competition, to delay commercialization, so it may be in the public interest to accelerate development. A slight modification of the calculus shows how this may be done in a public setting.

(4)

dNPV ∂NPV ∂T ∂T >0 = = −r ⎡⎣θ ( K 0 ) [VT ( K1 , KT ) − KT ] e− rT ⎤⎦ dθ ∂T ∂θ ∂θ

8

In (4) the change in the NPV with respect to risk is enhanced by the public objective that

∂T < 0 . In other words, there exists an overt policy to invest public funds immediately to reduce ∂θ risk sufficiently to encourage follow-on financing and hence move the project closer to commercialization. This has the additional benefit of increasing the NPV of the project, which in turn makes it more attractive to external investors. It is in fact an explicit statement that the initial public investment in the early stage is in part a means to purchase, albeit indirectly, a portion of the option to wait by venture capitalists and other follow-on financing agents. Given the potentially valuable role of initial public investment, an obvious question is: why should government undertake this investment in the first place? There are two economic justifications for a public role in early-stage investment. The first justification is based on positive externalities or knowledge spillovers resulting from the commercialization of the technology (Nelson 1959, Arrow 1962). Public agencies should provide subsidies for projects where the private return is insufficient to induce investment but the social return exceeds the cost of investment.

Griliches (1992) reviews the R&D literature studying agriculture and

manufacturing. He finds the social returns to R&D investment are substantially higher than private returns. Public investors, however, must judge whether the indirect or intangible benefits from the investment, in whatever form it takes including health or employment, exceeds the cost of public investment. This political decision will ultimately require a judgment that the direct marginal social benefits from an investment of K0 be at least as great as the left hand side of (2). The second justification is based on capital market imperfections as described in Section 2 above. In sum, the model highlights four effects from initial public investment that influence the private decision to undertake follow-on investment. First, public investment resolves some of

9

the technical and market uncertainties.

Second, for projects that remain feasible, public

investment directly increases the NPV of the investment opportunity, a social return. Third, to the extent that public and private investments are complementary, private follow-on investors increase their returns by leveraging their funding on the initial public investment. Fourth, public investment may speed up the commercialization process. In the empirical section, we use data from the SBIR program to explore the first and third of these effects. Unfortunately, the data are not rich enough to observe the social return from SBIR investment or the extent to which SBIR investment accelerates the commercialization process. 6

4 Empirical Setting: SBIR Program and Follow-on Venture Capital In the remainder of the paper, we empirically examine how reduced uncertainties surrounding an early-stage technology, the number of SBIR awards, and the magnitude of initial public investment influence the likelihood of follow-on private investment by venture capitalists. We start with a sample of non-venture backed firms that have sought and successfully obtained initial public investment from the SBIR program. Next, we identify which of these firms received follow-on venture capital (VC) investment after their first SBIR award. Using a decision-probability model, we relate an indicator of successful follow-on VC investment, VCafter, to SBIR program participation characteristics, available firm characteristics, as well as SBIR agency and geographic control variables. We focus on SBIR participating firms for two reasons. First, the SBIR program is the largest commercialization program supporting technology development in small U.S. firms. According to the Small Business Administration, the program awarded $8.6 billion in direct 6

Audretsch et al. (2002) examine the social return to forty-four Department of Defense SBIR awards and find the expected social rate of return to be at least 84 percent. Laidlaw (1998) presents evidence that the Advanced Technology Program awards accelerated private development and commercialization.

10

subsidies between 1983 and 1996. Beginning in 1997, annual awards across all agencies exceeded $1 billion and a recent figure from the National Research Council estimates the total value of awards made in 2003 to be over $1.6 billion (NRC 2004). Second, the structure of the program allows us to construct a proxy for the reduction of uncertainties surrounding an earlystage technology. The original SBIR legislation, the Small Business Innovation Development Act of 1982, established a three phase structure for the program. All applicants must start with a Phase I proposal. The Phase I project is intended to test the feasibility of a new idea. Phase I projects focus primarily on establishing the technical feasibility of an early-stage technology. It lasts from six to twelve months and awards can be up to $100,000. If the results of the feasibility study are favorable, firms may apply for a Phase II grant to move their idea into product development. The Phase II award is up to $750,000 and lasts for a two-year period. Finally, there is a Phase III to the SBIR program. This is an un-funded phase in which the companies are expected to commercialize their product (or process) and obtain follow-on private investment. 7 This multiphase structure permits us to empirically examine the effect of initial public investment aimed at reducing technical and/or market uncertainties on follow-on private venture capital investment. By design, a Phase I project focuses on resolving uncertainties related to the technology. These projects require the successful completion of specific technical objectives, but also allow firms time to reduce uncertainties about the potential market for their innovations. When this phase is successful, firms may apply for a Phase II award. The receipt of the Phase II award indicates that some the uncertainties surrounding the early-stage technology have been resolved, at least in the judgment of the SBIR program administrators. The receipt of a Phase II award serves as our proxy for reduced uncertainty. 7

See Wallsten (1998) and Audretsch (2003) for additional discussion of the SBIR program.

11

Firm participation in the SBIR program is not random but reflects an agency specific selection process. Firms must be eligible, submit a proposal, and be selected through a review process. Each federal agency with an annual extramural R&D budget of more than U.S. $10 billion is required to implement the SBIR program. 8 The agencies are given significant latitude for determining the topics, form of financing (grant, contract, cooperative agreement), proposal format, and evaluation criteria. 9 As described below, our sample covers the population of firms winning at least one Phase I award from any participating agency. This means that all of our findings are conditional SBIR participation and this limits our ability to generalize the empirical results (in a statistically legitimate way) to the larger underlying population of small technology-intensive firms. However, we do not ignore agency-specific selection. Except for the Department of Defense (DOD), the empirical analysis includes a dummy variable for each federal agency. For the DOD, we include eleven dummy variables, one for each of its separate funding components. This methodology allows follow-on venture capital investment to differ across agencies due to the agency selection of firms. For instance, venture capitalists may view the choice of firms and their associated technologies differently when considering follow-on investment for a DOD SBIR firm versus a National Institutes of Health SBIR firm. 10,11 Perhaps more important than the SBIR agency selection process is the selection process used by venture capitalists and firms seeking external capital investment. As stated above, all of the firms considered in this paper were not venture backed before entering the SBIR program. 8

There are currently eleven federal agencies implementing the SBIR program. Further detail is available in the U.S. Small Business Administration’s Handbook for SBIR Proposal Preparation, available at http://www.sba.gov/gopher/Innovation-And-Research/SBIR-Pro-Prep/. 10 The agency dummy variables also account for the situation in which firms win awards from multiple agencies. The U.S. National Institutes of Health is dominant part of the U.S. Department of Health and Human Services. 11 Data on SBIR applicants are not systematically available. Applicant data would allow more sophisticated statistical methods to be used to account for agency selection into the SBIR program. 9

12

These firms may not be venture backed for essentially four reasons. First, the firm had access to venture capital investors but failed to pass through the VC screening process. 12 Second, the firm did not have access to venture capital investors. Third, the firm had access to VC investors but chose to wait before beginning the screening process. Fourth, the firm did not want venture capital investment. Since these alternatives are not observable in the data, it is important to sort out how these VC and firm selection choices influence identification of real options investment behavior and SBIR certification. Recall that real options investment behavior takes place when investors delay current investment in a project to wait until uncertainties are sufficiently resolved before investing. To provide compelling evidence for this type of investment behavior, we would need to observe the following investment sequence: firms enter and fail the VC screening process due to excessive uncertainties, firms enter the SBIR program which allows them to reduce technical and/or market uncertainties surrounding their early-stage technologies, firms re-enter and pass the VC screening process. 13 This corresponds with the first reason given in the previous paragraph for not already having venture capital backing prior to SBIR. Given our data, we cannot directly test for real options behavior on the part of venture capitalists. However, the results we present in the next section are consistent with this type of investment behavior. The concept of SBIR certification involves the flow of new information to venture capitalists after firms participate in the SBIR program. For firms desiring venture capital, this could take two forms. First, for those firms with no prior access to VCs (the second reason given

12

See Fried and Hisrich (1994), Kaplan and Stromberg (1999, 2000) for a description of the venture capital screening process. 13 This sequence assumes all venture capitalists use a homogeneous screening process over time. While not entirely realistic, evidence presented by Fried and Hisrich (1994) suggests it may not be unreasonable. It also assumes the nature of the uncertainties causing venture capital investors to delay investment can be addressed by participating in the SBIR program.

13

above), SBIR participation may alert the venture capitalists to the existence of the firm. Getting noticed by venture capitalists may facilitate the opportunity to enter the VC screening process and obtain investment. Second, as emphasized by Lerner (1999), the information flow may contain a quality or credibility component. This possibility applies to those firms which had prior access to venture capitalists but decided to go into the SBIR program first in order to better prepare for the VC screening process (the third reason given above). 14 Both of these forms of SBIR certification, individually or together, suggest that non-venture backed firms entering the SBIR program will have an increased likelihood of VC investment. The “get noticed” form of SBIR certification suggests the likelihood of follow-on venture capital investment should be higher after winning a Phase I award, but not after a Phase II award. 15 In the case where firms decide to wait until after wining a Phase II award before approaching venture capital investors, SBIR certification works through reduced risk stemming from further development of their technologies. This suggests the likelihood of VC investment will increase after a Phase II award. Real options behavior by VCs suggests winning a Phase II award decreases the value of waiting through reduced uncertainties. This also suggests the likelihood of follow-on venture capital will increase after a Phase II award. Based on the NPV model presented in section 3 we expect the reduced risk effects to dominate:

14

The presence of firms that never wanted VC investment (reason four) in the sample is not a cause for concern since their investor preferences should not be systematically related to their own firm’s success in the SBIR program such as winning a Phase II award. 15 Observing this in the data depends on efficient information communication to the venture capital community once a Phase I award is obtained. Researchers studying the venture capital decision making process note that some venture capitalists “aggressively seek out deals” (Fried and Hisrich 1994, p 32). Moreover, the SBA Handbook states, “…some SBIR program managers send abstracts of Phase I awardees to large companies and venture capitalists…” (SBA 2007).

14

Hypothesis #1:

Winning an SBIR Phase II award will increase a firm’s probability of

follow-on venture capital investment since SBIR participation has reduced the technical and/or market uncertainties surrounding their early-stage technology.

In prior work on this topic, Lerner points out that SBIR certification also has implications about the effect of multiple SBIR awards received by a firm. In particular, the information content in the SBIR certification signal should be contained in the first SBIR awards. He notes the incremental value of additional awards will be “declining sharply” since each award provides less and less certification information (Lerner 1999, p. 312). This suggests the following hypothesis:

Hypothesis #2:

A firm’s probability of follow-on venture capital investment will fall as

the number of SBIR Phase I and Phase II awards increases.

In addition to reduced risk and SBIR certification, the model in section 3 suggests follow-on private investment will also be more likely as the magnitude of initial public investment increases. A larger initial public investment allows greater leveraging of private funds by follow-on investors.

Hypothesis #3:

The probability of follow-on venture capital investment will increase with the

size of the SBIR investment since this improves the leveraging effect of private funds.

15

We examine these three hypotheses using a database constructed from two sources of information. The primary source is the U.S. Small Business Administration (SBA). SBA is the coordinating agency for the SBIR program and the public data we use covers the 1983 to 1999 period for all eleven participating agencies. For each SBIR participating firm, these data provide the firm name, street address, city, state, SBIR phase, year of award, awarding agency, award amount, topic, and indicators for minority or woman owned. To link initial public investment to follow-on private investment, we use Securities Data Corporation’s (SDC) VentureXpert database (1977-2005) to identify which SBIR firms received venture capital and the date of their first round. Our database is composed of a cross-section of firms that received at least one SBIR award over the period 1983-1999 from any one of the participating U.S. agencies. We have information on 10,914 SBIR firms over this period compiled from the SBA. Cross referencing with the VentureXpert database identified 1,035 SBIR participant firms that ever received venture capital investment in this period, either before or after the firm’s first SBIR award. The dependent variable in our analysis is an indicator that takes the value 1 if the SBIR firm received its first infusion of venture capital after it received its first SBIR Phase I award and 0 otherwise. There are firms that receive both SBIR and venture capital in the same year. We code the VC_after indicator as 1 when the venture capital investment was received after June 1st of that year since the SBIR proposal review process takes about six months. 16 There are 374 SBIR firms that received VC after SBIR.

16

Given that our data are annual, we do not observe the exact dates of regarding when the firms submit their SBIR proposals. Further, to increase the confidence that follow-on venture capital investment is associated with SBIR participation, we restricted the lag between first SBIR award and follow-on VC to be less than nine years. It is also important to drop any SBIR awards won by firms after receiving venture capital since this could reflect the influence of venture capital investors.

16

Table 1 presents descriptive statistics for our variables. 3.4% of SBIR participants received follow-on venture capital investment. Lerner (1999) found 2.5% of his sample had received follow-on venture capital. Bhidé (2000) notes that only 4% of the companies on the Inc. 500 lists from 1982 to 1989 were venture backed. Overall, the small percentage reflects the uniqueness of venture capital investment. Nevertheless, venture capital investment remains an important source of follow-on private investment. 17 Similar to Lerner (1999) and consistent with the model of section 3, we analyze the probability of follow-on venture capital investment using a Probit model. We estimate the following reduced form model for our cross-section of SBIR participants:

Pr(VCafter = 1 | X ) = Φ [ β 0 + β 1( PII _ Dum ) + β 2 ( PhaseI $) + β 3( PhaseII $) + β 4 ( Num _ PI ) + β 5( Num _ PII ) + β 6( Minority ) 33

19

j =1

k =1

+ β 7 (Woman ) + β 8( Biotech ) + ∑ αj (US _ State ) + ∑ δk ( SBIR _ Agency )

where PII_Dum is a dummy variable that indicates the SBIR firm won at least one SBIR Phase II award in the sample period. If initial public investment resolved some of the technical and/or market uncertainties, then we expect β1 to be positive as stated in hypothesis #1. PhaseI$ and PhaseII$ are the cumulative values of Phase I and Phase II SBIR investment into the firm (adjusted for inflation), respectively. If the leveraging effect is important, we expect β2 and β3 to be positive as stated in hypothesis #3. Num_PI and Num_PII are counts of the number of Phase I

17

While we do not focus on the “downstream” performance of firms receiving follow-on VC investment, Hsu (2006) explored this and found these firms perform better in terms of cooperative commercialization agreements and initial public offerings than SBIR firms without follow-on VC investment.

17

and Phase II awards won by the firm. Following Lerner (1999), we expect both β4 and β5 to be negative reflecting the diminishing marginal value to venture capitalists of the SBIR signal. We include the other variables as controls. The minority and woman owned indicators may capture any differences in VC investment by firm ownership. To help control for the popularity of venture capital investment into biotechnology, we constructed a dummy, biotech, identifying those firms with project awards using biotechnologies. We also include U.S. state dummies to account for the regional concentration of venture capital investment. Table 2 shows the concentration of venture capital after SBIR across states (including the District of Columbia and Puerto Rico). Only 33 states show any VC investment into SBIR participating firms with Minnesota having the highest percentage of VC after SBIR. California, not surprisingly, has the highest absolute number of such investments. We also include SBIR agency dummies to account for differences in agency technology focus and selection. Table 3 shows the concentration of venture capital after SBIR across agencies. The Department of Health and Human Services, which houses the National Institutes of Health, has the greatest absolute total and percentage. Note the column totals in Tables 2 and 3 do not add to the firm total given above since firms may have locations in more than one state or win awards from multiple agencies.

5 Empirical Results Table 4 reports our Probit results using the regression coefficients. Model A in Table 4 includes the standard SBIR program participation characteristics. According to the leveraging hypothesis #3, the probability of follow-on venture capital investment should increase with the total initial public investment. The results show a positive and significant effect of total Phase I

18

dollars and a positive but insignificant effect of total Phase II dollars. The result for Phase I investment supports the leveraging effect; however, venture capitalists do not appear to be leveraging their investments on Phase II monies. If venture capitalists are investing early in the Phase II award period, Phase II dollars will not be as important for leveraging as Phase I dollars. Following Lerner (1999), hypothesis #2 suggests the SBIR certification signal will diminish as firms win multiple Phase I and Phase II awards. The results are consistent with this hypothesis. The coefficients on the total number of Phase I and Phase II awards are negative and significant. This indicates the probability a firm will receive follow-on venture capital falls as it wins a greater number of each type of award. Model B in Table 4 introduces the Phase II dummy variable indicating if the firm ever won a Phase II award. Since progressing into Phase II shows the firm has reduced at least some of the uncertainties surrounding its early-stage technology, hypothesis #1 holds that the probability of venture capital investment should increase as VCs view lower risk technologies more favorably. The Phase II dummy is positive and significant and supports hypothesis #1. Except for the coefficient on the number of Phase II awards, the other variables for program characteristics do not change much. The coefficient on the total number of Phase II awards becomes more negative relative to model A. It is now clear that the negative effect of winning multiple Phase II awards found in model A swamped the positive effect of ever progressing into Phase II. 18 As discussed in section 4, this finding is consistent with real options behavior by venture capital investors but is also consistent with traditional investment theory predicting increased investment as risks are decreased. Without knowing whether firms chose to wait before approaching VCs, we cannot differentiate between these two alternatives. Our results are 18

Of course, interpreting the Phase II dummy in the opposite direction indicates firms which never progress out of Phase I have significantly lower chances for follow-on VC investment. This makes sense because these firms never successfully reduce the technical and/or market uncertainties surrounding their new technologies.

19

consistent with prior work by Lerner (1999), who studied follow-on venture capital in the early years of the SBIR program, and Toole and Czarnitzki (2007), who only analyze the SBIR program administered by the National Institutes of Health. Adding our remaining control variables in models C and D in Table 4 does not change our findings. Model C indicates the probability of follow-on venture capital investment is greater if the firm uses biotechnologies. It also shows that woman-owned firms are less likely to receive follow-on VC investment. In model D, the SBIR agency dummies and U.S. state dummies are included. Both groups of variables are jointly significant as indicated by the chisquare statistic at the bottom of Table 4.

6 Conclusion One of the rationales underlying public financing programs for small firms is that initial public investment creates incentives for follow-on private investment. Focusing on external private investors, we present a two-stage net present value model which suggests four effects influencing their investment decision. First, public investment resolves some of the technical and market uncertainties. Second, for projects that remain feasible, public investment directly increases the NPV of the investment opportunity, a social return. Third, to the extent that public and private investments are complementary, private follow-on investors increase their returns by leveraging their funding on the initial public investment. Fourth, public investment may speed up the commercialization process. Using a sample of non-venture backed firms entering the SBIR program, we empirically examine how reduced risk, the number of SBIR awards, and size of initial public investment influence the likelihood of follow-on venture capital investment. We find the probability of

20

follow-on venture capital investment is more likely when firms reach Phase II of the program (reduced risk), is less likely as firms win multiple Phase I and Phase II awards (declining signal value), and is more likely as the size of initial public investment in Phase I increases (leveraging). Our findings have several policy implications. First, there is a sound basis for the policy argument that public investment creates incentives for follow-on private investment by other external financing agents. However, it is important to keep in mind that we have not analyzed the efficacy of the policymaker’s choice about which early-stage technologies to support. Initial public investment must still be justified by the expectation of knowledge spillovers or as an effort address capital market imperfections. When follow-on investment does take place, there is some research suggesting these firms perform better. Hsu (2006) presents evidence that SBIR firms receiving follow-on venture capital investment are more likely to engage in cooperative commercialization activities and reach initial public offering than those SBIR firms without follow-on VC. Second, our results on SBIR certification suggest that “getting noticed” is not sufficient for follow-on venture capital investment. Progressing into Phase II of the program is the important step. Third, when our results are interpreted in the context of the existing SBIR research, there is some indication that the legislative limit on Phase I SBIR funds may be too low. Previous research suggests the amount of Phase I investment increases the likelihood firms will progress into Phase II (Toole and Czarnitzki 2007). Our results show the amount of Phase I investment has a significant leveraging effect on follow-on venture capital. If the investment criteria used by other external financing agents are similar to those of venture capitalists, increasing the Phase I limit may improve commercialization outcomes. At the very least, the legislative limit on Phase I SBIR funds deserves further research.

21

Our empirical analysis is subject to a number of caveats. Given our data, we were unable to implement formal statistical methods to adjust for potential selection issues related to SBIR program participation and to account for the timing of choices made by venture capitalists and firm managers. Not adjusting for SBIR program selection means our results do not generalize in a statistically legitimate way to the larger population of small technology-intensive firms in the U.S. Knowing the timing of choices by VCs and firm managers is necessary if one wishes to provide compelling evidence for real options investment behavior by VCs. Advancing our research in that direction must wait until proprietary data become available. Finally, richer data on firm characteristics would increase confidence that our results are not driven by some omitted characteristic. Since firms participating in the SBIR program are primarily small and private, systematic data on firm characteristics are difficult to collect even when firms are willing to divulge their private information. We believe including more accessible measures, such as firm size and age, would not change our qualitative results.

22

TABLE 1 Descriptive Statistics SBIR Firms (N=10,914) Variable

Mean

Std. Dev.

Min

Max

VCafter

0.034

0.181

0

1

PII_Dum

0.483

0.500

0

1

0.286

0.707

0

19.695

(mil., 2000 base yr)

0.738

2.164

0

71.938

Num_PI

3.484

8.536

1

241

Num_PII

1.226

3.418

0

111

Minority

0.127

0.333

0

1

Woman

0.114

0.318

0

1

Biotech

0.024

0.152

0

1

PhaseI$

(mil., 2000 base yr) PhaseII$

23

TABLE 2: Firms Receiving Follow-on Venture Capital Investment By US State*# US States VC After No VC After Total Firms % VC After Minnesota 14 166 180 7.8% North Carolina 12 170 182 6.6% Connecticut 11 179 190 5.8% Wisconsin 7 116 123 5.7% Rhode Island 3 51 54 5.6% Massachusetts 50 916 966 5.2% Washington 14 292 306 4.6% Oregon 6 127 133 4.5% California 92 2,071 2163 4.3% Utah 6 147 153 3.9% New York 23 590 613 3.8% Pennsylvania 17 438 455 3.7% Alabama 6 157 163 3.7% Illinois 10 264 274 3.6% Colorado 13 347 360 3.6% Nebraska 1 27 28 3.6% Maryland 21 618 639 3.3% Texas 15 449 464 3.2% New Mexico 6 180 186 3.2% New Jersey 11 342 353 3.1% Idaho 1 34 35 2.9% Florida 10 348 358 2.8% Michigan 7 257 264 2.7% Virginia 16 599 615 2.6% Missouri 2 78 80 2.5% Iowa 1 41 42 2.4% Tennessee 3 132 135 2.2% New Hampshire 2 90 92 2.2% Indiana 2 96 98 2.0% Oklahoma 1 52 53 1.9% Arizona 3 166 169 1.8% Georgia 2 124 126 1.6% Ohio 2 411 413 0.5% Alaska 0 14 14 0.0% Arkansas 0 20 20 0.0% District of Columbia 0 76 76 0.0% Delaware 0 26 26 0.0% Hawaii 0 37 37 0.0% Kansas 0 58 58 0.0% Kentucky 0 35 35 0.0% Louisiana 0 37 37 0.0% Maine 0 38 38 0.0% Mississippi 0 26 26 0.0% Montana 0 31 31 0.0% North Dakota 0 21 21 0.0% Nevada 0 28 28 0.0% Puerto Rico 0 5 5 0.0% South Carolina 0 33 33 0.0% South Dakota 0 23 23 0.0% Vermont 0 45 45 0.0% West Virginia 0 14 14 0.0% Wyoming 0 18 18 0.0% * Table provides firm-state cross tabulations but firms may be associated with more than one state # State listings include the District of Colombia and Puerto Rico

24

TABLE 3: Firms Receiving Follow-on Venture Capital Investment By SBIR Awarding Agency* VC No VC Total After After Firms SBIR Agency Department of Health and Human Services (HHS) 181 3,328 3,509 DOD - Chemical and Biological Defense Program (CBD) 2 40 42 DOD - Defense Advanced Research Projects Agency (DARPA) 42 1,049 1,091 DOD - Ballistic Missile Defense Organization (BMDO) 30 750 780 National Science Foundation (NSF) 58 1,467 1,525 Nuclear Regulatory Commission (NRC) 3 86 89 Department of Energy (DOE) 37 1,066 1,103 DOD - Defense Special Weapons Agency (DSWA) 5 151 156 DOD - Air Force 72 2,182 2,254 DOD - ARMY 55 1,772 1,827 DOD - NAVY 63 2,032 2,095 National Aeronautics and Space Administration (NASA) 49 1,611 1,660 Department of Commerce (DOC) 8 303 311 Environmental Protection Agency (EPA) 7 277 284 DOD - Office of the Secretary of Defense (OSD) 5 201 206 Department of Agriculture (DOA) 13 560 573 Department of Transportation (DOT) 7 348 355 DOD - Standards Of Conduct Office (SOCO) 1 62 63 Department of Education (ED) 4 300 304 DOD - National Imagery and Mapping Agency (NIMA) 0 2 2 DOD - Defense Threat Reduction Agency (DTRA) 0 15 15 * Table provides firm-agency cross tabulations but firms may be associated with more than one agency

% VC After 5.4% 5.0% 4.0% 4.0% 4.0% 3.5% 3.5% 3.3% 3.3% 3.1% 3.1% 3.0% 2.6% 2.5% 2.5% 2.3% 2.0% 1.6% 1.3% 0.0% 0.0%

TABLE 4: Follow-on Venture Capital Investment (Firm-level Regressions) Dependent variable: VCafter Variable A B C 0.196 *** 0.196 *** PII_Dum (0.058) (0.059) 1.383 *** 1.412 *** 1.280 *** PhaseI$ (0.293) (0.295) (0.302) 0.064 0.087 0.087 PhaseII$ (0.080) (0.083) (0.084) -0.092 *** -0.083 *** -0.073 *** Num_PI (0.026) (0.027) (0.027) -0.118 ** -0.184 *** -0.180 *** Num_PII (0.056) (0.061) (0.0662) -0.025 Minority (0.080) -0.432 *** Woman (0.106) 0.644 *** Biotech (0.107) Intercept -1.815 *** -1.892 *** -1.886 *** (0.027) (0.035) (0.036) Joint significance of SBIR Agency dummies: χ2(19) Joint significance of US State dummies: χ2(33) # of obs. 10,914 10,914 10,914 Log-Likelihood -1611.307 -1605.487 -1577.740 McFadden-R2 0.011 0.015 0.0316

D 0.167 *** (0.063) 1.234 *** (0.324) 0.076 (0.091) -0.099 *** (0.031) -0.163 ** (0.067) -0.015 (0.083) -0.435 *** (0.109) 0.565 *** (0.114) -2.672 *** (0.154) 40.92*** 55.43*** 10,313 -1502.493 0.0654

Notes: Standard errors in parentheses. *** (**,*) indicate a significance level of 1% (5 ,10%).

REFERENCES Archibald, R.B., D.H. Finifter, 2003, ‘Evaluating the NASA small business innovation research program: preliminary evidence of a trade-off between commercialization and basic research,” Research Policy, 32, 605-919. Arrow, K.J., 1962, ‘Economic Welfare and the Allocations of Resources of Invention,’ in: Nelson, R.R. (ed.), The Rate and Direction of Inventive Activity: Economic and Social Factors, Princeton. Audretsch, D.B., 2003, ‘Standing on the Shoulders of Midgets: The U.S. Small Business Innovation Research Program (SBIR),’ Small Business Economics, 20, 129-135. Audretsch, D.B., A.N. Link and J.T. Scott, 2002, ‘Public/private technology partnerships: evaluating SBIR-supported research,’ Research Policy, 31, 145158. Bhidé, A.V., 2000, The Origins and Evolution of New Business, Oxford: Oxford University Press. Branscomb, L.M., P.E. Auerswald, 2001, Taking Technical Risks: How Innovators, Managers, and Investors Manage Risk in high-Tech Innovations, Cambridge, MA: MIT Press. David, P.A., B.H. Hall, and A.A. Toole, 2000, ‘Is public R&D a complement or substitute for private R&D? A review of the econometric evidence,’ Research Policy, 29(4-5), 497-529. De Bettignies, J-E., J.A. Brander, 2007, “Financing entrepreneurship: Bank finance versus venture capital,” Journal of Business Venturing, 22, 808-832.

27

Dixit, A., 1992, ‘Investment and Hysteresis,’ Journal of Economic Perspectives, 6(1), 107-132. Dixit, A.K., R.S. Pindyck, 1994, Investment under uncertainty. Princeton: Princeton University Press. Feldman, M.P., M.R. Kelley, 2003, ‘Leveraging research and development: Assessing the impact of the U.S. Advanced Technology Program,’ Small Business Economics, 20, 153-165. Fried, V.H., R.D. Hisrich, 1994, ‘Toward a model of venture capital investment decision making,’ Financial Management, 23(3), 28-37. Griliches, Z., 1992, ‘The search for R&D spillovers,’ Scandinavian Journal of Economics, 94 (0), S29-47. Hall, B.H., 2006, ‘The financing of innovation,’ Forthcoming in: Shane, S. (Ed.), Handbook of Technology and Innovation Management. Blackwell Publishers, Ltd.: Oxford. Hsu, D.H., 2006, ‘Venture capitalists and cooperative start-up commercialization strategy,’ Management Science, 52(2), 204-219. Hubbard, R.J., 1998, Capital-market Imperfections and investment,’ Journal of Economic Literature, 36, 193-225. Kaplan, S.N., P. Stromberg, 1999, ‘Venture capitalists as principals: contracting, screening, and monitoring,’ The American Economic Review Papers and Proceedings, 91(2), 426-430. Kaplan, S.N., P. Stromberg, 2000, ‘How do venture capitalists choose investments?’ Working paper, University of Chicago, September 2000.

28

Laidlaw, F.J., 1998, ‘ATP’s impact on accelerating the development and commercialization of advanced technology,’ Journal of Technology Transfer, 23, 33-41. Lerner, J., 1999, ‘The Government as Venture Capitalist: The Long-run Impact of the SBIR program,’ Journal of Business, 72(3), 285-318. National Research Council, 2004, An Assessment of the Small Business Research Program: Project Methodology. Committee on Capitalizing on Science, Technology, and Innovation. Washington, D.C.: National Academy Press. Nelson, R.R. (1959), The Simple Economics of Basic Scientific Research, Journal of Political Economy 67, 297-306. Pindyck, R.S., 1991, ‘Irreversibility, Uncertainty, and Investment,’ Journal of Economic Literature, 29(3), September 1991, 1110-1148. Sine, W.D, S. Shane, and D. Di Gregoria, 2003, ‘The halo effect and technology licensing: The influence of institutional prestige on licensing of university inventions,’ Management Science, 29(4), 478-496. Toole, A.A., D. Czarnitzki, 2007, ‘Biomedical Academic Entrepreneurship through the SBIR Program,’ Journal of Economic Behavior and Organization, 63, 716-738. US Small Business Administration, Office of Technology, 2007, Handbook for SBIR Proposal Preparation, http://www.sba.gov/gopher/Innovation-AndResearch/SBIR-Pro-Prep/. Wallsten, S., 1998, ‘Rethinking the Small Business Innovation Research Program,’ in: Branscomb, L.M, and Keller, J.H (eds.), Investing in Innovation, Cambridge: The MIT Press.

29

Suggest Documents