When Does Funding Research By Smaller Firms Bear Fruit?: Evidence

When Does Funding Research by Smaller Firms Bear Fruit?: Evidence from the SBIR Program*

by

Joshua S. Gans and Scott Stern** First Draft: April 2, 1999 This Version: November 8, 1999

COMMENTS WELCOME!

*

The cooperation and assistance of the firms that contributed data used in this study is greatly appreciated. Josh Lerner, Scott Wallsten, Scott Shane, seminar participants at MIT, the FTC and the NBER, and especially Iain Cockburn, provided thoughtful suggestions and advice. David Hsu provided outstanding research assistance. We also gratefully acknowledge funding for this research by the MIT Center for Innovation in Product Development under NSF Cooperative Agreement # EEC-9529140. Any remaining errors or omissions are our responsibility. ** Melbourne Business School, University of Melbourne, and MIT Sloan School & NBER, respectively. All correspondence to Scott Stern, Sloan School of Management, MIT, Cambridge (MA), 02142; mailto:[email protected]. The latest version of this paper is available at http://www.mbs.unimelb.edu.au/jgans/research.htm.

When Does Funding Research in Smaller Firms Bear Fruit?: Evidence from the SBIR Program

Abstract This paper evaluates whether the relative concentration of funding for small, researchoriented firms in some important high-tech industries is related to the differences across industries in the appropriability regime facing small firms. We explore this hypothesis by analyzing the performance of projects funded by the Small Business Innovation Research (SBIR) program, a Federal subsidy that provides R&D funds for small businesses. By design, the cost of R&D capital is (approximately) equalized for firms funded by the SBIR. As a result, we neutralize capital market imperfections as a direct source of variation in our sample. In contrast, the SBIR does not affect sector-level differences in the appropriability regime or in the underlying level of technological opportunity. The main contribution of this paper results from exploiting this difference to evaluate the salience of capital market imperfections, the appropriability regime facing small firms, and the overall level of technology opportunity. Specifically, if the SBIR fund projects on the margin (as it should under an optimal subsidy regime), then a cross-sectional comparison identifies the relative importance of capital versus product market imperfections across markets. Our principal empirical result is that project-level performance is highest for those technologies that are in industrial segments that attract high rates of venture capital investment. As well, there is weak but positive evidence that performance is related to the overall level of scientific opportunity. We interpret these findings as suggesting that an important difference between industrial sectors is the degree of appropriability for research-oriented small businesses; variation in the appropriability regime helps explain the concentrated nature of venture capital activity in the economy.

I.

Introduction Since Arrow (1962), economists have been aware that competitive markets may

fail to provide the socially optimal level of R&D investment. Innovation is beset by uncertainty, imperfect monitoring and imperfect property rights; the combination of these factors leads many to conclude that the realized level of R&D investment by private firms is too low (Bush, 1945; Griliches, 1992; Romer, 1990). In this context, special attention is paid to the role of the small, research-oriented firm. While many suggest that these firms have organizational advantages enhancing their research productivity (Schumpeter, 1934; Foster, 1986; Acs and Audretsch, 1996), others highlight that smaller, start-up firms may be particularly susceptible to the constraints identified by Arrow (Holmstrom, 1989; Teece, 1986; Kamien and Schwartz, 1982; Himmelberg and Petersen, 1994). Specifically, smaller firms may both have difficulty raising capital for R&D projects and be less well positioned to extract the social value of their innovations in the marketplace. Perhaps in part because of the salience of these constraints, private financing of small firm research is quite concentrated in a small number of industrial sectors. Historically, areas such as biotechnology and software development have attracted a disproportionate share of financing, particularly when compared with industries such as industrial equipment, transportation or environmental technologies (while all of these sectors experience high rates of innovation, the composition of funding and performance differs across sectors). Concentrated financing of entrepreneurial activity suggests that venture financiers have somehow overcome the capital market imperfections and appropriability constraints in a small number of distinct market and technological

2

environments. Put another way, understanding why small firm research is financed in only a few sectors involves discerning how the funded sectors differ in terms of the salience of agency and/or appropriability. By undertaking such an exercise, this paper enhances understanding of how agency and appropriability affect innovation incentives as well as providing policy guidance in terms of how to better encourage small firm formation and innovation. The paper evaluates the variation across sectors in the performance of projects funded by the Small Business Innovation Research (SBIR) program, a Federal subsidy that provides R&D funds for small businesses. By design, the SBIR program effectively equalizes the type of capital market imperfections that theoretically raise the costs of capital to small firm R&D. As a result, the salience of capital market imperfections is (roughly) equalized across our sample. In contrast, the SBIR does not affect sector-level differences in the appropriability regime or in the underlying level of technological opportunity. The main contribution of this paper arises from exploiting this difference to evaluate how sectors differ in terms of capital market imperfections, the appropriability regime facing small firms, and the overall level of technological opportunity. We construct a theoretical model demonstrating that, if the SBIR funds projects that would not have been funded in the absence of a subsidy, then a cross-sectional comparison identifies the relative importance of capital versus product market imperfections across industries. The intuition is straightforward. Private financiers will be deterred from funding innovations with relatively high ex post returns in those industries where capital market imperfections are particularly binding. If projects are funded on the margin, then one would expect that grants provided to firms in capital-constrained

3

industries would tend to perform most highly. On the other hand, grantee performance will be highest in those sectors that have the ability to earn the highest returns even in the absence of the subsidy. If differences across sectors are driven by differences in the appropriablity regime, this will manifest itself as a positive correlation between performance and the stock of private venture capital fundraising, conditional on the level of technological opportunity. Alternatively, if the key correlation is between performance and technological opportunity itself, this may suggest that the critical differences in funding across sectors simply result from variation in technological opportunity. Using a novel dataset constructed from a survey by the authors of SBIR funded projects, our principal empirical finding is that project-level performance is highest for those technologies that are in industrial segments that attract high rates of venture capital investment. As well, there is weak, but positive, evidence that performance is related to the overall level of scientific opportunity. We interpret these findings as suggesting that an important difference between industrial sectors is the degree of appropriability for research-oriented

small businesses, and this variation helps explain the relative

concentration of venture capital activity in some industrial segments. By examining performance differences across subsidized projects, this paper differs from most prior treatments that have attempted to evaluate intersectoral differences in the environment for small-firm R&D. By and large, such studies have analyzed samples that are subject to an important selectivity – they have indeed been funded by the private sector (Mansfield, 1995; Griliches, 1998; Hall, 1988; 1993; Himmelberg and Petersen, 1994). As such, it is difficult to isolate the differences in capital market imperfections from more downstream issues such as the appropriablity

4

regime. As well, our approach differs from prior work on the SBIR (in particular Lerner (1996) and Wallsten (1997)). Both of these prior studies seek to identify the incremental benefits of an SBIR grant relative to privately funded research; in contrast, we examine the differences across funded firms in order to understand how different segments differ in terms of the environment for innovation for smaller firms. The rest of the paper proceeds as follows. In the next section, we review the theoretical and empirical literature on the constraints facing small research-oriented firms in capital and product markets. Section III then describes the SBIR program and the features that allow it to provide an experiment to control for capital market imperfections across sectors. Section IV presents a model that captures the essence of capital and product market imperfections and we use this as a basis to formulate testable hypotheses about the determinants of venture capital funding levels across industries. Section V presents an overview of our dataset while in Section VI we present our main empirical results and tests for the robustness of these. A final section concludes.

II.

The Concentrated Nature of R&D and Venture Capital Expenditures: An Economic Puzzle? Among the most distinctive facts about the innovative process is the concentrated

nature of formal R&D investment across the economy and the even further concentrated nature of venture capital finance or small-firm R&D financing. Consider first the concentration of privately funded R&D expenditure. Figure 1A presents the Lorenz distribution of R&D funding relative to value added across the manufacturing sector of

5

the U.S. economy. 1 Relative to the distribution of aggregate economic activity (even confined to the manufacturing sector), a small share of sectors (such as the pharmaceutical, computer equipment, and transportation industries) account for most of privately financed R&D.2 This relative concentration of innovative investment is even more pronounced when one examines the venture capital financing or the financing of R&D in small and medium-sized firms. Using data drawn from Kortum and Lerner (1999), Figure 1B compares the relative distribution of venture financing by industrial segment relative to the (already concentrated) privately financed R&D investment. Though there are substantial difficulties in uniquely assigning venture capital investments into specific SIC-oriented sectors (a concern which we address directly in our empirical work), Figure 1B suggests that, even among R&D-intensive sectors, a very small number receive a very high share of the overall funds from venture capital sources.3 This latter finding about the concentrated nature of investment by small firms suggests that differences across sectors are not merely driven by difference in raw technological opportunity (which would be reflected in the aggregate R&D distribution) but depend on the ability of small firms to both finance risky but potentially valuable innovation investments and their ability to translate their technological success into economic returns.

1

All data are drawn from the 1992 NSF Science and Engineering Indicators and are presented in terms of R & Dj their rank of . VALUE ADDED j 2

Indeed, the Gini coefficient of inequality is over .6, a relatively high rate of inequality across sectors in terms of their R&D expenditures. 3 Indeed, while economists have relatively ignored the consequences of this observation, its validity is a widely accepted and well-documented feature of the venture finance industry (REFS, www.ventureone.com). As well, one could replicate our result by shifting the analysis away from venture capital in particular and towards the somewhat more comparable concept of small-to-medium sized R&D investment (tabulations available from the authors).

6

Although prior research has acknowledged that these distributions are in fact skewed (Griliches, 1986; Hall, 1992; Kortum and Lerner, 1999), few attempts have been made to distinguish the different potential drivers of the heterogeneity across sectors. Specifically, while there exist studies which examine a single hypothesis and exploit variation across sectors to evaluate its relative importance (e.g., Cockburn and Griliches, 1988; Himmelstein, 1993), there has been little systematic analysis of the full set of potential drivers of this observed heterogeneity across sectors in terms of the intensity of the innovative or entrepreneurial process. In particular, our analysis will consider three broad

sources

of

non-mutually-exclusive

differences

among

sectors

which

may

contribute to differences in terms of their R&D investment levels: the salience of capital market imperfections, the degree of appropriability, and the level of technological opportunity. Each of these potential sources has the potential to create variance in the environment across different industrial sectors that manifests itself in terms of differences in R&D investment. Accordingly, we review each of these areas informally before turning to an equilibrium analysis of how each shapes industry-level R&D funding and how evaluation of government-subsidized research may provide insight into the relative salience of these different forces.

Level of Technological Opportunity Perhaps the simplest explanation for differences across sectors is fundamental differences in the level of technological opportunity or demand across sectors (Schmookler, 1967; Rosenberg, 1974). Indeed, to the extent that most studies of the determinants of R&D investment focus on the activities of established incumbent firms

7

within given sectors, the opportunity hypothesis is perhaps a persuasive baseline hypothesis for considering the sources of differences across sectors.

Salience of Capital Market Imperfections From a theoretical perspective, research on the salience of capital market imperfections in shaping R&D investment has focused on the presence of information asymmetries between financier and research firm and the consequential incompleteness of financial contracts (Kamien and Schwartz, 198*). One strand of this literature focuses on the potential conflict between equity or debt finance and the incentives of small research firms to expend non- or partly contractible effort in innovative activity. Holmstrom (1989) notes the difficulty in measuring either the inputs or outputs of research firms while Aghion and Tirole (1994) extend this to demonstrate the adverse incentive consequences of in-house R&D and of equity participation by third party financiers such as venture capitalists. Related to this are concerns that innovators themselves may be able to hold-up financiers. Anand and Galetovic (1999) examine the potential concerns of corporate and venture capitalists in having multi-project research firms realize the returns from financed innovations in opportunistic ways. Hellmann (1998a) notes that innovators may have incentives that are not purely commercial and hence, may not act in ways that maximize investor returns. These effects can make finance contracts infeasible or require excessive control and monitoring by financiers thereby diluting the innovation incentives of small firms.

8

A final strand of the literature concerns the ability of financiers to identify potentially successful projects ex ante. The adverse selection literature goes all the way back to Stiglitz and Weiss (1981) and their analysis of the difficulties of using debt contracts to identify good projects. Ownership arrangements and other forms of corporate governance

can

mitigate

these

issues.

However,

the

potential

for

small

firm

overstatement of the potential feasibility of their proposed research remains (see Hubbard, 1998).

Degree of Appropriability Finally, there may be differences across sectors in terms of the ability of innovators to appropriate the economic returns from the innovative process. Innovation involves spillovers (Bush, 1949; Nelson, 1959) and the pricing of indivisibilities is imperfect (Arrow, 1962; Romer, 1990) so that private investors can never appropriate the full social returns on any sunk R&D costs. These challenges face all private research activity.

Indeed,

the

role

of

appropriability

in

shaping

both

the

mode

of

commercialization and the incentives for R&D investment is the central focus of the work by Teece (1987) and others. There is also a specific set of constraints that face research-oriented start-ups. Schumpeter (1942) noted that the competitive process dissipates innovative rents favoring monopoly over entry (see also Gilbert and Newbery, 1982). However, even when these firms are looking to license or otherwise sell their innovations to product market incumbents, there are important challenges. As noted earlier, such licensing contracts can rarely be agreed upon ex ante. Either the output of innovative activity is

9

hard to specify (Aghion and Tirole, 1994) or it is difficult to identify good projects ex ante (Anton and Yao, 1994). The result of this is that, even when intellectual property rights are relatively secure, small firms may only appropriate a fraction of the private returns from innovation. Only when they pose a sufficient competitive threat to existing incumbents or play multiple incumbents off against on another, can they improve their bargaining position (Teece, 1987; Gans and Stern, 1998). Furthermore, there are transactions costs associated with the exchange of ideas ex post. Potential purchasers of tacit knowledge or innovations with weak property rights can expropriate small firm innovators deterring them from seeking profitable partnerships (Anton and Yao, 1994). This is fundamental because it is rarely the case the innovations can be sold without revealing their nature to potential buyers (Arrow, 1962).4 This means that small research firms may face costly product market entry as the only feasible commercialization route, thereby, reducing their returns further (Gans and Stern, 1998).

III.

A Simple Model of Sector-Level Variation in Innovative Expenditure In this section, we integrate the insights of the previous two sections and develop

a simple model of the impact of an R&D subsidy focused on small mostly researchoriented firms. We demonstrate that, if capital market imperfections are important constraints on small firm research financing, the average performance of SBIR grantees is likely to be negatively associated with observed levels of venture capital funding. This

4

Anton and Yao (1994) demonstrate that the presence of multiple incumbents or low entry barriers can mitigate this exp ropriation problem. In these cases, the small research firm can credibly threaten to destroy the potential for an exclusive licensing arrangement in order to avoid expropriation and improve their potential license fee (see also Rasmusen, 1988).

10

is because the funding mechanism identifies high potential performers only in industries where those are not otherwise likely to be funded. Alternatively, a positive association between average performance and venture capital funding will be indicative of the relative importance of appropriability as explaining differences across industries in terms of small firm research financing. Either way, this discussion will highlight the fact that we need to control for differences across industries in appropriability and technological opportunity when attempting to examine the importance of capital market imperfections in constraining small firm research activity.

Players, Payoffs and Choice Variables Our focus here is on the provision of capital to a research unit (RU) that engages in innovative activity. That capital is considered a critical input for innovative activity. We suppose that a project requires a unit of capital. 5 This capital may be provided by a venture capitalist (VC) or by a subsidy from the government. The primary difference between the two sources is that the former requires, while the latter does not strictly use, potential private returns in selecting projects to fund. We will consider the VC capital source first and discuss the role of a government subsidy later. It is assumed that the VC’s capital contribution is contractible and that VC financing is competitive. Having received the requisite capital for the project, the probability that a commercializable innovation is generated is p. This probability may be endogenous depending on the ability (type) or effort of the RU.

5

This unit is the minimum level of capital required for a project. The unit assumption is a normalization; essentially all our return variables are in per unit of capital terms for notational ease.

11

An innovation, if generated, has a social value of V of which the RU (and their equity partners) can only appropriate a fraction, γ (< 1). We denote the RU’s returns by v = γV. As discussed earlier (in Section II), previous research has identified many potential determinants of γ including the degree of product market competition and the ease of entry to it, the strength of intellectual property rights, and transactions costs involved in selling ideas. Each of these represents a potential imperfection at the product, rather than capital, market level affecting the private returns to innovation. While our discussion of the model here will focus on an individual project with social return, V, we expect that these returns will vary among projects. For our exposition here we will assume that a given project is drawn from an industry-level distribution that is uniform over the space [ 0,V ] . The parameter, V , is, therefore, a measure of the level of technological opportunity in an industry. Hence, one would expect average performance to be higher in industries with larger V .

Determining VC Funding The social cost of the total level of industry funding, F, is assumed to be

1 2

F2;

that is, it is increasingly difficult to attract capital. For a given project, the marginal private cost of capital is F + θ, where θ is the difference between the private and social marginal cost of capital. We assume that θ > 0 capturing the potential for capital market imperfections. This marginal cost determines the supply of industry funding for small firm research. This model is admittedly a reduced form. In the appendix, we provide three alternative models of why, in this context, the private cost of capital may be greater than the social cost.

12

On the demand-side, the expected private return from a project is pv. The total industry demand for these projects – that is, the number of projects funded, F – is determined by the distribution of project value in the industry. That is, if for F projects are funded, the marginal project has social value V determined by F = V − V . The expected

marginal

private

return

when

F

projects

are

funded

is,

therefore,

pγ V = pγ (V − F ) . This is the (inverse) demand for funding of RUs in the industry. Figure 2 graphically depicts the supply and demand for RU funding in an industry. In equilibrium, the level of industry funding is: F=

1 2

( pγ V −θ )

(1)

Notice that F is increasing in V and γ but falling in θ. That is, an increase in V or γ shifts the demand and hence F, upwards, while an increase in θ, shifts supply upwards and reduces F. Thus, industries with greater appropriability, technological opportunity and lower capital costs, will see more funding than other industries. However, other things being equal, the marginal project funded will have a higher expected value in industries that have higher capital costs as well as greater degrees of appropriability and technological opportunity.

13

Figure 2: Equilibrium Industry Funding

pγ V

F+θ

Expected Project Value

pγ (V − F )

0

F

Total Funding

Notice that a project will be undertaken only if pv ≥ F + θ . This differs from the social criteria for project desirability:

pV ≥ F . Consequently, the level of private

investment in R&D may be less than the socially desirable level both because of difficulties in appropriability through the product market (γ < 1) and a greater private cost of capital.

The Effect of a Capital Subsidy Our empirical goal is to determine the broad explanation of differences in observed VC funding across industries. Above we derived that, for industry i, Fi = f (Vi ,γ i ,θ i ) . That is the random variable, Fi, is a function of the industry levels of

technological opportunity ( Vi ), appropriability (γi) and capital market imperfections (θi). Each of these variables determines either the demand or supply of funds to the industry. Therefore, to discover which of these might account for variations in industry VC

14

funding, we need to determine whether different inter-industry funding levels are consistent with movements along the demand or supply curve. The key issue with this empirical exercise is that, while the quantity variable (F) is readily observable, there are difficulties in determining the price variable (the expected profitability of the marginal project) in a way that is comparable across industries. We propose that SBIR program recipients provide a means of identifying the marginal project. The SBIR program provides capital grants that effectively equalize the type of capital market imperfections modeled here. Specifically, these grants do not require equity participation by third parties or any reference to the private returns from capital investment. As such, the effect of the SBIR program is to (roughly) equalize any differences in capital market imperfections across grantees in different industries. The private cost of capital is simply irrelevant in terms of the direct objectives of the subsidy program. Hence, SBIR recipients include those that would not otherwise meet private funding criteria. It is the SBIR funding mechanism that assists in identifying marginal projects. A laudable goal of any government subsidy program is to encourage economic activity where it does not currently exist. For the financing of small firm investment, this goal could be realised if the projects that were selected were the most promising projects that missed out on private funding. In our model, these projects would be those that just failed to attract funding. For industry i, the marginal project would be characterized by the condition: vi =

pγ iVi + θ i . 2p

15

If this project is granted a subsidy, this condition characterizes the private return it would realize. Observe that this private return is increasing in the degree of capital market imperfection (θi), the space of technological opportunities ( Vi ), and the level of appropriability (γi). Importantly, given its role in determining the supply of funds, changes in θi move Fi and v i in opposite directions. Hence, if differences in capital market perfections were an important driving force in the concentration of VC funding, we would expect observed Fi and v i to be negatively correlated. The following proposition characterizes the correlation between vi and Fi across industries. Proposition 1. cov[vi ,Fi ] < 0 if and only if the variance of θi exceeds the variance of pγ iVi . PROOF: Recall that, from

(1), Fi =

1 2

( pγ V −θ ) . Then: i i

i

 pγ V + θ i   pγ iVi +θ i  1 cov[ F , vi ] = E  12 ( pγ iVi − θ i ) i i  − E  2 ( pγ iVi −θ i )  E   <0 2p    2p  2 2 1 1 2 ⇒ p2 E  (γ iVi )  − E θi2  < p2 E γ iVi  − E [θi ]     4p 4p

( ) ( ⇒ p ( E ( γ V )  − E γ V  ) < E θ  − E [θ ]   2

2

2

i i

i i

2 i

)

2

i

Intuitively, variations in θi move the supply curve while variations pγ iVi move the demand curve. Thus if the variance of θi exceeded that for pγ iVi , one would observe ‘price’ and ‘quantity’ variables across industries tracking the demand curve. As this is downward sloping those variables would be negatively correlated. On the other hand, if the opposite were true and θi varied relatively little across industries, observed pairs would track the supply curve that is upward sloping. Hence, the variables would be positively correlated.

16

In summary, by looking to the SBIR program for information regarding the returns of marginal projects (v i), we can use the correlation between this and observed VC funding (Fi) to determine the relative strengths of capital market imperfections and product market variables in accounting for differing levels of VC funding across industries.

Funding Mechanism In proposing the above empirical test of the relative salience of capital and product market imperfections in determining the level of VC funding we have postulated a marginal funding mechanism for selecting SBIR recipients. It may be objected that the SBIR could use other mechanisms for allocating funds. In particular, projects may be selected to ensure political success of the program. Consequently, the SBIR may cherry pick the best possible project rather than projects that would otherwise not receive private funding (Wallsten, 1997). If cherry picking is the selection mechanism, then if the main source of variation across industries is in the capital market imperfection (θi) then the level of industry VC funding is unlikely to explain the average performance of grant recipients. While the grant alleviates the capital market imperfection, it equalizes its expected effect on individual grant recipients, as those recipients are not selected on the basis of V alone. This is in contrast to the marginal rule that implicitly considers θ when it searchers for projects that are not likely to be funded by the VC. Under cherry picking, projects that might otherwise by funded, are selected for a government subsidy. Hence, the concerns

17

that subsidy programs, such as the SBIR, may not be allocating resources to where they are currently scarce (Wallsten, 1997). Nonetheless, grantees in industries with greater technological opportunity ( Vi ) or higher product market appropriability (γi), will perform better under a cherry picking rule. Hence, if either of these were the main source of structural variation across industries, this would still be reflected in a positive correlation between industry-average grantee performance and the level of industry VC funding. Otherwise that correlation would be close to zero.

IV.

The SBIR Program and R&D Investment by Small Firms This paper uses Proposition One in order to evaluate how an evaluation of the

performance of government-subsidized small R&D-intensive firms may inform us about the relative salience of different forces which determine the funding and intensity of small-firm research-oriented innovation across industries and technology segments. To do so, we exploit data about projects funded by the Federal government’s SBIR, the largest individual source of R&D financing for small to medium-sized firms in the United States. SBIR funding provides a unique and informative source of variation in the funding of small, research-oriented firms. Specifically, from the perspective of funded firms, the SBIR program (roughly) equalizes the cost of capital for R&D regardless of industrial

sector,

while

keeping

the

nature

of

downstream

appropriability

and

technological opportunity constant in specific industries and technology sectors. We exploit this variation to evaluate the relative importance of capital market imperfections, product market imperfections, and technological opportunity in determining differences

18

in the level and composition of R&D investment across industries and technological sectors. Such a test depends, of course, on drawing a precise mapping between the institutional details of the SBIR program and the assumed economic consequence of the program. We, therefore, first briefly review the SBIR program, highlighting the potential economic effects of the subsidy program. The SBIR Program, first authorized in 1982, requires that all Federal agencies who support a minimal level of R&D activity are required to set aside a certain percentage of their funds for extramural grants to fund R&D projects by small business.6 The principal legislative goals of the program are to: (a) increase the rate of commercialization of innovations derived from Federal research, (b) enhance

the

‘competitiveness’

of

small

firms

in

technology-intensive

industries; and (c) enhance the participation of small firms as well as women and minorityowned businesses in the Federal contracting process (GAO, 1995). From a political perspective, the program’s support seems to derive in part from (everincreasing) political demand to focus R&D expenditures towards more near-term development projects and to areas which have a clear relationship to medium-term economic growth (Cohen and Noll, 1993). Combined with political rhetoric which simply assumes that research-oriented start-up firms are the engine of long-term economic growth, and that such firms face a particularly severe “funding gap” arising from the

6

This percentage has varied over time starting at X in 1990 and reaching a stable level of Y in 1997. For the purposes of the program, a small business is defined as an US-owned firm with less than 500 employees. Further details of the SBIR are discussed extensively in GAO (1995), Wallsten (1995, 1997), and Lerner (1999).

19

unwillingness of investors to shoulder the risks of early-stage financing, the SBIR has been a popular program whose scope has been consistently expanded since its inception (REFS). Indeed, the SBIR is now the single largest source of early–stage R&D financing for small firms in the United States with 1997 expenditures of over $1 billion. Despite its political support, the program has been somewhat controversial. In part, this is because neither the legislative nor regulatory rules governing the program mandate the program fund projects on the margin (which would determine performance according to the first part of Proposition One): indeed, as dramatically highlighted by Wallsten (1997), the program’s funding guidelines seem to focus on funding the most attractive grant applications from either a technical or commercialization perspective (thus suggesting that performance may reflect the “cherry-picking” regime).7 Perhaps as a consequence of this selectivity, there is some evidence that firms funded by the program do tend to have an accelerated rate of growth compared to similar firms (Lerner, 1999). However, Lerner emphasizes that this “boost” to firm-level growth seems to be localized according to the location or technology focus of the firm, a contention which we explore in much further detail in our empirical work. Indeed, the main focus of this prior research is on the assessment of the marginal contribution of SBIR funding (and government venture capital more generally). Towards this goal, each of these analyses is framed in terms of the counterfactual of the expected behavior or performance of these firms, in the absence of the subsidy program. Indeed, both Lerner (1999) and Wallsten (1997) focus their results around the comparison of a 7

An additional critique is that a small number of firms have been able to win a disproportionate share of overall grants. While the absolute importance of these so-called “SBIR mills” is limited, the potential for regulatory capture by these firms has led to interest in limits on the total number of grants for which a

20

group of SBIR-funded firms and a “matched sample” of firms who are observationally similar ex ante but who do not receive SBIR funding. For example, Wallsten concludes that SBIR award winners do not necessarily grow faster than other firms (in terms of employment growth) by comparing a group of SBIR award winners with a group of firms who had applied for but been rejected in the SBIR grant applications. Lerner also examines both SBIR grantees and a comparison group in a regression analysis. Once again his focus is on the incremental differences at the firm level associated with receiving an SBIR award. He concludes that the awardees do tend to grow faster though this effect is mostly localized around firms in locations that also fund venture capital. This paper refocuses analysis of the program away from program assessment. Instead, our analysis exploits three central features of SBIR funding: (1) the wide dispersion of these funds across industries and technology segments; (2) the program’s ability to alleviate capital market imperfections in the funding of research; and (3) the program’s inability to directly impact the nature of appropriability or technology opportunity in a given field or technology area. We use these features together to pose and analyze our empirical test which compares the behavior and performance of SBIR grantees across industrial areas and technology segments. First, because the program is administered through a variety of Federal agencies (Agriculture, Defense, HHS, etc.) whose missions span the scope of the economy’s activities, the expenditures of the SBIR program are much more widely dispersed (across

single firm might be eligible (GAO, 1995). While we do not specifically address mini-mills per se, we do demonstrate that our results are robust to their inclusion or exclusion.

21

industries and technological areas) than privately funded R&D. This wedge between SBIR expenditures and private funding is particularly salient if one focuses on investment or R&D expenditures specifically for small (or venture-backed) firms. In other words, a specific if unintended contribution of the SBIR program is to expand the supply of capital to research-oriented, smaller firms to a set of industries and technologies which, for one reason or another, do not currently attract such funds. The second element of the program that we utilize is that, as a “hands-off” subsidy, the SBIR program substantially alleviates the salience of capital market imperfections. In contrast to the tradeoff that emerges between the provision of incentives and the taking of equity in the analysis of private investment in a research-oriented organization, the SBIR does not extract equity (or debt) from the grantee. In fact, quite the opposite. The incentives to perform the research are preserved and incentives to divert the fruits of research away from equity holders are minimized. Of course, SBIR funds do not completely alleviate capital market issues or incentive problems – we only claim (or need to claim) that the SBIR program reduces the salience of capital market imperfections relative to the provision of private venture capital. Finally, as a hands-off subsidy, SBIR grants cannot alleviate product market imperfections or the lack of technological opportunity in a given field. If intellectual property rights are weak in a given area prior to the SBIR grant, there is nothing in the funding of that research which will overcome this general bias. Indeed, while venture capitalist funding may provide non-pecuniary value to the firm through the experience of the managers in translating research results or novel ideas into appropriable technologies

22

or business models, the SBIR granting process is focused on funding technology development per se and contributes essentially no additional commercialization services.8 The remainder of this paper is devoted to exposing and exploring the economic test embedded in these three features of the SBIR. By being more diverse than privately funded investment, data from grantees provides information about the differences in the returns structure across industries where there is currently venture capital funding and also where such funding is more rare. Differences among SBIR grantees across industrial areas may reflect the differential impact of reducing the salience of capital market imperfections. For example, if capital market imperfections are particularly strong in a given industry, then the provision of public funds may hold particularly high marginal productivity there. Alternatively, perhaps capital market imperfections are somewhat similar across sectors, but sectors differ substantially in terms of their appropriability regime (or overall technological opportunity). In this case, the subsidy will be most productive (in terms of project performance) in those sectors that already offer a favorable appropriability (or technological opportunity) environment. As such, in contrast to the earlier hypothesis, project performance should be highest in those sectors that already are funded by venture capitalists. In evaluating the performance of the SBIR program, we are examining a interesting empirical context where we observe an equalizing shift in the capital constraint, holding the appropriability regime constant. We can examine this more carefully by first describing the data we have gathered for this study before turning to our empirical results.

8

In recent years, there has been legislation encouraging the funding of consultants for grantees to assist in the commercialization process. By all accounts, such efforts or activities are very rare empirically and pale compared to the involvement of venture capital management.

23

V.

Data This paper presents results from a novel dataset of 100 projects funded by the

SBIR since 1990. The data were gathered via a field-based proprietary survey conducted by the authors. This project-level data was then supplemented with public data on each firm’s patenting behavior, SBIR grant history, and covariates related to the industry, business segment, and scientific underpinning associated with each project. In this section, we first review our procedure and the elements of our survey (highlighting the sample selection and data gathering process) and then review the summary statistics for the sample (Table 1 includes the definitions of all the variables used in the analysis; Table 2 provides means and standard deviations).

Survey Data Sources and Sample Selection Method The data is drawn from several sources; most importantly a survey conducted between December and February 1999 (see Appendix B for a copy). Along with a similar survey of the commercialization histories of venture-backed firms, the data from this survey are being used to study a variety of phenomena associated with the incentives, strategies, and performance of research-oriented start-ups and, in particular, on the impact of the Federal SBIR subsidy on each of these issues. The sample is drawn from a list (compiled by the Small Business Administration) of the (approximately) 200 largest historical beneficiaries of SBIR grants. For survey participants, we requested information about their most successful project funded by the SBIR (for most firms, there was only one (or at most two) SBIR-funded projects which the firm considered technologically successful). By focusing on the projects for each firm

24

that overcame the substantial technological hurdles associated with innovation, the data provide information about the relative economic returns of projects in different industries that are at least successful in a technical sense. Approximately 50% of the surveys were conducted over the telephone. The remainder was completed through fax and regular mail (many surveys required follow-up telephone conversations to clarify ambiguities or to fill in missing data fields). The overall response rate to the survey was approximately 50%. While sample selection was not ideal, the degree of non-response seemed correlated with the level of effort devoted to identifying the individual at the firm who could answer the questions on the survey. Only a minority of non-responders reported either that no SBIR-funded project was commercialized successfully or that their non-response was based on concerns relating to secrecy or confidentiality (all respondents were ensured that their individual responses would be kept confidential). The surveys provide information both about the company who received the SBIR grant as well as the details of the SBIR-funded project. The first part of the survey gathers information about the size and background of the workforce (e.g., share of workforce with a Ph.D.) as well as some information about the criteria used to promote scientists and engineers. Second, the survey asks about the financial structure of the firm (e.g., share of the firm’s equity owned by venture capitalists or by top management) as well as the rules used to fund research and development activities (as well as other investment activities such as advertising). Data is also gathered about the internal authority structure of the firm (the composition of the board of directors) and the firm’s overall strategy and perception of its competitive advantage.

25

Firm-level information is complemented with detailed information about the commercialization

history

of

the

SBIR-funded

project.

These

questions

include

information about the nature of the technology and the underlying innovation (e.g., product or process innovation, development of novel system, the intellectual property and appropriability environment) as well as information about the product development strategy of the firm (e.g., time from conception of the new product to market introduction, upgrades and modifications since product introduction). Further, we break out the revenues associated with the project into several different categories, distinguishing direct product market sales from licensing revenue and the sale of intellectual property assets (e.g., patent exchanges and the like). For both product market sales and licensing revenue, the survey gathers information about the current-year revenue (which seemed to be much more reliable, consistent and complete for most firms) and the total revenues accruing to this product since its commercial introduction. Finally, the survey gathers information about the nature of the licensing (if it occurs) as well as the structure of decision-making associated with choosing how to commercialize the technology. From this survey, we construct both project-specific and firm-specific variables (see Table 1 for definitions). First, we define the project-level variables. We define performance in terms of the aggregate annual revenues from product sales, licensing, and intellectual property exchanges (REVENUE 98).9 For each project, we identify the number of PATENTS awarded since the grant as well as the project-level SBIR AWARD SIZE (from the USPTO and the Small Business Administration, respectively). In

9

All of the qualitative results are robust to using the data on total project revenues (indeed, they are highly correlated). However, this would further reduce the number of usable surveys by approximately one-third.

26

addition, we identify all firms who receive at least some of their overall revenues not through direct product market sales but through licensing arrangements or intellectual property sales. We denote these firms COOPERATORS insofar as nearly all of these firms are involved in cooperative contracting with more established product market incumbents; it is useful to note that this cooperative behavior is in lieu of competition with the same firms who have arranged to “buy out” the SBIR awardees’ product market position (Anton and Yao, 1995; Gans and Stern, 1998). We also include several dummy variables, which denote the technology, product and customer base types of each project (UPGRADE, MADE-TO-ORDER, NOVEL SYSTEM, and LARGE CUSTOMER BASE).

Finally, we calculate one measure of overall product development efficiency

(TIME-TO-MARKET) that is simply the overall time from initial product conception to the first sale of this product to any customer (either directly or through licensing). As well, we include several firm-specific variables that describe the more general financial and organizational structure of the firm. In terms of the measurement of performance, a portion of our analysis is organized around the determinants of current firm size (EMPLOYMENT 98). We use the initial value of this measure as a key control in most of our empirical analysis (BASELINE EMPLOYEES), in order to control for the initial size of the firm. As well, the analysis examines variables related to financial structure, VC EQUITY SHARE and INSIDER EQUITY SHARE. The difference between these is while the first confirms the potential importance of the “certification” hypothesis (SBIR grants lead to VC funding which leads to overall performance), the latter allows us to capture the pure associated between performance and maintaining a

27

closely

held

organization.

Similarly,

FOUNDER

CEO

proxies

for

the

overall

entrepreneurial culture and authority structure of the firm.

Sources and Definitions of Industry and Segment-Level Variables A critical element of the analysis is the relationship of project-level performance to measures of private and public investments in the industries, technology segments or scientific areas associated with each project. Specifically, we want to distinguish three concepts: overall investment in businesses or R&D by small firms, aggregate private R&D in the project’s industrial area, and the scientific and engineering opportunities present in the technological areas inherent in the project. To capture the first, we assigned each project to one of eleven technology segments identified by Venture One (see Figure 3A). For each segment, we measure the VC FUNDING STOCK as the (undiscounted) sum of venture investing in that segment between 1985 – 1992. We also present results for the VC CAPITAL FLOW, which are composed exclusively of the 1992 disbursements. We also assign each project (firm) to a single three-digit SIC; the NSF Science and Engineering Indicators provides data for each SIC on SIC-LEVEL R&D EXPENDITURES, SIC SALES, and SIC-LEVEL SMALL FIRM R&D EXPENDITURES. Finally, we measure the project’s association with scientific or engineering opportunity by constructing a SCIENCE STOCK for each project. First, we assigned each project to one or more scientific and engineering fields (out of a total of 14). In contrast to the mutually exclusive nature of the prior variables, the science base of each project can be composed from multiple sources. For each area (e.g., physics or chemical engineering), we computed the (discounted) Federal Funding

28

stock in that area (equal to 1990 FUNDING + .8 1988 FUNDING + .6 1986 FUNDING). Each project’s SCIENCE STOCK is simply the sum of the individual field-level stocks for fields associated with that project. In addition, we attempt to provide additional controls for the environmental heterogeneity among firms by including a number of variables from the survey that capture, at least to some degree, the firm’s perception of the appropriability environment relevant for the project under consideration. Specifically, we include several variables that relate to the relative importance of differences source of intellectual property (PATENTS, SECRECY, SPEED TO MARKET), each of which is measured as a 5-point Likert scale variable. As well, we include several assessments of the degree of importance

associated

with

the

control

of

different

complementary

assets

(MANUFACTURING, DISTRIBUTION, BRANDING, SERVICING). As with the IP measures, each is a five-point Likert Scale measure as reported in the survey.

Summary Statistics Out of a total 100 responses, 74 were fully usable for the empirical work reported here.10 However, for most of the analysis, we exclude the small number of observations (3) that were clearly so-called SBIR “mills”; these organizations have a much more diverse research portfolio from the rest of the sample and also have a different relationship with funding agencies. This leaves us with 71 total observations for the majority of our analysis.

10

Most of the non-responses were because of non-reporting of all 1998 revenues (e.g., sales reported but not licensing revenue. We have checked our results including all results, even those for whom the data was incomplete; the qualitative results are unchanged.

29

Among these 71, the average project reports approximately 6.5 million dollars in revenue, compared with an average award size of 1.5 million (but note the high standard deviations associated with each). As well, these organizations seem to be highly productive in a technical sense. On average, each has been issued over 9 patents that were applied for since the first SBIR grant associated with the specific project here (all chosen projects have their first funding date after 1990). In addition, almost a third of the sample earns at least some its revenue through cooperative licensing or IP exchanges (most often with product market incumbents). As well, the average commercialization length of projects is a little over four years, though some outliers in part drive this (from biotechnology and the like). Finally, at the firm level, while most of these organizations are quite small at inception (35 employees on average), they tend to experience substantial growth over time (the average firm has nearly 90 employees by 1998).11 As suggested earlier, the principal empirical exercise of this paper will be to relate project-level performance measures (REVENUE 98 and EMPLOYMENT 98) to segment-level measures of private investment activity and the technological environment. Figure 3 presents the distributions of the three principal measures we use to capture these effects: 1992 VC FUNDING STOCK, SCIENCE STOCK, and SIC-LEVEL R&D EXPENDITURES. Each of these measures provides a distinct way of capturing the degree of environmental heterogeneity facing different SBIR-funded projects in terms of the level of technological opportunity and realized private investment activity. Whereas the VC FUNDING STOCK (Figure 3A) measures the skewed distribution of 11

As well, in terms of the internal organization and finance of the firms, over 50% of equity is retained by insiders; perhaps surprisingly, while 25% of the sample had attracted some form of VC or “angel” financing, the equity share of these investors is quite small (the average level of equity held by these outside investors across the full sample is less than 7%). In addition, in nearly 60% of the firms, the CEO is

30

entrepreneurial activity and is divided into several segments which do not precisely map into the traditional SIC classification, Figure 3B and 3C measure differences in investment in alternative scientific and engineering fields upon which many of these innovations draw and differences across sectors ni terms of aggregate R&D expenditures, respectively. Whereas SCIENCE STOCK and R&D EXPENDITURES will tend to be sensitive to differences across sectors in terms of technological opportunity, only VC FUNDING STOCK will be sensitive to the differences across sectors in terms of the degree of appropriability or the salience of capital market constraints facing small research-oriented firms. Consistent with the SIC-level motivating statistics we presented in Table II, Figure 3 highlights both the skewed nature of both venture financing and R&D funding (e.g., telecommunications, medical care technologies, and software make up over 60% of the aggregate 1992 VC FUNDING STOCK). In terms of their overall summary statistics, while the VC FUNDING STOCK is measured as the cumulative investment in given areas over 8 years, one can see that the aggregate average level of SCIENCE STOCK and R&D EXPENDITURES are considerably higher, particularly given that SCIENCE STOCK depends on the summation across three years of investment and R&D EXPENDITURES is measured from a single cross-section. As well, the Appropriability Mechanism measures (control vraiables for the environment) are each Likert measures; not surprisingly, the mean of each varies from a little above 3 to just over 4. With these summary statistics in mind, we can turn the heart of our empirical analysis, to which we now turn.

one of the original founders of the firm. Finally, the firms are reasonably concentrated geographically(nearly half the companies are located in either California or Massachusetts).

31

VI.

Empirical Results We now turn to our evaluation of the principal hypotheses of this paper (recall

Proposition 1). Specifically, we are concerned with the degree of variation across sectors in technological opportunity (V(V )), the degree of appropriability (V(γ)), and the relative salience of capital constraints (V(θ)). Our empirical test provides evidence about the relative salience of these factors in explaining the large differences in R&D funding for small firms across industries that we highlighted in Section II. Specifically, we test for the relationship between project-level performance and the level of private investment in each project’s industry or technological segment, controlling for project-specific or firmspecific factors that also may affect performance. In Tables 3-5, we present our main finding

from

this

analysis:

controlling

for

project-level,

firm-level

and

other

environmental characteristics, revenue is increasing in the level of venture capital funding for small firms in a given technology segment and, more tentatively, in the science stock associated with a given technology. In contrast, performance is not statistically related to overall R&D investment in a given area or the overall size of an industrial segment. These basic findings are robust to several quite stringent tests of robustness, including the inclusion of additional controls for appropriability, project-level and firm-level factors, as well as variation in the measurement of the level of financing for small research-oriented firms (some of these results are presented in Appendix C). As well, in Table 6, we show that there is also a positive relationship between an alternative measure of performance (EMPLOYMENT 98) and the industry-specific level of venture financing.

32

The Performance of SBIR-funded Development Projects We begin in Tables 3 with the relationship between L REVENUE 98 and each of the alternative measures of private investment and technological opportunity. The most striking result is that, whereas SCIENCE STOCK, SIC R&D EXPENDITURES, and SIC SIZE are uncorrelated with project-level performance, 1992 VC FUNDING STOCK is both statistically significant and quantitatively important: doubling the level of venture funding in an industrial segment is associated with over a 50% increase in measured revenue (see (3-1)).

As well, in (3-5), we demonstrate that our principal result

concerning the VC FUNDING STOCK is robust to the inclusion of the alternative measures of the segment-level environment (which remain themselves insignificant and lower in magnitude). When considering these results, it is useful to recall that, relative to the

VC

FUNDING

EXPENDITURES

are

STOCK more

measure,

closely

SCIENCE

associated

with

STOCK variation

and in

SIC

R&D

technological

opportunity than the variation in the structural parameters associated with appropriability or capital market imperfection impacting small, entrepreneurial firms. While we defer our overall interpretation of this result until our conclusions, it is useful to note that, under the logic of Proposition 1, this result suggests that there is a salient difference across industrial segments in terms of their appropriability environment (i.e., V(γ) > 0), even after incorporating some more direct measures which would capture variation in technological opportunity (V(V )). Table 4 extends this analysis of the determinants of REVENUE 98 by focusing on the robustness of the VC FUNDING STOCK result to the inclusion of various projectlevel, firm-level, or alternative environmental controls. In (4-1), we include the simplest

33

“control,” the level of initial employment at each firm. In some sense, this very simple measure should capture some degree of the heterogeneity among firms in terms of their initial conditions in terms of generating products associated with a given level of revenue; however, its inclusion does not impact our main result at all. We then extend our analysis is (4-2) and (4-3) by including a number of alternative control variables, associated with the firm’s perception of the appropriability environment and project-level controls, respectively. At one level, the results from this analysis are interesting in their own right: while revenues are increasing in the technological quality and product development efficiency of the project (PATENTS, UPGRADE and TIME TO MARKET respectively) and whether firms “cooperate” with more established firms (note this could be in part the result of the selectivity of projects which are attractive for licensing), revenues are relatively unrelated to the Likert measures of the appropriability environment facing individual firms. However, the more important finding from Table 4 is that the coefficient associated with VC FUNDING STOCK remains roughly the same magnitude and statistically significant, despite the large number of control variables included and given the relatively small number of observations in the sample. Indeed, in (4-4), we include all of the control measures simultaneously with no impact on the underlying result regarding the VC FUNDING STOCK variable. Once again, given the logic of Proposition 1, these results suggest that there exists an important source of variation across sectors in terms of the degree of appropriability facing small firms, and that such differences are reflected in the ability of subsidized firms to earn returns on their innovations in specific industrial segments.

34

In Table 5, we turn back to the more general comparison between the VC FUNDING STOCK variable and measures which more closely correspond to technological opportunity (SCIENCE STOCK and SIC R&D EXPENDITURES). Specifically, in (5-1) and (5-2), we regress REVENUE 98 on each of the alternative measures (SCIENCE STOCK and SIC R&D EXPENDITURES, respectively), including all of the controls considered in Table 4. Interestingly, with these controls in place, the magnitude of the coefficient associated with SCIENCE STOCK increases substantially and is statistically significant (SIC R&D EXPENDITURES remains both relatively small and statistically insignificant). However, when, in (5-3), we include all of these variables together, the principal result associated with VC FUNDING STOCK remains robust, as does the SCIENCE STOCK result. In other words, even after we control simultaneously for various project-level controls as well as measures associated with technological opportunity and aggregate R&D expenditures, the highest performing SBIR-funded projects tend to be associated with those industrial segments with high rates of venture capital funding.12 While such a result may be interesting in terms of SBIR evaluation per se, perhaps its more important implication is that such covariation is only possible when differences across sectors are powerfully shaped by differences in the appropriability conditions facing small, entrepreneurially oriented firms.

12

Appendix Tables A-1 and A-2 further establish the robustness of the VC FUNDING result. In A-1, we include additional firm-level covariates, including measures of the financial organization of the firm, as well as additional environmental covariates associated with the geographic location of the firm. As before, while the SCIENCE STOCK variable is only significant in some specifications, the VC FUNDING STOCK remains significant at a similar magnitude. Finally, in A-2, we consider alternative sample selection schemes such as mandating that we only include projects whose 98 REVENUES are greater than the size of their SBIR grant or including the SBIR “mills” in the analysis. As before, our main qualitative result is confirmed: project-level revenues are strongly associated with segment-specific level of venture financing.

35

Finally, in Table 6, we compare our results regarding REVENUE 98 to an alternative

analysis

focused

on

the

determinants

of

firm-level

employment

(EMPLOYMENT 98). In some sense, this analysis is more directly comparable to the analysis pursued by Lerner (1999), in that Table 6 focuses on the determinants of firmlevel performance among SBIR-funded firms. The analysis is straightforward and echoes our earlier finding: while employment growth seems to be relatively unrelated to environmental

measures

EXPENDITURES,

there

associated is

a

with

quantitatively

SCIENCE important

STOCK and

or

SIC

statistically

R&D

significant

relationship between EMPLOYMENT 98 and VC FUNDING STOCK, even after controlling for BASELINE EMPLOYMENT (as well, this result is relatively robust to the inclusion of various controls, similar to the earlier analysis). In conjunction with our earlier findings regarding REVENUE 98, Table 6 provides additional evidence that there is a robust positive relationship between the performance of SBIR-funded projects and segment-specific levels of venture activity, a result which can be tied to the potential presence of differences across sectors in the appropriability environment facing small research-oriented firms.

VII. Conclusions and Interpretation Before turning to our interpretations, we emphasize the fragile nature of our results. The sample is small and based on an imperfect survey. Even with this limitation, however, several generalizations are possible. First and most importantly, our results suggest a very specific interpretation of the extreme concentration of venture capital financing in a small number of technology segments. Rather than simply being a

36

“herding” response to a diffuse sense of technological opportunity, such financing behavior seems to reflect that the economy offers small “pockets” where the appropriability regime facing small, research-oriented start-ups is particularly favorable. Understanding how such pockets emerge seems to be a promising area for further study. More broadly, our results can be tied to policy. While recent policy activity has focused on overcoming the “funding gap” for small, research-oriented firms, our results suggest

that

benefits

might

arise

from

turning

attention

to

strengthening

the

appropriability regime facing small firms. Finally, we return to our contention that our results are suggestive but not the final word. Beyond the small sample size, we can imagine that a more conclusive approach to this type of research would be to more carefully integrate the earlier approaches of Lerner, Wallsten and others (which compared the incremental benefits of SBIR grants) with the intra-grantee analysis presented here. Such a dataset would allow for more careful distinction between variation resulting from overall technological conditions, the fact of being an SBIR grantee, and policy-sensitive issues such as the appropriability regime facing small firms.

37

Appendix A: Sources of Capital Market Imperfections In Section III, we considered the effect of a capital subsidy. We argued that a capital subsidy effectively eliminates problems caused by capital market imperfections or at least equalizes them for grant recipients across industries. It is, therefore, important to consider in more detail the reasons why θ > 0 . We consider three alternative models based on moral hazard, adverse selection and expropriation. In each the capital market imperfection will arise when a particular parameter, φ, is positive. We will demonstrate that θ = 0 whenever the parameter φ is zero.

(a) Moral Hazard Suppose that the RU employs effort in innovative activity and that this can influence the likelihood of generating a successful innovation. We assume that this effort has a contractible and a non-contractible component. Using only contractible effort, the probability of generating a successful innovation is pL. We normalize the cost of this effort to 0. However, by expending non-contractible effort for a marginal cost of 1, the RU can raise this innovation probability to pH > pL. We make two simplifying assumptions. First, we assume that research and capital provision to the RU is socially desirable for some projects, i.e., p HV > F + 1 . Second, we assume that there exists F beyond which it is never worthwhile providing capital to the RU if it is not expected to expend a high level of effort in innovative activity, i.e., p Lγ V = F . The former assumption raises the possibility that some socially desirable projects may go unfunded because of a lack of appropriability. The latter assumption, however, means that the VC may not fund some privately profitable projects. This is essentially because of a capital market imperfection that dilutes the role of VC-equity when RU effort is non-contractible. Turning to the determinants of VC funding let α denote the level of equity the RU retains in its own firm. This level of equity (and any capital forthcoming from the VC) is determined prior to the RU engaging in any innovative activity. The minimum level of equity the RU can have and still expend a high level of effort ex post is given by the α that just satisfies the RU’s incentive constraint: pH αv − 1 ≥ pL v ⇒ α ≥

1 . v ( pH − pL )

Given the competitive nature of VC capital markets, VC capital will be forthcoming for the project so long as it is still profitable at this minimum RU-equity level.

p H (1 − α ) v ≥ F

38

Substituting for α, we can re-write the participation constraint for the VC as:

pH v −1 ≥ F +φ , where φ = pL /( pH − pL ) . Notice that, in this model, θ = φ ; meaning that the private cost of capital exceeds the social cost of capital whenever, φ > 0.

(b) Adverse Selection In the previous model, the capital market imperfection arose because of the detrimental effect of VC equity on the incentives of the RU to expend non-contractible effort in innovative activity. Here we consider an alternative model where the probability of a successful innovation depends on the RU’s ‘type’ that is private information to the RU. We suppose that there are two types of RU’s. RU’s with high ability generate a successful innovation with probability pH while for those with lower ability this probability is reduced to pL. There is no non-contractible RU effort. It is assumed that a given RU is a high type with probability 1-φ and this is commonly known. The potential value of the project, v, is, however, known. Finally, we maintain the assumption of the previous section that p HV > K and p Lγ V < K . Recall that the VC market is competitive, so RUs can demand the maximum level of equity for a given project, v. However, as they cannot signal their type, the maximum RU equity that either the high or low type can demand is:

( (1 − φ ) pH + φ p L ) (1 − α ) v = F ⇒ α =1 −

F ( (1 − φ ) pH + φ p L ) v

This means that the lowest value project that can be funded in an industry is determined by the v that results in α = 0; i.e., v = (1−φ ) pKH +φ pL . This implies that the total level of VC funding in the industry will be:  F F =  V − ( pH − φ ( p H − p L ) ) γ 

  

Notice that this is decreasing in φ implying that the inability of VCs to distinguish good RUs from bad results in a lower level of VC funding in the industry. Thus, for this model, θ = (1φ−(φp) HpH+ −pφL )pL . This equals 0 if φ = 0.

39

(c) Expropriation Our final model of capital market perfections is based on the potential for expropriation of ideas by a VC. When an RU approaches a given VC for funding they must reveal their potential idea. In some situations an RU will provide inputs that are required for that idea to become commercially viable. In other situations, however, the RU does not add any value in this sense. If property rights over the idea are weak (as they may be prior to any patents or copyright), then a VC may expropriate the idea and refuse to let the RU share in any returns. This is a fundamental difficulty in any trade in ideas (Arrow, 1962; and Anton and Yao, 1994) and it constrains RU appropriability in at every stage of innovation. We suppose that, provided a unit of capital is expended, the idea becomes commercializable with probability, p, and the potential return, v, is common knowledge. However, it is possible that the VC simply expropriate the full private return v if approached by the RU. Anticipating this, the RU may not develop the idea of approach the RU (although here there is no incentive for the RU not to bring the idea to a VC13 ). However, as Anton and Yao (1994) demonstrate, a capital constrained RU may have some alternative means of raising capital. This is certainly the case here given our assumption of a competitive VC market. If one VC were to expropriate the idea, the idea may be financed by another VC. To see this, recall that product market appropriation (γ) depends in part on the degree of competition an innovation faces. That is, are there products that are close substitutes? If the RU were to turn to another VC, it may be able to reduce product market appropriation to (1 − φ )γ ; even though it would not receive any ex post rents itself in this eventuality. Fearing the RU’s competitive threat from disclosure to others, the initial VC may not expropriate the RU; instead giving the RU some equity in the venture. This means that RU-equity will be determined through a bilateral negotiation between themselves and the first VC they approach. This equity is, therefore, not determined competitively but through a bilateral monopoly (Williamson, 19??). Essentially, both the RU and VC can use the threat of disclosure to others or expropriation to bind themselves to reach agreement. Following disclosure of the idea to the VC, the two parties negotiate over the equity level given to the RU. Assuming Nash bargaining with equal bargaining power this yields:

p (1 − α )γ V − F − ( p (1 − φ )γ V − F ) = pαγ V ⇒ α = 12 φ A project with social value, V, will be funded if and only if the VC’s participation constraints (expecting this level of equity) are satisfied. This participation constraint is 13

One could imagine that the RU has some sunk expenditures prior to bringing an idea to a VC that would be lost in the face of expropriation. This could be added to the model here but it would not alter the basic insight below about the high private cost of capital; it would merely complicated the cost of capital function.

40

pγ V ≥

θ=

φ 2 −φ

2 2 −φ

F . Consequently, the private cost of capital for this model is given by

F and it clearly exceeds 0 for φ ∈ (0,2] .

41

Figure 1A. Concentration of Private R&D Expenditures vs. Value-added (Lorenz curve, millions of 1992 dollars) Gini Coefficient = 0.604 1.000

0.900

0.800

0.700

0.600

0.500

0.400

0.300

0.200

0.100

0.000 0.000

0.100

0.200

0.300

0.400

0.500

0.600

Cumulative % Value-added

0.700

0.800

0.900

1.000

42

Figure 1B. Concentration of VC Disbursement vs. Private R&D Expenditures (Lorenz curve, millions of 1992 dollars) Gini Coefficient = 0.560

1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0 0

0.1

0.2

0.3

0.4

0.5

0.6

Cumulative % private R&D expenditures

0.7

0.8

0.9

1

43

3A. VC Funding Stock, 1992 14000 12000 10000 8000 6000 4000 2000 0

Industrial Segment

3B. R&D Expenditures, 1993 16000 14000 12000 10000 8000 6000 4000 2000 0 metal prod.

industrial prod.

electric. equip.

instruments

biotech

Industrial Segment (Note: R&D expenditures unreported for software & business services)

3C. Science Stock, 1990 8000 7000 6000 5000 4000 3000 2000 1000 0

Academic Discipline

transportation

44

4A. Performance by VC Stock VC Funding Stock 1992

Revenue 1998

18.00 16.00

12000

12.00

8000

10.00

6000

8.00 6.00

4000

4.00 2000

2.00 0.00 Tra ns po rta tion Bu sin es sS erv ice s Ind us tria lP rod uc ts Se m ico nd uc tor s Bio tec hn olo gy M ed ica lD ev ice s

Industrial Segment

Co m pu ter /IT

0

1998 Revenue ($M)

14.00 10000

En erg y

1992 VC Funding Stock ($M)

14000

45

4B. Performance by R&D Expenditures R&D Expenditures ($M)

Revenue 1998

12.00

14000

8.00

10000 8000

6.00

6000

4.00

4000 2.00

2000

tra ns po rta tio n

bio tec h

ins tru m en ts

ele ctr ic.

eq uip .

0.00 ind us tria lp rod .

0

Industrial Segment (Note: R&D expenditures unreported for software & business services)

1998 Revenues ($M)

10.00

12000

me tal pro d.

1993 R&D Expenditures ($M)

16000

46

TABLE 1 VARIABLES * & DEFINITIONS VARIABLE

DEFINITION

SOURCE

PROJECT-LEVEL PERFORMANCE REVENUE 98 EMPOYMENT 98

Total 1998 Revenues from Project Total Employees in 1998

MIT Survey MIT Survey

PROJECT- AND FIRM-LEVEL CONTROL VARIABLES PATENTS COOPERATOR

UPGRADE

TIME-TO-MARKET MADE-TO-ORDER SBIR AWARD SIZE BASELINE EMPLOYEES

Patents awarded since SBIR grant Dummy = 1 if Project Revenues include licensing revenues, intellectual property sales, or merger and acquisition, 0 else Dummy = 1 if Project Technology has been “substantially upgraded” since market introduction or prototype, 0 else Months From Conception of Intial Project Idea Until Product Market Introduction Dummy = 1 if Firm Sells “Made-to-order” Technologies, 0 else SBIR Grant Awards Related to Project Technology (both Phase I and Phase II included) Total Employees in Firm at Start of Project

USPTO MIT Survey

MIT Survey

MIT Survey MIT Survey MIT Survey MIT Survey

INDUSTRY OR SEGMENT-LEVEL VARIABLES 1992 VC FUNDING STOCK SCIENCE STOCK

SIC-LEVEL R&D EXPENDITURES

Total (undiscounted) stock of venture capital investment in technology segment most closely associated with project Total (discounted) stock of Federal expenditures on scientific and engineering areas closely related to project technology. See text for details. NB: Stock is linear in all areas related to project technology 1993 R&D Expenditures in Project 3-digit SIC

MIT Survey; Venture One MIT Survey; NSF S&E Indicators

MIT Survey; Corptech Directory; NSF S&E Indicators NSF S&E Indicators

SIC SIZE

1993 Total SIC Sales

LIKERT OF IMPORTANCE OF PATENT FOR APPROPRIABILITY LIKERT OF IMPORTANCE OF SECRECY FOR APPROPRIABILITY LIKERT OF IMPORTANCE OF BEING FIRST TO MARKET FOR APPROPRIABILITY

5-Point Managerial Likert Scale Rating of Importance of Patents for Appropriating Returns

MIT Survey

5-Point Managerial Likert Scale Rating of Importance of Secrecy for Appropriating Returns

MIT Survey

5-Point Managerial Likert Scale Rating of Importance of being First to Market to Appropriate Returns

MIT Survey

APPROPRIABILITY VARIABLES

47

COMPLEMENTARY ASSET VARIABLES LIKERT OF 5-Point Managerial Likert Scale Rating of the IMPORTANCE OF Importance of Control over Manufacturing in MANUFACTURING Earning Returns from the Project AS A COMPLEM. ASSET LIKERT OF 5-Point Managerial Likert Scale Rating of the IMPORTANCE OF Importance of Control over Distribution Channels DISTRIBUTION AS in Earning Returns from the Project A COMPLEM. ASSET LIKERT OF 5-Point Managerial Likert Scale Rating of the IMPORTANCE OF Importance of Control over Branding Resources BRANDING AS A in Earning Returns from the Project COMPLEMENTARY ASSET LIKERT OF 5-Point Managerial Likert Scale Rating of the IMPORTANCE OF Importance of Control over Servicing Resources SERVICING AS A in Earning Returns from the Project COMPLEMENTARY ASSET * The natural logarithm of a variable, X, will be denoted L X

MIT Survey

MIT Survey

MIT Survey

MIT Survey

48

TABLE 2

MEANS & STANDARD DEVIATIONS

VARIABLE

MEAN

STD. DEVIATION

PROJECT-LEVEL PERFORMANCE REVENUE 98

6.453

13.145

87.784

97.973

PATENTS

8.928

13.104

COOPERATOR

0.351

0.481

UPGRADE

0.838

0.371

TIME-TO-MARKET

45.000

42.142

MADE-TO-ORDER

0.622

0.488

SBIR AWARD SIZE

1.745

1.458

35.554

45.477

1992 VC FUNDING STOCK

2361.752

1380.537

SCIENCE STOCK

7129.043

4336.470

SIC-LEVEL R&D EXPENDITURES

4564.783

3014.836

62722.370

38679.860

EMPLOYMENT 98

PROJECT- AND FIRM-LEVEL CONTROL VARIABLES

BASELINE EMPLOYEES INDUSTRY OR SEGMENT-LEVEL VARIABLES

SIC SIZE APPROPRIABILITY MECHANISM VARIABLES LIKERT PATENT

3.617

1.344

LIKERT SECRECY

3.723

1.363

LIKERT SPEED

3.655

1.261

LIKERT MANUFACTURING

4.191

1.056

LIKERT DISTRIBUTION

3.490

1.214

LIKERT BRANDING

3.191

1.313

LIKERT SERVICING

3.426

1.229

49

TABLE 3 PROJECT-LEVEL PERFORMANCE EQUATIONS (NO PROJECT-LEVEL CONTROLS)

L 1992 VC FUNDING STOCK L SCIENCE STOCK

(3-1) VC STOCK only 0.567 (0.255)

Dependent Variable = L REVENUE 98 N= 71 observations, excludes “mills” (3-2) (3-3) (3-4) SCI STOCK SIC R&D SIC SIZE only only only

0.383 (0.267)

L SIC-LEVEL R&D EXPENDITURES L SIC SIZE Constant R-Squared Adjusted R-Squared

0.463 (0.347)

-3.680 (1.931) 0.067

-2.703 (2.308) 0.029

-3.220 (2.931) 0.041

0.251 (0.504) -2.112 (5.605) 0.020

0.054

0.015

0.013

-0.009

(3-5) Combination Model 0.586 (0.265) 0.281 (0.271) 0.278 (0.347) 0.400 (0.507) -12.975 (6.700) 0.125 0.058

50

TABLE 4 PROJECT-LEVEL PERFORMANCE AS A FUNCTION OF VC

Dependent Variable = L REVENUE 98 N= 71 observations, excludes “mills” (4-1) (4-2) (4-3) Control for Initial (4-1) with Likert (4-1) with ProjectFirm Size Appropriability and level Controls Complementary Asset Controls

(4-4) Combination Model

L 1992 VC FUNDING STOCK

0.537 (0.265)

0.574 (0.286)

0.550 (0.258)

0.510 (0.747)

BASELINE EMPLOYEES

0.125 (0.152)

0.096 (0.175) -0.249 (0.248) -0.150 (0.262) 0.039 (0.260) 0.446 (0.352) 0.060 (0.222) 0.034 (0.249) 0.348 (0.299)

0.117 (0.143)

0.135 (0.169) -0.094 (0.237) 0.028 (0.237) -0.141 (0.258) 0.266 (0.344) 0.182 (0.215) 0.069 (0.242) 0.200 (0.287) 0.030 (0.017) 0.711 (0.428) 0.961 (0.551) -0.006 (0.005) 0.387 (0.485) -0.286 (0.176) -7.717 (3.439) 0.411 0.176

LIKERT SERVICING LIKERT DISTRIBUTION LIKERT BRANDING LIKERT MANUFACTURING LIKERT PATENT LIKERT SECRECY LIKERT SPEED PATENTS

-3.763 (1.953) 0.081

-6.221 (2.830) 0.181

0.037 (0.015) 0.836 (0.380) 1.007 (0.499) -0.008 (0.004) 0.430 (0.415) -0.213 (0.165) -5.730 (2.353) 0.347

0.040

0.012

0.225

COOPERATOR UPGRADE TIME-TO-MARKET MADE-TO-ORDER L SBIR AWARD SIZE Constant R-Squared Adjusted R-Squared

51

TABLE 5 ALTERNATIVE INDUSTRY-LEVEL CORRELATES OF SBIR PERFORMANCE Dependent Variable = L REVENUE 98 N= 71 observations, excludes “mills” (5-1) (5-2) (5-3) Science Stock and Industry-level Combination Opportunity Expenditures and Model Measures Opportunity Measures

L SCIENCE STOCK

0.603 (0.279)

L SIC-LEVEL R&D EXPENDITURES

0.178 (0.360)

0.655 (0.295) -0.079 (0.348) 0.613 (0.278)

L 1992 VC FUNDING STOCK

FIRM-LEVEL CONTROLS PROJECT-LEVEL CONTROLS

YES YES

YES YES

YES YES

COMPL. ASSET CONTROLS

YES

YES

YES

APPROPRIABILITY REGIME CONTROLS Constant

YES

YES

YES

-7.605 (3.245) 0.420

-4.493 (3.746) 0.374

-12.656 (4.445) 0.474

0.188

0.106

0.217

R-Squared Adjusted R-Squared

52

TABLE 6 FIRM EMPLOYMENT EQUATIONS (NO PROJECT-LEVEL CONTROLS)

(6-1) VC STOCK only BASELINE EMPLOYMENT L 1992 VC FUNDING STOCK L SCIENCE STOCK

0.324 (0.010) 0.334 (0.174)

Dependent Variable = L EMPLOYMENT 98 N= 71 observations, excludes “mills” (6-2) (6-3) (6-4) SCI STOCK SIC R&D SIC SIZE only only only 0.377 (0.010)

R-Squared Adjusted R-Squared

0.373 (0.100)

0.156 (0.178)

L SIC-LEVEL R&D EXPENDITURES L SIC SIZE Constant

0.364 (0.102)

0.047 (0.235)

0.377 (1.279) 0.217

1.394 (1.591) 0.184

2.379 (1.965) 0.175

0.306 (0.338) -0.661 (3.792) 0.184

0.182

0.147

0.125

0.135

(6-5) Combination Model 0.339 (0.103) 0.380 (0.182) 0.128 (0.183) -0.065 (0.236) 0.421 (0.344) -5.252 (4.523) 0.244 0.160

53

TABLE A-1 EQUITY & GEOGRAPHY CONTROLS Dependent Variable = L REVENUE 98 N= 71 observations, excludes “mills” (A1-1) (A1-2) (A1-3) Control for VC Control for Exploring EQUITY SHARE Management and Geographic Effects Employee Equity Share L SCIENCE STOCK L SIC-LEVEL R&D EXPENDITURES L 1992 VC FUNDING STOCK

VC EQUITY SHARE

0.532 (0.328) -0.090 (0.352) 0.606 (0.283)

0.615 (0.315) -0.086 (0.347) 0.512 (0.285)

0.689 (0.304) 0.004 (0.369) 0.636 (0.284)

-0.012 (0.014)

INSIDER EQUITY SHARE

-0.010 (0.007)

LOCATED IN NY

1.560 (1.035) 0.251 (0.502) 0.050 (0.653)

LOCATED IN CA LOCATED IN MA

FIRM-LEVEL CONTROLS

YES

YES

YES

PROJECT-LEVEL CONTROLS

YES

YES

YES

COMPL. ASSET CONTROLS

YES

YES

YES

APPROPRIABILITY REGIME CONTROLS Constant

YES

YES

YES

-11.111 (5.254) 0.486

-10.529 (5.198) 0.501

-13.022 (4.476) 0.502

0.200

0.224

0.208

R-Squared Adjusted R-Squared

54

TABLE A-2 VARYING THE SAMPLE & THE VC MEASURE

Dependent Variable = L REVENUE 98 (A2-1) (A2-2) Excludes “mills” Includes “mills” in and imposes the sample product revenue threshold for inclusion in the sample N=43

N=74

(A2-3)

Excludes ”mills”

and

imposes

product

revenue

threshold

for inclusion in the sample

N=71 L SCIENCE STOCK

0.739 (0.502) -0.053 (0.504) 1.242 (0.520)

0.670 (0.288) -0.010 (0.349) 0.567 (0.268)

0.626 (0.299) -0.084 (0.357) 0.440 (0.253)

PROJECT-LEVEL CONTROLS

0.028 (0.308) 0.021 (0.021) YES

0.228 (0.170) 0.030 (0.017) YES

0.244 (0.176) 0.031 (0.018) YES

COMPL. ASSET CONTROLS

YES

YES

YES

APPROPRIABILITY REGIME CONTROLS Constant

YES

YES

YES

-16.940 (7.144) 0.546

-13.168 (4.446) 0.463

-10.159 (4.161) 0.455

0.003

0.216

0.189

L SIC-LEVEL R&D EXPENDITURES L 1992 VC FUNDING STOCK

BASELINE EMPLOYEES PATENTS

R-Squared Adjusted R-Squared

55

References

Aghion, P. and J. Tirole (1994), “The Management of Innovation,” Quarterly Journal of Economics, 109 (4), pp.1185-1210. Anand, B. and A. Galetovic (1999), “Weak Property Rights and Hold-up in R&D,” Working Paper, No.99-007, Harvard Business School. Anton, J.J. and D.A. Yao (1994), “Expropriation and Inventions: Appropriable Rents in the Absence of Property Rights,” American Economic Review, 84 (1), pp.190209. Arrow, K.J. (1962), “Economic Welfare and the Allocation of Resources for Invention,” in The Rate and Direction of Inventive Activity, Princeton University Press: Princeton, pp.609-625. Bush, V. (1945), Science: The Endless Frontier, US GPO. Cockburn, I. and Z. Griliches (1988). “Industry Effects and Appropriability Measures in the Stock Market's Valuation of R&D and Patents,” American Economic Review. 78 (2). p 419-23. Cockburn, I. and R. Henderson (1994), “Racing to Invest? The Dynamics of Competition in Ethical Drug Discovery,” Journal of Economics and Management Strategy, 3 (3), pp.481-519. Cohen, W. and D. Levinthal (1990), “Absorptive Capacity: A New Perspective on Learning and Innovation,” Administrative Science Quarterly, 35(1), pp. 128-152. Cohen, L. and R. Noll (1993), The Technology Pork Barrel, Brooking Institution: Washington (DC). Crepon, B., Duguet, E. and J. Mairesse (1998). “Research, Innovation and Productivity: An Econometric Analysis at the Firm Level,” Economics of Innovation & New Technology. 7 (2) p 115-58. General Accounting Office (1995), Federal Research: Interim Report on the Small Business Innovation Research Program. Gans, J.S. and S. Stern (1998), “Incumbency and R&D Incentives: Licensing the Gale of Creative Destruction,” mimeo., MIT. Gans, J.S. and S. Stern (1999), “Models of Corporate and Venture Capital,” mimeo., Melbourne.

56

Geroski, P.A. (1994). Market Structure, Corporate Performance and Innovative Activity.. Oxford University Press. Geroski, P., Machin, S. and J. Van Reenen (1993). “The Profitability of Innovating Firms,” Rand Journal of Economics 24 (2). p 198-211. Gilbert, R. and D. Newbery (1982), “Preemptive Patenting and the Persistence of Monopoly,” American Economic Review, 72 (3), pp.514-526. Gompers, P.A. and J. Lerner (1999). “What Drives Venture Capital Fundraising?,” NBER WP. Griliches, Z (1986). “Productivity, R&D, and the Basic Research at the Firm Level in the 1970's” American Economic Review. 76 (1). p 141-54. Griliches, Z. (1998). R&D and Productivity. University of Chicago Press. Griliches, Z., Hall, B.H. and A. Pakes (1991). “R&D, Patents, and Market Value Revisited: Is There a Second (Technological Opportunity) Factor?,” Economics of Innovation & New Technology. 1 (3). p 183-201. Hall, B.H. (1992). “Investment and Research and Development at the Firm Level: Does the Source of Financing Matter?,” NBER WP 4096. Hall, Bronwyn H. (1988). “The Effect of Takeover Activity on Corporate Research and Development.. Corporate takeovers: Causes and consequences.” Auerbach, Alan J., ed, NBER Project Report series, University of Chicago Press. p 69-96. Hall, B.H. and J. van Reenen (1999). “How Effective are Fiscal Incentives for R&D? A Review of the Evidence,” NBER WP ***. Hao, K.Y. and A.B. Jaffe (1993). “Effect of Liquidity on Firms' R&D Spending,” Economics of Innovation & New Technology. 2 (4) p 275-82. Hellmann, T. (1998a), “The Allocation of Control Rights in Venture Capital Contracts,” RAND Journal of Economics, 29 (1), pp.57-76. Hellmann, T. (1998b), “A Theory of Corporate Venture Investing,” mimeo., Stanford. Hellmann, T. and M. Puri (1998), “The Interaction Between Product Market and Financing Strategy: The Role of Venture Capital,” mimeo., Stanford. Henderson, R. (1993), “Underinvestment and Incompetence as Responses to Radical Innovation: Evidence from the Photolithographic Alignment Equipment Industry,” RAND Journal of Economics, 24 (2), pp.248-270. Himmelberg, C.P. and B. Petersen (1994). “R&D and Internal Finance: A Panel Study of Small Firms in High-Tech Industries,” Review of Economics & Statistics. Vol. 76 (1). p 38-51.

57

Holmstrom, B. (1989), “Agency Costs and Innovation,” Journal of Economic Behavior and Organization, 12 (2), pp.305-327. Hubbard, R.G. (1998), “Capital-Market Imperfections and Investment,” Journal of Economic Literature, 36 (1), pp.192-225. Kortum, S. and J. Lerner (1998), “Does Venture Capital Spur Innovation,” Working Paper No.6846, NBER. Klenow, P.J. (1996). “Industry Innovation: Where and Why,” Carnegie-Rochester Conference Series on Public Policy. 44 (0). p 125-50. Klette, T. J., Moen, J. and Z. Griliches (1999). “Do Subsidies to Commercial R&D Reduce Market Failures? Microeconomic Evaluation Studies,” NBER WP Lerner, J. (1996), “The Government as Venture Capitalist: The Long-Run Impact of the SBIR Program,” Working Paper, No. 5753, NBER. Lerner, J. (1997), “An Empirical Analysis of a Technology Race,” RAND Journal of Economics, 28 (3), pp.228-247. Lerner, J. and R. Merges (1997), “The Control of Technology Alliances: An Empirical Analysis of the Biotechnology Industry,” NBER Working Paper No. 6014. Launjouw, J. (1994), “Economic Consequences of a Changing Litigation Environment: The Case of Patents,” NBER Working Paper No. 4835. Levin, R., et. al. (1987). “Appropriating the Returns from Industrial Research and Development,” Brookings Papers on Economic Activity, 3, pp. 783-831. Mansfield, Edwin (1995). Innovation, technology and the economy: Selected essays of Edwin Mansfield. Economists of the Twentieth Century series. Elgar (UK). Nelson, R.R. (1959), “The Simple Economics of Basic Scientific Research,” Journal of Political Economy, pp.297-306. Rasmusen, E. (1988), “Entry for Buyout,” Journal of Industrial Economics, 36 (3), pp.281-299. Reinganum, J.F. (1983), “Uncertain Innovation and the Persistence of Monopoly,” American Economic Review, 73 (4), pp.741-748. Reinganum, J.F. (1989), “On the Timing of Innovation,” in R. Schmalansee and R. Willig (eds.), Handbook of Industrial Organization, Vol.1, Elsevier, pp. 849-908. Romer, P. (1990), “Endogenous Technological Change,” Journal of Political Economy, Schumpeter, J.A. (1942), Capitalism, Socialism and Democracy, Harper: New York. Stiglitz, J.E. and A. Weiss (1981), “Credit Rationing in Markets with Imperfect Information,” American Economic Review, 71 (3), pp.393-410.

58

Teece,

D.J. (1987), “Profiting from Technological Innovation: Implications for Integration, Collaboration, Licensing, and Public Policy,” in D.J. Teece (ed.), The Competitive Challenge: Strategies for Industrial Innovation and Renewal, Ballinger: Cambridge (MA), pp.185-220.

Wallsten, S.J. (1997), “The Small Business Innovation Research Program: Encouraging Technological Innovation and Commercialization in Small Firms?” mimeo., Stanford.