when does funding research by smaller firms bear fruit?: evidence

8/14/2019 When Does Funding Research by Smaller Firms Bear Fruit?: Evidence

1/63

When Does Funding Research by Smaller Firms Bear Fruit?:

Evidence from the SBIR Program*

by

Joshua S. Gans andScott Stern**

First Draft: April 2, 1999This Version: November 24, 2009

This paper evaluates whether the relative concentration of funding for small, research-oriented firms in some important high-tech industries is related to the differences acrossindustries in the appropriability regime facing small firms. We explore this hypothesis byanalyzing the performance of projects funded by the Small Business Innovation Research(SBIR) program, a Federal subsidy that provides R&D funds for small businesses. Bydesign, the cost of R&D capital is (approximately) equalized for firms funded by theSBIR. As a result, we neutralize capital market imperfections as a direct source ofvariation in our sample. In contrast, the SBIR does notaffect sector-level differences inthe appropriability regime or in the underlying level of technological opportunity. Themain contribution of this paper results from exploiting this difference to evaluate thesalience of capital market imperfections, the appropriability regime facing small firms,and the overall level of technology opportunity. Specifically, if the SBIR fund projects onthe margin (as it should under an optimal subsidy regime), then a cross-sectional

comparison identifies the relative importance of capital versus product marketimperfections across markets. Our principal empirical result is that project-levelperformance is highest for those technologies that are in industrial segments that attracthigh rates of venture capital investment. As well, there is weak but positive evidence thatperformance is related to the overall level of scientific opportunity. We interpret thesefindings as suggesting that an important difference between industrial sectors is thedegree of appropriability for research-oriented small businesses; variation in theappropriability regime helps explain the concentrated nature of venture capital activity inthe economy. Journal of Economic Literature Classification Numbers:

* 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, andespecially Iain Cockburn, provided thoughtful suggestions and advice. David Hsu provided outstandingresearch assistance. We also gratefully acknowledge funding for this research by the MIT Center forInnovation in Product Development under NSF Cooperative Agreement # EEC-9529140. Any remainingerrors or omissions are our responsibility.** Melbourne Business School, University of Melbourne, and MIT Sloan School & NBER, respectively. Allcorrespondence to Scott Stern, Sloan School of Management, MIT, Cambridge (MA), 02142;mailto:[email protected]. The latest version of this paper is available athttp://www.mbs.unimelb.edu.au/jgans/research.htm.
mailto:[email protected]://www.mbs.unimelb.edu.au/jgans/research.htmmailto:[email protected]://www.mbs.unimelb.edu.au/jgans/research.htm


2/63

Keywords: innovation, appropriability, subsidy, capital constraints, technologicalopportunity, venture capital.


3/63

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


4/63

2

appropriability constraints in a small number of distinct market and technological

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


5/63

3

margin, then one would expect that grants provided to firms in capital-constrained

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 industrialsegments.By examining performance differences across subsidized projects, this paper

differs from most prior treatments that have attempted to evaluate inter-sectoral

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


6/63

4

capital market imperfections from more downstream issues such as the appropriablity

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 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 IV then describes the SBIR program and the features that allow it to provide an

experiment to control for capital market imperfections across sectors. 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


7/63

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

their rank of& j

j

R D

VALUE ADDED.

2 Indeed, the Gini coefficient of inequality is over .6, a relatively high rate of inequality across sectors interms of their R&D expenditures.3 Indeed, while economists have relatively ignored the consequences of this observation, its validity is awidely 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 venturecapital in particular and towards the somewhat more comparable concept of small-to-medium sized R&Dinvestment (tabulations available from the authors).


8/63

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


9/63

7

within given sectors, the opportunity hypothesis is perhaps a persuasive baseline

hypothesis for considering the sources of differences across sectors. Indeed, one of the

Discussion of Adams (1990), Levin et al (1987), Mowery and Rosenberg (198*),

Toole (1999), others (????).

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


10/63

8

control and monitoring by financiers thereby diluting the innovation incentives of small

firms.

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).

DISCUSSION OF DISCLOSURE

DISCUSSION OF EMPIRICAL LITERATURE (SEE REFERNCES

LIST!!)

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 and others (????).


11/63

9

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

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).

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


12/63

10

CONCLUSION (DO WE WANT TO DRAW FROM THE OLD END (NEXT

TWO PARAS) TO THE SBIR DESCRIPTION SECTION PRIOR TO GOING TO THE

MODEL SECTION AS BELOW?

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 which 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 the next section, we refine this intuition by demonstrating how each of these

effects depends in part on how the SBIR selects projects. In particular, we show that the

relationship between sector and performance depends critically on whether the SBIR

chooses projects which are the best available (a la Wallsten) or in fact funds projects on

the margin of profitability at the industry level (as would be optimal from the viewpoint


13/63

11

of theory). Consequently, the wedge between SBIR funding and private venture funding

provides an experiment insomuch as we observe a shift in the capital constraint,

holdingthe appropriability regime constant.

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 research-

oriented 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

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.


14/63

12

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

VCs 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.

An innovation, if generated, has a social value of Vof which the RU (and their

equity partners) can only appropriate ( , )iv V where i parameterizes the distance

between V and v in industry i. v(.) is assumed to be non-decreasing in V and non-

increasing in i. i is, therefore, a measure of the lack of appropriability in industry i.

As discussed earlier (in Section II), previous research has identified many potential

determinants of i 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.

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.


15/63

13

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, ]iV . The parameter, iV , is, therefore, a measure of the level

of technological opportunity in industry i. Hence, one would expect average performance

to be higher in industries with larger iV .

Determining VC Funding

Suppose that funding a research product in industry i costs ( , )i ic F where Fi is

the total level of industry funding, and i parameterizes the difference between the

private and social cost of capital for industry i. We assume that c(.) is non-decreasing in

both of its arguments. That is, the supply curve of capital for small firm research in the

industry slopes upward and is increase in the level of capital market imperfections, i.

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

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, Fi is

determined by the distribution of project value in the industry. So, ifF projects are

funded, the marginal project has social value Vdetermined by i iF V V = . The expected

marginal private return when Fi projects are funded is, therefore, ( , )i i i i pv pv V F = .

This is the (inverse) demand for funding of RUs in the industry.


16/63

14

Figure 2 graphically depicts the supply and demand for RU funding in an

industry. In equilibrium, the level of industry funding, iF, is determined by:

( , ) ( , )i i i i ipv V F c F =

Notice that iF is increasing in iV but falling in i and i. That is, an increase in iV or a

decrease in shifts demand and hence iF, upwards, while an increase in i, shifts

supply upwards and reduces iF. Thus, industries with high appropriability, greater

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.

Figure 2: Equilibrium Industry Funding

i

FTotal Funding

ExpectedProject Value

0

c(Fi, i)

( , )i i i pv V F

iv


17/63

15

Notice that a project will be undertaken only if ( , ) ( , )i i ipv V c F . This differs

from the social criteria for project desirability: ( ,0)ipV c 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 ( i > 0) and a greater private

cost of capital.

At this point, it is important to note that some potential financiers of the project

might have a superior ability to appropriate the rents from innovative activity. For

example, larger incumbent corporations have a superior product market position to

smaller firms (Gans and Stern, 2000). Such organizations may be able to integrate with

the RU (owning any innovative output produced) and finance the project even though this

would reduce the probability an innovation is generated (Aghion and Tirole, 1994). This

means that total funding of research and development in an industry is likely to be driven

by V than by v, as the latter may carry distributional consequences that rule out these

alternative funding sources (see Hellman, 1998b; Gans and Stern, 1999).

6

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,

( , , )i i i iF f V = . That is the random variable, iF, is a function of the industry levels of

technological opportunity ( iV), appropriability ( i) and capital market imperfections

( i). Each of these variables determines either the demand or supply of funds to the

6 For example, incumbents may not finance innovations that cannibalize their existing assets (Arrow, 1962).


18/63

16

industry. Therefore, to discover which of these might account for variations in industry

VC 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 ( iF)

is readily observable, there are difficulties in determining the price variable (the expected

profitability of the marginal project, iv ) 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:


19/63

17

( , ) i i

i

c Fv

p

=

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 ( i) and the space of technological opportunities ( iV) but decreasing in the

lack of appropriability ( i). Importantly, given its role in determining the supply of

funds, changes in i move iF and iv 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 iF and iv to be negatively correlated.

The following proposition summarizes this basic intuition.

Proposition 1. Suppose that Var( ) Var( ) 0i iV = = while Var( ) 0i > , then cov[ , ] 0i iv F while Var( ) 0i = , then cov[ , ] 0i iv F > .

This result follows from the fact that if the only variation is in i, then only the c(.) curve

differs across industries while the pv(.) curve remains the same. As such, ( , )i iv F pairs

will lie along the pv(.) curve which is downward sloping. Hence, observed iF and iv

would be negatively correlated. In contrast, if the only variation was in iV or i, then the

c(.) would be unchanged across industries and the only differences would be in the

position of thepv(.). This means that ( , )i iv F pairs will lie along the upward sloping c(.)

curve. Consequently, iF would be positively correlated with iv .


20/63

18

By assuming a specific functional form for c(.) and v(.) we can go further and

clarify the relationship between the covariance of iF and iv and the variances of iV , i

and i.

Proposition 2. Suppose that ( , )i i i ic F F = + and ( , )i iv V V = and that iV , i and

i are independent. Then [ ]( ) [ ]( )21(1 )cov[ , ] Var Var Var i i i i ipv F p V + = .

PROOF: For this case, from , 1 ( )pi i i ipF V += and, from ,

( )1(1 ) ( )i i i ip pv p V += + . Then:

( ) ( )

[ ] [ ]( )2

2

1 1(1 ) 1 (1 ) 1

2 2 22 2 21

(1 )

1

(1 )

cov[ , ] ( ) ( ) ( ) ( )

Var

p p

i i i i i i i i i i i i i i p p p p p p

i i i i i ip

ip

v F E p V V E p V E V

pE V pE E pE V E E

p V

+ + + +

+

+

= + +

= + + +

= [ ]( ) [ ]( )Var Var i i

where the second step follows from the independence assumption.

For the linear case, therefore, the covariance of iF and iv is negatively correlated if and

only if the variance of i exceeds the variance inpv(.).

In summary, by looking to the SBIR program for information regarding the

returns of marginal projects ( iv ), we can use the correlation between this and observed

VC funding ( iF) 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


21/63

19

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 ( i) 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 Valone.

This is in contrast to the marginal rule that implicitly considers i 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

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 ( iV) or

higher product market appropriability ( i), 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.


22/63

20

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 governments 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 offunded

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

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


23/63

21

percentage of their funds for extramural grants to fund R&D projects by small business.7

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 minority-

owned businesses in the Federal contracting process (GAO, 1995).

From a political perspective, the programs support seems to derive in part from (ever-

increasing) 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

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 earlystage 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

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


24/63

22

Wallsten (1997), the programs 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).8 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

group of SBIR-funded firms and a matched sample of firms who are observationally

similarex 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

8 An additional critique is that a small number of firms have been able to win a disproportionate share ofoverall grants. While the absolute importance of these so-called SBIR mills is limited, the potential forregulatory capture by these firms has led to interest in limits on the total number of grants for which asingle firm might be eligible (GAO, 1995). While we do not specifically address mini-mills per se, we dodemonstrate that our results are robust to their inclusion or exclusion.


25/63

23

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 programs ability to alleviate capital market imperfections in the funding

of research; and

(3) the programs 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 economys

activities, the expenditures of the SBIR program are much more widely dispersed (across

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


26/63

24

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

or business models, the SBIR granting process is focused on funding technology

development per se and contributes essentially no additional commercialization services.9

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

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


27/63

25

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, holdingthe 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.

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

firms 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


28/63

26

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

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


29/63

27

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 firms 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 firms

overall strategy and perception of its competitive advantage.

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,


30/63

28

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).10 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

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

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


31/63

29

(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

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


32/63

30

R&D in the projects 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 projects 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

stock in that area (equal to 1990 FUNDING + .8 1988 FUNDING + .6 1986 FUNDING).

Each projects 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 firms 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


33/63

31

(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.

11

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.

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

11 Most of the non-responses were because of non-reporting of all 1998 revenues (e.g., sales reported butnot licensing revenue. We have checked our results including all results, even those for whom the data wasincomplete; the qualitative results are unchanged.


34/63

32

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).12

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

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 in 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

12 As well, in terms of the internal organization and finance of the firms, over 50% of equity is retained byinsiders; perhaps surprisingly, while 25% of the sample had attracted some form of VC or angelfinancing, the equity share of these investors is quite small (the average level of equity held by theseoutside investors across the full sample is less than 7%). In addition, in nearly 60% of the firms, the CEO isone of the original founders of the firm. Finally, the firms are reasonably concentratedgeographically(nearly half the companies are located in either California or Massachusetts).


35/63

33

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.

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 (Var(V )), the degree of appropriability (Var( ) ), and the

relative salience of capital constraints (Var() ). 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 projects industry or technological segment, controlling for project-

specific or firm-specific factors that also may affect performance. In Tables 3-5, we


36/63

34

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.

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


37/63

35

lower in magnitude). When considering these results, it is useful to recall that, relative to

the VC FUNDING STOCK measure, SCIENCE STOCK and SIC R&D

EXPENDITURES are more closely associated with variation in 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.,

Var( ) > 0), even after incorporating some more direct measures which would capture

variation in technological opportunity (Var(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 project-

level, firm-level, or alternative environmental controls. In (4-1), we include the simplest

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 firms 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


38/63

36

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 ofsubsidized firms to earn returns on

their innovations in specific industrial segments.

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


39/63

37

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.13 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.

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 firm-

level 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 associated with SCIENCE STOCK or SIC R&D

EXPENDITURES, there is a quantitatively important and statistically 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

13 Appendix Tables A-1 and A-2 further establish the robustness of the VC FUNDING result. In A-1, weinclude additional firm-level covariates, including measures of the financial organization of the firm, aswell 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 FUNDINGSTOCK remains significant at a similar magnitude. Finally, in A-2, we consider alternative sampleselection schemes such as mandating that we only include projects whose 98 REVENUES are greater thanthe size of their SBIR grant or including the SBIR mills in the analysis. As before, our main qualitativeresult is confirmed: project-level revenues are strongly associated with segment-specific level of venturefinancing.


40/63

38

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

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)


41/63

39

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.


42/63

40

Appendix A: Sources of Capital Market Imperfections

NEED TO CHANGE ALL OF THIS TO REFLECT NEW NOTATION ... InSection III, we considered the effect of a capital subsidy. We argued that a capitalsubsidy effectively eliminates problems caused by capital market imperfections or at leastequalizes them for grant recipients across industries. It is, therefore, important to considerin more detail the reasons why 0> . We consider three alternative models based onmoral hazard, adverse selection and expropriation. In each the capital market

imperfection will arise when a particular parameter, , is positive. We will demonstrate

that = 0 whenever the parameter is zero.

(a) Moral Hazard

Suppose that the RU employs effort in innovative activity and that this caninfluence the likelihood of generating a successful innovation. We assume that this efforthas a contractible and a non-contractible component. Using only contractible effort, theprobability of generating a successful innovation is pL. We normalize the cost of thiseffort to 0. However, by expending non-contractible effort for a marginal cost of 1, theRU can raise this innovation probability topH>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., 1Hp V F > + . Second, weassume 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.,

Lp V F = . The former assumption raises the possibility that some socially desirableprojects 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 isessentially because of a capital market imperfection that dilutes the role of VC-equitywhen RU effort is non-contractible.

Turning to the determinants of VC funding let denote the level of equity theRU retains in its own firm. This level of equity (and any capital forthcoming from theVC) is determined prior to the RU engaging in any innovative activity. The minimumlevel of equity the RU can have and still expend a high level of effort ex post is given by

the that just satisfies the RUs incentive constraint:

p v p vv p p

H L

H L

11

( ).

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


43/63

41

(1 )Hp v F

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

1Hp v F + ,

where /( ) L H Lp p p= . Notice that, in this model, = ; meaning that the private costof capital exceeds the social cost of capital whenever, > 0.

(b) Adverse Selection

In the previous model, the capital market imperfection arose because of thedetrimental effect of VC equity on the incentives of the RU to expend non-contractibleeffort in innovative activity. Here we consider an alternative model where the probabilityof a successful innovation depends on the RUs type that is private information to theRU. We suppose that there are two types of RUs. RUs with high ability generate asuccessful innovation with probability pH while for those with lower ability thisprobability is reduced topL. There is no non-contractible RU effort. It is assumed that a

given RU is a high type with probability 1- and this is commonly known. The potentialvalue of the project, v, is, however, known. Finally, we maintain the assumption of the

previous section that Hp V K > and Lp V K < .

Recall that the VC market is competitive, so RUs can demand the maximum levelof equity for a given project, v. However, as they cannot signal their type, the maximumRU equity that either the high or low type can demand is:

( )

( )

(1 ) (1 )

1(1 )

H L

H L

p p v F F

p p v

+ = =

+

This means that the lowest value project that can be funded in an industry is determined

by the v that results in = 0; i.e., (1 ) H LK

p pv

+= . This implies that the total level of VC

funding in the industry will be:

( )( ) H H L

FF V

p p p

=

Notice that this is decreasing in implying that the inability of VCs to distinguish goodRUs from bad results in a lower level of VC funding in the industry. Thus, for this model,

( )

(1 )H L

H L

p p

p p

+

= . This equals 0 if = 0.


44/63

42

(c) Expropriation

Our final model of capital market perfections is based on the potential forexpropriation of ideas by a VC. When an RU approaches a given VC for funding theymust 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, theRU does not add any value in this sense. If property rights over the idea are weak (as theymay be prior to any patents or copyright), then a VC may expropriate the idea and refuseto 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 everystage of innovation.

We suppose that, provided a unit of capital is expended, the idea becomescommercializable 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 ifapproached by the RU. Anticipating this, the RU may not develop the idea of approachthe RU (although here there is no incentive for the RU not to bring the idea to a VC14).However, as Anton and Yao (1994) demonstrate, a capital constrained RU may havesome alternative means of raising capital. This is certainly the case here given ourassumption of a competitive VC market. If one VC were to expropriate the idea, the ideamay be financed by another VC.

To see this, recall that product market appropriation ( ) depends in part on thedegree of competition an innovation faces. That is, are there products that are closesubstitutes? If the RU were to turn to another VC, it may be able to reduce product

market appropriation to (1 ) ; even though it would not receive any ex post rents itselfin this eventuality. Fearing the RUs competitive threat from disclosure to others, theinitial 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, notdetermined competitively but through a bilateral monopoly (Williamson, 19??).Essentially, both the RU and VC can use the threat of disclosure to others orexpropriation to bind themselves to reach agreement.

Following disclosure of the idea to the VC, the two parties negotiate over theequity level given to the RU. Assuming Nash bargaining with equal bargaining powerthis yields:

( )1

2

(1 ) (1 )p V F p V F p V

=

=

A project with social value, V, will be funded if and only if the VCs participationconstraints (expecting this level of equity) are satisfied. This participation constraint is

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

8/14/2019 When Does Funding Research by Smaller Firms Bea

when does funding research by smaller firms bear fruit?: evidence

Documents