WHY DO VENTURE

CAPITAL FIRMS EXIST?

THEORY AND CANADIAN
EVIDENCE
RAPHAEL AMIT, JAMES BRANDER,
AND CHRISTOPH ZOTT
University of British Columbia, Vancouver,
British Columbia, Canada

This paper investigates the role of venture capitalists. We view their “raison
EXECUTIVE d’être” as their ability to reduce the cost of informational asymmetries. Our
SUMMARY theoretical framework focuses on two major forms of asymmetric informa-
tion: “hidden information” (leading to adverse selection) and “hidden ac-
tion” (leading to moral hazard). Our theoretical analysis suggests four em-
pirical predictions.

1. Venture capitalists operate in environments where their relative efficiency in selecting and monitoring
investments gives them a comparative advantage over other investors. This suggests strong industry
effects in venture capital investments. Venture capitalists should be prominent in industries where infor-
mational concerns are important, such as biotechnology, computer software, etc., rather than in “rou-
tine” start-ups such as restaurants, retail outlets, etc. The latter are risky, in that returns show high
variance, but they are relatively easy to monitor by conventional financial intermediaries.
2. Within the class of projects where venture capitalists have an advantage, they will still prefer projects
where monitoring and selection costs are relatively low or where the costs of informational asymmetry
are less severe. Thus, within a given industry where venture capitalists would be expected to focus,
we would also expect venture capitalists to favor firms with some track records over pure start-ups.

Address correspondence to Professor Raphael Amit, Faculty of Commerce, 2053 Main Mall, University of
British Columbia, Vancouver, BC V6T 1Z2, Canada (e-mail: ⬍[email protected]⬎; phone: (604) 822-8481).
We thank two anonymous referees and the editor for very helpful comments. We also thank Paul Gompers,
who served as the discussant on the paper at the Economic Foundations of Venture Capital Conference held at
Stanford University in March 1997. In addition, we owe a substantial debt to Mary Macdonald and Ted Liu of Mac-
donald & Associates Ltd. for providing access to the data. (Individual data records were provided on an anonymous
basis.) We gratefully acknowledge financial support from Social Sciences and Humanities Research Council of Can-
ada (SSHRC) grant no. 412-93-0005, and from Industry Canada. C. Zott also acknowledges financial support from
Doktorandenstipendium aus Mitteln des zweiten Hochschulsonderprogramms (DAAD). The authors are associated
with the W. Maurice Young Enterpreneurship and Venture Capital (EVC) Research Centre at UBC. The EVC
web page is located at http://pacific.commerce.ubc.ca/evc/

Journal of Business Venturing 13, 441–466

 1998 Elsevier Science Inc. All rights reserved. 0883-9026/98/$19.00
655 Avenue of the Americas, New York, NY 10010 PII S0883-9026(97)00061-X
442 R. AMIT ET AL.

To clarify the distinction between point 1 and point 2, note that point 1 states that if we look across
investors, we will see that venture capitalists will be more concentrated in areas characterized by sig-
nificant informational asymmetry. Point 2 says that if we look across investment opportunities, venture
capitalists will still favor those situations which provide better information (as will all other investors).
Thus venture capitalists perceive informational asymmetries as costly, but they perceive them as less
costly than do other investors.
3. If informational asymmetries are important, then the ability of the venture capitalist to “exit” may be
significantly affected. Ideally, venture capitalists will sell off their share in the venture after it “goes
public” on a stock exchange. If, however, venture investments are made in situations where informa-
tional asymmetries are important, it may be difficult to sell shares in a public market where most invest-
ors are relatively uninformed. This concern invokes two natural reactions. One is that many “exits”
would take place through sales to informed investors, such as to other firms in the same industry or to
the venture’s own management or owners. A second reaction is that venture capitalists might try to acquire
reputations for presenting good quality ventures in public offerings. Therefore, we might expect that the
exits that occur in initial public offerings would be drawn from the better-performing ventures.
4. Finally, informational asymmetries suggest that owner-managers will perform best when they have
a large stake in the venture. Therefore, we can expect entrepreneurial firms in which venture capitalists
own a large share to perform less well than other ventures. This is moral hazard problem, as higher values
of a venture capitalist’s share reduce the incentives of the entrepreneur to provide effort. Nevertheless, it
might still be best in a given situation for the venture capitalist to take on a high ownership share, since
this might be the only way of getting sufficient financial capital into the firm. However, we would still
expect a negative correlation between the venture capital ownership share and firm performance.
Our empirical examination of Canadian venture capital shows that these predictions are consistent with
the data. In particular, there are significant industry effects in the data, with venture capitalists having
disproportionate representation in industries that are thought to have high levels of informational asym-
metry. Secondly, venture capitalists favor later stage investment to start-up investment. Third, most exit
is through “insider” sales, particularly management buyouts, acquisitions by third parties, rather than
IPOs. However, IPOs have higher returns than other forms of exit. In addition, the data exhibit the nega-
tive relationship between the extent of venture capital ownership and firm performance predicted by our
analysis.  1998 Elsevier Science Inc.

INTRODUCTION
In both Canada and the United States, venture capital finance is a significant form of
financial intermediation. There is no strict regulatory definition of the venture capital
industry, unlike commercial banking or insurance but, generally speaking, venture capi-
tal firms provide privately held “entrepreneurial” firms with equity, debt, or hybrid
forms of financing, often in conjunction with managerial expertise. In Canada these
firms are playing an increasingly important role. As reported in Macdonald & Associ-
ates (1996), between the end of 1991 and the end of 1995, the amount of capital under
management by Canadian venture capital firms grew from C$3.2 billion (or about $3.8
billion in 1995 dollars) to C$6 billion, implying an annualized real growth rate of about
12% per year. The rate of new investment by venture capital firms grew even more
rapidly, rising from C$290 million in 1991 (or C$306 million 1995 dollars) to C$669 mil-
lion in 1995, which corresponds to real growth of more than 20% per year.
Despite its growing importance, the venture capital industry has received much
less academic scrutiny than other parts of the financial sector.1 This applies both to the-
1
The venture capital industry is more difficult to study than other financial industries such as banking, insur-
ance, stock markets, etc. Little of the relevant information is in the public domain, since the firms financed by venture
capitalists are privately held and therefore do not have the same public reporting requirements as publicly traded
 VENTURE CAPITAL INVESTMENT 443

ory and to empirical investigation. At the theoretical level, perhaps the most fundamen-
tal question to ask about the venture capital industry is why it exists at all. Why have
a set of specialized firms that focus on financing the entrepreneurial sector? Even if
there were no dedicated venture capital firms, a combination of commercial banks, in-
vestment banks, private investors, and stock exchanges providing the necessary inter-
mediation could still be imagined. In fact, among entrepreneurial firms, most finance
is provided by banks and private investors (including family members), and many young
entrepreneurial firms “go public” on stock exchanges without first seeking venture capi-
tal finance. In seeking to understand venture capital finance, it therefore seems impor-
tant to ask what exactly is the niche filled by venture capital firms.
The primary objective of this paper is to present a theory explaining the existence of
the venture capital industry and investigate the consistency of this theory with empirical
observations. Our basic hypothesis is that informational asymmetries are the key to un-
derstanding the venture capital industry. Previous papers have focused on the impor-
tance of asymmetric information in venture capital markets, and several authors have
suggested that a central distinction between venture capitalists and other financial inter-
mediaries is that venture capitalists operate in situations where asymmetric information
is particularly significant. In this paper we provide a simple formal model that distin-
guishes venture capitalists from other potential investors on the basis of their ability
to deal with informational asymmetries. This model is also used to draw inferences
about how venture capital financing would be expected to work. These predictions are
then compared with the actual pattern of venture capital investment in Canada. This
link between theory and empirical evidence is the main contribution of the paper.
There are two major forms of informational asymmetry. One type, sometimes re-
ferred to as “hidden information,” occurs when one party to a transaction knows rele-
vant information that is not known to the other party. For example, an entrepreneur
developing a new product may have a much better idea about whether the product will
actually work than does the venture capitalist who may finance the venture. The prob-
lem arises because the informed party typically has an incentive to misrepresent the
information. The entrepreneur, for example, may have an incentive to overstate the
likelihood of successful product development. Furthermore, the market may become
crowded with “low-quality” projects, precisely because it is hard for investors to distin-
guish between good-quality and poor-quality projects. This phenomenon is called ad-
verse selection. Potential investors understand that adverse selection exists and may
therefore be wary of funding such entrepreneurial endeavours.2
The other type of informational asymmetry is often described as “hidden action.”
In this situation one party to a transaction cannot observe relevant actions taken by
the other party (or at last cannot legally verify these actions). For example, an investor
in an entrepreneurial firm might not be able to observe whether the entrepreneur is
working hard and making sensible decisions, or whether the entrepreneur is planning
to “take the money and run.” This problem leads to what is called “moral hazard.” The

firms. Also, regulatory scrutiny of the industry is modest compared to other financial services, therefore relatively
little information arises from regulatory activities. Finally, as there are no organized exchanges for venture capital
investments, no information derives from that source.
2
A local financial advisor summed up a typical reaction: “You can meet ten enterpreneurs at a party and each
one will tell such a good story that you will want to invest your life savings. Remember, however, that you will lose
money on at least 7 out of the 10. My policy is never to invest in entrepreneurial ventures.”
444 R. AMIT ET AL.

informed party then has an incentive to act out of self interest, even if such actions im-
pose high costs on the other party.
Both adverse selection and moral hazard may arise in any investment environment,
but they seem particularly acute in entrepreneurial finance. With large established firms,
investments are made safer by the use of existing assets as collateral, and the develop-
ment of reputation. Collateral and reputation effects can mitigate the negative effects
of both adverse selection and moral hazard. Because entrepreneurial firms lack assets
to provide as collateral, and because they lack the “track record” necessary to establish
their reputation, the effects of informational market failures are more severe in entre-
preneurial finance than in financing established firms.
Our central hypothesis is that venture capitalists emerge because they develop spe-
cialized abilities in selecting and monitoring entrepreneurial projects. In other words,
venture capitalists are financial intermediaries with a comparative advantage in working
in environments where informational asymmetries are important. This is their niche.3
The next section of our paper provides a brief review of relevant literature, fol-
lowed by a section that sets out a formal model of venture capital finance with associated
empirical predictions. The fourth section describes the data set obtained from Macdon-
ald & Associates, and the fifth section compares the theoretical predictions with the
data. The final section contains concluding remarks.

LITERATURE REVIEW
Akerlof (1970) is normally taken as the starting point of the formal analysis of informa-
tional asymmetry. Akerlof describes a situation where sellers of used cars have private
information about the quality of their cars, but buyers cannot discern quality differences
before purchase. In this setting, low-quality cars or “lemons” dominate the market, thus
the market “selects” adversely. Akerlof showed that this adverse selection is inefficient
in that potentially efficient (i.e., Pareto-improving) trades will not take place.
Adverse selection problems can arise in many circumstances. For example, in insur-
ance markets, buyers may know their true risk better than insurance companies (as in
Pauly (1974)), and in labor markets, workers may be more aware of their abilities than
potential employers are (as in Spence (1973)). Spence points out that one natural market
response to adverse selection is “signalling,” where an informed party (usually the seller
of the high-quality item) provides some signal of high quality. Thus, for example, prod-
uct warranties may be signals of high quality. Rothschild and Stiglitz (1976) emphasize
the rolle of screening, under which the uninformed party offers a contract or set of con-
tracts that cause informed parties to self-select into different groups.
Hidden action (and moral hazard) was first discussed in insurance markets, where
insured parties can take actions that either decrease or increase the risk of hazard. For
example, after purchasing auto insurance, the insured party can either drive safely or
dangerously. Early influential work on moral hazard includes Arrow (1973) and Pauly
(1974), who showed that moral hazard causes market failure. Moral hazard problems
are particularly important in many situations where one party acts as an agent for an-
other party, such as when a client hires a lawyer, or the seller of a house hires a sales
agent. In these situations, the “principal” cannot perfectly observe the effort (or other
actions) of the agent. Jensen and Meckling (1976) argue that agency relationships are

3
This analysis focuses on the venture capitalist’s role as a buyer of entrepreneurial assets. Venture capitalists
must also be good at selling these assets. That is, they must also exit effectively from their investments.
 VENTURE CAPITAL INVESTMENT 445

the key to understanding the modern firm. Thus, for example, the managers of the firm
can be viewed as the agents of the owners, who might in turn be viewed as the agents
of other investors in the firm.
Adverse selection and moral hazard are often viewed as crucial determinants of
venture capital financing. Sahlman (1990), for example, postulates that contracting prac-
tices in the venture capital industry reflect informational asymmetries between venture
capitalists and entrepreneurs, and argues that the lack of operational history aggravates
the adverse selection problem. MacIntosh (1994) also asserts the basic idea that informa-
tional asymmetries are fundamental in the venture capital sector, and this point is also
emphasized in Amit, Glosten, and Muller (1993). Various other papers implicitly recog-
nize the importance of informational issues. For example, MacMillan, Zemann, and Nara-
shima (1987) provide a valuable discussion of how venture capitalists screen new projects.
Chan (1983) highlights the role of venture capitalists in reducing the adverse selec-
tion problem in the market for entrepreneurial capital. He shows that an adverse selec-
tion result derives from the absence of any informed venture capitalists in the sense that
only inferior projects are offered to investors. However, the introduction of informed
investors may overcome this problem, leading to a Pareto-preferred solution. Amit,
Glosten, and Muller (1990) present an agency model in which investors are uncertain
about the entrepreneur’s type when submitting investment bids. The authors relate the
venture capital financing decision to the entrepreneur’s skill level and predict which
entrepreneurs will decide to enter into an agreement with venture capitalists.
Sahlman (1990) notes that staged investment, which creates an option to abandon
the project, is an important means for venture capitalists to minimize agency costs.4 The
role of staged investment as a monitoring device is also examined by Gompers (1995). In
addition, the active involvement of venture capitalists in the operation of their investee
companies might mitigate the moral hazard problem. The empirical significance of the
role of venture capitalists as monitors is supported by Barry et al. (1990) and by Lerner
(1995). In addition, Lerner (1994) suggests the use of syndication (i.e., coordinated in-
vestment by two or more venture capitalists) as a method of reducing problems caused
by informational asymmetries. Two other useful papers that describe actions that ven-
ture capitalists can take to reduce problems arising from informational asymmetries
include Tyebjee and Bruno (1984) and Fried and Hisrich (1994).
Chan, Siegel, and Thakor (1990) seek to explain various “rules of thumb” in venture
capital contracting practices as a response to informational asymmetries and, in a related
paper, Hirao (1993) assumes that the entrepreneur’s unobservable actions affect the
venture capitalist’s learning process, and uses this context to study the effects of differ-
ent contracts. A more general overview of research challenges in the venture capital
area is given by Low and MacMillan (1988).
Despite a number of empirical and descriptive studies on venture capital practices
and activities, including some of those already mentioned and also MacMillan, Siegel,
and Narashima (1985), Bygrave and Timmons (1992), and Gompers and Lerner (1994),
among others, empirical work on venture capital finance is still relatively modest in
scope compared to the analysis of other financial intermediaries. Our paper seeks to
add to this literature. Specifically, we provide a formal model that uses asymmetric infor-
mation to explain the existence of venture capitalists, then compare the predictions of
this theoretical structure with evidence on venture capital finance in Canada.

4
Admati and Pfleiderer (1994) and Hellmann (1994) provide formal models of staged finance in the venture
capital context.
446 R. AMIT ET AL.

A THEORY OF VENTURE CAPITAL FINANCE

An entrepreneur has a potential project and seeks potential investors. To keep the anal-
ysis simple we assume that the project requires fixed financial input I from an investor.
The expected cash flow from the project, net of production costs, is denoted R (for “net
operating revenue”). This expected net operating revenue depends in part on the effort,
e, provided by the entrepreneur and it depends in part on the underlying project quality,
q. In addition, the outcome depends on a random variable, u, with expected value 0.
The realized net cash flow is therefore
R(e,q) ⫹ u (1)
where the expected operating revenue is R(e,q). We assume that entrepreneurs and
investors are risk-neutral expected value maximizers. We, therefore, ignore u and work
with R. Variable u plays one important role, however. Given unobservable random un-
certainty, as represented by u, it is not possible for an investor who knows project quality
q to infer effort e from the cash flow realization.
If e cannot be observed by the investor, then it is a hidden action and gives rise
to a moral hazard (or “agency”) problem. If q is known to the entrepreneur, but not
to the investor, then it is hidden or private information and gives rise to potential adverse
selection. The presence of exogenous uncertainty, as represented by random variable
u, does not in itself cause market failure. R is taken to be increasing in e and q. We
also assume that there are decreasing marginal returns to effort. The effort effects can
be written formally as
Re ⬎ 0, Ree ⬍ 0 (2)
where subscripts denote (partial) derivatives.
Let the share of the proceeds that go to the investor (possibly a venture capital
firm) be denoted ␣5. The expected return V to the investor is
V ⫽ ␣R(e,q) ⫺ I (3)
The expected return to the entrepreneurial firm, denoted ␲ (for “profit”), is its share
of the proceeds, net of the costs of effort e.
␲ ⫽ (1 ⫺ ␣)R(e,q) ⫺ e (4)
Variable e is normalized so that providing e units of effort imposes cost e on the entre-
preneurial firm.

Moral Hazard
To demonsrate the moral hazard problem, assume initially that q is known to both par-
ties. A profit maximizing entrepreneur will maximize (4) with respect to e, leading to
the following first order condition:
␲e ⫽ (1 ⫺ ␣)Re ⫺ 1 ⫽ 0 or Re ⫽ 1/(1 ⫺ ␣) (5)

5
The implied contract has a linear structure. The results in this section are therefore predicated on the assump-
tion of linear contracts. Non-linearities, such as buyback options for entrepreneurs, are not considered here, but
might be useful in mitigating some of the addressed informational problems. We thank seminar participants at MIT
for pointing this out.
 VENTURE CAPITAL INVESTMENT 447

The second order condition for a maximum is (1 ⫺ ␣)Ree ⬍ 0. Noting that the factor
(1 ⫺ ␣) is presumed to be strictly positive and using (2), this second order condition
must hold.
The efficient or “first-best” level of effort e* is determined by maximizing the sum
of (3) and (4) with respect to e. This sum, denoted S, is
S ⫽ R(e,q) ⫺ I ⫺ e. (6)
Maximizing (6) with respect to e yields the following first order condition
Re ⫽ 1 (7)
It follows form (5), (7), and (2) that the entrepreneur will choose less than the efficient
level of effort as long as ␣ is strictly positive. This is the moral hazard problem. It is
illustrated in Figure A1 in Appendix 1. It follows from the corresponding algebra and
Figure A1 that effort is declining in ␣.
de/d␣ ⬍ 0 (8)
It is possible that the moral hazard problem might render the project infeasible.
The inveestment is attractive to the investor only if the return equals or exceeds the
alternative value that can be obtained by investing I elsewhere. Let this required return
or opportunity cost be denoted r. Then feasibility requires
(1 ⫹ r)I ⱕ ␣R(e(␣),q) (9)
The problem is that there may be no value of ␣ that allows (9) to be satisfied. If the
expected return to the investor is too low, this suggests raising ␣, but then e will fall
(from (8)), reflecting the idea that the entrepreneur will provide less effort as his stake
in the firm falls.
Feasibility for the entrepreneur requires that the expected profit given by (4) ex-
ceed the return from the entrepreneur’s best alternative, which can be normalized to
equal 0. It is possible that effort level e* would in principle allow feasibility for both
investor and entrepreneur, but that the actual effort relationship, e(␣) would not allow
the project to be financed. Thus the moral hazard problem may cause the market to fail.
We now introduce the idea that investors can monitor the entrepreneur and, at
some cost, induce the entrepreneur to provide additional effort. Denote the monitoring
cost m. The expected return to the investor is therefore
V ⫽ ␣R(e(␣,m),q) ⫺ I ⫺ m (10)
If the responsiveness of e to m is low, then the investor will not bother to monitor, as
the cost will exceed the benefit. Some investments may be worthwhile, without monitor-
ing, in spite of the moral hazard problem, but many projects will be abandoned. If e
is highly responsive to monitoring, then the investor will undertake monitoring and will
elicit an effort level closer to “first-best” level e*. Projects that are not financed by other
investors will be feasible for investors who are good at monitoring (i.e., those for whom
the responsiveness of e to m is high).
It is also possible that the investor provides valuable services, s, to investee compa-
nies. These services (e.g., providing strategic and operational advice, aid in fundraising,
adding reputation, etc.) are observable by the entrepreneur. Ignoring monitoring for
the moment and normalizing the cost of providing s to 1 per unit, the expected return
to the investor is now
448 R. AMIT ET AL.

FIGURE 1 Effects of services on expected net revenues.

V ⫽ ␣R(e(␣),q,s) ⫺ I ⫺ s (11)
We can think of the effect of s ⬎ 0 on the operating revenues R in the following
way. Services s can produce a direct (positive) effect on R through Rs ⬎ 0 (case 1), or
can have an indirect (positive) effect on R through enhancing the marginal productivity
of the entrepreneur’s effort, or Res ⬎ 0 (case 2). When both effects are present, Rs ⬎
0 and Res ⬎ 0, we have case 3. Figure 1 illustrates these different cases and compares
them with the benchmark case where s ⫽ 0.
Case 1 is defined as the case in which the investor’s provision of s does not affect
the entrepreneur’s productivity of effort, Re, but raises revenues directly. Let us assume
that this effect is additive. For each effort level e expanded by the entrepreneur, the
provision of s ⬎ 0 by the investor will increase the venture’s revenues by DR. This is
expressed in Figure 1 as a parallel upward shift of the graph of R(e) from the benchmark
case to case 1. With respect to the moral hazard problem this means that, relative to
the benchmark case, Re and thus the entrepreneur’s incentive constraint (5) remain un-
changed in case 1. Therefore our basic analysis for s ⫽ 0 still holds (see equations (1)–(9)
and Appendix 1). In other words, the moral hazard problems in the benchmark case
and case 1 are identical.
In cases 2 and 3, however, the provision of s improves the productivity of e, and
Re is consequently shifted upward. This results in steeper curves for cases 2 and 3 in
Figure 1. The entrepreneur’s incentive constraint (5) is affected by this change, and
therefore a new analysis of the moral hazard problem is required. Let us denote the
case where s ⫽ 0 with superscript 0 and cases 2 or 3 where s ⫽ k ⬎ 0 with superscript
s. “First-best” effort levels are denoted e*, “second-best” effort levels e⬘. The new situa-
tion is depicted in Figure 2.
 VENTURE CAPITAL INVESTMENT 449

FIGURE 2 First- and second-best effort levels in base case (s ⫽ 0) and under Res ⬎ 0.

Figure 2 is based on the different possibilities of how s can affect R; it also draws
on the previous discussion of the standard moral hazard problem (without monitoring
or services rendered). It allows us to conclude that the moral hazard problem persists
for s ⬎ 0 even if Res ⬎ 0. In this case, the “second-best” effort level es⬘ is still smaller than
the “first-best” effort es*. However, relative to the base case scenario, the entrepreneur is
now willing to put forth more effort (es⬘ ⬎ e0⬘).
Thus, the provision of s might contribute to the realization of projects which other-
wise would have been abandoned, as they did not fulfill the investor’s original feasibility
constraint (9). Considering s, the investor’s feasibility constraint now becomes
(1 ⫹ r) I ⫹ s ⱕ ␣R(e(␣),q,s) (12)
If s is not prohibitively high, then it might relax this constraint through its direct and
indirect positive effect on R. Thus, investors who are skilled at providing value-creating
services to their portfolio companies will undertake certain projects which other, less
skilled investors will shun.
There is ample evidence that venture capitalists provide valuable services to their
portfolio companies. Gorman and Sahlman (1989) compiled a list of such services from
a survey of venture capital investors. The five highest ranked and most frequently used
activities can either be interpreted as directly enhancing investee reveneus (e.g., introduc-
tion to potential customers and suppliers, assistance in obtaining additional financing) or
as enhancing the entrepreneur’s productivity of effort and thus indirectly boosting in-
vestee revenues (e.g., strategic planning, management recruitment, operational planning).
We now turn to the case in which both monitoring and services are considered.
450 R. AMIT ET AL.

The effects of s on R might be important enough to render projects feasible which were
infeasible even with optimal monitoring. In fact, it seems natural to assume that a combi-
nation of monitoring and the provision of services constitutes a powerful tool in the
hands of specialized investors to reduce moral hazard problems. Note, for example, that
the entrepreneur’s “second-best” effort provided in the case where s ⬎ 0 and m ⬎ 0
might be higher than the “first-best” effort in the benchmark case where s ⫽ 0 and m ⫽
0. (Refer to Figure 2 and recall that if e is sufficiently responsive to m, es⬘ might get
fairly cose to es* under an optimal monitoring regime.)
Another point worth emphasizing is that providing services to entrepreneurs might
make it easier and thus cheaper for investors to monitor them. Denoting M(m|s) as the
monitoring costs at a given level of s, it is very likely, for example, that M(m|s ⬎ 0) ⬍
M(m|s ⫽ 0) ⫽ m. Thus, the return to the investor given monitoring and services is
V ⫽ ␣R(e(␣,m),q,s) ⫺ I ⫺ s ⫺ M(m|s) (13)
We note that investors who are good at monitoring and providing valuable services to
their portfolio companies are likely to invest in firms with more severe moral hazard
problems, as their feasibility constraint is more likely to be fulfilled.

Adverse Selection
A similar pattern emerges when adverse selection is considered. Assume that the ven-
ture capitalist chooses the optimal amount of services rendered and the optimal amount
of monitoring effort, giving rise to associated values of e and s for any given ␣. Quality
level q is now unobservable to the investor. Suppose that the range of q is such that
the average quality project does not yield enough expected returns (for any value of ␣)
to allow both (13) and (4) to be positive. Thus, the average project is not worth funding.
Formally, we can write the investor’s expected return as
EV ⫽ 冮q[␣R(e(␣,m(␣)),q,s(␣)) ⫺ I ⫺ s(␣) ⫺ M(m(␣)|s(␣))]f(q)dq ⬍ 0 (14)

where f(q) is the probability density function for project quality. To simplify this expres-
sion, we subsume the terms that do not bear directly on the analysis of the hidden infor-
mation problem into investor’s costs C. With
C ⫽ I ⫹ s(␣) ⫹ M(m(␣)|s(␣)) (15)
inequality (14) reduces to
EV ⫽ 冮q[␣R(q) ⫺ C]f(q)dq ⬍ 0 (16)

Inequality (16) says that the expected value across all projects is negative. However,
some of the individual projects (those in the upper end of the quality distribution) may
be very valuable. Suppose, for example, that the top 40% of projects could generate
a positive net profit. Unfortunately, the entire market will normally fail in this situation,
as it will typically not be worthwhile for investors to provide financing, even though
many individual projects are worthwhile.
Now suppose that an investor can acquire information about the quality of an indi-
vidual project by spending d before making the actual investment I. Parameter d can
be interpreted as the cost of “due diligence.” This cost determines the probability, p(d),
with which an investor can establish whether the quality of a certain project exceeds a
 VENTURE CAPITAL INVESTMENT 451

FIGURE 3 Venture capital investment process.

threshold level of quality. We denote this threshold level of quality as q0. Let us implicitly
define q0 as follows:

V ⫽ ␣R(q) ⫺ C ⫽ 0 for q ⫽ q0
V⬎0 for q ⬎ q0 (17)
V⬍0 for q ⬍ q0

The “detection function” p(d) is assumed to have the following properties:

p(d ⫽ 0) ⫽ 0, p(d ⫽ ∞) ⫽ 1, (18)

p⬘(d) ⬎ 0 and p″(d) ⬍ 0

Let us restate the assumptions concerning the sequence of events in the above
model. Investment in an entrepreneurial firm is a one period, multi-stage process as
illustrated in Figure 3. In the first stage, the investor incurs an up-front cost of d in order
to assess the quality of a potential investment. With probability p(d) the investor will
become informed about q and will, therefore, find out whether q ⱖ q0 or q ⬍ q0. Only
in the former case an investment will be made. With probability (1 ⫺ p(d)), however,
the investor will remain uninformed about q and, due to (16), refrain from investing.
Stage 3, in which the entrepreneur displays effort and is monitored and supported by
the investor, and stage 4, in which the benefits from the investment are reaped and dis-
tributed, occur only if in stage 1 q is found to be greater than q0.
The expected net return to the investor can therefore be expressed as

EV ⫽ p(d) 冮q ⬎ q0 (␣R(q) ⫺ C)f(q)dq ⫺ d (19)

Feasibility now requires that

r(I ⫹ d) ⱕ EV (20)

It follows immediately from (17), (18), and (20) that investors who are good at doing
due diligence in the sense that low values of d yield a given value of p are likely to engage
in due diligence, select high quality projects (i.e., projects with positive expected return),
452 R. AMIT ET AL.

and make investments.6 These are the investors that become venture capitalists. (For
further formal analysis of the advise selection case, see Appendix 2).
We should emphasize that we assume that the efforts undertaken by the venture
capitalist are not subject to free riding. That is, another investor cannot simply observe
the venture capitalist who has undertaken diligence and then underbid him. Typically
venture capitalists are able to keep the results of diligence and monitoring confidential
until after financial contracts have been signed. Free riding does occur but, given the
informational asymmetries in the venture capital sector, it seems plausible to abstract
from free riding here.

Implications
The above formulation provides the simplest configuration that reflects the idea that
venture capitalists are those investors who become skilled at selecting good projects
in environments with hidden information and are good at monitoring and advising en-
trepreneurs who might otherwise be vulnerable to moral hazard problems. The implica-
tions of this modeling framework are outlined below.
1. Venture capitalists will operate in environments where their relative efficiency in
selecting and monitoring investments and providing value-enhancing services gives
them a comparative advantage over other investors. For example, as we have seen
in the “hidden action” case, it may take effective monitoring m and specific services
s to make a project attractive for an investor. In the “hidden information” case, on
the other hand, market failure can be avoided if the probability of detecting whether
a project is worth supporting is high enough for sufficiently low due diligence costs.
This suggests strong industry effects in venture capital investments. We would expect
venture capitalists to be prominent in industries where informational concerns are
important, such as biotechnology, computer software, etc., rather than in “routine”
start-ups such as restaurants, retail outlets, etc. The latter are risky, in the sense that
random variable u has high variance, but they are situations that are relatively easy
to monitor by conventional financial intermediaries, whereas the former draw much
of their value from idiosyncratic knowledge that is much harder to assess. In princi-
ple, in-depth knowledge of traditional industries, such as retailing, is not less advanta-
geous than in-depth knowledge of high-tech industries, but there is some evidence
that such wisdom is harder to obtain for knowledge-based industries where informa-
tional asymmetries are, therefore, likely to be higher. (See Industry Canada (1994) on
the particular difficulties and challenges that investors and lenders face with regard to
the assessment of knowledge-based small- and medium-sized enterprises.)
2. Within the class of projects where venture capitalists have an advantage, venture
capitalists will still prefer projects where selection, monitoring, and service costs are
relatively low or, in other words, where the costs of informational asymmetry are
less severe. In the presence of moral hazard, investors would prefer projects for which
e is more responsive to m, and/or for which R and/or Re are more responsive to s.

6
We acknowledge that the structure of the venture capital investment process as depicted in Figure 3 is a
simplification. Venture capitalists make investments even if they are not completely certain that q ⬎ q0 and therefore
they may earn negative returns on individual investments. However, taking this fact into account does not change
the analysis except to add some additional algebra. (Note that in our model expected returns to venture capital
investments are positive, but actual returns can be negative if u ⬍ 0.)
 VENTURE CAPITAL INVESTMENT 453

In the presence of adverse selection, projects with a highly responsive p(d) would
be favored over those where the detection of quality is more difficult and thus more
costly. Thus, within a given industry where venture capitalists would be expected to
focus, we would expect venture capitalists to favor firms with some track record over
pure start-ups. To clarify the distinction between point 1 and point 2, note that point
1 states that if we look across investors, we will see that venture capitalists will be
more concentrated in areas characterized by significant informational asymmetry.
Point 2 says that if we look across investment opportunities, venture capitalists will
still favor those situations that provide better information (as will all other investors).
Thus venture capitalists perceive informational asymmetries as costly, but they per-
ceive them as less costly to deal with than do other investors.
3. If informational asymmetries are important, then the ability of the venture capitalist
to “exit” may be significantly affected. Ideally, the venture capitalists might wish to
sell off their share in the venture after it “goes public” on a stock exchange. If, how-
ever, these investments are made in situations where informational asymmetries are
important, it may be difficult to sell shares in a public market where most investors
are relatively uninformed. Public investors probably have a less responsive function
p(d) and therefore (19) could be negative for them. This concern invokes two natural
reactions: One is that many “exits” would take place through sales to informed inves-
tors, such as other firms in the same industry as the venture or to the venture’s own
management or owners. These informed investors probably have similar, if not better
detection functions p(d) than the venture capitalist. A second reaction is that venture
capitalists might try to acquire reputations for only presenting good quality ventures
in public offerings. (However, this is an argument drawing on a multiperiod scenario
and would therefore require an extension of our model). Therefore, we might expect
that the exits that occur in initial public offerings would be drawn from the better-
performing ventures.7
4. The model implies that dR/de (⫽ Re) ⬎ 0 and de/d␣ ⬍ 0. Together these two proper-
ties imply dR/d␣ ⬍ 0. Other things equal, we can expect entrepreneurial firms in
which venture capitalists own a large share to generate lower net returns. This would
be due to the moral hazard problem. Higher values of ␣ reduce the incentives of
the entrepreneur to provide effort. Nonetheless, it still might be optimal in a given
situation for the venture capitalist to take on a high ownership share, as this might
be the only way of getting sufficient financial capital into the firm. However, we would
still expect a negative correlation between the venture capital ownership share and
firm performance.
We note, however, that the model also suggests a negative relationship between R
and ␣ for another reason. Specifically, the selection constraint for investors is that ␣R ⱖ
(1 ⫹ r)I or R ⱖ (1 ⫹ r)I/␣. If the venture capital market were very competitive so that
investors earned no rents, then this selection constraint would hold with equality, and
there would be an exact negative relationship between expected net operating revenues
and ␣, whether or not moral hazard was present. Even if venture capitalists earn some
expected rents, this selection constraint will still rule out combinations of low ␣ and
low R, which will tend to induce a negative correlation between R and ␣. The basic

7
Empirical work by Megginson and Weiss (1991) and Gompers (1996) is consistent with the idea that the
reputation of venture capitalists is very important at the IPO stage.
454 R. AMIT ET AL.

logic is that, for a given investment I, investors will need to be compensated by a large
ownership share ␣ if the expected net operating revenues are relatively low.

THE DATA SET

The data used for this study were collected by Macdonald & Associates Ltd. and made
available to us on a confidential and anonymous basis. In addition, no individual firm-
specific information is reported or discussed in our analysis. The data are derived from
two surveys. The first survey, referred to as the “investment survey,” began as an annual
survey in 1991 and became quarterly in 1994. It asks just over 100 Canadian venture
capital firms to identify their investees and provide some information about each invest-
ment and divestiture. Investees are recorded in the database and follow-up information
is requested in subsequent surveys. The investment survey asks about the amount and
stage of each investment and also seeks information about the venture capitalist’s ulti-
mate divestiture of its holdings in each investee.8
This survey, which covers the period from 1991 through the first quarter of 1996,
seeks to obtain comprehensive information from all Canadian venture capital providers.
In an effort to get full information about the investee firms, the survey is sent to venture
capital companies (as just noted) and other investors who have investments in the ven-
ture-backed investees. However, some relevant venture capital providers may have
been overlooked in the survey, and some surveyed venture capitalists may not report
all of their investments. Nonetheless, Macdonald & Associates Ltd. estimate that the
investment survey identifies 90–95% of the underlying population of Canadian firms
supported by Canadian venture capitalists.
The second survey, referred to as the “economic impact” survey, began in 1993
and is conducted annually. It seeks additional information about the investees identified
in the investment survey. Thus, economic impact information is sought about each in-
vestee that received an investment in or after 1991. Retrospective information is also
requested. Suppose, for example, that an investee received an investment in 1991. The
venture capitalist making the investment would have received a 1993 economic impact
questionnaire asking for information about this investee going back as far as 1987. In
many cases not much retrospective information can be provided, but the database con-
tains economic information on a reasonable number of investees going back as far as
1987. The date of the investee’s original startup (which in some cases is well before 1987)
is also reported.
The response rate for the economic impact survey over its three year life has varied
between 56% and 74% (i.e., information has been received on 56% to 74% of the tar-
geted investee firms). If the investment survey identifies 90–95% of the relevant under-
lying population, then the effective sample coverage is between 50% (.9 times 56%)
and 70% of the underlying population. The economic impact survey collects balance
sheet and income statement information on the investees (including revenues and taxes
paid). It also collects information on the structure and amount of their employment,
and the nature of their industry.
A typical investee enters the data set when it receives its first investment from a

8
This “exit” information, which is obtained on a regular basis from the investment survey, was complemented
in November 1995 by an additional survey addressed to venture capital providers who had previously reported
on divestitures.
 VENTURE CAPITAL INVESTMENT 455

venture capitalist. It may receive investments from additional venture capitalists as well.
Subsequent rounds of investment may also occur. Eventually, an investee leaves the
sample. This occurs when all venture capitalists have either written off (in the case of
failure) or “cash in” their holdings in the investee. Thus, the data set contains a series
of “life histories” for venture capital-backed firms.
A “record” refers to information for one particular investee firm for one particular
year. There are 387 investee firms in the data availabale from the economic impact sur-
vey, but information on about 18 of these firms is significantly incomplete. The re-
maining 369 firms provide 1,298 reasonably complete records, and, therefore, have an
average of about 3.5 records each. The investment survey data includes information
on 1,086 Canadian investees. For some purposes, complete matched records are neces-
sary,9 but much interesting and relevant information is available from just the economic
impact data (1,298 records on 369 companies) or just the investment data (2,017 records
on 1,086 companies).
These data sets target Canadian investees supported by the Canadian venture capi-
tal industry. A Canadian entrepreneurial company that received support exclusively
from venture capitalists based in the United States or Asia and had no support from
Canadian venture capitalists would not be in our data set. This set of firms is probably
fairly small, but there is no data available on its magnitude. It seems unlikely that this
omission introduces much systematic bias over most subjects of interest in the data.
Despite some possible selection bias in the economic impact data, the data set as a whole
remains an important and unique data source.

INVESTMENT PRACTICES OF CANADIAN

VENTURE CAPITALISTS
We now present some empirical evidence that addresses the predictions of the theoreti-
cal framework outlined in section 3. Some of this data, together with other empirical
information on the Canadian venture capital industry is provided by Amit, Brander,
and Zott (1997). Before considering the implications of informational asymmetries, we
provide a general characterization of important financial variables in the data, as shown
in Table 1. All relevant table entries are in thousands of 1995 Canadian dollars. As this
table implies, the size of investee companies varies substantially, with a few large firms
that make the average values much larger than the median values. The median investee
has about 50 Canadian employees and annual revenues of C$6 million. A typical owner-
ship share for the venture capitalist is approximately 30%.
The data in Table 1 also imply that firms in the data set spend, on average, about
3.5% of their revenues on R&D. This is about the same as the overall ratio of R&D
spending to revenues for the Canadian economy as a whole. We should note, however,
that these rather moderate R&D expenses may be due to different accounting standards
that prevail in small and relatively young companies in contrast to large and established
firms. Revenues per Canadian employee are $148,800, and the average long term debt
to equity ratio is a conservative 0.77. (The long term debt to equity ratios derived from
Canadian COMPUSTAT data is estimated to be 1.75 for companies of all sizes, and

9
Matching the two data bases, we obtain 408 complete records on 302 investee companies. These numbers
are low primarily because there are only 339 investee companies with records in both databases and because, for
each investee, matches occur only in years when investments were undertaken.
456 R. AMIT ET AL.

TABLE 1 Summary Financial Data: 1987–94 (in Real $1995)

Mean Median Standard No. of
($000s) ($000s) deviation records
Total assets 22,928 5,540 70,707 1,277
Total equity 8,777 1,893 25,254 1,274
VC-share of equity (%) 34 30 30 1,218
Retained earnings 848 154 10,098 1,127
Total fixed assets 10,745 1,996 52,353 1,257
Long-term debt 6,729 1,056 28,122 1,157
Revenue 23,657 6,177 56,077 1,290
Investments in property, plant 1,954 222 8,180 1,161
and equipment
R&D expenditures 837 79 2,098 1,067
Taxes paid 461 25 1,315 1,027
# of Canadian employees 159 50 301 1,293
Source: Macdonald & Associates Ltd. Economic Impact Database.

0.90 for companies with annual sales less than $100 million.) The low debt-equity ratio
may reflect the limited borrowing capacity of entrepreneurial firms. We note also that
the average investee is profitable enough to pay nontrivial amounts of tax.
We now consider the implications of the information-based model described in Sec-
tion 3. One of the implications was that venture capital would be focused on industries
where the importance of monitoring and due diligence expertise is particularly great.
Table 2 presents information about the industry breakdown of the investee companies,
and compares these investment shares with the shares of these industries in total output
(as measured by Canadian gross domestic product (GDP)).

TABLE 2 Industry Classification

Early stage Total
investment* investment* % of early % of total % of total
(no. of investees) (no. of investees) investment investment output
Biotechnology 95.4 121.5 17 6 0
(43) (51)
Communications 83.7 225.1 15 10 5
(32) (63)
Manufacturing and 78.7 461.6 13 21 24
industrial equipment (82) (261)
Computer (hardware 70.0 314.4 12 14 3
and software) (100) (182)
Miscellaneous 67.1 314.7 12 15 34
(58) (178)
Medical/health 58.4 176.1 10 8 3
(34) (59)
Energy/environmental 57.4 134.6 10 6 4
technology (33) (68)
Consumer related 31.7 296.3 6 14 26
(27) (109)
Electrical components 25.0 125 4 6 2
and instruments (42) (89)
Total: 567.2** 2169.3 99** 100 101**
(451) (1060)
* In C$ mill; ** Due to rounding. Sources: Macdonald & Associates Ltd. Investment Database. Output shares are based
on estimates from Statistics Canada “Gross Domestic Product by Industry,” 1996, cat. no. 15-001-XPB.
 VENTURE CAPITAL INVESTMENT 457

TABLE 3 Age of Venture-Backed Companies

Year founded # of companies % of total
1994 23 6
1993 22 6
1992 20 5
1991 28 7
1984–1990 163 42
1974–1983 85 22
Before 1974 38 12
Total 379 100
Source: Macdonald & Associates Ltd. Economic Impact Database.

As can be seen from Table 2, venture capital is much more heavily represented
in biomedical areas, computers, and communications than would be implied by overall
output shares of these industries in the economy as a whole. Venture capital has a
slightly smaller share of manufacturing and industrial equipment than the economy as
a whole, and a much lower share of “consumer related” and “miscellaneous” industries.
The main components of these categories are the retail sector and various services. This
picture is even more pronounced when only early stage venture capital investments are
considered. It seems very plausible that the industries where venture capitalists concen-
trate the most are those where informational asymmetries are most severe. It is, of
course, possible that venture capitalists invest relatively heavily in high-tech industries
for reasons unrelated to information. For example, the high-tech sector may simply have
a disproportionately large number of new investment opportunities. More specifically,
it is a growth sector, and any growth sector will appear to have high levels of new invest-
ment from most financial intermediaries, including venture capitalists. Even so, venture
capitalists have a heavier relative investment in high-tech industries than other financial
intermediaries, and informational reasons offer a plausible explanation for this. Thus,
Table 2 is consistent with our theoretical expectations.
The second major implication of the information-based theory developed in Sec-
tion 3 is that within the sectors where venture capitalists operate, they still prefer to
invest in firms where the adverse selection and moral hazard problems are least severe.
The following information is consistent with this expectation. First, Table 3 shows the
age structure of the investee firms.
As shown on Table 3, quite a few investee companies are surprisingly old. Fully
12% of the 379 companies for whom information on age is available were founded prior
to 1974. Since the data set is limited to firms that received at least one infusion of venture
capital in 1991 or later, some firms obtain venture capital financing long after being
founded. (We note, however, that these firms might have obtained earlier venture capi-
tal infusions. Our data suggests that many recorded investments are indeed follow-
up investments.)
Furthermore, this information suggests that it takes longer than commonly per-
ceived, and perhaps more venture capital than originally anticipated, to bring some in-
vestee firms to the stage at which exit is feasible. A company may be founded well before
it obtains its first venture capital investment. These data raise the possibility that venture
capital focuses on expansion of existing small companies rather than on the start-up
phase. Tables 4 and 5 provide more information on this point.
Table 4 shows how many investments correspond to each stage in the entrepreneur-
ial firm’s life. It is based on investment records of investee companies that are in the
458 R. AMIT ET AL.

TABLE 4 Number of Investments by Stage and Year

Early stages Later stages
SE ST ES EX AC TU WC OT Count
1991 3 100 — 85 12 22 — 36 258
1992 15 111 — 65 23 41 2 50 307
1993 5 116 — 125 18 23 25 37 349
1994 3 128 11 206 12 23 — 15 398
1995 8 130 112 241 11 21 2 44 569
1996(Q1) 5 42 12 54 3 11 — 9 136
Total 39 627 135 776 79 141 29 191 2017
Key: SE ⫽ seed; ST ⫽ start-up; ES ⫽ other early stage investments; EX ⫽ expansion; AC ⫽ acquisition; TU ⫽
turnaround; WC ⫽ working capital; and OT ⫽ other. Source: Macdonald & Associates Ltd. Investment Database.

Investment Database and includes investments made betwen 1991 and the first quarter
of 1996. A given investee may obtain financing from multiple venture capitalists, and
may also receive multiple rounds of investment from a given venture capitalist. Each
investment, which may include debt, equity, or both, is recorded separately. We observe
that a full 60% of the investments made over the period covered by our sample are
late stage investments. As early stage investments are both smaller (from Table 5) and
less numerous (Table 4) than late stage investments, we can infer that the venture capital
industry seems to focus more on growth and development of firms than on start-up activ-
ity. Tables 3 to 5 show that venture capitalists focus on firms with a long enough track
record to provide significant information about the underlying quality of the venture.
Pure start-up activity, where adverse selection and moral hazard problems are most
severe, is less significant than later stage investment.
Figure 4 depicts the relative importance of debt and equity in an average or repre-
sentative investment by stage. There are, for example, 39 seed investments in total. The
total equity in these 39 investments is $21.89 million, giving an average of $561,000,
while the total debt is $2.34 million, resulting in an average of only $60,000 (note that
most seed investments have no debt). Figure 4 shows that equity is relatively more im-
portant at the early stages, and debt becomes more significant later, although equity
remains more important in abolute terms for every stage except working capital.
The third major implication of our information-based approach is that we might
expect exit to be dominated by “insider” activity rather than by public offerings. Figure
5 shows the pattern of exits in the data and indicates that only about 16% of exits occur
after initial public offerings (IPOs). About 10% are third party acquisitions, often by

TABLE 5 Average Size of Investments by Stage and Year (in C$000’s)

Early stages Later stages
SE ST ES EX AC TU WC OT Total
1991 489 678 — 1165 2003 1424 — 1374 1058
1992 900 617 — 1104 1283 628 480 1480 925
1993 836 1101 — 1714 1665 1620 362 1662 1394
1994 425 677 854 1227 2338 1521 — 2391 1128
1995 414 688 1005 1300 2341 436 1378 1564 1098
1996(Q1) 101 1034 847 1297 2260 1601 — 890 1151
1991–96 621 771 997 1316 1824 1107 378 1559 1127
Source: Macdonald & Associates Ltd. Investment Database.
 VENTURE CAPITAL INVESTMENT 459

FIGURE 4 Average debt and equity by investment stage 1991–1996(Q1).

a firm in the same industry as the venture. The largest category of exit is company buy-
outs, in which the venture capitalist’s holding is sold to officers or managers of the in-
vestee. Fully 37% of exits are in this category. Secondary purchases are purchases of
the venture capitalist’s holding by a third party in a private transaction that is not an
overall acquisition. The “other” category consists of exits for which the exit mode was
not identified, but we believe that most of these are company buyouts. Approximately
17% of exits were in the “write-off” category. If informational asymmetries are impor-
tant, it is not surprising that IPOs account for only a small share of exits while company
buyouts are much more important. We wish to note, however, that the small share of IPOs
may also partly reflect a minimum scale necessary to sustain a public market in a stock.
Our theoretical framework also suggests that returns would differ by exit vehicle
and that, in particular, IPOs would have high returns precisely because venture capital-
ists seek to reduce the adverse selection problem confronted by buyers of IPOs by only
“going public” with relatively strong investee firms. These returns shown in Table 6 are
consistent with our expectations. Write-offs, of course, represent a 100% loss over the
holding period. Among the other forms of exit, IPOs are relatively profitable. Secondary
460 R. AMIT ET AL.

FIGURE 5 Distribution of venture capital exits (percentage of exits).

purchases (i.e., secondary sales from the exiting venture capitalist’s point of view) are
similarly profitable in aggregate, although with only 11 observations, it is difficult to
regard the return to secondary purchases as highly meaningful. In any case, the high
return to IPOs is consistent with our expectations.
The final prediction of our model is that the venture capitalist’s ownership share
should be negatively associated with the firm’s performance. This derives both from
moral hazard and the venture capitalist’s participation constraint that expected returns
should at least equal the return from alternative investments. In addition, it is possible
that a negative correlation between a venture capitalist’s ownership share ␣ and a mea-
sure of firm performance could arise from dilution in a multi-period process (i.e., the
possibility that low performance leads to high ␣). Unfortunately, we do not have ade-
quate data, such as data on a venture capitalist’s ownership share in the start-up phase,
to correct for dilution.
It is difficult to measure firm performance directly, but revenues per unit asset and
taxes paid should both be good measures of performance. Table 7 reports the results

TABLE 6 Estimated Real Annual Returns by Exit Type

Standard
Mean of deviation of Real annual
individual real individual No. of return of sum
annual returns* returns observations of investments**
IPO 43% 62% 26 26%
Acquisition 36% 61% 16 9
Secondary purchase 23% 41% 11 29%
Company buyout 2% 15% 37 0%
Writeoff 100% loss over — 24 100% loss over
holding period holding period
Other 2% 18% 7 13%
* Individual annual returns are calculated as: {(Proceeds from investment ⫺ cost of investment) ⵩ (1/holding period) ⫺
1}. ** This number is calculated as: {(Sum of proceeds from investment ⫺ sum of costs of investment) ⵩ (1/average holding
period) ⫺ 1}. Source: Macdonald & Associates Ltd. Investment Database.
 VENTURE CAPITAL INVESTMENT 461

TABLE 7 Effect of Venture Capital Share on Performance (Tobit Regressions)

Dependent vbl. Expl. variable Coefficient Std. error T-stat P-value
TaxesPaid VCshare ⫺10.59 1.95 ⫺5.44 .000
log(Age) 454 61 7.44 .000
Const. ⫺696 155 ⫺4.50 .000
TaxesPaid/Assets VCshare ⫺28.27 5.94 ⫺4.76 .000
(⫻10000) log(Age) 488 187 2.61 .009
Const. ⫺578 474 ⫺1.22 .223
Revenues/Assets VCshare ⫺46 20 ⫺2.30 .021
(⫻1000) log(Age) ⫺386 649 ⫺0.60 .55
Const. 8958 1604 5.59 .000
Source: Macdonald & Associates Ltd. Economic Impact Database.

arising from regressing these measures of firm performance on the venture capital own-
ership share, correcting for age of the firm.
As can be seen from these regressions, there is a statistically strong negative rela-
tionship between the venture capitalist’s ownership share and these measures of firm
performance. Ideally, we would like to use profit as a measure of success, but profit is
not available in the data. However, profit is closely related to taxes paid, so taxes should
normally be a good proxy for profit. We acknowledge, however, that for emerging
growth companies, taxes paid may be a poor preditor of their value creation potential.
Note that taxes are truncated from below at 0. (Firms do not pay negative taxes no
matter how poor their performance.) Accordingly the estimation is done using Tobit
estimation rather than ordinary least squares. The basic finding is that there is a strong
negative relationship between whatever measure of performance we use and the share
of the venture owned by the venture capitalist. This could be the result of either the
moral hazard or the venture capitalist’s self-selection constraint. It is also possible that
ventures for which ␣ is high pay out more earnings to the venture capitalist, and, there-
fore, have lower future earnings. However, normalizing for asset size should mitigate
this concern.
We emphasize that the amount of variation explained by the venture capital share
is low. Thus, while the coefficient on the venture capital share is significant, variations
in this share are, at most, a minor determinant of performance. It is also important that
these results not be interpreted as suggesting that venture capital investment should
be viewed as a negative influence, or that other sources of finance are better than venture
capital. Venture capital investments could be an imortant positive influence on every
firm in the data set, and could be the best source of financial capital available, and we
would still expect to observe a negative correlation between venture capital ownership
and performance. What the negative correlation tells us is that the best performing com-
panies tend to be those in which the venture capital ownership share is not too high.
However, if financial requirements are high and the owner’s sources are meagre, then
a substantial venture capital share might be the best option, even if there is an associated
moral hazard problem, as the alternative might be outright failure of the company.

CONCLUDING REMARKS
The theoretical framework we offer focuses on informational issues. Specifically, we
view asymmetric information as the central feature of venture capital investment. Both
462 R. AMIT ET AL.

major forms of asymmetric information, “hidden information” (leading to adverse se-

lection) and “hidden action” (leading to moral hazard) are included in our analysis.
While the model abstracts from some important elements of the venture investment
process (such as bargaining, syndication, etc.), we believe that the informational issues
are perhaps the most central issues to focus on at this stage.
We have shown that this information-based approach is consistent with the data
on Canadian venture capital investments. Moral hazard and adverse selection create
a market failure in entrepreneurial financing, which might lead many worthwhile proj-
ects to be unfunded or underfunded. The more skilled the venture capitalist is in reduc-
ing these sources of market failure, the more effectively this sector will function. Ven-
ture capitalists exist because they are better at this function than unspecialized investors.
However, venture capitalists cannot eliminate adverse selection and moral hazard. Fur-
thermore, these problems are more acute for younger firms, and most acute for start-
ups. This explains why venture capitalists focus on later stage entrepreneurial firms.
Later stage firms have a track record that provides information to the entrepreneur,
and they have enough assets to reduce the problem associated with limited collateral
under limited liability. By virtue of their expertise, venture capitalists are better at deal-
ing with informational problems than are other investors (on average), but this advan-
tage shows up most in later stage entrepreneurial firms rather than at the start-up stage.
This theoretical structure is also consistent with the pattern of exit. If asymmetric
information is important, and remains important even at the exit stage, then outside
public investors will not be in the best position to evaluate the assets of the entrepreneur-
ial firm, and insiders will be in a better situation to buy out the venture capitalist’s posi-
tion. These insiders might be management or officers of the investee, or they might be
other firms in a related business. Thus, it is not surprising that IPOs account for only
a modest fraction of exit. In addition, our model predicts a negative relationship be-
tween the extent of venture capital ownership and firm performance. This relationship
is found in the data.
There are several natural extensions to the line of reasoning presented in the paper.
One complicating factor is the possibility that a venture capitalist’s cost of monitoring an
entrepreneur might vary with the venture capitalist’s ownership share. It is sometimes
suggested that it is easier for the venture capital firm to monitor if it has a larger owner-
ship share. In our model, this would suggest that m would exogenously depend on ␣.
Furthermore, we recognize that many aspects of venture capital activity have not been
captured in our analysis. In particular, we abstract from staged investment, which is a
common feature in venture capital finance and can serve to ameliorate problems caused
by asymmetric information. It would be interesting to extend our model to a multi-
period analysis.
The challenge we and other researchers face is to develop theoretical structures
that can be subject to empirical investigation. Ideally such theories should also provide
normative implications for practice. Our paper is a small but hopefully useful step in
this direction.

REFERENCES
Admati, A., and Pfleiderer, S. 1994. Robust financial contracting and the role of venture capitalists.
Journal of Finance 49(2):371–402.
 VENTURE CAPITAL INVESTMENT 463

Akerlof, G. 1970. The market for lemons: Quality uncertainty and the market mechanism. Quarterly
Journal of Economics 84(3):488–500.
Amit, R., Brander, J., and Zott, C. 1997. Venture capital financing of entrepreneurship in Canada. In
Paul Halpern, ed., Capital Market Issues in Canada. Industry Canada, Ottawa.
Amit, R., Glosten, L., and Muller, E. 1990. Entrepreneurial ability, venture investments, and risk shar-
ing. Management Science 36(10):1232–1245.
Amit, R., Glosten L., and Muller, E. 1993. Challenges to theory development in entrepreneurship re-
search. Journal of Management Studies 30(5):815–834.
Arrow, K. 1973. The Limits of Organization. New York: Norton.
Barry, C., Muscarella, C., Peavy, J., and Vetsuypens, M. 1990. Venture capital in the creation of public
companies: evidence from the going-public process. Journal of Financial Economics 27(2):
447–71.
Business Development Bank of Canada (Annual, 1993–95), Economic Impact of Venture Capital
(Montreal).
Bygrave, W., and Timmons, J. 1992. Venture Capital at the Crossroads. Boston, MA: Harvard Business
School Press.
Chan, Y. 1983. On the positive role of financial intermediation in allocations of venture capital in a
market with imperfect information. Journal of Finance 38(5):1543–1561.
Chan, Y., Siegel, D., and Thakor, A. 1990. Learning, corporate control and performance requirements
in venture capital contracts. International Economic Review 31(2):365–381.
Fried, V., and Hisrich, R. 1994. Toward a model of venture capital investment decision making. Finan-
cial Management 23(3):28–37.
Gompers, P. 1995. Optional investment, monitoring and the staging of venture capital. Journal of Fi-
nance 50(5):1461–1489.
Gompers, P. 1996. Grandstanding in the venture capital industry. Journal of Financial Economics
42(1):133–156.
Gompers, P., and Lerner, J. 1994. A note on the venture capital industry. Harvard Business School
Note N9-295–065.
Gorman, M., and Sahlman, W.A. 1989. What do venture capitalists do? Journal of Business Ventur-
ing 4(4):231–248.
Hellmann, T. 1994. Financial structure and control in venture capital. Stanford University Work-
ing Paper.
Hirao, Y. 1993. Learning and incentive problems in repeated partnerships. International Economic
Review 34(1):101–119.
Industry Canada, June 1994. Financing the new economy: towards a positive conspiracy. Project Report.
Jensen, M., and Meckling, W. 1976. Theory of the firm: managerial behavior, agency costs, and owner-
ship structure. Journal of Financial Economics 3(4):305–360.
Lerner, J. 1994. The syndication of venture capital investment. Financial Management 23(3):16–27.
Lerner, J. 1995. Venture capitalists and oversight of private firms. Journal of Finance 50(5):301–318.
Low, M., and MacMillan, I. 1988. Entrepreneurship: past research and future challenges. Journal of
Management 14(2):139–161.
Macdonald & Associates Ltd. (annual) Venture Capital in Canada: Annual Statistical Review and Direc-
tory, (Association of Canadian Venture Capital Companies: Toronto).
MacIntosh, Jeffrey G. 1994. Legal and Institutional Barriers to Financing Innovative Enterprise in Can-
ada, Discussion Paper 94–10, School of Policy Studies, Queen’s University, Kingston.
MacMillan, I., Siegel, R., and Narashima, P. 1985. Criteria used by venture capitalists to evaluate new
venture proposals. Journal of Business Venturing 1(1):119–128.
MacMillan, I., Zemann, L., and Narashima, P. 1987. Criteria distinguishing successful from unsuccessful
ventures in the venture screening process. Journal of Business Venturing 2(2):123–137.
Megginson, W., and Weiss, K. 1991. Venture capitalist certification in initial public offerings. Journal
of Finance 46(3):879–903.
464 R. AMIT ET AL.

Pauly, M. 1974. Overinsurance and public provision of insurance: the roles of moral hazard and adverse
selection. Quarterly Journal of Economics 88:(1):44–54.
Rothschild, M., and Stiglitz, J. 1976. Equilibrium in competitive Insurance Markets: an essay on the
economics of imperefect information. Quarterly Journal of Economics 90(9):629–649.
Sahlman, W. 1990. The structure and governance of venture-capital organizations. Journal of Financial
Economics 27(2):473–521.
Spence, M. 1973. Job market signalling. Quarterly Journal of Economics 87(3):355–374.
Tyebjee, T., and Bruno, A. 1984. A model of venture capital investment activity. Management Sci-
ence 30(9):1051–1066.

APPENDIX 1
Illustration of the Basic Moral Hazard Problem
Figure A1 shows the marginal cost of effort (a horizontal line) and the marginal
expected benefit of effort (given by Re). The efficient amount of effort occurs where
marginal benefit equals marginal cost, and is denoted e* in the diagram. The marginal
benefit perceived by the entrepreneur is only (1 ⫺ ␣)Re, which is strictly below Re.
It follows that the amount of effort actually chosen, denoted e⬘, is less than the efficient
amount. The basic problem is that the entrepreneur cannot precommit to provide
effort level e*. Once financing is obtained and share ␣ of the firm has been sold to
the investor, the entrepreneur will only provide effort level e⬘. If the investor and the
entrepreneur could contract over e, then they could agree that e* would be provided,
but this is impossible under the assumption that e cannot be observed (or at least
legally verified) by the investor.

FIGURE A1 Moral hazard.

 VENTURE CAPITAL INVESTMENT 465

APPENDIX 2
Formal Analysis of the Adverse Selection Problem
From (18) we derive the first-order condition
∞
EVd ⫽ p⬘(d) 冮q0 (␣R(q) ⫺ C)f(q)dq ⫺ 1 ⫽ 0 (A2.1)

where F is the cumulative distribution of q. Let

∞
K⫽ 冮q0 (␣R(q) ⫺ C)f(q)dq (A2.2)

Then (A2.1) simplifies to

1
p⬘(d) ⫽ (A2.3)
K
To derive the second-order condition, EVd is differentiated with respect to d, yielding
∞
EVdd ⫽ p″(d) 冮q0 (␣R(q) ⫺ C)f(q)dq
⫽ p″(d) K (A2.4)
It follows from (17) and (18) that (A2.4) is strictly negative, which is the precondi-
tion for (A2.3) to yield a maximum.
(A2.3) has interesting implications. Suppose that R(q) is such that there are many
worthwhile projects and a few projects that have very low negative expected returns.
q0 is therefore low. Specifically, assume that K is relatively large, resulting in a rather
low value of p⬘(d), which in term implies a relatively large optimal value of d (if a
solution to (A2.3) exists at all). Thus, with such a constellation of parameters, it pays
to invest high d in due diligence. On the other hand, if R(q) is such that K is relatively
small (which may happen if there are only few attractive projects and many ‘lemons,’
i.e., if q0 is high), this will result in a relatively small optimal value of d (depending,
of course, on the shape of p(d)).
In order to illustrate the point that an investor with a highly responsive detection
function p(d) (say, investor h with a detection function p(dh)) is more likely to invest
in projects with high asymmetry of information than an investor with a less responsive
p(d) (say, investor l with a detection function p(dl)), let us consider the following case.
Assume that q0 is high and K is small, resulting, according to (A2.3), in a large p″(d).
This is fairly realistic, as the pattern of returns of venture capitalists is usually skewed
with most investments generating either disappointing or negative returns and only a
few becoming ‘stars’.
It may happen that investor h finds it worthwhile to spend dh* ⬎ 0 (which is the
value of dh that satisfies A2.3) and go ahead with project q ⱖ q0, while investor l finds
that the optimal value of dl is dl* ⫽ 0 and thus refrains from investing. (Of course,
even if dh* ⬎ 0, the investor’s feasibility constraint (20) has to hold before investment
I is made.) These points are illustrated in Figure A2.
Note that for some values of K, both dl* and dh* can be positive in our example.
Then it pays even for investor l to do due diligence. Again, it also depends on constraint
(20) whether either investor l or h or both find the investment attractive.
466 R. AMIT ET AL.

FIGURE A2 Optimal due diligence for different detection functions p(d).