Description
In this particular brief description relating to management science product market competition and the financing of new ventures.
This article was downloaded by: [130.15.189.191] On: 18 March 2015, At: 06:57
Publisher: Institute for Operations Research and the Management Sciences (INFORMS)
INFORMS is located in Maryland, USA
Management Science
Publication details, including instructions for authors and subscription information:http://pubsonline.informs.org
Product Market Competition and the Financing of New
Ventures
Jean-Etienne de Bettignies, Anne Duchêne
To cite this article:
Jean-Etienne de Bettignies, Anne Duchêne (2015) Product Market Competition and the Financing of New Ventures.
Management Science
Published online in Articles in Advance 29 Jan 2015
.http://dx.doi.org/10.1287/mnsc.2014.2041
Full terms and conditions of use:http://pubsonline.informs.org/page/terms-and-conditions
This article may be used only for the purposes of research, teaching, and/or private study. Commercial use
or systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisher
approval, unless otherwise noted. For more information, contact [email protected].
The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitness
for a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, or
inclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, or
support of claims made of that product, publication, or service.
Copyright © 2015, INFORMS
Please scroll down for article—it is on subsequent pages
INFORMS is the largest professional society in the world for professionals in the fields of operations research, management
science, and analytics.
For more information on INFORMS, its publications, membership, or meetings visithttp://www.informs.org
MANAGEMENT SCIENCE
Articles in Advance, pp. 1–19
ISSN 0025-1909 (print) ISSN 1526-5501 (online)
http://dx.doi.org/10.1287/mnsc.2014.2041
©2015 INFORMS
Product Market Competition and the
Financing of New Ventures
Jean-Etienne de Bettignies
Queen’s School of Business, Queen’s University, Kingston, Ontario K7L 3N6, Canada,
[email protected]
Anne Duchêne
IPAG Business School, 75006 Paris, France; and Lebow School of Business, Drexel University,
Philadelphia, Pennsylvania 19104, [email protected]
T
his paper examines the interaction between venture risk, product market competition, and the entrepreneur’s
choice between bank ?nancing and venture capital (VC) ?nancing. Under bank ?nancing, a debt-type con-
tract emerges as optimal, which allows the entrepreneur to retain full control of the venture and thus yields
strong effort incentives, as long as the entrepreneur can service the debt repayment; however, this leads to liqui-
dation in the case of default, making the venture’s success quite sensitive to exogenous, even temporary, shocks
that may hinder debt repayment. Under VC ?nancing, an equity-type contract emerges as optimal. Although it
is not sensitive to exogenous shocks, this contract requires the entrepreneur to share a fraction of the rents with
the ?nancier, thus yielding lower effort incentives for the entrepreneur. There exists a threshold level of venture
risk such that bank ?nancing is optimal if and only if venture risk is below that threshold. Product market
competition increases the value of stronger entrepreneurial incentives and thus increases the maximum level of
risk the entrepreneur is willing to take before switching from bank ?nancing to VC ?nancing. This is a robust
result that is shown to hold in various models of competition, including the Hotelling, Salop, Dixit–Stiglitz, and
Cournot-to-Bertrand switch.
Keywords: competition; venture risk; bank ?nancing; venture capital
History: Received June 17, 2010; accepted July 17, 2014, by Lee Fleming, entrepreneurship and innovation.
Published online in Articles in Advance.
1. Introduction
Small ?rms are of critical importance to the econ-
omy.
1
In the United States, they represent 99.9% of
all ?rms, employ half of all private sector employ-
ees, produce 16 times more patents per employee than
large patenting ?rms (U.S. Small Business Adminis-
tration 2014), and have a signi?cantly positive impact
on urban growth (Glaeser et al. 2014).
The creation and growth of these entrepreneurial
ventures often requires external capital, and ?nan-
cial intermediaries such as banks and venture cap-
italists play a key role in that regard. What factors
affect the entrepreneur’s choice between the two types
of ?nancing? In practice it is sometimes argued that
entrepreneurs prefer bank-provided debt ?nancing
over venture capital and thus would always choose
bank ?nancing as long as it is feasible, that is, as
long as the venture has enough assets to collateralize
the loan. Under that hypothesis, a key factor in the
1
The Of?ce of Advocacy of the U.S. Small Business Administration
(SBA) de?nes a small business as an independent business with
fewer than 500 employees.
entrepreneur’s choice would be some ratio of collat-
eral value to investment requirement. This is a plausi-
ble, but likely incomplete, hypothesis: it relies on the
argument that unsecured debt is very dif?cult if not
impossible to obtain in new ventures, an argument
somewhat inconsistent with the fact that 40% of loans
taken by small ?rms are unsecured (Leeth and Scott
1989). In this paper, we suggest that two key factors—
venture risk and product market competition—may
affect the ?nancing strategy of new ventures (see the
discussion of the related empirical evidence in the
next section), and we examine the interaction between
these two factors and the trade-off between bank
?nancing and venture capital (VC) ?nancing.
We propose a two-period, incomplete contracting
model in which a wealth-constrained entrepreneur
chooses between bank ?nancing and VC ?nancing,
anticipating that a negative shock may or may not
occur at date 1. In our simple setup, under bank
?nancing a debt-type contract emerges as optimal.
On one hand, conditional on repaying the debt, the
entrepreneur retains full control of the venture and, as
a result, full access to the venture’s realized pro?t. She
1
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
2 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
thus has strong incentives to exert effort and create
value.
2
On the other hand, default on the debt leads
to liquidation, making the venture’s success quite sen-
sitive to exogenous, even temporary, shocks that may
hinder debt repayment.
We argue that unlike the bank, which receives none
of the realized pro?t when the entrepreneur controls
the ?rm, the venture capitalist can use expertise, social
networks, and ?nancial resources to extract a fraction
of ex post pro?t even when the entrepreneur con-
trols the venture. Accordingly, under VC ?nancing, an
equity-type contract emerges as optimal. This contract
is not sensitive to exogenous shocks but requires the
entrepreneur to share a fraction of the pro?t with the
?nancier, yielding lower entrepreneurial effort. Thus
our ?rst key result emerges: there exists a threshold
level of venture risk (probability of a negative shock
occurring) such that below that threshold the entrepre-
neurial incentive advantage of the debt contract more
than offsets the disadvantage associated with expected
default and liquidation, and bank ?nancing is pre-
ferred to VC ?nancing. Above that risk threshold, the
elimination of default risk associated with equity-type
contracts more than offsets the incentive disadvantage,
and VC ?nancing is optimal.
In our model, the entrepreneur’s effort increases the
probability of gaining a cost advantage over her prod-
uct market rival. We show that competition, measured
by the degree of substitutability between products,
increases the value of a cost reduction—the marginal
product of entrepreneurial effort. The intuition is sim-
ple: in markets with more homogeneous products,
consumers become more sensitive to prices, and this
increases the demand advantage of the cost-leading
entrepreneur charging a lower price than her rival.
The value of gaining a cost advantage thus increases
with competition, and this disproportionately favors
bank ?nancing, where (as discussed) entrepreneurial
effort and the resulting likelihood of a cost reduction
are greater in equilibrium.
Thus, competition increases the appeal of bank
?nancing relative to VC ?nancing—it raises the max-
imum level of risk the entrepreneur is willing to take
before switching from bank ?nancing to VC ?nanc-
ing. This second key result is robust and is shown
to hold in various models of competition, includ-
ing the Hotelling, Salop, Dixit–Stiglitz, logit, and
Cournot-to-Bertrand switch. It also is consistent with
anecdotal evidence,
3
and with the empirical evidence
discussed in the next section.
2
For clarity purposes, throughout the text we refer to entrepreneurs
as female and to ?nanciers as male.
3
Consider, for example, the opinion of Andras Forgacs, managing
director at Richmond Global, a New York VC ?rm, on looking
at investment opportunities in Chengdu, China: “There are some
interesting investment opportunities where there is less competition
There exists a fairly large theoretical literature on
the impact of debt ?nancing on the competitive
behavior of rivals in the product market.
4
Brander
and Lewis (1986) pioneered this line of research when
they examined the impact of debt on competitive
behavior in a Cournot duopoly. They argued that
debt forces ?rms to focus their attention on the good
states of the world where the marginal return to out-
put is highest, and hence can serve as a commitment
to increase output, resulting in a reduction in the
rival’s own output choice. Showalter (1995) showed
that similarly, under price competition, debt could
serve as a commitment device to raise one’s own
prices and those of rivals. In contrast, Bolton and
Scharfstein (1990) argued—in an optimal contracting
framework—that debt may lead to predatory behav-
ior by deep-pocketed rivals. In turn, Faure-Grimaud
(2000) and Povel and Raith (2004) brought together
strategic elements of the Cournot approach of Brander
and Lewis (1986) and optimal contracting elements of
Bolton and Scharfstein (1990).
5
More recently, a new theoretical literature has
emerged that examines the entrepreneur’s choice
between VC ?nancing and bank ?nancing more
speci?cally. Factors affecting this ?nancing trade-off
are shown to include intellectual property protec-
tion and the “high-techness” of projects (Ueda 2004),
entrepreneur/investor input complementarity (Bet-
tignies 2008), entrepreneurial “stigma of failure”
(Landier 2003), and “strategic uncertainty” (Winton
and Yerramilli 2008).
6
Although these two literatures provide valuable
insights into the impact of ?nancial structure on com-
petitive behavior and into entrepreneurial ?nancing
[emphasis added] and more patient development of companies
and talent” (Knowledge@Wharton 2011).
4
See, for example, Brander and Lewis (1986), Maksimovic (1988),
Poitevin (1989), Bolton and Scharfstein (1990), Glazer (1994),
Showalter (1995), Faure-Grimaud (2000), and Povel and Raith
(2004). See also Peress (2010), who examines the more general inter-
action between product market competition and ?rms’ behavior in
?nancial markets.
5
There also exists a small body of literature on venture capital and
start-up ?nancing in a competitive context. In particular, Inderst
and Mueller (2004) consider the impact of competition among start-
ups on VC investment decisions and ?nancing contracts: competi-
tion shifts the bargaining power between the entrepreneur and the
VC and thus affects ownership shares of the two parties and valu-
ation of the project. Additionally, Fulghieri and Sevilir (2009) study
how market conditions affect the composition of a venture capi-
talist’s portfolio (in particular the trade-off between larger/smaller
portfolios and diversi?ed/concentrated portfolios).
6
Other trade-offs have been examined in the entrepreneurial
?nance literature, for example, the entrepreneur’s choice between
venture capital and angel ?nancing (Chemmanur and Chen 2014),
or between private and public ownership (Boot et al. 2006). See also
Renucci (2014), who examines bargaining with a venture capitalist
when bank ?nancing is an option.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 3
trade-offs, neither can address the issue examined
here, that is, the interaction among product mar-
ket competition, venture risk, and the entrepreneur’s
choice between bank ?nancing and venture capi-
tal.
7
Indeed—and this is the key contribution of this
paper—we build a simple model of entrepreneurial
?nance where relatively risky ventures use VC ?nanc-
ing while safer ventures use bank ?nancing, and
where the positive impact of competition on the
appeal of bank ?nancing relative to VC ?nancing
emerges as a robust result.
8
One exception in the prior literature, and one more
closely related to this paper, is the recent work of
Inderst and Mueller (2009), which offers a compelling
analysis of the interaction between product market
competition and the entrepreneur’s choice between
active investors (e.g., venture capitalists) and passive
ones (e.g., banks). In their model, active investors may
enable ventures to “‘strategically overinvest’ early on,
thus forestalling their rivals’ future investment and
growth” (p. 276), a strategic advantage for active
investors that becomes more valuable as competition
intensi?es. In our paper, in contrast, the distinction
between bank and VC ?nancing does not rest on
strategic overinvestment, but rather on differences in
entrepreneurial incentives, and it generates a different
outcome: here, product market competition leads to
less VC ?nancing and more bank ?nancing, not the
other way round.
This paper is organized as follows: Section 2 dis-
cusses empirical evidence related to VC ?nancing and
bank ?nancing, and to the effects of venture risk and
product market competition on the trade-off between
the two types of ?nancing. Section 3 describes the
basic setup, and §4 presents a simple trade-off between
bank ?nancing and VC ?nancing. Section 5 exam-
ines the impact of product market competition on
the ?nancing trade-off. Section 6 discusses empiri-
cal implications of the model, and §7 concludes. All
proofs are in the appendix. Extensions to the main
model are presented in the online supplement to this
paper (available on the authors’ websites).
2. Empirical Motivation
Although a variety of sources of ?nancing are avail-
able to entrepreneurs (Fraser 2005, Industry Canada
7
The concept of “strategic uncertainty” used in Winton and
Yerramilli (2008) is somewhat related to the concept of venture risk
used here. However, the trade-off between bank and VC ?nancing
is very different in their model, as are the effects of strategic uncer-
tainty on that trade-off. As well, product market competition is
altogether absent from their model.
8
In a differentiated Cournot extension of their model, Povel and
Raith (2004) brie?y discuss the impact of product homogeneity on
the competitive behavior of ?rms. However they do not examine
the impact of competition on the entrepreneur’s choice between
bank and VC ?nancing.
2009, Robb and Robinson 2014), two ?nancial inter-
mediaries emerge as particularly relevant. First, banks
play a key role for venture creation and growth.
Cosh et al. (2009) point out that among the UK ?rms
that sought external ?nancing during 1996–1997, 81%
approached banks, far more than any other ?nancial
intermediary. Fraser’s (2005) and Robb and Robin-
son’s (2014) analyses suggest, respectively, that banks
were also the most frequent capital provider for UK
start-ups in 2004 and for U.S. start-ups over the
2004–2007 period.
Another ?nancial intermediary playing a key role
in the creation and development of new ?rms is the
VC fund. Although VC funds are used by a nar-
rower set of entrepreneurs (Fraser 2005, Cosh et al.
2009), they provide much more capital to the ventures
that they do back. In the United States, the average
amount of funding from VC funds over 2004–2007
was $1,162,898, whereas the average funding in busi-
ness bank loans was $261,358 (Robb and Robinson
2014). Moreover, VC-?nanced ?rms helped to create
many successful companies, including Apple, Intel,
Federal Express, and Microsoft (Sahlman 1990), and
VC-backed companies generate 5%–7% of employ-
ment in the United States (Puri and Zarutskie 2012).
Thus, both banks and VC funds are essential to
the ?nancing of new ventures and indeed are substi-
tute forms of ?nancing in that, as shown by Berger
and Schaeck (2011), VC-backed ?rms are less likely
to rely on bank ?nancing. So what factors affect
the entrepreneur’s choice between these two types
of ?nancing? Two factors in particular may have an
impact on the bank/VC trade-off.
2.1. Venture Risk
Anecdotal evidence suggests that “riskier” ventures—
de?ned as ventures that have a lower probability
of success, but a higher payoff in case of success—
are relatively more likely to be ?nanced by venture
capital, whereas “safer” ventures are more likely to
be ?nanced by banks. The vast majority of entre-
preneurial ventures are so-called “lifestyle” ventures
(e.g., hair salons, auto-repair shops, etc.), which have
a relatively low probability of failure but will never
yield very high growth or pro?t,
9
and these ventures
are predominantly bank ?nanced (Fraser 2005). On
the other hand, lifestyle ventures are never ?nanced
by venture capital. VC funds focus instead on high-
growth ventures,
10
which are riskier ventures: they
9
The terms “high-growth” and “lifestyle” ventures are commonly
used to divide entrepreneurial ventures into two broad categories.
See, for example, Hisrich et al. (2006, pp. 19–20) and Harvard Busi-
ness School (2006).
10
To the extent that (the inverse of) the time to bring a product
to market is a proxy for growth, this is consistent with Hellmann
and Puri’s (2000) ?nding that VC-backed ?rms enjoy a signi?cantly
lower time to market.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
4 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
have a relatively high probability of failure but can be
extremely pro?table when successful. Of course, some
high-growth ventures are bank ?nanced, and taken
together these facts suggest that riskier high-growth
ventures are more likely to be ?nanced by venture
capital, and safer lifestyle ventures are more likely to
be ?nanced by banks.
Cosh et al. (2009) explicitly examined the hypothe-
sis that riskier ventures are more likely to be ?nanced
by VC funds than by banks and found some evi-
dence consistent with it. For example, they found that
banks provide male-founded ?rms with 18% less of
their desired capital relative to female-founded ?rms.
Citing Jianakopolos and Bernasek (1998) to under-
line evidence that women are more risk averse than
men, they argue that this is illustrative of a nega-
tive connection between bank ?nancing and venture
risk. They also ?nd that ?rms that recently devel-
oped a new innovation and hence are—as implicitly
argued—riskier are more likely to seek VC ?nancing.
Similarly, Hellmann and Puri (2000) found that
innovator ?rms are signi?cantly more likely to obtain
VC ?nancing, and obtain VC ?nancing earlier in the
life cycle, than imitator ?rms. To the extent that inno-
vators are riskier than imitators, and/or that there is
more risk earlier in the venture life cycle, this is con-
sistent with a positive relationship between venture
risk and VC ?nancing.
2.2. Product Market Competition
A vast empirical literature exists that examines the
connection between product market competition, on
one hand, and decisions, behavior, and performance
in organizations on the other. Perhaps the most
compelling ?nding to emerge from this broad line
of research is that competition affects organiza-
tions in many ways.
11
What about the interaction
between competition and ?nancing strategy specif-
ically? Importantly here, Cosh et al. (2009) show
that entrepreneurs are more likely to choose bank
?nancing, and less likely to choose VC ?nancing, as
the degree of competition, measured by the number
and/or size of competitors, increases.
Moreover, as discussed above, both Hellmann and
Puri (2000) and Cosh et al. (2009) ?nd evidence of
a positive relationship between venture innovative-
ness and VC ?nancing. Although we highlighted the
11
The evidence suggests a positive impact of competition on inno-
vation (Geroski 1990, Bertschek 1995, Blundell et al. 1999, Bloom
et al. 2011, Teshima 2010, Hashmi 2013) and productivity (Caves
and Barton 1990, Green and Mayes 1991, Haskel 1991, Nickell 1996,
Grif?th 2001, Syverson 2004a, b), the power of managerial incen-
tives (Burgess and Metcalf 2000, Cuñat and Guadalupe 2005, Baggs
and Bettignies 2007), collective bargaining and wage negotiations
(Abowd and Lemieux 1993), mergers and acquisitions (Hoberg and
Phillips 2010), and IPO decisions (Chemmanur and He 2011).
strong positive correlation between venture innova-
tiveness and venture risk, it is also important to
underline the negative correlation between venture
innovativeness and product market competition. In
Hellmann and Puri (2000, p. 969), for example, “the
innovator variable captures the notion of ?rms that
introduce a new product that is considered not to be
a close substitute to any product or service already
offered on the market; ?rms that introduce a new
product or service that is considered to perform an
order of magnitude better than any substitute prod-
ucts already offered in the market; or ?rms that
are developing new technologies that could lead to
products satisfying either of the two criteria above.”
Indeed this corresponds quite closely to a description
of lack of competition in the product market.
12
To the
extent that innovation captures (lack of) competition,
the evidence presented in these two papers is consis-
tent with a negative connection between competition
and VC ?nancing.
Related to this, we note that patents, which are
associated with less substitutable products and con-
fer signi?cant monopoly power to their holders, may
arguably capture lack of competition. Kortum and
Lerner’s (2000) ?nding that VC-?nanced ?rms have
higher quality patents than non-VC-?nanced ?rms
then also suggests a negative connection between
competition and VC ?nancing.
Finally, the evolution of the ?nancing strategy over
the product life cycle—namely the relative prevalence
of VC ?nancing early in the product life cycle and of
bank ?nancing later (Berger and Udell 1998)—could
also be the result of a connection between competition
and ?nancing choice. Early in the life cycle, when the
entrepreneur introduces a new product and creates a
new market, competition is low and VC ?nancing is
chosen. As the life cycle progresses, however, similar
products emerge, competition intensi?es, and bank
?nancing eventually becomes the preferred choice.
The above discussion yields two key implications
regarding the bank/VC ?nancing trade-off. It sug-
gests ?rst that riskier ventures are more likely to be
VC ?nanced than bank ?nanced, and second that
a greater degree of product market competition is
likely to favor bank ?nancing over VC ?nancing. In
what follows we propose a theory of entrepreneurial
?nancing that highlights the interaction among the
bank/VC ?nancing trade-off, venture risk, and prod-
uct market competition.
12
The degree of substitutability between products (e.g., Sutton
1992), and the similarity between perceived product performance
and/or between ?rms’ costs (e.g., Bertrand competition with asym-
metric costs or performance components), are often used as mea-
sures of competition in industrial organization.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 5
3. Basic Model
An entrepreneur c has a novel venture idea that may
open a new and untapped market. Entrepreneur c is
wealth constrained with no initial wealth and is solic-
iting a (small) cash investment í from a ?nancier ]
to start her venture, which is to last for two periods.
Both c and ] are risk neutral, and ] ’s cost of capital
is normalized to zero.
At the beginning of the game, c makes a take-it-or-
leave-it contractual offer to ] ,
13
specifying a credible
way in which ] is to receive a nonnegative expected
payoff (net of í ) from the investment, thus ensur-
ing his participation. Once the initial investment is
made, the venture is started, but it requires c’s active
management for its success. The key elements of the
model can be described as follows.
First-Period Pro?t and Exogenous Shocks. Under c’s
management, the venture is launched in the ?rst
period, offering goods/services at marginal cost ¯ c ?
+
in a new monopolistic market. One of two states
of the world emerges. In the good state, which occurs
with exogenous probability (1 ?í
1
), with í
1
? |0, 1),
the venture generates monopoly pro?t H
1
(í
1
). In the
bad state, which occurs with probability í
1
, however,
a negative and temporary
14
shock affects the venture,
which generates zero pro?t in that period as a result.
This shock captures an exogenous randomness that
is inherent to all ventures, for example, random con-
sumer preferences, or macroeconomic or technologi-
cal shocks.
The good-state pro?t H
1
(í
1
) is strictly increas-
ing in í
1
such that the expected ?rst-period pro?t
(1 ?í
1
)H
1
(í
1
) =
¯
H
1
is a constant for all í
1
? |0, 1).
Conveniently, one can easily verify that the variance
of the ?rst-period pro?t is a strictly increasing func-
tion of the shock probability í
1
, and in what follows
we refer to í
1
as the “short-term risk” of the venture.
Thus, two projects with different shock probabilities
yield the same expected short-term pro?t but face dif-
ferent levels of risk.
Second-Period Pro?t and Product Market Competition.
In the second period, a rival r enters the market, offer-
ing related but distinct goods/services at marginal
cost ¯ c. Thus emerges competition in the product mar-
ket. Parameter 0 ? |0
min
, 0
max
], with 0
max
> 0
min
> 0,
captures the degree of competition in the industry,
which, as discussed in §5, will depend on the way
demand/competition is modeled: it may represent the
inverse of the transport cost in a Hotelling (1929)
13
The entrepreneur has all ex ante bargaining power: there are
many more ?nanciers wishing to invest than there are good entre-
preneurs (good projects to be funded).
14
This negative shock is temporary in that it does not affect second-
period payoffs. The case of a long-term shock that yields zero pro?t
in the second period as well (or that conveys negative information
about second-period pro?tability prospects) is less interesting (as
discussed in Footnote 33).
or Salop (1979) model, or the inverse of variance of
errors in a logit model, for example.
Bene?tting from a ?rst-mover advantage, under c’s
management the venture may—through innovation—
lower the marginal cost of production from ¯ c to c,
where c - ¯ c, thus gaining a cost advantage over
the entrant and generating “high” pro?t H
l
(0) =
H(c, ¯ c, 0) > 0.
15
The successful innovation required
for gaining a cost advantage takes place with endoge-
nous probability ¡ ? |0, 1). With probability 1 ?¡ the
marginal cost remains unchanged at ¯ c, and in the sec-
ond period the venture yields “status quo” or “sym-
metric” pro?t H
s
(0) = H( ¯ c, ¯ c, 0) > 0. Accordingly, in
the second period, c’s venture generates the expected
duopoly pro?t H
2
(¡, 0) = H
s
(0) + ¡(H
l
(0) ? H
s
(0)),
which can be expressed as the sum of the status quo
pro?t H
s
(0) and the expected gain from a cost advan-
tage ¡(H
l
(0) ?H
s
(0)).
Note that, similar to the ?rst period, we could have
an exogenous negative second-period shock occurring
with probability í
2
and leading to zero pro?t in that
period. However, unlike the ?rst-period shock, which
will be shown to play a key role in our model, a
second-period shock plays little role here, and indeed
our results can be shown to hold for all í
2
? |0, 1).
Hence, for simplicity, and without loss of generality,
we assume í
2
=0.
Entrepreneurial Participation and Effort. Entrepre-
neur c affects the venture in two primary ways.
First, as mentioned above, her active management
is required for the venture to operate successfully.
Absent the entrepreneur in period 1, the venture
yields zero pro?t in that period. Absent the entre-
preneur in period 2, the venture is liquidated and
yields liquidation value |, with | ? (0, H
s
(0
max
)).
16
The
entrepreneur’s personal opportunity cost of remain-
ing involved in the venture is a > 0 with a ?0, and
she must thus expect a payoff of at least a in period i
to remain involved in the venture in that period.
The second way in which c can affect the venture
is by exerting effort u at the beginning of the second
period, which increases the probability ¡ of successful
innovation and of decreasing the marginal cost from
¯ c to c.
17
For simplicity we assume ¡ = u, and use ¡
15
In any pro?t function H(·, ·, 0), price–cost margin function
P(·, ·, 0), or demand function x(·, ·, 0), the ?rst argument refers to
the marginal cost of the ?rm generating the pro?t under study, and
the second argument refers to the rival’s marginal cost. Equilibrium
prices have been determined as functions of costs and substituted
out of these functions.
16
This condition is suf?cient to ensure that liquidation is inef?cient:
it generates less surplus than keeping the venture as a going con-
cern, regardless of entrepreneurial effort or degree of competition.
17
We could model entrepreneurial effort in both periods rather than
in the second period only. However, as will become clear below
(see expression (5)), ?rst-period pro?ts do not affect the trade-off
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
6 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
to capture both entrepreneurial effort and innovation
probability. Effort comes at a personal cost K(¡) =
(|,2)¡
2
to the entrepreneur. It is observable but not
veri?able in court and thus noncontractible.
Contractual Incompleteness and Property Rights. Real-
ized pro?ts at the end of periods 1 and 2 are assumed
to be observable by both parties, but not veri?able in
court, and thus cannot be contracted upon.
18
To quote
Bolton and Scharfstein (1996, p. 5), “this assumption
is meant to capture the idea that managers have some
ability to divert corporate resources to themselves at
the expense of outside investors,” and that “such perk
consumption and investment may be dif?cult to dis-
tinguish from appropriate business decisions and thus
impossible to control through contracts.” This has two
important consequences. First, c and ] must bargain
at the end of each period, over pro?t H
1
, and H
l
or H
s
, as they arise. Second, the contract designed at
date 0 can only specify the (possibly contingent) allo-
cation of property rights over the venture, which can
be of two types. Under entrepreneur control (|), c has
full ownership of the venture and complete control
over the assets of the venture. At the other end of
the ownership spectrum, under ?nancier control (í ),
] owns the venture and controls the assets.
Bank vs. Venture Capital. Entrepreneur c can seek
?nancing from one of two types of investors: a ven-
ture capitalist (or angel investor) or a banker. We
posit that banks and venture capitalists share simi-
larities and exhibit differences in the way in which
control-right allocations affect bargaining with the
entrepreneur. Financier control, for example, is likely
to affect ?nancier/entrepreneur bargaining similarly
whether the ?nancier is a bank or a venture capital-
ist. In both cases, ] has complete control over the
assets of the venture and can extract 100% of the real-
ized pro?t through bargaining at the end of a period.
This is consistent with Nash bargaining where, in case
negotiations break down, ] can simply exclude c from
accessing the venture and continue to enjoy all rents.
Entrepreneur c cannot interfere and gets nothing.
In contrast, we argue that entrepreneur control may
lead to different bargaining outcomes under bank
?nancing and VC ?nancing. Under bank ?nancing,
between bank ?nancing and VC ?nancing; hence neither does
potential ?rst-period entrepreneurial effort affecting these pro?ts.
To avoid unnecessary complications and keep the model as simple
as possible, we do not model ?rst-period effort explicitly here.
18
The assumption of nonveri?able cash ?ows ?ts squarely within
the recent literature on corporate/entrepreneurial ?nance, where
it is commonly used. See, for example, Hart and Moore (1989,
1994, 1995), Bolton and Scharfstein (1990, 1996), Berglöf (1994),
Gertner et al. (1994), Fluck (1998), Myers (2000), Bascha and Walz
(2001), Dybvig and Wang (2003), Repullo and Suarez (2004), and
Bettignies (2008).
entrepreneur control is the converse of ?nancier con-
trol: c has complete control over the assets of the ven-
ture and can extract 100% of realized pro?t through
bargaining at the end of a period. In case negotiations
break down during Nash bargaining, c can simply
exclude ] from accessing the venture and continue to
enjoy all rents. Financier ] cannot interfere and gets
nothing.
Under VC ?nancing, however, this bargaining out-
come is unlikely to happen even under entrepreneur
control. Indeed, a key characteristic of venture cap-
italists here is that—given their level of expertise,
their social networks, and their ?nancial resources—
they may be able to extract second-period rents even
when c owns the venture and controls its assets.
In the second period, once the venture capitalist is
familiar with the idea and its implementation, he
may be in a stronger bargaining position relative to
the entrepreneur because—as argued by Ueda (2004)
previously—in the case of a bargaining breakdown he
may be able to successfully “steal” the entrepreneur’s
idea with some probability \ ? |\
min
, 1] with \
min
?
(0, 1). In that case, as is easily shown, the Nash bar-
gaining equilibrium split would lead the entrepreneur
and the venture capitalist to obtain shares 1 ?\ and
\ of realized pro?t, respectively. Indeed, even with-
out any “stealing” by the venture capitalist, as long as
he has the ability to interfere with the entrepreneur’s
access to second-period rents in the case of a bar-
gaining breakdown,
19
even with a small probability,
then he will be able to extract a share of pro?t in
equilibrium.
20
Accordingly in what follows we assume that entre-
preneurial control under bank ?nancing leads the
entrepreneur to extract 100% of realized pro?ts
in both periods, whereas under VC ?nancing the
entrepreneur extracts all realized pro?ts in the ?rst
period, but only a fraction 1 ? \ of these realized
19
For example, suppose that in the case of a bargaining breakdown,
the venture capitalist can use his clout in entrepreneurial circles
and his access to related portfolio companies to interfere with the
venture’s pro?ts, and that as a result the entrepreneur ends up with
only a fraction q of the pro?ts she would have obtained without
interference. One can easily show that, in that case, the Nash bar-
gaining split would lead the entrepreneur and the venture capitalist
to obtain shares 1 ?\ =(1 +q),2 and \ =(1 ?q),2, respectively.
20
Note an implicit distinction between the venture capitalist and
the entrepreneur. On one hand, in the case of VC ?nancing, the ven-
ture capitalist can extract rents from the entrepreneur, even under
entrepreneur control, by essentially threatening to “steal” her idea
or interfere with her business in the case of a bargaining break-
down. In contrast, under bank ?nancing, the entrepreneur cannot
extract rents from the banker under ?nancier control. The reason
for this is that the entrepreneur, unlike the venture capitalist, does
not have the ?nancial resources required to credibly threaten the
banker to “steal” the idea (she could not ?nance a new venture
with the “stolen” idea), nor does she have the clout/networks to
credibly interfere with the banker’s access to rents.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 7
pro?ts in the second period.
21
As we shall see below,
this distinction between bank and VC ?nancing leads
to a signi?cant trade-off between these two ?nancing
possibilities.
Timing of the Game. We summarize the timing of the
game as follows:
22
• At date 0, c makes a contractual offer to ] , who
is either a banker or a venture capitalist. The contract
only speci?es the (possibly contingent) allocation of
property rights over the venture, as well as possible
date 0 monetary transfers. The entrepreneur decides
whether or not to manage the venture in the ?rst
period, at personal cost a, with a ?0.
• At date 1, under c’s management, the ven-
ture generates monopoly pro?t H
1
with probability
(1 ?í
1
), and zero pro?t with probability í
1
. Absent
c, the venture generates zero pro?t.
• At the beginning of the second period, rival r
enters the market. The entrepreneur decides whether
or not to manage the venture in the second period,
at personal cost a, with a ? 0. In the former
case, she exerts innovation effort ¡, at personal cost
K(¡) =|¡
2
,2.
• At date 2, under c’s management, the venture
generates expected pro?t H
2
(¡, 0) with probability 1?
í
2
=1. Absent c, the venture generates zero pro?t.
Regularity Condition. Throughout the paper we as-
sume that
í ?\
min
H
s
(0
max
). (1)
This is a suf?cient condition to ensure that the
project can be ?nanced under VC ?nancing for all
\ ? |\
min
, 1] and 0 ? |0
min
, 0
max
], and that there exists
an í
max
(0) > 0 such that for all 0 ? |0
min
, 0
max
] bank
?nancing is feasible if and only if (henceforth iff) í
1
?
|0, í
max
(0)]. Because our primary interest lies with the
trade-off between VC ?nancing and bank ?nancing,
we henceforth focus on situations where condition (1)
holds, and refer to it where relevant.
23
We relax this
21
We assume that the venture capitalist can extract a fraction \
of realized pro?ts even under entrepreneur control, but only in
the second period. This assumption is made to capture the idea,
important in our view, that it requires learning about the venture
before the venture capitalist can gain real bargaining power over
the entrepreneur, and for analytical simplicity. All results of the
model continue to hold if we assume the venture capitalist can
extract \ in both periods rather than in the second period only.
22
Our model shares some of the model timing elements, as well
as the assumptions of contractual incompleteness and entrepre-
neurial effort, with Bettignies (2008). Despite these similarities, the
two models differ signi?cantly. Bettignies’ (2008) focus is on the
effects of investor cost of capital and entrepreneur–investor effort
complementarity on an entrepreneur’s optimal contracting envi-
ronment. It cannot address issues of venture risk and competition
that are central to this paper, nor does it examine speci?cally the
entrepreneur’s choice between bank and VC ?nancing.
23
If parameter í is such that only one type of ?nancing is feasible,
the ?nancing decision problem becomes exceedingly simple and
obviously less interesting.
regularity condition in §A.4 in the appendix and show
that the main results of the model continue to hold.
First-Best Benchmark. We consider the ?rst-best
scenario where the entrepreneur is not wealth
constrained and can ?nance the venture herself. Pro-
ceeding by backward induction, we ?nd that in the
second period, if she actively manages the venture,
she exerts effort ¡
FB
to maximize her expected payoff
H
2
(¡, 0) ? |¡
2
,2 = H
s
(0) + ¡|H
l
(0) ? H
s
(0)] ? |¡
2
,2,
yielding ?rst-best effort ¡
FB
(0) = |H
l
(0) ? H
s
(0)],|.
Her equilibrium expected payoff, H
2
(¡
FB
(0), 0) ?
||¡
FB
(0)]
2
,2 is larger than a+|,
24
and anticipating this
she chooses to be actively involved in the venture in
the ?rst place. In the ?rst period, her expected payoff
¯
H
1
is also larger than her cost a of managing the ven-
ture, and hence she chooses to be actively involved in
the venture in that period. Thus, from a date 0 point
of view, c’s expected payoff, and the ?rst-best total
surplus created, can be expressed as
U
FB
(0) =
¯
H
1
+V
FB
(0) ?í , with
V
FB
(0) = H
2
(¡
FB
(0), 0) ?||¡
FB
(0)]
2
,2, (2)
where V
FB
(0) captures the venture’s (expected) contin-
uation surplus in the second period. Regularity con-
dition (1) ensures that the venture is worth ?nancing
in the ?rst-best scenario.
25
4. Optimal Financing Choice
We now turn to bank ?nancing and VC ?nancing and
examine how these sources of capital compare to the
?rst-best benchmark and to each other.
4.1. Bank Financing
The entrepreneur has two control-rights allocations
available to design the optimal contract for the
?nancier/banker: under ?nancier control in period i,
i = 1, 2, c anticipates that end-of-period bargaining
will result in ] extracting all rents, and in her getting
nothing. This zero expected payoff is less than her
opportunity cost a of actively managing the venture,
and accordingly she decides not to remain involved
in the venture during that period.
Under entrepreneurial control, in contrast, c antic-
ipates that end-of-period bargaining will result in
24
We know from c’s maximization program that her equilibrium
expected payoff in the second period is strictly larger than H
s
(0),
which itself is larger than a +|.
25
To see this, note that V
FB
(0) =H
2
(¡
FB
(0), 0) ?||¡
FB
(0)]
2
,2 ?H
s
(0)
for any given 0 ? |0
min
, 0
max
]. This is because c could generate a
payoff H
s
(0) if she chose effort ¡ =0, and must thus do at least as
well by choosing equilibrium effort ¡
FB
(0). Moreover, as discussed
below in §5, symmetric pro?t H
s
(0) is decreasing in the degree
of competition, and hence H
s
(0) ? H
s
(0
max
) for all 0 ? |0
min
, 0
max
].
From these two points, it follows directly that V
FB
(0) ?H
s
(0
max
) >í ,
which in turn implies U
FB
(0) ?0 for all 0 ? |0
min
, 0
max
].
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
8 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
her extracting all rents, as in the ?rst-best sce-
nario described above. Thus, entrepreneurial con-
trol in period i, i = 1, 2 leads to the ?rst-best out-
come in that period. In period 2, c chooses to be
actively involved in the management of the venture,
exerts the ?rst-best effort ¡
|
(0) = |H
l
(0) ? H
s
(0)],|
= ¡
FB
(0), and generates equilibrium expected payoff
H
2
(¡
|
(0), 0) ?||¡
|
(0)]
2
,2. In period 1, c also chooses
to be actively involved in the management of the ven-
ture, generating expected payoff
¯
H
1
in that period.
Following Hart and Moore (1989) and Bolton and
Scharfstein (1990, 1996), we know that in a framework
with these two feasible control allocations, the opti-
mal “pure strategy”
26
contract for c to offer to ] is
a simple debt contract, which speci?es at date 0 the
debt repayment D to be made at date 1. If c pays D to
] at date 1, she retains full control in period 2. On the
other hand, if c does not repay D at date 1, ] obtains
full control over the venture.
If the negative shock does not occur in period 1
and the venture generates pro?t H
1
, c faces two
options at date 1. First, she can repay D and retain
control over the venture in period 2. In that case,
as mentioned she exerts effort ¡
|
(0), and generates
expected second-period continuation surplus V
|
(0) =
H
2
(¡
|
(0), 0)?||¡
|
(0)]
2
,2. Alternatively, she can strate-
gically default (see, e.g., Bolton and Scharfstein 1990).
Doing so would lead to renegotiation since ?nancier
control and the ensuing liquidation yield a lower pay-
off (|) than entrepreneur control, and c’s payoff in
that case would depend on the outcome of renego-
tiation. Without loss of generality we assume that
] has full bargaining power in renegotiation follow-
ing a default
27
and requires payment D
max
(0) =V
|
(0)
from c in exchange for letting her retain venture con-
trol. Anticipating this, c makes debt repayment D at
date 1 if D?V
|
(0) but would default and renegotiate
if D >V
|
(0).
In contrast, if the negative shock does occur in
period 1, c gets no cash and must default at date 1. As
a result, ] takes control over the assets at that time,
and c quits the venture, and this in turn leads to liq-
uidation payoff |. (There can be no scope for renego-
tiation here since | >0 implies that ] is strictly better
off with venture control than without, and c has no
cash with which to renegotiate.)
26
For expositional simplicity, we focus on “pure strategy” contracts
where an allocation of control is assigned with probability 0 or 1
contingent on debt repayment or default. In the online supplement
to this paper, we allow for “mixed strategy” debt contracts whereby
c retains control with probability p
D
if she repays the debt, and
with probability p
0
if she does not. The main results of the model
continue to hold.
27
The main results of our model do not depend on the allocation
of bargaining power in renegotiation and indeed would continue
to hold under alternative bargaining allocations. See discussion in
§A.4 in the appendix.
Naturally, at date 0 ] foresees that he will get
at most V
|
(0) in the good state and | in the bad
state. Thus bank ?nancing will be feasible only if
í ?(1?í
1
)V
|
(0)+í
1
|, or í
1
?í
max
(0) with í
max
(0) =
(V
|
(0)?í ),(V
|
(0)?|). Regularity condition (1) ensures
that í
max
(0) >0 for all 0 ? |0
min
, 0
max
].
28
Since c has full bargaining power at date 0, she sets
D so that she extracts all rents from ] , i.e., such that
(1 ?í
1
)D +í
1
| ?í =0, for all í
1
? í
max
(0). Accord-
ingly, from a date 0 point of view, c’s expected payoff
from bank ?nancing is the total surplus generated by
the venture under that type of ?nancing:
U
|
(0) =
¯
H
1
+(1 ?í
1
)V
|
(0) +í
1
|?í , with
V
|
(0) = H
2
(¡
|
(0), 0) ?||¡
|
(0)]
2
,2.
(3)
Thus, absent a negative shock in the ?rst period,
bank ?nancing is ef?cient: it enables c to retain full
control over the venture and generates ?rst-best entre-
preneurial effort ¡
|
(0) = ¡
FB
(0) and in turn ?rst-best
continuation surplus V
|
(0) = V
FB
(0). On the other
hand, with probability í
1
a negative shock does occur
in the ?rst period, and this leads to inef?ciency:
although this shock is temporary and does not sig-
nal anything about the prospects of the venture in
the second period, it leads to the inef?cient liquida-
tion of a venture which should instead—in a ?rst-
best world—continue business as usual in the second
period. More formally, the inef?ciency associated with
bank ?nancing can be expressed as U
FB
(0) ?U
|
(0) =
í
1
|V
|
(0) ?|].
29
We summarize the foregoing results in the follow-
ing proposition.
Proposition 1. Conditional on short-term venture suc-
cess, bank ?nancing generates ?rst-best entrepreneurial
effort ¡
|
(0) =¡
FB
(0) and continuation surplus V
|
(0) =
V
FB
(0). On the other hand, conditional on short-term
failure—even when this failure is temporary and uninfor-
mative about long-term prospects—bank ?nancing leads to
inef?cient liquidation of the venture at date 1.
28
As demonstrated in Footnote 25, V
|
(0) = V
FB
(0) > H
s
(0
max
).
Together with our condition that | - H
s
(0
max
), this implies
V
|
(0) >|, and together with regularity condition (1) this implies
V
|
(0) >í . From this it immediately follows that í
max
(0) >0 for all
0 ? |0
min
, 0
max
].
29
Note that these characteristics of bank ?nancing are similar to
the classic models of debt mentioned above (Hart and Moore 1989;
Bolton and Scharfstein 1990, 1996). Here, however, second-period
pro?ts are driven by endogenous efforts, and (as shall become clear
in §§4.3 and 5.2) this will play a crucial role in the analysis of
the trade-off between bank ?nancing and VC ?nancing, and of the
effects of product market competition, neither of which were exam-
ined in these prior models.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 9
4.2. VC Financing
With VC ?nancing, the entrepreneur again has two
control-rights allocations available to design the opti-
mal contract: ?nancier control in period i, i = 1, 2,
yields the same outcome as that under bank ?nanc-
ing: c anticipates getting nothing in end-of-period
renegotiation and decides not to remain involved in
the venture during that period.
Entrepreneurial control, on the other hand, is differ-
ent with VC ?nancing. In period 2, c anticipates that
end-of-period bargaining will result in her extracting
a fraction 1 ?\ of the rents. If she manages the ven-
ture, she exerts effort ¡
.c
to maximize her expected
payoff (1 ? \)H
2
(¡, 0) ? |¡
2
,2 = (1 ? \)|H
s
(0) +
¡|H
l
(0) ?H
s
(0)]] ?|¡
2
,2. She has lower effort incen-
tives than under entrepreneurial control with bank
?nancing, since she anticipates extracting a fraction
(1 ?\), rather than 100%, of the rents and thus exerts
a second-best effort:
¡
.c
(0) =(1 ?\)
H
l
(0) ?H
s
(0)
|
-¡
FB
(0).
Her equilibrium expected payoff,
(1 ?\)H
2
(¡
.c
(0), 0) ?
||¡
.c
(0)]
2
2
,
is larger than a+(1–\)| (for reasons provided in Foot-
note 24), and anticipating this she chooses to remain
actively involved in the venture in that period. In
period 1, again her expected payoff
¯
H
1
is larger than
her cost a of managing the venture, and she remains
actively involved.
Entrepreneurial control enables c to commit to
return some rents to the venture capitalist through
bargaining at the end of period 2. In this environment,
it follows directly that it is optimal for c to offer the
following contract to the venture capitalist: entrepre-
neurial control in both periods regardless of the state
of the world and a payment í +T from the venture
capitalist to the entrepreneur at date 0, with T ? 0
chosen so as to leave no ex ante rents to the ?nancier
(i.e., such that \H
2
(¡
.c
(0), 0) ?í ?T =0). Since c can
use date 0 transfer T to extract all ex ante rents,
her contractual objective is to maximize total sur-
plus. Additionally, since entrepreneurial control leads
to more effort—and hence, ceteris paribus, to more
pro?t—than investor control, it is optimal to assign
entrepreneurial control regardless of the state of the
world.
30
30
In the online supplement to this paper, we examine an extension
of the main model where the venture capitalist exerts effort as well.
In that case the optimal allocation of control rights depends on the
ef?ciency of c relative to ] . Indeed, if the venture capitalist is suf?-
ciently more ef?cient than c in exerting effort, giving control rights
to him may, by increasing his incentives, generate more surplus
than giving control rights to c. In that case c may do just that: she
may sell the venture to the venture capitalist.
Note that the maximum gross payoff that ] can
expect to get under VC ?nancing is \H
2
(¡
.c
(0), 0).
Hence VC ?nancing is feasible iff í ? \H
2
(¡
.c
(0), 0).
Regularity condition (1) ensures that VC ?nancing is
feasible for all \ ? |\
min
, 1] and all 0 ? |0
min
, 0
max
].
31
From a date 0 point of view, c’s expected payoff
from VC ?nancing is the total surplus generated by
the venture under that type of ?nancing:
U
.c
(0) =
¯
H
1
+V
.c
(0) ?í , with
V
.c
(0) = H
2
(¡
.c
(0), 0) ?||¡
.c
(0)]
2
,2,
(4)
where V
.c
(0) captures the venture’s continuation sur-
plus in the second period.
As mentioned above, anticipating she will have
to share rents with the venture capitalist mutes
c’s incentives, leading to second-best entrepreneurial
effort ¡
.c
(0) - ¡
FB
(0) and, in turn, to second-best
continuation surplus V
.c
(0) - V
FB
(0). On the posi-
tive side, however, the venture capitalist’s bargaining
power and rent extraction ability serve as commit-
ment devices for c to return some of the rents to ] ,
which in turn enables c to ?nance the venture with-
out resorting to debt and facing the associated liqui-
dation inef?ciencies discussed above. More formally,
the inef?ciency associated with VC ?nancing can be
expressed as U
FB
(0) ?U
.c
(0) =V
FB
(0) ?V
.c
(0).
We summarize the foregoing results in the follow-
ing proposition.
Proposition 2. Under VC ?nancing, short-term ven-
ture failure, resulting from a negative ?rst-period shock,
has no lasting impact on the venture, which continues to
operate in the second period regardless. On the other hand,
VC ?nancing generates second-best entrepreneurial effort
¡
.c
(0) -¡
FB
(0) and continuation surplus V
.c
(0) -V
FB
(0).
Importantly, although in our model an equity con-
tract per se would not yield any payoff to the investor
(since by assumption cash ?ows are not veri?able),
under VC ?nancing, even when the entrepreneur
retains overall control of the venture, the venture cap-
italist is still able to extract a stream of payoffs—a
fraction \ of pro?t—that is similar to the one typically
obtained in a standard voting equity contract. Indeed
the contract offered to the venture capitalist is in effect
an “equity-type” contract.
In practice, VC contracts include both control rights
and cash ?ow rights (Kaplan and Strömberg 2003).
Cash ?ow rights do have a role in practice because
pro?ts are veri?able to some extent. They are not per-
fectly veri?able, however, in that entrepreneurs have
the ability to divert some, if not all, of the cash ?ows.
Indeed, that is an important reason why control rights
are also used in VC contracts: they complement cash
31
This follows directly from the logic used in Footnote 25.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
10 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
?ow rights, controlling opportunistic behavior by the
entrepreneur. In this model, we assume that pro?ts
are not veri?able at all.
32
In doing so, we focus on con-
trol rights and on end-of-period bargaining, rather
than on the cash ?ow rights, and this enables us to
introduce an equity-like contract in the model (sim-
ilar to equity contracts observed in practice), all the
while keeping our assumption of nonveri?ability of
cash ?ows necessary for debt to emerge as optimal
under bank ?nancing. The modeling of equity ?nanc-
ing under the assumption of non-veri?ability of cash
?ows is neither novel nor speci?c to this paper. See,
for example, Fluck (1998), Myers (2000), and Dybvig
and Wang (2003).
4.3. Financing Trade-off
Comparing c’s expected payoffs, it is easy to see that
she chooses bank ?nancing over VC ?nancing iff
U
|
(0) ?U
.c
(0) ?0,
iff |V
|
(0) ?V
.c
(0)] ?í
1
|V
|
(0) ?|] ?0. (5)
Expression (5) yields two key results. First, it illus-
trates the main trade-off between bank ?nancing and
VC ?nancing.
Proposition 3. The entrepreneur’s ?nancing trade-off
is the following: on the upside, bank ?nancing leads to a
larger continuation surplus than VC ?nancing, V
|
(0) >
V
.c
(0). On the downside, with probability í
1
, a negative
shock occurs that prevents continuation under bank ?nanc-
ing but not under VC ?nancing.
The intuition follows directly from Propositions (1)
and (2): conditional on repaying the debt, bank ?nanc-
ing enables c to retain full access to the venture’s
rents, giving her ?rst-best effort incentives and yield-
ing a surplus larger than what could be obtained
under VC ?nancing, where rent sharing mutes entre-
preneurial incentives. On the other hand, in the event
of a negative short-term shock, lack of cash ?ow pre-
vents the repayment of bank debt, leading to inef?-
cient liquidation—an inef?ciency that does not arise
under VC ?nancing.
33
32
In the online supplement to this paper, we examine an extension
to the main model where venture capitalists can verify cash ?ows
with some probability. We show that the main results of the model
continue to hold in that case.
33
Note that if the negative ?rst-period shock were not temporary
and instead led to zero pro?ts in the second period as well as in the
?rst period, the trade-off between bank ?nancing and VC ?nanc-
ing would become exceedingly simple and much less interesting.
Following a negative shock, the prospects of future pro?ts would
disappear and the venture would be liquidated regardless of the
?nancing choice. In that case, bank ?nancing has no inef?cient liq-
uidation disadvantage relative to VC ?nancing but retains its incen-
tive advantage and thus unambiguously dominates VC ?nancing.
The second key result emerging from expression (5)
is that there exists a threshold level of venture risk í
1
,
denoted í
?
1
(0) =|V
|
(0) ?V
.c
(0)],|V
|
(0) ?|], such that
U
|
(0) ?U
.c
(0) ?0 iff í
1
?í
?
1
(0). (6)
We state this result in the following proposition.
Proposition 4. There exists a threshold level of ven-
ture risk í
?
1
(0) =|V
|
(0) ?V
.c
(0)],|V
|
(0) ?|], such that
the entrepreneur’s optimal ?nancing choice is bank ?nanc-
ing if í
1
?í
?
1
(0), and VC ?nancing otherwise.
Intuitively, for relatively risky ventures (i.e., ven-
tures with a relatively high probability of failure
(í
1
>í
?
1
(0)) but a relatively high pro?t H
1
(í
1
) in case
of success), bank ?nancing is not optimal, because
the advantage of this type of ?nancing in terms of
continuation incentives is more than offset by the
risk of liquidation associated with short-term failure.
VC ?nancing, which is not subject to this liquidation
risk, is preferred in that case. Conversely, for rela-
tively safe ventures (i.e., ventures with a relatively
low probability of failure (í
1
?í
?
1
(0)) but a relatively
low pro?t H
1
(í
1
) in case of success), bank ?nancing
makes sense because the downside risk of liquidation
is relatively low relative to the upside continuation
potential associated with high incentives.
Note that regularity condition (1) ensures that bank
?nancing is always feasible at the venture risk thresh-
old í
?
1
(0). In other words, it ensures that í
?
1
(0) -
í
max
(0) for all 0 ? |0
min
, 0
max
].
34
5. Product Market Competition
In his classic work on sunk costs and market struc-
ture, Sutton (1992, p. 9) argued that for a given con-
centration level “such features of the market as the
physical nature of the product (homogeneous versus
differentiated products) and the climate of competi-
tion policy (a strict or acquiescent approach to price
coordination by ?rms)” will affect toughness of compe-
tition in the industry. Many of the workhorse models
of competition used in the industrial organization lit-
erature today are consistent with that view: a decrease
in the transport cost | on a Hotelling (1929) line or a
Similarly, if the negative ?rst-period shock were to convey nega-
tive information about second-period prospects, this would have
no impact under bank ?nancing (where liquidation occurs whether
or not this information is present) but would reduce expected pay-
offs under VC ?nancing. It would therefore tilt the trade-off in favor
of bank ?nancing.
34
Strict bank ?nancing feasibility at í
?
1
(0) requires í - (1 ?
í
?
1
(0))V
|
(0) +í
?
1
(0)|, which simpli?es to í - V
.c
(0). (Equivalently,
one can verify that í
?
1
(0) -í
max
(0) iff í -V
.c
(0).) Using logic sim-
ilar to that of Footnote 25, one can readily verify that V
.c
(0) >
\
min
H
s
(0
max
). Together with condition (1), this implies í -V
.c
(0) for
all 0 ? |0
min
, 0
max
].
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 11
Salop (1979) circle, a decrease in the variance of errors
in a logit model, or an increase in the constant elastic-
ity of substitution in a Dixit–Stiglitz framework (Dixit
and Stiglitz 1977), for example, all capture an increase
in the degree of homogeneity across products. In
this paper as well we use the degree of homogene-
ity across products as our main measure of competi-
tion 0.
35
Rather than focusing on a particular demand
speci?cation, however, we highlight commonalities
across speci?cations and show that because of these
commonalities, under these various demand frame-
works, competition has qualitatively similar effects in
our model.
5.1. Status Quo Pro?t and the Value of
Cost Advantage
In this section we highlight two characteristic effects
of competition that are common to the Hotelling
line, Salop circle, Dixit–Stiglitz, logit, and Cournot-
to-Bertrand switch demand speci?cations, namely a
reduction in the status quo pro?t H
s
(0) and an increase
in the value of a cost advantage H
l
(0) ?H
s
(0). In the
next section we use these simple results to explain
why competition increases the overall appeal of bank
?nancing relative to VC ?nancing.
5.1.1. Status Quo Pro?t. Generated when the ven-
ture has failed to innovate and faces the same
marginal cost ¯ c as the entrant, status quo pro?t can
be expressed as the product of markup and demand:
H( ¯ c, ¯ c, 0) = P( ¯ c, ¯ c, 0)x( ¯ c, ¯ c), where P( ¯ c, ¯ c, 0) repre-
sents the markup and x( ¯ c, ¯ c) represents demand. The
only effect of competition here is to lower equilibrium
prices. Intuitively, as the degree of homogeneity across
products increases, consumers become more sensitive
to prices, putting downward pressure on prices and in
turn on markups: dP( ¯ c, ¯ c, 0),d0 -0. Since c’s venture
and the entrant have identical costs, equilibriumprices
are identical in both ?rms, and the two rivals share
the market equally regardless of the degree of compe-
tition: x( ¯ c, ¯ c) =1,2. Accordingly, the overall impact of
competition on the status quo pro?t is dH( ¯ c, ¯ c, 0),d0 =
(dP( ¯ c, ¯ c, 0),d0)x( ¯ c, ¯ c) -0.
5.1.2. Value of a Cost Advantage. This is the dif-
ference between pro?t obtained when the venture
successfully innovates and gains a cost advantage
over the entrant, and the status quo pro?t: H(c, ¯ c, 0)?
H( ¯ c, ¯ c, 0) = P(c, ¯ c, 0)x(c, ¯ c, 0) ? P( ¯ c, ¯ c, 0)x( ¯ c, ¯ c). The
main effect of competition here works through
demand: again, in markets with more homogeneous
products, consumers become more sensitive to prices,
35
Another common measure of competition is the number of com-
petitors. We consider this in an extension to the main model in
the online supplement to this paper. The main results of the model
continue to hold.
and the cost-leading venture (in the event of suc-
cessful innovation), which charges a lower equilib-
rium price than the higher-cost entrant, increases
its demand advantage over its rival. By increasing
demand for the venture when it gains a cost advan-
tage, competition increases the value of gaining such
a cost advantage.
As discussed above in the analysis of the status
quo pro?t, competition also puts downward pressure
on prices. However, this affects pro?t whether or not
the innovation takes place, and the net effect on the
value of innovation is, if not negligible, always domi-
nated by the ?rst-order demand effect just described.
Thus, overall competition unambiguously increases
the value of a cost advantage:
d|H(c, ¯ c, 0) ?H( ¯ c, ¯ c, 0)]
d0
=
dx(c, ¯ c, 0)
d0
P(c, ¯ c, 0)
+
dP(c, ¯ c, 0)
d0
x(c, ¯ c, 0) ?
dP( ¯ c, ¯ c, 0)
d0
x( ¯ c, ¯ c)
>0.
Indeed, we show in the appendix that these effects
of competition on the status quo pro?t and on the
value of innovation hold under all of demand speci-
?cations listed below.
Lemma 1. Under demand speci?cations such as the Ho-
telling line, Salop circle, Dixit–Stiglitz, logit, and Cournot-
to-Bertrand switch, competition unambiguously reduces
the status quo pro?t, dH
s
(0),d0 - 0, and unambiguously
increases the value of a cost advantage, d|H
l
(0) ?H
s
(0)],
d0 >0.
5.2. Effects of Competition on the
Financing Trade-off
Consider a product market where the degree of com-
petition is initially 0
0
and a venture with risk í
0
1
=
í
?
1
(0
0
). We know from Propositions 3 and 4 that, at
that threshold risk level, the upside of bank ?nancing
over VC ?nancing—the gain V
|
?V
.c
in continuation
surplus—is exactly offset by the downside í
?
1
|V
|
?|]
of bank ?nancing; and accordingly, the entrepreneur
is indifferent between the two types of ?nancing. In
what follows we show that competition unambigu-
ously increases the appeal of bank ?nancing over VC
?nancing: a marginal entrepreneur initially indifferent
between the two types of ?nancing will begin to strictly
prefer bank ?nancing as competition intensi?es.
Using (3) and (4), we can express the continu-
ation surplus under ?nancing of type ¡, ¡ = |, .c,
as V
¡
(0) = H
s
(0) + ¡
¡
(0)|H
l
(0) ? H
s
(0)] ? ||¡
¡
(0)]
2
,2.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
12 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
Differentiating V
|
(0) and V
.c
(0), and using the enve-
lope theorem, we obtain
dV
|
(0)
d0
=
dH
s
(0)
d0
+¡
|
(0)
d|H
l
(0) ?H
s
(0)]
d0
,
dV
.c
(0)
d0
=
dH
s
(0)
d0
+¡
.c
(0)
d|H
l
(0) ?H
s
(0)]
d0
+\
d¡
.c
d0
|H
l
(0) ?H
s
(0)].
(7)
Competition affects continuation surpluses in the
two ways described above. It reduces the status
quo pro?t H
s
(0), which has a negative impact on
V
¡
(0), and it increases the value of a cost advan-
tage H
l
(0) ?H
s
(0), which has a positive, direct impact
on V
¡
(0). This latter effect also increases entrepre-
neurial effort ¡
¡
, Which, as shown in §4, is an increas-
ing function of H
l
(0) ?H
s
(0); under VC ?nancing this
effort increase has a positive effect on the continua-
tion surplus (under bank ?nancing, this effect can be
shown to be null by the envelope theorem).
36
5.2.1. Competition and the Upside of Bank Fi-
nancing. Subtracting dV
.c
(0),d0 from dV
|
(0),d0, one
can readily establish that the upside V
|
(0) ?V
.c
(0) of
bank ?nancing over VC ?nancing is strictly increasing
in competition:
d|V
|
(0) ?V
.c
(0)]
d0
= |¡
|
(0) ?¡
.c
(0)]
d|H
l
(0) ?H
s
(0)]
d0
?
d¡
.c
d0
\|H
l
(0) ?H
s
(0)] >0. (8)
The main intuition for this result comes from the
?rst factor in (8): competition increases the value of a
cost advantage H
l
(0) ?H
s
(0), and this increase has a
larger effect on the continuation surplus under bank
?nancing because it occurs with a higher probabil-
ity (¡
|
(0) > ¡
.c
(0)). Another way to explain this is
to point out that competition increases the marginal
product of effort H
l
(0) ?H
s
(0), and that the impact
on the continuation surplus is larger under bank
?nancing, where superior entrepreneurial incentives
lead to greater equilibrium effort. The second factor,
increased entrepreneurial effort under VC ?nancing,
works the other way but can readily be shown to be
of second-order importance relative to the ?rst fac-
tor for all \ ? |\
min
, 1].
37
Indeed the upside of bank
36
As discussed above, competition increases the value of gaining a
cost advantage, and this in turn increases the entrepreneur’s incen-
tive to lower cost and her equilibrium effort level. This is similar
to the positive “escape competition effect” of competition on incen-
tives highlighted previously by Aghion et al. (2005).
37
The second factor in (8) can easily be rewritten as
\(1 ?\)¡
|
(0)d|H
l
(0) ?H
s
(0)],d0,
whereas the ?rst factor can be expressed as \¡
|
(0)d|H
l
(0) ?
H
s
(0)],d0. Clearly the latter is strictly larger than the former for all
\ ? |\
min
, 1].
?nancing over VC ?nancing is strictly increasing in 0:
d|V
|
(0) ?V
.c
(0)],d0 >0.
5.2.2. Competition and the Overall Appeal of
Bank Financing Relative to VC Financing. Let us
start by using (7) to express the effects of competition
on the downside of bank ?nancing í
?
1
(0
0
)|V
|
(0) ?|]
as follows:
í
?
1
(0
0
)
dV
|
(0)
d0
=í
?
1
(0
0
)
dH
s
(0)
d0
+¡
|
(0)
d|H
l
(0)?H
s
(0)]
d0
. (9)
On one hand, the positive effect of competition on
the value of a cost advantage H
l
(0) ?H
s
(0) has a pos-
itive impact on continuation surplus V
|
(0) and on the
downside of bank ?nancing. Most importantly, how-
ever, competition reduces the status quo pro?t H
s
(0),
which has a negative effect on the continuation sur-
plus and the downside of bank ?nancing. As shown
in the appendix, under some demand speci?cations
(e.g., Hotelling or Salop duopoly) this latter effect
strictly dominates the former for all 0 ? |0
min
, 0
max
],
and competition unambiguously reduces the down-
side of bank ?nancing. At the very least, this neg-
ative latter effect signi?cantly weakens the strength
of the former effect, ensuring that the overall impact
of competition on the downside of bank ?nancing,
regardless of its sign, is always strictly dominated by
its positive effect on the upside of bank ?nancing.
Proposition 5 then follows immediately.
Proposition 5. Competition increases the appeal of
bank ?nancing relative to VC ?nancing.
At a given degree of competition 0
0
, the marginal
entrepreneur managing a venture associated with risk
í
?
1
(0
0
), who is indifferent between the two types of
?nancing, will begin to strictly prefer bank ?nancing
over VC ?nancing if competition intensi?es. Equiv-
alently, the threshold risk level í
?
1
(0) at which the
entrepreneur remains indifferent between the two
?nancing types will increase with competition, as will
the domain of risk í
1
over which the entrepreneur
chooses bank ?nancing over VC ?nancing.
This result works primarily through the effect of
competition on the upside of bank ?nancing, and the
intuition for this important result, which comes from
two key elements of the model, is worth reiterat-
ing here. First, in allowing the entrepreneur to keep
full control of the venture in continuation and full
access to ex post pro?t, bank ?nancing yields stronger
entrepreneurial incentives and higher effort in con-
tinuation than VC ?nancing. Second, under the var-
ious demand speci?cations examined in this paper,
competition increases the value of a cost advan-
tage, and the impact of this increase on the con-
tinuation surplus is greater under bank ?nancing,
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 13
where the probability of gaining such a cost advan-
tage (i.e., entrepreneurial effort) is greater. In that
way competition increases the upside of bank ?nanc-
ing (d|V
|
(0) ?V
.c
(0)],d0>0) and the overall appeal of
bank ?nancing over VC ?nancing.
Note that the conditions of Lemma 1—that com-
petition reduces the status quo pro?t (dH
s
(0),d0
- 0) and increases the value of a cost advantage
(d|H
l
(0) ?H
s
(0)],d0 > 0)—are suf?cient to generate
Proposition 5.
38
This key result is thus robust to any
demand speci?cation satisfying these two conditions.
6. Empirical Implications
Our model yields several empirical predictions, which
we describe and discuss below.
Prediction 1. Absent negative shocks in the short
term, bank ?nancing provides stronger entrepreneurial
incentives and superior effort and performance. In the
case of negative short-term shock, it is VC ?nancing that
provides stronger entrepreneurial incentives and superior
effort and performance.
This prediction is a corollary to Propositions 1 and 2
and highlights a simple trade-off between entrepre-
neurial incentives and inef?cient liquidation affecting
the choice between bank ?nancing and VC ?nancing.
It has three implications. First, it rests on the ideas
that stronger incentives lead to superior entrepre-
neurial effort and that in turn higher entrepreneurial
effort increases venture performance. This is indeed
consistent with the recent work of Bitler et al. (2005),
who use unique data on entrepreneurial effort and
wealth in privately held ?rms to examine the connec-
tion between entrepreneurial incentives, effort, and
venture performance. They ?nd that entrepreneurial
effort, measured by hours worked, responds posi-
tively to incentives and that it has a positive impact
on venture performance.
The second implication is that bank ?nancing yields
stronger entrepreneurial incentives following short-
term success. In our opinion this implication makes
good sense: conditional on short-term success, under
bank ?nancing the entrepreneur can repay her debt
and retain 100% of all subsequent pro?t and hence has
“?rst-best” incentives, certainly stronger incentives
than under VC ?nancing where she must inevitably
relinquish a signi?cant fraction of pro?t to the venture
capitalist.
39
38
Moreover, these conditions are not necessary for our results to
hold. We show in an extension to the main model in the online
supplement to this paper that Proposition 5 holds even under
demand speci?cations where the conditions of Lemma 1 are not
satis?ed.
39
In practice, even if the entrepreneur has not yet repaid the debt
entirely, as long as short-term success ensures a small probability
of default in the future, she essentially retains ?rst-best incentives.
The third implication is that VC ?nancing gener-
ates stronger incentives following short-term failure,
because the entrepreneur retains access to at least
some fraction of pro?t, unlike under bank ?nancing
where following default the entrepreneur loses con-
trol of the venture. It might be surprising at ?rst
sight that the venture capitalist does not liquidate or
exit the venture following short-term failure, although
evidence suggests that venture capitalists are aggres-
sive in that regard (e.g., Kaplan and Strömberg 2003).
Our goal here is to capture what we view as a com-
pelling distinction between debt contracts and equity-
type contracts,
40
namely that even a ?rm with good
long-term prospects is more likely to fail under bank
?nancing, where for instance temporary cash ?ow
shortages might be enough to trigger default and
liquidation, than under VC ?nancing, where such
inef?cient liquidations would not occur.
41
This is con-
sistent with the recent empirical work of Puri and
Zarutskie (2012), which suggests that VC-?nanced
ventures are less likely to shut down, and have
lower pro?t when they do shut down, than similar
non-VC-backed ventures.
Prediction 2. Relatively risky ventures should be ?-
nanced with venture capital, and relatively safe ventures
should be ?nanced by banks.
This prediction is a corollary to Proposition 4 and
can be explained easily: ventures that are more sen-
sitive to exogenous shocks are more likely to have to
default on bank debt and face liquidation, even when
cash ?ow problems are temporary. This raises the
expected cost of bank ?nancing relative to VC ?nanc-
ing, which is immune to such inef?cient liquidation.
Thus, the model yields a prediction that is consistent
with, and an intuitive explanation for, the ?rst key
empirical result discussed in §2, namely that riskier
ventures are more likely to be VC-?nanced than bank-
?nanced.
Note that in the main model we have assumed
that there are no negative shocks in the second
40
Recall that in our model short-term failure says nothing about
the intrinsic quality of the venture; indeed both the banker and
the venture capitalist know that loss of entrepreneurial control
in the second period is inef?cient. Although VC ?nancing’s equity-
type contract avoids the problem, bank ?nancing’s debt con-
tract unavoidably leads to loss of entrepreneurial control and to
inef?ciency.
41
One could add a third state of the world at the end of period 1
where the venture would be revealed to be a total failure with no
future prospects. In that case liquidation/exit would occur under
both bank ?nancing and VC ?nancing. Even in this richer environ-
ment, our prediction would still hold: incentives following failure
would remain (weakly) superior under bank ?nancing. Thus, we
feel that the complications associated with such a change would
not be offset by additional insights and choose to focus on a model
with only two states.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
14 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
period: í
2
=0. Suppose we now assume that negative
shocks í
1
and í
2
can occur in periods 1 and 2, respec-
tively: in period 1 a pro?t H
1
is generated with proba-
bility (1?í
1
), and in period 2 the continuation payoff
V
i
, i = |, .c, is generated with probability (1 ? í
2
).
Moreover, suppose that these shocks are negatively
correlated: let us assume 1 ?í
2
=í
1
. Two points are
worth making here. First, the trade-off between the
two sources of ?nancing (and the effects of compe-
tition on that trade-off), remains exactly the same as
in the main model: bank ?nancing is preferred over
VC ?nancing iff |V
|
(0) ? V
.c
(0)] ? í
1
|V
|
(0) ? |] ? 0.
Second, we now have a new interpretation of í
1
, with
lower values of í
1
capturing shorter cash ?ow hori-
zons and larger values of í
1
capturing longer hori-
zons. Prediction 2 can thus be reinterpreted within
this context: bank ?nancing works well for shorter
cash ?ow horizons, and VC ?nancing works better for
longer horizons.
Prediction 3. Product market competition should in-
crease the probability that entrepreneurs choose bank
?nancing over VC ?nancing.
This prediction is a corollary to Proposition 5.
Important to this result is the idea that bank ?nanc-
ing yields stronger entrepreneurial incentives than
VC ?nancing. Then product market competition, by
increasing the value of stronger entrepreneurial incen-
tives, increases the bene?t of bank ?nancing rela-
tive to VC ?nancing. With this prediction, the model
provides an intuitive explanation for the second key
empirical result discussed in §2, namely the apparent
positive correlation between product market competi-
tion and the preponderance of bank ?nancing relative
to VC ?nancing.
Note that in our model, lack of competition is
closely associated with highly differentiated products
and a relatively low value of cost reduction; in con-
trast, intense competition is de?ned by highly sub-
stitutable products and a relatively high value of
cost reduction. Another corollary follows immediately
from Proposition 5.
Prediction 4. Firms that focus on product innovation
(i.e., they develop new, highly differentiated products) may
prefer VC ?nancing, and ?rms that focus on process inno-
vation (i.e., they produce less novel products albeit at a
lower cost) may prefer bank ?nancing.
To the extent that ?rms tend to rely more on VC
?nancing early in the life cycle of their product and
on bank ?nancing later in the life cycle (see, e.g.,
Berger and Udell 1998), this prediction is consistent
with Klepper (1996), who ?nds that ?rms devote
an increasing share of their R&D to process innova-
tion rather than product innovation over the product
life cycle.
A related implication concerns the distinction be-
tween radical innovations, which create new products
or completely change technologies, and incremental
innovations, which are minor modi?cations of exist-
ing products or processes. Radical innovations might
give a ?rm a ?rst-mover advantage or a domi-
nant position with little competition (Lieberman and
Montgomery 1988), as opposed to incremental inno-
vations that still compete with other technologies.
In a study of the ?nancing and R&D decisions of UK
public ?rms between 1990 and 2002, Casson et al.
(2008) ?nd that radical innovation is usually associ-
ated with equity ?nancing, whereas incremental inno-
vation is associated with debt ?nancing. This ?nding
is consistent with Atanassov et al. (2007) who examine
the interaction between capital structure and patent-
ing activity and ?nd a positive correlation between
equity ?nancing and more novel (drastic) patents.
42
Note that this distinction between radical and incre-
mental innovation is also related to Prediction 2,
since radical innovation has a higher level of risk
than incremental innovation, due to the complex-
ity associated with new product requirements and
to the added uncertainty regarding technology, mar-
ket needs, and the actions of competitors (Song and
Montoya-Weiss 1998).
Finally, noting the one-to-one correspondence be-
tween bank ?nancing and debt-type contracts on one
hand, and VC ?nancing and equity-type contracts on
the other, our model suggests that competition could
have an impact not only on the choice between bank
?nancing and VC ?nancing in entrepreneurial ven-
tures, but also on the ?nancial structure of more estab-
lished corporations.
Prediction 5. Product market competition should in-
crease the probability that managers choose debt ?nancing
over equity ?nancing.
This prediction is consistent with ?ndings in empir-
ical corporate ?nance literature. Titman and Wessels
(1988), for example, ?nd that debt levels are nega-
tively related to product uniqueness, which is a good
measure of the degree of horizontal differentiation, or
lack of competition 0 as used in our model. Using
U.S. newspaper industry data, Schargrodsky (2002)
?nds that oligopolies have higher debt ratios than
monopolies. Lord and McIntyre (2003) provide evi-
dence that leverage increases with import competition
in the textile and apparel industry. Similarly, Baggs
42
See also Titman and Wessels (1988), who ?nd a positive correla-
tion between equity ?nancing and the level of R&D expenditures;
Kortum and Lerner (2000), who show that ?rms that receive VC
?nancing innovate more; Hellmann and Puri (2000), who ?nd that
?rms pursuing an innovator rather than an imitator strategy are
more likely to obtain VC ?nancing and that obtaining VC is asso-
ciated with the ability to secure a ?rst-mover advantage.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 15
and Brander (2006) use Canada—United States Free
Trade Agreement data to show that increased compe-
tition, measured as an exogenous reduction in import
tariffs, increases leverage. More recently, Beiner et al.
(2011) ?nd signi?cantly higher leverage ratios for
?rms operating in more competitive environments.
43
To the extent that competition can be adequately
captured by market concentration, Prediction 5 sug-
gests that managers should be less likely to choose
debt over equity ?nancing in more concentrated mar-
kets. Since market concentration tends to be asso-
ciated with higher pro?tability (Azzam 1997, Slade
2004), this is indeed consistent with the negative
correlation between leverage and pro?tability docu-
mented in much previous empirical work (Titman
and Wessels 1988, Rajan and Zingales 1995, Fama and
French 2002).
In this paper we offer novel insights into the work-
ings of the connections among product market com-
petition, venture risk, and entrepreneurial ?nance,
and we propose several new, empirically testable pre-
dictions. We hope that our model will instigate further
empirical interest and can serve as a theoretical basis
for new empirical work, in what we view as a fruitful
avenue for future research.
7. Conclusion
Empirical evidence suggests possible connections
among venture risk, product market competition, and
the ?nancing of start-ups. Because of a lack of the-
oretical work on the subject, however, the microe-
conomic foundations necessary for more systematic
empirical analysis of these possible interactions were
missing. In this paper we ?ll this gap. We propose
a general theory of entrepreneurial ?nance in imper-
fectly competitive environments that (i) can be used
to explain these connections, (ii) suggests that venture
risk and product market characteristics may be impor-
tant factors to take into account in empirical work on
entrepreneurial ?nance, and (iii) yields new empirical
predictions that we hope will spur further empirical
work on the subject.
Our model highlights a simple trade-off between
the two types of ?nancing whereby bank ?nancing
yields stronger incentives and greater entrepreneurial
effort than VC ?nancing but is more sensitive to neg-
ative shocks affecting the venture, potentially lead-
ing to inef?cient liquidation. Two main results, both
consistent with available empirical evidence, emerge
from this simple trade-off. First, riskier ventures,
which are more sensitive to negative shocks, ought
43
See also Fosu (2013), who shows that the bene?ts of leverage are
improved by product market competition, measured by the Boone
indicator (using panel data consisting of 257 South African ?rms
over the period from 1998 to 2009).
to be VC-?nanced, whereas for safer ventures bank
?nancing is optimal. Second, product market compe-
tition, by increasing the marginal product of entrepre-
neurial effort, raises the advantage of superior effort
associated with bank ?nancing, thus increasing the
appeal of bank ?nancing relative to VC ?nancing.
In order to provide clear insights and tractable
results, we propose a highly stylized model of ?nanc-
ing and competition. In the online supplement to this
paper (available on the authors’ websites), we address
robustness concerns by proposing a number of exten-
sions to the main model, including allowing for com-
petition in the ?rst period, for effort by the venture
capitalist, for veri?able pro?ts, for staged ?nancing,
for mixed strategies in contracting, for entrants with
either high or low marginal cost, and for product mar-
ket competition between two entrepreneur–?nancier
pairs. We ?nd that the main results of our model
continue to hold in a variety of environments. Yet
there is still more interesting work to be done on the
interactions among competition, venture risk, and the
?nancing of new ventures. For example, it would be
interesting to consider the bank/VC ?nancing trade-
off in an in?nite-horizon setting rather than in the
simpler two-period setting we chose here; that is an
extension beyond the scope of this paper but one we
look forward to tackling in future work.
Acknowledgments
The authors thank Ingela Alger, Patrick Bolton, Gilles
Chemla, Jean Dermine, Landis Gabel, Bob Gibbons, Denis
Gromb, Thomas Hellmann, Martin Holmén, Rich Mathews,
Joel Peress, Antoine Renucci, Joel Shapiro, Lars Stole,
Tim Van Zandt, and Jano Zabojnik, as well as semi-
nar participants at Carleton University, Copenhagen Busi-
ness School, European University Institute, Hitotsubashi
University, INSEAD, Queen’s University, Université Paris-
Dauphine, University of Melbourne, University of Ottawa,
University of Tokyo, the International Industrial Organi-
zation Conference 2009 (Boston), the European Financial
Management Association Meetings 2009 (Milan), and the
Conference on Entrepreneurship and Finance 2013 (Lund,
Sweden) for helpful comments. Jing Liang and Huafang Liu
provided excellent research assistance. Jean-Etienne de Bet-
tignies gratefully acknowledges ?nancial support from the
Social Sciences and Humanities Research Council [Grant
435-2013-1863].
Appendix
A.1. Proofs of Propositions 1, 2, 3, and 4
These proofs follow directly from the text.
A.2. Proof of Lemma 1
A.2.1. Example 1: Hotelling Model. The two start-ups
are located at each end of a Hotelling (1929) line and com-
pete in both periods. Venture 1 is located at x =0 and
venture 2 is at x =1. There is a unique consumer who is
uniformly distributed along the Hotelling line. Located at ç,
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
16 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
the consumer incurs a transport cost |ç for traveling to
?rm 1, and a cost |(1 ?ç) for visiting ?rm 2. In each period,
the consumer enjoys conditional indirect utility V
1
= j ?
p
1
?|ç from product 1 and V
2
=j ?p
2
?|(1 ?ç), where j
represents income, and chooses the product that gives the
highest utility. In this paper we measure the degree of com-
petition as the degree of substitutability between the two
competing products. Conveniently, on a Hotelling line, the
transport cost | captures the degree of product differentia-
tion, so we use the degree of substitutability between prod-
ucts, 0 =1,|, as a measure of the toughness of competition,
to use Sutton’s (1992, p. 9) terminology.
The equilibrium pro?t of ?rm i is H
i
(c
i
,c
¡
,0) =(1,(20))·
(1+(0(c
¡
?c
i
)),3)
2
. The pro?ts for the different outcomes
are H
s
(0) =1,(20) and H
l
(0) =1,(20)(1+0( ¯ c?c),3)
2
. This
yields H
l
(0)?H
s
(0) =(( ¯ c?c),6)(2+0( ¯ c?c),3). It is straight-
forward that dH
s
(0),d0=?1,(20
2
) -0, and d(H
l
(0)?
H
s
(0)),d0=( ¯ c?c)
2
,18>0.
A.2.2. Example 2: Salop Model. The market is repre-
sented by a unit circumference. There is a mass 1 of consu-
mers who are uniformly distributed along the circumference
of the circle on which n ?rms are located symmetrically.
Firm 1 produces at constant marginal cost c
i
. Locations in
the market are indexed by z ? |0, 1). Each consumer buys
either one unit of output or no output at all. A consumer
located at point d buying a unit of output from ?rm i
derives utility u(d, i) =.?p
i
?|D(d, i), where . is the com-
mon valuation of output, p
i
is the price that ?rm i charges,
D(z, i) is the shortest arc length separating consumer d
from ?rm i, and | ?0 is the transport cost per unit of dis-
tance. The demand for ?rm i is given by D
i
(p
i
, p
i?1
, p
i+1
) =
1,n?p
i
,| +(p
i?1
+p
i+1
),(2|). Pro?ts are maximized for p
i
=
|,(2n) + c
i
,2 + (p
i?1
+ p
i+1
),4, which yields to the unique
Bertrand equilibrium price p
?
i
(c
i
, c
¡
) = |,n +
n?1
|=0
|
|
c
i?|
,
where |
|
= |(2 +
?
3)
|
+(2 +
?
3)
n?|
],|
?
3((2 +
?
3)
n
?1)].
44
When the venture’s n ? 1 competitors have a cost ¯ c (c
¡
=
¯ c ?¡ = 1), the equilibrium price is p
?
i
(c
i
) = |,n + c
i
+ ( ¯ c ?
c
i
)(1 ? |
0
), where |
0
= |1 + (2 +
?
3)
n
],|
?
3((2 +
?
3)
n
? 1)].
Therefore the pro?ts for the different outcomes are H
s
(0) =
1,(0n
2
) and H
l
(0) = (1,0)(1,n + 0( ¯ c ? c)(1 ? |
0
))
2
, where
0 =1,| measures the degree of competition in the market.
This yields H
l
(0) ?H
s
(0) =( ¯ c ? c)(1 ?|
0
)(2,n+0( ¯ c ? c)(1 ?
|
0
)). It is straightforward that dH
s
(0),d0 = ?1,(0
2
n
2
) - 0,
and d(H
l
(0) ?H
s
(0)),d0 =( ¯ c ? c)
2
(1 ?|
0
)
2
>0.
A.2.3. Example 3: Constant Elasticity of Substition
(CES) (Dixit–Stiglitz) Model. Assume that the two start-
ups face demand of the form p
i
(q
i
, q
¡
) = (1,q
1?0
i
),(q
0
i
+q
0
¡
)
with 0 - 0 - 1. This demand function is derived from a
CES utility function u(q
1
, q
2
) = (q
0
1
+ q
0
2
)
1,0
, where 0 mea-
sures the degree of substitutability between the goods of
?rms 1 and 2. Firm i has a marginal cost c
i
. In this
example, competition increases as 0 increases so goods
become closer substitutes. The pro?t of ?rm i can be writ-
ten as H
i
(c
i
, c
¡
, 0) =(1 +(1 ?0)(c
i
,c
¡
)
0
),((1 +(c
i
,c
¡
)
0
)
2
). The
pro?ts for the different outcomes are H
s
(0) = (2 ?0),4,
and H
l
(0) = (1 + (1 ? 0)c
0
),(1 +c
0
)
2
, where c = c,¯ c -1.
It is straightforward that dH
s
(0),d0 = ?1,4 -0. More-
over, d(H
l
(0) ?H
s
(0)),d0 = (1,(4(1 +c
0
)
3
))(?4c
0
(lnc)(1 +
0 +(1 ?0)c
0
) +(1 +c
0
)(1 ?c
0
)
2
) >0.
44
See Lin and Wu (2013) for the proof.
A.2.4. Example 4: Logit Model. Consider a market
with N consumers and two ?rms. Each consumer in the
market purchases one unit of just one of the two products
that the ?rms offer. A consumer chosen at random from the
population of N consumers has a conditional indirect util-
ity function V
i
=A ?p
i
+a
i
for i =1, 2. The parameter A
stands for consumers’ income, p
i
is the price of good i =
1, 2, and a
i
represents consumers’ idiosyncratic tastes about
good i (horizontal differentiation). The purchase probabil-
ity for product i is the probability that a randomly selected
consumer derives the largest utility from purchasing it. In
the absence of an outside (no-purchase) option, the expected
demand of ?rm i is given by the following logit formula
(see Anderson et al. 1992) d
i
(p) =exp(?p
i
,j),(exp(?p
1
,j)+
exp(?p
2
,j)). If j = 0, goods are perfect substitutes, but if
j = , goods are independent. So competition increases
as j decreases. Therefore we use 0 = 1,j as a mea-
sure of the toughness of competition. Firms compete in
price, which leads to equilibrium prices such that p
i
=
c
i
+1,0 +(1,0)(exp(?0p
i
),exp(?0p
¡
)). Pro?ts for the differ-
ent outcomes are the following: H
s
(0) = N,0 and H
l
(0) =
(N,0)(exp(?0p
?
1
),exp(?0p
?
2
)), where p
?
1
and p
?
2
are the equi-
librium prices when c
1
= c and c
2
= ¯ c.
45
It follows directly
that dH
s
(0),d0 -0.
In what follows we show that d(H
l
(0) ? H
s
(0)),d0 >0.
Let X = exp(?0p
?
1
),exp(?0p
?
2
). Since ¯ c > c, p
?
1
- p
?
2
so
X >1. Moreover, lnX =0(p
?
2
?p
?
1
) =0( ¯ c ? c) +1,X ?X. By
differentiating this equality with respect to 0, we obtain
(dX,d0)(X
2
+X +1),X
2
= ¯ c ? c. We have H
l
(0) ?H
s
(0) =
N((X ?1),0), so (1,N)d(H
l
(0) ?H
s
(0)),d0 =?(X ?1),0
2
+
(1,0)(dX,d0) = ?(X ?1),0
2
+X
2
(lnX +X ?1,X),(0
2
(X
2
+
X +1)) = (X
2
lnX ? X +1),(0
2
(X
2
+ X +1)) > 0 ? X > 1.
Therefore, d(H
l
(0) ?H
s
(0)),d0 >0.
A.2.5. Example 5: Switch from Cournot to Bertrand.
Let us analyze a shift from a Cournot duopoly to a Bertrand
duopoly, where Bertrand competition is often seen as more
competitive than Cournot competition. Assume that the two
ventures face a demand of the form P =u ?Q.
Under a Cournot duopoly, the equilibrium pro?t of ven-
ture i is H
i
(c
i
, c
¡
) =(u?2c
i
+c
¡
)
2
,9. The pro?ts for the differ-
ent outcomes are H
s
C
=(u ? ¯ c)
2
,9 and H
l
C
=(u ?2c + ¯ c)
2
,9.
Under a Bertrand duopoly, the equilibrium pro?t of ven-
ture i is H
i
(c
i
, c
¡
) =0 if p
i
>p
¡
and H
i
(c
i
, c
¡
) =(p
i
?c
i
)(u?p
i
)
if p
i
-p
¡
. The pro?ts for the different outcomes are H
s
8
=0
and H
l
8
=( ¯ c ? c)(u ? ¯ c).
As competition increases, we switch from Cournot com-
petition to Bertrand competition. Since H
s
C
> H
s
8
, H
s
de-
creases. Moreover, H
l
C
? H
s
C
= (4,9)( ¯ c ? c)(u ? ¯ c) whereas
H
l
8
?H
s
8
=( ¯ c ? c)(u ? ¯ c), so H
l
?H
s
increases.
A.3. Proof of Proposition 5
Let us de?ne A ? H
l
(0) ?H
s
(0), A
? d(H
l
(0) ?H
s
(0)),d0,
and H
s
?dH
s
(0),d0. We know from the main text that ¡
|
=
A,| and ¡
.c
=(1 ?\)A,|, which immediately implies ¡
|
?
¡
.c
=\A,| and d¡
.c
,d0 =(1 ?\)A
,|.
45
More speci?cally, these equilibrium prices are p
?
1
= c + 1,0 +
(1,0)(exp(?0p
?
1
),exp(?0p
?
2
)) and p
?
2
= ¯ c + 1,0 + (1,0)(exp(?0p
?
2
),
exp(?0p
?
1
)).
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 17
Using expression (8), and substituting the relevant ex-
pressions above, we can express the impact of competition
on the upside of bank ?nancing as follows:
d|V
|
(0) ?V
.c
(0)]
d0
= (¡
|
?¡
.c
)A
?\
d¡
.c
d0
A
=
\
2
|
AA
>0. (10)
For a given degree of competition 0, the associated
threshold venture risk level í
?
1
(0) can be expressed as
í
?
1
(0) = 1 ?
V
.c
(0) ?|
V
|
(0) ?|
=1 ?
H
s
+¡
.c
A?(|,2)¡
2
.c
?|
H
s
+¡
|
A?(|,2)¡
2
|
?|
=
(\
2
,(2|))A
2
H
s
+(1,(2|))A
2
?|
. (11)
Since we know from (7) that dV
|
,d0 =H
s
+¡
|
A
=H
s
+
AA
,|, we can express the downside of bank ?nancing as
dV
|
d0
í
?
1
(0) =
H
s
+
AA
|
(\
2
,(2|))A
2
H
s
+(1,(2|))A
2
?|
. (12)
Subtracting (12) from (10) and simplifying yields
d|V
|
(0) ?V
.c
(0)]
d0
?
dV
|
d0
í
?
1
(0) =¹|2(H
s
?|)A
?AH
s
],
with
¹=
(\
2
,(2|))A
2
H
s
+(1,(2|))A
2
?|
>0. (13)
Clearly, the results of Lemma 1, namely A
> 0 and
H
s
-0, are suf?cient to ensure a strictly positive impact
of competition on the appeal of bank ?nancing relative to
VC ?nancing.
A.4. Relaxing the Regularity Condition (1)
Throughout the paper we have imposed a regularity con-
dition (1) on the model: í ? \
min
H
s
(0
max
). As discussed
in §4, this condition is suf?cient to ensure that for all 0 ?
|0
min
, 0
max
], (a) VC ?nancing is feasible for all \ ? |\
min
, 1];
(b) there exists an í
max
(0) = 1 ? (í ? |),(V
|
(0) ? |) > 0
such that bank ?nancing is feasible iff í
1
? |0, í
max
(0)]; and
(c) there exists a threshold level of venture risk í
?
1
(0) =1 ?
(V
.c
(0) ?|),(V
|
(0) ?|) - í
max
(0) such that bank ?nancing
is optimal for all í
1
? í
?
1
(0), and VC ?nancing is optimal
otherwise. We now relax this condition.
Recall that VC ?nancing is feasible iff í ?\H
2
(¡
.c
(0), 0),
and that bank ?nancing is feasible iff í ?(1?í
1
)V
|
(0)+í
1
|.
Note that for any \ ? |\
min
, 1] and 0 ? |0
min
, 0
max
], we have
\H
2
(¡
.c
(0), 0) -V
.c
(0) -V
|
(0). (14)
Thus, for given values of í , \, and 0, there are three cases
to examine.
Case 1. í >V
|
(0).
Bank ?nancing is not feasible: there exists no venture risk
í
1
? 0 such that í ? (1 ? í
1
)V
|
(0) + í
1
|. VC ?nancing is
not feasible either, since (14) implies that if í > V
|
(0) then
í >\H
2
(¡
.c
(0), 0). Hence the venture is not ?nanced in the
?rst place.
Case 2. \H
2
(¡
.c
(0), 0) -í ?V
|
(0).
Bank ?nancing is feasible: there exists a í
max
(0) = 1 ?
(í ?|),(V
|
(0) ?|) ? 0 such that bank ?nancing is feasible
iff í
1
? |0, í
max
(0)]. On the other hand, \H
2
(¡
.c
(0), 0) - í
implies that VC ?nancing is not feasible in that case. Thus,
ventures with risk í
1
?í
max
(0) are bank ?nanced, and ven-
tures with risk í
1
>í
max
(0) are not ?nanced.
Case 3. í ?\H
2
(¡
.c
(0), 0) -V
|
(0).
Bank ?nancing is feasible for all ventures with í
1
?
|0, í
max
(0)] and í
max
(0) = 1 ? (í ? |),(V
|
(0) ? |) > 0.
VC ?nancing is also feasible since í ?\H
2
(¡
.c
(0), 0). More-
over, note that since í ?\H
2
(¡
.c
(0), 0) and \H
2
(¡
.c
(0), 0) -
V
.c
(0) (see expression (14)), we must have í -V
.c
(0), which
in turn implies í
?
1
(0) =1?(V
.c
(0)?|),(V
|
(0)?|) -í
max
(0).
Thus the results in Case 3 are the same as in the main
model. When both bank ?nancing and VC ?nancing are
available and c faces a “real”
46
?nancing choice, (1) for
low-risk ventures (í
1
? í
?
1
(0)), bank ?nancing is optimal,
whereas for high risk ventures (í
1
>í
?
1
(0)) VC ?nancing is
optimal; (2) the relative appeal of bank ?nancing increases
with competition (oí
?
1
,o0 >0).
Note that bank ?nancing is strictly more feasible than
VC ?nancing in our Model, a feature consistent with the
empirical evidence discussed in §2, which suggests that
banks are a much more frequent capital provider than ven-
ture capitalists.
This feature comes in part from our assumption that ]
has full bargaining power in renegotiation. This assumption
enables c to commit to a relatively large debt repayment at
date 1, making bank ?nancing more feasible at date 0. If
we allow c to have some bargaining power in renegotiation,
then the more bargaining power she has in renegotiation,
the lower her ability to commit to debt repayments, and the
lower her ability to borrow from a bank in the ?rst place.
Alternatively, the venture capitalist may be able to extract
rents in the ?rst period as well as in the second period (in
the model we assume that the venture capitalist extracts
rents in the second period only). In both of these cases,
the feasibility of VC ?nancing relative to bank ?nancing
would increase, and an additional (fourth) case may emerge
whereby for some values of í , \, and 0, only VC ?nancing
may be available to the entrepreneur. While one cannot rule
out the possibility that this would occur at higher values
of 0, one cannot tell a priori for which values of 0 this would
be more likely to occur. In any case, the main results of the
model, which pertain to the scenario where both sources of
?nancing are available, as described in Case 3 above, con-
tinue to hold.
References
Abowd J, Lemieux T (1993) The effects of product market compe-
tition on collective bargaining agreements: The case of foreign
competition in Canada. Quart. J. Econom. 108:983–1014.
Aghion P, Bloom N, Blundell R, Grif?th R, Howitt P (2005) Com-
petition and innovation: An inverted-U relationship. Quart. J.
Econom. 120:701–728.
46
In contrast to Cases 1 and 2 where c faces either one choice or no
choice at all.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
18 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
Atanassov J, Nanda VK, Seru A (2007) Finance and innovation: The
case of publicly traded ?rms. Working Paper 970, Ross School
of Business, University of Michigan, Ann Arbor.
Azzam AM (1997) Measuring market power and cost ef?ciency
effects of industrial concentration. J. Indust. Econom. 45:
377–385.
Baggs J, Bettignies (de) J-E (2007) Product market competition and
agency costs. J. Indust. Econom. 55:289–323.
Baggs J, Brander J (2006) Trade liberalization, pro?tability, and
?nancial leverage. J. Internat. Bus. Stud. 37:196–211.
Bascha A, Walz U (2001) Convertible securities and optimal exit
decisions in venture capital ?nance. J. Corporate Finance 7:
285–306.
Beiner S, Schmid M, Wanzenried G (2011) Product market compe-
tition, managerial incentives and ?rm valuation. Eur. Financial
Management 17:331–366.
Berger A, Schaek K (2011) Small and medium-sized enterprises,
bank relationship strength, and the use of venture capital.
J. Money Credit Banking 43:461–490.
Berger A, Udell G (1998) The economics of small business ?nance:
The roles of private equity and debt markets in the ?nancial
growth cycle. J. Banking Finance 22:613–673.
Berglöf E (1994) A control theory of venture capital ?nance. J. Law,
Econom. Organ. 10:247–267.
Bertschek I (1995) Product and process innovation as a response
to increasing imports and foreign direct investment. J. Indust.
Econom. 43:341–357.
Bettignies (de) J-E (2008) Financing the entrepreneurial venture.
Management Sci. 54:151–166.
Bitler M, Moskowitz T, Vissing-Jorgensen A (2005) Testing agency
theory with entrepreneur effort and wealth. J. Finance 60:
539–576.
Bloom N, Draca M, Van Reenen J (2011) Trade induced technical
change? The impact of chinese imports on IT and productivity.
Working paper, Stanford University, Stanford, CA.
Blundell R, Grif?th R, Van Reenen J (1999) Market share, mar-
ket value, and innovation in a panel of british manufacturing
?rms. Rev. Econom. Stud. 66:529–554.
Bolton P, Scharfstein D (1990) A theory of predation based on
agency problems in ?nancial contracting. Amer. Econom. Rev.
80:94–106.
Bolton P, Scharfstein D (1996) Optimal debt structure with multiple
creditors. J. Political Econom. 104:1–25.
Boot A, Gopalan R, Thakor A (2006) The entrepreneur’s choice
between private and public ownership. J. Finance 61:803–836.
Brander J, Lewis T (1986) Oligopoly and ?nancial structure: The
limited liability effect. Amer. Econom. Rev. 80:93–106.
Burgess S, Metcalfe P (2000) Incentive pay and product market
competition. CMPO Working Paper 00/28, University of Bris-
tol, Bristol, United Kingdom.
Casson PD, Martin R, Nisar TM (2008) The ?nancing decisions of
innovative ?rms. Res. Internat. Bus. Finance 22:208–221.
Caves RE, Barton D (1990) Ef?ciency in U.S. Manufacturing Industries
(MIT Press, Cambridge, MA).
Chemmanur T, Chen Z (2014) Venture capitalists versus angels:
The dynamics of private ?rm ?nancing contracts. Rev.Corporate
Finance Stud.. 3:39–86.
Chemmanur T, He J (2011) IPO waves, product market competition,
and the going public decision: Theory and evidence. J. Financial
Econom. 101:382–412.
Cosh A, Cumming D, Hughes A (2009) Outside entrepreneurial
capital. Econom. J. 119:1494–1533.
Cuñat V, Guadalupe M (2005) How does product market com-
petition shape incentive contracts? J. Eur. Econom. Assoc. 3:
1058–1082.
Dixit AK, Stiglitz J (1977) Monopolistic competition and optimal
product diversity. Amer. Econom. Rev. 67:297–308.
Dybvig P, Wang Y (2003) Debt and equity. Working paper, Wash-
ington University in St. Louis, St. Louis.
Fama EF, French KR (2002) Testing tradeoff and pecking order pre-
dictions about dividends and debt. Rev. Financial Stud. 15:1–33.
Faure-Grimaud A (2000) Product market competition and opti-
mal debt contracts: The limited liability effect revisited. Eur.
Econom. Rev. 44:1823–1840.
Fluck S (1998) Optimal ?nancial contracting: Debt versus outside
equity. Rev. Financial Stud. 11:383–418.
Fosu S (2013) Capital structure, product market competition and
?rm performance: Evidence from South Africa. Quart. Rev.
Econom. Finance 53:140–151.
Fraser S (2005) Finance for small and medium-sized enterprises:
A report on the 2004 UK survey of SME ?nances. Working
paper, University of Warwick, Coventry, United Kingdom.
Fulghieri P, Sevilir M (2009) Size and focus of a venture capitalist’s
portfolio. Rev. Financial Stud. 22:4643–4680.
Geroski P (1990) Innovation, technological opportunity, and market
structure. Oxford Econom. Papers 42:586–602.
Gertner R, Scharfstein D, Stein J (1994) Internal versus external cap-
ital markets. Quart. J. Econom. 109:1211–1230.
Glaeser EL, Kerr SP, Kerr WR (2014) Entrepreneurship and urban
growth: An empirical assessment with historical mines. Rev.
Econom. Statist., ePub ahead of print March 21,http://dx.doi
.org/10.1162/rest_a_00456.
Glazer J (1994) The strategic effects of long-term debt in imperfect
competition. J. Econom. Theory 62:428–443.
Green A, Mayes D (1991) Technical inef?ciency in manufacturing
industries. Econom. J. 101:523–538.
Grif?th R (2001) Product market competition, ef?ciency and agency
costs: An empirical analysis. IFS Working paper, Institute for
Fiscal Studies, London.
Hart O, Moore J (1989) Default and renegotiation: A dynamic model
of debt. Working paper, Massachusetts Institute of Technology,
Cambridge. (Published in Quart. J. Econom. 1998, 113:1–41.)
Hart O, Moore J (1994) A theory of debt based on the inalienability
of human capital. Quart. J. Econom. 109:841–880.
Hart O, Moore J (1995) Debt and seniority: An analysis of hard
claims in constraining management. Amer. Econom. Rev. 85:
567–585.
Harvard Business School (2006) Financing the business: Where is
the money? Harvard Business Essentials: Entrepreneur’s Toolkit
(Harvard Business School Press, Boston), 1–43.
Hashmi A (2013) Competition and innovation: The inverted-U rela-
tionship revisited. Rev. Econom. Statist. 95:1653–1668.
Haskel J (1991) Imperfect competition, work practices and produc-
tivity growth. Oxford Bull. Econom. Statist. 53:265–279.
Hellmann T, Puri M (2000) The interaction between product market
and ?nancing strategy: The role of venture capital. Rev. Finan-
cial Stud. 13:959–984.
Hisrich R, Peters M, Shepherd D, Mombourquette P (2006)
Entrepreneurship, Canadian ed. (McGraw-Hill Ryerson, Whitby,
Canada).
Hoberg G, Phillips G (2010) Product market synergies and compe-
tition in mergers and acquisitions: A text-based analysis. Rev.
Financial Stud. 23:3773–3811.
Hotelling H (1929) Stability in competition. Econom. J. 39:41–57.
Inderst R, Mueller H (2004) The effect of capital market charac-
teristics on the value of start-up ?rms. J. Financial Econom. 74:
319–356.
Inderst R, Mueller H (2009) Early-stage ?nancing and ?rm growth
in new industries. J. Financial Econom. 93:276–291.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 19
Industry Canada (2009) Canada Small Business Financing Act com-
prehensive review report, 2004–2009. Last accessed Decem-
ber 18, 2014,http://www.publications.gc.ca/site/eng/367008/
publication.html.
Jianakoplos NA, Bernasek A (1998) Are women more risk-averse?
Econom. Inquiry 36:620–630.
Kaplan S, Strömberg P (2003) Financial contracting meets the real
world: An empirical analysis of venture capital contracts. Rev.
Econom. Stud. 70:281–316.
Klepper S (1996) Entry, exit, growth and innovation over the prod-
uct life cycle. Amer. Econom. Rev. 86(3):562–583.
Knowledge@Wharton (2011) When markets are hot: Opportunities
for venture capitalists across the globe. (June 24), http://
knowledge.wharton.upenn.edu/article/where-markets-are-hot
-opportunities-for-venture-capitalists-across-the-globe/.
Kortum S, Lerner J (2000) Assessing the contribution of venture
capital to innovation. RAND J. Econom. 31:674–692.
Landier A (2003) Start-up ?nancing: From banks to venture capital.
Working paper, University of Chicago, Chicago.
Leeth J, Scott J (1989) The incidence of secured debt: Evidence
from the small business community. J. Financial Quant. Anal. 24:
379–394.
Lieberman MB, Montgomery DB (1988) First mover advantages.
Strategic Management J. 9:41–58.
Lin M, Wu R (2013) Production cost heterogeneity in a circular-
city model. Working paper, Singapore Management University,
Singapore.
Lord RA, McIntyre JE Jr (2003) Leverage, imports, pro?tability,
exchange rates, and capital investments: A panel data study
of the textile and apparel industries 1974–1987. Internat. Rev.
Financial Anal. 12:287–310.
Maksimovic V (1988) Capital structure in repeated oligopolies.
RAND J. Econom. 19:389–407.
Myers S (2000) Outside equity. J. Finance 55:1005–1037.
Nickell SJ (1996) Competition and corporate performance. J. Politi-
cal Econom. 104:724–746.
Peress J (2010) Product market competition, insider trading and
stock market ef?ciency. J. Finance 65:1–43.
Poitevin M (1989) Collusion and the banking structure of a duopoly.
Canadian J. Econom. 22:263–277.
Povel P, Raith M (2004) Financial constraints and product market
competition: Ex ante vs. ex post incentives. Internat. J. Indust.
Organ. 22:917–949.
Puri M, Zarutskie R (2012) On the lifecycle dynamics of venture-
capital- and non-venture-capital-?nanced ?rms. J. Finance 67:
2247–2293.
Rajan RG, Zingales L (1995) What do we know about capital
structure? Some evidence from international data. J. Finance
50:1421–1460.
Renucci A (2014) Bargaining with venture capitalists: When
should entrepreneurs show their ?nancial muscle? Rev. Finance
18:2197–2214.
Repullo R, Suarez J (2004) Venture capital ?nance: A security design
approach. Rev. Finance 8:75–108.
Robb AM, Robinson DT (2014) The capital structure decisions of
new ?rms. Rev. Financial Stud. 27:153–179.
Sahlman WA (1990) The structure and governance of venture cap-
ital organization. J. Financial Econom. 27:473–521.
Salop SC (1979) Monopolistic competition with outside goods. Bell
J. Econom. 10:141–156.
Schargrodsky E (2002) The effect of product market competition
on capital structure: Empirical evidence from the newspaper
industry. Business School Working paper, Universidad Torcu-
ato Di Tella, Buenos Aires.
Showalter D (1995) Oligopoly and ?nancial structure: Comment.
Amer. Econom. Rev. 85:647–653.
Slade M (2004) Competing models of ?rm pro?tability. Internat. J.
Indust. Organ. 22:289–308.
Song XM, Montoya-Weiss MM (1998) Critical development activi-
ties for really new versus incremental products. J. Product Inno-
vation Management 15:124–135.
Sutton J (1992) Sunk Costs and Market Structure (MIT Press, Cam-
bridge, MA).
Syverson C (2004a) Product substitutability and productivity dis-
persion. Rev. Econom. Statist. 86:534–550.
Syverson C (2004b) Market structure and productivity: A concrete
example. J. Political Econom. 112:1181–1222.
Teshima K (2010) Import competition and innovation at the
plant level: Evidence from Mexico. Working paper, ITAM,
Mexico City.
Titman S, Wessels T (1988) The determinants of capital structure
choice. J. Finance 43:1–19.
Ueda M (2004) Banks versus venture capital: Project evaluation,
screening, and expropriation. J. Finance 59:601–621.
U.S. Small Business Administration, Of?ce of Advocacy (2014) Fre-
quently asked questions—Advocacy: The voice of small busi-
nesses in government.http://www.sba.gov/sites/default/
?les/sbfaq.pdf.
Winton A, Yerramilli V (2008) Entrepreneurial ?nance: Banks ver-
sus venture capital. J. Financial Econom. 88:51–79.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
doc_770096523.pdf
In this particular brief description relating to management science product market competition and the financing of new ventures.
This article was downloaded by: [130.15.189.191] On: 18 March 2015, At: 06:57
Publisher: Institute for Operations Research and the Management Sciences (INFORMS)
INFORMS is located in Maryland, USA
Management Science
Publication details, including instructions for authors and subscription information:http://pubsonline.informs.org
Product Market Competition and the Financing of New
Ventures
Jean-Etienne de Bettignies, Anne Duchêne
To cite this article:
Jean-Etienne de Bettignies, Anne Duchêne (2015) Product Market Competition and the Financing of New Ventures.
Management Science
Published online in Articles in Advance 29 Jan 2015
.http://dx.doi.org/10.1287/mnsc.2014.2041
Full terms and conditions of use:http://pubsonline.informs.org/page/terms-and-conditions
This article may be used only for the purposes of research, teaching, and/or private study. Commercial use
or systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisher
approval, unless otherwise noted. For more information, contact [email protected].
The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitness
for a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, or
inclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, or
support of claims made of that product, publication, or service.
Copyright © 2015, INFORMS
Please scroll down for article—it is on subsequent pages
INFORMS is the largest professional society in the world for professionals in the fields of operations research, management
science, and analytics.
For more information on INFORMS, its publications, membership, or meetings visithttp://www.informs.org
MANAGEMENT SCIENCE
Articles in Advance, pp. 1–19
ISSN 0025-1909 (print) ISSN 1526-5501 (online)
http://dx.doi.org/10.1287/mnsc.2014.2041
©2015 INFORMS
Product Market Competition and the
Financing of New Ventures
Jean-Etienne de Bettignies
Queen’s School of Business, Queen’s University, Kingston, Ontario K7L 3N6, Canada,
[email protected]
Anne Duchêne
IPAG Business School, 75006 Paris, France; and Lebow School of Business, Drexel University,
Philadelphia, Pennsylvania 19104, [email protected]
T
his paper examines the interaction between venture risk, product market competition, and the entrepreneur’s
choice between bank ?nancing and venture capital (VC) ?nancing. Under bank ?nancing, a debt-type con-
tract emerges as optimal, which allows the entrepreneur to retain full control of the venture and thus yields
strong effort incentives, as long as the entrepreneur can service the debt repayment; however, this leads to liqui-
dation in the case of default, making the venture’s success quite sensitive to exogenous, even temporary, shocks
that may hinder debt repayment. Under VC ?nancing, an equity-type contract emerges as optimal. Although it
is not sensitive to exogenous shocks, this contract requires the entrepreneur to share a fraction of the rents with
the ?nancier, thus yielding lower effort incentives for the entrepreneur. There exists a threshold level of venture
risk such that bank ?nancing is optimal if and only if venture risk is below that threshold. Product market
competition increases the value of stronger entrepreneurial incentives and thus increases the maximum level of
risk the entrepreneur is willing to take before switching from bank ?nancing to VC ?nancing. This is a robust
result that is shown to hold in various models of competition, including the Hotelling, Salop, Dixit–Stiglitz, and
Cournot-to-Bertrand switch.
Keywords: competition; venture risk; bank ?nancing; venture capital
History: Received June 17, 2010; accepted July 17, 2014, by Lee Fleming, entrepreneurship and innovation.
Published online in Articles in Advance.
1. Introduction
Small ?rms are of critical importance to the econ-
omy.
1
In the United States, they represent 99.9% of
all ?rms, employ half of all private sector employ-
ees, produce 16 times more patents per employee than
large patenting ?rms (U.S. Small Business Adminis-
tration 2014), and have a signi?cantly positive impact
on urban growth (Glaeser et al. 2014).
The creation and growth of these entrepreneurial
ventures often requires external capital, and ?nan-
cial intermediaries such as banks and venture cap-
italists play a key role in that regard. What factors
affect the entrepreneur’s choice between the two types
of ?nancing? In practice it is sometimes argued that
entrepreneurs prefer bank-provided debt ?nancing
over venture capital and thus would always choose
bank ?nancing as long as it is feasible, that is, as
long as the venture has enough assets to collateralize
the loan. Under that hypothesis, a key factor in the
1
The Of?ce of Advocacy of the U.S. Small Business Administration
(SBA) de?nes a small business as an independent business with
fewer than 500 employees.
entrepreneur’s choice would be some ratio of collat-
eral value to investment requirement. This is a plausi-
ble, but likely incomplete, hypothesis: it relies on the
argument that unsecured debt is very dif?cult if not
impossible to obtain in new ventures, an argument
somewhat inconsistent with the fact that 40% of loans
taken by small ?rms are unsecured (Leeth and Scott
1989). In this paper, we suggest that two key factors—
venture risk and product market competition—may
affect the ?nancing strategy of new ventures (see the
discussion of the related empirical evidence in the
next section), and we examine the interaction between
these two factors and the trade-off between bank
?nancing and venture capital (VC) ?nancing.
We propose a two-period, incomplete contracting
model in which a wealth-constrained entrepreneur
chooses between bank ?nancing and VC ?nancing,
anticipating that a negative shock may or may not
occur at date 1. In our simple setup, under bank
?nancing a debt-type contract emerges as optimal.
On one hand, conditional on repaying the debt, the
entrepreneur retains full control of the venture and, as
a result, full access to the venture’s realized pro?t. She
1
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
2 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
thus has strong incentives to exert effort and create
value.
2
On the other hand, default on the debt leads
to liquidation, making the venture’s success quite sen-
sitive to exogenous, even temporary, shocks that may
hinder debt repayment.
We argue that unlike the bank, which receives none
of the realized pro?t when the entrepreneur controls
the ?rm, the venture capitalist can use expertise, social
networks, and ?nancial resources to extract a fraction
of ex post pro?t even when the entrepreneur con-
trols the venture. Accordingly, under VC ?nancing, an
equity-type contract emerges as optimal. This contract
is not sensitive to exogenous shocks but requires the
entrepreneur to share a fraction of the pro?t with the
?nancier, yielding lower entrepreneurial effort. Thus
our ?rst key result emerges: there exists a threshold
level of venture risk (probability of a negative shock
occurring) such that below that threshold the entrepre-
neurial incentive advantage of the debt contract more
than offsets the disadvantage associated with expected
default and liquidation, and bank ?nancing is pre-
ferred to VC ?nancing. Above that risk threshold, the
elimination of default risk associated with equity-type
contracts more than offsets the incentive disadvantage,
and VC ?nancing is optimal.
In our model, the entrepreneur’s effort increases the
probability of gaining a cost advantage over her prod-
uct market rival. We show that competition, measured
by the degree of substitutability between products,
increases the value of a cost reduction—the marginal
product of entrepreneurial effort. The intuition is sim-
ple: in markets with more homogeneous products,
consumers become more sensitive to prices, and this
increases the demand advantage of the cost-leading
entrepreneur charging a lower price than her rival.
The value of gaining a cost advantage thus increases
with competition, and this disproportionately favors
bank ?nancing, where (as discussed) entrepreneurial
effort and the resulting likelihood of a cost reduction
are greater in equilibrium.
Thus, competition increases the appeal of bank
?nancing relative to VC ?nancing—it raises the max-
imum level of risk the entrepreneur is willing to take
before switching from bank ?nancing to VC ?nanc-
ing. This second key result is robust and is shown
to hold in various models of competition, includ-
ing the Hotelling, Salop, Dixit–Stiglitz, logit, and
Cournot-to-Bertrand switch. It also is consistent with
anecdotal evidence,
3
and with the empirical evidence
discussed in the next section.
2
For clarity purposes, throughout the text we refer to entrepreneurs
as female and to ?nanciers as male.
3
Consider, for example, the opinion of Andras Forgacs, managing
director at Richmond Global, a New York VC ?rm, on looking
at investment opportunities in Chengdu, China: “There are some
interesting investment opportunities where there is less competition
There exists a fairly large theoretical literature on
the impact of debt ?nancing on the competitive
behavior of rivals in the product market.
4
Brander
and Lewis (1986) pioneered this line of research when
they examined the impact of debt on competitive
behavior in a Cournot duopoly. They argued that
debt forces ?rms to focus their attention on the good
states of the world where the marginal return to out-
put is highest, and hence can serve as a commitment
to increase output, resulting in a reduction in the
rival’s own output choice. Showalter (1995) showed
that similarly, under price competition, debt could
serve as a commitment device to raise one’s own
prices and those of rivals. In contrast, Bolton and
Scharfstein (1990) argued—in an optimal contracting
framework—that debt may lead to predatory behav-
ior by deep-pocketed rivals. In turn, Faure-Grimaud
(2000) and Povel and Raith (2004) brought together
strategic elements of the Cournot approach of Brander
and Lewis (1986) and optimal contracting elements of
Bolton and Scharfstein (1990).
5
More recently, a new theoretical literature has
emerged that examines the entrepreneur’s choice
between VC ?nancing and bank ?nancing more
speci?cally. Factors affecting this ?nancing trade-off
are shown to include intellectual property protec-
tion and the “high-techness” of projects (Ueda 2004),
entrepreneur/investor input complementarity (Bet-
tignies 2008), entrepreneurial “stigma of failure”
(Landier 2003), and “strategic uncertainty” (Winton
and Yerramilli 2008).
6
Although these two literatures provide valuable
insights into the impact of ?nancial structure on com-
petitive behavior and into entrepreneurial ?nancing
[emphasis added] and more patient development of companies
and talent” (Knowledge@Wharton 2011).
4
See, for example, Brander and Lewis (1986), Maksimovic (1988),
Poitevin (1989), Bolton and Scharfstein (1990), Glazer (1994),
Showalter (1995), Faure-Grimaud (2000), and Povel and Raith
(2004). See also Peress (2010), who examines the more general inter-
action between product market competition and ?rms’ behavior in
?nancial markets.
5
There also exists a small body of literature on venture capital and
start-up ?nancing in a competitive context. In particular, Inderst
and Mueller (2004) consider the impact of competition among start-
ups on VC investment decisions and ?nancing contracts: competi-
tion shifts the bargaining power between the entrepreneur and the
VC and thus affects ownership shares of the two parties and valu-
ation of the project. Additionally, Fulghieri and Sevilir (2009) study
how market conditions affect the composition of a venture capi-
talist’s portfolio (in particular the trade-off between larger/smaller
portfolios and diversi?ed/concentrated portfolios).
6
Other trade-offs have been examined in the entrepreneurial
?nance literature, for example, the entrepreneur’s choice between
venture capital and angel ?nancing (Chemmanur and Chen 2014),
or between private and public ownership (Boot et al. 2006). See also
Renucci (2014), who examines bargaining with a venture capitalist
when bank ?nancing is an option.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 3
trade-offs, neither can address the issue examined
here, that is, the interaction among product mar-
ket competition, venture risk, and the entrepreneur’s
choice between bank ?nancing and venture capi-
tal.
7
Indeed—and this is the key contribution of this
paper—we build a simple model of entrepreneurial
?nance where relatively risky ventures use VC ?nanc-
ing while safer ventures use bank ?nancing, and
where the positive impact of competition on the
appeal of bank ?nancing relative to VC ?nancing
emerges as a robust result.
8
One exception in the prior literature, and one more
closely related to this paper, is the recent work of
Inderst and Mueller (2009), which offers a compelling
analysis of the interaction between product market
competition and the entrepreneur’s choice between
active investors (e.g., venture capitalists) and passive
ones (e.g., banks). In their model, active investors may
enable ventures to “‘strategically overinvest’ early on,
thus forestalling their rivals’ future investment and
growth” (p. 276), a strategic advantage for active
investors that becomes more valuable as competition
intensi?es. In our paper, in contrast, the distinction
between bank and VC ?nancing does not rest on
strategic overinvestment, but rather on differences in
entrepreneurial incentives, and it generates a different
outcome: here, product market competition leads to
less VC ?nancing and more bank ?nancing, not the
other way round.
This paper is organized as follows: Section 2 dis-
cusses empirical evidence related to VC ?nancing and
bank ?nancing, and to the effects of venture risk and
product market competition on the trade-off between
the two types of ?nancing. Section 3 describes the
basic setup, and §4 presents a simple trade-off between
bank ?nancing and VC ?nancing. Section 5 exam-
ines the impact of product market competition on
the ?nancing trade-off. Section 6 discusses empiri-
cal implications of the model, and §7 concludes. All
proofs are in the appendix. Extensions to the main
model are presented in the online supplement to this
paper (available on the authors’ websites).
2. Empirical Motivation
Although a variety of sources of ?nancing are avail-
able to entrepreneurs (Fraser 2005, Industry Canada
7
The concept of “strategic uncertainty” used in Winton and
Yerramilli (2008) is somewhat related to the concept of venture risk
used here. However, the trade-off between bank and VC ?nancing
is very different in their model, as are the effects of strategic uncer-
tainty on that trade-off. As well, product market competition is
altogether absent from their model.
8
In a differentiated Cournot extension of their model, Povel and
Raith (2004) brie?y discuss the impact of product homogeneity on
the competitive behavior of ?rms. However they do not examine
the impact of competition on the entrepreneur’s choice between
bank and VC ?nancing.
2009, Robb and Robinson 2014), two ?nancial inter-
mediaries emerge as particularly relevant. First, banks
play a key role for venture creation and growth.
Cosh et al. (2009) point out that among the UK ?rms
that sought external ?nancing during 1996–1997, 81%
approached banks, far more than any other ?nancial
intermediary. Fraser’s (2005) and Robb and Robin-
son’s (2014) analyses suggest, respectively, that banks
were also the most frequent capital provider for UK
start-ups in 2004 and for U.S. start-ups over the
2004–2007 period.
Another ?nancial intermediary playing a key role
in the creation and development of new ?rms is the
VC fund. Although VC funds are used by a nar-
rower set of entrepreneurs (Fraser 2005, Cosh et al.
2009), they provide much more capital to the ventures
that they do back. In the United States, the average
amount of funding from VC funds over 2004–2007
was $1,162,898, whereas the average funding in busi-
ness bank loans was $261,358 (Robb and Robinson
2014). Moreover, VC-?nanced ?rms helped to create
many successful companies, including Apple, Intel,
Federal Express, and Microsoft (Sahlman 1990), and
VC-backed companies generate 5%–7% of employ-
ment in the United States (Puri and Zarutskie 2012).
Thus, both banks and VC funds are essential to
the ?nancing of new ventures and indeed are substi-
tute forms of ?nancing in that, as shown by Berger
and Schaeck (2011), VC-backed ?rms are less likely
to rely on bank ?nancing. So what factors affect
the entrepreneur’s choice between these two types
of ?nancing? Two factors in particular may have an
impact on the bank/VC trade-off.
2.1. Venture Risk
Anecdotal evidence suggests that “riskier” ventures—
de?ned as ventures that have a lower probability
of success, but a higher payoff in case of success—
are relatively more likely to be ?nanced by venture
capital, whereas “safer” ventures are more likely to
be ?nanced by banks. The vast majority of entre-
preneurial ventures are so-called “lifestyle” ventures
(e.g., hair salons, auto-repair shops, etc.), which have
a relatively low probability of failure but will never
yield very high growth or pro?t,
9
and these ventures
are predominantly bank ?nanced (Fraser 2005). On
the other hand, lifestyle ventures are never ?nanced
by venture capital. VC funds focus instead on high-
growth ventures,
10
which are riskier ventures: they
9
The terms “high-growth” and “lifestyle” ventures are commonly
used to divide entrepreneurial ventures into two broad categories.
See, for example, Hisrich et al. (2006, pp. 19–20) and Harvard Busi-
ness School (2006).
10
To the extent that (the inverse of) the time to bring a product
to market is a proxy for growth, this is consistent with Hellmann
and Puri’s (2000) ?nding that VC-backed ?rms enjoy a signi?cantly
lower time to market.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
4 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
have a relatively high probability of failure but can be
extremely pro?table when successful. Of course, some
high-growth ventures are bank ?nanced, and taken
together these facts suggest that riskier high-growth
ventures are more likely to be ?nanced by venture
capital, and safer lifestyle ventures are more likely to
be ?nanced by banks.
Cosh et al. (2009) explicitly examined the hypothe-
sis that riskier ventures are more likely to be ?nanced
by VC funds than by banks and found some evi-
dence consistent with it. For example, they found that
banks provide male-founded ?rms with 18% less of
their desired capital relative to female-founded ?rms.
Citing Jianakopolos and Bernasek (1998) to under-
line evidence that women are more risk averse than
men, they argue that this is illustrative of a nega-
tive connection between bank ?nancing and venture
risk. They also ?nd that ?rms that recently devel-
oped a new innovation and hence are—as implicitly
argued—riskier are more likely to seek VC ?nancing.
Similarly, Hellmann and Puri (2000) found that
innovator ?rms are signi?cantly more likely to obtain
VC ?nancing, and obtain VC ?nancing earlier in the
life cycle, than imitator ?rms. To the extent that inno-
vators are riskier than imitators, and/or that there is
more risk earlier in the venture life cycle, this is con-
sistent with a positive relationship between venture
risk and VC ?nancing.
2.2. Product Market Competition
A vast empirical literature exists that examines the
connection between product market competition, on
one hand, and decisions, behavior, and performance
in organizations on the other. Perhaps the most
compelling ?nding to emerge from this broad line
of research is that competition affects organiza-
tions in many ways.
11
What about the interaction
between competition and ?nancing strategy specif-
ically? Importantly here, Cosh et al. (2009) show
that entrepreneurs are more likely to choose bank
?nancing, and less likely to choose VC ?nancing, as
the degree of competition, measured by the number
and/or size of competitors, increases.
Moreover, as discussed above, both Hellmann and
Puri (2000) and Cosh et al. (2009) ?nd evidence of
a positive relationship between venture innovative-
ness and VC ?nancing. Although we highlighted the
11
The evidence suggests a positive impact of competition on inno-
vation (Geroski 1990, Bertschek 1995, Blundell et al. 1999, Bloom
et al. 2011, Teshima 2010, Hashmi 2013) and productivity (Caves
and Barton 1990, Green and Mayes 1991, Haskel 1991, Nickell 1996,
Grif?th 2001, Syverson 2004a, b), the power of managerial incen-
tives (Burgess and Metcalf 2000, Cuñat and Guadalupe 2005, Baggs
and Bettignies 2007), collective bargaining and wage negotiations
(Abowd and Lemieux 1993), mergers and acquisitions (Hoberg and
Phillips 2010), and IPO decisions (Chemmanur and He 2011).
strong positive correlation between venture innova-
tiveness and venture risk, it is also important to
underline the negative correlation between venture
innovativeness and product market competition. In
Hellmann and Puri (2000, p. 969), for example, “the
innovator variable captures the notion of ?rms that
introduce a new product that is considered not to be
a close substitute to any product or service already
offered on the market; ?rms that introduce a new
product or service that is considered to perform an
order of magnitude better than any substitute prod-
ucts already offered in the market; or ?rms that
are developing new technologies that could lead to
products satisfying either of the two criteria above.”
Indeed this corresponds quite closely to a description
of lack of competition in the product market.
12
To the
extent that innovation captures (lack of) competition,
the evidence presented in these two papers is consis-
tent with a negative connection between competition
and VC ?nancing.
Related to this, we note that patents, which are
associated with less substitutable products and con-
fer signi?cant monopoly power to their holders, may
arguably capture lack of competition. Kortum and
Lerner’s (2000) ?nding that VC-?nanced ?rms have
higher quality patents than non-VC-?nanced ?rms
then also suggests a negative connection between
competition and VC ?nancing.
Finally, the evolution of the ?nancing strategy over
the product life cycle—namely the relative prevalence
of VC ?nancing early in the product life cycle and of
bank ?nancing later (Berger and Udell 1998)—could
also be the result of a connection between competition
and ?nancing choice. Early in the life cycle, when the
entrepreneur introduces a new product and creates a
new market, competition is low and VC ?nancing is
chosen. As the life cycle progresses, however, similar
products emerge, competition intensi?es, and bank
?nancing eventually becomes the preferred choice.
The above discussion yields two key implications
regarding the bank/VC ?nancing trade-off. It sug-
gests ?rst that riskier ventures are more likely to be
VC ?nanced than bank ?nanced, and second that
a greater degree of product market competition is
likely to favor bank ?nancing over VC ?nancing. In
what follows we propose a theory of entrepreneurial
?nancing that highlights the interaction among the
bank/VC ?nancing trade-off, venture risk, and prod-
uct market competition.
12
The degree of substitutability between products (e.g., Sutton
1992), and the similarity between perceived product performance
and/or between ?rms’ costs (e.g., Bertrand competition with asym-
metric costs or performance components), are often used as mea-
sures of competition in industrial organization.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 5
3. Basic Model
An entrepreneur c has a novel venture idea that may
open a new and untapped market. Entrepreneur c is
wealth constrained with no initial wealth and is solic-
iting a (small) cash investment í from a ?nancier ]
to start her venture, which is to last for two periods.
Both c and ] are risk neutral, and ] ’s cost of capital
is normalized to zero.
At the beginning of the game, c makes a take-it-or-
leave-it contractual offer to ] ,
13
specifying a credible
way in which ] is to receive a nonnegative expected
payoff (net of í ) from the investment, thus ensur-
ing his participation. Once the initial investment is
made, the venture is started, but it requires c’s active
management for its success. The key elements of the
model can be described as follows.
First-Period Pro?t and Exogenous Shocks. Under c’s
management, the venture is launched in the ?rst
period, offering goods/services at marginal cost ¯ c ?
+
in a new monopolistic market. One of two states
of the world emerges. In the good state, which occurs
with exogenous probability (1 ?í
1
), with í
1
? |0, 1),
the venture generates monopoly pro?t H
1
(í
1
). In the
bad state, which occurs with probability í
1
, however,
a negative and temporary
14
shock affects the venture,
which generates zero pro?t in that period as a result.
This shock captures an exogenous randomness that
is inherent to all ventures, for example, random con-
sumer preferences, or macroeconomic or technologi-
cal shocks.
The good-state pro?t H
1
(í
1
) is strictly increas-
ing in í
1
such that the expected ?rst-period pro?t
(1 ?í
1
)H
1
(í
1
) =
¯
H
1
is a constant for all í
1
? |0, 1).
Conveniently, one can easily verify that the variance
of the ?rst-period pro?t is a strictly increasing func-
tion of the shock probability í
1
, and in what follows
we refer to í
1
as the “short-term risk” of the venture.
Thus, two projects with different shock probabilities
yield the same expected short-term pro?t but face dif-
ferent levels of risk.
Second-Period Pro?t and Product Market Competition.
In the second period, a rival r enters the market, offer-
ing related but distinct goods/services at marginal
cost ¯ c. Thus emerges competition in the product mar-
ket. Parameter 0 ? |0
min
, 0
max
], with 0
max
> 0
min
> 0,
captures the degree of competition in the industry,
which, as discussed in §5, will depend on the way
demand/competition is modeled: it may represent the
inverse of the transport cost in a Hotelling (1929)
13
The entrepreneur has all ex ante bargaining power: there are
many more ?nanciers wishing to invest than there are good entre-
preneurs (good projects to be funded).
14
This negative shock is temporary in that it does not affect second-
period payoffs. The case of a long-term shock that yields zero pro?t
in the second period as well (or that conveys negative information
about second-period pro?tability prospects) is less interesting (as
discussed in Footnote 33).
or Salop (1979) model, or the inverse of variance of
errors in a logit model, for example.
Bene?tting from a ?rst-mover advantage, under c’s
management the venture may—through innovation—
lower the marginal cost of production from ¯ c to c,
where c - ¯ c, thus gaining a cost advantage over
the entrant and generating “high” pro?t H
l
(0) =
H(c, ¯ c, 0) > 0.
15
The successful innovation required
for gaining a cost advantage takes place with endoge-
nous probability ¡ ? |0, 1). With probability 1 ?¡ the
marginal cost remains unchanged at ¯ c, and in the sec-
ond period the venture yields “status quo” or “sym-
metric” pro?t H
s
(0) = H( ¯ c, ¯ c, 0) > 0. Accordingly, in
the second period, c’s venture generates the expected
duopoly pro?t H
2
(¡, 0) = H
s
(0) + ¡(H
l
(0) ? H
s
(0)),
which can be expressed as the sum of the status quo
pro?t H
s
(0) and the expected gain from a cost advan-
tage ¡(H
l
(0) ?H
s
(0)).
Note that, similar to the ?rst period, we could have
an exogenous negative second-period shock occurring
with probability í
2
and leading to zero pro?t in that
period. However, unlike the ?rst-period shock, which
will be shown to play a key role in our model, a
second-period shock plays little role here, and indeed
our results can be shown to hold for all í
2
? |0, 1).
Hence, for simplicity, and without loss of generality,
we assume í
2
=0.
Entrepreneurial Participation and Effort. Entrepre-
neur c affects the venture in two primary ways.
First, as mentioned above, her active management
is required for the venture to operate successfully.
Absent the entrepreneur in period 1, the venture
yields zero pro?t in that period. Absent the entre-
preneur in period 2, the venture is liquidated and
yields liquidation value |, with | ? (0, H
s
(0
max
)).
16
The
entrepreneur’s personal opportunity cost of remain-
ing involved in the venture is a > 0 with a ?0, and
she must thus expect a payoff of at least a in period i
to remain involved in the venture in that period.
The second way in which c can affect the venture
is by exerting effort u at the beginning of the second
period, which increases the probability ¡ of successful
innovation and of decreasing the marginal cost from
¯ c to c.
17
For simplicity we assume ¡ = u, and use ¡
15
In any pro?t function H(·, ·, 0), price–cost margin function
P(·, ·, 0), or demand function x(·, ·, 0), the ?rst argument refers to
the marginal cost of the ?rm generating the pro?t under study, and
the second argument refers to the rival’s marginal cost. Equilibrium
prices have been determined as functions of costs and substituted
out of these functions.
16
This condition is suf?cient to ensure that liquidation is inef?cient:
it generates less surplus than keeping the venture as a going con-
cern, regardless of entrepreneurial effort or degree of competition.
17
We could model entrepreneurial effort in both periods rather than
in the second period only. However, as will become clear below
(see expression (5)), ?rst-period pro?ts do not affect the trade-off
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
6 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
to capture both entrepreneurial effort and innovation
probability. Effort comes at a personal cost K(¡) =
(|,2)¡
2
to the entrepreneur. It is observable but not
veri?able in court and thus noncontractible.
Contractual Incompleteness and Property Rights. Real-
ized pro?ts at the end of periods 1 and 2 are assumed
to be observable by both parties, but not veri?able in
court, and thus cannot be contracted upon.
18
To quote
Bolton and Scharfstein (1996, p. 5), “this assumption
is meant to capture the idea that managers have some
ability to divert corporate resources to themselves at
the expense of outside investors,” and that “such perk
consumption and investment may be dif?cult to dis-
tinguish from appropriate business decisions and thus
impossible to control through contracts.” This has two
important consequences. First, c and ] must bargain
at the end of each period, over pro?t H
1
, and H
l
or H
s
, as they arise. Second, the contract designed at
date 0 can only specify the (possibly contingent) allo-
cation of property rights over the venture, which can
be of two types. Under entrepreneur control (|), c has
full ownership of the venture and complete control
over the assets of the venture. At the other end of
the ownership spectrum, under ?nancier control (í ),
] owns the venture and controls the assets.
Bank vs. Venture Capital. Entrepreneur c can seek
?nancing from one of two types of investors: a ven-
ture capitalist (or angel investor) or a banker. We
posit that banks and venture capitalists share simi-
larities and exhibit differences in the way in which
control-right allocations affect bargaining with the
entrepreneur. Financier control, for example, is likely
to affect ?nancier/entrepreneur bargaining similarly
whether the ?nancier is a bank or a venture capital-
ist. In both cases, ] has complete control over the
assets of the venture and can extract 100% of the real-
ized pro?t through bargaining at the end of a period.
This is consistent with Nash bargaining where, in case
negotiations break down, ] can simply exclude c from
accessing the venture and continue to enjoy all rents.
Entrepreneur c cannot interfere and gets nothing.
In contrast, we argue that entrepreneur control may
lead to different bargaining outcomes under bank
?nancing and VC ?nancing. Under bank ?nancing,
between bank ?nancing and VC ?nancing; hence neither does
potential ?rst-period entrepreneurial effort affecting these pro?ts.
To avoid unnecessary complications and keep the model as simple
as possible, we do not model ?rst-period effort explicitly here.
18
The assumption of nonveri?able cash ?ows ?ts squarely within
the recent literature on corporate/entrepreneurial ?nance, where
it is commonly used. See, for example, Hart and Moore (1989,
1994, 1995), Bolton and Scharfstein (1990, 1996), Berglöf (1994),
Gertner et al. (1994), Fluck (1998), Myers (2000), Bascha and Walz
(2001), Dybvig and Wang (2003), Repullo and Suarez (2004), and
Bettignies (2008).
entrepreneur control is the converse of ?nancier con-
trol: c has complete control over the assets of the ven-
ture and can extract 100% of realized pro?t through
bargaining at the end of a period. In case negotiations
break down during Nash bargaining, c can simply
exclude ] from accessing the venture and continue to
enjoy all rents. Financier ] cannot interfere and gets
nothing.
Under VC ?nancing, however, this bargaining out-
come is unlikely to happen even under entrepreneur
control. Indeed, a key characteristic of venture cap-
italists here is that—given their level of expertise,
their social networks, and their ?nancial resources—
they may be able to extract second-period rents even
when c owns the venture and controls its assets.
In the second period, once the venture capitalist is
familiar with the idea and its implementation, he
may be in a stronger bargaining position relative to
the entrepreneur because—as argued by Ueda (2004)
previously—in the case of a bargaining breakdown he
may be able to successfully “steal” the entrepreneur’s
idea with some probability \ ? |\
min
, 1] with \
min
?
(0, 1). In that case, as is easily shown, the Nash bar-
gaining equilibrium split would lead the entrepreneur
and the venture capitalist to obtain shares 1 ?\ and
\ of realized pro?t, respectively. Indeed, even with-
out any “stealing” by the venture capitalist, as long as
he has the ability to interfere with the entrepreneur’s
access to second-period rents in the case of a bar-
gaining breakdown,
19
even with a small probability,
then he will be able to extract a share of pro?t in
equilibrium.
20
Accordingly in what follows we assume that entre-
preneurial control under bank ?nancing leads the
entrepreneur to extract 100% of realized pro?ts
in both periods, whereas under VC ?nancing the
entrepreneur extracts all realized pro?ts in the ?rst
period, but only a fraction 1 ? \ of these realized
19
For example, suppose that in the case of a bargaining breakdown,
the venture capitalist can use his clout in entrepreneurial circles
and his access to related portfolio companies to interfere with the
venture’s pro?ts, and that as a result the entrepreneur ends up with
only a fraction q of the pro?ts she would have obtained without
interference. One can easily show that, in that case, the Nash bar-
gaining split would lead the entrepreneur and the venture capitalist
to obtain shares 1 ?\ =(1 +q),2 and \ =(1 ?q),2, respectively.
20
Note an implicit distinction between the venture capitalist and
the entrepreneur. On one hand, in the case of VC ?nancing, the ven-
ture capitalist can extract rents from the entrepreneur, even under
entrepreneur control, by essentially threatening to “steal” her idea
or interfere with her business in the case of a bargaining break-
down. In contrast, under bank ?nancing, the entrepreneur cannot
extract rents from the banker under ?nancier control. The reason
for this is that the entrepreneur, unlike the venture capitalist, does
not have the ?nancial resources required to credibly threaten the
banker to “steal” the idea (she could not ?nance a new venture
with the “stolen” idea), nor does she have the clout/networks to
credibly interfere with the banker’s access to rents.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 7
pro?ts in the second period.
21
As we shall see below,
this distinction between bank and VC ?nancing leads
to a signi?cant trade-off between these two ?nancing
possibilities.
Timing of the Game. We summarize the timing of the
game as follows:
22
• At date 0, c makes a contractual offer to ] , who
is either a banker or a venture capitalist. The contract
only speci?es the (possibly contingent) allocation of
property rights over the venture, as well as possible
date 0 monetary transfers. The entrepreneur decides
whether or not to manage the venture in the ?rst
period, at personal cost a, with a ?0.
• At date 1, under c’s management, the ven-
ture generates monopoly pro?t H
1
with probability
(1 ?í
1
), and zero pro?t with probability í
1
. Absent
c, the venture generates zero pro?t.
• At the beginning of the second period, rival r
enters the market. The entrepreneur decides whether
or not to manage the venture in the second period,
at personal cost a, with a ? 0. In the former
case, she exerts innovation effort ¡, at personal cost
K(¡) =|¡
2
,2.
• At date 2, under c’s management, the venture
generates expected pro?t H
2
(¡, 0) with probability 1?
í
2
=1. Absent c, the venture generates zero pro?t.
Regularity Condition. Throughout the paper we as-
sume that
í ?\
min
H
s
(0
max
). (1)
This is a suf?cient condition to ensure that the
project can be ?nanced under VC ?nancing for all
\ ? |\
min
, 1] and 0 ? |0
min
, 0
max
], and that there exists
an í
max
(0) > 0 such that for all 0 ? |0
min
, 0
max
] bank
?nancing is feasible if and only if (henceforth iff) í
1
?
|0, í
max
(0)]. Because our primary interest lies with the
trade-off between VC ?nancing and bank ?nancing,
we henceforth focus on situations where condition (1)
holds, and refer to it where relevant.
23
We relax this
21
We assume that the venture capitalist can extract a fraction \
of realized pro?ts even under entrepreneur control, but only in
the second period. This assumption is made to capture the idea,
important in our view, that it requires learning about the venture
before the venture capitalist can gain real bargaining power over
the entrepreneur, and for analytical simplicity. All results of the
model continue to hold if we assume the venture capitalist can
extract \ in both periods rather than in the second period only.
22
Our model shares some of the model timing elements, as well
as the assumptions of contractual incompleteness and entrepre-
neurial effort, with Bettignies (2008). Despite these similarities, the
two models differ signi?cantly. Bettignies’ (2008) focus is on the
effects of investor cost of capital and entrepreneur–investor effort
complementarity on an entrepreneur’s optimal contracting envi-
ronment. It cannot address issues of venture risk and competition
that are central to this paper, nor does it examine speci?cally the
entrepreneur’s choice between bank and VC ?nancing.
23
If parameter í is such that only one type of ?nancing is feasible,
the ?nancing decision problem becomes exceedingly simple and
obviously less interesting.
regularity condition in §A.4 in the appendix and show
that the main results of the model continue to hold.
First-Best Benchmark. We consider the ?rst-best
scenario where the entrepreneur is not wealth
constrained and can ?nance the venture herself. Pro-
ceeding by backward induction, we ?nd that in the
second period, if she actively manages the venture,
she exerts effort ¡
FB
to maximize her expected payoff
H
2
(¡, 0) ? |¡
2
,2 = H
s
(0) + ¡|H
l
(0) ? H
s
(0)] ? |¡
2
,2,
yielding ?rst-best effort ¡
FB
(0) = |H
l
(0) ? H
s
(0)],|.
Her equilibrium expected payoff, H
2
(¡
FB
(0), 0) ?
||¡
FB
(0)]
2
,2 is larger than a+|,
24
and anticipating this
she chooses to be actively involved in the venture in
the ?rst place. In the ?rst period, her expected payoff
¯
H
1
is also larger than her cost a of managing the ven-
ture, and hence she chooses to be actively involved in
the venture in that period. Thus, from a date 0 point
of view, c’s expected payoff, and the ?rst-best total
surplus created, can be expressed as
U
FB
(0) =
¯
H
1
+V
FB
(0) ?í , with
V
FB
(0) = H
2
(¡
FB
(0), 0) ?||¡
FB
(0)]
2
,2, (2)
where V
FB
(0) captures the venture’s (expected) contin-
uation surplus in the second period. Regularity con-
dition (1) ensures that the venture is worth ?nancing
in the ?rst-best scenario.
25
4. Optimal Financing Choice
We now turn to bank ?nancing and VC ?nancing and
examine how these sources of capital compare to the
?rst-best benchmark and to each other.
4.1. Bank Financing
The entrepreneur has two control-rights allocations
available to design the optimal contract for the
?nancier/banker: under ?nancier control in period i,
i = 1, 2, c anticipates that end-of-period bargaining
will result in ] extracting all rents, and in her getting
nothing. This zero expected payoff is less than her
opportunity cost a of actively managing the venture,
and accordingly she decides not to remain involved
in the venture during that period.
Under entrepreneurial control, in contrast, c antic-
ipates that end-of-period bargaining will result in
24
We know from c’s maximization program that her equilibrium
expected payoff in the second period is strictly larger than H
s
(0),
which itself is larger than a +|.
25
To see this, note that V
FB
(0) =H
2
(¡
FB
(0), 0) ?||¡
FB
(0)]
2
,2 ?H
s
(0)
for any given 0 ? |0
min
, 0
max
]. This is because c could generate a
payoff H
s
(0) if she chose effort ¡ =0, and must thus do at least as
well by choosing equilibrium effort ¡
FB
(0). Moreover, as discussed
below in §5, symmetric pro?t H
s
(0) is decreasing in the degree
of competition, and hence H
s
(0) ? H
s
(0
max
) for all 0 ? |0
min
, 0
max
].
From these two points, it follows directly that V
FB
(0) ?H
s
(0
max
) >í ,
which in turn implies U
FB
(0) ?0 for all 0 ? |0
min
, 0
max
].
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
8 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
her extracting all rents, as in the ?rst-best sce-
nario described above. Thus, entrepreneurial con-
trol in period i, i = 1, 2 leads to the ?rst-best out-
come in that period. In period 2, c chooses to be
actively involved in the management of the venture,
exerts the ?rst-best effort ¡
|
(0) = |H
l
(0) ? H
s
(0)],|
= ¡
FB
(0), and generates equilibrium expected payoff
H
2
(¡
|
(0), 0) ?||¡
|
(0)]
2
,2. In period 1, c also chooses
to be actively involved in the management of the ven-
ture, generating expected payoff
¯
H
1
in that period.
Following Hart and Moore (1989) and Bolton and
Scharfstein (1990, 1996), we know that in a framework
with these two feasible control allocations, the opti-
mal “pure strategy”
26
contract for c to offer to ] is
a simple debt contract, which speci?es at date 0 the
debt repayment D to be made at date 1. If c pays D to
] at date 1, she retains full control in period 2. On the
other hand, if c does not repay D at date 1, ] obtains
full control over the venture.
If the negative shock does not occur in period 1
and the venture generates pro?t H
1
, c faces two
options at date 1. First, she can repay D and retain
control over the venture in period 2. In that case,
as mentioned she exerts effort ¡
|
(0), and generates
expected second-period continuation surplus V
|
(0) =
H
2
(¡
|
(0), 0)?||¡
|
(0)]
2
,2. Alternatively, she can strate-
gically default (see, e.g., Bolton and Scharfstein 1990).
Doing so would lead to renegotiation since ?nancier
control and the ensuing liquidation yield a lower pay-
off (|) than entrepreneur control, and c’s payoff in
that case would depend on the outcome of renego-
tiation. Without loss of generality we assume that
] has full bargaining power in renegotiation follow-
ing a default
27
and requires payment D
max
(0) =V
|
(0)
from c in exchange for letting her retain venture con-
trol. Anticipating this, c makes debt repayment D at
date 1 if D?V
|
(0) but would default and renegotiate
if D >V
|
(0).
In contrast, if the negative shock does occur in
period 1, c gets no cash and must default at date 1. As
a result, ] takes control over the assets at that time,
and c quits the venture, and this in turn leads to liq-
uidation payoff |. (There can be no scope for renego-
tiation here since | >0 implies that ] is strictly better
off with venture control than without, and c has no
cash with which to renegotiate.)
26
For expositional simplicity, we focus on “pure strategy” contracts
where an allocation of control is assigned with probability 0 or 1
contingent on debt repayment or default. In the online supplement
to this paper, we allow for “mixed strategy” debt contracts whereby
c retains control with probability p
D
if she repays the debt, and
with probability p
0
if she does not. The main results of the model
continue to hold.
27
The main results of our model do not depend on the allocation
of bargaining power in renegotiation and indeed would continue
to hold under alternative bargaining allocations. See discussion in
§A.4 in the appendix.
Naturally, at date 0 ] foresees that he will get
at most V
|
(0) in the good state and | in the bad
state. Thus bank ?nancing will be feasible only if
í ?(1?í
1
)V
|
(0)+í
1
|, or í
1
?í
max
(0) with í
max
(0) =
(V
|
(0)?í ),(V
|
(0)?|). Regularity condition (1) ensures
that í
max
(0) >0 for all 0 ? |0
min
, 0
max
].
28
Since c has full bargaining power at date 0, she sets
D so that she extracts all rents from ] , i.e., such that
(1 ?í
1
)D +í
1
| ?í =0, for all í
1
? í
max
(0). Accord-
ingly, from a date 0 point of view, c’s expected payoff
from bank ?nancing is the total surplus generated by
the venture under that type of ?nancing:
U
|
(0) =
¯
H
1
+(1 ?í
1
)V
|
(0) +í
1
|?í , with
V
|
(0) = H
2
(¡
|
(0), 0) ?||¡
|
(0)]
2
,2.
(3)
Thus, absent a negative shock in the ?rst period,
bank ?nancing is ef?cient: it enables c to retain full
control over the venture and generates ?rst-best entre-
preneurial effort ¡
|
(0) = ¡
FB
(0) and in turn ?rst-best
continuation surplus V
|
(0) = V
FB
(0). On the other
hand, with probability í
1
a negative shock does occur
in the ?rst period, and this leads to inef?ciency:
although this shock is temporary and does not sig-
nal anything about the prospects of the venture in
the second period, it leads to the inef?cient liquida-
tion of a venture which should instead—in a ?rst-
best world—continue business as usual in the second
period. More formally, the inef?ciency associated with
bank ?nancing can be expressed as U
FB
(0) ?U
|
(0) =
í
1
|V
|
(0) ?|].
29
We summarize the foregoing results in the follow-
ing proposition.
Proposition 1. Conditional on short-term venture suc-
cess, bank ?nancing generates ?rst-best entrepreneurial
effort ¡
|
(0) =¡
FB
(0) and continuation surplus V
|
(0) =
V
FB
(0). On the other hand, conditional on short-term
failure—even when this failure is temporary and uninfor-
mative about long-term prospects—bank ?nancing leads to
inef?cient liquidation of the venture at date 1.
28
As demonstrated in Footnote 25, V
|
(0) = V
FB
(0) > H
s
(0
max
).
Together with our condition that | - H
s
(0
max
), this implies
V
|
(0) >|, and together with regularity condition (1) this implies
V
|
(0) >í . From this it immediately follows that í
max
(0) >0 for all
0 ? |0
min
, 0
max
].
29
Note that these characteristics of bank ?nancing are similar to
the classic models of debt mentioned above (Hart and Moore 1989;
Bolton and Scharfstein 1990, 1996). Here, however, second-period
pro?ts are driven by endogenous efforts, and (as shall become clear
in §§4.3 and 5.2) this will play a crucial role in the analysis of
the trade-off between bank ?nancing and VC ?nancing, and of the
effects of product market competition, neither of which were exam-
ined in these prior models.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 9
4.2. VC Financing
With VC ?nancing, the entrepreneur again has two
control-rights allocations available to design the opti-
mal contract: ?nancier control in period i, i = 1, 2,
yields the same outcome as that under bank ?nanc-
ing: c anticipates getting nothing in end-of-period
renegotiation and decides not to remain involved in
the venture during that period.
Entrepreneurial control, on the other hand, is differ-
ent with VC ?nancing. In period 2, c anticipates that
end-of-period bargaining will result in her extracting
a fraction 1 ?\ of the rents. If she manages the ven-
ture, she exerts effort ¡
.c
to maximize her expected
payoff (1 ? \)H
2
(¡, 0) ? |¡
2
,2 = (1 ? \)|H
s
(0) +
¡|H
l
(0) ?H
s
(0)]] ?|¡
2
,2. She has lower effort incen-
tives than under entrepreneurial control with bank
?nancing, since she anticipates extracting a fraction
(1 ?\), rather than 100%, of the rents and thus exerts
a second-best effort:
¡
.c
(0) =(1 ?\)
H
l
(0) ?H
s
(0)
|
-¡
FB
(0).
Her equilibrium expected payoff,
(1 ?\)H
2
(¡
.c
(0), 0) ?
||¡
.c
(0)]
2
2
,
is larger than a+(1–\)| (for reasons provided in Foot-
note 24), and anticipating this she chooses to remain
actively involved in the venture in that period. In
period 1, again her expected payoff
¯
H
1
is larger than
her cost a of managing the venture, and she remains
actively involved.
Entrepreneurial control enables c to commit to
return some rents to the venture capitalist through
bargaining at the end of period 2. In this environment,
it follows directly that it is optimal for c to offer the
following contract to the venture capitalist: entrepre-
neurial control in both periods regardless of the state
of the world and a payment í +T from the venture
capitalist to the entrepreneur at date 0, with T ? 0
chosen so as to leave no ex ante rents to the ?nancier
(i.e., such that \H
2
(¡
.c
(0), 0) ?í ?T =0). Since c can
use date 0 transfer T to extract all ex ante rents,
her contractual objective is to maximize total sur-
plus. Additionally, since entrepreneurial control leads
to more effort—and hence, ceteris paribus, to more
pro?t—than investor control, it is optimal to assign
entrepreneurial control regardless of the state of the
world.
30
30
In the online supplement to this paper, we examine an extension
of the main model where the venture capitalist exerts effort as well.
In that case the optimal allocation of control rights depends on the
ef?ciency of c relative to ] . Indeed, if the venture capitalist is suf?-
ciently more ef?cient than c in exerting effort, giving control rights
to him may, by increasing his incentives, generate more surplus
than giving control rights to c. In that case c may do just that: she
may sell the venture to the venture capitalist.
Note that the maximum gross payoff that ] can
expect to get under VC ?nancing is \H
2
(¡
.c
(0), 0).
Hence VC ?nancing is feasible iff í ? \H
2
(¡
.c
(0), 0).
Regularity condition (1) ensures that VC ?nancing is
feasible for all \ ? |\
min
, 1] and all 0 ? |0
min
, 0
max
].
31
From a date 0 point of view, c’s expected payoff
from VC ?nancing is the total surplus generated by
the venture under that type of ?nancing:
U
.c
(0) =
¯
H
1
+V
.c
(0) ?í , with
V
.c
(0) = H
2
(¡
.c
(0), 0) ?||¡
.c
(0)]
2
,2,
(4)
where V
.c
(0) captures the venture’s continuation sur-
plus in the second period.
As mentioned above, anticipating she will have
to share rents with the venture capitalist mutes
c’s incentives, leading to second-best entrepreneurial
effort ¡
.c
(0) - ¡
FB
(0) and, in turn, to second-best
continuation surplus V
.c
(0) - V
FB
(0). On the posi-
tive side, however, the venture capitalist’s bargaining
power and rent extraction ability serve as commit-
ment devices for c to return some of the rents to ] ,
which in turn enables c to ?nance the venture with-
out resorting to debt and facing the associated liqui-
dation inef?ciencies discussed above. More formally,
the inef?ciency associated with VC ?nancing can be
expressed as U
FB
(0) ?U
.c
(0) =V
FB
(0) ?V
.c
(0).
We summarize the foregoing results in the follow-
ing proposition.
Proposition 2. Under VC ?nancing, short-term ven-
ture failure, resulting from a negative ?rst-period shock,
has no lasting impact on the venture, which continues to
operate in the second period regardless. On the other hand,
VC ?nancing generates second-best entrepreneurial effort
¡
.c
(0) -¡
FB
(0) and continuation surplus V
.c
(0) -V
FB
(0).
Importantly, although in our model an equity con-
tract per se would not yield any payoff to the investor
(since by assumption cash ?ows are not veri?able),
under VC ?nancing, even when the entrepreneur
retains overall control of the venture, the venture cap-
italist is still able to extract a stream of payoffs—a
fraction \ of pro?t—that is similar to the one typically
obtained in a standard voting equity contract. Indeed
the contract offered to the venture capitalist is in effect
an “equity-type” contract.
In practice, VC contracts include both control rights
and cash ?ow rights (Kaplan and Strömberg 2003).
Cash ?ow rights do have a role in practice because
pro?ts are veri?able to some extent. They are not per-
fectly veri?able, however, in that entrepreneurs have
the ability to divert some, if not all, of the cash ?ows.
Indeed, that is an important reason why control rights
are also used in VC contracts: they complement cash
31
This follows directly from the logic used in Footnote 25.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
10 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
?ow rights, controlling opportunistic behavior by the
entrepreneur. In this model, we assume that pro?ts
are not veri?able at all.
32
In doing so, we focus on con-
trol rights and on end-of-period bargaining, rather
than on the cash ?ow rights, and this enables us to
introduce an equity-like contract in the model (sim-
ilar to equity contracts observed in practice), all the
while keeping our assumption of nonveri?ability of
cash ?ows necessary for debt to emerge as optimal
under bank ?nancing. The modeling of equity ?nanc-
ing under the assumption of non-veri?ability of cash
?ows is neither novel nor speci?c to this paper. See,
for example, Fluck (1998), Myers (2000), and Dybvig
and Wang (2003).
4.3. Financing Trade-off
Comparing c’s expected payoffs, it is easy to see that
she chooses bank ?nancing over VC ?nancing iff
U
|
(0) ?U
.c
(0) ?0,
iff |V
|
(0) ?V
.c
(0)] ?í
1
|V
|
(0) ?|] ?0. (5)
Expression (5) yields two key results. First, it illus-
trates the main trade-off between bank ?nancing and
VC ?nancing.
Proposition 3. The entrepreneur’s ?nancing trade-off
is the following: on the upside, bank ?nancing leads to a
larger continuation surplus than VC ?nancing, V
|
(0) >
V
.c
(0). On the downside, with probability í
1
, a negative
shock occurs that prevents continuation under bank ?nanc-
ing but not under VC ?nancing.
The intuition follows directly from Propositions (1)
and (2): conditional on repaying the debt, bank ?nanc-
ing enables c to retain full access to the venture’s
rents, giving her ?rst-best effort incentives and yield-
ing a surplus larger than what could be obtained
under VC ?nancing, where rent sharing mutes entre-
preneurial incentives. On the other hand, in the event
of a negative short-term shock, lack of cash ?ow pre-
vents the repayment of bank debt, leading to inef?-
cient liquidation—an inef?ciency that does not arise
under VC ?nancing.
33
32
In the online supplement to this paper, we examine an extension
to the main model where venture capitalists can verify cash ?ows
with some probability. We show that the main results of the model
continue to hold in that case.
33
Note that if the negative ?rst-period shock were not temporary
and instead led to zero pro?ts in the second period as well as in the
?rst period, the trade-off between bank ?nancing and VC ?nanc-
ing would become exceedingly simple and much less interesting.
Following a negative shock, the prospects of future pro?ts would
disappear and the venture would be liquidated regardless of the
?nancing choice. In that case, bank ?nancing has no inef?cient liq-
uidation disadvantage relative to VC ?nancing but retains its incen-
tive advantage and thus unambiguously dominates VC ?nancing.
The second key result emerging from expression (5)
is that there exists a threshold level of venture risk í
1
,
denoted í
?
1
(0) =|V
|
(0) ?V
.c
(0)],|V
|
(0) ?|], such that
U
|
(0) ?U
.c
(0) ?0 iff í
1
?í
?
1
(0). (6)
We state this result in the following proposition.
Proposition 4. There exists a threshold level of ven-
ture risk í
?
1
(0) =|V
|
(0) ?V
.c
(0)],|V
|
(0) ?|], such that
the entrepreneur’s optimal ?nancing choice is bank ?nanc-
ing if í
1
?í
?
1
(0), and VC ?nancing otherwise.
Intuitively, for relatively risky ventures (i.e., ven-
tures with a relatively high probability of failure
(í
1
>í
?
1
(0)) but a relatively high pro?t H
1
(í
1
) in case
of success), bank ?nancing is not optimal, because
the advantage of this type of ?nancing in terms of
continuation incentives is more than offset by the
risk of liquidation associated with short-term failure.
VC ?nancing, which is not subject to this liquidation
risk, is preferred in that case. Conversely, for rela-
tively safe ventures (i.e., ventures with a relatively
low probability of failure (í
1
?í
?
1
(0)) but a relatively
low pro?t H
1
(í
1
) in case of success), bank ?nancing
makes sense because the downside risk of liquidation
is relatively low relative to the upside continuation
potential associated with high incentives.
Note that regularity condition (1) ensures that bank
?nancing is always feasible at the venture risk thresh-
old í
?
1
(0). In other words, it ensures that í
?
1
(0) -
í
max
(0) for all 0 ? |0
min
, 0
max
].
34
5. Product Market Competition
In his classic work on sunk costs and market struc-
ture, Sutton (1992, p. 9) argued that for a given con-
centration level “such features of the market as the
physical nature of the product (homogeneous versus
differentiated products) and the climate of competi-
tion policy (a strict or acquiescent approach to price
coordination by ?rms)” will affect toughness of compe-
tition in the industry. Many of the workhorse models
of competition used in the industrial organization lit-
erature today are consistent with that view: a decrease
in the transport cost | on a Hotelling (1929) line or a
Similarly, if the negative ?rst-period shock were to convey nega-
tive information about second-period prospects, this would have
no impact under bank ?nancing (where liquidation occurs whether
or not this information is present) but would reduce expected pay-
offs under VC ?nancing. It would therefore tilt the trade-off in favor
of bank ?nancing.
34
Strict bank ?nancing feasibility at í
?
1
(0) requires í - (1 ?
í
?
1
(0))V
|
(0) +í
?
1
(0)|, which simpli?es to í - V
.c
(0). (Equivalently,
one can verify that í
?
1
(0) -í
max
(0) iff í -V
.c
(0).) Using logic sim-
ilar to that of Footnote 25, one can readily verify that V
.c
(0) >
\
min
H
s
(0
max
). Together with condition (1), this implies í -V
.c
(0) for
all 0 ? |0
min
, 0
max
].
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 11
Salop (1979) circle, a decrease in the variance of errors
in a logit model, or an increase in the constant elastic-
ity of substitution in a Dixit–Stiglitz framework (Dixit
and Stiglitz 1977), for example, all capture an increase
in the degree of homogeneity across products. In
this paper as well we use the degree of homogene-
ity across products as our main measure of competi-
tion 0.
35
Rather than focusing on a particular demand
speci?cation, however, we highlight commonalities
across speci?cations and show that because of these
commonalities, under these various demand frame-
works, competition has qualitatively similar effects in
our model.
5.1. Status Quo Pro?t and the Value of
Cost Advantage
In this section we highlight two characteristic effects
of competition that are common to the Hotelling
line, Salop circle, Dixit–Stiglitz, logit, and Cournot-
to-Bertrand switch demand speci?cations, namely a
reduction in the status quo pro?t H
s
(0) and an increase
in the value of a cost advantage H
l
(0) ?H
s
(0). In the
next section we use these simple results to explain
why competition increases the overall appeal of bank
?nancing relative to VC ?nancing.
5.1.1. Status Quo Pro?t. Generated when the ven-
ture has failed to innovate and faces the same
marginal cost ¯ c as the entrant, status quo pro?t can
be expressed as the product of markup and demand:
H( ¯ c, ¯ c, 0) = P( ¯ c, ¯ c, 0)x( ¯ c, ¯ c), where P( ¯ c, ¯ c, 0) repre-
sents the markup and x( ¯ c, ¯ c) represents demand. The
only effect of competition here is to lower equilibrium
prices. Intuitively, as the degree of homogeneity across
products increases, consumers become more sensitive
to prices, putting downward pressure on prices and in
turn on markups: dP( ¯ c, ¯ c, 0),d0 -0. Since c’s venture
and the entrant have identical costs, equilibriumprices
are identical in both ?rms, and the two rivals share
the market equally regardless of the degree of compe-
tition: x( ¯ c, ¯ c) =1,2. Accordingly, the overall impact of
competition on the status quo pro?t is dH( ¯ c, ¯ c, 0),d0 =
(dP( ¯ c, ¯ c, 0),d0)x( ¯ c, ¯ c) -0.
5.1.2. Value of a Cost Advantage. This is the dif-
ference between pro?t obtained when the venture
successfully innovates and gains a cost advantage
over the entrant, and the status quo pro?t: H(c, ¯ c, 0)?
H( ¯ c, ¯ c, 0) = P(c, ¯ c, 0)x(c, ¯ c, 0) ? P( ¯ c, ¯ c, 0)x( ¯ c, ¯ c). The
main effect of competition here works through
demand: again, in markets with more homogeneous
products, consumers become more sensitive to prices,
35
Another common measure of competition is the number of com-
petitors. We consider this in an extension to the main model in
the online supplement to this paper. The main results of the model
continue to hold.
and the cost-leading venture (in the event of suc-
cessful innovation), which charges a lower equilib-
rium price than the higher-cost entrant, increases
its demand advantage over its rival. By increasing
demand for the venture when it gains a cost advan-
tage, competition increases the value of gaining such
a cost advantage.
As discussed above in the analysis of the status
quo pro?t, competition also puts downward pressure
on prices. However, this affects pro?t whether or not
the innovation takes place, and the net effect on the
value of innovation is, if not negligible, always domi-
nated by the ?rst-order demand effect just described.
Thus, overall competition unambiguously increases
the value of a cost advantage:
d|H(c, ¯ c, 0) ?H( ¯ c, ¯ c, 0)]
d0
=
dx(c, ¯ c, 0)
d0
P(c, ¯ c, 0)
+
dP(c, ¯ c, 0)
d0
x(c, ¯ c, 0) ?
dP( ¯ c, ¯ c, 0)
d0
x( ¯ c, ¯ c)
>0.
Indeed, we show in the appendix that these effects
of competition on the status quo pro?t and on the
value of innovation hold under all of demand speci-
?cations listed below.
Lemma 1. Under demand speci?cations such as the Ho-
telling line, Salop circle, Dixit–Stiglitz, logit, and Cournot-
to-Bertrand switch, competition unambiguously reduces
the status quo pro?t, dH
s
(0),d0 - 0, and unambiguously
increases the value of a cost advantage, d|H
l
(0) ?H
s
(0)],
d0 >0.
5.2. Effects of Competition on the
Financing Trade-off
Consider a product market where the degree of com-
petition is initially 0
0
and a venture with risk í
0
1
=
í
?
1
(0
0
). We know from Propositions 3 and 4 that, at
that threshold risk level, the upside of bank ?nancing
over VC ?nancing—the gain V
|
?V
.c
in continuation
surplus—is exactly offset by the downside í
?
1
|V
|
?|]
of bank ?nancing; and accordingly, the entrepreneur
is indifferent between the two types of ?nancing. In
what follows we show that competition unambigu-
ously increases the appeal of bank ?nancing over VC
?nancing: a marginal entrepreneur initially indifferent
between the two types of ?nancing will begin to strictly
prefer bank ?nancing as competition intensi?es.
Using (3) and (4), we can express the continu-
ation surplus under ?nancing of type ¡, ¡ = |, .c,
as V
¡
(0) = H
s
(0) + ¡
¡
(0)|H
l
(0) ? H
s
(0)] ? ||¡
¡
(0)]
2
,2.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
12 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
Differentiating V
|
(0) and V
.c
(0), and using the enve-
lope theorem, we obtain
dV
|
(0)
d0
=
dH
s
(0)
d0
+¡
|
(0)
d|H
l
(0) ?H
s
(0)]
d0
,
dV
.c
(0)
d0
=
dH
s
(0)
d0
+¡
.c
(0)
d|H
l
(0) ?H
s
(0)]
d0
+\
d¡
.c
d0
|H
l
(0) ?H
s
(0)].
(7)
Competition affects continuation surpluses in the
two ways described above. It reduces the status
quo pro?t H
s
(0), which has a negative impact on
V
¡
(0), and it increases the value of a cost advan-
tage H
l
(0) ?H
s
(0), which has a positive, direct impact
on V
¡
(0). This latter effect also increases entrepre-
neurial effort ¡
¡
, Which, as shown in §4, is an increas-
ing function of H
l
(0) ?H
s
(0); under VC ?nancing this
effort increase has a positive effect on the continua-
tion surplus (under bank ?nancing, this effect can be
shown to be null by the envelope theorem).
36
5.2.1. Competition and the Upside of Bank Fi-
nancing. Subtracting dV
.c
(0),d0 from dV
|
(0),d0, one
can readily establish that the upside V
|
(0) ?V
.c
(0) of
bank ?nancing over VC ?nancing is strictly increasing
in competition:
d|V
|
(0) ?V
.c
(0)]
d0
= |¡
|
(0) ?¡
.c
(0)]
d|H
l
(0) ?H
s
(0)]
d0
?
d¡
.c
d0
\|H
l
(0) ?H
s
(0)] >0. (8)
The main intuition for this result comes from the
?rst factor in (8): competition increases the value of a
cost advantage H
l
(0) ?H
s
(0), and this increase has a
larger effect on the continuation surplus under bank
?nancing because it occurs with a higher probabil-
ity (¡
|
(0) > ¡
.c
(0)). Another way to explain this is
to point out that competition increases the marginal
product of effort H
l
(0) ?H
s
(0), and that the impact
on the continuation surplus is larger under bank
?nancing, where superior entrepreneurial incentives
lead to greater equilibrium effort. The second factor,
increased entrepreneurial effort under VC ?nancing,
works the other way but can readily be shown to be
of second-order importance relative to the ?rst fac-
tor for all \ ? |\
min
, 1].
37
Indeed the upside of bank
36
As discussed above, competition increases the value of gaining a
cost advantage, and this in turn increases the entrepreneur’s incen-
tive to lower cost and her equilibrium effort level. This is similar
to the positive “escape competition effect” of competition on incen-
tives highlighted previously by Aghion et al. (2005).
37
The second factor in (8) can easily be rewritten as
\(1 ?\)¡
|
(0)d|H
l
(0) ?H
s
(0)],d0,
whereas the ?rst factor can be expressed as \¡
|
(0)d|H
l
(0) ?
H
s
(0)],d0. Clearly the latter is strictly larger than the former for all
\ ? |\
min
, 1].
?nancing over VC ?nancing is strictly increasing in 0:
d|V
|
(0) ?V
.c
(0)],d0 >0.
5.2.2. Competition and the Overall Appeal of
Bank Financing Relative to VC Financing. Let us
start by using (7) to express the effects of competition
on the downside of bank ?nancing í
?
1
(0
0
)|V
|
(0) ?|]
as follows:
í
?
1
(0
0
)
dV
|
(0)
d0
=í
?
1
(0
0
)
dH
s
(0)
d0
+¡
|
(0)
d|H
l
(0)?H
s
(0)]
d0
. (9)
On one hand, the positive effect of competition on
the value of a cost advantage H
l
(0) ?H
s
(0) has a pos-
itive impact on continuation surplus V
|
(0) and on the
downside of bank ?nancing. Most importantly, how-
ever, competition reduces the status quo pro?t H
s
(0),
which has a negative effect on the continuation sur-
plus and the downside of bank ?nancing. As shown
in the appendix, under some demand speci?cations
(e.g., Hotelling or Salop duopoly) this latter effect
strictly dominates the former for all 0 ? |0
min
, 0
max
],
and competition unambiguously reduces the down-
side of bank ?nancing. At the very least, this neg-
ative latter effect signi?cantly weakens the strength
of the former effect, ensuring that the overall impact
of competition on the downside of bank ?nancing,
regardless of its sign, is always strictly dominated by
its positive effect on the upside of bank ?nancing.
Proposition 5 then follows immediately.
Proposition 5. Competition increases the appeal of
bank ?nancing relative to VC ?nancing.
At a given degree of competition 0
0
, the marginal
entrepreneur managing a venture associated with risk
í
?
1
(0
0
), who is indifferent between the two types of
?nancing, will begin to strictly prefer bank ?nancing
over VC ?nancing if competition intensi?es. Equiv-
alently, the threshold risk level í
?
1
(0) at which the
entrepreneur remains indifferent between the two
?nancing types will increase with competition, as will
the domain of risk í
1
over which the entrepreneur
chooses bank ?nancing over VC ?nancing.
This result works primarily through the effect of
competition on the upside of bank ?nancing, and the
intuition for this important result, which comes from
two key elements of the model, is worth reiterat-
ing here. First, in allowing the entrepreneur to keep
full control of the venture in continuation and full
access to ex post pro?t, bank ?nancing yields stronger
entrepreneurial incentives and higher effort in con-
tinuation than VC ?nancing. Second, under the var-
ious demand speci?cations examined in this paper,
competition increases the value of a cost advan-
tage, and the impact of this increase on the con-
tinuation surplus is greater under bank ?nancing,
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 13
where the probability of gaining such a cost advan-
tage (i.e., entrepreneurial effort) is greater. In that
way competition increases the upside of bank ?nanc-
ing (d|V
|
(0) ?V
.c
(0)],d0>0) and the overall appeal of
bank ?nancing over VC ?nancing.
Note that the conditions of Lemma 1—that com-
petition reduces the status quo pro?t (dH
s
(0),d0
- 0) and increases the value of a cost advantage
(d|H
l
(0) ?H
s
(0)],d0 > 0)—are suf?cient to generate
Proposition 5.
38
This key result is thus robust to any
demand speci?cation satisfying these two conditions.
6. Empirical Implications
Our model yields several empirical predictions, which
we describe and discuss below.
Prediction 1. Absent negative shocks in the short
term, bank ?nancing provides stronger entrepreneurial
incentives and superior effort and performance. In the
case of negative short-term shock, it is VC ?nancing that
provides stronger entrepreneurial incentives and superior
effort and performance.
This prediction is a corollary to Propositions 1 and 2
and highlights a simple trade-off between entrepre-
neurial incentives and inef?cient liquidation affecting
the choice between bank ?nancing and VC ?nancing.
It has three implications. First, it rests on the ideas
that stronger incentives lead to superior entrepre-
neurial effort and that in turn higher entrepreneurial
effort increases venture performance. This is indeed
consistent with the recent work of Bitler et al. (2005),
who use unique data on entrepreneurial effort and
wealth in privately held ?rms to examine the connec-
tion between entrepreneurial incentives, effort, and
venture performance. They ?nd that entrepreneurial
effort, measured by hours worked, responds posi-
tively to incentives and that it has a positive impact
on venture performance.
The second implication is that bank ?nancing yields
stronger entrepreneurial incentives following short-
term success. In our opinion this implication makes
good sense: conditional on short-term success, under
bank ?nancing the entrepreneur can repay her debt
and retain 100% of all subsequent pro?t and hence has
“?rst-best” incentives, certainly stronger incentives
than under VC ?nancing where she must inevitably
relinquish a signi?cant fraction of pro?t to the venture
capitalist.
39
38
Moreover, these conditions are not necessary for our results to
hold. We show in an extension to the main model in the online
supplement to this paper that Proposition 5 holds even under
demand speci?cations where the conditions of Lemma 1 are not
satis?ed.
39
In practice, even if the entrepreneur has not yet repaid the debt
entirely, as long as short-term success ensures a small probability
of default in the future, she essentially retains ?rst-best incentives.
The third implication is that VC ?nancing gener-
ates stronger incentives following short-term failure,
because the entrepreneur retains access to at least
some fraction of pro?t, unlike under bank ?nancing
where following default the entrepreneur loses con-
trol of the venture. It might be surprising at ?rst
sight that the venture capitalist does not liquidate or
exit the venture following short-term failure, although
evidence suggests that venture capitalists are aggres-
sive in that regard (e.g., Kaplan and Strömberg 2003).
Our goal here is to capture what we view as a com-
pelling distinction between debt contracts and equity-
type contracts,
40
namely that even a ?rm with good
long-term prospects is more likely to fail under bank
?nancing, where for instance temporary cash ?ow
shortages might be enough to trigger default and
liquidation, than under VC ?nancing, where such
inef?cient liquidations would not occur.
41
This is con-
sistent with the recent empirical work of Puri and
Zarutskie (2012), which suggests that VC-?nanced
ventures are less likely to shut down, and have
lower pro?t when they do shut down, than similar
non-VC-backed ventures.
Prediction 2. Relatively risky ventures should be ?-
nanced with venture capital, and relatively safe ventures
should be ?nanced by banks.
This prediction is a corollary to Proposition 4 and
can be explained easily: ventures that are more sen-
sitive to exogenous shocks are more likely to have to
default on bank debt and face liquidation, even when
cash ?ow problems are temporary. This raises the
expected cost of bank ?nancing relative to VC ?nanc-
ing, which is immune to such inef?cient liquidation.
Thus, the model yields a prediction that is consistent
with, and an intuitive explanation for, the ?rst key
empirical result discussed in §2, namely that riskier
ventures are more likely to be VC-?nanced than bank-
?nanced.
Note that in the main model we have assumed
that there are no negative shocks in the second
40
Recall that in our model short-term failure says nothing about
the intrinsic quality of the venture; indeed both the banker and
the venture capitalist know that loss of entrepreneurial control
in the second period is inef?cient. Although VC ?nancing’s equity-
type contract avoids the problem, bank ?nancing’s debt con-
tract unavoidably leads to loss of entrepreneurial control and to
inef?ciency.
41
One could add a third state of the world at the end of period 1
where the venture would be revealed to be a total failure with no
future prospects. In that case liquidation/exit would occur under
both bank ?nancing and VC ?nancing. Even in this richer environ-
ment, our prediction would still hold: incentives following failure
would remain (weakly) superior under bank ?nancing. Thus, we
feel that the complications associated with such a change would
not be offset by additional insights and choose to focus on a model
with only two states.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
14 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
period: í
2
=0. Suppose we now assume that negative
shocks í
1
and í
2
can occur in periods 1 and 2, respec-
tively: in period 1 a pro?t H
1
is generated with proba-
bility (1?í
1
), and in period 2 the continuation payoff
V
i
, i = |, .c, is generated with probability (1 ? í
2
).
Moreover, suppose that these shocks are negatively
correlated: let us assume 1 ?í
2
=í
1
. Two points are
worth making here. First, the trade-off between the
two sources of ?nancing (and the effects of compe-
tition on that trade-off), remains exactly the same as
in the main model: bank ?nancing is preferred over
VC ?nancing iff |V
|
(0) ? V
.c
(0)] ? í
1
|V
|
(0) ? |] ? 0.
Second, we now have a new interpretation of í
1
, with
lower values of í
1
capturing shorter cash ?ow hori-
zons and larger values of í
1
capturing longer hori-
zons. Prediction 2 can thus be reinterpreted within
this context: bank ?nancing works well for shorter
cash ?ow horizons, and VC ?nancing works better for
longer horizons.
Prediction 3. Product market competition should in-
crease the probability that entrepreneurs choose bank
?nancing over VC ?nancing.
This prediction is a corollary to Proposition 5.
Important to this result is the idea that bank ?nanc-
ing yields stronger entrepreneurial incentives than
VC ?nancing. Then product market competition, by
increasing the value of stronger entrepreneurial incen-
tives, increases the bene?t of bank ?nancing rela-
tive to VC ?nancing. With this prediction, the model
provides an intuitive explanation for the second key
empirical result discussed in §2, namely the apparent
positive correlation between product market competi-
tion and the preponderance of bank ?nancing relative
to VC ?nancing.
Note that in our model, lack of competition is
closely associated with highly differentiated products
and a relatively low value of cost reduction; in con-
trast, intense competition is de?ned by highly sub-
stitutable products and a relatively high value of
cost reduction. Another corollary follows immediately
from Proposition 5.
Prediction 4. Firms that focus on product innovation
(i.e., they develop new, highly differentiated products) may
prefer VC ?nancing, and ?rms that focus on process inno-
vation (i.e., they produce less novel products albeit at a
lower cost) may prefer bank ?nancing.
To the extent that ?rms tend to rely more on VC
?nancing early in the life cycle of their product and
on bank ?nancing later in the life cycle (see, e.g.,
Berger and Udell 1998), this prediction is consistent
with Klepper (1996), who ?nds that ?rms devote
an increasing share of their R&D to process innova-
tion rather than product innovation over the product
life cycle.
A related implication concerns the distinction be-
tween radical innovations, which create new products
or completely change technologies, and incremental
innovations, which are minor modi?cations of exist-
ing products or processes. Radical innovations might
give a ?rm a ?rst-mover advantage or a domi-
nant position with little competition (Lieberman and
Montgomery 1988), as opposed to incremental inno-
vations that still compete with other technologies.
In a study of the ?nancing and R&D decisions of UK
public ?rms between 1990 and 2002, Casson et al.
(2008) ?nd that radical innovation is usually associ-
ated with equity ?nancing, whereas incremental inno-
vation is associated with debt ?nancing. This ?nding
is consistent with Atanassov et al. (2007) who examine
the interaction between capital structure and patent-
ing activity and ?nd a positive correlation between
equity ?nancing and more novel (drastic) patents.
42
Note that this distinction between radical and incre-
mental innovation is also related to Prediction 2,
since radical innovation has a higher level of risk
than incremental innovation, due to the complex-
ity associated with new product requirements and
to the added uncertainty regarding technology, mar-
ket needs, and the actions of competitors (Song and
Montoya-Weiss 1998).
Finally, noting the one-to-one correspondence be-
tween bank ?nancing and debt-type contracts on one
hand, and VC ?nancing and equity-type contracts on
the other, our model suggests that competition could
have an impact not only on the choice between bank
?nancing and VC ?nancing in entrepreneurial ven-
tures, but also on the ?nancial structure of more estab-
lished corporations.
Prediction 5. Product market competition should in-
crease the probability that managers choose debt ?nancing
over equity ?nancing.
This prediction is consistent with ?ndings in empir-
ical corporate ?nance literature. Titman and Wessels
(1988), for example, ?nd that debt levels are nega-
tively related to product uniqueness, which is a good
measure of the degree of horizontal differentiation, or
lack of competition 0 as used in our model. Using
U.S. newspaper industry data, Schargrodsky (2002)
?nds that oligopolies have higher debt ratios than
monopolies. Lord and McIntyre (2003) provide evi-
dence that leverage increases with import competition
in the textile and apparel industry. Similarly, Baggs
42
See also Titman and Wessels (1988), who ?nd a positive correla-
tion between equity ?nancing and the level of R&D expenditures;
Kortum and Lerner (2000), who show that ?rms that receive VC
?nancing innovate more; Hellmann and Puri (2000), who ?nd that
?rms pursuing an innovator rather than an imitator strategy are
more likely to obtain VC ?nancing and that obtaining VC is asso-
ciated with the ability to secure a ?rst-mover advantage.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 15
and Brander (2006) use Canada—United States Free
Trade Agreement data to show that increased compe-
tition, measured as an exogenous reduction in import
tariffs, increases leverage. More recently, Beiner et al.
(2011) ?nd signi?cantly higher leverage ratios for
?rms operating in more competitive environments.
43
To the extent that competition can be adequately
captured by market concentration, Prediction 5 sug-
gests that managers should be less likely to choose
debt over equity ?nancing in more concentrated mar-
kets. Since market concentration tends to be asso-
ciated with higher pro?tability (Azzam 1997, Slade
2004), this is indeed consistent with the negative
correlation between leverage and pro?tability docu-
mented in much previous empirical work (Titman
and Wessels 1988, Rajan and Zingales 1995, Fama and
French 2002).
In this paper we offer novel insights into the work-
ings of the connections among product market com-
petition, venture risk, and entrepreneurial ?nance,
and we propose several new, empirically testable pre-
dictions. We hope that our model will instigate further
empirical interest and can serve as a theoretical basis
for new empirical work, in what we view as a fruitful
avenue for future research.
7. Conclusion
Empirical evidence suggests possible connections
among venture risk, product market competition, and
the ?nancing of start-ups. Because of a lack of the-
oretical work on the subject, however, the microe-
conomic foundations necessary for more systematic
empirical analysis of these possible interactions were
missing. In this paper we ?ll this gap. We propose
a general theory of entrepreneurial ?nance in imper-
fectly competitive environments that (i) can be used
to explain these connections, (ii) suggests that venture
risk and product market characteristics may be impor-
tant factors to take into account in empirical work on
entrepreneurial ?nance, and (iii) yields new empirical
predictions that we hope will spur further empirical
work on the subject.
Our model highlights a simple trade-off between
the two types of ?nancing whereby bank ?nancing
yields stronger incentives and greater entrepreneurial
effort than VC ?nancing but is more sensitive to neg-
ative shocks affecting the venture, potentially lead-
ing to inef?cient liquidation. Two main results, both
consistent with available empirical evidence, emerge
from this simple trade-off. First, riskier ventures,
which are more sensitive to negative shocks, ought
43
See also Fosu (2013), who shows that the bene?ts of leverage are
improved by product market competition, measured by the Boone
indicator (using panel data consisting of 257 South African ?rms
over the period from 1998 to 2009).
to be VC-?nanced, whereas for safer ventures bank
?nancing is optimal. Second, product market compe-
tition, by increasing the marginal product of entrepre-
neurial effort, raises the advantage of superior effort
associated with bank ?nancing, thus increasing the
appeal of bank ?nancing relative to VC ?nancing.
In order to provide clear insights and tractable
results, we propose a highly stylized model of ?nanc-
ing and competition. In the online supplement to this
paper (available on the authors’ websites), we address
robustness concerns by proposing a number of exten-
sions to the main model, including allowing for com-
petition in the ?rst period, for effort by the venture
capitalist, for veri?able pro?ts, for staged ?nancing,
for mixed strategies in contracting, for entrants with
either high or low marginal cost, and for product mar-
ket competition between two entrepreneur–?nancier
pairs. We ?nd that the main results of our model
continue to hold in a variety of environments. Yet
there is still more interesting work to be done on the
interactions among competition, venture risk, and the
?nancing of new ventures. For example, it would be
interesting to consider the bank/VC ?nancing trade-
off in an in?nite-horizon setting rather than in the
simpler two-period setting we chose here; that is an
extension beyond the scope of this paper but one we
look forward to tackling in future work.
Acknowledgments
The authors thank Ingela Alger, Patrick Bolton, Gilles
Chemla, Jean Dermine, Landis Gabel, Bob Gibbons, Denis
Gromb, Thomas Hellmann, Martin Holmén, Rich Mathews,
Joel Peress, Antoine Renucci, Joel Shapiro, Lars Stole,
Tim Van Zandt, and Jano Zabojnik, as well as semi-
nar participants at Carleton University, Copenhagen Busi-
ness School, European University Institute, Hitotsubashi
University, INSEAD, Queen’s University, Université Paris-
Dauphine, University of Melbourne, University of Ottawa,
University of Tokyo, the International Industrial Organi-
zation Conference 2009 (Boston), the European Financial
Management Association Meetings 2009 (Milan), and the
Conference on Entrepreneurship and Finance 2013 (Lund,
Sweden) for helpful comments. Jing Liang and Huafang Liu
provided excellent research assistance. Jean-Etienne de Bet-
tignies gratefully acknowledges ?nancial support from the
Social Sciences and Humanities Research Council [Grant
435-2013-1863].
Appendix
A.1. Proofs of Propositions 1, 2, 3, and 4
These proofs follow directly from the text.
A.2. Proof of Lemma 1
A.2.1. Example 1: Hotelling Model. The two start-ups
are located at each end of a Hotelling (1929) line and com-
pete in both periods. Venture 1 is located at x =0 and
venture 2 is at x =1. There is a unique consumer who is
uniformly distributed along the Hotelling line. Located at ç,
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
16 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
the consumer incurs a transport cost |ç for traveling to
?rm 1, and a cost |(1 ?ç) for visiting ?rm 2. In each period,
the consumer enjoys conditional indirect utility V
1
= j ?
p
1
?|ç from product 1 and V
2
=j ?p
2
?|(1 ?ç), where j
represents income, and chooses the product that gives the
highest utility. In this paper we measure the degree of com-
petition as the degree of substitutability between the two
competing products. Conveniently, on a Hotelling line, the
transport cost | captures the degree of product differentia-
tion, so we use the degree of substitutability between prod-
ucts, 0 =1,|, as a measure of the toughness of competition,
to use Sutton’s (1992, p. 9) terminology.
The equilibrium pro?t of ?rm i is H
i
(c
i
,c
¡
,0) =(1,(20))·
(1+(0(c
¡
?c
i
)),3)
2
. The pro?ts for the different outcomes
are H
s
(0) =1,(20) and H
l
(0) =1,(20)(1+0( ¯ c?c),3)
2
. This
yields H
l
(0)?H
s
(0) =(( ¯ c?c),6)(2+0( ¯ c?c),3). It is straight-
forward that dH
s
(0),d0=?1,(20
2
) -0, and d(H
l
(0)?
H
s
(0)),d0=( ¯ c?c)
2
,18>0.
A.2.2. Example 2: Salop Model. The market is repre-
sented by a unit circumference. There is a mass 1 of consu-
mers who are uniformly distributed along the circumference
of the circle on which n ?rms are located symmetrically.
Firm 1 produces at constant marginal cost c
i
. Locations in
the market are indexed by z ? |0, 1). Each consumer buys
either one unit of output or no output at all. A consumer
located at point d buying a unit of output from ?rm i
derives utility u(d, i) =.?p
i
?|D(d, i), where . is the com-
mon valuation of output, p
i
is the price that ?rm i charges,
D(z, i) is the shortest arc length separating consumer d
from ?rm i, and | ?0 is the transport cost per unit of dis-
tance. The demand for ?rm i is given by D
i
(p
i
, p
i?1
, p
i+1
) =
1,n?p
i
,| +(p
i?1
+p
i+1
),(2|). Pro?ts are maximized for p
i
=
|,(2n) + c
i
,2 + (p
i?1
+ p
i+1
),4, which yields to the unique
Bertrand equilibrium price p
?
i
(c
i
, c
¡
) = |,n +
n?1
|=0
|
|
c
i?|
,
where |
|
= |(2 +
?
3)
|
+(2 +
?
3)
n?|
],|
?
3((2 +
?
3)
n
?1)].
44
When the venture’s n ? 1 competitors have a cost ¯ c (c
¡
=
¯ c ?¡ = 1), the equilibrium price is p
?
i
(c
i
) = |,n + c
i
+ ( ¯ c ?
c
i
)(1 ? |
0
), where |
0
= |1 + (2 +
?
3)
n
],|
?
3((2 +
?
3)
n
? 1)].
Therefore the pro?ts for the different outcomes are H
s
(0) =
1,(0n
2
) and H
l
(0) = (1,0)(1,n + 0( ¯ c ? c)(1 ? |
0
))
2
, where
0 =1,| measures the degree of competition in the market.
This yields H
l
(0) ?H
s
(0) =( ¯ c ? c)(1 ?|
0
)(2,n+0( ¯ c ? c)(1 ?
|
0
)). It is straightforward that dH
s
(0),d0 = ?1,(0
2
n
2
) - 0,
and d(H
l
(0) ?H
s
(0)),d0 =( ¯ c ? c)
2
(1 ?|
0
)
2
>0.
A.2.3. Example 3: Constant Elasticity of Substition
(CES) (Dixit–Stiglitz) Model. Assume that the two start-
ups face demand of the form p
i
(q
i
, q
¡
) = (1,q
1?0
i
),(q
0
i
+q
0
¡
)
with 0 - 0 - 1. This demand function is derived from a
CES utility function u(q
1
, q
2
) = (q
0
1
+ q
0
2
)
1,0
, where 0 mea-
sures the degree of substitutability between the goods of
?rms 1 and 2. Firm i has a marginal cost c
i
. In this
example, competition increases as 0 increases so goods
become closer substitutes. The pro?t of ?rm i can be writ-
ten as H
i
(c
i
, c
¡
, 0) =(1 +(1 ?0)(c
i
,c
¡
)
0
),((1 +(c
i
,c
¡
)
0
)
2
). The
pro?ts for the different outcomes are H
s
(0) = (2 ?0),4,
and H
l
(0) = (1 + (1 ? 0)c
0
),(1 +c
0
)
2
, where c = c,¯ c -1.
It is straightforward that dH
s
(0),d0 = ?1,4 -0. More-
over, d(H
l
(0) ?H
s
(0)),d0 = (1,(4(1 +c
0
)
3
))(?4c
0
(lnc)(1 +
0 +(1 ?0)c
0
) +(1 +c
0
)(1 ?c
0
)
2
) >0.
44
See Lin and Wu (2013) for the proof.
A.2.4. Example 4: Logit Model. Consider a market
with N consumers and two ?rms. Each consumer in the
market purchases one unit of just one of the two products
that the ?rms offer. A consumer chosen at random from the
population of N consumers has a conditional indirect util-
ity function V
i
=A ?p
i
+a
i
for i =1, 2. The parameter A
stands for consumers’ income, p
i
is the price of good i =
1, 2, and a
i
represents consumers’ idiosyncratic tastes about
good i (horizontal differentiation). The purchase probabil-
ity for product i is the probability that a randomly selected
consumer derives the largest utility from purchasing it. In
the absence of an outside (no-purchase) option, the expected
demand of ?rm i is given by the following logit formula
(see Anderson et al. 1992) d
i
(p) =exp(?p
i
,j),(exp(?p
1
,j)+
exp(?p
2
,j)). If j = 0, goods are perfect substitutes, but if
j = , goods are independent. So competition increases
as j decreases. Therefore we use 0 = 1,j as a mea-
sure of the toughness of competition. Firms compete in
price, which leads to equilibrium prices such that p
i
=
c
i
+1,0 +(1,0)(exp(?0p
i
),exp(?0p
¡
)). Pro?ts for the differ-
ent outcomes are the following: H
s
(0) = N,0 and H
l
(0) =
(N,0)(exp(?0p
?
1
),exp(?0p
?
2
)), where p
?
1
and p
?
2
are the equi-
librium prices when c
1
= c and c
2
= ¯ c.
45
It follows directly
that dH
s
(0),d0 -0.
In what follows we show that d(H
l
(0) ? H
s
(0)),d0 >0.
Let X = exp(?0p
?
1
),exp(?0p
?
2
). Since ¯ c > c, p
?
1
- p
?
2
so
X >1. Moreover, lnX =0(p
?
2
?p
?
1
) =0( ¯ c ? c) +1,X ?X. By
differentiating this equality with respect to 0, we obtain
(dX,d0)(X
2
+X +1),X
2
= ¯ c ? c. We have H
l
(0) ?H
s
(0) =
N((X ?1),0), so (1,N)d(H
l
(0) ?H
s
(0)),d0 =?(X ?1),0
2
+
(1,0)(dX,d0) = ?(X ?1),0
2
+X
2
(lnX +X ?1,X),(0
2
(X
2
+
X +1)) = (X
2
lnX ? X +1),(0
2
(X
2
+ X +1)) > 0 ? X > 1.
Therefore, d(H
l
(0) ?H
s
(0)),d0 >0.
A.2.5. Example 5: Switch from Cournot to Bertrand.
Let us analyze a shift from a Cournot duopoly to a Bertrand
duopoly, where Bertrand competition is often seen as more
competitive than Cournot competition. Assume that the two
ventures face a demand of the form P =u ?Q.
Under a Cournot duopoly, the equilibrium pro?t of ven-
ture i is H
i
(c
i
, c
¡
) =(u?2c
i
+c
¡
)
2
,9. The pro?ts for the differ-
ent outcomes are H
s
C
=(u ? ¯ c)
2
,9 and H
l
C
=(u ?2c + ¯ c)
2
,9.
Under a Bertrand duopoly, the equilibrium pro?t of ven-
ture i is H
i
(c
i
, c
¡
) =0 if p
i
>p
¡
and H
i
(c
i
, c
¡
) =(p
i
?c
i
)(u?p
i
)
if p
i
-p
¡
. The pro?ts for the different outcomes are H
s
8
=0
and H
l
8
=( ¯ c ? c)(u ? ¯ c).
As competition increases, we switch from Cournot com-
petition to Bertrand competition. Since H
s
C
> H
s
8
, H
s
de-
creases. Moreover, H
l
C
? H
s
C
= (4,9)( ¯ c ? c)(u ? ¯ c) whereas
H
l
8
?H
s
8
=( ¯ c ? c)(u ? ¯ c), so H
l
?H
s
increases.
A.3. Proof of Proposition 5
Let us de?ne A ? H
l
(0) ?H
s
(0), A
? d(H
l
(0) ?H
s
(0)),d0,
and H
s
?dH
s
(0),d0. We know from the main text that ¡
|
=
A,| and ¡
.c
=(1 ?\)A,|, which immediately implies ¡
|
?
¡
.c
=\A,| and d¡
.c
,d0 =(1 ?\)A
,|.
45
More speci?cally, these equilibrium prices are p
?
1
= c + 1,0 +
(1,0)(exp(?0p
?
1
),exp(?0p
?
2
)) and p
?
2
= ¯ c + 1,0 + (1,0)(exp(?0p
?
2
),
exp(?0p
?
1
)).
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 17
Using expression (8), and substituting the relevant ex-
pressions above, we can express the impact of competition
on the upside of bank ?nancing as follows:
d|V
|
(0) ?V
.c
(0)]
d0
= (¡
|
?¡
.c
)A
?\
d¡
.c
d0
A
=
\
2
|
AA
>0. (10)
For a given degree of competition 0, the associated
threshold venture risk level í
?
1
(0) can be expressed as
í
?
1
(0) = 1 ?
V
.c
(0) ?|
V
|
(0) ?|
=1 ?
H
s
+¡
.c
A?(|,2)¡
2
.c
?|
H
s
+¡
|
A?(|,2)¡
2
|
?|
=
(\
2
,(2|))A
2
H
s
+(1,(2|))A
2
?|
. (11)
Since we know from (7) that dV
|
,d0 =H
s
+¡
|
A
=H
s
+
AA
,|, we can express the downside of bank ?nancing as
dV
|
d0
í
?
1
(0) =
H
s
+
AA
|
(\
2
,(2|))A
2
H
s
+(1,(2|))A
2
?|
. (12)
Subtracting (12) from (10) and simplifying yields
d|V
|
(0) ?V
.c
(0)]
d0
?
dV
|
d0
í
?
1
(0) =¹|2(H
s
?|)A
?AH
s
],
with
¹=
(\
2
,(2|))A
2
H
s
+(1,(2|))A
2
?|
>0. (13)
Clearly, the results of Lemma 1, namely A
> 0 and
H
s
-0, are suf?cient to ensure a strictly positive impact
of competition on the appeal of bank ?nancing relative to
VC ?nancing.
A.4. Relaxing the Regularity Condition (1)
Throughout the paper we have imposed a regularity con-
dition (1) on the model: í ? \
min
H
s
(0
max
). As discussed
in §4, this condition is suf?cient to ensure that for all 0 ?
|0
min
, 0
max
], (a) VC ?nancing is feasible for all \ ? |\
min
, 1];
(b) there exists an í
max
(0) = 1 ? (í ? |),(V
|
(0) ? |) > 0
such that bank ?nancing is feasible iff í
1
? |0, í
max
(0)]; and
(c) there exists a threshold level of venture risk í
?
1
(0) =1 ?
(V
.c
(0) ?|),(V
|
(0) ?|) - í
max
(0) such that bank ?nancing
is optimal for all í
1
? í
?
1
(0), and VC ?nancing is optimal
otherwise. We now relax this condition.
Recall that VC ?nancing is feasible iff í ?\H
2
(¡
.c
(0), 0),
and that bank ?nancing is feasible iff í ?(1?í
1
)V
|
(0)+í
1
|.
Note that for any \ ? |\
min
, 1] and 0 ? |0
min
, 0
max
], we have
\H
2
(¡
.c
(0), 0) -V
.c
(0) -V
|
(0). (14)
Thus, for given values of í , \, and 0, there are three cases
to examine.
Case 1. í >V
|
(0).
Bank ?nancing is not feasible: there exists no venture risk
í
1
? 0 such that í ? (1 ? í
1
)V
|
(0) + í
1
|. VC ?nancing is
not feasible either, since (14) implies that if í > V
|
(0) then
í >\H
2
(¡
.c
(0), 0). Hence the venture is not ?nanced in the
?rst place.
Case 2. \H
2
(¡
.c
(0), 0) -í ?V
|
(0).
Bank ?nancing is feasible: there exists a í
max
(0) = 1 ?
(í ?|),(V
|
(0) ?|) ? 0 such that bank ?nancing is feasible
iff í
1
? |0, í
max
(0)]. On the other hand, \H
2
(¡
.c
(0), 0) - í
implies that VC ?nancing is not feasible in that case. Thus,
ventures with risk í
1
?í
max
(0) are bank ?nanced, and ven-
tures with risk í
1
>í
max
(0) are not ?nanced.
Case 3. í ?\H
2
(¡
.c
(0), 0) -V
|
(0).
Bank ?nancing is feasible for all ventures with í
1
?
|0, í
max
(0)] and í
max
(0) = 1 ? (í ? |),(V
|
(0) ? |) > 0.
VC ?nancing is also feasible since í ?\H
2
(¡
.c
(0), 0). More-
over, note that since í ?\H
2
(¡
.c
(0), 0) and \H
2
(¡
.c
(0), 0) -
V
.c
(0) (see expression (14)), we must have í -V
.c
(0), which
in turn implies í
?
1
(0) =1?(V
.c
(0)?|),(V
|
(0)?|) -í
max
(0).
Thus the results in Case 3 are the same as in the main
model. When both bank ?nancing and VC ?nancing are
available and c faces a “real”
46
?nancing choice, (1) for
low-risk ventures (í
1
? í
?
1
(0)), bank ?nancing is optimal,
whereas for high risk ventures (í
1
>í
?
1
(0)) VC ?nancing is
optimal; (2) the relative appeal of bank ?nancing increases
with competition (oí
?
1
,o0 >0).
Note that bank ?nancing is strictly more feasible than
VC ?nancing in our Model, a feature consistent with the
empirical evidence discussed in §2, which suggests that
banks are a much more frequent capital provider than ven-
ture capitalists.
This feature comes in part from our assumption that ]
has full bargaining power in renegotiation. This assumption
enables c to commit to a relatively large debt repayment at
date 1, making bank ?nancing more feasible at date 0. If
we allow c to have some bargaining power in renegotiation,
then the more bargaining power she has in renegotiation,
the lower her ability to commit to debt repayments, and the
lower her ability to borrow from a bank in the ?rst place.
Alternatively, the venture capitalist may be able to extract
rents in the ?rst period as well as in the second period (in
the model we assume that the venture capitalist extracts
rents in the second period only). In both of these cases,
the feasibility of VC ?nancing relative to bank ?nancing
would increase, and an additional (fourth) case may emerge
whereby for some values of í , \, and 0, only VC ?nancing
may be available to the entrepreneur. While one cannot rule
out the possibility that this would occur at higher values
of 0, one cannot tell a priori for which values of 0 this would
be more likely to occur. In any case, the main results of the
model, which pertain to the scenario where both sources of
?nancing are available, as described in Case 3 above, con-
tinue to hold.
References
Abowd J, Lemieux T (1993) The effects of product market compe-
tition on collective bargaining agreements: The case of foreign
competition in Canada. Quart. J. Econom. 108:983–1014.
Aghion P, Bloom N, Blundell R, Grif?th R, Howitt P (2005) Com-
petition and innovation: An inverted-U relationship. Quart. J.
Econom. 120:701–728.
46
In contrast to Cases 1 and 2 where c faces either one choice or no
choice at all.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
18 Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS
Atanassov J, Nanda VK, Seru A (2007) Finance and innovation: The
case of publicly traded ?rms. Working Paper 970, Ross School
of Business, University of Michigan, Ann Arbor.
Azzam AM (1997) Measuring market power and cost ef?ciency
effects of industrial concentration. J. Indust. Econom. 45:
377–385.
Baggs J, Bettignies (de) J-E (2007) Product market competition and
agency costs. J. Indust. Econom. 55:289–323.
Baggs J, Brander J (2006) Trade liberalization, pro?tability, and
?nancial leverage. J. Internat. Bus. Stud. 37:196–211.
Bascha A, Walz U (2001) Convertible securities and optimal exit
decisions in venture capital ?nance. J. Corporate Finance 7:
285–306.
Beiner S, Schmid M, Wanzenried G (2011) Product market compe-
tition, managerial incentives and ?rm valuation. Eur. Financial
Management 17:331–366.
Berger A, Schaek K (2011) Small and medium-sized enterprises,
bank relationship strength, and the use of venture capital.
J. Money Credit Banking 43:461–490.
Berger A, Udell G (1998) The economics of small business ?nance:
The roles of private equity and debt markets in the ?nancial
growth cycle. J. Banking Finance 22:613–673.
Berglöf E (1994) A control theory of venture capital ?nance. J. Law,
Econom. Organ. 10:247–267.
Bertschek I (1995) Product and process innovation as a response
to increasing imports and foreign direct investment. J. Indust.
Econom. 43:341–357.
Bettignies (de) J-E (2008) Financing the entrepreneurial venture.
Management Sci. 54:151–166.
Bitler M, Moskowitz T, Vissing-Jorgensen A (2005) Testing agency
theory with entrepreneur effort and wealth. J. Finance 60:
539–576.
Bloom N, Draca M, Van Reenen J (2011) Trade induced technical
change? The impact of chinese imports on IT and productivity.
Working paper, Stanford University, Stanford, CA.
Blundell R, Grif?th R, Van Reenen J (1999) Market share, mar-
ket value, and innovation in a panel of british manufacturing
?rms. Rev. Econom. Stud. 66:529–554.
Bolton P, Scharfstein D (1990) A theory of predation based on
agency problems in ?nancial contracting. Amer. Econom. Rev.
80:94–106.
Bolton P, Scharfstein D (1996) Optimal debt structure with multiple
creditors. J. Political Econom. 104:1–25.
Boot A, Gopalan R, Thakor A (2006) The entrepreneur’s choice
between private and public ownership. J. Finance 61:803–836.
Brander J, Lewis T (1986) Oligopoly and ?nancial structure: The
limited liability effect. Amer. Econom. Rev. 80:93–106.
Burgess S, Metcalfe P (2000) Incentive pay and product market
competition. CMPO Working Paper 00/28, University of Bris-
tol, Bristol, United Kingdom.
Casson PD, Martin R, Nisar TM (2008) The ?nancing decisions of
innovative ?rms. Res. Internat. Bus. Finance 22:208–221.
Caves RE, Barton D (1990) Ef?ciency in U.S. Manufacturing Industries
(MIT Press, Cambridge, MA).
Chemmanur T, Chen Z (2014) Venture capitalists versus angels:
The dynamics of private ?rm ?nancing contracts. Rev.Corporate
Finance Stud.. 3:39–86.
Chemmanur T, He J (2011) IPO waves, product market competition,
and the going public decision: Theory and evidence. J. Financial
Econom. 101:382–412.
Cosh A, Cumming D, Hughes A (2009) Outside entrepreneurial
capital. Econom. J. 119:1494–1533.
Cuñat V, Guadalupe M (2005) How does product market com-
petition shape incentive contracts? J. Eur. Econom. Assoc. 3:
1058–1082.
Dixit AK, Stiglitz J (1977) Monopolistic competition and optimal
product diversity. Amer. Econom. Rev. 67:297–308.
Dybvig P, Wang Y (2003) Debt and equity. Working paper, Wash-
ington University in St. Louis, St. Louis.
Fama EF, French KR (2002) Testing tradeoff and pecking order pre-
dictions about dividends and debt. Rev. Financial Stud. 15:1–33.
Faure-Grimaud A (2000) Product market competition and opti-
mal debt contracts: The limited liability effect revisited. Eur.
Econom. Rev. 44:1823–1840.
Fluck S (1998) Optimal ?nancial contracting: Debt versus outside
equity. Rev. Financial Stud. 11:383–418.
Fosu S (2013) Capital structure, product market competition and
?rm performance: Evidence from South Africa. Quart. Rev.
Econom. Finance 53:140–151.
Fraser S (2005) Finance for small and medium-sized enterprises:
A report on the 2004 UK survey of SME ?nances. Working
paper, University of Warwick, Coventry, United Kingdom.
Fulghieri P, Sevilir M (2009) Size and focus of a venture capitalist’s
portfolio. Rev. Financial Stud. 22:4643–4680.
Geroski P (1990) Innovation, technological opportunity, and market
structure. Oxford Econom. Papers 42:586–602.
Gertner R, Scharfstein D, Stein J (1994) Internal versus external cap-
ital markets. Quart. J. Econom. 109:1211–1230.
Glaeser EL, Kerr SP, Kerr WR (2014) Entrepreneurship and urban
growth: An empirical assessment with historical mines. Rev.
Econom. Statist., ePub ahead of print March 21,http://dx.doi
.org/10.1162/rest_a_00456.
Glazer J (1994) The strategic effects of long-term debt in imperfect
competition. J. Econom. Theory 62:428–443.
Green A, Mayes D (1991) Technical inef?ciency in manufacturing
industries. Econom. J. 101:523–538.
Grif?th R (2001) Product market competition, ef?ciency and agency
costs: An empirical analysis. IFS Working paper, Institute for
Fiscal Studies, London.
Hart O, Moore J (1989) Default and renegotiation: A dynamic model
of debt. Working paper, Massachusetts Institute of Technology,
Cambridge. (Published in Quart. J. Econom. 1998, 113:1–41.)
Hart O, Moore J (1994) A theory of debt based on the inalienability
of human capital. Quart. J. Econom. 109:841–880.
Hart O, Moore J (1995) Debt and seniority: An analysis of hard
claims in constraining management. Amer. Econom. Rev. 85:
567–585.
Harvard Business School (2006) Financing the business: Where is
the money? Harvard Business Essentials: Entrepreneur’s Toolkit
(Harvard Business School Press, Boston), 1–43.
Hashmi A (2013) Competition and innovation: The inverted-U rela-
tionship revisited. Rev. Econom. Statist. 95:1653–1668.
Haskel J (1991) Imperfect competition, work practices and produc-
tivity growth. Oxford Bull. Econom. Statist. 53:265–279.
Hellmann T, Puri M (2000) The interaction between product market
and ?nancing strategy: The role of venture capital. Rev. Finan-
cial Stud. 13:959–984.
Hisrich R, Peters M, Shepherd D, Mombourquette P (2006)
Entrepreneurship, Canadian ed. (McGraw-Hill Ryerson, Whitby,
Canada).
Hoberg G, Phillips G (2010) Product market synergies and compe-
tition in mergers and acquisitions: A text-based analysis. Rev.
Financial Stud. 23:3773–3811.
Hotelling H (1929) Stability in competition. Econom. J. 39:41–57.
Inderst R, Mueller H (2004) The effect of capital market charac-
teristics on the value of start-up ?rms. J. Financial Econom. 74:
319–356.
Inderst R, Mueller H (2009) Early-stage ?nancing and ?rm growth
in new industries. J. Financial Econom. 93:276–291.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
Bettignies and Duchêne: Product Market Competition and the Financing of New Ventures
Management Science, Articles in Advance, pp. 1–19, ©2015 INFORMS 19
Industry Canada (2009) Canada Small Business Financing Act com-
prehensive review report, 2004–2009. Last accessed Decem-
ber 18, 2014,http://www.publications.gc.ca/site/eng/367008/
publication.html.
Jianakoplos NA, Bernasek A (1998) Are women more risk-averse?
Econom. Inquiry 36:620–630.
Kaplan S, Strömberg P (2003) Financial contracting meets the real
world: An empirical analysis of venture capital contracts. Rev.
Econom. Stud. 70:281–316.
Klepper S (1996) Entry, exit, growth and innovation over the prod-
uct life cycle. Amer. Econom. Rev. 86(3):562–583.
Knowledge@Wharton (2011) When markets are hot: Opportunities
for venture capitalists across the globe. (June 24), http://
knowledge.wharton.upenn.edu/article/where-markets-are-hot
-opportunities-for-venture-capitalists-across-the-globe/.
Kortum S, Lerner J (2000) Assessing the contribution of venture
capital to innovation. RAND J. Econom. 31:674–692.
Landier A (2003) Start-up ?nancing: From banks to venture capital.
Working paper, University of Chicago, Chicago.
Leeth J, Scott J (1989) The incidence of secured debt: Evidence
from the small business community. J. Financial Quant. Anal. 24:
379–394.
Lieberman MB, Montgomery DB (1988) First mover advantages.
Strategic Management J. 9:41–58.
Lin M, Wu R (2013) Production cost heterogeneity in a circular-
city model. Working paper, Singapore Management University,
Singapore.
Lord RA, McIntyre JE Jr (2003) Leverage, imports, pro?tability,
exchange rates, and capital investments: A panel data study
of the textile and apparel industries 1974–1987. Internat. Rev.
Financial Anal. 12:287–310.
Maksimovic V (1988) Capital structure in repeated oligopolies.
RAND J. Econom. 19:389–407.
Myers S (2000) Outside equity. J. Finance 55:1005–1037.
Nickell SJ (1996) Competition and corporate performance. J. Politi-
cal Econom. 104:724–746.
Peress J (2010) Product market competition, insider trading and
stock market ef?ciency. J. Finance 65:1–43.
Poitevin M (1989) Collusion and the banking structure of a duopoly.
Canadian J. Econom. 22:263–277.
Povel P, Raith M (2004) Financial constraints and product market
competition: Ex ante vs. ex post incentives. Internat. J. Indust.
Organ. 22:917–949.
Puri M, Zarutskie R (2012) On the lifecycle dynamics of venture-
capital- and non-venture-capital-?nanced ?rms. J. Finance 67:
2247–2293.
Rajan RG, Zingales L (1995) What do we know about capital
structure? Some evidence from international data. J. Finance
50:1421–1460.
Renucci A (2014) Bargaining with venture capitalists: When
should entrepreneurs show their ?nancial muscle? Rev. Finance
18:2197–2214.
Repullo R, Suarez J (2004) Venture capital ?nance: A security design
approach. Rev. Finance 8:75–108.
Robb AM, Robinson DT (2014) The capital structure decisions of
new ?rms. Rev. Financial Stud. 27:153–179.
Sahlman WA (1990) The structure and governance of venture cap-
ital organization. J. Financial Econom. 27:473–521.
Salop SC (1979) Monopolistic competition with outside goods. Bell
J. Econom. 10:141–156.
Schargrodsky E (2002) The effect of product market competition
on capital structure: Empirical evidence from the newspaper
industry. Business School Working paper, Universidad Torcu-
ato Di Tella, Buenos Aires.
Showalter D (1995) Oligopoly and ?nancial structure: Comment.
Amer. Econom. Rev. 85:647–653.
Slade M (2004) Competing models of ?rm pro?tability. Internat. J.
Indust. Organ. 22:289–308.
Song XM, Montoya-Weiss MM (1998) Critical development activi-
ties for really new versus incremental products. J. Product Inno-
vation Management 15:124–135.
Sutton J (1992) Sunk Costs and Market Structure (MIT Press, Cam-
bridge, MA).
Syverson C (2004a) Product substitutability and productivity dis-
persion. Rev. Econom. Statist. 86:534–550.
Syverson C (2004b) Market structure and productivity: A concrete
example. J. Political Econom. 112:1181–1222.
Teshima K (2010) Import competition and innovation at the
plant level: Evidence from Mexico. Working paper, ITAM,
Mexico City.
Titman S, Wessels T (1988) The determinants of capital structure
choice. J. Finance 43:1–19.
Ueda M (2004) Banks versus venture capital: Project evaluation,
screening, and expropriation. J. Finance 59:601–621.
U.S. Small Business Administration, Of?ce of Advocacy (2014) Fre-
quently asked questions—Advocacy: The voice of small busi-
nesses in government.http://www.sba.gov/sites/default/
?les/sbfaq.pdf.
Winton A, Yerramilli V (2008) Entrepreneurial ?nance: Banks ver-
sus venture capital. J. Financial Econom. 88:51–79.
D
o
w
n
l
o
a
d
e
d
f
r
o
m
i
n
f
o
r
m
s
.
o
r
g
b
y
[
1
3
0
.
1
5
.
1
8
9
.
1
9
1
]
o
n
1
8
M
a
r
c
h
2
0
1
5
,
a
t
0
6
:
5
7
.
F
o
r
p
e
r
s
o
n
a
l
u
s
e
o
n
l
y
,
a
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
doc_770096523.pdf