Journal of Business Venturing xxx (xxxx) xxxx

Contents lists available at ScienceDirect

Journal of Business Venturing

journal homepage: www.elsevier.com/locate/jbusvent

Strategic exits in secondary venture capital markets

Guillaume Andrieua, Alexander Peter Grohb,∗
a
Montpellier Business School, 2300 avenue des Moulins, Montpellier Cedex 4 34185, France
b
EMLYON Business School, 23 Avenue Guy de Collongue, Ecully 69134, France

ARTICLE INFO ABSTRACT

Keywords: The market for secondary venture capital transactions provides a way for investors to obtain
Secondary venture capital market liquidity if the investee start-up corporation has not yet reached sufficient maturity for an IPO. It
Entrepreneurial finance also creates divestment opportunities for badly performing ventures that investors would rather
Venture capital abandon. We analyze the way in which the opportunistic behavior and liquidity constraints of
Moral hazards
venture capital funds channel deal flow into the secondary venture capital market. Such op-
portunistic behavior leads to the strategic exit of seed financing venture capitalists. These exits
JEL classification:
enlarge investors' opportunity set of strategies and therefore affect the deal terms with en-
D81
G24 trepreneurs. In this paper, we show that two contracts are possible in a world with financially
G32 constrained venture capital investors, staged investments, and premature divestment opportu-
L26 nities. Both contracts have their disadvantages. With the first, the venture capitalist will never
liquidate a project, even if it is a lemon, but will instead engage in a secondary transaction. With
the second contract, although lemons will be systematically abandoned, high-quality ventures
may also be liquidated. Entrepreneurs need to consider these effects when aiming to maximize
their benefits and must trade off the contract parameters accordingly. Our model provides gui-
dance to entrepreneurs in this respect, helps to explain deal flow into the secondary venture
capital market, and offers several empirical predictions.

1. Executive summary

The market for secondary venture capital (VC) transactions provides VC funds with an opportunity to relatively simply divest
their assets. In this market, VCs' claims in particular ventures, or fund stakes, or even complete VC portfolios are traded among
financial investors. It has substantially grown in recent years in volume and numbers. The limited partners of VC funds appreciate the
opportunity to liquidate their exposure prior to the “usual” VC holding period and the transactions allow seed funding VCs to exit
early and to stay focused on seed investments. New players emerge who specialize as “secondary” acquirers and thus develop a
market niche which is fueled by increasing capital commitments from institutional investors.
Nevertheless, the relationship between VC funds and their investees was traditionally based on a long-term tight partnership. It
was considered important for VC funds to bring ventures to maturity until they would qualify for an initial public offering or a “trade
sale” (i.e. a transaction where a VC fund sells its claim to an incumbent firm that has a strategic interest). The literature emphasizes
the importance of such exits to deliver appropriate returns to investors. Other exit channels are supposedly not to pay-off appro-
priately, relative to the typical VC investment risk. This points to the puzzle why VC funds engage in secondary transactions and sell
their claims prematurely and why others buy in the secondary market. One may assume that VC funds use this exit channel to divest
lemons because they would stay with the promising ventures and develop them to maturity. However, if this was the case then the

∗
Corresponding author.
E-mail addresses: g.andrieu@montpellier-bs.com (G. Andrieu), groh@em-lyon.com (A. Peter Groh).

https://doi.org/10.1016/j.jbusvent.2019.105999
Received 19 December 2018; Received in revised form 10 December 2019; Accepted 15 December 2019
0883-9026/ © 2019 Elsevier Inc. All rights reserved.

Please cite this article as: Guillaume Andrieu and Alexander Peter Groh, Journal of Business Venturing,
https://doi.org/10.1016/j.jbusvent.2019.105999
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

secondary market would probably already have turned into a market for lemons and dried out. An alternative explanation for the
disposal of VC assets prior to their IPO/Trade Sale maturity is the requirement to raise a new fund or a liquidity shock. A venture
capitalist could be forced to sell-off its portfolio at the end of a fund's lifetime, but especially if its investors are not satisfied with its
performance.
Literature has not yet elaborated on the secondary VC market puzzle other than descriptive or empirically addressing venture
success. Our paper is the first to theoretically focus on this market segment contributing to our understanding of the mechanisms
which drive secondary exits. A liquid secondary VC market incentivizes opportunistic behavior which we refer to as “strategic exits”
based on Faure-Grimaud and Gromb (2004): A seed financing venture capitalist can use his insider information on the quality of a
project to dispose of a lemon. Our main hypothesis is that an acquirer cannot infer why the seed financing VC fund is selling his claim
rather than bringing the venture to an initial public offering or trade sale. The buying rationale is that although the venture is a
prospective good quality, the seller has a financial constraint and is therefore required to divest.
We contribute to the entrepreneurship literature and derive important implications for practice in several ways. We prove that a
contract in which the venture capitalist makes optimal continuation decisions is impossible in a world with staged financing, a liquid
secondary VC market, and liquidity constraints. Only two contracts are possible, each of which has particular disadvantages. The first
contract channels not only promising ventures but also lemons into the secondary VC market. The second contract is a lock-up
contract that forces liquidations. It would dry out the secondary market and would instead lead to the abandonment of favorable
ventures which is certainly an unwarranted feature. The discussion of strategic exits is particularly important in the entrepreneurial
finance context where information asymmetries are large and where it is difficult for outsiders to evaluate investees. Our model
parameters correspond with important entrepreneurial finance contract terms. An entrepreneur needs to select among the two types
of contracts and trades off the terms to maximize her NPV. Our paper provides guidance in this respect. It also helps in explaining the
growing deal flow in the secondary VC market.

2. Introduction

The investors of a venture capital (VC) fund (i.e. its limited partners) are generally expected to lock in their money for about ten
years, with entrepreneurs assuming that VC funds will be long-term strategic partners. While both factors are true in principle, the
activity in the “secondary VC market” is currently rising remarkably. Both individual claims of VCs1 and even fund interests or
complete portfolios are traded in this market and thus allow to liquidate VC assets in a rather simple manner. Limited partners
appreciate this liquidity gain and even commit capital to specialized “secondary VC funds” acting as acquirers. This way, “second-
aries” develop to a sub-category within the VC asset class.2
Cumming and MacIntosh (2003) define a secondary deal as a transaction where “...only the shares of the VC are sold to the third
party; the entrepreneur and other investors will retain their investments.” The secondary market is an unquoted equity market where
transactions are quickly settled. Institutional investors and VCs both as buyers and sellers, accordingly. This characterizes a pecu-
liarity of secondary transactions: There are no strategic buyers involved. Such buyers have a long-term interest when acquiring young
ventures and usually expect synergies. They are therefore able to price-in these synergies to the benefit of a venture's current
shareholders. However, strategic buyers would also typically aim to gain majorities and to oust the entrepreneur from the firm, or
would at least reduce the entrepreneur's scope of authority, e.g. by removing her from her position as CEO of the firm she founded.
Such behavior is described in Cumming and Johan (2008b).
It is obvious that the possibility of obtaining liquidity is appealing to institutional investors of VC assets and reduces their
illiquidity premium. However, the desire to gain such liquidity contradicts the economic virtue of VC at the same time. This virtue
stems from tight and long-term financing relationships with innovative ventures. The nature of these relationships is highly illiquid. It
has further been pointed out, e.g. by Black and Gilson (1998), that VC exposure only generates appropriate returns if a venture
“reaches maturity” and can be divested via an initial offering in the public equity market (IPO) or a trade sale (i.e. selling to a
strategic buyer). Cumming and MacIntosh (2003) and Cumming et al. (2006) prove this empirically.3
Prior to the existence of a liquid secondary market, VCs needed to bring good quality ventures to maturity and to abandon lemons
in a timely fashion. With the rise of secondary market liquidity, VCs now have the option of divesting exposure prematurely, as
described in Cumming (2008) and Cumming and Johan (2008a,b). However, as VC funds aim to generate higher returns, they would
probably not engage in a secondary transaction if the investee were successful unless there is a particular reason. It is therefore
important to understand the drivers and characteristics of deal flow into the secondary VC market. In principle, there can only be two
causes for a VC to initiate a secondary transaction. The first one is that he needs to sell a claim in a promising venture because he is

1
Note: We also use the abbreviation VC to denote “venture capitalist”.
2
Secondary transactions always existed in the unquoted equity market as a channel for divesting exposure. However, the volume and number of
secondary transactions has risen substantially, as reported, e.g., in: “Secondaries deals hit record $58bn”, Financial Times, February 26, 2018, or
“Alternative exits: The rise of secondary deals in venture capital”, Pitchbook, October 26, 2018. In its 2013 and 2018 yearbooks, the European
Venture Capital Association (“InvestEurope”, formerly EVCA) reports an increase in the secondary market exit funnel from 7.2% to 15% of invested
capital.
3
There are additional VC exit routes, namely share buy-backs and write-offs. In a share buy-back, the venture and/or the entrepreneur buys back a
VC's claims. As suggested by the term “write-off”, in this type of exit, the VC abandons the transaction without expecting significant liquidation
proceeds. The exposure will thus be written-off, ultimately to zero value. As would be expected, these exit channels tend to yield low or even
negative (in case of write-offs) returns, as noted in Cumming and Johan (2008b) and Johan and Zhang (2016a).

2
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

facing a liquidity constraint. The second reason is that he is trying to dispose of a lemon, passing it off as a good quality venture. This
is possible because secondary transactions exhibit substantial information asymmetry between deal partners. Following Admati and
Pfleiderer (1994), a lemon can be sold in a secondary transaction if the insider is able to hide the true reason for his disposal.
Accordingly, acquirers in secondary transactions cannot infer the quality of the venture unless they know whether the originator is
facing a liquidity constraint or not. Regardless of the reason for a secondary transaction, the rising liquidity in this exit channel adds
to the VC's opportunity set of divestment strategies. It must therefore also affect the deal terms with the entrepreneur. To the best of
our knowledge, no paper has yet considered this new exit route in a formalized way by focusing on the potential entrepreneurial
finance contracts. This is our contribution.
We present a model based on the possibility that a seller in the secondary VC market intentionally benefits from the opacity
surrounding the true reason for his divestment. We consider an entrepreneur endowed with an innovative project who solicits outside
financing. She is irreplaceable due to her specific knowledge and will manage the venture. Two capital contributions are required to
back the investee through a seed and a development stage. At seed stage, a contract is negotiated with a VC, and immediately after
closing, the fund injects seed money into the young company. The entrepreneur determines her level of effort according to her
incentives as defined in the contract. During the seed period, the VC actively supports and monitors the venture. This encourages the
entrepreneur and thus increases the likelihood of success. It also allows the VC to gain private information about the quality of the
venture so that he can determine its state of nature, which is either good or bad. At the same time, he has private information on
whether he needs to exit the relationship. This requirement may result from fundraising pressure at the end of his fund's lifetime, even
if the venture is promising. The lifetime of a VC fund is typically limited to ten years. It is split into two stages, an investment and a
liquidation period. When the liquidation period is reached, the VC will start divesting his exposure and will need to raise a new fund
to stay in business. A fund's vintage, and therefore the beginning of its liquidation period, are known by outsiders. For this reason,
fundraising pressure is predictable and a common feature in the VC asset class. However, the duration of the venture's development
period remains unpredictable, which could shift a potential IPO or trade sale window beyond the fund's lifecycle. Additionally, it is
possible that the VC fund's other portfolio investments are not successful. Such information on fund performance is not usually
available to the public but only to the fund's investors. If current limited partners are not satisfied with the fund's performance they
might be reluctant to accept a lifetime extension or to receive the securities held by the fund, and may instead insist on its liquidation.
In this case, the VC will need to exit, either by winding up the venture, or by selling his claim to an outside investor in a secondary
transaction.
There may be several other reasons for a VC to prematurely dispose of claims, even in successful ventures. Shafi et al. (forth-
coming) comprehensively discuss, for example, the fact that investment discontinuation decisions can also be motivated by financial
constraints caused by crisis situations, shifts in industry or geographic investment focus, or a fund's focus on seed-stage transactions,
thus avoiding style drifts. The key feature with respect to our model is that all of these causes are based on private information and,
hence, outsiders cannot infer, even after due diligence, the true reason for a premature disposal via a secondary transaction. VC funds
are not required to disclose any information on potential liquidity constraints to the public. Furthermore, it is common practice not to
disclose such information for competitive reasons. Such reporting behavior of VC funds is comprehensively discussed in Johan and
Zhang (2016b). Their paper emphasizes that VC fund managers tend to be opaque even towards their own investors.
In any case, unless the venture has been wound up, it will need expansion capital. This must be provided by the acquirer if the
venture goes through a secondary market transaction. If there is no need for the seed phase VC to exit and if the development of the
venture is promising, the VC will keep his claim in the venture. Nevertheless, in this second situation, we assume that the initial VC is
restricted to seed-financing transactions and that its fund covenants and fund size do not allow additional exposure in a single
venture. Therefore, the seed financing VC cannot inject additional capital and a new outside investor needs to provide expansion
capital in a new financing round. Gompers and Lerner (1996) emphasize that such restrictions on follow-on investments and exposure
in single ventures are common in the VC industry. Cumming et al. (2009) discuss the potential severe negative consequences for VC
managers from not adhering to their funds' covenants. For the end of the development period, our model requires that the venture
reaches maturity and can be divested via an IPO or a trade sale.
In this setting, we refer to the principle of strategic exits as defined in Faure-Grimaud and Gromb (2004) and analyze how a seed
financing VC can use his insider information on the quality of a project to dispose of a claim in a bad venture on the secondary VC
market. Our main hypothesis is that the acquirer cannot infer why the seed financing VC is selling his stake rather than bringing the
venture to an IPO or trade sale. His buying rationale is that although the venture is a prospective good quality, the seller has a
financial constraint and is therefore required to liquidate certain assets. Our model yields several important results. In particular, we
show that only two contracts are possible, corresponding to two semi-pooling equilibria. The first contract supports the perception
that the VC secondary market is a market for both good quality ventures and lemons. Market participants dispose of good and bad
ventures, contingent on several parameters, as discussed in our model. It is impossible to design a contract forcing the VC to disclose
the true quality of a project in a secondary transaction. The seed financing VC hides this information and thus benefits because his
only viable choice may be to exit. We then derive the second contract, which provides incentives that rule out strategic exits. Such a
contract would force bad projects to be liquidated instead of channeling them into the secondary VC market. However, the drawback
here is the liquidation of successful projects. This is certainly an unwarranted feature and we prove that it is therefore an imperfect
solution for mitigating opportunistic insider behavior. Understanding this type of opportunistic behavior is particularly important in
the entrepreneurial finance context where information asymmetries are large and where it is difficult for outsiders to evaluate
investees.
We contribute to the entrepreneurship literature and derive important implications for practice in several ways. Our model helps
to explain deal flow in the secondary VC market. We prove that it is not possible to design a contract in which the VC makes optimal

3
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

continuation decisions in a world with staged financing, premature exit opportunities, and potential liquidity constraints. Only two
contracts are possible, each of which has particular disadvantages. The first contract channels not only promising ventures but also
lemons into the secondary VC market. The second contract is a lock-up contract that forces liquidations. It would dry out the
secondary market and would instead lead to the abandoning of favorable ventures. Our model's primitives correspond to en-
trepreneurial finance contract parameters. An entrepreneur needs to select one of the contracts and it is important for her to trade off
the contract terms to maximize her NPV. Our paper provides guidance in this respect.
The paper is organized as follows: We first review the related literature and then present the model and its assumptions. In the
subsequent section we derive the first-best situation and the optimal contracts. We then compare these contracts and discuss par-
ticular cases. Finally, we provide empirical predictions and conclude. All proofs are provided in the appendix.

3. Related literature

A comprehensive overview of VC contracting problems is provided in Cumming and Johan (2013). Our model builds on several
papers from the entrepreneurship and economics literature. For example, Schwienbacher (2013) compares “generalist” and “spe-
cialist investor” startup financing. The specialist provides better support but incurs additional costs in later financing rounds. He
investigates the mechanisms that entrepreneurs can use to trade off these two types of investors. De Bettignies and Brander (2007)
also compare the two types of seed investors. They highlight the potentially positive impact of VC support on entrepreneurial effort.
Enhanced entrepreneurial effort nonetheless comes at some cost because the entrepreneur must relinquish control and cash flow
rights. The arguments of Andrieu and Groh (2018) are closely related to those of De Bettignies and Brander (2007) and of
Schwienbacher (2013), but focus on the allocation of control rights to entrepreneurs or investors and their impact on venture
continuation decisions. However, none of the contributions consider the existence of the secondary VC market and the effect of
asymmetric information if VCs divest prematurely. The model in this paper bridges this gap. It is similar to Andrieu and Groh (2012),
who compare bank-affiliated and independent VC funds, although the VC exit in their model is triggered by an exogenous shock. Our
paper elaborates on the endogeneity of a premature exit decision and its relevance for entrepreneurs and acquirers in the secondary
VC market.
A related model is also proposed in Ueda (2004), where an entrepreneur benefits from investor support but may see her idea
expropriated. Our paper does not focus on expropriation but instead examines opportunistic investor behavior. Elitzur and Gavious
(2003) and Schwienbacher (2007) elaborate on the ways in which entrepreneurs trade off angel and VC financing. The angel and the
seed stage VC in the Elitzur and Gavious (2003) model have characteristics similar to the VCs in our model. However, the authors do
not consider the possibility of premature exits and the inevitable moral hazard. Schwienbacher (2007) focuses on an option to wait,
which can remove the need for an initial angel round. In our model, we consider seed financing to be deterministic, in order to avoid
additional complexity. Hellmann and Thiele (2015) consider a world where an entrepreneur successively solicits angel and VC
financing. They analyze the link between the angel and VC markets and examine how the bargaining power between the two types of
investors frames the sharing of rents. Renucci (2014) explicitly models bargaining power as being a consequence of the investor's
personal wealth. Cumming and Johan (2008b) empirically demonstrate the influence of bargaining power on cash flow allocation
and control rights. Our model does not elaborate on the allocation of bargaining power but is robust to alternative allocations.
The model presented in Faure-Grimaud and Gromb (2004) is close to the model we present here. They focus on the impact of an
active majority shareholder's exit on the stock price of a firm. In their model, a majority shareholder may be hit by a liquidity shock
that forces him to exit before the effect of his value-adding activity is observed by the market. In our setting, exit is not an exogenous
requirement but the result of a decision to liquidate assets. Cumming and Johan (2010) model the exit decision as a conflict between
the entrepreneur and the VC. Exit will be triggered if the VC's projected marginal cost reaches the level of the projected marginal
value added through further VC backing. This approach is related to the divestment probability in our model. The rising liquidation
pressure from a fund's limited partners may be considered to be part of the projected marginal cost. Nevertheless, our model is based
on the liquidity constraint of a VC and not on potential conflicts with the entrepreneur. Aghion et al. (2004) also consider a fra-
mework in which an investor may exit an investment before its maturity. In their model, the investor monitors the entrepreneur's
effort and coinvests with uninformed outsiders. As is the case in our model, the investor may be forced to exit prematurely, with this
necessity being only privately observed. The authors show that while it may be important to control the exit rights of an investor,
compensation is required for the resulting illiquidity. Kandel et al. (2011) study VC funds organized as limited partnerships. They
show that bad projects may be continued to conceal them from the partnerships' investors. At the same time, good projects may be
liquidated if their duration is expected to be too long. These inefficiencies are socially undesirable. In this paper, we model this
inefficiency in the particular case of ex ante uncertainty regarding the premature exit of a VC. This ex ante uncertainty has not yet
been formalized. One reason is that this was not necessary prior to the substantial rise of secondary VC market activity because
premature exits of VC exposure were difficult to achieve and therefore rather rare. Out contribution to the literature is therefore to
model the rationale for such divestments and their impact on start-up financing contracts.

4. The model

We consider a risk-neutral world with no discounting. An entrepreneur endowed with an innovative project requires outside
equity financing. As in Ueda (2004), we assume that VCs compete à la Bertrand, i.e., they are competitive and the entrepreneur has
full bargaining power due to the uniqueness of her idea. We also assume that VCs stage their capital contributions. The project
requires two investments I0 and I1 at the beginning of the seed and development periods, respectively. We consider two financiers: a

4
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

seed stage VC fund and an expansion stage VC fund. The goal of the investors is to sell their claims through an IPO or trade sale at the
end of the development period. This provides exit proceeds R.
The entrepreneur exerts effort. We consider a unit cost of effort parameter cI to assess the full cost of entrepreneurial effort. These
costs are affected by the support and monitoring quality of the seed stage VC. A higher management support quality yields a lower
cost of effort, and hence lower values for cI. The level of entrepreneurial effort is e, such that 0 ≤ e ≤ 1. The cost of effort is defined
e2
by cI 2 . The entrepreneur's effort plays an important role in our model. It determines the venture's state of nature at the end of the
seed phase, which will be either good or bad. We assume that the probability of reaching the good state is equal to e. At this interim
stage, the VC is privately informed about the project's state of nature. If the state is good, the exit proceeds R at the end of the
expansion period are certain. If the state is bad, there is still a chance of receiving R given probability q, or 0 with probability (1 − q).
The duration of the project is ex ante uncertain. It is possible that the VC realizes during the seed phase that it will take longer than
initially anticipated to bring the venture to maturity. At the same time, the seed financing VC fund may arrive at its liquidation
period. This is publicly observable, in principle, by the VC fund's vintage. However, the fund's overall investment portfolio might not
be appealing to its limited partners - and this information is not publicly disclosed. The limited partners may therefore be reluctant to
extend the fund's lifetime and may insist on its liquidation. This information is also not publicly available. Consequently, the seed VC
is the only party to be privately informed of the potential need to exit at interim state.4 We denote the probability of this happening
by p. If the seed financing VC is required to exit, he can either wind up the venture or sell his claim in a secondary transaction. If he is
not forced to exit, he has the same alternatives, plus the option of staying with the investee until its maturity. Winding up the venture
yields liquidation proceeds L and the VC has a liquidation preference LI. We assume that these proceeds do not cover the initial
investment, i.e. L < I0. If the venture is continued, it will require expansion capital. If the seed VC has not exited, then a development-
stage VC will inject expansion money in a new financing round and dilute the first VC and the entrepreneur. If the venture instead
goes through a secondary transaction, then the acquirer will inject the capital required and dilute the entrepreneur.
The timeline is presented in Fig. 1:
We firstly focus on the seed VC's opportunity set of strategies. If he is forced to exit, with probability p, he will either sell his claim
in a secondary transaction at the interim stage or will wind up the venture. If he is not forced to exit, with probability (1 − p), he can
either liquidate, sell, or continue backing the venture. All contract parameters are set ex anteand we assume that the entrepreneur has
full bargaining power. She makes a take-it-or-leave-it offer to a seed financing VC, as in Hellmann (2002). For simplicity, we assume
that the parameters of the contract with the expansion VC are also negotiated ex ante.5
We summarize the strategy set in Table 1:
If the venture goes through a secondary transaction, the model requires the acquirer to inject I1 straight afterwards and also
requires that qR > L. This implies that there is no benefit to be gained from buying the stakes of the seed VC and liquidating the
venture. Since the acquirer knows neither the true reason for the secondary transaction, nor the quality of the venture, there is no way
to set the contract parametersex ante contingent on the nature of the event.
We define the following contract parameters:6

- δ denotes the seed VC's cash flow claim, which is sold in the secondary transaction;
- δ1 and δ2 denote the cash flow claims of the seed and expansion capital VCs, respectively, if the venture is not sold in a secondary
transaction;
- PI denotes the price for the claim δ in the secondary transaction; and
- L denotes the liquidation proceeds of the venture. LI is the liquidation preference of the seed VC, the residual (L − LI) flows to the
entrepreneur.

5. Optimal contracts

We subsequently derive the first-best situation and the optimal contracts corresponding to the game. We assume that agents are
utility maximizing if the entrepreneur has full bargaining power, but address the model's robustness with respect to different bar-
gaining power allocations further below.
In the first-best equilibrium, the revenue minus the cost of all agents is maximized. The net present value of the project is:

e2
I0 + e (R I1) + (1 e) L cI .
2

To reach the first-best equilibrium, the entrepreneur should set her level of effort such that:

4
Shafi et al. (forthcoming) discuss additional rationale why a VC could be forced to divest exposure prematurely. The important feature to be in
line with our model is that the exit motivation is not visible to outsiders. This should be the case in all circumstances described in their paper and
thus provide further motivation for our model.
5
Given that the contract parameters are defined ex ante, they perfectly reflect the information set and respect the expansion VC's participation
constraint.
6
For simplicity, we assume that δ,δ1,δ2, and PI are negotiated in the initial contract between the seed stage VC and the entrepreneur. This
assumption is purely formal. The parameters are calculated according to the acquirer's information set while buying the seed VC's claim. A ne-
gotiation at an interim stage would yield exactly the same values.

5
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

Fig. 1. Timeline of the model.

Table 1
Seed VC's opportunity set of strategies.
/ Seed VC is forced to exit Seed VC is not forced to exit

Good state Secondary transaction/Liquidation Secondary transaction/Liquidation/Continuation

Bad state Secondary transaction/Liquidation Secondary transaction/Liquidation/Continuation

e2
max I0 + e (R I1) + (1 e) L cI .
e 2

The level of effort that maximizes the NPV is hence e FB = .7

1 R I L
cI
We now consider the case in which entrepreneurs maximize their utility when VCs are competitive. The seed capital VC makes a
continuation decision at the end of the seed period, contingent on the requirement to exit or not. Using backward induction
methodology to find the equilibria, we successively investigate the two alternatives (exit or not) and determine which of the two
related contracts maximizes the entrepreneur's profit, solving the optimization program.
Let us first assume that the limited partners of the seed VC force him to liquidate his fund and he therefore has to exit. He either
engages in a secondary transaction or winds up the venture. If liquidated, he faces a loss because, at best, he recovers L < I0. He
therefore sells, as long as PI > LI, and only liquidates if LI > PI.
Let us here assume PI > LI, which simply becomes true by setting LI = 0. In this way, the seed VC will always initiate a secondary
transaction.
We now assume the second case, where the seed VC is not forced to exit and therefore has three options: continuing, selling, or
liquidating. Liquidating is not appropriate, however, because PI > LI. The VC may therefore decide to sell regardless of the state of the
venture, reaching a pooling equilibrium. Or, he may decide to stay regardless of the state of the venture. Nevertheless, in the
appendix we show that this is not an optimal strategy and should therefore be avoided. The intuition is that the seed VC should not
continue if the venture is a lemon but would be better off with a secondary transaction or with a liquidation. An alternative strategy is
to continue if the venture's state is good or to sell if it is a lemon. We prove in the appendix that i) this strategy dominates the pooling
equilibrium (selling whatever the state of nature) and that: ii) the pooling equilibrium replicates the other strategy (continuing in
good states and selling in bad) if p = 1. As a result, a contract (which we denote as contract number one) is possible, where the seed
VC continues in good states if he is not forced to exit and sells via a secondary transaction in any other case.
If we now assume that LI > PI, which is simply true by setting PI = 0, then the seed VC is always better off liquidating rather than
selling. It is then also possible to find a second contract such that the VC has an incentive to liquidate. We refer to this as contract
number two.
To summarize, there are two possible situations. In the first, the project is never liquidated even if it is a lemon. The seed financing
VC always sells the venture via a secondary transaction - in a good or bad state of nature - if he is required to exit. If he is not forced to
exit, then he continues if it is successful or he initiates a secondary market transaction if it is a lemon. In the second situation, the seed
VC always liquidates the venture except in the case where he does not need to exit and where the venture is successful. Both situations
correspond to a semi-separating (or pooling) equilibrium. It is impossible to implement a contract from which the true nature of the
secondary transaction could be inferred.
In the next section, we elaborate on the contracts' optimal parameters with respect to these two situations. We then analyze the
determinants of the entrepreneur's choice between the two contracts.

5.1. Contract that induces secondary transactions

We focus on contract number one, where the VC always engages in a secondary transaction, except if the venture is successful and
he is not forced to exit. The resulting opportunity set of strategies is presented in Table 2:

7 R I1 L
It is assumed that R − I1 − L > 0 and R − I1 − L < cI. Then, 1 > cI
> 0.

6
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

Table 2
Seed VC's opportunity set of strategies with contract number one, which rules out liquidation.
/ Seed VC is forced to exit Seed VC is not forced to exit

Good state Secondary transaction Continuation

Bad state Secondary transaction Secondary transaction

We subsequently derive the contract parameters that maximize the entrepreneur's NPV. The index S (S for secondary transaction)
denotes the parameters for contract one.
Under this contract, the venture is never liquidated. It therefore only requires the following parameters: δ1 and δ2 (i.e. the cash
flow rights of both VCs if there is no secondary transaction), δ (the cash flow claim of the expansion VC if the first enters into a
secondary transaction because he is forced or decides to exit), and PS (the price for δ in the secondary transaction).
The entrepreneur maximizes her utility function:
e2
p (e (1 ) R + (1 e )(1 ) qR) + (1 p)(e (1 1 2) R + (1 e )(1 ) qR) cI
2

e2
= (1 e )(1 ) qR + pe (1 ) R + (1 p) e (1 1 2) R cI .
2
The seed VC does not liquidate if:
PS > LS .
This can by verified by simply setting LS = 0.
The seed VC continues with the venture in its good state of nature, if he is not forced to exit and iff:
1R PS .
We assume that, even if the inequality is not strict, the seed VC always prefers continuing to selling.8
He sells a lemon in a secondary transaction if he can continue backing the venture and iff:9
1 qR PS .

To motivate his investment in the venture in the first place, he requires:

[p + (1 p)(1 e )] PS + (1 p) e 1 R I0 0.
Given the seed VC's incentive constraints, he has no interest in deviating from the defined strategy, which corresponds to a Perfect
Bayesian Equilibrium. The expansion VC's participation constraint is determined by his beliefs regarding the seed VC's behavior:
(1 e )( qR I1 PS ) + ep ( R I1 PS ) + ( 2 R I1)(1 p) e 0.
He can also infer the good quality of the venture if the seed VC continues. The expansion VC also requires:
2R I1 0,
from which we deduce 2 = R1 to maximize the entrepreneurial profit. If this was not true, the expansion VC would make a loss and
I

would refuse to invest in the first place. We can therefore substitute the expansion VC's participation constraint with the following
term:
(1 e )( qR I1 PS ) + ep ( R I1 PS ) 0.

Using
I1
2 = R
, the maximization program is:
e2
max (1 e)(1 ) qR + pe (1 ) R + (1 p ) e (R 1R I1) cI ,
, e, 1, PS 2

e2
s. t . e arg max(1 e )(1 ) qR + pe (1 ) R + (1 p ) e (R 1R I1) cI ,
2

1R PS 0,

PS 1 qR > 0,

(1 e )( qR I1 PS ) + ep ( R I1 PS ) 0,

[p + (1 p)(1 e )] PS + (1 p) e 1 R I0 0.

8
We prove in the appendix that a pooling equilibrium strategy in which the VC always sells is not optimal, and therefore not possible.
9
We assume that the inequality is strict and prove in the appendix that it is the case.

7
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

The subsequent proposition defines the parameters of the optimal contract, which rules out liquidation.10 We set I = I0 + I1.
Proposition 1. The entrepreneur can raise capital under a first contract setting the following parameters:
PS = 1R = I0,

I1
2 = ,
R

cI ( s )
= ,
2(q p)2R2

R (q p)[(p q) R + (1 p)(R 2I )]
with = qR + ,
cI

2 (q p)2R2 (p q) R + (1 p)(R I)
and s = 4 I+ [I (1 p)] .
cI cI

Three conditions for the exogenous parameters need to be satisfied for the contract to be feasible:
s 0,

(p q) R + (1 p)(R I)
0or I+ [I (1 p)] 0,
cI

(q p) R (qR + I0 L) + p (qR + I ) 0.

Under this contract, we obtain an equilibrium that is partially, but not completely, separating. The expansion VC can only infer
that the venture is good quality if the seed VC stays. If the seed VC divests in a secondary transaction, the expansion VC will not
receive any information about the venture's state. The problem is that lemons are not liquidated, thus leading to inefficient con-
tinuations. The expansion VC needs to recover losses from lemons with profits from ventures in good states. The entrepreneurial
effort is lower than first-best.
Several conditions need to be met for the contract to be feasible. Revenues must be sufficiently high compared to the investment
cost. In addition, the chance q of achieving an IPO or trade sale exit, even in a bad state, needs to be low compared to the probability p
that the seed VC will be forced to exit. This prevents the entrepreneur from shirking. Finally, the unit cost of effort should be
reasonable.

5.2. Contract that forces liquidation

We focus now on the parameters of contract two, where the seed VC always liquidates unless the venture is of good quality and he
is not forced to exit. Index L (L for liquidation) denotes the parameters of this contract. To incentivize liquidation, we simply set
PL = 0, yielding PL < LL and define the liquidation proceeds allocated to the entrepreneur to be LE = L − LL. Table 3 summarizes the
seed VC's opportunity set of strategies:
Next, we derive the optimal contract parameters.
The entrepreneur's profit is defined according to the following function:
e2
LE (p + (1 p)(1 e )) + (1 p) e (1 1 2) R cI .
2
Here, the expansion VC only invests if the seed VC continues to back the venture. The expansion VC updates his beliefs accordingly
and requires:

2R I1 0
We set 2 = to maximize the entrepreneur's profit. It is also necessary to have LL > δ1qR and δ1R > LL11 to meet the incentive
I1
R
constraints of the seed VC.
The seed VC's participation constraint is:
LL (p + (1 e)(1 p)) + e (1 p) 1 R I0 0.

10
The proposition derives the parameters of the optimal contract which induces a secondary transaction. It also determines the price of the
transaction with the outsider at the interim stage. This price will be equal to the seed financing VC's initial investment. It further solves for the
percentage stake held by the outsider. Since the VC will stay with the venture if he can and if its state is good, the proposition also derives his
ownership share in this case. Finally, the proposition sets conditions for some of the exogeneous parameters for the contract to be feasible. We
discuss these conditions in the following.
11
We prove in the appendix that these inequalities are strict.

8
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

Table 3
Seed VC's opportunity set of strategies with contract two, which incentivizes liquidation.
/ Seed VC is forced to exit Seed VC is not forced to exit

Good state Liquidation Continuation

Bad state Liquidation Liquidation

The maximization program is hence the following:

e2
max (p + (1 p)(1 e ))(L LL ) + (1 p) e (1 1R I1) cI ,
, e, 1, LL 2

e2
s. t . e arg max(p + (1 p)(1 e ))(L LL) + (1 p) e (1 1R I1) cI ,
2

1R LL 0,

LL 1 qR 0,

(p + (1 p)(1 e))(LL) + (1 p) e 1 R I0 0,

The subsequent proposition defines the optimal contract parameters.12

Proposition 2. The entrepreneur may raise capital under a second contract with the following parameters:
LL = L,

R+L I1 [R + L I1 ]2 + 4
1 = ,
2R

I1
2 = .
R
A necessary condition is:

R+L I1 2 ,

(L I0) cI
with = (R I1) L .
(1 p) 2

Compared to the first contract, the second contract avoids inefficient continuations. If the venture is a lemon, it will be liquidated.
However, the serious drawback of this contract is that successful ventures will also be liquidated if the seed VC is required to exit. It is
impossible to design a contract without this shortcoming: the seed VC will prefer to hide information about the venture's state of
nature in order to reduce losses in bad states. In addition, if the seed VC is required to exit in a good state, it is impossible to deduce
the real reason for his exit decision.
With this contract, the entrepreneur will receive zero proceeds if the venture is liquidated. This boosts her effort. She will only be
compensated in good states and if the seed VC does not exit. To be feasible, this contract must exhibit certain characteristics. First,
final revenue R and the liquidation proceeds L should be high enough to compensate for the second investment I1. Second, the
recovery rate L − I0 needs to be low. In addition, the probability p that the seed financing VC will exit should not be high, because
this would increase the likelihood of liquidation losses.
This optimal contract is similar to a lock-up agreement, where there are no secondary transactions. It forces the seed VC to
continue until the project's maturity. We note that with p = 0, the contract replicates the first-best equilibrium.

6. Determinants of the entrepreneur's choice

In this section, we compare the determinants of the entrepreneur's choice between the two possible contracts and discuss par-
ticular cases. We then make empirical predictions.

12
The proposition defines the parameters of the optimal contract which forces liquidations. It derives the liquidation proceeds that will be
allocated to the VC, who will receive the full amount. Since the VC maintains his claim in the venture if he can and if its state is good, the proposition
also defines the ownership stakes of both investors. The proposition sets one condition on the exogenous parameters of the contract in order to make
the contract feasible. This condition is that the final revenue R plus the opportunity gain from continuation must be sufficiently high to compensate
for the cost involved. These costs stem directly from the expansion investment and from a function of the recovery rate, of the probability that the
VC is required to exit, and of the cost of effort.

9
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

6.1. Choice between an optimal contract inducing secondary transactions or, alternatively, forcing liquidations

The following proposition establishes the determinants of the entrepreneur's choice under the assumption that it is not possible to
know the true reason for the seed VC's exit. The entrepreneur weighs the benefits and drawbacks of the two possible contracts and
decides accordingly. Again, we denote by subscripts S and L the parameters of the first and second contract, respectively, and note
that the latter is only possible if the final revenue R and the potential liquidation proceeds L are sufficiently large compared to the
initial investment I0.
Proposition 3. The entrepreneur prefers contract two under the following conditions:13

I (1 p) 2 (R I1 L) 2 2cI (eS ) ,
(1 p)(R I1 L) (1 p)2 (R I1 L) 2 2cI (eS )
II eL cI
,
eS2
with (eS ) = eS R + (1 eS ) qR L I1 cI 2
.
Several parameters determine the entrepreneur's choice, with the most important of these being the amount of the final IPO or
trade sale proceeds R, the amount of liquidation proceeds L, and the expansion money I1. The entrepreneur needs to expect a
sufficiently high R − I1 − L for the second contract to be attractive. This amount is the wealth generated by the decision to continue,
net of its opportunity cost.
The second important parameter is the seed VC's exit likelihood. The higher the value of p, the less appealing contract two
becomes. If the seed VC is unable to continue (at probability p) the venture will always be liquidated, independent of its state of
nature. This creates a loss since L < I0. In contrast, if there is no need to exit (at probability (1 − p)), an optimal continuation decision
will be made. This means that the venture will be wound up in its bad state and continued in its good state. We note that the function
(eS ) captures the expected value obtained by systematic continuation: the higher it is, the more attractive contract one becomes. In
particular, contract two becomes less attractive if the probability q of reaching maturity for an IPO or trade sale in a bad state is high.
This is intuitive because the propensity for a successful exit increases with q even if the entrepreneur shirks.
The unit cost of effort cI also affects the trade off. These costs are specific to the project and to the entrepreneur. However, they
can be reduced by the VC's management support and monitoring activities. High costs make contract two more favorable because they
lower the entrepreneurial effort and therefore increase the likelihood of the venture becoming a failure.
To summarize, the entrepreneur needs to thoroughly consider all contract parameter combinations in order to trade off the
potential NPVs with the different contracts.

6.2. Choice when the investor can provide seed and expansion capital

If we assume that the entrepreneur finds a “generalist” investor, e.g. a bank, as in Schwienbacher (2013), who can provide seed
and expansion stage financing, then we can replicate the first-best equilibrium. If such a generalist enters into a secondary trans-
action, any outside investor will infer that the generalist is trying to sell a lemon. Since such an investor will always stay if the project
is in a good state, he cannot hide the fact that he only sells projects in bad states. A generalist investor is therefore required to wind up
bad ventures and to only continue with good ventures.
We can easily derive the contract with a generalist. It is sufficient to set parameter p = 0 in contract two, with systematic
liquidation. However, as also argued in Schwienbacher (2013), or in Andrieu and Groh (2012), a generalist investor may not be
expected to be as good as a VC fund in providing management support to a venture. In our model, selecting a generalist investor
would therefore result in a higher c parameter, i.e. a higher unit cost of effort. We subsequently compare our two aforementioned
contracts with a contract with a generalist investor.

6.2.1. Comparison with the contract that forces liquidation

Let cG be the exogenous unit cost of effort with a generalist investor.
The entrepreneur has to decide between a seed financing VC or a generalist investor. We assume that the contract that forces
liquidation is feasible and more attractive, according to the conditions of Proposition 3. The subsequent proposition defines the
determinants of the entrepreneur's choice between the two investors:
Proposition 4. The entrepreneur prefers the VC compared to a generalist if the two following conditions are true:14

I (1 − p)2(R − I1 − L)2⩾2cIφG(eG),
(1 p)(R I1 L) (1 p)2 (R I1 L) 2 2cI G (eG )
II eL cI
,

with .
eG
G (eG ) = eG (R I1 L) I0 cG 2

The contract with the VC is advantageous under several conditions. The most important is the resulting level of entrepreneurial

13
The proposition derives the two necessary conditions for the contract with liquidation to be more attractive to the entrepreneur.
14
The proposition directly defines the two necessary conditions for the contract with liquidation to be more attractive to the entrepreneur than a contract
offered by a generalist investor.

10
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

effort. A large difference in the quality of management support will favor the VC because support increases both the likelihood of
success and the project's NPV. The venture's gain if continued R − I1 − L (i.e. the final revenue net of the expansion investment and
the opportunity cost of the continuation decision) is ambiguous. The probability p that the seed financing VC will exit has a strong
impact and must be low. Otherwise the losses stemming from the liquidation of good projects where the VC is required to exit become
too high.
We note that if a generalist provided the same level of support quality as a VC, then the generalist would perfectly replicate the
first-best equilibrium and would always be the best alternative. However, if this were to be the case in practice, then VCs would most
probably not exist.

6.2.2. Comparison with the contract that induces secondary transactions

We now assume that the contract forcing liquidation is not possible or not the appropriate alternative, i.e., the conditions defined
in Proposition 3 are not verified. This results in the systematic continuation of the venture independent of its state of nature. We
compare this outcome and that of a contract with a generalist investor in the following proposition.15
Proposition 5. The entrepreneur prefers the VC compared to a generalist if the following two conditions are true:

I (R)2(1 − q)2⩾2cIφG(eG),
R (1 q) (R2 (1 q)2 2cI G (eG )
II eS cI
,

with .
eG
G (eG ) = eG (R I1 L) qR + I1 cG 2

The seed VC generates an inefficiency because he never liquidates, but instead engages in secondary transactions. Again, several
characteristics determine the entrepreneur's choice between the two investors. The first is the unit cost of effort with the generalist cG.
When the cost is lower, the generalist becomes more attractive for the entrepreneur. The unit cost of effort also influences the level of
entrepreneurial effort and, hence, the likelihood of a successful project. We emphasize again that if the support quality of the
generalist is as good as that of the VC, then the generalist will dominate, given that all other exogenous model parameters are equal.
The second important characteristic is the level of the liquidation proceeds L. Higher proceeds render the generalist more at-
tractive. Conversely, higher chances of success even in a bad state of nature q, will favor the VC because the expected pay-offs will be
higher even if the entrepreneur shirks.
To summarize, the entrepreneur will choose one of the two investors by weighing up the benefits stemming from the VC's better
support quality compared to the generalist's appropriate liquidation decisions.

6.3. Allocation of bargaining power and renegotiation

Contingent on market dynamics, venture particularities, and the characteristics of the entrepreneur and the VC, negotiation power
may switch from the entrepreneur to the VC. We therefore consider the alternative scenario where bargaining power is allocated to
the seed VC. In this case, the VC is no longer in perfect competition but recovers the project's NPV. He can then set the liquidation
preference, the purchase price in the secondary transaction, and his cash flow claim. However, a shift in bargaining power does not
change the general trade-off established previously.
If the seed VC is required to exit, he will be able to choose between liquidating or entering into a secondary transaction. When the
entrepreneur holds the bargaining power, she receives 0 liquidation proceeds to boost her effort. When the VC holds the bargaining
power, he will fix the liquidation proceeds allocation in exactly the same way to boost effort and to maximize the project's value. If
the liquidation proceeds are not too high, contract number one can be replicated: the VC will always engage in a secondary trans-
action to avoid liquidation losses, except that he will not be forced to divest and the venture will reach a good state of nature in the
seed period. If the liquidation proceeds are high he can set ex ante the contract parameters to favor liquidation, exactly as described in
contract number two. To summarize, a change in the allocation of bargaining power does not affect the model's two equilibria.
The contract partners may also ask to renegotiate at the interim stage to avoid inefficient continuation decisions. However,
renegotiations can only take place if all parties agree, i.e. if they will be better off by signing a new contract. Since the VC already has
the full liquidation preference, he cannot improve his position, which could avoid channeling a lemon into the secondary VC market
with contract one. Similarly, renegotiating contract number two would require a solution that prevents the VC abandoning good
projects if he is forced to exit. In this case, the price of the secondary transaction needs to be above the venture's liquidation proceeds.
However, this price cannot be defined contingent on the exit requirement of the VC because it is private information. Setting the
secondary market transaction price ex ante above the VC's liquidation preference would rule out any liquidation and, thus, replicate
the first contract. Renegotiation is therefore also impossible with the second contract.

6.4. Empirical predictions and discussion

Our model allows us to derive several empirical predictions. The setting reveals that only two contracts are possible with a seed

15
The proposition directly defines the two necessary conditions for the contract which induces secondary transactions to be more attractive to the
entrepreneur than a contract offered by a generalist investor.

11
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

financing VC in a world where the VC funds are exposed to certain liquidity constraints. Both contracts have specific drawbacks. The
first contract inefficiently continues bad projects while the second contract induces systematic liquidation, even of good projects. We
should therefore expect characteristic divestment behavior in function of VC funds' financial constraints. Funds that are more likely to
liquidate assets for their investors, e.g. because they are structured as limited partnerships, should have a lower propensity for
abandoning ventures and will instead engage in the secondary VC market. Conversely, investors with a large pool of resources or VC
funds structured in a different way (e.g. corporate, affiliated, quoted, open end, government, or simply large funds) are expected to be
less financially constrained and, hence, less active in the secondary VC market. They should have a higher propensity for either
abandoning transactions, even if liquidation proceeds are low, or keeping non-successful exposures on their books without taking any
action. Empirical research on secondary VC market activity could test this prediction by analyzing the impact of affiliations and
vendor fund size. It also seems important to address the role of expected liquidation values, e.g. via the ratio of tangible to total assets
as in Rajan and Zingales (1995), with respect to exit timing and the exit channels of VC transactions. Our model predicts that
transactions with low expected liquidation proceeds should be channeled into the secondary VC market more often.
In addition, our theory contributes to understanding the behavior of individual seed financing VCs. In a perfect world, an un-
constrained VC will not sell a good project at the interim stage but will develop it to maturity or liquidate it in a timely fashion if it is
bad. However, if a fund is constrained, then our model proposes that he will engage in a secondary transaction independent of the
project quality. The secondary VC market is thus fueled by financial constraints and deal flow includes both promising ventures and
lemons. Nevertheless, these constraints are not equally distributed over the lifetime of a VC fund, but instead increase with its age. We
therefore hypothesize that older vintage year funds engage in the secondary market more frequently. Additionally, we hypothesize
that a fund's performance induces secondary market activity. If fund investors are satisfied with the performance they may allow
extensions of its lifetime or the transfer of securities held by the fund. The VC is therefore not required to liquidate assets.
Accordingly, we predict that successful VC funds are less constrained in the sense of our model and therefore less active players in the
secondary VC market.
Our model also improves our understanding of the rising secondary VC market activity. Increasing competition among VC funds
and tighter monitoring by their limited partners may yield higher pressure on VC managers. We expect that this increasing pressure
incentivizes short-termism and stage specialization in the VC industry and thus spurs secondary VC transactions. Seed financing VCs
find a sufficiently liquid demand to divest exposure prematurely and to return proceeds to their limited partners. This signals deal
making and professional capabilities and also allows them to devote their attention and management capacity to seed stage trans-
actions. In addition, expansion investors get easily access to deal flow and can focus on later stage transactions. Both deal partners in
the secondary market benefit from a clear cut with respect to their funds' stage orientation and the “typical” risk profiles of their
investments. We therefore hypothesize that the rise of the secondary VC market is driven by a higher liquidity demand of institutional
investors and the resulting pressure on VC managers who search for stronger investment stage orientation and who aim to establish
“well-defined” portfolio risk profiles along the timeline from seed stage to maturity of innovative ventures.

7. Conclusion

The secondary VC market mitigates the liquidity concerns of limited partners who have an interest in turning the long-term
illiquid assets typically held by VC funds into “marketable securities”. While this desire is understandable and can be used to
discharge an illiquidity premium from the required return on VC investments, it appears to be at odds with the long-term and tight
relationships typically observed between a VC and an entrepreneurial venture. It is also questionable whether investors are able to
capture appropriate returns if exposure is divested prematurely in the secondary VC market. Nevertheless, this market is growing at a
strong pace and its liquidity adds to the opportunity set of strategies of seed phase VCs. This must affect the deal terms with
entrepreneurs and our paper aims to contribute to an understanding of these effects. We therefore model the conflict between a
potentially financially constrained VC fund and the funding needs of a start-up venture. An entrepreneur endowed with an innovative
project solicits seed and expansion capital from two respective VCs. It is possible, however, that the seed VC will be unable to stay
until the project's maturity and will need to liquidate his fund's assets. As a consequence, he either winds up the venture or sells his
claim in a secondary transaction. If sold, the related uncertainty corresponds to a moral hazard situation for the buyer because the
VC's requirement to divest is not observable by outsiders. This introduces the possibility for the insider to take advantage by making
strategic exits via secondary transactions. Our model shows that only two situations are possible in this world.
In the first situation, the project is never liquidated, even if it is a failure. The VC instead engages in a secondary deal and sells a
lemon. However, the VC will also sell a promising project if his investors ask him to liquidate the current fund. This contract therefore
explains the flow of good and bad deals into the secondary VC market.
The second situation is similar to a lock-up agreement. The venture will be systematically liquidated, even in a good state of
nature, if the VC is forced to exit. Good projects will only be continued if the seed VC is able to stay.
The two resulting optimal contracts have strengths and drawbacks and the entrepreneur will determine her contract choice by
trading off several important parameters to maximize her NPV. Our model proves that an ideal early stage investor would be able to
provide high quality management support and would have sufficient capacity to provide subsequent financing. The model is re-
negotiation-proof and robust with respect to different allocations of bargaining power. We predict different characteristic divestment
behavior in the secondary market for VC funds, contingent on their fund life cycle, structure, affiliation, and fund performance. These
are interesting avenues for subsequent empirical research.

12
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

Acknowledgments

Montpellier Business School (MBS) is a founding member of the public research center Montpellier Research in Management, MRM
(EA 4557, Univ. Montpellier). G. Andrieu is a member of the LabEx Entrepreneurship (University of Montpellier, France), funded by
the French government (Labex Entreprendre, ANR-10-Labex-11-01).

Appendix A

Proof of Proposition 1. First, we assume that entrepreneurial effort is inferior to the first-best effort, checking this assumption
afterwards. This is equivalent to the assumption that spending too much effort is not a problem.
The entrepreneur sets her level of effort so that she maximizes her utility function:
e2
(1 e )(1 ) qR + pe (1 ) R + (1 p) e (1 1 2) R cI 2 .
Her level of effort is then:
(p q)(1 ) R + (1 p)(1 1 2) R
e* = .
cI
A necessary incentive constraint for the seed VC to continue backing the venture in a good state of nature, and if he has the possibility
to continue, is:
1R PS 1 qR.

We investigate which one of the two constraints is binding in order to maximize the entrepreneur's revenue (previously we supposed
PS > L).
Let us assume that δ1R > PS. The seed VC's participation constraint is the following:
PS (p + (1 p)(1 e )) + e 1 R = I0 .

Let us increase PS in ε > 0. Then, δ1 reduces in (p + (1 p)(1 e)) , and e* increases (we have assumed that e* is inferior to first-best
eR
effort). The entrepreneurial profits increase. To maximize these profits, it is necessary to set PS = δ1R. In the same way, we have
2 = R . By replacing the terms in the equation, we find PS = I0 > L. The condition 1 qR
I1
PS is impossible since δ1qR = qPS and
q < 1. The VC cannot continue when the state is bad.
By transforming the seed VC's participation constraint, we find PS I0 . As PS = δ1R, PS = I0 to maximize entrepreneurial profits.
Let us define I = I1 + I0. We check that 1 R + 2 R = I R .
We now investigate the optimal value of δ*. We know that:
(1 e )( qR I1 PS ) + ep ( R I1 PS ) = qR I e ( R (q p) I (1 p)) 0,

(p q)(1 ) R + (1 p)(R I)
e* = .
cI
The previous inequation may be rewritten as follows:
(p q)(1 ) R + (1 p)(R I)
qR I [ R (q p) I (1 p)] 0.
cI
We note that:
(p q)(1 ) R + (1 p)(R I)
[ R (q p) I (1 p)]
cI
[(p q) R + (1 p)(R I )]( I (1 p)) R (q p)((p q) R + (1 p)(R I )) + (p q) RI (1 p) 2
= + +
cI cI
(q p) 2R2
.
cI

We rewrite the previous equation as:

2 (q p)2R2 R (q p)((p q) R + (1 p)(R I )) + (p q) RI (1 p)

+ qR + +
cI cI
(p q) R + (1 p)(R I)
I+ [I (1 p)] .
cI

The characteristics of the polynomial function aδ2 + bδ + c are:

(q p) 2R2
a= 0,
cI

13
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

R (q p)((p q) R + (1 p)(R 2I ))
b = qR + ,
cI

(p q) R + (1 p)(R I)
c= I+ [I (1 p)].
cI
A first necessary condition for the existence of δ is:

s = b2 4ac 0.
If c⩾0, the condition is always true.
The two roots corresponding to the polynomial function are b ± s . The lowest is b s
.
2a 2a
We must find a solution δ such that 0 1. δ should be as low as possible to maximize entrepreneurial profits. a > 0, it is
therefore impossible that δ* = 0 (otherwise the expansion VC would make a loss). It is not possible that
b
* = min(0; max(0, 2a
s
)) .
b+
The only possible solution is * = 2a
s
.
A necessary condition is:
b s b 0or 4ac 0 b 0orc 0.

Proof of initial assumptions:

(p q)(1 ) R + (1 p)(R I)
First, we check that e* = is inferior to the first-best effort e FB =
R I1 L
cI cI
. This is true if:
e* e FB (p q)(1 ) R + (1 p)(R I) R I1 L (p q)(1 ) R + (1 p)(R I) R I1 L

q (1 )R pR + R I0 + pI0 + pI1 R L q (1 )R pR + (L I0) + pI 0.

As δR − I ≥ 0 (incentive constraint), this condition is always true.
Second, we check that the seed VC cannot decide to sell when he has to exit, and always continues (e.g., in bad states), i.e.
1 qR PS .
P
To maximize the entrepreneur's profits, we would have δ1qR = PS, meaning that 1 = qRS .
The seed VC makes zero profit on average. His participation constraint would be:
pPS + (1 p)(e 1 R + (1 e ) 1 qR) I0 0

I0
PS e (1 p)
.
q
+ (1 e ) + ep
e (1 p)
The latter condition is binding in order to maximize the entrepreneur's profits. Given that q + (1 e) + ep > 1 1 q > 0,
PS < I0.
In this case, the price would be below the price with the contract defined in Proposition 1 and the condition is strict.
Third, we check that the pooling equilibrium strategy (selling whatever happens) is not dominant.
To derive the optimal contract parameters with this equilibrium, we simply need to set p = 1 and calculate the parameters
according to Proposition 1.
In this case, if the seed VC sells even if he could continue and the venture's state is good, then he obtains PS = I0. In good states, if
he stayed he would have an equilibrium δ1R = I0. Allocating ε > 0 to him creates an incentive to stay.
The entrepreneur is better off if the seed VC continues backing the venture. Suppose the levels of effort are the same in both
situations. The difference in NPV between the final contract and the pooling equilibrium contract in the two situations would be
(1 − e)p(L − qR + I1). It is positive since qR > L. Since the entrepreneur's effort is lower than first-best, it is boosted by a higher NPV
and the entrepreneur rejects the pooling equilibrium contract.
Proof of Proposition 2. We investigate the contract under which the seed VC liquidates the venture. A necessary condition for the
viability of the contract is LL L , and the cash flow rights of VCs should be such that: δ1 + δ2⩽1 and δ⩽1. We check if it is verified in
a subsequent step.
e2
The entrepreneur's profit is: LE (p + (1 p)(1 e )) + (1 p) e (1 1 2) R cI 2 .
The entrepreneur would set her effort such that:
(1 p)(LE ) + (1 p)(1 1 2) R (1 p)[(1 1 2) R LE ]
e* = = .
cI cI
A necessary condition for e*⩾0 is LE (1 1 2) R .
The expansion VC's participation constraint is:
2R I1 0,
in which we set 2 = R1 to maximize the entrepreneur's profits.
I

Other conditions are LL 1 qR and 1 R LL . We assume here that even if the inequalities are strict, the seed VC always prefers

14
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

liquidating or staying. We then prove that they are strict.

The seed VC's participation constraint is:
LL (p + (1 e)(1 p)) + e (1 p) 1 R I0 0.
Let us assume that δ1R = LL. This is equivalent to allocating the maximal amount of liquidation proceeds to the seed VC. It would
result in the fact that the seed VC's participation constraint cannot be satisfied since this would imply LL − I0⩾0 (which contradicts
one initial assumption). We then have δ1R > LL.
Consequently, the only constraints relative to LL are: LL L and LL 1 qR .
Let us assume LL < L. Let us increase LL in ε > 0. Then, in the seed VC's participation constraint, it can be seen that δ1 diminishes
in (p + (1 e)(1 p)) . In addition, e* increases (i.e. decreasing with δ1). We now assume that the entrepreneurial effort is lower than first
e (1 p) R
best (this assumption needs to be checked afterwards). The entrepreneur's profit therefore increases. Consequently, LL = L and
LE = 0. We also verify here our initial condition LL L .
We will subsequently check that L > δ1qR.
The seed VC's participation constraint now becomes:
L (p + (1 e )(1 p)) + e (1 p) 1 R I0 0.
(1 p)(R 1R I1)
As e* = cI
, we have:
L I0 + e (1 p)( L + 1 R) 0,

(1 p) 2 (R 1R I1)( L + 1 R)
L I0 + 0,
cI

2 R2 (1 p)2 (1 p)2LR + R (1 p ) 2 (R I1) (1 p ) 2 (R I1) L

1 + 1 +L I0 0
cI cI cI

2 (L I0) cI
1 [ R2] + 1 [LR + R (R I1)] + (R I1) L 0.
(1 p)2
The characteristics of the polynomial function are:
a= R2 < 0,

b = LR + R (R I1) > 0,

(L I0) cI
c= = (R I1) L 0 < 0.
(1 p)2
The discriminant is Δ = b2 − 4ac. It is positive iff: b2⩾4ac, otherwise the function is negative and the contract is not feasible. We set it
as a condition of the contract.
This condition can be rewritten as b 2 R2 R + L I1 2 .
The first polynomial's root is 2a > 0 and the second one is 2a > 0 , which is the lowest.
b b+

To maximize her profits, the entrepreneur will set:

b+ [LR + R (R I1)] + [LR + R (R I1)]2 + 4R2

1 = =
2a 2R2

LR + R (R I1) [LR + R (R I1)]2 + 4

1 = ,
2R
with = (1 p0 )2I
(L I )c
(R I1) L .
Proof of initial assumptions:
q (R I1) L
We first check that δ1qR < LL = L. One sufficient condition would be 2
< 0 , which is true since qR − I1 − L < 0 by
assumption.
We then check that 1 R + 2 R R . This is equivalent to:

R I1 + L [LR + R (R I1)]2 + 4
R I1 0
2

R I1 L+ [LR + R (R I1)]2 + 4
0.
2
The latter is always true since R − I1 − L⩾0.
Last, we check that e* is inferior to first-best effort:
(1 p)(R 1R I1) R I1 L
e* =
cI cI

15
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

(1 p)(R 1R I1) R I1 L

( p)(R 1R I1) 1R +L 0.
This is always true since δ1R > L.
Proof of Proposition 3. We compare the global NPV obtained with the two contracts. We denote by S the parameters of the first
contract inducing secondary transactions and by L the parameters of the second contract forcing liquidations. The contract with
liquidation generates higher NPV iff:

eL2 eS2
pL + (1 p)(1 eL) L + eL (1 p)(R I1) I0 cI eS R + (1 eS ) qR I0 I1 cI 0
2 2

cI eS2
eL2 ( ) + eL (1 p)(R I1 L) eS R + (1 eS ) qR L I1 cI 0.
2 2

The polynomial function ax2 + bx + c is such that a < 0; b > 0 and c > 0 (one may easily note in c the NPV of the project's
systematic continuation decision).
The function is positive if the discriminant Δ is positive. A first necessary condition is:

eS2
(1 p) 2 (R I1 L) 2 2cI eS R + (1 eS ) qR L I1 cI .
2
b+ b b (1 p)(R I1 L) (1 p)[R (1 1) I1 L + LS ]
A second necessary condition is: 2a
eL 2a
. As 2a
= cI
and eL = cI
, we have:
b (1 p)[R ( 1) LS ]
eL = > 0,
2a cI
because the seed VC prefers to continue rather than to liquidate in a good state with this contract. Consequently, the condition
b
eL 2a
is always true, as 2a = c > 0 . The other condition with respect to eL is not always true.
I

Proof of Proposition 4. The liquidation contract with the seed VC is more attractive than the contract with a generalist investor iff:
eL2 eG
pL + (1 p)(1 eL) L + eL (1 p)(R I1) I0 cI eG (R I1) + (1 eG ) L I0 cG 0
2 2

cI eG
eL2 ( ) + eL (1 p)(R I1 L) eG (R I1 L) I0 cG 0.
2 2
The polynomial function is ax2 + bx + c with a < 0; b > 0.
The value of c is:
eG2
c = eG (R I1 L) I0 cG
2

(R I1 L ) 2 + (R I1 L ) (R I1 L) 2 + 4cG (L I0)
c= I0
2cG

(R I1 L ) 2 + (R I1 L ) (R I1 L)2 + 4cG (L I0) + 2cG (L I0)

4cG

(R I1 L ) 2 + (R I1 L ) (R I1 L) 2 + 4cG (L I0) I0 L
c= .
4cG 2
We note that c has an undetermined sign.
For the polynomial function to be positive, it must have a positive discriminant Δ. A first necessary condition is:
eG
(1 p) 2 (R I1 L) 2 2cI eG (R I1 L) I0 cG .
2
b+ b b (1 p)(R I1 L) (1 p)[R (1 1) I1 L + LL ]
A second necessary condition is: 2a
eL 2a
. We know that 2a
= cI
and eL = cI
. We have:
b (1 p)[R ( 1) LL ]
eL = > 0,
2a cI
because the seed VC prefers to continue rather than to liquidate in a good state if he is able to continue backing the venture with this
b
contract. Consequently, the condition eL 2a
is always true, as 2a = c > 0 . The second condition with respect to eL is not
I
always true.
To summarize, the two conditions that need to be simultaneously satisfied are:

16
G. Andrieu and A. Peter Groh Journal of Business Venturing xxx (xxxx) xxxx

eG2
I (1 p) 2 (R I1 L) 2 2cI eG (R I1 L) I0 cG 2
,
e 2
(1 p)(R I1 L) (1 p)2 (R I1 L) 2 2cI eG (R I1 L) I0 cG G
2
II eL cI
.

Proof of Proposition 5. The liquidation contract with the seed VC is more attractive than the contract with a generalist investor iff:

eL2 eG2
eL R + (1 eL) qR I0 I1 cI eG (R I1) + (1 eG ) L I0 cG 0
2 2

cI eG2
eL2 ( ) + eL (R (1 q)) + qR I1 eG (R I1 L) cG 0.
2 2

The polynomial function is ax2 + bx + c with a < 0; b > 0.

For the polynomial function to be positive, it must have a positive discriminant Δ. A first necessary condition is:

eG2
(R) 2 (1 q) 2 2cI eG (R I1 L) qR + I1 cG .
2
b R (1 q) (p q)(1 ) R + (1 p)(1 2) R
A second necessary condition is:
b+ b
2a
eC 2a
. As 2a
= cI
and eC = cI
1
, we have:
b R (p + q ) (1 p)(I0 + I1)
eC = > 0,
2a cI

as pδR − I0 − I1 > 0 (otherwise the expansion VC makes a loss if the seed VC exits and this would violate his participation
b
constraint). The condition eL 2a
is always true, as 2a = c > 0 . The other condition with respect to eL is not always true.
I

References

Admati, A.R., Pfleiderer, P., 1994. Robust financial contracting and the role of venture capitalists. J. Financ. 49 (2), 371–402.
Aghion, P., Bolton, P., Tirole, J., 2004. Exit options in corporate finance: liquidity versus incentives. Rev. Finance 8 (3), 327–353.
Andrieu, G., Groh, A.P., 2012. Entrepreneurs' financing choice between independent and bank-affiliated venture capital firms. J. Corp. Finan. 18 (5), 1143–1167.
Andrieu, G., Groh, A.P., 2018. Specialist versus generalist investors: trading off support quality, investment horizon and control rights. Eur. Econ. Rev. 101, 459–478.
Black, B., Gilson, R., 1998. Venture capital and the structure of capital markets: banks versus stock markets. J. Financ. Econ. 47 (3), 243–277.
Cumming, D.J., 2008. Contracts and exits in venture capital finance. Rev. Financ. Stud. 21 (5), 1948–1982.
Cumming, D.J., Johan, S., 2008a. Information asymmetries, agency costs and venture capital exit outcomes. Ventur. Cap. 10 (3), 197–231.
Cumming, D.J., Johan, S., 2008b. Preplanned exit strategies in venture capital. Eur. Econ. Rev. 52, 1209–1241.
Cumming, D.J., Johan, S., 2010. Venture capital investment duration. J. Small Bus. Manag. 48 (2), 228–257.
Cumming, D.J., Johan, S., 2013. Venture Capital and Private Equity Contracting: an International Perspective, second ed. Elsevier.
Cumming, D.J., MacIntosh, G.J., 2003. A cross-country comparison of full and partial venture capital exits. J. Bank. Financ. 27, 511–548.
Cumming, D.J., Fleming, G., Schwienbacher, A., 2006. Legality and venture capital exits. J. Corp. Finan. 12, 214–245.
Cumming, D.J., Fleming, G., Schwienbacher, A., 2009. Style drift in private equity. J. Bus. Financ. Acc. 36(5) (6), 645–678.
De Bettignies, J.E., Brander, J.A., 2007. Financing entrepreneurship: bank finance versus venture capital. J. Bus. Ventur. 22 (6), 808–832.
Elitzur, R., Gavious, A., 2003. Contracting, signaling, and moral hazard: a model of entrepreneurs, ‘angels,’ and venture capitalists. J. Bus. Ventur. 18 (6), 709–725.
Faure-Grimaud, A., Gromb, D., 2004. Public trading and private incentives. Rev. Financ. Stud. 17 (4), 985–1014.
Gompers, P., Lerner, J., 1996. The use of covenants: an empirical analysis of venture capital partnership agreements. J. Law Econ. 39, 463–498.
Hellmann, T., 2002. A theory of strategic venture investing. J. Financ. Econ. 64 (2), 285–314.
Hellmann, T., Thiele, V., 2015. Friends or foes? The interrelationship between angel and venture capital markets. J. Financ. Econ. 115 (3), 639–653.
Johan, S., Zhang, M., 2016a. Private equity exits in emerging markets. Emerg. Mark. Rev. 29, 133–153.
Johan, S., Zhang, M., 2016b. Reporting bias in private equity: reporting frequency, endowments, and governance. Unpublished working paper, available at SSRN:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2826839.
Kandel, E., Leshchinskii, D., Yuklea, H., 2011. VC funds: aging brings myopia. J. Financ. Quant. Anal. 46 (02), 431–457.
Rajan, R.G., Zingales, L., 1995. What do we know about capital structure? Some evidence from international data. J. Financ. 50 (5), 1421–1460.
Renucci, A., 2014. Bargaining with venture capitalists: when should entrepreneurs show their financial muscle? Eur. Finan. Rev. 18 (6), 2197–2214.
Schwienbacher, A., 2007. A theoretical analysis of optimal financing strategies for different types of capital-constrained entrepreneurs. J. Bus. Ventur. 22 (6), 753–781.
Schwienbacher, A., 2013. The entrepreneur's investor choice: the impact on later-stage firm development. J. Bus. Ventur. 28 (4), 528–545.
Shafi, K., Mohammadi, A., Johan, S., 2019. Investment ties gone awry. Acad. Manage. J. https://journals.aom.org/doi/10.5465/amj.2017.0103 in press.
Ueda, M., 2004. Banks versus venture capital: project evaluation, screening, and expropriation. J. Financ. 59 (2), 601–621.

17