Venture Capital Contracting Under Asymmetric Information

26
Venture capital contracting under asymmetric information Jerey J. Trester * Financial Institutions Center, The Wharton School, University of Pennsylvania, 3301 Steinberg Hall- Deitrich Hall, Philadelphia, PA 19103, USA Abstract A model is developed wherein entrepreneurs and venture capitalists contract under symmetric information. Asymmetric information may arise following ®rst contracting. It is shown this can lead to debt infeasibility and preferred equity usage. Control is linked to choice between common and preferred. Results are robust to multiperiod ex- tensions. Roles of convertible preferred, retained equity, and debt in IPOs are consid- er ed. An empi ri cal surv ey of ve nture capi tal ®r ms is pres ented demons tr ating preferred dominates in early ®nancing. Debt and common are used far less ± generally at later stages under lower probability of asymmetric information. These results agree with the theory's implications. Ó 1998 Elsevier Science B.V. All rights reserved. JEL classi®cation: G3; G2; D4; D8; L2 Keywords: Venture capital; Entrepreneurship; Preferred stock; Debt; Corporate ®nance 1. Introduction Venture capital is often the source of ®nancing in the crucial stages of the early development of many ®rms. Despite its critical role, a relatively small Journal of Banking & Finance 22 (1998) 675±699 * Tel.: 1 215 898 9555; fax: 1 215 573 8084. 0378-4266/98/$19.00 Ó 1998 Elsevier Science B.V. All rights reserved. PII S 0 3 7 8 - 4 2 6 6 ( 9 8 ) 0 0 0 1 3 - 2

Transcript of Venture Capital Contracting Under Asymmetric Information

Page 1: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 1/25

Venture capital contracting under asymmetric

informationJe�rey J. Trester *

Financial Institutions Center, The Wharton School, University of Pennsylvania, 3301 Steinberg Hall-

Deitrich Hall, Philadelphia, PA 19103, USA

Abstract

A model is developed wherein entrepreneurs and venture capitalists contract under

symmetric information. Asymmetric information may arise following ®rst contracting.

It is shown this can lead to debt infeasibility and preferred equity usage. Control is

linked to choice between common and preferred. Results are robust to multiperiod ex-

tensions. Roles of convertible preferred, retained equity, and debt in IPOs are consid-

ered. An empirical survey of venture capital ®rms is presented demonstrating

preferred dominates in early ®nancing. Debt and common are used far less ± generally

at later stages under lower probability of asymmetric information. These results agree

with the theory's implications. Ó 1998 Elsevier Science B.V. All rights reserved.

JEL classi®cation: G3; G2; D4; D8; L2

Keywords: Venture capital; Entrepreneurship; Preferred stock; Debt; Corporate ®nance

1. Introduction

Venture capital is often the source of ®nancing in the crucial stages of the

early development of many ®rms. Despite its critical role, a relatively small

Journal of Banking & Finance 22 (1998) 675±699

* Tel.: 1 215 898 9555; fax: 1 215 573 8084.

0378-4266/98/$19.00 Ó 1998 Elsevier Science B.V. All rights reserved.

PII  S 0 3 7 8 - 4 2 6 6 ( 9 8 ) 0 0 0 1 3 - 2

Page 2: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 2/25

amount of attention has been paid to this form of ®nancing in the academic

literature. Much of what does exist has been written only in the last few years.

Much of the theoretical literature has been concerned with the staged process

by which project information is revealed and venture capital is invested. Ques-

tions related to this area have been explored in the ®rm formation and labor eco-

nomics literature, for example Kihlstrom and La�ont (1979) and Allen (1985).

Others have explicitly tied asymmetric information to capital structure in corpo-

rate ®nance, e.g., Boot and Thakor (1993). This concept has also been a frequent

theme in the banking literature; a review may be found in Santomero (1984) or

Allen and Santomero (1997). In the venture capital area, work generally focuses

on the project continuation decision, and often on debt optimality along the

lines of Jensen and Meckling (1976). For example, Cooper and Carleton(1979) study a model in which both the entrepreneur and the venture capitalist

learn information about project quality at stages of ®rm development, and each

must make continuation decisions, contracting with debt. Gompers (1992) ties

stage timing to agency cost and the creation and value of a strategic option,

using an equity contract. Hansen (1992) derives another model of debt optimal-

ity in venture ®nance, and explores issues of conversion and stopping. Chan et

al. (1990) create a model where entrepreneur skill is revealed to a venture capi-

talist making equity investments, and that skill level determines who ultimately

controls the ®rm. Admati and P¯eiderer (1991) create a two-period asymmetric

information model in which they examine the dual role of the venture capitalistas a provider of ®nancing and a sender of project quality signals. They determine

that a constant fractional holding of equity sends an incentive compatible signal

to the markets regarding project quality. Boot and Thakor (1993) present a

model in which splitting ®rm capital structure into debt and equity increases

®rm value. The split increases the volatility of the risky security and the pro®ts

of informed investors, encouraging investors to become informed.

Discussions of the relation between venture capital and capital structure are

rare. Sahlman (1991) considers the importance of the staged structure of ven-

ture capital ®nance. He mentions the choice between preferred and common

equity, and conjectures that preferred stock may serve to shift more risk awayfrom venture capitalists and to entrepreneurs. He suggests this greater risk

might have the e�ect of ``smoking out'' poorer projects and giving the entrepre-

neur an incentive to perform well. But as discussed below, debt would seem to

perform a similar function, casting doubt on this explanation of the role of pre-

ferred stock.

Given this paper's consideration of the preferred versus debt choice under

asymmetric information, it is natural to also consider the literature on debt

contract optimality with costly state veri®cation. Example of this literature in-

clude Townsend (1979), Diamond (1984) and Gale and Hellwig (1985). In these

models a ®nancier pays a costly auditor (such as a banker) to verify the ex-poststate of the project. Debt optimality arises from the minimization of auditing

676 J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699

Page 3: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 3/25

costs under such contracts. This suggests the optimality of preferred might

arise from the impossibility of auditing and/or a market failure associated with

ine�cient liquidation. In the model presented here, there exists the possibility

of information asymmetries which cannot be lifted by costly auditing, the na-

ture of the projects being such that auditing is either prohibitively costly or im-

possible. This creates the possibility of opportunistic behavior by the

entrepreneur, a mode of behavior encouraged by the venture capitalist's fore-

closure option. This leads to ine�cient liquidation and a market failure under

debt which preferred is not subject to.

This paper expands the literature by addressing the role asymmetric infor-

mation plays in determining why venture capitalists use equity, and preferred

equity in particular, rather then debt, to ®nance entrepreneurial projects. Itis suggested that the risk of the entrepreneur observing information about

the project type before the venture capitalist may play a role in the contract

type selection, as well as the project quality uncertainty itself.

The paper develops a multiperiod framework in which optimal contract

choice is driven by considerations of monitoring di�culty in an environment

of potentially asymmetric information with moral hazard. Conditions of asym-

metric information may arise between the venture capitalist and the entrepre-

neur. That is, at various stages of ®rm development there is a positive

probability of a condition of asymmetric information. As a result, it is shown

that the foreclosure option embedded in a debt contract may actually create anincentive for the entrepreneur to ``behave opportunistically'', taking whatever

project payo�s are available and defaulting on the debt. As a result, for su�-

ciently high probabilities of asymmetric information, debt contracting may be

neither feasible nor desirable. Under these circumstances, equity contracting

may be a feasible alternative precisely because of the lack of foreclosure rights.

In particular, preferred or convertible preferred stock may be the dominating

contract, because this contract eliminates the foreclosure option while preserv-

ing some seniority in the event of bankruptcy due to current liabilities should

information turn out to be symmetric. It is argued that this asymmetric infor-

mation mechanism may be responsible for the predominance of preferred stockin venture capital contracting. That predominance is demonstrated in an em-

pirical study which extends the previous empirical literature.

Empirical inquiries into the nature of venture capital contracts have also

been limited in number. The approach of Sahlman (1991) is of the case-study

type, with emphasis on the importance of the option to abandon the project in

a state of distress, or alternatively to re-®nance. Speci®c examples analyzed are

MCI and Federal Express. Lerner (1992a) analyzes venture capital ®nancings

in two industries: biotechnology and Winchester hard drives. He ®nds that

the likelihood of later round venture capital ®nancing increases with the ``tech-

nological progress'' (i.e. patents) obtained by the ®rm, and with rising liquidityin the IPO market through which venture capitalists often ``cash out''. In a

J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699 677

Page 4: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 4/25

second paper, Lerner (1992b) examines the venture capitalist role in the IPO

process, and determines that ®rms backed by ``seasoned'' venture capitalists

are more likely to successfully ``time'' the market, going public when valuations

are highest. He also ®nds only such experienced venture capitalists consider in-

tangible assets in the IPO decision, and that they tend to retain the most ``rep-

utable'' underwriters. Nowhere, however, has there been any study which

documents the types of ®nancing seen in venture capital. In particular, the

presence of preferred stock in such ®nancings has not been discussed except

in passing. For example, Lerner (1992a) mentions in a footnote that the biotech

and disk drive ®rms he examines were ®nanced with either equity or convertible

debt, and most of the equity was preferred. No mention of the relative amounts

of preferred versus debt is made.It would appear that there exists a need for investigation of the evolution of 

venture capital contracts through the life of the ®rm, particularly to comple-

ment dynamic capital structure models like the one developed here. To con-

struct a database which demonstrates the stylized facts which this theory

implies, a survey of venture capital ®rms is conducted whose results indicate

that preferred stock is by far the most common contract type used in early-

stage ®nancings. Debt and common equity are used far less frequently. Where

they do appear it is usually in later stage ®nancings and/or in industries where

monitoring is less di�cult, and hence the probability of asymmetric informa-

tion is much lower. These results are in accord with the theoretical structureI present.

The plan of the remainder of this paper is as follows. In Section 2 the model

is developed. It is a four-period model of venture capital contracting in which

the entrepreneur and venture capitalist contract under symmetric information.

However, a condition of asymmetric information may arise subsequent to the

®rst contract. It is shown that this can lead to infeasibility of debt contracts and

the use of preferred equity contracts. Issues of control and ownership percent-

ages are linked to a similar choice between common and preferred equity. In

Section 3 issues of multiperiod extension and risk aversion are explored. In

Section 4 empirical evidence is presented which supports the theory's assertionthat preferred equity is more likely to be used in instances where conditions of 

information asymmetry are more probable. Section 5 o�ers some concluding

remarks.

2. The model

 2.1. Agents, projects and endowments

I consider a world with risk-neutral venture capitalists and entrepreneurs.The world has four periods t 0 ` t 1 ` t 2 ` t 3Y (Fig. 1). All agents seek to

678 J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699

Page 5: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 5/25

maximize their expected wealth at t3. There is free entry into the venture capital

business, which exhibits perfect competition. Each venture capitalist is en-

dowed with capital at t0. At t0 a project is owned by the entrepreneur who

has no other assets. The project may be one of three types S �L, M, or H rep-

resenting low, medium and high value projects. However, the type is not

known to the entrepreneur or the venture capitalists at t0. Only the probabili-

ties of the project being of each type, denoted P L, P M, P H ( � L � � M � � H � 1),

are known in the t0 information set (dashed line). While the entrepreneur has

no other assets, at t0 he does have an employment opportunity paying wages w1

at t1 and w3 at t3 which will constitute reservation wages.To move forward the project requires an initial infusion of capital I 0 at t0.

The only sources for this capital are the venture capitalists. The entrepreneurs

use the funds obtained in the following way. A fraction q of the initial capital I 0is used as ®xed capital qI 0 at t0 (e.g. plant and equipment) the rest is used as

operating capital. It is also assumed that in the course of operations the project

takes on current liabilities d  payable at t2 (these can be phone bills, rent, pay-

roll etc.).

If the project is funded by a venture capitalist at t0 (see below) then the ®rst

payout from the project is R1(S ) where �1

�H

�P qs 0

�d Y �1

�M

� �qs 0

�d Y

qs 0T �1�L� ` qs 0 � d . It is assumed that R1(S ) is generated by the project att1. However, while the entrepreneur observes this payo� at t1 the venture cap-

Fig. 1. The four period model: Game form. t0: venture capitalist can invest I 0; entrepreneur learns S 

at t1, venture capitalist learns S  at tl; t1: venture capitalist can foreclose under debt if  l� 1, entre-

preneur can behave opportunistically if  l� 2; t2: if no opportunism at t1, venture capitalist may

invest I 2 or foreclose if debt used; t3: ®nal payo�s if any.

J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699 679

Page 6: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 6/25

italist observes the payo� at tl where l� 1 with probability 1 A p and l� 2

with probability p. It is also assumed that the entrepreneur observes S  at t1

while the venture capitalist observes S  at tl. Thus there is a probability p of 

a state of asymmetric information existing between the venture capitalist and

the entrepreneur. Note S  is assumed here to be learned by some mechanism

other than receiving R1(S ). If  R1(H) ¹ R1(M) then R1(S ) could conceivably

serve as a signal itself, but in practice the signal is more likely the result of fac-

tors like lab tests. A signaling mechanism distinct from the t1 payo� allows for

R1(S ) to be a�ected by factors other then S , which will permit the possibility of 

default under a debt contract even in the H state. It is assumed that

 �1�L� ` �1�M�T �1�H� to permit the existence of a project type or state of 

the world L in which current liabilities d  cannot be serviced at t2. This will givevalue to contract seniority during project development, and, as will be seen,

have a role in the selection of preferred over common equity.

To continue the project until t3, when a second payo�  R3(S ) is made, it

is necessary for the project to have capital k � qs 0 available at t2; that is, the

®xed capital must be preserved and an added amount of working capital is re-

quired. As in the previous period, it is assumed current liabilities d must be paid

at t3.

The venture capitalist learns S  at tl through a costless auditing process

which cannot be completed until then. On the other hand, the entrepreneur

learns S  at t1. Therefore, if l� 2 the entrepreneur learns S  before the venturecapitalist and has operating control of the project. In this case the entrepreneur

may choose to pay himself any amount g e1T �l�� � at t1, regardless of any con-

tractual obligations at t2, and abandon the project. In practice this may be

done by legally increasing the project's ``burn rate''. That is, by t2, when the

venture capitalist learns S , the entrepreneur is no longer accessible. Alternative-

ly he may choose to stay with the project because of the payo�  C e3 which he

now knows he will receive at t3. If l� 1 then the entrepreneur and the venture

capitalist observe S  and the payo�  R1(S ) simultaneously. Then it is assumed

the venture capitalist has at his disposal the legal enforcement mechanism nec-

essary to assure compliance with any contractual obligations. The payment theventure capitalist receives at t1 is labeled C v1.

If the project is to continue until t3, the venture capitalist must make a sec-

ond-round investment I 2 where

 s 2 � k � qs 0 À � �1�� � À d À g e1 À g v2�X �1�Note that if the endowment of the venture capitalist is restricted to s 0 � k  then

the quantity �1�� � À d À g e1 À g v2 must exceed qI 0 for venture capital to be

able to fund the second stage. In the scenarios examined the entrepreneur will

not have the ability to ®nd a replacement venture capitalist between t1 and t2.

In the scenario we will consider a ``no-exit'' clause arises endogenously to pre-vent such a move by the entrepreneur, because the possibility of such an action

680 J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699

Page 7: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 7/25

would reduce the venture capitalist's expected return at t0 below the level of 

feasibility. Hence participation by the initial venture capitalist at t2 is necessary

to continue the project if added external capital is needed. In essence the infor-

mation conveyed in the valuation process that the venture capitalist performed

at t0 gives him a ``lock'' on the second-stage ®nancing option at t2. Further, it is

assumed that if the entrepreneur behaves opportunistically at t1 the project is

necessarily terminated.

 2.2. Contracts

In this study I seek to compare three types of contracts with varying embed-

ded options and legal rights: debt, preferred equity and common equity. I pro-pose to explore the conditions under which these securities, particularly

preferred equity, may be feasible and dominating among the three contracts.

It is worth noting that the stronger claim of a particular contract being optimal

over all possible contracts, including but not exclusively the three considered

here, is far more di�cult, if not impossible, to show. This is because such an

optimization must be performed over all possible project types and legal ar-

rangements between parties, and it is di�cult to know all of these. Therefore

I con®ne my attention to the three contracts stated above, and assume that

their widespread usage is some demonstration of the likely superiority of the

group.I de®ne the rights associated with the three contracts as follows: Under a

debt contract the venture capitalist loans the entrepreneur I 0 and receives a

®xed payment g v2 � g dv2 at t2 if there is no default. A covenant exists allowing

the venture capitalist to foreclose at t1 if under symmetric information (l� 1)

he learns the entrepreneur will default at t2. Thus, here the right of foreclosure

is the right of the venture capitalist, under symmetric information, to seize the

entire t1 payo� in any state other than the best (H) state. The venture capitalist

then takes whatever assets are available less the current liabilities, which are as-

sumed for the moment to be senior to all contracts between the venture capi-

talist and the entrepreneur. If information is asymmetric (l� 2) theentrepreneur may choose to ``behave opportunistically'', taking all the assets

available. On the other hand, he may choose to remain committed to the pro-

 ject through t3. Should the project be continued until t3, a second debt contract

is created in which the venture capitalist loans the entrepreneur I 2 and receives

total payment (for both debt contracts) g v3 � g dv3 at t 3. Note since informa-

tion is always symmetric at t3, the venture capitalists right of foreclosure is en-

forceable with certainty at that time.

A preferred equity contract is essentially equivalent to a debt contract but

without the right of foreclosure. The venture capitalist loans the entrepreneur

I 0 and receives a ®xed payment g v2 � g pv2 at t2 if there is no default. In theevent of default under symmetric information, the venture capitalist may not

J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699 681

Page 8: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 8/25

foreclose, but the entrepreneur may not take any payment for himself unless

the amount due the venture capitalist is paid in full. Under asymmetric

information the entrepreneur may or may not choose to behave opportunisti-

cally as in the debt contract. Should the project be continued until t3, the ven-

ture capitalist invests I 2 and is to receive total payment (for both contracts) C pv3

at t3. Note that although preferred has no foreclosure right, because its divi-

dend must be paid before the entrepreneur if information is symmetric, under

the always symmetric information of  t3 the preferred stock payo� structure is

identical to that of debt.

Lastly, in a common equity contract the venture capitalist pays the entrepre-

neur I 0 and receives a fraction of equity f 0 at t0. It is assumed that control of 

the ®rm rests with whoever owns more then Z % of the common equity. With-out loss of generality, it can be assumed that � � 12. If the venture capitalist and

the entrepreneur each own 12

of common stock the ®rm is jointly controlled so

that both parties must agree on payout decisions. In the case where the entre-

preneur does not behave opportunistically the payment to the venture capitalist

at t2 is g v2 � g ev2P 0. If the entrepreneur does behave opportunistically

then C v2 � 0. Should the project be continued until t3 the venture capitalist in-

vests I 2 and, if this second contract is an equity contract, receives additional

equity ownership f 2 at t2. The payment he receives at t3 is again labeled

g v3 � g ev3.

It is assumed there exists free entry and perfect competition among the ven-ture capitalists, and that the entrepreneur circulates his business plan among all

of them. Further, a zero risk-free rate is assumed. Under risk neutrality this has

the e�ect of driving the venture capitalist's expected return down to zero. That

is, if  W v0 and W v3 are the wealth of the venture capitalist at t0 and t3, respec-

tively, then the t0 expectation of the return at t3 is E0W v3 A W v0 � 0.

This expected value of the venture capitalist's return at t3 may be calculated

as follows: for the debt contract

E0�  v3 À �  v0 �  � t0 À s 0 � �� �LYMYH

 � �� ��min�g d

v2Y max� �1�� �@Àd Y 0���1 À p� � � t 1 min�g dv2Y max� �1�� � À d Y 0��p�

�� t 2� t 1 � t 1�À s 2 � min�g dv3Y max� �3�� � À d Y 0���AY �2�

where � t 0 � 1 if the venture capitalist invests at t0 and zero otherwise; if l� 2

and the entrepreneur behaves opportunistically then � t 1 � 0, otherwise Z t1 � 1;

X t2

�1 if the venture capitalist does not foreclose at t1 and invests at t2 and zero

otherwise; Z t2 � 1 if the entrepreneur continues with the project at t2 and zerootherwise, and again s 2 � k � qs 0 À � �1�� � À d À g e1 À g dv2�X

682 J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699

Page 9: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 9/25

Under a preferred equity contract,

E0�  v3 À �  v0 �� � t 0 À s 0 ��

� �LYMYH

 � �� ��g v2�1 À p� � � t 1g v2p�@

�� t 2� t 1 � t 2�À s 2 � min�g pv3Y max� �3�� � À d Y 0���AY �3�

where C v2 T C pv2, and

if  l � 2 then g e1T �1�� � À � t 1�d � g v2�Yif  l � 1 then g e1T �1�� � À �d � g v2�X

Here � t 0Y � 

t 1Y � 

t 2and � 

t 2are dependent on S and each other in chronological

order. Note again that under a preferred contract with symmetric information

at t1 the entrepreneur can pay g v2Y 0 ` g v2Tg pv2Y so long as C e1 � 0 if 

C v1 ` Cpv1. This is to say that the entrepreneur can choose to pay the venture

capitalist any amount between zero and the preferred coupon so long as

he pays himself nothing if the full coupon is not met. Of course if  l� 1

and the project is not continued, (Z t1 and/or X t2 equal zero) then

g v2 � min�g pv2Y max� �2�� � À d Y 0� and therefore the entrepreneur receives

g e2 � max� �2�� � À d À g pv2Y 0�. That is, after debt service, the venture capitalist

receives all funds up to the preferred coupon, and the entrepreneur receives

whatever remains or zero, whichever is larger.In the case of common equity, the form of E0W v3 A W v0 is dependent upon

the equity ownership percentages f 0 and f 2. For the moment I will consider the

case where f 0T12; I do not wish to treat buyouts at this point. It may be that

entrepreneurial control is necessary in the ®rst stage for development to be suc-

cessful, or it may be impossible to construct an equity contract with f 0 b 12

which gives the entrepreneur an expected return greater than the sum of his res-

ervation wages w1 + w3 (these variables are exogenous so they can always be

set so that this is the case). Further, there may be some non-pecuniary bene®t

to the freedom of control which the entrepreneur seeks, alternatively this may

be factored in as a higher set of reservation wages.Further for both parties to get paid at t3 it may be necessary for each to have

an equity ownership interest equal to 12

(this restriction will be lifted in a sub-

sequent analysis of a multiperiod extension with an IPO cash-out).

If the above assumptions hold then under a two period common equity con-

tract the venture capitalist's expected t3 return is

E0�  v3 À �  v0 �  � t 0 À s 0 � � t 1�

� �LYMYH

 � �� ���g ev2 � � t 2� t 2�@

À s 2 � 12

max� �3�� � À d Y 0����AY �4�

J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699 683

Page 10: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 10/25

where g ev2T �1�� � À d À g e1Yg e1T �1�� � À d  if  l� 2 and g e1T �1�� � if  l� 1.

Of course if  l�

1 and the entrepreneur chooses to behave opportunistically,

then g e1T �1�� � and C ev1 � 0.

 2.3. Feasible and dominant contracting 

For a given type of contract to be feasible here means it is possible to de-

sign a contract of the type that the relation E0W v3 A W v0 � 0 holds. I am

now in a position to compare the feasibility of the above contracts under var-

ious projects as parameterized by � s 0Y s 2Y k Y qY d Y pY � �� �Y �1�� �Y �3�� �Y l � 1Y 2Y

� � LY MY H�. In particular, I analyze the parameter subspace in which the pre-

ferred contract is feasible while the common stock and debt contracts are infea-sible.

What will be illustrated here is the following point. States may exist in which

the venture capitalist would prefer to foreclose as opposed to continuing the

project, but under asymmetric information such foreclosure will not be possible

because the entrepreneur will anticipate the foreclosure and act opportunisti-

cally. Without the foreclosure option the entrepreneur might choose to contin-

ue the project if the venture capitalist makes another investment. For the

venture capitalist, this might be preferable to getting nothing, which is what

he receives under asymmetric information with debt and its foreclosure option.

If the probability of a condition of asymmetric information is su�ciently high,the foreclosure option's existence reduces the value of the debt contract to the

venture capitalist, making it inferior to preferred equity contract and indeed

not even feasible.

To further illuminate this point, consider the following two propositions.

Proposition 1. For any set of parameter values I 0, I 2, k, q, d, P(S), R1(S),

R3(S); l� 1, 2; S �L, M, H there exists some value of p � pÃY 0TpT 1, above

which the debt contract's option to foreclose is worthless.

Proof. First note that if p� 1 the foreclosure option is always worthless to thelender, because the entrepreneur has the R1(S ) payo� in his possession at t1

with certainty and may behave opportunistically if debt service is not possible,

making the foreclosure right unenforceable ± e�ectively the entrepreneur now

has a put on the project with a strike price of zero. Also note that if p� 0 then

information is symmetric and the entrepreneur is certain not to be able to

behave opportunistically, making the foreclosure option always enforceable

and hence of positive value. Therefore there must exist pÃY 0TpÃT 1, such that

the foreclosure option is worthless.

Corollary 1. If it can be shown that at p� 1 the existence of the foreclosure

option is actually of negative value to the venture capitalist, then 0Tpà ` 1, the

684 J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699

Page 11: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 11/25

critical asymmetric information probability is strictly less than 1, by the

intermediate value theorem.

Proposition 2. For p� 1, there exist projects for which the existence of the debt

 foreclosure option makes a debt contract infeasible, while a preferred contract is

 feasible. In such cases the foreclosure option is of negative value.

The proof of this is given in Appendix A. Essentially, a project is construct-

ed so that a preferred equity contract is just feasible, only because it receives

some payo� in the M state. Then since the debt contract receives no payo� 

in the M or L states, and can receive no more payo� than the preferred con-tract in the H state due to the entrepreneur's reservation wage, the debt con-

tract is not feasible. It should be noted that since the contract is just feasible,

an endogenous ``no exit'' clause arises at t2; otherwise the opportunity to fully

participate in the project's upside state H could be lost to the venture capitalist,

and the relation E0W v3 A W v0 � 0 could not be satis®ed.

The point here is that there can exist states where under symmetric informa-

tion the venture capitalist would prefer to foreclose, but if information is asym-

metric the existence of the foreclosure option causes the entrepreneur to behave

opportunistically, leaving the venture capitalist with nothing. It may also be the

case that in such states the venture capitalist would prefer to make another in-vestment than receive nothing, and the entrepreneur would rather continue

with the project than behave opportunistically, given that he will not be fore-

closed on. Then the foreclosure option's existence will, under asymmetric infor-

mation, drive the entrepreneur to behave opportunistically when he would

rather continue the project. This leaves the venture capitalist with nothing

when he would rather have made another investment. When the probability

of such states under asymmetric information is high enough, the foreclosure

option actually serves to reduce the expected return to the venture capitalist

to the point where it is below that of a preferred equity contract. Indeed under

such conditions a debt contract may not be feasible or desirable, while a pre-ferred contract may o�er an acceptable rate of return.

What has been demonstrated here is that the existence of the foreclosure op-

tion embedded in the debt contracts can create a condition of intertemporal in-

centive incompatibility under asymmetric information. This may result in the

entrepreneur taking actions which reduce the expected return to the venture

capitalist to the point that debt is no longer a feasible form of contracting.

Note that this is in contrast to Sahlman's ``smoking out'' or downside limita-

tion suggestions for the role of preferred versus common equity. In the case of 

the preferred versus debt choice the decision is driven by a state of asymmetric

information which may occur subsequent to the time of contracting, ratherthan before.

J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699 685

Page 12: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 12/25

It is interesting to look at the above result in the context of the costly state

veri®cation/debt optimality literature. In models like those of Townsend

(1979), Diamond (1984) and Gale and Hellwig (1985), the ®nancier can verify

the ex-post project state through a costly auditing process. Debt optimality

arises from the minimization of the auditing cost. In the model presented here,

it is assumed that there is a possibility of an information asymmetry which can-

not be eliminated through auditing. We may think of such auditing as either

prohibitively expensive or impossible. This possibility makes venture capital

projects distinct from those funded by banks and the bond market, and drives

the contracting choice toward preferred. It is a severe di�culty in monitoring

intrinsic to the sort of projects ®nanced through venture capital, a point ex-

plored in the following empirical section. Without auditing, the possibility of asymmetric information and the foreclosure option of debt gives rise to the po-

tential for opportunistic behavior on the part of the entrepreneur. Indeed, the

foreclosure option incentivises the entrepreneur to such behavior in a debt con-

tract. The result is ine�cient liquidation and a ``market failure'' under debt.

With no foreclosure option, the preferred contract is not a�icted with this in-

tertemporal incentive incompatibility.

In the context of this model, consideration of the choice of preferred equity

versus common equity is necessarily somewhat qualitative, with some features

beyond the scope of this paper. However, the forms of Eqs. (3) and (4) certain-

ly suggest a role for downside protection, in that as current liabilities d  rise itwill take a greater fraction f 0 of the ®rm's equity to compensate the venture

capitalist for giving up his priority position in a bankruptcy liquidation. This

would be necessary as noted above: because in the L state under symmetric in-

formation liquidation is certain, and therefore opportunistic behavior is certain

under asymmetric information. Either way the venture capitalist is left with

nothing here since under opportunistic behavior the entrepreneur takes the en-

tire payo�.

It may be that the amount of equity required by the venture capitalist to

compensate him for this zero payo� in the L state leaves the entrepreneur with

a fractional interest such that his expected return is less than the sum of his res-ervation wages w2 + w3. It may also be the case that there is some non-pecu-

niary bene®t to the entrepreneur from having common equity fractional

ownership above some critical point, notably the control level 12, which results

in a substantially higher ``e�ective'' reservation wage. Finally entrepreneurial

control may be necessary to facilitate the advancement of the project through

the development process. Any of these factors or some combination of them

may be responsible for a common stock contract's infeasibility for a given pro-

 ject and, in conjunction with the argument against debt given above, the choice

of a preferred equity contract becomes the dominating one.

A convertible preferred contract will allow the venture capitalist to partici-pate in any cash-out via an IPO. Therefore under the above framework there is

686 J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699

Page 13: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 13/25

no reason why the venture capitalist need give up the downside protection of 

the preferred for common (nor the entrepreneur give up necessary control).

3. Extensions of the model

In this section two extensions of the basic model are considered to investi-

gate the robustness of its results. In the ®rst extension, the basic framework

is extended to N  periods and in the second the introduction of risk aversion

is considered. Let us examine each of these in turn.

The above model has been formulated in a four period setting. However, it

is relatively straightforward to extend the model into more periods.

Proposition 3. The above model may be extended to N periods while preserving 

subgame perfection.

Proof. Replace the t3 terminal payo�s R3(S ), S �L, M, H, with games of the

same structure as the t0 ± t3 game presented above whose expected payo�s are

equal to those previously terminal values. This self-similar extension procedure

may be repeated for an arbitrary n iterations to produce a 4n period model. So

long as the assumption of risk neutrality hold the agent will view the new

extensions exactly as they viewed the original t3 terminal payo�s (by the law of iterated expectations). Then the solution discussed in Proposition 1 will be a

sub-game perfect Nash equilibrium.

What does alter the analysis somewhat is the introduction of risk aversion

on the part of the agents. Such curvature of preferences may give rise to some

mixing of contracts, which is to say the use of some debt in addition to pre-

ferred or common equity. A way to see the intuition behind this point is to con-

sider the above treatment in the larger context of standard (von Neuman± 

Morgenstern) expected utility theory. The venture capitalist's preferences con-

sidered above correspond to a risk-neutral utility function of consumption U with � H b 0 and � HH � 0; in this case U �C . Under risk neutrality only the

overall expected consumption matters without regard to a projects risk, which

is to say any hedge against the risk of a state M under symmetric information

has no value if it reduces the overall expected consumption, which we have seen

it does.

If some risk aversion is introduced, � H b 0 and � HH ` 0, then a hedge

against the risk of an M state under symmetric information acquires value.

If  p� 1 then there is no possibility of such a state, and therefore there is no

such risk to hedge against. However, as p is reduced from 1 by some value

D b 0, so that 0 ` p ` 1, then there exists the possibility of such a state, andby the standard result the value of debt as a hedge against this possibility

J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699 687

Page 14: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 14/25

may result in some contract mixing, with both debt and equity used. A more

formal analysis of this mixing could be performed with the speci®cation of a

particular utility function, but this would introduce unnecessary mathematical

complexity ± the intuition remains as presented above. As is shown in the fol-

lowing empirical section, there seems to be in practice very little such contract

mixing. This result is consistent with the risk neutral model presented above

under a high probability of asymmetric information, p % 1. To the extent that

risk aversion exists, this would seem to imply that there is an even higher prob-

ability of asymmetric information in the projects funded by venture capitalists.

There are three other points that bear mentioning:

· There may be reasons for the use of preferred stock even under certain

symmetric information (p� 1 A l� 2). If for example the entrepreneurhas customers expecting delivery of some good or service at t2, the poten-

tial for foreclosure by the venture capitalist at that time may make those

customers reticent to do business with the entrepreneur if debt is used.

A similar argument could be used for current liability holders unsure of 

the seniority of their claim or concerned that the venture capitalist foreclo-

sure option will, under asymmetric information, cause the entrepreneur to

behave opportunistically and leave them with nothing. If however the ®rm

has no customers in the initial development stage the former argument is

not valid, of course.

· Just as debt may provide a poor incentive structure in a venture capital ®-nancing under high probabilities of asymmetric information, so a similar in-

centive inconsistency may develop if equity is later sold to the public by the

venture capitalist in an IPO, if that high probability of asymmetric informa-

tion persists. This would tend to make high debt levels undesirable in such

situations, if an IPO is desired. Conversely, under the above framework

the existence of debt may indicate a low probability of asymmetric informa-

tion, and might even serve as a signal of this to the market.

· A logical question in the context of this model is what distinguishes ven-

ture capital ®rms from other intermediaries, i.e. banks, which use debt.

In terms of contracting, the key di�erence is the di�culty of monitoring,which is to say the likelihood of asymmetric information, p in the model.

It is the high likelihood of asymmetric information in the projects venture

capitalists ®nance which causes them to use preferred equity. Banks and

other institutions contracting with debt are ®nancing projects where mon-

itoring is easier and the probability of asymmetric information is smaller.

Should banks choose to become involved in ®nancing the sort of projects

funded by venture capitalists, they would presumably switch to preferred

equity (anecdotally, one of the ®rms responding to the survey described be-

low is the venture capital arm of a money center bank, and its investments

were made along the same pattern as the other ®rms, that is, with preferredstock).

688 J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699

Page 15: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 15/25

4. Empirical evidence

The above model gives us a context under which to examine patterns in ven-

ture capital ®nance. This context is di�erent from that previously presented in

the literature in that prior studies have focused on the ``cash-out'' IPO end

stage of the venture capital process. Previously the availability of IPO data mo-

tivated the theoretical focus (e.g. questions of signals to the market) to the ex-

clusion of the initial investment issues considered here. This analysis deals with

the ``cash-in'' initial investment stages of venture capital. Therefore I will in-

vestigate the contracts o�ered by these ®rms to seek support for the above

model. This study will center on the nature of these contracts and how they

vary with di�ering probabilities of asymmetric information.The theory presented makes the following predictions, which constitute null

hypotheses here. In situations where monitoring is di�cult and hence the prob-

ability of a state of asymmetric information is high, preferred equity should

tend to be the dominant form of contracting. The use of debt should become

more pronounced in situations where monitoring is easier and the probability

of information asymmetries is lower.

Not surprisingly, venture capital ®rms tend to be unwilling to reveal detailed

information on the ®nancings on a deal by deal basis. Therefore, data was ob-

tained through a direct survey of ®rm principals asked to provide information

on the aggregate ®nancing patterns of their ®rms. Such a method is necessarilycruder than direct examination of ®nancial records, but at present may be the

only hope for acquiring any meaningful data on the crucial initial stages of the

venture capital process.

A survey of eight venture-capital ®rms was performed, the sample having

been provided by partners leading venture ®rms directly. All of the ®rms had

at least 250 million dollars under management and had done in excess of 

$100 deals. Firms were asked the following questions:

1. In ®nancing companies in initial product development phases (de®ned as

®rms with no current customers) what is the estimated percentage of deals

in which the following are used: (a) preferred stock (b) common stock (c)debt?

2. In ®nancing companies in later phases (de®ned as ®rms with current custom-

ers) what is the estimated percentage of deals in which the following are used

(a) preferred stock (b) common stock (c) debt?

3. What percentage of early stage ®nancings are of ®rms with pre-existing debt

(including and excluding current liabilities)?

4. What percentage of later stage ®nancings are of ®rms with pre-existing debt

(including and excluding current liabilities)?

Questions 1±4 were then each repeated for speci®c industries: biotech, soft-

ware, retailing and manufacturing. These industries were chosen for their dif-ferent developmental processes and hence variations in the probability of 

J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699 689

Page 16: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 16/25

asymmetric information (see below). Lastly, venture capitalist were asked what

percentage of preferred-®nanced deals utilized convertible preferred, and in

what percentage of venture capital ®nanced deals control was shifted from

the entrepreneur to the venture capitalist.

Turning to the industry groups considered, note that biotech ®rms go

through a product development process in which the initial stages, consisting

of in-house lab tests, are non-public and extremely di�cult for anyone but

the researchers involved to interpret. By contrast, later stages of testing, e.g.

the FDAs Phase I±III tests, have results which are generally made public, mak-

ing monitoring easier and reducing the likelihood of asymmetric information.

Similarly, software ®rms have an initial product development stage known

as ``alpha'' testing, in which the software is tested in-house and the resultsare known only to the programmers. A later phase of testing, known as the

``beta'' phase, involves sending software out to user groups and industry peri-

odicals for review. The results of this phase are widely observed by industry

participants, lowering the probability of asymmetric information.

In the case of retailing, no high-tech barrier exists to determining the state of 

the project in early phases of development, although some practical di�culties

may remain in monitoring sales or the reliability of suppliers, etc. Once reve-

nues exist, however, they may be used as a monitoring device.

As would be expected under the above theory with a high probability of 

asymmetric information, the data demonstrates that preferred stock representsthe dominant mode of contracting. In early stage ®nancings, the means and

standard deviations of the contract types are (Table 1) 95.4% (�1.3%) for pre-

ferred, 2.0% (�1.0%) for common, and 2.1% (�1.0%) for debt. No venture

®rms indicated any signi®cant change in their estimated percentage use of these

contracts for the four industry groups (Tables 2±5), although two ®rms indicat-

ed that as a matter of policy they do not invest in retail ®rms, which accounts

for a slight di�erence in the breakdown for this category: (Table 4) 95.5%

(�1.7%), 2.5% (�1.3%), 2.0% (�1.1%).

The 95% con®dence intervals are calculated assuming a sample size of 800

entrepreneurial ®nancings (600 in retail) with each venture ®rm representing100 of these transactions. In fact it was only possible to determine that each

venture ®rm had done at least 100 deals, so the con®dence intervals given here

are for a lower bound on the sample size, and may be somewhat distorted by

the assumption of equal weights for the venture ®rms. However, given the sim-

Table 1

All deals ± Early stage, # venture ®rms � 8

Mean (95%) C.I. (n� 800) Min Max

% Preferred 96.4 �1.3 90 100

% Common 2.0 �1.0 0 10% Debt 2.1 �1.0 0 10

690 J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699

Page 17: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 17/25

ilarity in responses among the venture capital ®rms, the weighting assumptionis of very limited importance. Con®dence intervals are therefore calculated as

follows. For sample size n and fraction of a�rmative responses p, i.e. fraction

of deals in which a given contract type is used, the 95% C.I. is

Æ� 0X025

p � p �1 À p �an�. Here Z 0X025 � 1.96 is the standard normal, and note here

the central limit theorem has been applied so that in the large sample limit

Z 0X025 equals and replaces t0X025, the student t at 0.025.

In later stage ®nancings, there appears to be some increase in the use of debt

and common, although the great majority of deals still utilize preferred. The

percentage use breakdown is (Table 6) 87.8% (�2.3%) for preferred, 3.9%

(�1.3%) for common, and 10.1% (�2.1%) for debt. As can be seen from Ta-bles 7±10, there is little variation from this pattern for the four industry groups,

Table 2

Biotech ± Early stage, # venture ®rms

�8

Mean (5%) C.I. (n� 800) Min Max

% Preferred 96.4 �1.3 90 100

% Common 2.0 �1.0 0 10

% Debt 2.1 �1.0 0 10

Table 3

Software ± Early stage, # venture ®rms� 8

Mean (95%) C.I. (n� 800) Min Max

% Preferred 96.4 �1.3 90 100% Common 2.0 �1.0 0 10

% Debt 2.1 �1.0 0 10

Table 4

Retail ± Early stage, # venture ®rms� 6

Mean (95%) C.I. (n� 600) Min Max

% Preferred 95.5 �1.7 90 100

% Common 2.5 �1.3 0 10

% Debt 2.0 �1.1 0 10

Table 5

Manufacturing ± Early stage, # venture ®rms� 8

Mean (95%) C.I. (n� 800) Min Max

% Preferred 96.4 �1.3 90 100

% Common 2.0 �1.0 0 10

% Debt 2.1 �1.0 0 10

J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699 691

Page 18: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 18/25

with only retailing indicating any meaningful deviation, the breakdown being

(Table 9) 83.2.8% (�3.0%), 5.0% (�1.7%), 13.2% (�2.7%).

Under the theoretical framework presented here such a shift would be con-sistent with easier monitoring, and hence a lower probability of asymmetric in-

Table 6

All deals ± Later stage, # venture ®rms

�8

Mean 95% C.I. (n� 800) Min Max

% Preferred 87.8 �2.3 50 100

% Common 3.9 �1.3 0 10

% Debt 10.1 �2.1 0 50

Table 7

Biotech ± Later stage, # venture ®rms � 8

Mean (95%) C.I. (n� 800) Min Max

% Preferred 88.4 �2.2 50 100

% Common 3.9 �1.3 0 10

% Debt 9.5 �2.0 0 50

Table 8

Software ± Later stage, # venture ®rms� 8

Mean (95%) C.I. (n� 800) Min Max

% Preferred 88.4 �2.2 50 100

% Common 3.9 �1.3 0 10

% Debt 9.5 �2.0 0 50

Table 9

Retail ± Later stage, # venture ®rms� 6

Mean (95%) C.I. (n� 600) Min Max

% Preferred 83.2 �3.0 50 100

% Common 5.0 �1.7 0 10

% Debt 13.5 �2.7 0 50

Table 10

Manufacturing ± Later stage, # venture ®rms� 8

Mean (95%) C.I. (n� 800) Min Max

% Preferred 88.4 �2.2 50 100

% Common 3.9 �1.3 0 10

% Debt 9.5 �2.0 0 50

692 J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699

Page 19: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 19/25

formation, in retail. However, given the smaller sample size and large standard

deviations here, this slightly lower use of preferred, in favor of debt, in retail

deals may not be meaningful. Nonetheless, retailing does not involve advanced

technologies which may be tested in-house, and thus out of the view of the ven-

ture capitalist. While ®elds like biotech, software and manufacturing involve

advanced technologies which only experts (e.g., the entrepreneur) may be in

a position to evaluate, the evaluation of a retail operations health is a simpler

matter, largely involving the monitoring of inventory, sales and the like. Thus it

seems plausible to argue that a higher percentage of retail ®rms would present a

low enough probability of asymmetric information to make debt ®nancing fea-

sible.

Of course it is possible that entrepreneurial ®rms are receiving debt ®nancingfrom sources other than venture capital ®rms. Questions 3 and 4 are designed

to address this issue. In the case of early stage ®nancing and excluding current

liabilities (which virtually all ®rms have e.g., utility bills, rent etc.) the mean

percentage of deals in which some debt existed on the books of the ®rm at

the time of venture capital ®nancing is (Table 11) 8.2% (�1.9%). This ®gure

is constant across the four industry groups with the exception of retail, which

is slightly higher at 10.8% (�2.5%). For later stage ®nancings (Table 12), the

percentage of deals with debt is considerably higher, at a mean of 41.9%

(�3.4%). Note that the mean for biotech software and manufacturing are

40.0.3% (�3.4%), 43.1% (�3.4%) and 46.9% (�3.5%), respectively. Only retailis signi®cantly higher, at 69.2% (�4.0%).

The higher instance of debt usage in later stage ®rms may be indicative

of easier monitoring; for example the health of a ®rm with revenues may be

Table 11

% Deals with some debt on books ± Early stage

Mean (95%) C.I. Min Max # venture ®rms

% All deals 8.2 �1.9 0 35 8

% Biotech 8.1 �1.9 0 35 8

% Software 8.1 �1.9 0 35 8

% Retail 10.8 �2.5 0 35 6

% Manufacturing 8.1 �1.9 0 35 8

Table 12

% Deals with some debt on books ± Later stage

Mean (95%) C.I Min Max # venture ®rms

% All deals 41.9 �3.4 0 75 8

% Biotech 40.0 �3.4 0 80 8

% Software 43.1 �3.4 0 75 8

% Retail 69.2 �4.0 20 100 6% Manufacturing 46.9 �3.5 0 80 8

J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699 693

Page 20: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 20/25

partially evaluated through examination of those revenues. As noted above,

more mature high-tech ®rms go through more public product testing, e.g.,

FDA tests that biotech ®rms go through, and the beta testing of software. De-

pending on the product, manufactured items may also have to go through gov-

ernment safety tests with public results (e.g., in the defense or auto industries).

All of these factors would tend to lower the probability of asymmetric informa-

tion in later stage ®rms and make the use of debt more attractive under the

probability of asymmetric information-driven framework constructed above.

Likewise the higher instance of debt tolerated in retail ®rms may be a function

of the greater ease of monitoring, and hence lower probability of asymmetric

information, associated with this industry. Under the theory presented, this

lower probability of asymmetric information would make the use of debt moreattractive.

Lastly, it is argued that the use of a conversion option in a preferred con-

tract should cause that contract to dominate common in the pre-IPO period,

eliminating any argument for the use of common to make possible a cash-

out in an IPO. Indeed, the mean of the surveyed ®rms' estimates of the percent-

age of preferred deals which used convertible preferred is (Table 13) 88.8%

(�2.2%). These options generally do not result in a transfer of control (de®ned

as e�ective control, or the ability to convert to voting shares capable of seizing

control) from entrepreneurs to the venture capitalists ± the mean percentage of 

deals in which such a transfer occurred was (Table 14) 28.0% (�3.1%). Thus itseems plausible that these options are simply facilitating the cash-out process.

Certain caveats are in order in drawing interpretations from the estimates

garnered in a survey of the type performed here. The precision which can be

reasonably hoped for in the estimates made by the venture capitalists may be

lower than would be indicated by the standard deviations given. However,

the overall pattern revealed here, that being the dominance of the use of pre-

ferred in deals where the potential for asymmetric information is high, seems

so pronounced that it is unlikely to be in any sense spurious.

The data presented here is of particular value because, despite its necessarily

rough quality, it is the only data on the aggregate percentages of contract types

Table 13

% Preferred deals where preferred convertible

Mean (95%) C.I. Min Max # venture ®rms

% Pref./conv. pref. 88.8 �2.2 30 100 8

Table 14

% Deals where venture syndicate could take control

Mean (95%) C.I. Min Max # venture ®rms% All deals 28.0 �3.1 1 100 8

694 J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699

Page 21: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 21/25

used in venture capital which has ever been obtained. Further, the massive

dominance of preferred equity as the contract type, not only as compared to

common equity but to debt as well, is of particular importance given that

the previous venture capital models have centered on the use of common stock,

e.g. Admati and P¯eiderer (1991), Chan et al. (1990), or debt e.g. Hansen

(1992). As illustrated in the model presented here, the use of preferred stock

in venture capital is crucial. It is driven not only by the risk inherent in the en-

trepreneurial projects themselves but by the di�culty in monitoring those pro-

 ject which the venture capitalist is likely to encounter.

Venture capital is not just a matter of making a risky loan or purchasing a

portion of a risky project. It involves extraordinary monitoring problems lead-

ing to a high probability of asymmetric information, particularly in the hightech ®elds which venture capital has ®nanced in recent years. It is this high

probability of asymmetric information which appears to account for the choice

of preferred stock as a ®nancing tool, and renders the foreclosure option of 

debt a hindrance. Analysis of venture capital ®nancing without consideration

of this point is fundamentally incomplete.

5. Conclusions

This study has extended the venture capital literature by analyzing the roleof asymmetric information conditions. Such conditions may arise subsequent

to the time of contracting, can play in the choice of contract type. In particular,

the choice of preferred equity over debt has been explored and shown to be po-

tentially motivated by intertemporal incentive inconsistencies due to the possi-

bility of information asymmetries in certain states of intermediate return. The

venture capitalist would foreclose under symmetric information. However, the

entrepreneur has an incentive to behave opportunistically under asymmetric in-

formation. If the probability of asymmetric information is high, this e�ect may

make the foreclosure feature of debt a large enough drag on the expected re-

turn of the venture capitalist as to make debt contracting infeasible. Undersuch circumstances preferred equity may be desirable, because without foreclo-

sure the entrepreneur is not pushed into behaving opportunistically and the

venture capitalist may receive some positive return, rather then a total loss.

This feature of my model may help to explain the widespread use of pre-

ferred stock in venture capital contracting, particularly in high-tech ®elds like

software and biotech. In these ®elds, the initial phases of development often in-

volve tests which only the entrepreneur and his sta� can observe and evaluate

(in-house ``alpha'' testing in software, private lab tests in biotech) while later

phases are more easily observed by outside parties (computer journal ``beta''

tests in software, FDA drug tests or EPA/Dept. of Agriculture ®eld tests in bio-tech).

J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699 695

Page 22: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 22/25

In the empirical results presented, the high instance of preferred stock usage

is documented and shown to be consistent with the model presented. This is

due to the fact that the industries funded by venture capitalists tend to exhibit

by their nature high probabilities of asymmetric information. The greater usage

of debt in retail, which presents less of a monitoring challenge and presumably

exhibits a lower probability of asymmetric information, is also consistent with

a framework in which a lower probability of asymmetric information makes

the use of debt more attractive. Finally, it is found that in later rounds of ®-

nancing the instance of debt use rises, although preferred is still the dominant

contract. This is consistent with a higher percentage of more developed ®rms

having a lower probability of asymmetric information, making a debt contract

feasible.These results suggest policy makers wishing to encourage the formation and

deployment of venture capital might consider actions conducive to preferred

stock investment. For example, these might include elimination of the double

taxation of dividends on preferred stock used in venture capital, and the place-

ment of such securities on at least an equal tax footing with debt. A cut in the

capital gains tax on such securities would also be helpful.

This paper also suggests future directions of inquiry. Among these are

the pursuit of a better understanding of the role of capital structure in the ven-

ture capitalist's exit strategy, e.g. IPOs and earn-outs. Also of interest is the

role of capital structure in markets with a less developed venture capital sector,e.g. the European market with its universal banks. The ability of such banks to

make equity investments may allow them to substitute for venture capital ®rms

to some degree, although other institutional features (size, con¯ict of interest,

bias toward debt, etc.) may limit the e�ectiveness of such banks in funding ven-

tures.

In summary, this study has demonstrated that the dominant contract in ven-

ture capital ®nancing is preferred stock. This is supportive of a model in which

the contract choice is driven by the high probability of information asymme-

tries between the entrepreneur and the venture capitalist. Venture capital ®-

nancing is not simply a matter of making a high-risk loan or equityinvestment. It involves extreme monitoring di�culties, and preferred equity

represents the method by which these can be dealt with in an incentive-compat-

ible manner. An understanding of the venture capital process without consid-

eration of this point in necessarily incomplete.

Acknowledgements

The author wishes to thank Anthony M. Santomero, Franklin Allen, Rich-

ard Herring, Richard E. Kihlstrom, and Donald F. Morrison for help and sup-port. This paper was presented at N.Y.U. Stern School's Conference on the

696 J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699

Page 23: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 23/25

Economics of Small Business Finance, and the author gratefully acknowledges

the discussant, Anjan V. Thakor, as well as those of Gregory F. Udell, Allen N.

Berger, an anonymous referee, and the conference participants. I also thank

Matthew Richardson, Michael Goldstein, Michel Habib, Jim Mahoney and

Mark Roomans for very useful discussions. This work was supported in part

by a Dean and a Unisys Fellowships. All errors are the responsibility of the au-

thor.

Appendix A. Proofs

A.1. Proof of Proposition 2

First, set p� 1 and construct a project such that �1�� � ` w1Y� � MY L and �3�L� ` w3. I further adjust � �� �Y �1�� � and �1�� � so that un-

der a preferred equity contract

E0�  v3 À �  v0 � 0 iff   g pv2 �  �1�H� À d À �w1 � d 1� set at t 0Y �5�

g pv3 �  �3�H� À d À �w3 � d 3� if  �  is revealed to be H at t lY �6�

p

v3 �  �3�M� À d À w3 if  �  is revealed to be M at t lX �7�Here d 1 and d 3 are premia over the reservation wages w1 and w3 necessary to

make the expected return to the entrepreneur at least as high as w1 + w3, given

 �1�� � ` w1Y � � MY L and �3�L� ` w3, as well as the probabilities � �� �Y� � LY MY HX

Such a project may be constructed if, in the L state �3�L� ` s 2 À d X Then if 

S �L it will not be possible for the venture capitalist to make a positive return

on his investment of s 2T k necessary to continue the project. Suppose this is the

case. Then since the project requires total capital of  qs 0 � k  to continue at t2

(including the ®xed asset and the new working capital) the forced liquidation

of the project by the current liability creditors is certain in the L state if thereis symmetric information or if the entrepreneur does not behave opportunisti-

cally. Therefore the entrepreneur is certain to behave opportunistically under

asymmetric information (which will be the case here since p� 1), but he will

receive less than his reservation wage in both the L and M states by the above

assumptions. Therefore he must receive more than that wage in the H state to

make his expected return equal that reservation wage. As for the venture cap-

italist, his payo�s in the H and M states must be high enough to make up for

the certain zero payo� in the L state in order for E0W v3 A W v0 � 0 to hold and

the contract to be feasible.

It is possible to construct such a project simply by adjusting P (S ), raisingR1(H) and R3(H) su�ciently above R1(M) and R3(M), and adjusting the

J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699 697

Page 24: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 24/25

capital requirements I 0 and K  appropriately. This is possible since these are all

free parameters in an underdetermined system of equations (the system of 

Eqs. (2), (3), (5)±(7)) plus the normalization�

� �LYMYH � �� � � 1 represents six

equations with 15 unknowns, admitting a nine dimensional free parameter sub-

space in which the desired project solution may be found). Note the w3 term in

Eq. (7) arises form the fact that unless the entrepreneur knows he will be paid

that amount at t3 there is no reason to continue with the project. Any new con-

tract at t2 in M must then promise such a wage at t3.

Now consider: for Eqs. (6) and (7) to play a necessary role in making

E0W v3 A W v0 � 0, the entrepreneur must not behave opportunistically in either

the M or H states, despite the certainty of asymmetric information at t1. There-

fore w3 b �1�H� b �1�M� must hold. This is to say that if the project is contin-ued, the entrepreneur stands to make more at t3 then he would by behaving

opportunistically at t1, l� 2. So long as this is the case the preferred equity

contract above may be feasible.

This scenario can be adjusted so that the preferred equity cannot be replaced

with a feasible debt contract. If debt were used, Eq. (5) would have to hold for

E0W v3 A W v0 � 0. But since debt carries a foreclosure right, the venture capi-

talist would have the right to foreclose on the ®rm's assets in state M, because

the entrepreneur would be unable to pay g v2 �  �1�H� À d À �w1 � d 1� in state

M. Indeed, if the condition

 �1�M� À d b �3�M� À d À w3 �8�the venture capitalist will prefer to foreclose should the entrepreneur not be-

have opportunistically at t1. Knowing this, however, given that p� 1 and hence

l� 2, the entrepreneur will be certain to behave opportunistically, setting

C e1 �R1(M) and leaving the venture capitalist (and the current liability credi-

tors) with nothing. Thus the venture capitalist is left with nothing in state M

under a debt contract. But by construction the venture capitalist must have

a non-zero payo� in state M for E0W v3 A W v0 � 0. Therefore the debt contract

is not feasible for this project while the preferred equity contract is feasible, and

given the proper values of d 1 and d 3, the preferred represents a Nash equilib-rium. It should be noted that under the preferred contract, for the venture cap-

italist to wish to fund the second stage the relation

0 (  �3�M� À d À w3 �9�must hold. Substitution of this into Eq. (8) yields, �1�M� À d P 0 which of 

course is consistent.

Thus, a project is constructed so that a preferred equity contract is just fea-

sible, and only because it receives some payo� in the M state. Then since the

debt contract receives no payo� in the M or L states, and con receive no more

payo� then the preferred contract in the H state due to the entrepreneur's res-ervation wage, the debt contract is not feasible. h

698 J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699

Page 25: Venture Capital Contracting Under Asymmetric Information

8/4/2019 Venture Capital Contracting Under Asymmetric Information

http://slidepdf.com/reader/full/venture-capital-contracting-under-asymmetric-information 25/25

References

Admati, A.R., P¯eiderer, P., 1991. Robust ®nancial contracting and the role of venture capitalists.

Journal of Finance 49 (2), 371±402.

Allen, F., Santomero, A.M., 1997. The theory of ®nancial intermediation. Journal of Banking and

Finance 21, 1461±1485.

Allen, F., 1985. On the ®xed nature of sharecropping contracts. The Economic Journal 95, 30±34.

Boot, A., Thakor, A.V., 1993. Security design. Journal of Finance 48, 1349±1378.

Chan, Y.-S., Siegel, D., Thakor, A.V., 1990. Learning, corporate control and performance

requirements in venture capital contracts. International Economic Review 34 (2), 365±381.

Diamond, D., 1984. Financial intermediation and delegated monitoring. Review of Financial

Studies 51, 393±414.

Gale, D., Hellwig, M., 1985. Incentive-compatible debt contracts: The one-period problem. Review

of Economic Studies 52, 547±563.Gompers, P.A., 1992. The structure of venture capital investing. Working paper. Harvard

University Cambridge, MA.

Hansen, E., 1992. Venture capital contracts with conversion options. Discussion paper 24. Reserve

Bank of New Zealand Wellington.

Jensen, M.C., Meckling, W.H., 1976. Theory of the ®rm: Managerial behavior, agency costs and

ownership structure. Journal of Financial Economics 3, 305±360.

Kihlstrom, R.E., La�ont, J.-J., 1979. A general equilibrium entrepreneurial theory of ®rm

formation based on risk aversion. Journal of Political Economy 87 (4), 719±748.

Lerner, J., 1992a. Venture capitalists and the decision to go public. Journal of Financial Economics

35 (3), 293±316.

Lerner, J., 1992b. Venture capitalists and the oversight of privately-held ®rms. Journal of Finance

50 (1), 301±318.Sahlman, W.A., 1991. Aspects of ®nancial contracting in venture capital. Journal of Applied

Corporate Finance 1, 23±36.

Santomero, A.M., 1984. Modeling the banking ®rm. Journal of Money Credit and Banking 16 (4),

576±602.

Townsend, R.N., 1979. Optimal contracts and competitive markets with costly state veri®cation.

Journal of Economic Theory 21, 1±29.

J.J. Trester / Journal of Banking & Finance 22 (1998) 675±699 699