Description
A floating interest rate, also known as a variable or adjustable rate, refers to any type of debt instrument, such as a loan, bond, mortgage, or credit, that does not have a fixed rate of interest over the life of the instrument.
CASE STUDY ON RECENT INNOVATIONS IN
FINANCE: MICROFINANCING AND FLOATING
RATE CONVERTIBLES
ABSTRACT
The first essay provides theory concerning the risk-taking incentives of microfinance
borrowers in varying cases: individual liability, group liability without social sanctions, and group
liability with social sanctions. The results provide insight into how a community's social capital and a
country's credit rights interact to induce recipients of microfinance programs to take risk. Consistent with
recent anecdotal evidence that suggests a "dark side" to microfinance, the results show that communal ties
among joint liability borrowing groups may not lead to higher repayment rates and may have worse
welfare effects on the recipients by making the poorest group members unwilling to take the risks
necessary to grow a business.
The second essay considers floating rate convertibles (FRCs). FRCs are a category of
PIPE securities that receive negative associations in both the academic and professional literature. This
study sheds light on the managerial relationship to the decision to issue FRCs and to the variation in
market response to these issues. One main result of the study identifies influence of the CFO relative to
the CEO as significant in the decision to issue FRCs and in the market's immediate reaction to the
issuance. Another main result is that FRC issuing firms with CFOs without prior public equity
issuance experience have significantly negative long run abnormal returns, whereas FRC issuing firms
with experienced CFOs do not
Table of Contents
Dedication..................................................... ii
Acknowledgements................................................ iii Table of
Contents................................................ iv
Essay 1: A Theory on the Risk-Taking Behavior of Microfinance Borrowers, Creditor
Rights, and Social Capital..................................................... 1
I. Introduction..............................2 II. Literature
Review...........................4 III. Individual
Lending.............................8 IV. Joint Liability
Lending........................11 V. Welfare Comparisons and
General Equilibrium Extension.........25 VI. Empirical
Hypotheses..........................28 VII.
Conclusion..............................28
Appendix.................................29
References...............................58
Essay 2: Issuances of Floating Rate Convertible Securities and Financial Manager
Characteristics..................................61
I. Introduction................................................... 62 II.
Literature..............................63 III. Data
Sample..............................69 IV.
Hypotheses...............................81 V.
Results...............................84 VI.
Conclusion...............................89
Appendix.................................91
References................................101
iv
Essay 1: A Theory on the Risk-Taking Behavior of Microfinance Borrowers, Creditor Rights,
and Social Capital
Abstract:
This paper provides theory concerning the risk-taking incentives of microfinance borrowers in varying
cases: individual liability, group liability without social sanctions, and group liability with social sanctions. The results
provide insight into how a community's social capital and a country's credit rights interact to induce recipients of
microfinance programs to take risk. Consistent with recent anecdotal evidence that suggests a "dark side" to microfinance,
the results show that communal ties among joint liability borrowing groups do not lead to higher repayment rates and does
have worse welfare effects on the recipients by making the poorest group members unwilling to take the risks necessary to
grow a business. This paper contributes to the literature on contract design and on financial development and growth,
extending them into the realm of finance for the poorest and least able to access formal financial markets.
1
I. Introduction
Microfinance is the popular economic development program aimed at the world's poorest entrepreneurs
in many developing countries. It normally consists of lending very small amounts, as little as $75, to
invest in self-owned enterprises, in order to provide more opportunities beyond wage labor.
Microfinance is intriguing because it provides financial services to a very large segment of the world's population, who
otherwise would borrow in the informal financial market from friends, family, and moneylenders. Furthermore, prior to its
first implementation by economics professor Mohammed Yunus in Bangladesh in the 1970s, very poor borrowers in
developing countries were not expected to be willing or at least able to repay unsecured business loans made by lenders
from outside their communities.
It is commonly believed that microfinance programs succeed in leading to high repayment rates because
of the strong social ties that exist among their clients. "Social capital" is considered an alternative to physical capital in
the developmental economics literature. While borrowers may not have physical collateral to secure a loan, they do live
in tight knit cultures where the social repercussions from defaulting on a loan could be as costly as losing material
possessions. The first microfinance programs were thought to tap into this social capital by giving loans to groups, where
each member of the group was liable for all the others' share of the loan (Van Bastalaer, 2000).
Because all members of the borrowing group are liable for each other, they have incentives to punish
noncontributing members and to encourage each other to succeed. They also have incentives to screen potential members
into the group as well as verify that each group member exerts effort in the projects so that they can repay their shares of
the loans. Harnessing social capital is particularly important in countries where lenders have little recourse to deal with
borrowers that are able to repay their loans but choose not to do so. This inability is often due to poor institutions like
creditor rights. (La Porta, et.
al., 1998)
Even if borrowers own land, they may not be able to use it as collateral because of laws that restrict the
use of land as collateral or make the ability to seize collateral very costly or impossible. Also, record of title may be
unavailable due to poor record keeping.
Despite the theoretical support for the value of social capital in providing financial services to
developing nations' poor entrepreneurs, studies of microfinance programs have also yielded some puzzling results. One
is the success of microloans made on an individual liability basis.
If individual liability loans have similar repayment rates as joint liability loans, what does this say
about either the need for social capital or the necessity of the joint liability contract in harnessing it? The other puzzle
deals with the problem of loans not being put to productive use. That is, many of microborrowers seem to hold onto their
loans or consume them rather than invest them in a business.
2
This is one finding that does not have much theoretical explanation. Intuitively, the reasons given is that the members of
the group receiving the loan are too afraid of the repercussions from failure from their peers that they prefer to use the
loan to smooth their consumption and payoff the loan with income from pre-existing sources such as a wage earning job
or even a business that they did not invest the loan in.
In this paper, I develop a theoretical model on the relationship between the repayment strategies of
microfinance borrowers and the types of projects they invest their microloans in. Besley and Coate (1995) model how
group lending mitigates the moral hazard problem of borrowers being unable to credibly contract with lenders to repay
their loans even when they are able. Besley and Coate address the question of how the lending schemes affect the
incentives of borrowers to repay their loans, but they assume that the borrowers' projects are exogenously given. This
paper extends their model by endogenizing project selection under various lending schemes. By modeling the choice of
what project a borrower will choose, I provide a theory to explain the relationship between financial contract design, a
country's creditor protections, investment choice, entrepreneurship, and poverty.
I present conditions for microfinance programs resulting in optimal project selection. I show that
certain entrepreneurs who are jointly liable for their loans will not take the necessary business risks that microfinance is
supposed to induce them to take while not improving the repayment rate over individual liability. This result is driven
by strong social ties that exist among borrowing group members, which is usually thought of as a positive effect on social
welfare.
The conclusions of this study contribute to the theory of microborrower behavior in two main ways. The
first theoretical contribution of this study is an extension of Besley and Coate's (1995) comparison of the repayment
behavior of microborrowers under varying assumptions of liability and existence of social sanctions. I show that
borrowers under joint liability without the threat of social sanctions are more likely to repay their loans than when under
the threat of social sanctions. By introducing project and peer selection into the model, I get this result because the threat
of social sanctions stifles risk sharing and encourages free-riding among borrowers with low upside potential to their
project options.
The second theoretical contribution is the identification of people who choose particular types of
contracts where there is an option between individual and joint liability contracts. If there are no social sanctions, then
people with high upside potential to their projects take the joint liability contract because the benefits of risk sharing
outweigh the costs of free-riding. People with low upside project potential, however, take the individual liability contracts
because they do not share any risk under the joint liability contract by selecting safe projects. The introduction of social
sanctions to the model, however, effectively eliminates the use of the joint liability contract by every microborrower.
3
Ultimately, studying this question is important because microfinance institutions (MFIs), governments, non-governmental
organizations (NGOs), and microfinance investors want to know what model of microfinance to follow. Should a MFI
offer joint liability contracts to borrowers who are able to sanction each other if one borrower does not contribute to
repayment? Should a social planner subsidize MFIs
that offer joint liability contracts or those that offer individual liability contracts?
The remainder of the paper is organized as follows. Section two reviews the relevant literature on
microfinance and positions this paper in the existing literature. Section three presents a model of individual liability
lending. Section four presents a model of joint liability lending without and with social sanctions. Section five discusses
the results of the model. Section six concludes with a summary and direction for further research.
II. Literature Review
The literature on group lending argues for social capital to impact the likelihood of repayment through various channels.
The most frequently cited categorization of models explaining how social capital impacts group repayment rates is by
Ghatak and Guinnane (1999). The first is in a superior screening ability of peers over delegated monitors because most
groups select which borrowers can join them (Ghatak and Guinnane 1999). The second is a superior monitoring ability of
peers to control ex ante moral hazard (Stiglitz 1990). The third is a superior auditing ability of peers to limit verification
costs. The fourth is a superior enforcement mechanism through imposition of social sanctions should a borrower default
to control ex post moral hazard (Besley and Coate 1995). As opposed to comparing the effects of each of these four
assumptions on what kind of problem social capital addresses in group lending, I compare the impact of various forms of
social capital on risk taking and repayment of borrowers.
Besley and Coate (1995) provide a model to predict how the group liability aspect of microfinance affects
the repayment behavior of borrowers when repayment of loans is not enforceable. They compare the model's predictions
of repayment rates among three lending systems: individual lending, group lending without social sanctions, and group
lending with social sanctions. The individual lending system describes the traditional arrangement in which the bank
lends to an individual who is solely liable. The group lending system without social sanctions describes an arrangement in
which the bank lends to a group of borrowers who divide the loan among themselves and invest their shares in their own
enterprises, which are independent from one another. In this system, the group as a whole is liable, each member of the
group decides whether or not she contributes her share to the group's repayment, and the group cannot penalize
noncontributing members. The group lending system with social sanctions describes the same arrangement, except the
group can exert peer pressure on the members to contribute to the repayment of the loan.
4
Their main conclusions are twofold. First, the impact of group lending on repayment rates over that of individual lending
are affected by two countervailing forces: risk-sharing and free-riding. On one hand, if one member of the group cannot
contribute her share of the loan because of poor project returns, the other group members can cover her share. Theref ore,
group lending may improve loan repayment rates through sharing risk. On the other hand, a borrower who would have
repaid her loan under individual lending might take advantage of the group's incentives to cover her share of the loan.
This free-riding incentive leads to a coordination failure whereby the group as a whole will default even though certain
individuals would have repaid their own shares if they were individually liable. Without social sanctions, it is unclear as
to which effect dominates. The second conclusion is that the free-rider effect can be lessened if the group is able to use
social sanctions to pressure the members if they stand to lose a significant amount of social capital. They show that if
social sanctions are great enough, then group lending does dominate individual lending in regard to the repayment rate.
An extension of Besley and Coate's model is Che (2002)'s dynamic model of repayment behavior in a
repeated game. Che does not include social sanctions other than exclusion from participation in future borrowing
opportunities. He shows that while the static model has ambiguous results concerning group lending's repayment rates
due to the free-rider problem, the dynamic model shows that group lending weakly dominates individual lending due to
the cost of exclusion from future opportunities.
In addition to extending their model to a dynamic setting, among the recommendations that Besley and
Coate make for further research is to model the effect of group lending on the type of project that each member chooses.
As this paper will show, Besley and Coate's model can be adapted to find interesting implications for selection of project
risk by different lending schemes. In addition, I extend their model to include self selection of borrowers into groups. The
outcome of endogenizing project and group member selection is domination of the risk sharing effect without social
sanctions and suboptimal risk taking with social sanctions.
In addition to Besley and Coate (1995)'s focus on the enforcement problem, other theories focus on the
problem of (i) screening out risky borrowers, (ii) monitoring borrower effort ("ex ante" moral hazard), and (iii) verification
of project outcomes. Ghatak (1999) and Ghatak and Guinnane (1999) provide the most cited model of how joint liability
microloans induce risky borrowers to select other risky borrowers to form a group and safe borrowers select other safe
borrowers. By this assortive matching process, lenders are more able to identify risky borrowers from safe borrowers "by
the company they keep". Other theories such as van Tassell (1999) and Laffont and N'Guessan (2000) also conclude that
borrowers will match with borrowers of similar riskiness. Guttman (2006, 2007), however, offers a model that predicts
the opposite: Borrowers of opposite risk types will match with each other. Guttman makes this prediction because he
assumes that borrowers can make side payments to each other to
5
attract group members. While both risky and safe borrowers prefer to have safe peers, the value of a safe peer to a risky
borrower is greater than to another safe borrower. Intuitively, this greater value comes from a diversification benefit of
matching a risky borrower with a safe one. This later conclusion is also supported by my model as well as experimental
evidence (Gine, Jakiela, Karlan, and Morduch, 2010).
Stiglitz (1990) finds group lending's advantage in improving repayment rates come from the peer
monitoring effect, thus limiting the ex ante moral hazard of how group members will use their loans and exert effort. He
focuses on group members being able to observe the effort each applies to her projects and to write enforceable contracts
among themselves.
Ghatak and Guinnane (1999) present a model where verification of states is costly in the spirit of
Townsend (1979), demonstrating that joint liability contracts induce truth telling by borrowers concerning their projects'
payoffs by delegating the auditing function to group members. Therefore, the lender only has to verify the state when the
entire group defaults, thus reducing the number of cases the bank has to incur auditing costs.
While most research deals with comparing available microfinance contracts' impact on welfare, some
offers new kinds of contracts that are currently not observed. Bubna and Chowdhry (2009), for example, offer a new
institution they call, "microfinance franchising" in which a single MFI offers a lending franchise to local capitalists who
compete with a single moneylender. Their model is currently being experimented in Samoa. Ayi (2007) suggests that
MFIs offer a "microequity" contract whereby the MFI has an equity stake in the microenterprises, similar to venture
capitalists.
A similarly titled contract is also suggested by Pretes (2002). The newest stage in the microfinance
movement is that of "microinsurance," which offers insurance contracts to the same people that
microloans are intended for (Morduch, 2004; McGuinness and Tounytsky, 2006; Leftley and Mapfumo).
While all the theory on microfinance loan performance assumes that social capital is pivotal, the way in
which social capital functions differs among the theories. The empirical research also varies in its conclusions regarding
social capital's effect on loan repayment and on borrower welfare.
How aspects of borrowers' relationship to each other and their culture add to or subtract from social
capital is complicated. Studies on the type of social ties show that certain aspects of relationships among borrowers in a
group differ in their effect on borrower behavior. Hermes and Lensink (2007) survey of empirical studies on social
capital's relationship to loan repayment identifies several characteristics of social capital that actually have negative
effects on loan repayment: family membership in group, distance between members, strong social ties, group
homogeneity, relatedness,
6
sharing within group, and high level of joint liability. They also find that some of these characteristics also have positive
effects on loan repayment in other studies, indicating that the relationship between social capital and loan repayment is
actually ambiguous. Similarly another survey concludes, "The results of available empirical studies are contradictory
with respect to virtually all potential determinants of repayment performance" (Guttman, 2006).
Studies show that the joint liability contract is not the only means of harnessing social capital. De
Aghion and Morduch (1998) find that microfinance borrowers in transition economies who borrow on an individual
liability basis still have incentives to repay their loans because of social stigma over default. Also, an individual liability
loan can be marketed to, purchased by, and collected in groups, thereby lowering transactions costs, having a good
information source, fostering education and training, and increasing individuals' comfort with banks. I n a randomized
controlled field experiment, Karlan and Gine (2010) show that repayment rates of borrowers of a Philippine MFI with
branches that are randomly converted from offering offer joint liability to loans to individual liability loans do not differ
from unconverted branches. They find that social capital plays an important role in influencing repayment after branches
are converted to offering individual liability loans.
Studies on the impact of microfinance programs on repayment and on poverty reduction are not
uniformly supportive. However, they find that the impact of microfinance programs is not uniform across borrowers.
Hulme and Mosley (1996) and Morduch (1998) find microfinance programs are less effective in increasing income
among those below the poverty line. In Hulme and Mosley (1996), the researchers suggest that microfinance programs'
impact decreases with falling income because the borrowers below the poverty line take less risks, invest less in
technology, and use their loans to protect their subsistence. In some cases, the loans lower income among the poorest of the
poor (Khawari 2004). Many microfinance borrowers take loans to reduce variation in consumption and not to increase
expected income.
Pretes (2002) criticizes the use of microfinance for certain borrowers. He cites cases where very poor
borrowers become worse-off because of business failure. They attempt to repay their portion of the loan by borrowing
from moneylenders, selling their household assets or food, or leaving their home to find wage labor. Taking these drastic
steps as opposed to just defaulting as would be the case in a country with developed bankruptcy laws may be due to "the
darker side of collective peer pressure." Social sanctions may be so strong that the borrower may be in fear of becoming
an outcast. Also, since most microfinance borrowers are women in countries where they are under a high degree of male
control, the husbands of the borrowers sometimes take the money away, leaving the wife to struggle to find a way to repay
the loan. So, the informal institutions of the community and the home may not allow for microfinance to succeed in
empowering borrowers to take the appropriate risks as an entrepreneur.
7
Pretes postulates that the high repayment rates of microfinance programs are misleading as to their effectiveness in
reaching the very poor, as "financial benefits disproportionately accrue to the middle
poor and do not reach the very poor."
This paper contributes to the literature on microfinance by combining Besley and Coate's model with
other models of microfinance to consider how the group lending system affects peer and project choices. Allowing for
projects of varying levels of risk is an important assumption due to the empirical evidence that project selection matters.
The model's results provide a theoretical explanation for the finding that group lending can have the opposite effect on
investment decisions than is intended. I show how the nature of social sanctions that are levied by the group play a role in
this result.
III. Individual Lending
Consider an entrepreneur's payoff function when liability is individual. In the first period, t = 0, the
borrower receives a loan of l at an interest rate, i. At t = 1, the borrower chooses whether to invest in a safe or risky
project. At t = 2, the investment returns are realized and the borrower chooses whether to repay the loan. Assume that
without a loan, the entrepreneur would not be able to invest in any project. All loans are for the same amount and require a
repayment of r. r = l(1+i). The project returns a random
variable,u . If the risky project is taken, then the outcome has either a high or low payoff (u
L
oru
H
) with ~
equal probability. If the safe project is taken, then the outcome has a payoff ofu
S
with probability of 1.
u
S
= (u
L
+u
H
)/2. After the payoffs are realized, the entrepreneur has the choice of repaying or not repaying the loan. Partial repayment is not a
possibility. If she repays the loan, she has a net payoff of
u
~
- r.
The assumption of either full repayment or total default is made by Besley and Coate (1995). It is a
realistic assumption because these are very small loans, where the amount to be voluntarily paid can be thought of as
discrete. Alternatively, the consequence to defaulting can be thought of as discrete, where any kind of default disqualifies
the borrower from borrowing in the future. Therefore, she would not have any motivation to make a partial repayment.
Also, the model can be adapted to there being
partial repayment in that the borrower faces a potential penalty for strategically defaulting. The
penalty may include collection of a portion of her project's payoffs when she defaults.
If she does not repay the loan, then she incurs costs of various forms. First, the lender may penalize her
for defaulting through refusing to make future loans or sharing this information with other
potential lenders (a credit bureau). Second, the lender may retrieve a portion of what is owed by
litigation. Third, the borrower may face loss of reputation in her community. Fourth, the borrower
8
may expend resources to hide her profits from the lender and her community. Fifth, the borrower
may inflict guilt on herself for not repaying the loan when she has the means to repay it.
Because I am considering lending in a developing country where institutions that allow for contract
enforcement are lacking, the bank is limited in how much it can penalize the borrower. The costs of default
may differ from those in a developed country. First, the loan officer may not be able to credibly commit to
refusing to make another loan and there may not be a credit bureau if there are other lenders in the
market. Second, bringing the case to court may be too costly to the lender for the amount that would be
recovered, and the laws governing collection of debts in developing countries often favor the debtors over
the creditors. Third, though the borrower is individually liable, she may be concerned with what her
community thinks of her if she does not repay a debt. Because the borrowers under consideration typically
live in tight knit communities, the knowledge and opinions of others concerning one's own affairs could be
quite significant. On the other hand, her community may have greater solidarity with her rather than the
bank, thereby causing it to be understanding of a defaulting borrower's non-repayment. Fourth, the borrower
may very easily hide the amount of her project's payoffs from observation if she lives in a remote village.
Fifth, whether borrowers' average conscience in developing countries differs from other borrowers is
unknown.
Assume that the penalty for default is increasing in her project's payoffs. Assume there is a fixed
cost and a variable cost that is increasing in project payoffs. For simplicity, I assume that penalty
function is an affine function in project payoffs. Let of represent the fixed cost and o e (0,1)
~
represent the variable cost per unit of project payoff. The penalty function is defined as p(u )=of +
~
ou . Assume for now that of =0 so that all of the penalty is linear in project payoffs. Making this
assumption will not change the analysis as long as of is not greater than r. I assume that penalties
increase in project payoffs for the following reasons. First, limitation on obtaining future financing
limits the ability of the successful but defaulting entrepreneur from being able to fully utilize her present
project's payoffs. That is, a portion of the value of project payoffs may be in the ability to invest them in
future projects. If additional external financing is also necessary for this future investment and there are
increasing returns to scale, then the value of these payoffs will be less. Second, the greater the payoff, the
greater is the amount that may be retrieved through litigation. Third, the community would probably think
worse of a defaulting borrower the more able she is of repayment. Fourth, the more project payoffs there
are, the more there is to hide; the more there is to hide, the more costly it is to hide. Fifth, the defaulting
borrower would probably think worse of herself the more she is able to repay.
9
The entrepreneur will repay only when the payoff from doing so is greater than not doing so, which
is when u
~
> r/o. See Proof 1.
Therefore, the individual's payoff function is P * (u
~
) = max(u
~
÷ r, (1 ÷ o )u
~
) from implementing the
optimal strategy. Figure 1 graphs the individual lending optimal strategy payoff function.
P(u
~
)
r/ o
u
~
Figure 1. Individual borrower's net payoff as a function of her project's payoffs
(u
~
), assuming penalty parameter (o) and amount owed (r).
Note that this payoff function is weakly convex in u
~
:
Proposition 1: An individual liability borrower's optimal repayment strategy (s*)
implies the utility (U
I
) of the individual liability loan to be convex in her project's
~
payoffs (u ). See Appendix for proof.
As can be seen in Figure 1, if the entrepreneur were to select between the safe and risky projects,
then she would certainly choose the risky project if u
H
> r/o> u
L
:
Proposition 2: An individual liability borrower's optimal investment decision (p*) is
to take the risky project (R) rather than the safe project (S). See Appendix for proof.
10
However, if u S> r/o, then the lender would prefer that the safe project be chosen, in which case its
expected repayment would be r, rather than the risky project, whose expected repayment is r/2:
Proposition 3: The expected repayment rate of an individual liability loan is 50
percent. See Appendix for proof.
Typically, in developed countries' markets for loans where there are significant information
asymmetries, credit rationing is understood to occur due to the moral hazard at the investment choice
stage induced by setting interest rates high enough to compensate for the prior riskiness of the borrower
(Stiglitz and Weis 1981). Because this agent's action in this moral hazard problem occurs before the project
payoffs are realized, this type of moral hazard is referred to as ex ante moral
hazard. This is demonstrated in Proof 2, where the borrower would choose to invest in the risky project that
has only a 50 percent chance of her repaying the loan instead of the safe project, which the bank would
prefer her to take. The bank is limited in raising r to compensate for this problem because the critical point,
r/o, for the borrower to choose to repay the loan increases in r. However, if o is higher, then the expectation of repayment is
higher because the critical point for the repayment decision, r/o, is lower and the fraction of payoffs recoverable in default is
higher. Therefore, the bank will limit its losses through controlling the quantity of credit supplied rather than in
price.
In undeveloped economies with weak institutions, another type of moral hazard is introduced:
whether or not to default when the borrower is in fact capable of repaying the loan (Besley and Coate 1995).
This other type of moral hazard is referred to as either ex post moral hazard, strategic default, or
unenforceability. Ifo is higher, then the expectation of repayment is higher because the critical point for the repayment
decision, r/o, is lower and the fraction of payoffs recoverable in default is higher. This can be seen in the proof for
Proposition 1.
IV. Joint Liability Lending
Next, consider an entrepreneur's payoff function when the loan is made to a group. I model group lending
in a two-player (like Besley and Coate), three-period (t = {0,1,2}) game in which both borrowers choose
whether or not to contribute to the repayment of their loan. The two borrowers are identified as "Borrower j"
where j = 1 or 2. In the first period, t = 0, the two-member group forms and receives a loan. The loan
agreement stipulates that the group must pay a total of 2r in principal and interest in the last period, t = 2.
(Under individual lending, each borrower would have to pay r.)
At t = 1, each borrower invests her share of the loan in a project that yields a random variable, u
~
j .
She chooses between two investments, safe or risky.
11
At t = 2, each group member decides whether or not to contribute to the repayment of the loan.
Following the assumption of Besley and Coate (1995), the group as a whole can either default on the entire
loan or repay the entire amount. If each member does contribute, then borrower j has a payoff
of u
~
j -r. If one contributes nothing, then the other borrower can either decide to also contribute
nothing, thereby allowing the group to default, or to cover both members' share and repay the entire
loan. If the group defaults, then, the borrowers are penalized by the same amount as under
~
individual liability (p(u
j
)). The penalties are increasing in u
~
j . Therefore, if the group defaults, then
each member gets a payoff of u
~
j (1-o).
IV. A. Joint Liability Lending without Social Sanctions
It has already been recognized that simply making borrowers jointly liable for each other's loan does not
have strictly positive net effects on borrower repayment (Besley and Coate 1995). Furthermore, it has been
shown empirically in the Philippines that joint liability is not the only method by which social capital can
impact the probability of repayment by groups (Karlan, et. al.). Therefore, the first kind of group lending that
I consider is one where the group members are jointly liable for each other but cannot impose any kind of
penalty for non-contribution to the repayment of the loan. I call this type of harnessing of social capital as
joint liability lending without social sanctions. By making this distinction, I can separate out the effects of
joint liability itself from other factors on the enforcement mechanism.
Optimal Strategies at t=2
The two group members have to choose a strategy of either contribute (C) or not contribute (NC).
Let sj e {C,NC} denote the strategy played by borrower j. Each borrower's net payoff depends on the
strategy played by the peer. Denote borrower 1's net payoff as U
J
1
(s,u
~
) ÷ U
J
1
[(s
1
, s
2
),(u
~
1,u
~
2 )] .
Their payoffs from each possible strategy are denoted in Figure 2.
Borrower 2 Strategy (s2)
Borrower 1
Strategy (s1)
C
NC
C
u
~
1 ÷ r , u
2
÷ r
u
~
1 , u
~
2 ÷
2r
NC
u
~
1 ÷ 2r , u
~
2
u
~
1(1 ÷ o ) , u
~
2 (1 ÷ o
)
12
Figure 2. Joint liability without social sanctions net payoffs
U
J
1
(s,u
~
) , U
J
2
(s,u
~
) under the four possible strategy combinations
of both borrowers in a group.
Project returns, u
~
1 and u
~
2 , determine the payoffs of each combination of strategies. Therefore, the
~ ~
Nash equilibria vary
by the realizations of u and u . The optimal strategies, then, proven in
1 2
Lemmas 1.1 - 1.4 in the Appendix. Figure 3 presents these optimal strategies in a table.
u
~
2
u
~
2 < r/o r/ os u
~
2 < 2r/o
u
~
2 > 2r / o
u
~
1
< r/o
(NC, NC) (NC, NC) (NC, C)
u
~
1
r/ os u
~
1
< 2r/o
(NC, NC) (C , C ) (NC, C)
u
~
1
> 2r/o
(C, NC) (C , N C ) (C, NC), (NC, C)
Figure 3. Nash equilibrium strategies (s1*, s2*) for various project under joint liability without social
sanctions.
The impact of joint liability without any social sanctions on bank repayment appears ambiguous. On
one hand, there is a positive effect of joint liability on borrower repayment. Suppose the bank makes two
loans on an individual liability basis to the two members in this group. If one's project payoffs were less
than r/o but the other's were greater than 2r/o, then only one loan would have been repaid. However, under joint liability,
the borrower with the project with the higher payoffs will pay for her peer's loan. The bank then has both
loan repaid under joint liability. This positive effect on the probability of repayment of loans is the risk-
sharing effect.
Consider now a different scenario: Borrower 1 has payoffs greater than r/o but less than 2r/o, and
Borrower 2 has payoffs less than r/o. Under individual liability one loan is repaid. Under joint liability, neither loan is
repaid because Borrower 1 will not cover for Borrower 2. This negative effect on the probability of
repayment of loans is the free-riding effect.
A borrower's expected payoff function under optimal strategies, then is more complicated than under
individual liability because her net payoffs and optimal strategies depend on the payoffs of the other
borrower's project and her strategy. The dominance of the risk sharing effect or the free riding effect
13
depends on the project payoff. If her peer contributes to the repayment of the loan, Borrower 1, in
~
contrast to the individual liability case, keeps more of her project's payoffs in low payoff states (u
1
<
r/o) due to the risk-sharing effect, has the same net payoff in intermediate payoffs (r/os u
~
1 <
2r/o),
~
and keeps all of her project's payoffs in high payoffs (u
1
> 2r/o) due to the free-riding effect.
However, if her peer does not contribute to the repayment of the loan, Borrower 1 keeps the same
amount of the loan for low payoffs (u1< r/o), has lower net payoffs under intermediate payoffs (r/os
u
~
1 < 2r/o) due to the coordination failure induced by free-riding, and also has lower net payoffs
under high payoffs states due to being exploited by free-riding. These payoffs are represented in
Figure 4.
U
J
1
[(s
1
*, s
2
),
(u ,u
~
)] ~
1 2
s
2
= C
r/o
2r/o
s
2
= NC
u
~
1
Figure 4. Optimal strategy net payoff functions of Borrower 1. The solid lines
represent P*1(u1|C), the optimal strategy net payoff if Borrower 2 contributes to repayment. The
dashed line represents P*1(u1|NC), the optimal strategy net payoff if Borrower 2 does not
contribute to repayment.
The uncertainty surrounding what a borrower's payoffs does not only stem from the payoffs of her
peer's project, but also on which equilibrium strategy they play. When both projects have high
levels of payoffs, they play either (C, NC) or (NC, C). This surprising result that one borrower allows the
other to free-ride on her comes from neither borrower credibly being able to commit to the group to default
This leaves a question of which equilibrium strategy will be played when project payoffs are in these
ranges. If the game were moved from static to dynamic or a convention were applied to
14
it, then the Pareto improving strategy may be played more often. For the purposes of a borrower's
prior beliefs about which equilibrium strategy will be played, I will assume that the probability of 0.5
for each strategy when there are two equilibrium strategies.
3
Optimal Strategies at t=1
Each borrower will uses her expectations of her own project's payoffs and her peers at t=1 when she
chooses what kind of project in which to invest. Assume that there are two projects to choose from, as
under the individual liability case. For simplicity, add the assumption that the safe investment's payoffs are
strictly less than 2r/o. Does Borrower 1 choose the risky or the safe investment if Borrower 2 invests in the safe
investment? What does Borrower 1 choose if Borrower 2 invests in the risky investment.
To address the first question, assume that Borrower 2 invests in the safe investment. This means
that Borrower 2's project yields u
S
, where 2r/o> u
S
> r/o. If Borrower 1 invests in the safe project,
then her project's payoff is in the same range, and they would play (C, C). Borrower 1's expected net
payoff, then, is u
S
- r. If Borrower 1 were to invest in the risky project, then the group will default if
her project paysu
L
, or it will repay the loan with her paying all of it if her project paysu
H
. Her
expected payoff is u
S
-ou
L
/2 - r, which is less than u
S
- r, the net payoff from taking the safe project.
She, therefore, prefers to take the safe project:
Proposition 4: Under a joint liability contract, if one borrower invests in the safe
project, then the other borrower will also invest in the safe project (S) rather than the risky
project (R). See Appendix for the proof.
Next, consider what Borrower 1's net payoffs would be under the two investment options if Borrower
2 takes the risky project. There are now five possible net payoffs if she takes the risky project shown in
Figure 5.
3
One possible rule of the game is for each borrower to flip a coin when they come
together at t=2 to determine who plays her strategy first. If both have payoffs greater
than 2r/ , then the loser of the coin flip goes second and therefore covers the entire
loan.
15
u
~
1
,u
~
2
Possibilities
NE Strategies
(NC, C)
~
P
1
*(u
1
|s
2
)
u
1
H
Probability
0.125
u
1
H
,
u
~
2
u
1
H
,u
2
H
(C, NC)
(C, NC)
u
1
H
- 2r
u
1
H
- 2r
0.125
0.25
u
~
1
u
1
H
, u
2
L
u
1
L
,u
2
H (NC, C) u
1
L 0.25
u
1
L
, u
~
2
u
1
L
,u
2
L (NC, NC) u
1
L
(1 -o) 0.25
Figure 5. Optimal strategy net payoffs to Borrower 1 when both borrowers take the risky project.
The expected payoff for Borrower 1 in taking the risky project when Borrower 2 does
the same isu
S
-ou
L
/4 - 3r/4. If Borrower 1 takes the safe project when Borrower 2 takes the risky, then
Borrower 1 has possible payoffs shown in Figure 6 instead. Her expected net payoff from
taking the risky project would beu
S
-ou
L
/2.
u
~
1
u
~
1
,u
~
2
u
1S
,u
~
2
Possibilities
u
1
S
,
u
1
S
,
u
2H
u
2L
NE Strategies
(NC, C)
(NC, NC)
~
P
1
*(u
1
|s
2
)
u
1
S
u
1
S
(1-o)
Probability
0.5
0.5
Figure 6. Optimal strategy net payoffs to Borrower 1 when Borrower 2 takes
the risky project and Borrower 1 takes the safe project.
16
Her payoff from choosing the safe project over the risky project isu
S
-ou
S
/2 - [u
S
-ou
L
/4 - 3r/4] =
3r/4 -ou
L
/4. Comparison of the expected utilities yields the following result:
Proposition 5: Under a joint liability contract, if one borrower invests in the risky project, then the
other borrower will invest in the risky project only if her projects' high state payoff is greater than
3r
.
Otherwise, she will invest in the safe project. See Appendix for the proof.
o
Each borrower, therefore, knows that if she takes the safe project, then the other borrower will take
the safe project, also. If she takes the risky project, then the other borrower will take the safe project only
ifu
H
s 3r/o and will take the risky project only ifu
H
> 3r/o. The borrowers investing in different
projects, however, is not an equilibrium strategy because if Borrower 2 first chooses a risky investment
and Borrower 1 responds by choosing a safe investment becauseu
H
s 3r/o, then Borrower 2 would reverse her
investment decision to the safe investment as shown in Proof 4. Therefore, ifu
H
s 3r/o, then both borrowers choose
to make the safe investments; and ifu
H
> 3r/o, then both
borrowers choose to make the risky investments. In other words, if the mean payoffs of the
investments are higher, then the borrowers will choose risky strategies.
Optimal strategies at t=0
If the payoffs of the possible projects differ between the group members, then borrowers withu
H
s3r/o, who
always prefer the safe project select other borrowers who would prefer the safe project, too. Likewise, the borrowers withu
H
>
3r/o select other borrowers with the same possible high payoff state. Therefore, when there are no social sanctions on
noncontributing group members, there is
assortive matching of borrowers, consistent with Ghatak and Guinnane (1999) and Ghatak (1999):
Corollary 1: Borrowers with high possible project payoffs (uH > 3r
)
select each
a
other to take joint liability loans and invest in risky projects. Other borrowers
(uH < 3r
)
select each other and invest in safe projects. See Appendix for the proof.
a
Optimal group project selection under joint liability without social sanctions is peer dependent. That
is, even if the project opportunity sets between the two borrowers differ, each borrower's strategy is
dependent on whether at least one borrower always prefers to take the safe investment.
Does the lender choose to make this joint liability loan rather than two individual liability loans? If
u
H
s 3r/o, then both borrowers play "safe", which implies a repayment probability of 1. Ifu
H
> 3r/o,
17
then both borrowers play risky, which implies a repayment rate of (2r * 0.75 + 0 * 0.25)/(2r) = 0.75.
These repayment rates are improvements over the repayment rate of making two individually liable
loans: (2r * 0.25 + r * 0.5 + 0 * 0.25)/2r = 0.5 because the individually liable borrower will always
choose the risky investment. Since the lender has a higher probability of being repaid under joint liability,
borrowers benefit from reduced credit rationing and reduced interest rates. This is shown
formally in the proof for the following proposition in the Appendix:
Proposition 6: The expected repayment rate of a joint liability loan is between 75
and 100 percent.
Presuming the same availability and terms of credit, does the borrower choose to accept a joint
liability loan over an individual liability loan? Ifu
H
s 3r/o, then Borrower 1 has an expected payoff of u
S
- r if she has a joint
liability loan because both take the safe projects. This is less than the expected payoff from taking an individual liability
loan. See Proof 2. Ifu
H
> 3r/o, then Borrower 1 will have an expected payoff ofu
S
-ou
L
/4 - 3r/4. This net expected payoff
is also less than that under individual liability. The difference in net expected payoffs of the individual liability
loan over
the joint liability loan is [u
S
-ou
L
/2 - r/2] - [u
S
-ou
L
/4 - 3r/4] = (r -ou
L
)/4 > 0 sinceu
L
< r/o.
Therefore, if equal in terms and accessibility, microentrepreneurs would prefer the individual liability loan.
See Proposition 12 in the Appendix with proof.
The insight that the joint liability contract without social sanctions induces the borrowers to take
safe investments when the payoff possibilities are lower implies that the risk sharing effect does dominate
the free riding effect. This result contradicts Besley and Coate (1995), who argue that the ability of
borrowers to level punishments on one another is a necessary element to guarantee superior repayment
rates for joint liability loans. I find this different result because I relax their assumption of project choice
being exogenous.
IV. B. Joint Lending with Social Sanctions
Now, I allow group members to penalize each other if one does not contribute to the repayment of the
loan. Following Besley and Coate (1995), social sanctions are a function of payoffs of the
noncontributing borrower, and not on observation of effort or project selection. Social sanctions can be in
the form of loss of reputation in the community or inability to participate in future loans. "Social sanctions"
can also be seen as the group member's internal sense of obligation or guilt or for not contributing to the
repayment of the loan even if the community completely forgives her. With an outside, impersonal
institution, she may not have such guilt for defaulting. Social sanctions' sensitivity to realized returns are
assumed to be more punitive than the bank penalty functions; this
18
belief is the basis for arguments that the cultural realities of these borrowers can induce them to
repay their loans better than what a lender outside the community has at its disposal. The structure of the
social sanctions function, s(uj), is similar to that of the bank penalty function:
s(u
~
j ) = s
f +
o
u
~
j , where sf> af and
|
s|s 1. 1/| represents the increased degree by which the
contributing borrower can penalize the noncontributing borrower over that of the bank. A larger|
implies relatively greater leniency by the group.
Optimal Strategies at t=2
A group member's payoffs decrease by s(u
~
j ) only when the other group member repays the loan and
the member being sanctioned does not contribute. If the loan is not repaid (i.e., neither borrower
contributes), then they do not sanction each other, but they are both penalized by the bank. Continue to
assume that af = 0, and also assume for now that sf = 0. The borrowers' payoffs to each
pair strategies are given in Figure 7.
Borrower 2 Strategy (s2)
Borrower 1
C
C NC
u
~
-r, u
~
-r
1 2
u
~
-2r, u
~
(1-o/|)
Strategy (s1)
NC
u
~
1
(1-o/|),
u
~
2 -
2r
1
u
~
1
2
(1-o),
u
~
2 (1-
o)
~
Figure 7. Joint liability with social sanctions net payoffs (U
JS
1
(s,u ) ,
~
P2(u
2
|s1)) under the four possible strategy combinations of both
borrowers in a group.
Project returns, u
~
1 and u
~
2 , determine the payoffs of each combination of strategies as in the
previous considered case. Therefore, the Nash equilibria vary by the realizations of u
~
1 and u
~
2 .
The
optimal strategies are in Lemmas 3.1 - 3.4 in the Appendix with proofs and summarized in a table in
Figure 8. The optimal strategies when borrowers can sanction each other are less likely to be
dominated by free-riding. When both borrowers have payoffs in excess of|r/o, then the entire group
repays the loan. Furthermore, since|s 1, the lender is more likely to be repaid than in the
individual liability case, where the critical point for individual repayment is r/o>|r/o. However,
joint liability with social sanctions still suffers from the problem of a borrower with medium level
payoffs not contributing her portion when her peer's project has very low payoffs.
19
u
~
2
u
~
2 <|r/o |r/os u
~
2
< 2r/o u
~
2 > 2r / o
u
~
1
<|r/o
(NC, NC) (NC, NC) (NC, C)
u
~
1
|r/ os u
~
1
< 2r/o
(NC, NC) (C, C) (C, C)
u
~
1 > 2r/o (C, NC) (C, C) (C, C)
Figure 8. Nash equilibrium strategies (s1*, s2*) for various project payoffs under joint liability with
social sanctions.
Figure 9 graphs the payoffs of these optimal strategies. The new payoff functions with social
sanctions differ from those without. The joint liability loan without social sanctions does have
payoffs that exceed that of the social sanctions case and the individual liability case due to the reduction
in the free-riding effect. Borrowers are more likely to both contribute to the repayment of the loan for high
realizations of project payoffs. However, there are no net payoffs under joint liability with social sanctions
that exceed that of the individual liability case.
P*
1
(u
~
1 |s
2
)
s
2
= C
|r/o
r/o
2r/o
s
2
= NC
u
~
1
Figure 9. Optimal strategy net payoff functions of Borrower 1 under joint
liability with social sanctions. The solid line represents P*1(u
~
1 |C), the optimal
strategy net payoff if Borrower 2 contributes to repayment. The dashed line
represents P*1(u~
1
|NC), the optimal strategy net payoff if Borrower 2 does not
contribute to repayment.
20
Optimal Strategies at t=1
As in the other joint liability case, each borrower will use her expectations of her own project's payoffs and
her peers' at t =1 when choosing what kind of project in which to invest. Assume that there are two projects
to choose from as before and that that the safe investment's payoffs are strictly less than 2r/o. Does Borrower
1 choose the risky or the safe investment if Borrower 2 invests in the safe investment? What does Borrower 1
choose if Borrower 2 invests in the risky
investment?
First, consider what possible net payoffs Borrower 1 faces if Borrower 2 chooses the safe investment.
If Borrower 1 chooses the safe investment also, then they would play (C, C), and Borrower 1 has a net
payoff ofu
S
- r for sure. If Borrower 1 chooses the risky investment, then she has a net payoff of
u
H
- r if the project paysu
H
, and she has a net payoff of eitheru
L
- r oru
L
(1-o/|), depending on
whetheru
L
is greater than or less than|r/o, respectively. Ifu
L
>|r/o, then her expected net payoff is
(u
H
- r)/2 + (u
L
- r)/2 =u
S
- r, making her indifferent between taking the safe project and the risky
project. In either case, both she and her peer contribute to repayment. If u
L
<|r/o, then her
expected net payoff is (u
H
- r)/2 +u
L
(1-o/|)/2 =u
S
- ou
L
/(2|) - r/2. In this case, she will choose the
risky investment:
Proposition 7: If borrower 2 invests in the safe project, then borrower 1 will invest
in the risky project. See Appendix for the proof.
Next, consider what possible net payoffs Borrower 1 faces if Borrower 2 chooses the risky
investment. If Borrower 1 also chooses the risky investment, she faces four possible net payoffs. Figure 10
shows the net payoffs for that case. In this case, there are possibilities that Borrower 1 will either depend
on her peer to cover her portion of the loan or she will cover her peer's portion of the loan. The expected net
payoff from taking the risky investment when Borrower 2 also takes the
o (1÷ 1 )u
1L
+ 3r
risky investment is EU
JS
R1 R = u
1S
÷ |
4
. If Borrower 1 takes the safe project, however, then
she has possible net payoffs shown in Figure 11. The expected net payoff from taking the safe
EU
S
JS
1
= u
1S ÷
ou
1
+ r
.
S
investment when Borrower 2 takes the risky investment is
R
2
21
The difference in taking the risky project over taking the safe project when Borrower 2 takes the
EU
JS
R
1
÷ EU
S
JS
1
= 2ou
1
÷ (| + 2)r
,
which is greater than zero only if S
risky project is
R R
4
u
1S
> (| + 2)r . Therefore, Borrower 1 chooses the risky project when Borrower 2 chooses a risky
2o
project if
u
1S
> (| + 2)r
2o
and the safe project otherwise:
Proposition 8:
If
u
1L <
|r
and
o
borrower 2 invests in the risky project, then
borrower 1 will take the risky project only if
u
1S
>
(
2 + | )r
2o
. Otherwise, borrower 1
will take the safe project. See Appendix for the proof.
Project 1 Payoff NE Strategies P
1
*(u
1
|s
2
) Probability
u
1
H
,u
2
H (C, C) u
1
H
- r 0.25
u
1
H
,u
2
u
1
H
,u
2
L (C, NC) u
1
H
- 2r 0.25
u
1
u
1
L
,u
2
H (NC, C) u
1
L
(1-o/|) 0.25
u
1
L
,u
2
u
1
L
,u
2
L (NC, NC) u
1
L
(1 -o) 0.25
Figure 10. Optimal strategy net payoffs to Borrower 1 when both borrowers take the risky
project, there are social sanctions, andu
L
s|r/o.
22
Project 2 Payoff NE Strategies P
1
*(u
1
|s
2
) Probability
u
1
S
, u
2H
(C, C) u
1
S
- r 0.5
u
2
u
1
S
, u
2L
(NC, NC) u
1
S
(1-o) 0.5
Figure 11. Optimal strategy net payoffs to Borrower 1 when Borrower 2 takes the risky
project, Borrower 1 takes the safe project, there are social sanctions, andu
L
s|r/o.
Optimal Strategies at t=0
The following lemmas identify which peers each type of borrower prefers according to her expected
project payoff:
Lemma 4.1: For the borrower who will always choose the risky project, i.e. those
who have project expected payoffs greater than
(| + 2)r
,
it is preferable for her to
2o
find a peer who will choose the safe project. See Appendix for the proof.
Lemma 4.2: For the borrower who will invest in the safe project when her peer
invests in the risky project, it is preferable for her to find a peer with low expected
payoffs, i.e. between
r
and
(| + 2)r
.
See Appendix for the proof.
o 2o
By consequence of these preferences, borrowers with similar project expected payoffs will match
together, as stated in the following proposition:
Proposition 9: Borrowers will match with other borrowers with the same expected
project payoffs where there is a possibility of social sanctions. See Appendix for the proof.
23
As a consequence, there are is separation in investment strategies according to the groups' projects'
expected payoffs:
Proposition 10: The only investment strategies that will be played are P1=P2=R for
groups with high expected project payoffs and P1=P2 for groups with low
expected project payoffs. See Appendix for the proof.
Would we see groups who play opposite investment strategies? Guttman (2006, 2007) show that if
group members can select each other and they can make side payments to each other, then the risky borrowers will
match with safe borrowers. My model is consistent with this finding because groups of borrowers with low expected
payoffs play opposite investment strategies. However, my model does not identify which borrower would invest in
which type of project, though the possibility of side payments could resolve such a question as long as each borrower
has an equal probability of offering the side payment and receiving it.
In the above scenario, the repayment rate is 0.5:
Proposition 11: The expected repayment rate of a joint liability loan with the
possibility of social sanctions is between 50 percent and 75 percent. See Appendix for the proof.
Therefore, the repayment rate of joint liability contracts with social sanctions is greater than that of
the individual liability contract. However, if all borrowers have average project prospects that are sufficiently low, then
the repayment rate is no greater than the individual liability's. See the proof for
Proposition 11. Therefore, we see a case where a joint liability contract might be inferior to an
individual liability contract because both the lender will not be any better off, and the borrower will be worse off.
Note that the conditions for the mixed investment strategy of borrowers, for whomu1Ss (2 +|)r/(2o),
implies that the more stringent the social sanctions (lower|), the less likely it is that the mixed
investment strategy will occur because the minimum threshold for taking the (Risky, Risky) strategy
is lowered. Therefore, though joint liability with social sanctions leads to lower repayment rates
than a joint liability without them, stronger social sanctions do make it less likely that the strategies of (Risky, Safe)
will be played.
24
V. Welfare Comparisons and General Equilibrium Extension
The ultimate question for microfinance institutions, governments, and NGOs is what type of contract
maximizes welfare? The microfinance contract that maximizes societal welfare is one that increases the expected net
payoff to one borrower without decreasing the expected repayment to the lender.4 Recent experience with individual
liability contracts supports my contention below that the
individual liability contract is often superior to the joint liability contract. It has been found in
practice as well as in controlled experiments that MFIs that switched from joint liability to individual liability loans
did not experience a decrease in loan repayments.
Social capital, therefore, has a "dark side" of inhibiting some borrowers from taking more risk, which
does not improve repayment rates because of the enforceability problem that exists in the countries these loans are made
in. These results provide some counter predictions to the usual beliefs that group lending programs improve repayment
rates because the group is able to penalize its members by taking away some of their social capital. Furthermore, these
results give a theoretical explanation for anecdotal evidence of the negative welfare effects of microfinance programs on
very
poor borrowers (Khawari 2004, Hulme and Mosley 1996, and Pretes 2002). These borrowers in
particular seem to pick relatively overly safe projects. The empirical evidence shows it is the very poor for whom
microfinance does not work as well as intended. The results presented here show group lending produces the same
choice by borrowers facing project opportunity sets with a low expected outcome. If borrowers' incomes are positively
correlated with the project opportunity sets available to them, then these results may explain the disparate impact of
microfinance programs. Interestingly, social sanctions may work too well by making borrowers too scared to take on an
optimal amount of risk.
V.A. Only One Contract Is Offered
Inspection of Figures 1, 4, and 9 reveals that for the same loan availability and terms, borrowers' expected net payoffs
are higher under individual liability contracts for all projects than under either joint liability cases.
Proposition 12: In terms of borrower expected utility, the individual liability
contract weakly dominates the joint liability contracts assuming the same
4
An additional criterion for maximum societal welfare is the degree that business activity has a positive externality. If the
microentrepreneurs invest loans by purchasing capital and hiring labor within their community and the risky projects require
greater investment than the safe project, then this externality effect would also suggest that more risk taking by, Ceteris Paribus, would
increase welfare. I show later that if such an externality exists, then my conclusions are even stronger.
25
principal and interest across contracts (i.e., same r) and only one contract is offered.
See Appendix for the proof.
However, I also show that the repayment rate under joint liability without social sanctions is higher
by 25 to 50 percent than the repayment rate under individual liability. This increase in repayment probability is due to
the risk sharing benefit when risky projects are taken or to the incentive to take safe projects when the payoffs to the
risky investment are not very high. Because the lender can expect to be repaid with a higher probability, it will make
loans available to more borrowers by rationing credit less and/or reducing the interest rate.
If borrowers are able to sanction each other, however, there actually may be no effect on the
repayment rate over the individual liability contract if the sanctions are not sufficiently strong and average payoffs are
not sufficiently high. This surprising conclusion is due to the lack of risk-sharing benefit by groups forming whereby
only one borrower takes the risky project. This is worse than if the borrowers both chose to take the risky project
because if only one borrower takes the risky
project and the project paysuL, then there will be no chance of the borrower taking the safe project
being able to cover for her. The costs of being punished by one's peer are what drive this result.
The question has been raised in microfinance circles of whether the joint liability feature of group
lending is what harnesses the social capital that leads to high repayment rates. I also demonstrate how joint liability
contracts can induce greater repayment without peers being able to sanction each
other. The mechanism by which joint liability works to increase loan repayment is not through
inducing peers to punish each other, but rather through borrowers sharing risk or through cooperatively avoiding risk.
If peers can punish each other, however, joint liability does not necessarily work better than individual liability loans.
The dominance of the individual liability contract with the constraint that the lender can only offer
one contract is demonstrated in the proof of the following proposition, which is in the Appendix:
V.B. A Choice of Contract Is Offered
The microfinance industry has developed to the point that microborrowers have options of taking an
individual liability or joint liability contract. In this section, I allow the amount due, r, to vary
across contract types, which is dependent on the types of borrowers who may separate into either type of contract. It
also influences which borrower will take a particular contract type.
First, I compare the individual liability loan to the joint liability loan without the possibility of social
26
sanctions. The following proposition is obtained:
Proposition 14: If given the choice between an individual liability and joint liability
contract with the possibility of social sanctions, borrowers with
u
iH
< 3r
o
choose the
individual liability contract and those with
u
iH
> 3r
o
choose the joint liability contract.
The total principal and interest due on the individual liability contract is 150% of that of
the joint liability contract. See Appendix for the proof.
Proposition 14 implies that borrowers with better prospects (u
iH
>
3r
)
prefer the joint liability contract
o
because the interest rate is lower because the lender knows that peers share risk with them.
Borrowers with lesser prospects (u
iH
<
3r
)
prefer the individual liability contract because the cost
o
imposed by the lender is less than the cost of taking the safe project.
By offering the individual liability contract in addition to the joint liability contract, the borrowers
who otherwise would have invested in the safe projects under joint liability now invest in the individual liability
contract. Therefore, everyone invests in the risky project.
Next, I compare the individual liability loan to the joint liability loan where there is a possibility of
social sanctions. The following lemma and proposition are obtained:
Lemma 5: If given the choice between an individual and joint liability contract
where there will be social sanctions, borrowers for whom p*=(R,R) will choose the
individual liability contract. See Appendix for the proof.
Proposition 15: If given the choice between an individual liability and joint liability
loan where there is a possibility of social sanctions, no one will take the joint liabi lity
contract. See Appendix for the proof.
Therefore, the dominance of the individual liability contract is maintained where there would be
social sanctions under a joint liability contract under the general equilibrium assumptions and assumption A8 that
expected project payoffs are always greater than r/o. If this assumption is relaxed, however, then the lower interest rate
induced by the possibility of higher repayment under
27
joint liability would cause more people to take these loans who otherwise would not borrow at all.
VI. Empirical Hypotheses
The conclusions of this study imply some empirical hypotheses using data on the MFI level. Using
this level of data is newer in microfinance research because these data have only become available within the last
decade. One seminal paper is Cull, Demirguc-Kunt, and Morduch (2009), which studies the financial and customer
demographic ratios of MFIs as reported by www.mixmarket.org. They study such key questions in microfinance as
what is the potential tradeoff between MFI sustainability and reach to the poorest borrowers. Cull, et. al. (2009) find
that contract design substantially impacts MFI profitability, loan repayment, and costs. They find that MFIs that make
individual liability loans experience increases in portfolio quality and profitability when they raise interest rates.
Whereas they compare contract types of individual versus group based lending, the theory presented here predicts
different results among group based lenders based on the social ties of their customers. The difference arises from
varying strengths and types of informal institutions that impact how group members respond to non-paying peers.
The data used by Cull, et. al. (2009) could be augmented with measures of formal and informal
institutions within in each of the MFIs' markets. If these measures can be obtained and MFIs are identified by their mix
of making individual and joint liability contracts, then repayment rates across the three kinds of lending discussed here
could be compared.
This study also has implications for testing using micro level data, which is what the majority of
studies have used. If a measure of the project opportunities available to microborrowers can be collected, then there are
several more testable hypotheses. Microfinance borrowers with higher (lower) possible project payoffs are expected to
be more like likely to find peers with similarly higher (lower) possible project payoffs. Microfinance borrowers with
higher and more varied possible project payoffs are expected to be more likely to match with peers with lower and less
varied possible project payoffs. Joint liability borrowers are expected to switch to riskier projects if they switch to
receiving individual liability loans, especially if they have project opportunities with low upside payoffs.
The empirical testing of this model entails data collection challenges. First, measures of forbearance
by culture would require conducting surveys among every identifiable culture served by the MFIs
being studied. Second, measuring project possibilities directly also would require reliance on
surveying microborrowers.
VII. Conclusion
Microfinance has popularly been touted as a program that succeeds in improving its borrowers'
28
incomes by overcoming the moral hazard problem inherent in individual liability loans in countries with poor
institutions. This paper contributes to theory as to why microfinance does not always work as intended. Furthermore, it
shows that both formal and informal institutions matter, which could lead to certain policy prescriptions according to a
country's institutions.
Given that the whole purpose of microfinance is to promote entrepreneurship, which is inherently a
risky project, and if the borrower's prospects are low, then she will not use the funds to make a business grow. Rather,
she will use it for some other purpose, such as income smoothing. She may even forgo her entrepreneurial pursuits in
order to be certain to pay back the loan: There have been stories of borrowers taking on jobs in cities just to repay their
portions of loans rather than working on their own businesses. If the real need of these individuals is insurance, then
they may benefit more by microinsurance programs instead of microlending ones.
On the one hand, the presence of social sanctions in joint liability contracts may inhibit
entrepreneurial activity among people who would otherwise take business loans as individuals. On the other hand, joint
liability contracts can be made to groups with social capital but low payoff project opportunities would otherwise not
borrow money at all.
There are several testable empirical hypotheses implied by this model that may be tested. First,
MFIs that make individual liability loans should have lower repayment rates than those that make joint liability loans if
the group members are unlikely to punish one another for non-contribution. Second, MFIs that make individual liability
loans should have the same repayment rates as those that make joint liability loans where group members are likely to
punish one another for non- contribution and project payoffs are sufficiently high. To test these first two predictions,
one would regress repayment rate at the MFI level on the composition of individual liability contracts to total
contracts, the average measure of social ties among borrowers' communities, and the average measure of potential
project payoffs to the borrowers. Tests using this regression would be most powerful with a sample of MFIs that serve
identifiably specific types of borrowers by culture and economic status. Third, microfinance borrowers with higher
(lower) possible project payoffs will be
more likely to find peers with similarly higher (lower) possible project payoffs. Fourth,
microfinance borrowers with higher and more varied possible project payoffs will be more likely to match with peers
with lower and less varied possible project payoffs. Fifth, joint liability loan borrowers will be more likely to switch to
riskier projects if they switch to receiving individual liability loans. These third, fourth, and fifth predictions would be
tested using individual borrower data, which would depend more heavily on specialized surveys than the methodology
for testing the first two predictions.
29
Appendix
Assumptions:
1. Ordering of Payoff Possibilities for Borrower i:
0 s u
i L
< u
iS
< u
i
H (A1)
2. Payoff Probabilities Conditional on Investing in the Risky Project:
pr(u
~
i = u
iL
| p
i
= R) = pr(u
~
i = u
iL
| p
i
= R) = 1
2
3. Payoff Probability Conditional on Investing in the Safe Project:
pr(u
~
i = u
iS
| p
i
= S) = 1
4. Equivalence of Expected Payoffs of Both Projects:
(A2)
(A3)
E(u
~
i | p
i
= R)
=
u
i
+ u
i
= u
i
S
L
2
H
(A4)
5. Bank Penalty Parameter Bounds:
0 <o s1
6. Social Sanctions Parameter Bounds:
o s | s1
7. Low State Payoff Bounds:
(A5)
(A6)
0 s u
iL <
|r < r
(A7)
o o
8. Expected Payoff Bounds for Both Projects:
r < u
S
< 2r i
(A8)
o o
9. High State Payoff Bounds:
30
2r < uH
o i
(A9)
Proposition 1: An individual liability borrower's optimal repayment strategy (s*) implies the
~
utility (U
I
) of the individual liability loan to be convex in her project's payoffs (u ).
Proof:
~~
U
I
[s(u ),u ] is the utility of individual liability borrower from playing s e (C, NC) and realizing
outcome u
~
.
U
I
(C, u
~
) = u
~
- r (1)
U
I
(NC, u
~
) = u
~
(1-o) (2)
If
u
~
>
r
o
, then U
I
(C,
u
~
) - U
I
(NC, u
~
) > 0.
Therefore, s*(u ~
>
r
)
= C.
o
(3)
If
u
~
<
r
o
, then U
I
(C,
u
~
) - U
I
(NC, u
~
) < 0.
Therefore, s*(u ~
<
r
)
= NC.
o
(4)
Therefore, individual liability borrower's utility under her optimal repayment strategy is as follows
(combining (1) - (4)):
u
~
(1 ÷ o ) if u
~
< r
U
I
[s *
(u
~
),u
~
] =
o
~
(5)
u ÷ r
if u
~
> r
o
Since 0 < o< 1, (5) is a convex function.
Proposition 2: An individual liability borrower's optimal investment decision (p*) is to take the
risky project (R) rather than the safe project (S).
31
Proof:
E[U
Ip
(s*,u
~
)] is the expected utility of individual liability borrower from take project pe (R, S)
according to playing her optimal repayment strategy (s*).
The expected utilities from investing in the risky (R) and safe (S) project (p):
If p = R:
E[U
IR
(s*,u
~
)] = 1 U
I
(s*,u
L
) + 1 U
I
(s*,u
H
)
2 2
=
1
u
L
(1 ÷ o ) + 1 (u
H
÷ r) (6)
2 2
Recognizing that
1
u
L
+
1
u
H
= u
S
and suppressing the arguments of U
I
, (6) reduces to:
2 2
R
EU
IR
= u
S ÷
ou + r L
2
If p = S:
E[U
IS
(s*,u
~
)] = U
I
(s*,u
S
)
EU
IS
= u
S
÷ r
(7)
(8)
(9)
Subtracting (9) from (7) yields the difference in expected utility between the two investment
strategies:
EU
IR
÷ EU
IS
= r ÷ou
2
Since, by assumption,
L
u
L
< r
o
:
(10)
r ÷o( r )
EU
IR
÷ EU
SI
>
2
o
=
0
(11)
Therefore, the expected utility to the individual liability borrower of investing in the risky project
exceeds that of investing in the safe project, i.e., p* = R.
32
Proposition 3: The expected repayment rate of an individual liability loan is 50 percent.
Proof:
V
I
(s*,u
~
) is the ex post value to the lender from making an individual liability loan conditional on
the project payoffs and the borrower's optimal repayment strategy.
From Proposition 1:
0 if u
~
< r
V
I
(s*,u
~
) =
o
r if u
~
> r
(i)
o
EV
PI
(s*,u
~
) is the expected value to the lender of making an individual liability loan conditional on
the borrower taking investment P.
EV
P
I
*
(s*,u
~
) = EV
RI
(s*,u
~
) by Proposition 2.
EV
P
I
*
(s*,u
~
) = 1 ? 0 + 1 ? r = r
(i i)
2 2 2
Therefore, the expected repayment rate is EV
P
I
*
(s*,u
~
) = 1
=
50%.
r 2
Lemma 1.1: The unique Nash Equilibrium strategy (s*) for both borrowers under a joint liability
contract is to not contribute (NC) to the loan repayment when both projects' payoffs are less than
2r
.
a
Proof:
33
U
J
1
(s,u
~
) is the value to borrower 1 under a joint liability contract from playing s = (s1, s2), where s1
e (C1, NC1) , s2 e (C2, NC2), and u
~
= (u
~
,u
~
) .
U
J
1
[(NC, NC),u
~
] = u
~
1(1 ÷ o )
U
J
1
[(C, NC),u
~
] = u
~
1 ÷ 2r
1 2
(12)
(13)
Subtracting (13) from (12):
U
J
1
[(NC, NC),u
~
] ÷ U
J
1
[(C, NC),u
~
] > 0 since u
~
1 < 2r
.
The same
o
applies to
U
J
2
(s,u
~
) . Therefore, s* = (NC, NC) when u
~
1 < 2r
and
u
~
2 < 2r
.
o o
Lemma 1.2: The unique Nash Equilibrium strategy (s*) under joint liability contract is for the
borrower with the project payoffs greater than or equal to
2r
to
pay for both loans and the borrower
a
with project payoff less than
Proof:
2r
to
not contribute.
a
Let
u
~
1 > 2r
o
and
u
2
< 2r
.
o
From (12) and (13):
U
J
1
[(C, NC),u
~
] ÷ U
J
1
[(NC, NC),u
~
] = ou
~
1 ÷ 2r (14)
(14) is greater than zero since
u
~
1 > 2r
.
o
Therefore, borrower would not deviate from playing C.
Applying (12) and (13) to borrower 2
U
J
2
[(C, NC),u
~
] ÷ U
J
1
[(C,C),u
~
] = r (15)
(15) is greater than zero. Therefore, borrower 2 will not deviate from playing NC.
34
Therefore, s*=(C, NC) when
u
~
1 > 2r
o
and
u
~
2 < 2r
o
, and by the same reasoning, s*=(NC, C) when
u
~
1 < 2r
o
and
u
~
2 > 2r
.
o
Lemma 1.3: The unique Nash Equilibrium strategy (s*) under a joint liability contract is for both
r
borrowers to contribute (C,C) to repayment of the loan if both projects' payoffs are greater than
a
and less than
Proof:
2r
.
a
Let
r s u
~
< 2r
and
r s u
~
< 2r
.
o
1
o o
2
o
U
J
1
[(C,C),u
~
] = u
~
1 ÷ r (16)
U
J
1
[(NC,C),u
~
] = u
~
1 (17)
Subtracting (17) from (16):
U
J
1
[(NC, NC),u
~
] ÷ U
J
1
[(C, NC),u
~
] = ÷r (18)
Since (18) is less than zero, borrower 1 would like to deviate from playing C to NC if borrower 2 does
not respond by deviating. Borrower 1 anticipates how borrower 2 will respond by evaluating the
following:
U
J
2
[(NC, NC),u
~
] ÷ U
J
2
[(NC,C),u
~
] = 2r ÷ ou
~
2 (19)
Because
r s u
~
< 2r
,
the right hand side of (19) is strictly greater than zero: r > 2r ÷ ou
~
> 0 .
o
2
o
2
Therefore, borrower 2 would respond to borrower 1 deviating by also deviating. Because both
borrowers know that deviating from C to NC induces the peer to doing likewise, neither borrower
35
will deviate from playing C if the marginal value to each borrower from the group playing (C,C) over
(NC,NC) is positive:
From (12) and (16):
U
J
1
[(C,C),u
~
] ÷ U
J
1
[(NC, NC),u
~
] = ou
~
÷ r (20)
(20) is weakly greater than zero since
r s u
~
< 2r
.
Therefore, neither borrower will deviate from
o
1
o
C, and s*=(C,C) when
r s u
~
< 2r
and
r s u
~
< 2r
.
o
1
o o
2
o
Lemma 1.4: The only Nash Equilibrium strategies (s*) under a joint liability contract is for only one
borrower to contribute to repayment of the loan for both [(C,NC) or (NC,C)] if both projects' payoffs
are greater than or equal to
Proof:
From (12) and (13):
2r
.
a
U
J
1
[(C, NC),u
~
] ÷ U
J
1
[(NC, NC),u
~
] = ou
~
2 ÷ 2r (21)
Since
u
~
1 > 2r
,
the right hand side of (21) is greater than or equal to zero.
o
Therefore, borrower 1 will
not deviate from playing C.
From (13) and (16):
U
J
2
[(C, NC),u
~
] ÷ U
J
2
[(C, C),u
~
] = r (22)
Since the right hand side of (22) is positive, borrower 2 will not deviate from playing NC.
However, the same logic applies to show that it is also a Nash Equilibrium for borrower 1 to not
contribute and borrower 2 to contribute. Therefore, s* = [(C, NC), (NC,C)].
36
Proposition 4: Under a joint liability contract, if one borrower invests in the safe project, then the
other borrower will also invest in the safe project (S) rather than the risky project (R).
Proof:
E[U
J
11P
2
(s*,u
~
)] is the expected value to borrower 1 from borrower 1 taking project P1 and borrower 2 P
taking project P2 according to both players' optimal contribution strategies (s*) and payoff
~
possibilities (u ).
If P1 = R and P2 = S, then by Lemma 1.1 and Lemma 1.2:
s * (u
1L
,u
2S
) = (NC, NC) (23)
s * (u
1H
,u
2S
) = (C, NC) (24)
Therefore,
EU
J
1
= 1 U
J
1
[s*,(u
1L
,u
2S
)] + 1 U
J
1
[s*,(u
1H
,u
2S
)]
RS
2 2
=
1
u
1L
(1 ÷ o ) + 1 (u
1H
÷ 2r)
2 2
=u ÷ S
1
ou
1
L ÷ r
2
(25)
If P1 = R and P2 = S, then by Lemma 1.3:
s * (u
1S
,u
2S
) = (C,C) (26)
Therefore,
EU
S
JS
1
= U
J
1
[s*,(u
1S
,u
2S
)] = u
1S
÷ r (27)
To compare the marginal value to borrower 1 from taking the safe project over the risky project when
borrower 2 takes the safe project, subtract (25) from (27):
37
EU
S
JS
1
÷ EU
J
1
= u
1S
÷ r ÷ (u
1S
÷
ou
1
÷ r)
=
ou
1
RS
2
L
2
L
(28)
Because the right hand side of (28) is greater than or equal to zero, borrower 1 will invest in the safe
project if borrower 2 invests in the safe project. I.e., P
1
*
= S if P
2
= S .
Proposition 5: Under a joint liability contract, if one borrower invests in the risky project, then the
other borrower will invest in the risky project only if her projects' high state payoff is greater than
3r
.
Otherwise, she will invest in the safe project.
o
Proof:
If P1=P2=R, by Lemmas 1.1, 1.2, and 1.4:
s * (u
1L
,u
2L
) = (NC, NC) (29)
s * (u
1L
,u
2H
) = (NC,C) (30)
s * (u
1H
,u
2L
) = (C, NC) (31)
s * (u
1H
,u
2H
) = [(C, NC), (NC,C)] (32)
Therefore, the expected value to borrower 1 from taking the risky project when borrower 2 does
likewise is
EU
J
1
= 1 U
J
1
[s*,(u
1L
,u
2L
)] + 1 U
J
1
[s*,(u
1L
,u
2H
)] + 1 U
J
1
[s*,(u
1H
,u
2L
)]
RR
4 4 4
+ 1 {1 U
J
1
[(C, NC), (u
H
,u
H
)] + 1 U
J
1
[(NC, C), (u
H
,u
H
)]}
42
1 2
2
1 2
=
1
u
1L
(1 ÷ o )
+
1
u
1L +
1 (u
1H
÷ 2r)
+
1 [1 (u
1H
÷ 2r)
+
1
u
1H
]
4 4 4 42 2
38
= u
1S
÷
ou
1
+ 3r L
4
If P1=S and P2=R, by Lemmas 1.1 and 1.2:
s * (u
1S
,u
2L
) = (NC, NC)
s * (u
1S
,u
2H
) = (NC,C)
(32)
(33)
(34)
Therefore, the expected value to borrower 1 from taking the safe project when borrower 2 takes the
risky project is
EU
S
JR
1
= 1 U
J
1
[s*,(u
1S
,u
2L
)] + 1 U
J
1
[s*,(u
1S
,u
2H
)]
2 2
=
1
u
1S
(1 ÷ o ) +
1
u
1
S
2
= u
1S
÷
ou
1
S
2
2
(35)
To compare the marginal value to borrower 1 from taking the risky project over the safe project when
borrower 2 takes the risky project, subtract (35) from (32):
EU
R
J
1
÷ EU
S
JR
1
= u
1S
÷
ou
1
+ 3r ÷ (u
1S
÷
ou
1
) = a(2u
1
÷ u
1
) ÷ 3r
R
u
1S
=
u
1
+ u
1
L
4 2
S S
4
L
(36)
Using
L
2
H
in (36):
EU
J
1
÷ EU
S
JR
1
= ou
1
÷ 3r H
RR
4
(37)
39
The right hand side of equation (37) is greater than zero if u
1H
> 3r
and
less than zero if u
1H
< 3r
.
Therefore, P1* = R if P2 = R and
u
1H
> 3r
4
4
, and P1* = S if P2 = R and
u
1H
< 3r
4
.
4
Lemma 2: The expected value to one borrower of a joint liability contract when both borrowers
invest in the risky project exceeds the expected value when they both invest in the safe project.
Proof:
Subtract (27) from (32):
EU
J
1
÷ EU
S
JS
1
= r ÷ ou
1
L
RR
4
(38)
Because u
1L
< r , the right hand side of (38) is greater than zero. Therefore, both borrowers prefer
a
to choose the same investment strategies of P1=P2=S or P1=P2=R.
Corollary 1: Borrowers with high possible project payoffs (uH >
3
r
)
select each other to take joint
a
liability loans and invest in risky projects. Other borrowers (uH <
3
r
)
select each other and invest
a
in safe projects.
Proof:
Following Proposition 4 there is only a possibility of P1=P2=S or P1=P2=R because it is not possible for one
borrower to invest in the safe project and the other to invest in the risky project. Lemma 2 shows that both
borrowers would prefer that they both invest in the risky project. From Proposition 5, it is only the borrowers
who have high state payoffs under the risky investment who can commit to investing in the risky project
when her peer does the same. Therefore, borrowers who can commit to not deviating from investing in the
risky project when they both agree to do so would prefer to match with each other. The borrowers who
cannot make such a commitment will be forced to match with other such borrowers. Therefore, the
borrowers with the very high state payoff possibilities will
40
match with each other and invest in risky projects; the borrowers with the moderately high state
payoff possibilities will match with each other and invest in safe projects.
Proposition 6: The expected repayment rate of a joint liability loan is between 75 and 100 percent.
Proof:
Let ¢ e [0,1] be the fraction of borrowers with u
H
> 3r
.
a
EV
P
J
1
P
3
(s*,u
~
) is the expected value per borrower to the lender from making a joint liability loan to a
group conditional on borrowers' investment choices (P1 and P2).
By equation (26):
EV
S
J
S
(s*,u
~
) = 2r / 2 = r (39)
By equations (29) - (32):
EV
R
J
R
(s*,u
~
)
=
1
?
0
+
3
?
2r
=
3r / 2
=
3r
(40)
4 4 2 4
EV
J
is the expected value to the lender per borrower from making a joint liability loan to a group
without knowing what projects are available to them.
EV
J
= ¢EV
S
J
S
+ (1 ÷¢ )EV
R
J
R
= ¢r
+
3 (1 ÷¢ )r
4
= (3 +¢ )r
4
(41)
Therefore, the expected repayment rate is = (3 +¢ )r / r = 3 +¢
,
which is bounded between ¾ and
1 (75% and 100%) because
0 s ¢ s 1.
4 4
41
Lemma 3.1: The unique Nash Equilibrium strategy (s*) for both borrowers under a joint liability
contract with a possibility of social sanctions is to not contribute to the loan repayment when both
projects' payoffs are less than
Proof:
|r
.
o
JS ÷ joint liability contract with a possibility of social sanctions
U
JS
1
[(NC, NC)],u
~
] = u
~
1(1 ÷ o )
U
JS
1
[(C, NC)],u
~
] = u
~
1 ÷ 2r
To compare the value of not contributing over contributing, subtract (43) from (42):
U
JS
1
[(NC, NC)],u
~
] ÷ U
JS
1
[(C, NC)],u
~
] = 2r ÷ ou
~
1
(42)
(43)
(44)
Applying this case, i.e.,
u
~
1
<
|r
,
to equation (44):
o
U
JS
1
[(NC, NC)],u
~
] ÷U
JS
1
[(C, NC)],u
~
] > 2r ÷ |r > 0
The same logic applies to the borrower 2. Therefore, s*=(NC,NC) when
u
1 <
|r
o
and
u
2 <
|r
.
o
Lemma 3.2: The unique Nash Equilibrium strategy (s*) under a joint liability contract and
possibility of social sanctions is for the borrower with project payoffs greater than 2r
to
contribute
o
to the repayment of the loan for both borrowers and for the borrower with project payoffs less than
|r
to not contribute.
o
Proof:
Assume that
u
~
`
1
> 2r
o
and
u
~
`
2
<
|r
o
.
42
To compare the value of contributing over not contributing for borrower 1, subtract (42) from (43):
U
JS
1
[(C, NC)],u
~
] ÷U
JS
1
[(NC, NC)],u
~
] = ou
~
`
1
÷ 2r (45)
Applying this case, i.e.,
u
~
1 > 2r
,
o
to equation (45):
U
JS
1
[(C, NC)],u
~
] ÷ U
JS
1
[(NC, NC)],u
~
> 0 .
Therefore, borrower 1 will not deviate from contributing for both.
U
JS
2
[(C, NC)],u
~
] = u
~
2 (1
÷
o
)
|
U
JS
2
[(C, C)],u
~
] = u
~
2 ÷ r
(46)
(47)
To compare the value of not contributing over contributing for borrower 2, subtract (47) from (46:
U
JS 2
[(C, NC)],u ~] ÷ U
JS
2
[(C,C)],u
~
] = r
÷
ou
~
2
|
(48)
Apply this case, i.e.
u
~
2
<
|r
,
o
to (48):
U
JS
2
[(C, NC)],u
~
] ÷ U
JS
2
[(C,C)],u
~
] > 0 .
Therefore,
borrower 2 will not deviate from not contributing.
So, when
u
~
`
1
>
2
r
o
and
u
~
`
2
<
|r
,
s* = (C, NC) , and vice versa.
o
Lemma 3.3: The unique Nash Equilibrium strategy (s*) under a joint liability contract with
possibility of social sanctions is for both borrowers to contribute to the repayment of the loan if one's
project pays off
2r
or
more and the other pays off |r
or
more.
Proof:
o o
Let
u
~
1 >
2
r
o
and
u
~
2
>
|r
o
.
43
Apply (43) and (47) to borrower 1:
JS1
U
[(C,C),u ~] ÷ U
JS
1
[(NC,C),u
~
] = u
~
÷ r ÷ (u
~
÷ o
) =
ou
~
1 ÷ r
1 1
| |
(49)
Applying this case, i.e.
u
~
1 > 2r
o
to (49):
U
JS
1
[(C,C),u
~
] ÷ U
JS
1
[(NC,C),u
~
] > 0 . Therefore, borrower
1 will not deviate from contributing to repayment.
Apply (49) to borrower 2:
JS 2
U
[(C,C),u ~] ÷ U
JS
2
[(C, NC),u
~
] = u
~
÷ r ÷ (u
~
÷ o
) =
ou
~
2 ÷ r
2 2
| |
(50)
Applying this case, i.e.,
u
~
2
>
|r
o
t o (50):
U
JS
2
[(C,C),u
~
] ÷ U
JS
2
[(C, NC),u
~
] > 0 .
Therefore,
borrower 2 will not deviate from contributing to repayment.
So, when one project pays
2r
or
more and the other project pays |r
or
more, s*=(C,C).
o o
Lemma 3.4: The unique Nash Equilibrium strategy (s*) under a joint liability contract with the
possibility of social sanctions is for neither borrower to contribute to the repayment of the loan if one
|r
project payoffs less than
Proof:
o
and the other pays off less than
2r
.
o
Let
|r < u
~
s 2r
o
1
o
and
u
~
2
<
|r
.
o
Recall (44):
U
JS
1
[(NC, NC)],u
~
] ÷ U
JS
1
[(C, NC)],u
~
] = 2r ÷ ou
~
1
44
Applying
this
case,
i.e.,
|r < u
~
s 2r 1
to
the
above:
o
(2 ÷ | )r > U
JS
1
[(NC, NC)],u
~
] ÷ U
JS
1
[(C, NC)],u
~
] > 0 .
o
Apply
the
above
to
borrower
2
and
this
case,
i.e.
u
~
2
<
|r
:
o
U
JS
2
[(NC, NC)],u
~
] ÷ U
JS
2
[(C, NC)],u
~
] > 0 .
Therefore, neither borrower deviates from this strategy. So, s* = (NC,NC).
Proposition 7: If
in the risky project.
Proof:
u
1L <
|r
o
and borrower 2 invests in the safe project, then borrower 1 will invest
If P1=R and P2=S, then by Lemmas 3.4 and 3.3:
s * (u
1L
,u
2S
) = (NC, NC) (51)
s * (u
1H
,u
2S
) = (C,C) (52)
Therefore, the expected value to borrower 1 from taking the risky project when her peer takes the
safe project is:
EU
JS 1
= 1 [u
1L
(1 ÷ o )] + 1 (u
1H
÷ r )
(53)
RS
2 2
S u
1L
+ u
1
H
Use
u= 1 2
in (53):
EU
JS
S
1
= u
1S
÷
ou
1
+ r L
R
2
(54)
45
If P1= P2=S, then by Lemma 3.3:
s * (u
1S
,u
2S
) = (C,C) (55)
Therefore, the expected value to borrower 1 from taking the safe project when her peer does likewise
is:
EU
S
JS
1
= u
1S
÷ r S
(56)
To compare the marginal value to borrower 1 from taking the risky project over the safe project when
borrower 2 takes the safe project, subtract (56) from (54):
EU
JS
S
1
÷ EU
S
JS
1
= u
1S
÷
ou
1
+ r ÷ (u
1S
÷ r) = r ÷ u
1
R S L
2
2
L
(57)
Apply this case, i.e.,
u
1L <
|r
,
to (57): EU
JS
S
1
÷ EU
S
JS
1
> r(1 ÷ | ) > 0 .
o
R S
2
Therefore, borrower 1 will
invest in the risky project when her peer invests in the safe project and her risky project's low state
payoff is less than
Proposition 8: If
|r
.
o
u
1L <
|r
and
o
borrower 2 invests in the risky project, then borrower 1 will take
the risky project only if
Proof:
u
1S
> (2 + | )r
2o
. Otherwise, borrower 1 will take the safe project.
If P1=P2=R, then by Lemmas 3.1, 3.2, and 3.3:
s * (u
1L
,u
2L
) = (NC, NC) (58)
s * (u
1L
,u
2H
) = (NC,C) (59)
46
s * (u
1H
,u
2L
) = (C, NC) (60)
s * (u
1H
,u
2H
) = (C,C) (61)
Therefore, the expected value to borrower 1 from taking the risky project when her peer does the
same is:
EU
JS
R
1
= 1 U
JS
1
[s*,(u
1L
,u
2L
)] + 1 U
JS
1
[s*,(u
1L
,u
2H
)] + 1 U
JS
1
[s*,(u
1H
,u
2L
)]
R
4 4 4
+ 1 U
JS
1
[s*,(u
H
,u
H
)]
4
1 2
=
1
u
1L
(1 ÷ o ) +
1
u
1L
(1
÷
o
)
+ 1 (u
1H
÷ 2r) + 1 (u
1H
÷ r) (62)
4 4 | 4 4
S u
1L
+ u
1
H
Use u = 1 2
in (62) and reduce:
EU
R
JSR1
= u
1S
÷
o (1 ÷ 1 )u
1L
+ 3r
|
4
(63)
If P1=S and P2=R, then by Lemmas 3.4 and 3.3:
s * (u
1S
,u
2L
) = (NC, NC) (64)
s * (u
1S
,u
2H
) = (C,C) (65)
Therefore, the expected value to borrower 1 from taking the safe project when her peer takes the
risky project is:
EU
S
JS
1
= 1 U
JS
1
[s*,(u
1S
,u
2L
)] + 1 U
JS
1
[s*,(u
1S
,u
2H
)]
R
2 2
47
=
1
u
1S
(1 ÷ o ) + 1 (u
1S
÷ r)
2
= u
1S
÷
ou
1
+ r S
2
2
(66)
To compare the marginal value of borrower 1 taking the risky project over the safe project when her
peer takes the risky project, subtract (66) from (62):
EU
JS
R
1
÷ EU
S
JS
1
= 2ou
1
÷ (| + 2)r S
R R
4
(67)
A necessary and sufficient condition for the right hand side of (67) to be greater than or equal to zero
is determined by setting (67) greater than or equal to zero, which yields:
u
1S
> (| + 2)r
2o
Therefore, if borrower 1's projects' mean expected payoff exceeds
(68)
(| + 2)r
,
then she invests in the
2o
risky project when her peer does also. If her mean expected payoff is between r
and
(| + 2)r
,
o 2o
then she invests in the safe project.
Lemma 4.1: For the borrower who will always choose the risky project, it is preferable for her to
find a peer who will choose the safe project.
Proof:
If
u
1S
> (| + 2)r
2o
, then borrower 1 will invest in the risky project when her peer invests in the risky
project (Proposition 8) or if her peer invests in the safe project (Proposition 7). The relative value of
finding a peer who will invest in the safe risky project when she invests in the risky project over one
who will invest in the safe project is determined by subtracting (54) from (62):
48
o (1 ÷ 1 )u
1L
+ 3r o
| ÷1 u
1L
÷ r
S
|
S
ou
1
+ r =
|
S
EU
JS
R
1
÷ EU
JS
S1
R R
= u
1
÷
4
÷ u
1
÷
2
4
<0
(69)
Since (69) is negative, a borrower with an expected payoff on her projects greater than
values having a peer who will invest in the safe project.
( | + 2) r
2o
Lemma 4.2: For the borrower who will invest in the safe project when her peer invests in the risky
project, it is preferable for her to find a peer with low expected payoffs, i.e. between
( | + 2) r
.
2o
Proof:
r
and
o
Proposition 7 states that if one borrower invests in the safe project, then the other will invest in the risky
project. Proposition 8 states that if the expected project payoffs are low, then a borrower will maintain
investment in the safe project when her peer invests in the risky project. If both borrowers
have low expected project payoffs, then their optimal investment strategy, P* = (R, S) or
P* = (S, R) . There is no a priori reason for either borrower to expect that she will be the one to play
the risky investment strategy. Therefore, both will expect to play either strategy equally.
The expected value for a borrower with low expected payoffs (borrower 1) from having a peer with
similarly low expected payoffs is derived from (54) and (66):
1 EU
JS
1
+ 1 EU
JS
1
= u
S
÷ 2r + o (u
1L
+ u
1S
)
2
RS
2
SR
1
4
(70)
If the peer has high expected payoffs, however, the borrower with low expected payoffs is assured to
always invest in the safe project because she cannot credibly commit to the risky strategy and the high
expected payoff borrower will always invest in the risky project (Proposition 8).
The expected value for a borrower with low expected payoffs (borrower 1) from having a peer with
high expected payoffs is given in (64).
49
The marginal value to the low expected payoff borrower from having a peer with similarly a low
expected payoff project is the difference between (64) and (71):
(1 EU
JS
S
1
+ 1 EU
S
JS
1
) ÷ EU
S
JS
1
= o (u
1
÷ u
1
) > 0
2
R
2
R R S
4
L
(71)
Proposition 9: Borrowers will match with other borrowers with the same expected project payoffs
where there is a possibility of social sanctions.
Proof:
Borrowers with low expected project payoffs prefer to invest in the risky project (Lemma 4.2).
These borrowers cannot credibly commit to invest in the risky project with a peer with high project
payoffs because the peer would always invest in the risky project (Proposition 8).
Therefore, the only chance the low expected project payoffs borrower has to play the risky strategy is
to match with a borrower with low expected project payoffs, too (Proposition 7).
Proposition 10: The only investment strategies that will be played are P1=P2=R for groups with
high expected project payoffs and P1=P2 for groups with low expected project payoffs.
Proof:
This follows from Propositions 7, 8, and 9.
Proposition 11: The expected repayment rate of a joint liability loan with the possibility of social
sanctions is between 50 percent and 75 percent.
Proof:
Let
| e[0,1]
be the fraction of borrowers with
r < u
S
< (| + 2)r
.
o 2o
50
EV
P
J1SP
2
(s*,u
~
) is the expected value per borrower to the lender from making a joint liability loan to a
group conditional on borrowers' investment choices (P1 and P2).
EV
R
JR
S
(s*,u
~
) = (1 ? 0 + 3 ? 2r) / 2 = 3r
(72)
4 4 4
EV
R
JS
S
(s*,u
~
) = EV
S
JR
S
(s*,u
~
) = (1 ? 0 + 1 ? 2r) / 2 = r (73)
2 2 2
The expected value to the lender is the weighted average of (72) and (73) by the distribution of all
borrowers' expected project payoffs:
EV
JS
= (1 ÷ |)EV
R
JR
S
+ |EV
R
JS
S
= (3 ÷ |)r
4
(74)
The expected repayment rate is (74) divided by r and is bounded according to the value of |:
1 s 3 ÷|
s
3
(75)
2 4 4
Proposition 12: In terms of borrower expected utility, the individual liability contract weakly
dominates the joint liability contracts assuming the same principal and interest across contracts (i.e., same r).
Proof:
Case 1: Borrowers with low expected payoffs
From Proposition 2, Corollary 1, and Proposition 10 for borrower 1 with
u
1H
< 3r
:
EU
Ip
* = EU
IR
,
o
EU
Jp
1
= EU
S
JS
1
, and EU
J
pS*
1
e{EU
JS
S
1
, EU
S
JS
1
}.
*
R R
51
The value of the individual liability contract over the joint liability contract without the possibility of
social sanctions is given by subtracting (27) from (7):
EU
Ip
* ÷ EU
Jp
1
= r ÷ou
1
> 0 L
*
2
(76)
The value of the individual liability contract over the joint liability contract with the possibility of
social sanctions is given by subtracting (64) from (7) or (71) from (7):
EU
Ip
* ÷ EU
R
JSS
1
= 0 (77)
EU
Ip
* ÷ EU
S
JS
1
= o (u
1S
÷ u
1L
) > 0 R
Case 2: Borrowers with high expected payoffs
From Proposition 2, Corollary 1, and Proposition 10 for borrower 1 with
(78)
u
1H
> 3r
:
EU
Ip
* = EU
IR
,
o
EU
Jp
1
= EU
R
J
1
, and EU
J
pS*
1
= EU
R
JSR
1
.
*
R
The value of the individual liability contract over the joint liability contract without the possibility of
social sanctions is given by subtracting (32) from (7):
EU
Ip
* ÷ EU
Jp
1* = r ÷ ou
1
> 0 L
4
(79)
The value of the individual liability contract over the joint liability contract with the possibility of
social sanctions is given by subtracting (63) from (7):
r
÷
ou
1
L
EU
Ip
* ÷ EU
J
pS*
1
=
4
|
>
0
(80)
Proposition 14: If given the choice between an individual liability and joint liability contract
without the possibility of social sanctions, borrowers with
u
iH
< 3r
o
choose the individual liability
52
contract and those with
u
iH
> 3r
choose
the joint liability contract.
o
The total principal and interest
due on the individual liability contract is 150% of that of the joint liability contract.
Proof:
Let rI be the amount due for the individual contract and rJ be the amount due for the joint liability contract
without the possibility of social sanctions.
From Propositions 3 and 6:
EV
p
I
*
= r
I
2
EV
p
J
*
= (3 +¢ )r
J
4
(81)
(82)
If the lender were to choose rI relative to rJ such that the expected values to the lender is equal, then
set (81) equal to (82), which yields:
r
I
= (3 +¢ )r
J
2
(83)
The borrower will choose the individual liability loan over the joint liability loan if
(EU
Ip
* | r
I
) ÷ (EU
Jp
1* | r
J
) > 0 . Otherwise, she will choose the joint liability loan.
Case 1: p*=(R,R)
From Propositions 2 and 5:
(EU
Ip
* | r
I
) ÷ (EU
Jp
1* | r
J
) =
u
1S
÷
ou
1
+ r
I
L
÷
u
S
÷
ou
1L
+ 3r
J
Substitute rI with expression (83) in (84):
2
1
4
(84)
(EU
Ip
* | r
I
) ÷ (EU
Jp
1* | r
J
) =
÷
ou
1
+ (3 + 2¢ )r
J
< 0 L
(85)
4
53
Therefore, borrowers for whom p* = (R, R) (those with u
iH
> 3r
)
will always select the joint liability
o
contract.
Case 2: p*=(S,S)
From Propositions 2 and 4:
(EU
Ip
* | r
I
) ÷ (EU
Jp
1* | r
J
) =
u
1S ÷
ou
1
+ r
I
L
÷ u
S
÷r
2
(
1
J
)
(86)
Substitute rI with expression (83) in (86):
(EU
Ip
* | r
I
) ÷ (EU
Jp
1* | r
J
) = (7 +¢ )r
J
÷ 2ou
1
L
4
(87)
Use the assumption that
u
1L
< r
J
o
to show that the right hand side of (87) is greater than zero:
(7 +¢ )r
J
÷ 2ou
1
L > (7 +¢ )r
J
÷ 2r
J
= (5 +¢ )r
J
> 0 (88)
4 4 4
Therefore, borrowers for whom p*=(S, S) (those with
liability contract.
u
iH
< 3r
),
o
will always choose the individual
Since the borrowers will separate in this way, the lender believes that ¢ = 0. Therefore,
substituting in (83):
r
I
* = 3r
J
*
,
where rI* and rJ* are the equilibrium amounts due for both the
2
individual and joint liability contracts.
54
The lender values the joint liability contract at the general equilibrium amount due, rJ*, at (82)
evaluated with
¢ = 0 : EV
p
J
*
= 3r
J
*
.
4
The repayment rate in general equilibrium, therefore, is
EV
p
J
*
= 3r
J
* r = 3 = 75% .
4 4
Comment 1: Borrowers with better prospects (u
i
H
> 3r
)
prefer the joint liability contract because
o
the interest rate is lower because the lender knows that their peers share risk with them. Borrowers
with lesser prospects (u
i
H
< 3r
)
prefer the individual liability contract because the cost imposed by
o
the lender is less than the cost of taking the safe project.
Comment 2: Everyone invests in the risky project.
Lemma 5: If given the choice between an individual and joint liability contract where there will be
social sanctions, a borrowers for whom p*=(R,R) will choose the individual liability contract.
Proof:
Let rJS be the amount due under a joint liability contract with the possibility of social sanctions.
From Propositions 3 and 11:
EV
p
I
*
= r
I
2
EV
p
J*
S
= (3 ÷ |)r
JS
4
(89)
(90)
If the lender were to choose rI relative to rJS such that the expected values to the lender is equal,
then set (89) equal to (90), which yields:
r
I
= (3 ÷ |)r
JS
2
(91)
55
The borrower will choose the individual liability loan over the joint liability loan if
(EU
Ip
*) ÷ (EU
J
pS*
1
) > 0 . Otherwise, she will choose the joint liability loan.
If p*=(R,R), then:
1L
(EU
Ip
*) ÷ (EU
J
pS*
1
) =
u
1S ÷
ou
1
+ r
I
L
1
÷ u
S
÷ o (1
+
|
)
u
1
+ 3r
JS
(92)
2
2
Substitute rI with expression (91) in (92):
(EU
Ip
* | r
I
) ÷ (EU
R
JSS
1
| r
p
*
)
=
ou
1
(1 ÷ | ) + 4||r
JS
> 0 L
4|
Therefore, if p*=(R,R), then the borrowers prefer the individual liability contract.
(93)
Proposition 15: If given the choice between an individual liability and joint liability loan where
there is a possibility of social sanctions, no one will take the joint liability contract.
Proof:
If anyone does take the joint liability contract, it would be the borrowers with lower expected project payoffs
because Lemma 5 shows that the high expected project payoff borrowers will definitely choose the
individual liability contract. These lower expected project payoff borrowers are those who would play (R,S) or
(S,R) investment strategies. Therefore, if the joint liability contract is taken by
at least one group, | = 1.
Compare the expected values from taking the individual liability loan to the joint liability loan with
the possibility of social sanctions by subtracting (70) from (10) and substituting rI with expression
(91):
EU
Ip
* ÷
1
EU
JS
S
1
+ 1
EU
JS
1
= o (u
1
÷ u
1
) ÷ (1 ÷ |)r
JS
2
R
2
S
R
S
L
4
(
9
4
)
56
As explained above, if this contract is accepted by anyone, it is by the lower expected payoff
borrowers. Therefore, evaluate (94) with | = 1:
EU
Ip
* ÷ 1 EU
R
JSS
1
+ 1 EU
S
JS
1
=
o (u
1
÷ u
1
) > 0
2
2
R
S
4
L
(95)
Therefore, the borrowers with lower expected project payoffs would choose the individual liability
loan.
57
References
Ahlin, Christian and Townsend, Robert M., "Using Repayment Data to Test Across Models of Joint
Liability Lending" . Economic Journal, Vol. 117, No. 517, pp. F11-F51, February 2007 Available at
SSRN: http://ssrn.com/abstract=978564 or DOI: 10.1111/j.1468-0297.2007.02014.x
Armendariz de Aghion, Beatriz and Jonathan Morduch. 2000. "Microfinance Beyond Group Lending,"
Economics of Transition, Vol. 8 (2), 401-420.
Ayayi, Ayi. 2009. "Microfinance: Debt or Equity? What are the Implications for Social Welfare?" Global
Economy and Finance Journal, Vol. 2 (2), 64-80.
Besley, Timothy and Stephen Coate. 1995. "Group Lending, Repayment Incentives, and Social
Collateral," Journal of Developmental Economics, 46, 1-18.
Bubna, Amit and Bhagwan Chowdhry. 2009. "Franchising Microfinance," Review of Finance, Vol. 14
(3), 451-476.
Casssar, Alessandra, Luke Crowley, and Bruce Wydick. 2007. "The Effect of Social Capital on Group
Loan Repayment: Evidence from Field Experiments," The Economic Journal, Vol. 117 (February), F85- F106.
Che, Yeon-Koo. 2002. "Joint Liability and Peer Monitoring Under Group Lending," Contributions to
Theoretical Economics, 2, Article 3.
Conning, Jonathan. 2005. "Monitoring by Delegates or by Peers? Joint Liability Loans under Moral
Hazard," (June 2005). Unpublished working paper.
Cull, Robert, Asli Demirguc-Kunt, and Jonathan Morduch. 2006. "Financial Performance and Outreach:
A Global Analysis of Lending Microbanks," World Bank Policy Research Working Paper 3827, February 2006.
Cull, Robert, Asli Demirguc-Kunt, and Jonathan Morduch. 2009. "Microfinance Meets the Market,"
Journal of Economic Perspectives, Vol 23 (1), 167-192.
De Mel, Suresh, David McKenzie, and Christopher Woodruff. 2008. "Returns to Capital in
Microenterprises: Evidence from a Field Experiment," The Quarterly Journal of Economics, Vol. 123 (4), 1329-1372.
Ghatak, Maitreesh. 1999. "Screening by the Company You Keep: Joint Liability Lendings and the Peer
Selection Effect," Economic Journal.
58
Ghatak, Maitreesh and Timothy W. Guinnane. 1999. "The Economics of Lending with Joint Liability: Theory and
Practice," Journal of Development Economics, Vol. 60 (1999), 195-228.
Gine, Xavier and Dean Karlan. 2007. "Group versus Individual Liability: A Field Experiment in the
Phillipines" Center f or Global Development Working Paper Number 111,
http://ssrn.com/abstract981390.
Gine, Xavier and Dean Karlan. 2010. "Group Versus Individual Liability: Long Term Evidence from
Phillipine Microcredit Lending Groups," Yale Economics Working Paper No. 61; Yale University Economic Growth
Discussion Paper No. 970, http://ssrn.com/abstract=1407614.
Gine, Xavier, Pamela Jakiela, Dean Karlan, and Jonathan Morduch. 2010. "Microfinance Games,"
American Economic Journal, Forthcoming.
Gomez, Rafael and Eric Santor. 2003. "Do Peer Group Members Outperform Individual Borrowers? A
Test of Peer Group Lending Using Canadian Micro-Credit Data," Bank of Canada Working Paper 2003- 33.
Guttman, Joel M. 2006. "Repayment Performance in Group Lending Programs: A Survey," Networks
Financial Institute at Indiana State University Working Paper, 2006-WP-01, April 2006.
Guttman, Joel M. 2007. "Assortive Matching, Adverse Selection, and Group Lending," Journal of
Developmental Economics, Vol. 87 (1), 51-56.
Hermes, Niels and Robert Lensink. 2007. "The Empirics of Microfinance: What Do We Know?" The
Economic Journal, Vol. 117 (February), F1-F10.
Hulme, David and Paul Mosley. 1996. Finance against Poverty, Volume 1, Routlege Publications,
London.
Karlan, Dean S. and Jonathan Zinman. 2007. "Credit Elasticities in Less-Developed Economies:
Implications for Microfinance" (July 2007). (r&r American Economic Review)
Karlan, Dean S. and Jonathan Zinman. 2009. "Observing Unobservables: Identifying Information
Asymmetries with a Consumer Credit Field Experiment" Econometrica, Vol. 77 (6), 1993-2008.
Karlan, Dean and Jonathan Zinman. 2009. "Expanding Credit Access: Using Randomized Supply
Decisions to Estimate the Impacts," The Review of Financial Studies, Vol. 23 (6), 2560-2590.
Khawari, Aliya. 2004. "Microfinance: Does It Hold Its Promises? A Survey of Recent Literature,"
Hamburg Institute of International Economics Discussion Paper 276.
59
LaPorta, R., Lopez-de-Silanex, F., Shleifer, A., and Vishey, R. (1998), Law and Finance, 106 Journal of Political Economy,
1113 - 1115.
Leftley, Richard and Shadreck Mapfumo. "Effective Micro-Insurance Programs to Reduce
Vulnerability," Opportunity International Network
McGuinness, Elizabeth and Volodymyr Tounytsky. 2006. "The Demand for Microinsurance in
Pakistan." March 2006. Microfinance Opportunities
McLean, Kate. 2010. "Capitalizing on Women's Social Capital? Women-Targeted Microfinance in
Bolivia," Development and Change, Vol. 41 (3), 495-515.
Mersland, Roy. 2005. "The Agenda and Relevance of Recent Research in Microfinance." January 2005.
Unpublished working paper, Adger University College, Norway, http://ssrn.com/abstract=977138
Mersland, Roy and R. Øystein Strøm. 2009. "Performance and Governance in Microfinance
Institutions," Journal of Banking and Finance, Vol. 33 (4), 662-669.
"Morduch, Jonathan. 1998. "Does Microfinance Really Help the Poor? New Evidence from Flagship
Programs in Bangladesh," Research Program in Development Studies, Princeton University, Research Program in
Development Studies Working Paper 211.
Paxton, Julia, Douglas Graham, and Cameron Thraen. 2000. "Modeling Group Loan Repayment
Behavior: New Insights from Burkina Faso," Economic Development and Cultural Change, 48, 639-655.
Pellegrina, Lucia Dalla. 2006. "Microfinance and Investment: A Comparison Between Group Lending,
Bank Lending and Informal Credit" (December 15, 2006). Available at SSRN:
http://ssrn.com/abstract=953231
Pretes, Michael. 2002. "Microequity and Microfinance," World Development, 30, 1341-1353.
Stiglitz, Joseph E. 1990. "Peer Monitoring and Credit Markets," 4, World Bank Economic Review, 341-
366.
Townsend, R., 1979. "Optimal Contracts and Competitive Markets with Costly State Verification,"
Journal of Economic Theory, 21, 265-293.
Van Bastaleaer, Thierry. 2000. "Imperfect Information, Social Capital, and the Poor's Access to Credit,"
Center for Institutional Reform and the Informal Sector, University of Maryland Working Paper No. 234.
60
Essay 2: Issuances of Floating Rate Convertible Securities and Financial Manager
Characteristics
Abstract
Floating rate convertibles (FRCs) are a category of PIPE securities that receive negative associations in both the
academic and professional literature, earning monikers such as "death spirals" because of significant negative returns
to equity of firms subsequent to issuing them. This study sheds light on the managerial relationship to the decision to
issue FRCs and to the variation in market response to these issues. One main result of the study identifies influence of
the CFO relative to the CEO as significant in the decision to issue FRCs and in the market's immediate reaction to the
issuance. Another main result is that FRC issuing firms with CFOs without prior public equity issuance experience
have significantly negative long run abnormal returns, whereas FRC issuing firms with experienced CFOs do not.
Overall, I find support for the faulty contract design hypothesis for the firms with less experienced CFOs and a new
hypothesis consistent with optimal security design (OSD) for the firms with more experienced CFOs.
61
I. Introduction
Since 1995, a growing number of public firms have raised funds through issues of private investment
in public equity (PIPEs). PIPEs are common stock, preferred stock, convertible debt, convertible preferred stock,
stock warrants, and equity lines, which are sold to investors privately by public firms. While private issuances of stock
took place prior to 1995, that was the year in which the term, PIPE, came into use. PIPEs are typically more than
solely straight equity deals, but those that include asymmetric payoff features such as that of a warrant.
PIPEs typically have been seen as solutions to an information asymmetry problem that firms,
particularly small growth ones, face when needing to raise external funds publicly. Through the private negotiations
preceding a private issue, information can be conveyed to the investor at less cost than in a public issue. Nevertheless,
some private placement advisors are critical of a significant number of PIPE deals, contending that these deals' terms
are harmful to the issuing firms.
Given the theoretical benefit of PIPE issues in solving the information asymmetry problem and the
fact that firms choose to enter into deals that have been deemed harmful, why question the net benefit of any PIPE to
the firm? Investors in PIPEs that contain the arguably adverse terms for the firm are primarily hedge funds, which
make their profit on arbitrage opportunities and not on the long run performance of the firms. According to these
funds' detractors, they trick PIPE issuers to accept "bad" deals because issuing firm managers are unsophisticated or
have personal connections to the funds. The criticism says that the principal-agency conflict between shareholders
and management cause firms to issue PIPEs with "bad" terms.
A type of PIPE that has come under particular scrutiny is the floating rate convertible (FRC), which
is either a preferred stock or a bond that is convertible into common stock at a price determined by the future stock
price at the time of conversion. Chaplinsky and Haushalter (2003), Hillion and Vermaelen (2004), and others have
studied the firms issuing FRCs. They and finance practitioners have found a negative relationship between the
issuance of a floating rate convertible (FRC) and firm performance.
Existing evidence supports two theories for this relationship. The first, the faulty contract design
(FCD) hypothesis, states that firms mistakenly issue FRCs because they do not understand how they will impact long
term performance. The second theory, the financing of last resort (FLR) hypothesis, states that firms rationally issue
FRCs because they face severe information asymmetry problems and are unable to find financing elsewhere. This
second theory states that these firms' poor performances are simply anticipated by the FRCs themselves. Hindsight
indicates that certain FRCs
62
are not good financing options. Why then are they issued in the first place? How could have one predicted a poor
financing decision? And by implication, how can one predict poor financing decisions by firms now? I model the FRC
issue choice as a function of CFO characteristics. This approach yields interesting answers to the question of whether
FRC issues precipitate or simply anticipate poor future performance. It also addresses an interesting question of
whether the degree and type of sophistication of financial managers add value to the firm.
I show that the financial sophistication does indeed affect the FRC issuance decision and affects how
the market reacts to the issue. Firms where the CFOs are more highly compensated relative to their CEOs are more
likely to issue FRCs. Among FRC issuing firms, those with higher CFO-to-CEO compensation have lower abnormal
returns around the issuance announcements. In the long-term, FRC issuing firms whose CFOs do not have experience
accessing the public equity market have significantly negative stock returns. These findings support the FCD
hypothesis where the CFO has less experience. In addition, they support an alternative hypothesis of optimal security
design (OSD) where the CFO does have experience.
This paper is organized as follows. Section 2 reviews the literature on PIPEs and impact of financial
management on firm actions and performance. Section 3 discusses the sources of the data, methodology for
identifying control firms, and summary statistics. Section 4 presents hypotheses, including an introduction of the
optimal security design (OSD) hypothesis to the FRC literature. Section 5 analyzes the results of testing the
hypothesis. Section 6 concludes.
2. Literature
This study brings together two strands of financial research. The first is on the motivation and impact of PIPE
issues, particularly structured PIPE issues. The existing literature has come to support two major hypotheses for why
a firm would issue a floating rate convertible security (FRC): the faulty contract design hypothesis and financing of last
resort hypothesis. I join this research area to another strand, which is how CFOs' incentives and abilities along with
directors' financial ties affect financing actions of firms.
A. Private Investment in Public Equity (PIPEs)
How does a firm benefit from privately placing equity when it is already a registered public firm? In
a Myers and Majluf (1984) framework, a private issue could be a less costly way of conveying information to the
market while raising new equity. At the heart of Myers and Majluf's model is information asymmetry between a firm
and potential outside investors. In bad states, the firm knows that its stock is overvalued. In this case, the firm has the
reason to issue new equity because
63
it is overvalued. The market, then, would know whether the bad state of the world has been realized, and firm's stock
price would decline. If outside financing is necessary to invest in a positive net present value project, then some such
projects would not be taken. How, then, can this suboptimal investment policy be avoided? If information on the true
value of the assets and growth opportunities can be conveyed, then an issue of equity would not be a negative signal.
However, transmission of this information to the public is costly. For example, it may entail giving away trade secrets.
Theoretically, the certification benefit of PIPE issues is found in Hertzel and Smith (1993), which is
an extension of Myers and Majluf (1984)'s model of information asymmetry and security selection. Hertzel and Smith
(1993) suggest that private placements of equity can serve as a less costly way to convey information to the market and
thus allow for a more optimal investment decision rule in the Myers and Majluf framework. By issuing stock privately,
a firm can convey information through the negotiations with the investors, raise the capital it needs, and signal to the
market that it is issuing stock because of good growth prospects. Hertzel and Smith postulate and test that firms sell
their privately placed stock at a discount to pay for the due diligence costs.
The certification benefit of PIPEs is supported by the observation that the market reacts positively to
the announcement of PIPE issues even though they are typically issued at a discount to market price. Hertzel and
Smith (1993) strictly consider placements of common stock, and Chaplinsky and Haushalter (2003) consider all PIPEs
issued between 1995 and 2000. Consistent with Hertzel and Smith, Chaplinksy and Haushalter find positive
announcement returns for PIPEs. They also find that PIPEs with features that protect investors against declines in
stock prices have lower announcement returns than those with such protections. Examples of securities with
downside price protections are floating rate convertible debt and floating rate convertible preferred stock (FRCs).
They infer that the market interprets the terms as signals of the quality of the firms' prospects, because they indicate
that the issuing firms agree to the investors' concern that the firms' stock prices will decline in the future. The
convertible security holder may convert at a low stock price, forcing the firm to give more shares of stock than it
would at the time of the security's issuance. Terms with less downside protection indicates that private investors have
greater confidence in the firm's future performance. In other words, the structured nature of the PIPE signals that the
investor is unable to certify the value of the firm.
Studies on PIPE issuing firm performance that control for identity of the investor supports the
notion that there is a certification effect. Dai (2007) directly examines how VC's and hedge funds differ in their
relationship to PIPE issuing firms, finding that VCs tend to have seats on the issuing firms' boards of directors and to
have longer investment horizons than hedge funds. She finds
64
support for the argument that VCs provide certification to the PIPE issuers in which they invest. Similarly,
Krishnamurthy, Spindt, Subramaniam, and Woidtke (2004) find that the negative long run abnormal returns of firms
after issuing PIPEs are confined to the sample of nondistressed firms that issue to nonaffiliated investors. Distressed
firms, on the other hand, show a positive announcement effect and no long-run underperformance regardless of
whether the investors are affiliated or unaffiliated. This provides support to the hypothesis that the identity of the
investor provides a certification benefit.
Consistent with other studies (Hertzel and Smith, 1993; Dai, 2007; Krishnamurthy, et. al., 2004; and
Barclay, et. al. 2003), Brophy, et. al. (2006)¸ do not find support for a monitoring benefit in PIPE issues. The
monitoring hypothesis states that since PIPEs are issued to a block holder, the free- riding problem of equity
issuance to disperse shareholders is circumvented. The argument for a monitoring benefit is set forth by Wruck
(1989), who postulates that private placements enhance value by targeting a small number of investors rather than a
large number typical in a public issue. By concentrating ownership, investors are capable of monitoring the firm
more carefully. The discount that private investors usually receive, therefore, compensates for monitoring costs
according to Wruck's ownership concentration hypothesis. Hertzel and Smith, however, find that the ownership
concentration variable does not have statistical significance in a regression of abnormal returns in the presence of
certification variables. Barclay, Holderness, and Sheehan (2003) find that investors in PIPEs typically are passive, and
therefore they actually serve to entrench management, which they argue is why PIPE issues are typically followed by
long run negative abnormal returns.
Despite the support for a certification effect motivating the issuance of PIPEs at discounts, PIPE
issuers also tend to experience negative long run abnormal returns following the issue. This is documented by
Hertzel, et. al. (2002) and Chaplinsky and Haushalter (2003). Why then are PIPEs
issued?
One explanation is faulty contract design (FCD) hypothesis, which states that firms issue FRCs
because their managements do not recognize that the securities induce short selling in their firms' stock; the investors
of these FRC buy the shares in order to profit from anticipated manipulation of the stock price to the detriment of the
existing share holders. Even among nonstructured PIPE issues, Hertzel, et. al. (2002) find a negative post-event
performance, despite a positive announcement effect for private placements of equity. They interpret this result as
indicating that the issuing firm management is overly optimistic concerning growth opportunities. They speculate
that, rather than covering due diligence or monitoring costs, the discounts on privately placed stock indicate that the
investors in these issues are better informed concerning the firms' true lower value.
65
They do recognize such a story is inconsistent with an efficient market because, while privately issued stock are sold
at a discount, the market prices of existing stock are higher when the deals are announced.
Another explanation is the financing of last resort (FLR) hypothesis, which states that the issuing
firm needs external financing, and despite the costliness of the issuance, it will do so out of desperation. Indeed,
Chaplinsky and Haushalter (2003) find that over 80 percent of PIPE issuing firms have negative operating income
prior to an issue. They say that companies using PIPEs "appear to be highly distressed and have a high probability of
failure regardless of the actions taken by management. Therefore, it is difficult to judge the success of these contracts
based solely on the issuer's post issue performance." (Chaplinsky and Haushalter's study of PIPEs is restricted to the
1990s. Since 2000, however, PIPE financing has become more common among healthier firms.)
B. Structured PIPEs
If a non-structured PIPE provides a certification benefit, and the market reacts negatively to the issue of a structured
PIPE, then why would a firm opt to issue a structured PIPE? The development of hypotheses concerning relationship
between FRCs and their long run return and operating performance of their issuers is most comprehensively set forth
in Hillion and Vermalaen (2004). They test the faulty contract design (FCD) hypothesis, and financing of last resort
(FLR) hypothesis, as well as a third, the undervaluation hypothesis.
The undervaluation hypothesis states that firms issue FRCs because their price is undervalued.
Therefore, they issue a convertible security that converts at a future market price that the firm believes will match its
true value.
Using a dataset of all FRC issues from December 1994 - July 1998, Hillion and Vermaelen reject the
undervaluation hypothesis and find support for both the faulty contract design (FCD) and financing of last resort
(FLR) hypotheses. They reject the undervaluation hypothesis because firms that issue FRCs typically experience
significant negative abnormal returns in the years following issuance. Their support for the FCD hypothesis follows
from supporting several empirical predictions of this hypothesis. While negative abnormal returns would be expected
under both the FCD hypothesis and the FLR hypothesis, the reasons are different. One unique prediction of the FCD
hypothesis, for which the authors finds support, is that the conversion discount is negatively correlated with the
abnormal returns of the issuing firm and with the size of the issue. A unique prediction of the FLR hypothesis is that
the negative abnormal returns of an FRC issuing firm is accompanied by negative
66
abnormal operating performance because the issuance of the FRC is interpreted as a signal of future problems for the
firm. The authors find that this prediction holds also.
Chaplinsky and Haushalter (2003) consider all PIPE issues from 1995 - 2000. This study not only
considers more deals than Hillion and Vermaelen, but also provides further evidence of long run abnormal
performance of FRC issuers by comparing them to firms that issue PIPEs that do not provide the downside price
protection to the holders of the securities, such as straight equity and fixed rate convertible securities.
Brophy, Ouimet, and Sialm (2006) support the FLR hypothesis despite finding a relationship
between negative abnormal returns and the PIPE investor being more likely to be involved in short selling.
Specifically, they argue that firms issue FRCs as a last resort when they issue to hedge funds. They find that the long
run underperformance of traditional PIPEs is confined to the sample of firms issuing structured PIPEs to hedge funds.
This finding is consistent with the FCD hypothesis because hedge funds are not long term investors, but often engage
in significant amounts of short selling and operate by taking advantage of arbitrages (Dai, 2007). Nevertheless, the
authors do not accept the FCD hypothesis because the amount of short positions on firms issuing structured PIPEs to
hedge funds does not vary significantly from the short positions of other deals. Therefore, they argue that these
negative abnormal returns are due to these firms having the severest asymmetric information and agency problems,
risks that only hedge funds are able to hedge against. They also find that PIPE issuing firms that were backed by a
venture capitalist before IPOing do not experience the long run underperformance of other PIPE issuers. Their
interpretation of this finding is that VCs provide certification of these firms and therefore do not have the adverse
selection problems of firms without VC backing.
In summary of the literature on PIPEs, the motivation for the issuance is concluded to be despair
(FLR hypothesis) and/or ignorance (FCD hypothesis). A puzzling finding in the empirical literature, however, is that
there are long run negative abnormal returns to firms making private placements. Explanations for the puzzle have
been based on a lack firm management sophistication (ignorance; faulty contract design) or the market's slowness to
recognize that the underlying conditions of the issuer (despair; financing of last resort). Support has been provided
for both explanations. Observation of the level of sophistication and incentives of the issuing firm's financial
management would serve to disentangle the faulty contract design hypothesis from the financing of last resort
hypothesis. The literature has yet to incorporate such a control and therefore has not fully answered the question of
how much the security choice affects the value of the firm rather than merely signaling its existing value.
67
C. Financial Managers
Despite Hillion and Vermaelen (2004)'s observation that the amount of FRC's issuance has declined at the time of its
writing, Singh (2005) notes that there is still a significant amount of FRCs being issued. This observation either can be
indicative of the financing of last resort motivation functioning or a continued slowness of financial managers to
recognize the faulty design features of FRCs. According to Placement Tracker, there were 121 FRC deals in 2004.
While less than a third of issues made in 2000, this is still a significant amount of deals, amounting to $628 million
($3.2 billion in 2000). Though, Singh does find that more recently issued FRCs have terms that reduce the problems
that these FRCs have had in the past. Examples of terms that can control the price manipulation of the stock of FRC
issuers are (1) direct restriction on the investor from short selling, (2) floors on the conversion price, (3) restriction on
the number of shares convertible at one time, (4) reduced discounts relative to the reference price, and (5)
investigating whether the investor has engaged in price manipulation in the past. The change in these terms, therefore,
would support the view that these contracts were faulty and have merited being altered. However, not all issues
involve these
controls. Singh concludes,
"toxic convertibles represent a financial innovation that, through an
iterative and - unfortunately, for any investors - costly process, has improved its
design but is still used by the same types of firms as in the past..The rationale for use
of toxics appears to be driven by despair, ignorance, or both. However, with improvements
in contract design, as outlined earlier, smaller firms in need of capital that cannot
access the market for traditional securities are better
positioned to consider these securities"
The existence of research that cautions against the use of FRCs raises the question of why some
firms issue them and why others avoid them. The implication of the FCD hypothesis is that the financial
management of the issuers do not understand them. While there is a large literature on the impact of the chief
executive officer on the decisions and performance of the firm, there is an emerging literature on the impact of the
chief financial officer. CFOs do influence the performance of firms as evidenced by their removal following poor
performance in Mian (2001). The disciplinary removal of CFOs is robust to decision to retain or remove the CEO.
Chava and Purnanandam (2007) show that the incentives of the CFO, not the CEO, affect the firm's choice of floating
rate versus fixed rate straight debt. Brettel et. al. (2008) test Hackbarth (2004)'s model of firm leverage and CFO.
They find that firms with "overconfident" CFOs tend to have higher leverage.
While the above papers consider the interaction among firm performance, CFO actions, and CFO
incentives, the only area of research regarding direct measures of financial manager skill that the
68
author is aware of is in the mutual fund literature. Numerous studies document that mutual fund managers differ in
their stock picking skills (Wermers, 2000; Chen, et. al., 2001; Baker, et. al., 2005; and Harlow and Brown, 2007).
In a study of active fund managers, MBA school quality is positively and significantly related to fund
performance over 2000-2003 (Gottesman and Morey, 2006). This effect is particularly strong among
the top business schools, as ranked by Business Week. Other studies find a significance of the
undergraduate program from which the manager graduated (Chevalier and Ellison, 1999). Furthermore, other
certifications or degrees such as a CFA, other masters degrees, or Ph.Ds does not appear to correlate with fund
performance.
There also has been recent research regarding the financial connection and expertise of the board of
directors. Mitchell and Walker (2008) find that firms that are large and less likely to be in financial distress are more
likely to have commercial bankers on their boards. Also, firms that have higher leverage, less market value of equity,
and high investment / low Tobin's q or low investment / high Tobin's q are more likely to have bankers as directors.
Becker-Blease and Grein (2008) argue that
the advisory role of the board of directors needs to be considered in addition to the monitoring role.
III. Data Sample
A. Sources
All FRC deals made in 2001 and 2002 are considered. Deal data come from PrivateRaise. Stock
return and delisting information for each issuer come from CRSP, and financial statement data from Compustat.
Analyst coverage data come from I/B/E/S. CFO, CEO, and director data are hand collected primarily from Edgar
filings (10K's, 10Q's, and proxy statements mostly). Lexis Nexis and online business press articles are used where
Edgar and the companies web sites do not provide the information.
B. Pre-Issue Firm Characteristics
1. FRC Issuers
I identify the CFO characteristics of 61 firms issuing FRCs in 2001 and 2002. I am able to identify the CFO
characteristics of 43 firms that issued FRCs over the same period. Issuing firms tend have small size, high growth
opportunities, low leverage, high investment, operating losses, and high propensity to be in bankruptcy (median
Ohlson score of 0.88). See Table 1.
69
Table 1
Pre-Issue Firm Financial Characteristics
This table displays summary statistics of characteristics of FRC issuers and three groups of control firms, fixed price PIPE issuers, SEO issuers,
and match non-equity issuers measured in the year prior to the firms making these issues. (In the case of match non-issuers, FRC issuance
dates are imputed to them.) COVERED is an indicator that equals 1 if at least one stock analyst made a forecast for the firm. CFVOL is the
cash flow volatility, defined as the standard deviation of operating income up to twenty fiscal quarters before the announcement date. OSCORE is the probability of
becoming financially distressed defined by Ohlson (1980). FIRMVALUE is log of the market value of equity plus book values of preferred and total debt. TOBINQ is the
market value of the firm divided by the book value of the assets. LEVERAGE is the the long term debt divided by the book value of assets. INVESTMENT is the total of
R&D and advertising divided by the lagged property, plant, and equipment. PROFITABILITY is the operating cash flow before depreciation divided by lagged assets. FRC
issuers' variables' means are
significantly different from the fixed price PIPE issuers, SEO issuers, non-equity-issuers if denoted by an a, b, or c, respectively, at the 5% level.
Panel A: FRC Issuers
Variable
Statistic COVERED CFVO OSCORE FIRM- TOBINQ LEVERA INVEST- PROFIT-
L VALUE GE MENT ABILITY
N 43 43 43 43 43 43 43 43
Median 1.000 5.793 0.879 4.410 0.054 0.087 0.780 -0.192
Mean 0.581b 20.193 0.753 b 4.726b 0.180ab 0.470 2.563 -0.346bc
Standard Error 0.076 7.507 0.042 0.225 0.051 0.212 0.656 0.093
Panel B: Fixed Price PIPE Issuers
Variable
Statistic COVERED CFVOL OSCORE FIRM- TOBINQ LEVERA INVEST- PROFIT-
VALUE GE MENT ABILITY
N 40 40 40 40 40 40 40 40
Median 1.000 4.431 0.794 4.440 0.100 0.002 0.905 -0.139
Mean 0.600 10.784 0.715 4.245 0.135 0.100 2.942 -0.620
Standard Error 0.078 3.385 0.047 0.209 0.023 0.037 0.845 0.320
Panel C: SEO Issuers
Variable
Statistic COVERED CFVOL OSCORE FIRM- TOBINQ LEVERA INVEST- PROFIT-
VALUE GE MENT ABILITY
N 46 46 46 46 46 46 46 46
Median 1.000 18.352 0.530 6.792 0.015 0.555 0.076 0.150
Mean 0.934 25.127 0.529 6.693 0.030 0.562 1.787 -0.060
Standard Error 0.036 3.747 0.039 0.150 0.006 0.084 0.618 0.097
Panel D: Non-Issuers
Variable
Statistic COVERED CFVOL OSCORE FIRM- TOBINQ LEVERA INVEST- PROFIT-
VALUE GE MENT ABILITY
N 38 38 38 38 38 38 38 38
Median 1.000 14.222 0.870 5.047 0.035 0.022 0.183 -0.043
Mean 0.736 23.562 0.696 5.108 0.072 0.202 1.730 -0.057
Standard Error 0.072 4.221 0.053 0.308 0.018 0.043 0.624 0.043
The sample of FRC issuers tend to be in the information technology or pharmaceutical industries.
Using the 48 industry Fama-French industry definitions, the most represented industries among the 43 FRC issuing
firms are "Business Services" (ten), "Drugs" (five), "Medical Equipment" (four), and "Telecommunications" (four).
Only half of the firms have at least one analyst covering them in the year prior to issue. Therefore, I
mostly measure the degree of information asymmetry by using a dummy variable that equals one if the firm has at
least one analyst covering it in the prior year and zero otherwise. Among the 19
issuers that do have analyst coverage, the median earnings surprise is 30.80 percent. Am o n g
70
analysts covering the same firm, the median standard deviation in estimates of earnings is 16.76 percent. See Table 2.
Table 2
Pre-Issue Analyst Coverage Characteristics
This table displays summary statistics of analyst forecasts of FRC issuers and the three control group firms (fixed price PIPE
issuers, SEO issuers, and match non-equity-issuing firms). The mean surprise is the absolute percent difference in actual earnings from
forecasted earnings over the year. Dispersion is the standard deviation of forecasts among analysts. Maximum # of Analyst Coverage is the
maximum number of analysts covering the firm at the same quarter. FRC issuers' variables' means are significantly different from the fixed
price PIPE issuers, SEO issuers, non-equity-issuers if denoted by an a, b, or c, respectively, at the 5% level.
Panel A: FRC Issuers
Variables
Statistics
N
Median
Mean
Standard Error
Mean Surprise
19
0.308
0.396
0.065
Dispersion
19
0.005
10.103
9.472
Maximum # Analyst
Coverage
22
4.500
5.727b
0.947
Panel B: Fixed Price PIPE Issuers
Variables
Statistics
N
Median
Mean
Standard Error
Mean Surprise
21
0.302
0.709
0.234
Dispersion
20
0.059
1.041
0.895
Maximum # Analyst
Coverage
21
4.000
6.761
1.305
Panel C: SEO Issuers
Statistics
N
Median
Mean
Standard Error
Mean Surprise
45
0.082
0.291
0.107
Variables
Dispersion
45
0.049
0.109
0.032
Maximum # Analyst
Coverage
45
9.000
10.933
1.328
Panel D: Matched Non-Equity-Issuers
Variables
Statistics
N
Median
Mean
Standard Error
Mean Surprise
21
0.230
0.353
0.073
Dispersion
18
0.047
0.090
0.030
Maximum # Analyst
Coverage
21
5.000
10.761
3.110
These firm characteristics are consistent with prior studies of FRC issues and with the theory that
firms with severe information asymmetries and risk can use privately placed equity to find financing at a low enough
cost. Furthermore, the structured nature of FRCs indicates that investors have concerns that the values of the firms'
stock will drop. The apparent decline in number of FRC deals from 2001 to 2002 may also reflect the disproportionate
reluctance of investing in such firms as the stock market continued to cool.
71
2. Control Firm Identification
Finding the benchmark for these firms is a challenging task that some previous studies deal with.
Hillion and Vermaelen (2004) compare issuing firms to matched nonissuing firms using a propensity to issue matching
algorithm wherein they create a probit model for propensity to issue a FRC
(Dehejia and Wahba, 1998). They match every issuer with a nonissuer that has the closest propensity of
issuing a FRC.
Their probit model uses operating return on assets, profit margin, ROA, operating income / sales,
(capitalization expenditures + R&D) / assets, and market to book value of equity. I use the same or similar predictors
plus a measure for financial distress. I include the Ohlson financial distress variable ("O-score", or OSCORE)
(Ohlson, 1980) because I am considering a very special segment of stocks that are characterized by being in extreme
financial distress (Chaplinsky and Haushalter, 2003). These firms are likely to "go dark," i.e. cease trading on NYSE
or Nasdaq, which may occur because their stock price falls below the minimum levels allowed by the exchanges
(Leuz, Triantis, and Wang, 2006). Furthermore, the financing of last resort hypothesis states that it is precisely the
firms that are in financial distress that would issue a FRC.
In addition to OSCORE, I borrow variables used by Gomes and Phillips (2005) in their study of
public versus private security issuance choice. To measure risk, they use cash flow volatility
(CFVOL), defined as the standard deviation of operating income before depreciation (Compustat item data13) up to
twenty fiscal quarters before the announcement date. To measure information asymmetry, they use the mean earnings
surprise and dispersion of analyst earnings estimates discussed previously. Because half the sample of firms issuing
FRCs do not having any analyst coverage, I cannot use these variables without losing hal f the sample of firms,
leaving only 19. Therefore, I use a dummy variable, COVERED, that equals one if the firm is covered by at least one
analyst in at least one quarter prior to the issue, and zero otherwise. In addition to measures for risk and information
asymmetry, Gomes and Phillips (2005) use log of firm value, Tobin's q, leverage, investment in R&D, and
profitability as controls. The log of firm value (FIRMVALUE) is defined by the market value of equity plus book
values of preferred and total debt (in Compustat:
data24*data25 + data9 + data34 + data39). Tobin's q (TOBINQ) is defined by the market value of
the firm divided by the book value of the assets (exp(FIRMVALUE) / data6). The debt to asset ratio (LEVERAGE) is
defined by the long term debt divided by the book value of assets (data9t / data6t-1). Research and development
(INVESTMENT) is defined by the total of R&D and advertising divided
72
by the lagged property, plant, and equipment (data45t+data46t)/data8t-1. PROFITABILITY is defined
by the operating cash flow before depreciation divided lagged assets (data13t/data6t-1). 1
I regress the qualitative variable that equals one if a firm issues a PIPE of any kind and zero if it
does not on these variables with clustered standard errors by Fama-French 48 industry classification. The
regression model has an explanatory power of 9.86 percent. Consistent with previous research, firms that issue PIPEs
are more likely to be in financial distress and high R&D expenditure (all significant at the 5% or greater levels.) See
Table 3.
Table 3
Propensity to Issue a PIPE Logit Regression
Below are logit model estimates of the likelihood that a firm issues a PIPE. The standard errors are clustered by the Fama-
French 48 industry classification. COVERED is an indicator that equals 1 if at least one stock analyst made a forecast for the firm. CFVOL is
the cash flow volatility, defined as the standard deviation of operating income up to twenty fiscal quarters before the announcement date.
OSCORE is the probability of becoming financially distressed defined by Ohlson (1980). FIRMVALUE is log of the market value of equity
plus book values of preferred and total debt. TOBINQ is the market value of the firm divided by the book value of the assets. LEVERAGE is
the the long term debt divided by the book value of assets. INVESTMENT is the total of R&D and advertising divided by the lagged
property, plant, and equipment. PROFITABILITY is the operating cash flow before depreciation divided lagged assets. Estimates are
significant at the 10%, 5%, and 1% levels if denoted with *, **, or *** respectively.
Variable
COVERED
CFVOL
OSCORE
FIRMVALUE
TOBINQ
LEVERAGE
INVESTMENT
PROFITABILITY
CONSTANT
N
PSEUDO R-SQUARED
Coefficient Estimate
0.697***
-0.004
2.544***
0.094*
-0.017**
-0.106*
0.006***
-0.027*
-5.056***
10509
0.098
Robust Standard Error
0.132
0.002
0.206
0.051
0.008
0.061
0.002
0.014
0.302
Firms that are covered by at least one stock analyst are actually more likely to issue PIPEs. If
COVERED represents information asymmetry, then such a result is not expected if the regression were conditional on
issuing some type of equity security. However, the regression is based on all firms in the Compustat universe in 2001
and 2002. Therefore, firms that are not covered by an analyst at all, ceteris paribus, are less likely to issue a PIPE
(and in all probability, a SEO as well). If COVERED represents the costs to issuing equity publicly, then the fact that
PIPE issuers are likely to be covered by an analyst already may indicate that private costs borne by the managers of
the firm may be taken into consideration when making the issuance decision. Variables that are weakly significant in
the propensity equation are FIRMVALUE (positive sign), LEVERAGE (negative sign), and PROFITABILITY
(negative sign).
1 Gomes and Phillips (2005) use an alternative measure for financial distress, ALTMAN, a dummy variable that equals one if
the firm's Altman z-score is less than 1.81 (Altman, 2000). I also compute a propensity model using ALTMAN, and do not find qualitatively
different results. I choose the OSCORE because it produces a 1% higher pseudo-R2 in the logit regression.
73
Using the propensity scores computed on the universe of non-structured PIPE issuers, defined by
PIPEs that do not have downward price protections ("Fixed PIPEs"), SEO issuers ("SEOs"), and non- equity matched
issuers ("Non-Issuers"), I match the firms from each control group that have the closest propensity score
(PROPENSITY) to each FRC issuer.
Because the Non-Issuer control group has no issue date, for the purpose of the event study, I imput e
an event date to each Non-Issuer equivalent to the announcement date of a FRC issue of the firm with the closest
propensity score.
C. Deal Characteristics
1. FRC Deals
The median FRC deal is for $4.0 million and 12.1 percent of the issuing firm's market capitalization.
Table 4
PIPE Deal Characteristics
This table enumerates the characteristics of FRC deals and control set of fixed price PIPE deals.
Fixed Fixed
FRC PIPE FRC PIPE
Security Type Conversion Restriction
Common Stock 6 25 No 19 11
Convertible Debt 22 9 Unknown 14 25
Convertible Preferred 15 6 Yes 10 4
Stock
Conversion Type Selling Restriction
Fixed 0 40 No 27 26
Reset 22 0 Unknown 14 12
Variable 23 0 Yes 2 2
Warrants Included Hedge Restriction
No 15 23 No 23 25
Yes 28 17 Unknown 13 13
Yes 7 2
Antidilution Clause Forced Conversion
No 15 19 No 14 7
Unknown 12 11 Unknown 15 27
Yes 16 10 Yes 14 6
Hard Floor Price Investor Purchase
Rights
No 21 40 No 20 22
Unknown 7 0 Unknown 14 12
74
Yes 15 0 Yes 9 6
Soft Floor Price Mandatory Registration
No 33 40 No 6 8Unknown 7 0
Unknown 4 10 Yes 3 0 Yes 33
22
There is a wide range of deal sizes from $1.0 million to $1.5 billion, and percentages of market
capitalization from 1.7 percent to 108.8 percent. The median conversion price is calculated at 100.0 percent of the
reference stock price, with the lowest conversion price being 50 percent and the highest being 242 percent. These
deal characteristics are consistent with previous research finding that these deals are made at discounts and of
significant sizes. See Table 5.
Of the 43 deals, a significant number have at least one term that provides some type of limit on how
much investors can profit from stock price declines. See Table 4. Eighteen deals impose floors on the conversion
prices, and 10 deals restrict the amount of converting at one time. Two deals restrict short selling, and seven restrict
hedging by the investors, thus limiting their ability and incentive to exert selling pressure on the issuers' stocks. Ten
of the deals have forced conversion provisions, which would force the investors to convert their securities under
certain circumstances; this provision enables the firm to take advantage of a rising stock price by selling shares at the
future high market price.
Sixteen deals have anti-dilution clauses, which provide more shares to the investor if the firm issues
more shares in the future, so as to protect the investor from diluted value of shares. Thirty-three of the deals include
mandatory registration rights of the stock underlying the FRCs, making the investors able to sell the shares that they
convert. Nine of the deals have investor purchase rights, which gives the investor in the FRC the right of first refusal
when the firm attempts to issue future securities; thus, firms that issue these rights would be limited in its future
financing options.
Twenty-eight deals include warrants, which tend to have high exercise prices; median exercise price
is at a premium of 16 percent over the market price at issuance (See Table 5). Inclusion of warrants at premiums may
offset the interest the investors have in stock price declines. These warrants represent sizable stakes in the warrant
issuers, ranging from 4.4 percent to 150.0 percent of outstanding shares with a median of 33.3 percent.
75
2. Fixed-Price PIPE Deals
The deal sizes of the fixed PIPE issues of the control group are similar to those of the FRC issues.
The median deal size is $6.5 million and 16.0 percent of market capitalization. The median PIPE price is 97.0 percent
of the market stock price.
Fixed price PIPEs do not have the features that pressure the stock prices down as do FRCs.
Therefore, the net benefits of including specific terms differ between the two types of issuances. Fixed price PIPE
issuers are less likely to include conversion restrictions (4 out of 40 deals) and less likely to include hedge restrictions
(2 out of 40), indicating that some issuers who choose to issue FRCs over fixed price PIPEs negotiate terms that are
particularly important to moderating the impact of the FRC. However, other terms are equally present in the two types
of deals: short-selling restrictions and forced conversion clauses.
The fraction of deals with pro-investor terms is higher among FRC issues than fixed price PIPE
issues. Fixed price PIPE deals are less likely to have anti-dilution clauses (10 out of 40 deals) and less likely to have
registration rights (22 out of 40 deals). This pattern is contrary to that of the greater prevalence of pro-issuer rights
among FRC deals than fixed PIPE deals. Perhaps the investors in FRCs demand that they be protected from the
higher likelihood that the firm will need to raise more capital in the future via anti -dilution clauses and from the
likelihood that the investor will be unable to sell its shares if it waits too long to seek to have them registered.
Fixed PIPE issuers are also less likely to include warrants. Only 17 out of the 40 fixed price PIPE
deals include warrants. The lower rate of inclusion of warrants among fixed price PIPE issuers may indicate the lesser
need to provide rewards to the investor for upward movements in the stock price. The fixed price PIPE deals' median
warrant premium, 6.0 percent, and median warrant coverage amount, 50.0 percent, are comparable to the FRC issues.
3. SEO Deals
The size of SEO deals in comparison to the FRC deals is greater in absolute dollars but similar in percent or market
capitalization. The median SEO amount is $114.5 million, and the median deal amount as percent of market
capitalization is 16.8 percent. The median SEO price is 95.9 percent of the market stock price.
76
Table 5
Deal Characteristics
The pricing and quantity of deals among the FRC, fixed price PIPE, and SEO issuers are given below. The deal amount /
market cap is the amount raised divided by the market value of the stock at the time of issuance. The premium is the
percentage of the reference stock price. The warrant premium is in term of the stock price at the time of issuance. The warrant coverage is
the number of shares underlying the warrants as a percent of outstanding shares. FRC issuers' variables' means are significantly different
from the fixed price PIPE issuers, SEO issuers, non-equity-issuers if denoted by an a, b, or c, respectively, at the 5% level.
Panel A: FRCs
Statistics Deal Amount Deal Amount / Premium (%) Warrant Warrant
($000) Market Cap (%) Premium (%) Coverage (%)
N 43 43 43 28 27
Median 4.000 12.100 100.000 115.500 33.300
Mean 45.518b 21.041 103.504 117.642 49.644
Standard Error 34.694 4.035 5.554 8.740 7.447
Panel B: Fixed Price PIPEs
Statistics Deal Amount Deal Amount / Premium (%) Warrant Warrant
($000) Market Cap (%) Premium (%) Coverage (%)
N 40 40 40 17 17
Median 6.500 16.000 97.000 106.000 50.000
Mean 10.047 21.707 99.017 112.970 64.635
Standard Error 1.611 4.136 5.610 5.603 11.384
Panel C: SEOs
Statistics Deal Amount Deal Amount / Premium (%)
($000) Market Cap (%)
N 46 44 44
Median 114.550 16.760 95.893
Mean 171.258 36.238 101.490
Standard Error 33.836 11.037 4.294
D. CFO and Other Management Characteristics
CFO Characteristics. The median age of the CFOs of FRC issuers is 43.5 years. The median amount
that CFOs own of the firm is 1.0 percent. Fifty-one percent of CFOs have experience as a CFO, treasurer, or
comptroller of another public firm. However, only 27 percent were financial officers at public firms when those firms
issued either IPOs or SEOs. Among the more limited number of FRC issuers for which I could find educational data,
42 percent have CFOs who have MBAs. CFOs of FRC issuers are less likely to have been at the firm when it went
public than the control firms. See Table 6.
In addition to the CFO variables in Table 6, the tenure of the CFO at her firm is collected. Only two
of the CFOs of FRC issuers were at their firms when they made their initial public offerings. However eight fixed
price PIPE issuers, seven SEO issuers, and four non-equity issuers have CFOs at the time of issue who also were
CFOs at the time of IPO. See the Appendix for the biographical information provided by firms on their CFOs,
organized by CFO public equity experience.
77
Boards of Directors. The median board size of FRC issuers is six. The median percent of board members who are
also employees of the firm ("insiders") is 33.33. The median percent of board members who are financial
professionals is 16.67. Financial professional is defined as being an employee of a financial institution, such as an
investment bank or hedge fund, or being a financial officer, such as a CFO, of another firm. The median of the mean
age of the firms' directors is 54.8 years. The median ownership of the firm by all directors and managers is 18.1
percent. See Table 6 for summary statistics on all four groups of firms' managerial characteristics.
Table 6
Management Characteristics
The descriptive statistics of the CEOs, CFOs, and boards of directors of the FRC issuers and control firms are shown the
following panels. These values are collected in the period prior to the security issuances, or imputed issuances in the case of
the matched non-equity-issuers. CFO / CEO Comp is the fraction of CFO cash compensation to CEO cash compensation. % Board Insiders is
the percent of board members who are also managers of the same firms. % Board Financial Experts is the percent of board members who are
either employed by a financial institution or is a CFO. FRC issuers' variables' means are significantly different from the fixed price PIPE
issuers, SEO issuers, non-equity-issuers if denoted by an a, b, or c, respectively, at the 5% level.
Panel A: FRC Issuers
Standard
Variable N Median Mean Error
CFO with Prior Public Firm Experience 53 1 0.547b 0.069
CFO with Public Equity Offering Experience 54 0 0.203 b 0.055 Total Director and
Management Ownership 51 18.100 24.306 b 2.571
CFO/CEO Comp 44 0.622 0.692 b 0.051 %
Board Insiders 51 0.333 0.390 ab 0.041 %
Board Financial Experts 51 0.166 0.217 0.032
Avg. Board Age 51 54.80000 53.649 0.891
CFO with MBA 57 0.000 0.421a 0.065
CFO with Unknown Education 57 0.000 0.175 abc 0.050
CFO Ownership 47 0.010 0.020 b 0.005
CEO Ownership 51 4.4000 9.447 ac 1.939
CEO Compensation 50 325072 611264 236765
CFO Compensation 45 199615 229416 b 19358
CFO Age 36 43.500 42.250 b 1.257
Board Size 38 6.000 6.289 0.330
Panel B: Fixed Price PIPE Issuers
Standard
Variable N Median Mean Error
CFO with Prior Public Firm Experience 40 0.000 0.425 0.079
CFO with Public Equity Offering Experience 40 0.000 0.325 0.075
Total Director and Management Ownership 42 13.910 21.130 2.906
CFO/CEO Comp 41 0.606 1.978 1.323 %
Board Insiders 43 0.250 0.294 0.024 %
Board Financial Experts 43 0.222 0.271 0.030
Avg. Board Age 43 54.500 53.820 0.732
CFO with MBA 47 0.000 0.234 0.062
CFO with Unknown Education 47 0.000 0.489 0.073
CFO Ownership 42 0.010 0.012 0.003
CEO Ownership 42 0.0272 0.0626 0.016
CEO Compensation 43 310000 365279 49104
CFO Compensation 42 200053 210114 17327
CFO Age 34 42.500 43.882 1.452
Board Size 36 6.000 6.222 0.314
78
Panel C: SEO Issuers
Standard
Variable N Median Mean Error
CFO with Prior Public Firm Experience 41 1.000 0.804 0.062
CFO with Public Equity Offering Experience 41 1.000 0.951 0.034
Total Director and Management Ownership 40 7.850 16.052 3.024
CFO/CEO Comp 41 0.409 0.502 0.041 %
Board Insiders 39 0.285 0.290 0.024 %
Board Financial Experts 39 0.200 0.215 0.025
Avg. Board Age 38 55.160 55.832 0.949
CFO with MBA 47 0.000 0.361 0.070
CFO with Unknown Education 47 0.000 0.468 0.073
CFO Ownership 40 1.000 0.985 0.027
CEO Ownership 40 2.045 5.352 1.617
CEO Compensation 41 806077 1033665 120987
CFO Compensation 41 357180 389269 27236
CFO Age 44 45.000 45.750 1.112
Board Size 36 7.000 7.194 0.313
Panel D: Non-Issuers
Standard
Variable N Median Mean Error
CFO with Prior Public Firm Experience 39 0.000 0.435 0.080
CFO with Public Equity Offering Experience 39 0.000 0.282 0.073
Total Director and Management Ownership 39 21.400 25.801 3.421
CFO/CEO Comp 38 0.586 0.660 0.074 %
Board Insiders 39 0.333 0.366 0.039 %
Board Financial Experts 39 0.250 0.270 0.031
Avg. Board Age 39 53.571 52.761 0.992
CFO with MBA 47 0.000 0.319 0.068
CFO with Unknown Education 47 1.000 0.510 0.073
CFO Ownership 39 0.010 0.013 0.001
CEO Ownership 40 0.023 0.071 0.014
CEO Compensation 39 399250 547564 69821
CFO Compensation 39 206153 296634 52068
CFO Age 32 45.000 44.750 1.448
Board Size 31 7.000 7.129 0.421
PIPE Investor Identity. Investors in FRCs are unlikely to be firm managers. Only two FRC issues
are purchased by management, whereas eight fixed price PIPE issues were purchased by management. FRC and
fixed price PIPE issues are made to investors who hold seats on the issuers' boards at a rate less than ten percent. Four
FRC issuers and six fixed price PIPE issuers sold PIPEs to institutions with board seats.
E. Announcement Cumulative Abnormal Returns
A Fama-French three-factor plus momentum model was built to calculate abnormal returns. The CARs are also
robust to alternative market models: the capital asset pricing model and the Fama- French three factor model. The
coefficients for each model are determined using a 250 market day window prior to three months prior before the
announcement dates. Table 7 presents the cumulative abnormal returns (CARs) over an event windows covering five
days prior through five days after the announcement ([-5,+5]).
79
Table 7
Cumulative Abnormal Returns by Firm Group
Cumulative abnormal returns are calculated five days prior through five days after the issuance announcement (imputed
announcement in the case of the matched non-equity-issuers). The abnormal return is computed by using a four factor
market model: the excess market, small minus big portfolio, high minus low market to book portfolio, and the winners minus losers
momentum portfolio. The coefficients on each of these are calculated over the 250 market day period a month prior to
the events for each firm. Estimates are significant at the 10%, 5%, and 1% levels if denoted with *, **, or *** respectively.
Freque Standard
Firm Type ncy Minimum Median Maximum Mean Error
FRC Issuer 41 -0.387 -0.040 0.682 -0.011 0.035
Fixed Price PIPE Issuer 34 -0.421 0.004 2.339 0.152* 0.083
SEO Issuer 35 -0.188 0.024 0.889 0.063 0.030
Non-Equity Issuer 31 -0.301 0.016 0.473 0.028 0.030
Total 141 -0.421 0.013 2.339 0.055** 0.024
F. Post-Issue Firm Characteristics
In the year of the FRC issuance, the only financial ratios presented in Table I that significantly change is financial
distress (OSCORE) and the value of the firm (FIRMVALUE). The mean probability of financial distress increases
13.01 percent and the mean firms' value decreases by $0.50 million, both with significances greater than the one
percent level. From the year of issue to the following year, neither the level of financial distress nor firm value change.
However, from the year of the issue to the following year, Tobin's q increases by eight percent and R&D drops by 64
percent, both at the five percent level of significance. These patterns are also robust to industry adjustments. These
changes in operating performance over the period of a year prior to the year of issue are only found in the group of
FRC issuers. SEO issuers and non-equity issuers experience significant drops in the probability of financial distress
(SEO: -10.1 percent; Non-Issuers: -12.2 percent). SEO issuers' value significantly increases by 23.2 percent, and their
leverage significantly decreases by 16.4 percent. Fixed price PIPE issuers and non-equity-issuers decrease their levels
of R&D investment (Fixed PIPE: -127.6 percent; Non-Issuers: -51.2 percent). Among the control groups, there is little
change from the year of issue to the following year. Fixed price PIPE issuers are more likely to be in financial distress
by a mean of 10.5 percent. Non-equity issuers' profits increase by a mean of 6.7 percent.
While the level of R&D investment does not decrease significantly from the year prior to and year of
issue among FRC issuers, the fixed PIPE issuers and non-equity-issuers have significant decreases in R&D. However,
in the year following issue, only FRC issuers significantly reduce R&D investment. This finding is consistent with
Hillion and Vermaelen (2004)'s, who interpret the relative initial run up in R&D and subsequent decline in R&D as a
pre-issue confidence by issuing management in future returns to their investment.
80
Of the 43 FRC issuers, ten are delisted and one is acquired within one year following the issuance. However, only two
of the 40 fixed-price-PIPE issuers are delisted and three are acquired in one year following the announcement. None of
the 46 SEO issuers are delisted and one is acquired. The only control group that has a similar number of delistings to
the FRC issuer group is 38 non-equity issuers, of which seven are delisted.
The similarity between FRCs and non-issuers in terms of delisting rates indicates that the issuance
of a FRC does not increase its chances of remaining in business (or, is predictive of being able to stay in business). On
the other hand, fixed PIPE issuers and SEO issuers have a greater chance of remaining in business and being
acquired than FRC issuers and non-equity issuers. This pattern does not support the FLR hypothesis.
IV. Hypotheses
Hypotheses concerning the rationale for the existence of FRCs are presented in Hillion and
Vermaelen (2004): the undervaluation, faulty contract design, and financing of last resort
hypotheses. Two of these, the FCD and FLR hypotheses are not rejected by the authors, and they are two explanations
that are commonly accepted today for the existence of FRCs. Either of the two
hypotheses place FRCs in a negative light. In the case of the FCD hypothesis, the FRC is an
instrument of predation by unscrupulous investors who take advantage of firms whose management is either ignorant
or in collusion with the investors. In the case of the FLR hypothesis, the FRC is the only source of continued financing
for firms that would otherwise have to cease operations.
While the latter hypothesis offers a rational explanation for the existence of FRCs, it does not offer a
more positive explanation than that they are cheap enough for investors so that they will be willing to purchase them.
Why, however, is the floating conversion feature included rather than a more deeply discounted fixed conversion
price? The FLR hypothesis does not provide an answer.
Additional explanations appeal to fundamental concepts in corporate finance. FRCs may exist to
resolve problems that reduce firm value, which may be explained by an "optimal security design" (OSD) hypothesis.
Two possible examples of the OSD hypothesis are (i) the tradeoff between the
debt tax shield and financial distress and (ii) the problem of debt overhang:
(i)
Trade-Off: A firm determines an optimal leverage by maximizing the net benefit of the
debt tax shield minus bankruptcy costs. The costs of bankruptcy are higher with the likelihood of
default. If the market value of the stock falls after the issuance of debt and equity securities, the tax
shield benefits decrease due to lower likelihood of having any
81
positive pre-tax income before interest payments for the coming years and the bankruptcy costs
increase due to higher likelihood of bankruptcy. The FRC causes the investors to change the debt-equity
ratio in this scenario without the firm having to negotiate with bondholders.
(ii) Debt Overhang: A firm issues preferred stock or debt. After issuance, the real option
value of its growth prospects decrease. This is reflected in a declining stock price. If the capital structure
remains the same, the management, acting in the interest of the common equity holders, does not take
positive NPV projects because of the overhang
from debt or required dividends (Myers 1977). To avoid the renegotiation costs of
exchanging the preferred stock or debt for common stock, the firm issues a FRC so that when the real
option value goes down, the FRC is automatically converted into common shares because it is in the
interest of the FRC investors.
The hypotheses that this study tests are the FCD, FLR, and OSD hypotheses. The FLR hypothesis
and OSD hypotheses are similar in that they provide a rationale for FRCs that does not depend on behavioral
arguments. They differ in that the FLR hypothesis basically says that FRCs are issued because they are cheap for the
investor, but the OSD hypothesis provides rationales specifically for
the floating conversion price feature. Because this study's purview is empirical and the OSD
hypothesis is only introduced here, I only test the OSD hypothesis in the broad sense. Testing the specific
manifestations of the OSD hypothesis is reserved for future research.
All firms do not necessarily issue FRCs for the same cause. Some firms may be unwitting victims of
exploitive investors who intend to manipulate the stock price downward in order to expropriate a larger portion of the
firm's equity. Many firms could be fully aware of the costs of FRC issuance and make the decision to issue the FRC
wisely. Therefore, the empirical tests of this study do not seek to
accept one hypothesis as true for all firms and reject other hypotheses. Rather, the tests are
designed to test whether certain characteristics of firms' management can be linked to the various explanations for
FRC issuances.
The group of FRC issuing firms is compared to the three control groups: fixed price PIPE issuers,
SEO issuers, and no- equity issuing matched firms. Abnormal announcement returns are based on the Fama French
three-factor plus momentum model and are computed using the 250 market day window prior to the month of
issuance. The control groups are determined by using a score measuring the propensity to issue a PIPE. The non-
equity-issuing matched firms are imputed with
82
event dates the same as the FRC issuing firm with the closest propensity scores. Managerial
characteristics of FRC issuers relative to the control firms are related to the following three areas: propensity to issue
FRCs, stock market FRC announcement reaction, and long-term stock performance of FRC issuers.
In the first area, multinomial logit regression of the issuance type among the control groups is run
on managerial characteristics. Significant coefficients on sophistication and/or incentives provide
support to the FCD hypothesis and lack of support to the FLR and OSD hypotheses. If less
sophistication, greater conflicts of interest, and less monitoring of financial managers correlate with greater likelihood
of issuing a FRC, then the FCD hypothesis is supported. The FLR and OSD hypotheses would not necessarily be
refuted because it would be possible for certain FRC issuers to issue FRCs that benefit their firms if they have CFOs
who have above average sophistication than the entire sample of FRC issuers.
In the second area, the stock market cumulative abnormal return is regressed on the managerial
characteristics. If the market reacts negatively to firms issuing FRCs with unsophisticated or
conflicted management, then the FCD hypothesis may be true for those firms. In addition, I would interpret such a
relationship to support the OSD hypothesis because the market only would be reacting negatively when the FRC is
issued to the determinant of existing shareholders. I would not interpret such a finding to support the FLR hypothesis,
however, because if the market differentiates FRC issues by managerial characteristics, then it should also be able to
interpret the issuance as a negative signal immediately upon announcement rather than over a long period of ti me
after issuance.
In the third area, calendar time alphas of a portfolio long on experienced and short on inexperienced
FRC issuers is computed. If this portfolio is positive, then this supports the FCD and OSD
hypotheses. I would conclude that a non-equity-value maximizing choice was made where the firm management is
less experienced. On the other hand, among the firms where the firm management is experienced, I would conclude
that the FRC was issued to maximize existing shareholder value if
no long run abnormal returns are detected among these same firms. I would reject the FLR
hypothesis if only the FRC issuers with inexperienced CFOs have negative long run returns because the FLR
hypothesis predicts that all FRC issuers experience negative long run returns.
83
V. Results
A. Relationship of Managerial Characteristics on Issuance Choice
Unconditional Mean Comparison of Firm Managements. If FRCs are issued because they are faulty contracts rather
than financings of last resort, then the firms that issue FRCs are less financially savvy than similar firms. Or, their
management has conflicts of interest whereby they have interests with the FRC investors. I show that firms with CFOs
who have had experience placing equity
publicly and those with boards that are less composed of insiders are less likely to issue FRCs.
The unconditional means between the two groups show that 85.85 percent of the non-FRC issuing
firms have CFOs who are the CFO, treasurer, or comptroller over a firm when it makes either an IPO or SEO. On the
other hand, only 27.91 percent of the FRC issuing firms had CFOs with such experience. This is statistically
significant at the five percent level. (Compare Panel A with the other
panels of Table 6.) This difference in groups that are otherwise just as likely to issue a PIPE
suggests that the CFOs' experience issuing public equity predicts which kind of security is issued. This result supports
the argument underlying the FCD hypothesis that firms issue FRCs because their management is not comfortable with
alternative ways to issue equity.
The unconditional means of the two groups show that the average percent of insiders on the board of
directors is 29.84 percent among non-FRC issuing firms. (See Table 8.) On the other hand, the average insider
percentage is 40.33 percent among FRC issuers. The difference is statistically significant at the five percent level.
This difference in groups suggests that the quality of board monitoring affects the decision to issue FRCs. This result
also supports the belief that faulty contract design is an influential reason for firms to issue FRCs because if FRC
issuance were to maximize the value of existing equity in the firm, then firms whose boards are more independent
from management would permit the issuances of FRCs.
Other variables do not display statistically significant differences between the two groups. Whether
the CFO was previously a financial officer of a different public firm or a non-public firm does not make any
difference between the FRC issuers and the control firms. Officer and director ownership of the firm is not different
between the two groups. The percent of directors who are financial experts, i.e. those who are officers in a financial
institution or are CFOs themselves, is not different between the two groups either.
The data of one variable, MBA, is frequently less available than the others in the firms' filings.
Among the control group, 60.32 percent of firms' CFOs did have MBAs, and among FRC issuers, 44.44% had MBAs.
However, this difference is not statistically significant. Because this field has
84
more missing values, the regressions in the next section use a dummy variable that equals one when the education of
the CFO is unknown and zero otherwise, and the MBA variable equals one if the CFO is known to have an MBA and
equal to zero if otherwise.
Regression. A multinomial regression is run on the managerial characteristics and the PIPE
issuance propensity score. The dependent variable takes four values: 0 if the firm is a matched non- equity issuer, 1 if
a FRC issuer, 2 if a fixed price PIPE issuer, and 3 if a SEO issuer. The propensity score is included as a regressor as an
additional control beyond the fact that the firms are already similarly matched. The results of the regression ar e robust
to excluding the propensity score. I find that firms with CFOs with prior public equity experience are less likely to
issue FRCs at the ten percent level. The regression does not support the unconditional means finding that more board
insiders are positively correlated with FRC issuance in the presence of other management variables. I do find that the
ratio of CFO to CEO pay is positively correlated with PIPE issuance over SEO or no equity issuance at the five
percent significance level. This relationship may be saying that, all else equal, a firm where the CFO is more on par
with the CEO in influence is more likely to issue PIPEs.
Table 8
Issuance Choice: Multinomial Logit Regression
A multinomial regression model is presented. The left hand side takes four possible: FRC issuer, fixed price PIPE issuer,
SEO issuer, and match non-equity issuer. The non-issuer is the excluded class. The coefficients with standard errors in
parentheses are given below. Statistical significance at varying levels is denoted with a * (10%), ** (5%), and *** (1%).
Issuer Type
Variable
CFO with Prior Public Firm Experience
CFO with Public Equity Offering Experience
Total Director and Management Ownership
CFO/CEO Comp
% Board Insiders
% Board Financial Experts
Avg. Board Age
CFO with MBA
CFO with Unknown Education
CFO Ownership
FRC Issuer
1.253*
(0.745)
-1.451*
(0.864)
4.23E-05
(0.022)
3.707**
(1.599)
1.183
(1.597)
-1.768
(1.752)
0.032
(0.059)
-0.273
(0.817)
-2.117
(0.919)
Fixed Price PIPE
Issuer
-0.634
(0.704)
0.156
(0.758)
-0.016
(0.020)
3.420**
(1.555)
-2.984
(2.173)
0.881
(1.636)
0.077
(0.060)
-0.893
(0.879)
-0.687
(0.798)
-0.161
SEO Issuer
-0.138
(1.270)
5.249***
(1.550)
-0.188**
(0.075)
-3.253*
(1.807)
0.0662
(2.056)
2.562
(3.245)
-0.023
(0.089)
-0.143
(1.471)
-1.170
(1.615) -
0.942
85
CEO Ownership
CFO Compensation
Propensity to Issue PIPE
Intercept
Number of Observations
Pseudo R-squared
-0.071
(0.198)
0.040
(0.037)
-4.49E-06*
(2.41E-06)
-4.335
(8.496)
-2.125
(3.478)
112
0.419
(0.224)
0.022
(0.040)
-8.07E-06***
(3.08E-06)
-4.826
(8.173)
-1.935
(3.604)
(1.756)
0.188**
(1.756)
2.66E-06
(2.10E-06)
-8.718
(10.950)
1.521
(5.463)
B. Stock Market Reaction to FRC Announcement by Managerial Characteristics
The variation in the stock market reactions to FRC issuances is explained partly by management characteristics. Four
separate regressions are run, one for each group of firms. See Table 9. Among FRC issuers, the only characteristic at
the five percent level of significance in CARs around the announcement is the ratio of CFO compensation to CEO
compensation. The relationship between abnormal returns and this variable is negative, suggesting that the market
reacts negatively to FRC issuers by firms in which the CFO has more clout relative to the CEO.
The only other variable displaying significance in the FRC announcement reaction regression is the
unknown education dummy variable, having a positive sign. The interpretation for this result is unclear. Unavailable
educational data may be a proxy for either less reporting quality, which would
not support the argument that managerial quality impacts fi rm value. Alternatively, citing
educational background may be a substitute for citing experience, in which case the significantly positive sign on
unknown education would support the argument that managerial quality does impact firm value.
Number of board insiders appears as a significant variable in the SEO and Non-Issuer regressions,
but not the FRC issuer regression. This may indicate that corporate governance is a concern when SEOs are issued or
no issuance is made by firms similar to the issuing firms.
Apart from the CFO to CEO compensation ratio, I do not see market reaction being sensitive to
managerial characteristics in a clearly interpretable way. The market may be concerned in
particular by a FRC issuance announcement when the CFO is more highly compensated. This
relationship supports the FCD hypothesis where the CFO has less oversight because she holds a higher rank in the
firm.
86
Table 9
Regressions: 11 day CAR on Management Characteristics by Firm Type
Each column represents a separate regression of the CAR [-5,+5] around issuance announcements (or imputed
announcements) on the managerial variables. Cumulative abnormal returns are calculated five days prior through five days
after the issuance announcement (imputed announcement in the case of the matched non-equity-issuers). The abnormal return is computed
by using a four factor market model: the excess market, small minus big portfolio, high minus low market to book portfolio, and the winners
minus losers momentum portfolio. The coefficients on each of these are calculated over the 250 market day period a month prior to the
events for each firm. The managerial values are collected in the period prior to the security issuances, or imputed issuances in the case of the
matched non-equity-issuers. CFO / CEO Comp is the fraction of CFO cash compensation to CEO cash compensation. % Board Insiders is
the percent of board members who are also managers of the same firms. % Board Financial Experts is the percent of board membe rs who
are either employed by a financial institution or is a CFO. The coefficients with standard errors in parentheses are given below. Statistical
significance at varying levels is denoted with a * (10%), ** (5%), and *** (1%). FRC issuers' variables' means are significantlydifferent from
the fixed price PIPE issuers, SEO issuers, non-equity-issuers if denoted by an a, b, or c, respectively, at the 5% level.
Fixed Price
Variable
CFO with Prior Public Firm Experience
CFO with Public Equity Offering
Experience
Total Director and Management
Ownership
CFO/CEO Comp
% Board Insiders
% Board Financial Experts
Avg. Board Age
CFO with MBA
CFO with Unknown Education
CFO Ownership
CEO Ownership
CFO Compensation
Propensity to Issue PIPE
Intercept
FRC Issuer
-0.048
0.105
0.019
0.156
0.005
0.003
-0.396**c
0.179
-0.104 c
0.146
0.271
0.221
0.014
0.016
-0.008
0.132
0.364** ac
0.139
-0.090
0.057
0.005
0.005
3.44E-7
3.79E-7
2.159
2.162
-0.926 c
0.999
PIPE Issuer
0.027
0.416
-0.066
0.408
-4.69E-4
0.009
-0.611
0.857
0.100
0.975
0.120
0.848
0.001
0.028
-0.194
0.353
-0.256
0.273
-0.106
0.290
-0.001
0.015
2.53E-7
1.80E-7
4.299
3.540
-0.283
1.628
SEO Issuer
0.0839487
0.0526844
-0.1092252
0.1125912
0.0079937
0.0041605
-0.0588477
0.1246221
-0.3762595**
0.1701241
-0.0543524
0.1641585
-0.0050111
0.0053303
0.0768565
0.0938519
0.011357
0.1030467
0.3157607
0.2763186
-0.0056735
0.0045315
-1.18E-7
1.19E-7
1.204**
0.580
0.074
0.354
Non-Issuers
-0.1047058
0.0829212
-0.0378850
0.0892258
-0.0040279
0.0031122
0.0409362
0.1454364
-0.7817884**
0.2892039
-0.0922416
0.2475943
-0.0143507**
0.0068937
0.1083244
0.1307238
0.1142139
0.1187252
0.0272749
0.0456118
0.0154634**
0.0064923
1.73E-7
1.46E-7
2.255**
0.919
0.719
0.473
C. Long-Run Stock Performance
87
Four portfolios are created of firms that either issued each for each group of firms. The monthly return of these firms
is regressed on the Fama French plus momentum factors over 2001-2003. Each group's monthly portfolio is composed
of firms who issue (or are imputed with an issue) in the twelve months prior to the particular month. The non-issuers
have an issuance date imputed to them according to the issuance dates of the FRCs to which the non-issuer have the
closest propensity
scores. The significances of the estimates of alpha are evaluated by the mean return of 1,000
regressions using random samples of firms in the same size and book-to-market deciles as the FRC issuers (Mitchell
and Stafford, 2001). The portfolio returns of firm group p in month t are regressed
on a constant, the Fama French factors, and the momentum portfolio:
r
p
,
t
= o
p
+ |
p
,MKT r
MKT
+ |
p
,SMBr
SMB
+ |
p
,HMLr
HML
+ |
p
,MKT r
MKT
+ |
p
,UMDr
UMD
+ c
p
,t
Table 10
Calendar Time Portfolio Alphas
Each row corresponds to a separate regression that predicts the return of the row's portfolio description. The monthly returns
are regressed on the Fama French plus momentum factors over 2001-2003. Each group's monthly portfolio is composed of firms who i ssue
(or are imputed with an issue) in the twelve months prior to the particular month. The non-issuers have an issuance date imputed to them
according to the issuance dates of the FRCs to which the non-issuer have the closest propensity scores. The significances of the estimates of
alpha are evaluated by the mean return of 1,000 regressions using random samples of firms in the same size and book-to-market deciles as the
FRC issuers (Mitchell and Stafford, 2001).
Portfolio Alpha MKTRF SMB HML UMD R2
FRC Issuers -0.024 1.763*** 0.295 0.730 -0.267 0.54
(0.018) (0.495) (0.533) (0.570) (-0.400)
FRC Issuers - Fixed -0.013 -0.732 -1.135 0.763 -0.621 0.13
PIPE Issuers (0.025) (0.679) (0.731) (0.781) (0.548)
FRC Issuers - SEO -0.017 1.012 -0.658 0.919 -0.312 0.22
Issuers (0.022) (0.613) (0.660) (0.705) (0.495)
FRC Issuers - -0.050* 0.788 -0.651 0.624 -0.109 0.10
Nonissuers (0.025) (0.688) (0.741) (0.792) (0.555)
Inexperienced FRC -0.045** 1.441** 0.222 0.760 -0.291 0.42
Issuers (0.020) (0.538) (0.580) (0.620) (0.435)
Experienced FRC 0.097** 1.452 0.484 -0.824 0.480 0.12
Issuers - Inexperienced (0.040) (1.086) (1.169) (1.250) (0.876)
FRC Issuers
Experienced Fixed PIPE 0.008 -0.881 -0.265 0.467 -0.055 0.08
Issuers - Inexperienced (0.038) (1.016) (1.094) (1.169) (0.820)
Fixed PIPE Issuers
Experienced SEO 0.016 0.258 0.061 -0.675* 0.130 0.16
Issuers - Inexperienced (0.126) (0.339) (0.365) (0.390) (0.573)
SEO Issuers
Experienced Nonissuers -0.030 -0.765 1.207 -1.553 0.996 0.15
- Inexperienced (0.041) (1.106) (1.191) (1.273) (0.893)
Nonissuers
88
The regression of all FRC issuers shows a negative monthly alpha of -2.48 percent, which is
economically a large amount, but weak statistically at only the ten percent level. The weaker
significance contrasts with Hillion and Vermaelen (2004)'s stronger statistical finding of negative future abnormal
returns. Two differences between this study and theirs are the sample periods and the methodologies. First, Hillion
and Veramaelen (2004) use FRCs issued during the 1990s, and I use FRCs issued over 2001-2002. As the market had
more experience with FRCs, the firms may have become more careful with the terms to which they agreed, thus
reducing the likelihood that their stocks would enter "death spirals." Second, Hillion and Vermaelen (2004) use buy
and hold abnormal returns instead of computing calendar time alphas. This difference may be symptomatic of the
problems with the buy and hold return approach cited by Barber and Lyon (1997, 1999).
A portfolio with a long position in FRC issuers with CFOs who have prior experience issuing stock
publically and a short position in FRC issuers without CFOs with this experience show a more significant statistical
result. This "experienced minus inexperienced" FRC portfolio has a positive monthly alpha of 9.79 percent at the five
percent level of significance (Mitchell and Stafford t- statistic of 2.53). This striking difference among FRC issuing
firms provides support to the FCD hypothesis. The significantly greater returns FRC issuers with experienced CFOs
does not indicate these firms are struggling as the FLR hypothesis predicts. Rather, the non-negative returns to these
firms support the OSD hypothesis, that the FRC terms are appropriate for the firm and are optimal for existing
shareholders.
The same analysis performed on the three groups of control firms provides robust support to the
above findings. The portfolios long in FRC issues and short in either the fixed PIPE or SEO issuers both show
insignificant alphas, which does not support the FCD and FLR hypotheses, and does support the OSD hypothesis.
The portfolio long in FRC issuers and short in matched non-issuers, however, does show a negative alpha close to the
five percent significance level, which does not support the FLR hypothesis because the firms that do not obtain
additional equity financing perform better overall.
Portfolios long in experienced and short in inexperienced CFOs per control group display no
significant alphas. Therefore, the significant alpha of the FRC experienced minus inexperienced portfolio is all the
more compelling.
VI. Conclusion
This research adds to the body of knowledge on the motives for FRC issues by characterizing the types of issuers at
the managerial level. FRC issuing firms that have CFOs without prior experience
89
in making a public offering of equity have significantly poorer stock returns than FRC issuing firms
with CFOs who have prior experience making public offerings. The difference in stock returns
suggests that the experience of the CFO indicates which FRC deals will not maximize shareholder value. While some
firms may be taken advantage of when issuing FRCs, the FRC contract may be a rational security that is appropriate
for some firms by resolving the problems posed by classical corporate finance such as the trade-off between the debt
tax shield and bankruptcy costs and the overhang of debt on investment decisions.
I also find an interesting relationship between CFO compensation relative to the CEO and both the
decision to issue a FRC and the market's reaction to issuance announcements. Firms with higher CFO to CEO
compensation ratios are more likely to issue FRCs. Among FRC issue announcements, the market reacts negatively
when CFOs are more highly compensated. This finding merits further study on the relationship between CEO and
CFO with regard to financing decisions.
The OSD hypothesis is introduced to explain FRCs. Further research could formalize it and allow
for testing of how FRCs resolve issues that reduce firm value.
90
Appendix
FRC Issuers with CFOs with prior public issue experience
COMPUTER MOTION INC - GORDON L. ROGERS
GORDON L. ROGERS joined the Company as Vice President / Chief Financial Officer in March
2000. From 1999 to 2000, Mr. Rogers served as Vice President of Finance at ViroLogic, Inc. a medical biotechnology
company. Previously, he spent five years at Nellcor Puritan Bennett, Inc., one of the world's largest medical device
manufacturers, most recently as Controller for Worldwide Field Operations.
CRAY INC - KENNETH L. JOHNSON
Kenneth W. Johnson serves as Vice President - Legal, General Counsel and Secretary and has held those positions
since joining us in September 1997. From September 1997 to December 2001 he also served as our Vice President -
Finance and Chief Financial Officer. Prior to joining us, Mr. Johnson practiced law in Seattle for twenty years with
Stoel Rives LLP and predecessor firms, where his practice emphasized corporate finance. Mr. Johnson received an
A.B. degree from Stanford University and a J.D. degree from Columbia University Law School.
EXELIXIS - JOHN Y. S.ATO
Glen Y. Sato has served as the Company's Chief Financial Officer, Vice President of Legal Affairs and Secretary
since November 1999. From April 1999 to November 1999, Mr. Sato served as Vice President, Legal and General
Counsel for Protein Design Labs, Inc., a biotechnology company, where he previously served as the Associate General
Counsel and Director of Corporate Planning from July 1993 to April 1999. Mr. Sato holds a B.A. from Wesleyan
University and a J.D. and M.B.A. from the University of California, Los Angeles.
KEY3MEDIA GROUP INC - PETER B. KNEPPER
Peter B. Knepper was hired by Ziff-Davis in March 2000 to be our Executive Vice President and
Chief Financial Officer. From 1998 to March 2000, he was a private investor and consultant providing strategic
planning and financial management services. Mr. Knepper was previously Senior Vice President and Chief Financial
Officer of Ticketmaster Group, Inc., a position he held for more than ten years, from 1988 to 1998.
NEKTAR THERAPEUTICS - BRIGID A. MAKES
Brigid A. Makes has served as Vice President of Finance and Administration and Chief Financial Officer since June
1999. Ms. Makes has also served as Assistant Secretary since January 2001. From 1998 until joining Inhale, Ms. Makes
served as Vice President, Chief Financial Officer and Treasurer for Oravax, Inc., a life sciences company. From 1992
to 1998, Ms. Makes served in various management positions for Haemonetics Corporation, a developer of automated
blood processing systems, including, from 1995 to 1998, Vice President Finance, Chief Financial Officer and
Treasurer. Prior to Haemonetics Corporation, Ms. Makes held a number of financial management positions at Lotus
Development Corp. (now International Business Machines) and General Electric Co. Ms. Makes holds a Bachelor of
Commerce degree from McGill University in Finance and
International Business and an MBA from Bentley College.
RENTECH INC - JAMES P. SAMUELS
Mr. James P. Samuels, age 55, has served as Vice President-Finance, Treasurer and Chief Financial Officer of
Rentech since May 1, 1996. He has executive experience in general corporate management, finance, sales and
marketing, information technologies, and consulting for both large companies and
91
development stage businesses. From December 1995 through April 1998, he provided consulting
services in finance and securities law compliance to Telepad Corporation, Herndon, Virginia, a company engaged in
systems solutions for field force computing. From 1991 through August 1995, Mr. Samuels served as chief financial
officer, vice president-finance, treasurer and director of Top Source, Inc., Palm Beach Gardens, Florida, a
development stage company engaged in developing and commercializing state-of-the-art technologies for the
transportation, industrial and petrochemical markets. During that empl oyment, he also served during 1994 and 1995
as president of a subsidiary of Top Source, Inc. From 1989 to 1991, he was vice president and general manager of the
Automotive group of BML Corporation, Mississauga, Ontario, a privately-held company engaged in auto rentals, car
leasing, and automotive insurance. From 1983 through 1989, Mr. Samuels was employed by Purolator Products
Corporation, a large manufacturer and distributor of automotive parts. He was president of the Mississauga, Ontario
branch from 1985 to 1989; a director of marketing from 1984 to 1985; and director of business development and
planning during 1983 for the Rahway, New Jersey filter division headquarters of Purolator Products Corporation. From
1975 to 1983, he was employed by Bendix Automotive Group, Southfield, Michigan, a manufacturer of automotive
filters, electronics and brakes. He served in various capacities, including group director for management consulting
services on the corporate staff, director of market research and planning, manager of financial analysis and planning,
and plant controller at its Fram Autolite division. From 1973 to 1974, he was employed by Bowmar Ali, Inc., Acton,
Massachusetts, in various marketing and financial positions, and in 1974 he was managing director of its division in
Wiesbaden, Germany. He received a Bachelor's degree in Business Administration from Lowell Technological
Institute in 1970, and a Master of Business Administration degree in 1972 from Suffolk University, Boston,
Massachusetts. He completed an executive program in strategic market management through Harvard University in
Switzerland in 1984.
STAR TELECOMMUNICATIONS INC - KELLY D. ENOS
KELLY D. ENOS has served as our Chief Financial Officer since December 1996 and as Treasurer and Assistant
Secretary since April 1997. Prior to that time, Ms. Enos was an independent consultant in the merchant banking
field from February 1996 to November 1996 and a Vice President of Fortune Financial, a merchant banking firm,
from April 1995 to January 1996. Ms. Enos served as a Vice President of Oppenheimer & Co., Inc., an investment
bank, from July 1994 to March 1995 and a Vice President of Sutro & Co., an investment bank, from January 1991 to
June 1994.
TARGETTED GENETICS CORP - TODD E. SIMPSON
Todd E. Simpson has served as vice president, finance and administration, chief financial officer, treasurer and
secretary of Targeted Genetics since October 2001. From January 1996 to October 2001, Mr. Simpson served as vice
president, finance and administration and chief financial officer of Aastrom Biosciences, Inc., a public life science
company focused on the development of cell-based therapeutics. From August 1995 to December 1995, he served as
treasurer of Integra LifeSciences Corporation, a public biotechnology company, which acquired Telios
Pharmaceuticals, Inc. in August 1995. From 1992 until its acquisition by Integra, he served as vice president of
finance and chief financial officer of Telios and in various other finance-related positions. From 1983 to 1992, Mr.
Simpson practiced public accounting with the firm of Ernst & Young LLP. Mr. Simpson is a certified public
accountant. He received his B.S. in accounting and computer science from Oregon State University.
TIVO INC - DAVID COURTNEY
David Courtney was appointed by our Board to serve as a director in May 2002. Mr. Courtney joined TiVo in March
1999 as Vice President and Chief Financial Officer and in March 2000 was named Senior Vice President for Finance
and Administration. Mr. Courtney is currently Chief Financial Officer and Executive Vice President, Worldwide
Operations and Administration, serving in this capacity since October 2001. From May 1995 to July 1998, Mr.
Courtney served as a Managing
92
Director at J.P. Morgan, an investment banking firm, where he was responsible for building and
expanding the firm's high technology investment banking business in the United States. From 1986 to 1995, Mr.
Courtney was a member of the high technology investment banking group at Goldman, Sachs & Co., most recently
serving as Vice President. Mr. Courtney currently serves as a director of KQED Television, a non-profit affiliate of
the Public Broadcasting System in San Francisco, California. Mr. Courtney holds a B.A. degree in Economics from
Dartmouth College and an M.B.A. degree from Stanford University.
VIASOURCE COMMUNICATIONS INC. DOUGLAS J. BETLACH
DOUGLAS J. BETLACH has been our Executive Vice President, Chief Financial Officer, Treasurer and Secretary
since June 1999. Prior to joining Viasource, Mr. Betlach was Vice President, Chief Financial Officer and Treasurer of
Dycom Industries, Inc., a nationwide provider of engineering, construction and maintenance services to
telecommunications operators.
V-ONE CORP - MARGARET E. GRAYSON
MARGARET E. GRAYSON (54) was elected President and CEO in November 2000. She had served as the Company's
Senior Vice President and Chief Financial Officer since May 1999. Ms. Grayson was elected to the Board of
Directors in August 1999. Prior to joining V-ONE Corporation, Ms. Grayson served as Vice President of Finance and
Administration and Chief Financial Officer for SPACEHAB, Inc. (Nasdaq: SPAB) from September 1994 to October
1998. Immediately prior to joining SPAB, Ms. Grayson served as Chief Financial Officer for CD Radio, Inc. in
Washington, D.C., an early entrant in the satellite radio mobile communications market. Previously, Ms. Grayson
served as a senior executive and consultant to high technology start-up companies. Ms. Grayson is on the Board of
Directors of Ronbotics Corporation and the Advisory Board of Celsion Corporation. Ms. Grayson holds an M.B.A.
from the University of South Florida and a B.S. in Accounting from the State University of New York at Buffalo.
FRC Issuers with CFOs without public issue experience
ADEPT TECHNOLOGY - MICHAEL W. OVERBY
Michael W. Overby has served as Adept's Vice President of Finance and Chief Financial Officer since
March 2000. From December 1999 to March 2000, Mr. Overby held the position of Corporate Controller at Adept.
Prior to joining Adept, Mr. Overby was the financial executive for DG Systems, a leading provider of digital
distribution services to the broadcast advertising industry. From 1996 to 1998 he was Corporate Controller and
Director of Information Systems at Inprise Corporation, formerly Borland, a public software company. Mr. Overby
holds a B.S. in Business Administration from California Polytechnic State University.
ALKERMES INC. - JAMES M. FRATES
Mr. Frates has been Vice President, Chief Financial Officer and Treasurer of Alkermes since July 1998. From June
1996 to July 1998, he was employed at Robertson, Stephens & Company, most recently as a Vice President in
Investment Banking. Prior to that time he was employed at Robertson, Stephens & Company and at Morgan Stanley
& Co. In June 1996, he obtained his M.B.A. from Harvard University.
ALLIANCE PHARMACEUTICALS - TIM T. HART
TIM T. HART, C.P.A. Mr. Hart, who is 44, was appointed Vice President in May 1999 and Chief Financial Officer in
August 1998. He joined the Company in 1991 as Controller and has also served as Treasurer since 1994. Prior to
joining Alliance in 1991, he was employed in various financial management positions at Cubic Corporation for over
eight years. He was also employed by Ernst &Whinney in San Diego, California as a C.P.A.
93
ANTEX BIOLOGICS - GREGORY C. ZAKARIAN
GREGORY C. ZAKARIAN, CPA, age 52, has served as Vice President, Finance and Administration, Chief Financial
Officer and Treasurer of the Company since September 1992. He has served as Secretary of the Company since
November 1993, and as Assistant Secretary of the Company from September 1992 until October 1993. Prior to
September 1992, Mr. Zakarian was a partner with an international CPA firm.
APPIANT TECHNOLOGIES - DOUGLAS S. ZORN
DOUGLAS S. ZORN. Mr. Zorn has been our Chairman of the Board, Chief Executive Officer and
President since May 2000. Mr. Zorn served as Executive Vice President, Chief Financial Officer, Secretary and a
Director of the Company since our incorporation in October 1996 until May 2000. Mr. Zorn served as Executive Vice
President, Secretary and Treasurer, and Chief Financial and Operating Officer of BioFactors, Inc. from December
1993 until February 1997 and as a Director from June 1994 until February 1997.
CECO ENVIRONMENTAL CORP - MARSHALL J. MORRIS
Marshall J. Morris became the Chief Financial Officer of the Company on January 26, 2000. From
1996 to 1999 Mr. Morris was Treasurer of Calgon Carbon Corporation which stock trades on the New York Stock
Exchange and which is a worldwide producer of specialty chemicals and supplier of pollution control technologies
and services with annual sales of approximately $300 million. From 1995 to 1996 he served as a consultant with
respect to business management and strategic planning. From 1989 through 1995 Mr. Morris also served as the
Treasurer of Trico Products Corporation, an international manufacturer and distributor of original equipment
automative parts with annual sales of approximately $350 million.
CEL-SCI CORP - GEERT KERSTEN
Geert R. Kersten, Esq. Mr. Kersten was Director of Corporate and Investment Relations for the Company between
February 1987 and October 1987. In October of 1987, he was appointed Vice President of Operations. In December
1988, Mr. Kersten was appointed Director of the Company. Mr. Kersten also became the Company's Treasurer in
1989. In May 1992, Mr. Kersten was appointed Chief Operating Officer and in February 1995, Mr. Kersten became
the Company's Chief Executive Officer. In previous years, Mr. Kersten worked as a financial analyst with Source
Capital, Ltd., an investment advising firm in McLean, Virginia. Mr. Kersten is a stepson of Maximilian de Clara, who
is the President and a Director of the Company. Mr. Kersten at tended George Washington University in
Washington, D.C. where he earned a B.A. in Accounting and an M.B.A. with emphasis on International Finance. He
also attended law school at American University in Washington, D.C. where he received a Juris Doctor degree.
CHAMPION ENTERPRISES - ANTHONY C. CLEBURG
In 2000 Mr. Cleberg joined Champion from Washington Group International ("Washington Group"),
a publicly-held engineering and construction firm, where for the previous three years he was the Executive Vice
President and Chief Financial Officer. On May 14, 2001, subsequent to Mr. Cleberg's departure from Washington
Group, it filed a voluntary petition for bankruptcy under Chapter 11 of the U.S. Bankruptcy Code. On January 25,
2002, Washington Group completed its Plan of Reorganization and emerged from Chapter 11 bankruptcy protection.
Previous to Washington Group, Mr. Cleberg worked for Honeywell Inc. for 15 years in various senior financial
positions, leaving as
Corporate Vice President, Business Development.
CHELL GROUP - DON PAGNUTTI
Don Pagnutti was appointed our Vice President, Finance on September 19, 2000. Mr. Pagnutti has been our Chief
Financial Officer since September 1998, and was our Executive Vice President and Chief Operating Officer from
September 1997 to September 2000. From 1996 to 1997, he worked for Sullivan Entertainment Inc., as Executive Vice
President and Chief Financial Officer. From 1980 to
94
1996, he worked for Telemedia Communications Ltd., a large Canadian media company as Vice
President, Radio. Mr. Pagnutti is a Chartered Accountant and has a Masters Degree in Business Administration and a
Bachelor of Commerce Degree from the University of Toronto.
CLEAN HARBORS INC - ROGER A. KOENECKE
Roger A. Koenecke joined the Company as Senior Vice President and Chief Financial Officer in 1998.
From 1982 through 1997, Mr. Koenecke held a variety of management positions, including Senior Vice President and
Chief Financial Officer, with Millbrook Distribution Services, Inc. and its predecessor corporations, which are
engaged in the distribution of health and beauty care, general merchandise, and specialty food products. Prior to that,
he was an Audit Manager with Price Waterhouse & Co., an international accounting firm. Mr. Koenecke holds a BS
in Chemistry and MBA from the University of Wisconsin.
CYBERCARE - PAUL PERSHES
PAUL C. PERSHES (age 57) Class II, has served as a director since August 1996 and as our president since May
1997. In March 2001, Mr. Pershes assumed responsibilities as acting chief financial officer. Before joining us, Mr.
Pershes founded and served as an officer of Weinberg, Pershes & Company, P.A., an accounting firm, from July
1994 to May 1997. Before founding Weinberg, Pershes & Company, Mr. Pershes was a senior partner of the
international accounting firm Laventhol and Horvath for 18 years.
DALEEN TECHNOLOGIES - STEVEN M. WAGMAN
STEPHEN M. WAGMAN, 40, has served as chief financial officer of Daleen since June 2000 and has
served as an executive vice president of corporate development and secretary since June 1999. Mr. Wagman has over
12 years of finance, business and legal experience with high-growth software companies. Before joining Daleen, Mr.
Wagman served in various capacities with PowerCerv Corporation, an enterprise resource planning software
company, including Chief financial officer, treasurer, senior vice president of administration, general counsel and
secretary.
DATATEC SYSTEMS INC - ISAAC J. GAON
Isaac J. Gaon, Chairman of the Board since December 1997 and Director since 1992, has served as the Chief
Executive Officer since October 1994. He served as Chief Financial Officer from April 1992 until October 1994.
From September 1987 to December 1991, Mr. Gaon, a chartered accountant, served as President and Chief
Executive Officer of Toronto-based NRG, Inc., (a subsidiary of Gestetner International) an office equipment
supplier, and in several senior management roles within Gestetner Canada and Gestetner USA.
95
DIGITAL RECORDERS - LAWRENCE A. TAYLOR
Lawrence A. Taylor, age 55, has 12 years' experience in the transit industry, as well as extensive
knowledge and experience in auditing, merger and acquisition reporting, analysis and financial information-
technology systems. He has been the Company's secretary, chief financial officer and vice president since May 1998.
From March 1997 to June 1999, Mr. Taylor was a partner in the Dallas office of Tatum CFO, LLP, a professional
partnership of career CFOs. From March 1995 to August 1996, he was senior vice president of Precept Business
Products, Inc., a privately held holding company in Dallas that distributed business forms, construction and on-
demand courier services. From May 1991 to December 1994, he was vice president and group controller of Dallas -
based Mark IV Industries' Transportation Products Group, which included nine companies, subsidiaries and
operating units serving transit and transportation markets worldwide. Prior to 1991, he served in various financial
managerial capacities in the food processing, commercial construction and oil field supply industries, as well as other
manufacturing environments. A 1970 graduate of Wayne State University in Detroit, Mich., Mr. Taylor earned a
B.S. degree in Accounting. A Certified Public Accountant, he is a member of the Texas Society of CPAs and its
Dallas Chapter, the American Institute of CPAs, and Financial Executive International.
DYNTECK INC - JAMES A LINESCH
Since August 14, 2000, Mr. Linesch has served as the Chief Financial and Chief Accounting Officer,
Executive Vice President and Secretary, and since February 1997 Director, of TekInsight. Previously, Mr. Linesch
was the President, Chief Executive Officer and Chief Financial Officer of CompuMed, a public computer company
involved with computer assisted diagnosis of medical conditions, which he joined in April 1996 as Vice President and
Chief Financial Officer. Mr. Linesch served as a Vice President, Chief Financial Officer of the Company from August
1991 to April 1996. From May 1998 to August 1991, Mr. Linesch served as the Chief Financial Officer of Science
Dynamics Corp., a corporation involved in the development of computer Software. Mr. Linesch holds a CPA
certification in the State of California, where he practiced with Price Waterhouse from 1981 to 1984.
ECHOSTAR COMMUNICATIONS CORP - MICHAEL R. MCDONNEL
Mr. McDonnell joined EchoStar in August 2000 as Chief Financial Officer. Mr. McDonnell is responsible for all
accounting and finance functions of the Company. Prior to joining EchoStar, Mr. McDonnell was a Partner with
PricewaterhouseCoopers LLP, serving on engagements for companies in the technology and information
communications industries.
ELECTRIC CITY CORP - JEFFERY R. MISTARZ
Jeffrey R. Mistarz has been our chief financial officer since January 2000 and our treasurer since October 2000. From
January 1994 until joining us, Mr. Mistarz served as chief financial officer for Nucon Corporation, a privately held
manufacturer of material handling products and systems, responsible for all areas of finance and accounting,
managing capital and shareholder relations. Prior to joining Nucon, Mr. Mistarz was with First Chicago Corporation
(now Bank One Corporation) for 12 years where he held several positions in corporate lending, investment banking
and credit
strategy.
ELECTROGLAS INC - THOMAS E. BRUNTON
Thomas E. Brunton was appointed Vice President — Finance, Chief Financial Officer, Treasurer and Secretary of the
Company in November 2000. Prior to joining the Company, Mr. Brunton was Chief Financial Officer of Centigram
Communications from March 1998 to July 2000. He joined Centigram in March 1991 as Controller and also served as
Treasurer. Prior to his service at Centigram, he had financial management responsibilities at 3Com, Sun Microsystems,
and IBM/ Rolm.
96
EUROTECH LTD - JOHN W. DOWIE
JON W. DOWIE, 54, IS OUR VICE PRESIDENT, TREASURER AND CHIEF FINANCIAL OFFICER under an
employment agreement that expired February 6, 2001. He joined the Company in February 2000 after serving as Vice
President, Finance, and CFO for Research Planning, Inc., from September 1997. Prior to that, he served as Controller
for Automation Research Systems Ltd. from August 1992. He is a Certified Public Accountant and a Certified
Government Financial Manager. He holds a B.S. in Accounting and an MBA from Murray State University, and is a
Doctor of Business Administration candidate in Information Systems, Finance, and Marketing at Mississippi State
University.
GLOBAL TECHNOVATIONS - DAVID NATAN
David Natan - was appointed a director of the Company on April 16, 1998 in order to fill a vacancy. Currently, Mr.
Natan, a CPA, has been Vice President and Chief Financial Officer of the Company since June 1995 and Secretary
from August 1997. Mr. Natan previously served on the Company's Board from June 1995 to January 1997. Mr. Natan
brings nearly 20 years of management and analytical experience to his responsibilities. Prior to joining the Company,
from November 1992 through June 1995, Mr. Natan was Chief Financial Officer of MBf USA, Inc., which is a
Nasdaq listed subsidiary of MBf Holdings Berhad, a multi-national conglomerate. From August 1987 through
October 1992, Mr. Natan was Treasurer and Controller for Jewelmasters, Inc., an AMEX listed company.
HORIZON MEDICAL PRODUCTS INC - JULIE F. LANCASTER
Julie F. Lancaster has served as the Vice President — Finance since January 2001. Ms. Lancaster joined the
Company in 1994 as Assistant Controller and served in that capacity until 1996. From August 1996 through August
2000, Ms. Lancaster served as Controller of the Company. From August 2000 to January 2001, Ms. Lancaster served
as Director of Financial Reporting and Planning for the Company.
HYPERTENSION DIAGNOSTICS - JAMES S. MURPHY
James S. Murphy Mr. Murphy joined us as Vice President of Finance and Chief Financial Officer during May 1996.
In March 2000, his title was changed to Senior Vice President, Finance and Administration and Chief Financial
Officer. Mr. Murphy was Controller of Gaming Corporation of America from December 1992 through November
1995. From 1978 to 1988, he was a tax partner with Fox, McCue and Murphy, a certified public accounting firm
located in Eden Prairie, Minnesota. From 1970 to 1978, Mr. Murphy was employed by Ernst & Ernst (currently
named Ernst &Young LLP) with both audit (six years) and tax (two years) experience. Mr. Murphy is a member of
the American Institute of Certified Public Accountants as well as the Minnesota Society of CPAs. He holds a Bachelor
of Science degree from Saint John's University in Collegeville, Minnesota (1966) and a Master of Business
Administration degree (M.B.A.) from the University of Minnesota (1968).
INTERNATIONAL FIBERCOM INC - GREGORY B. HILL
Mr. Hill served as our Controller from September 1999 to March 2000 and became our Vice President -Finance in
April 2000. From June 1998 until June 1999 he was employed by All Star Telecom, an infrastructure development
subsidiary that we acquired in April 1999, where he served as chief financial officer and controller. From June to
September 1999, he served as Regional Controller of our Infrastructure Development Group. Mr. Hill is a certified
public accountant and served in the Technology Industry Group of Price Waterhouse providing audit, transaction
support, and business advisory services to technology companies from January 1992 through June 1998. He received
his bachelor of science in business administration from California State University Sacramento.
97
MEDWAVE INC - MARK T. BAKKO
MARK T. BAKKO is the Chief Financial Officer of the Company. He has served in this position since February 1996.
From 1984 to 1996, Mr. Bakko was with Deloitte & Touche LLP with his most recent position being a senior manager.
Mr. Bakko has been a Certified Public Accountant since 1985 in the State of Minnesota. Mr. Bakko holds a Masters of
Business Taxation and B.S.B.A. degree in Accounting from the University of Minnesota.
ONE VOICE TECHNOGOLOGIES - RAHOUL SHARAN
Rahoul Sharan holds a Bachelor of Commerce degree from the University of British Columbia and is a member of the
Institute of Chartered Accountants of British Columbia. Mr. Sharan was employed by Coopers & Lybrand (now
Pricewaterhouse Coopers) from 1984 to 1989. Since 1989, Mr. Sharan has been the President and a Director of KJN
Management Ltd., a private company that provides a broad range of administrative, management and financial
services to both private and public companies. Mr. Sharan has been a partner in S & P Group, a company that
specializes in investment financing for venture capital projects and real estate development and construction, since
1988. Mr. Sharan was also a Treasurer and Director.
ONSTREAM MEDIAN CORP - GAIL L. BABITT
GAIL BABITT, CPA. Ms. Babitt joined VDC as Chief Financial Officer in November 2000. From
1999 through October 2000 Ms. Babitt served as Vice President of Finance, North America and Corporate Controller
for TeleComputing ASA. TeleComputing ASA is a leading application service provider. From 1997 to 1999 Ms.
Babitt served as Manager-Transaction Services for Price Waterhouse Coopers LLP. During 1997 Ms. Babitt served
as Director of Finance for ToppTelecom, Inc. Topp Telecom is a prepaid cellular company based in Miami. From
1994 to 1997 Ms. Babitt worked in the audit group with Price Waterhouse Coopers LLP (formerly Price Waterhouse
LLP) and with Ernst & Young LLP from 1992 to 1994. Ms. Babitt has received a MBA from Boston University and
a B.S. from Nova Southeastern University.
P-COM INC - LEIGHTON J. STEPHENSON
Mr. Stephenson has served as Vice President and Chief Financial Officer since September 2000.
From 1993 to 2000 he served as Chief Financial Officer, Treasurer, and Secretary of Vallen Corporation, a Texas
company engaged in manufacturing and distribution of industrial safety products and services.
PENTON MEDIA INC. - JOSEPH A. NECASTRO
Joseph G. NeCastro, 44, Chief Financial Officer and Treasurer of Penton since June 1998. Before
joining Penton, Mr. NeCastro spent five years with Reader's Digest Association, Inc. Mr. NeCastro was Vice
President, Finance for Reader's Digest USA from 1995 until 1998 and Corporate Controller in 1994 and 1995.
RAMP CORP - GARY L. SMITH
Mr. Smith joined the Company as Executive Vice President and Chief Financial Officer in
December of 2000. From 1995 to 2000, Mr. Smith was with Provident Group, a financial advisory firm serving
companies operating in emerging market countries, where he was a principal. Previously, Mr. Smith was an
executive of American Express Bank, the international banking arm of the financial services conglomerate American
Express Corporation (NYSE: AXP), where he held various senior financial positions, most recently as Senior Director
and Commercial Banking Head, London Branch. He holds a BSc degree in Economics from the Wharton School and
an MSc in Accounting and Finance from the London School of Economics.
SCIENCE DYNAMICS CORP - ROBERT O'CONNOR
98
Robert O'Connor came to SciDyn from PricewaterhouseCoopers, L.L.P in Philadelphia, PA, where he
served as a manager of middle market advisory services. Mr. O'Connor brings with him a strong background in
corporate finance, including prior positions as Corporate Controller and Chief Financial Officer at three technology
companies. Mr. O'Connor received his MBA from Rutgers- Graduate School of Management, BS from Kean
University in Union, NJ, and he is a Certified Public Accountant.
STARBASE CORP - DOUGLAS S. NORMAN
Douglas S. Norman founded Starbase in September 1991. In February 2000, Mr. Norman was
appointed to serve as our Chief Financial Officer. From September 1997 to February 2000, Mr. Norman served as
our Chief Accounting Officer. In February 2002, Mr. Norman was elected Secretary. Mr. Norman has served as our
Assistant Secretary since February 1997 and Director of Finance from June 1996 to February 2000. Douglas S.
Norman is the son-in-law of William R. Stow III, a member of the Board or Directors. Mr. Norman holds a B.S. in
Business Administration from California State University and an M.B.A. from Loyola Marymount.
STORAGE COMPUTER CORP - PETER N. HOOD
Peter N. Hood, 60, has been the Company's Chief Financial Officer since May 16, 2000. Mr. Hood was previously
owner and Chief Executive Officer of Phoenix Custom Molders, Inc., a custom manufacturer of plastic parts from
1993 to 2000. Phoenix Custom Molders, Inc. filed Chapter 11 bankruptcy on September 4, 1996 in the U.S.
Bankruptcy Court for the District of New Hampshire. The Chapter 11 bankruptcy was entitled "In re Phoenix Custom
Molders, Inc." and docketed as Bk. 96-12443-MWV. Phoenix Custom Molders, Inc. emerged from Chapter 11
bankruptcy on August 17, 1997. He was also co-founder and Vice President of Phoenix Distributors, Inc., a business
involved in consolidating independent distributors of industrial gas and welding supplies from 1985 to 1993. From
1965 to 1985, he was with the accounting firm of Ernst & Young, becoming a partner in 1976.
He received his business degree from Northeastern University and is a certified public accountant.
TEAM COMMUNICATIONS GROUP - JAY J. SHAPIRO
Jay J. Shapiro became our President, Chief Operating Officer and acting Chief Financial Officer on March 16, 2001.
Mr. Shapiro will assist us in overseeing our corporate, financial and fiduciary activities worldwide. From 1993 to
2000, Mr. Shapiro, a certified public accountant, operated a private accounting and consulting practice specializing in
servicing the television industry. During such period, he served as a temporary corporate officer for several publicly
traded entertainment companies. Mr. Shapiro received his B.B.A. from the University of Wisconsin and a MBA (with
Distinction) in Accounting and Finance from Arizona State University Graduate School of Business Administration.
THINKPATH INC - KELLY HANKINSON
Kelly Hankinson has served as our Chief Financial Officer since May 1999 and as a Director since June 2000. Ms.
Hankinson served as our Controller from February 1994 to May 1999. Ms. Hankinson has a Masters Degree and a
Bachelors Degree from York University.
US PLASTIC LUMBER CORP - MICHAEL D. SCHMIDT
MICHAEL SCHMIDT is Treasurer and Vice President of Finance. Mr. Schmidt joined us in December 1997. Mr.
Schmidt has over 20 years of public and private accounting experience including
ten years in the environmental industry. Prior to joining us, Mr. Schmidt served as Chief
Financial Officer of Republic Environmental Systems, Inc., a publicly traded company and a leading environmental
service provider, headquartered in Blue Bell, Pennsylvania, a position he held for approximately ten years. Mr.
Schmidt has a B.S. degree in Business Administration from Rowan University and is a Certified Public Accountant in
the State of New Jersey.
VELOCITY EXPRESS CORP - MARK E. TIES
99
Mark E. Ties. Mr. Ties joined the Company in April 2000 as its Vice President of Finance. Mr. Ties is
also the Vice President of Finance for Velocity Express. Mr. Ties has more than 13 years of financial experience, of
which eight years have been at the executive level in a number of companies in varied industries. Since 1998 and prior
to joining the Company, Mr. Ties was a Manager and Senior Manager for Ernst & Young LLP in its entrepreneurial
services and mergers and acquisitions departments. From 1994 to 1998 Mr. Ties was the Chief Financial Officer of
Progressive Beauty Enterprises, Inc., a regional distribution company. Prior to 1994 Mr. Ties was the corporate
controller of MEI Salons, Inc. and prior to that he was a senior auditor for Coopers & Libran LLP. Mr. Ties is a
Certified Public Accountant.
WAVERIDER COMMUNICATIONS INC. - T. SCOTT WORTHINGTON
T. Scott Worthington has been a Vice President and the Company's chief financial officer since
January 1998. From 1988 to 1996, he worked at Dell Computer Corporation, in Canada, where he held numerous
positions including CFO of the Canadian subsidiary. From October 1996 to January 1998, he was a financial and
business consultant. Mr. Worthington is a Chartered Accountant.
100
References
Aier, J.K., J. Comprix, M.T. Gunlock, and D. Lee, 2005, The Financial Expertise of CFOs and
Accounting Restatements, Accounting Horizons, 19, 123-135.
Baker, M., L. Litov, J. Wachter, J., 2001, Can Mutual Fund Managers Pick Stocks?: Evidence from
Their Trades Prior to Earnings Announcements, Working Paper, New York University.
Barber, B. and J. Lyon, 1997, Detecting Long-Run Abnormal Stock Returns: The Empirical Power
and Specification of Test Statistics, Journal of Financial Economics 43, 341-372.
Barber, B. J. Lyon, and C. Tsai, 1999, Improved Methods for Tests of Long-Run Abnormal Stock
Returns, Journal of Finance 54, 165-201.
Barclay, M., C. Holderness, and D. Sheehan, 2003, Private Placements and Managerial
Entrenchment, William E. Simon Graduate School of Business, University of Rochester working paper.
Brophy D., Oiumet P., and Sialm C., 2006, Hedge Funds as Investors of Last Resort? Review of
Financial Studies
Chava, S. and A. Purnanandam, 2007, Determinants of the Fixed-to-Floating Rate Debt Structure of
Firms, Journal of Financial Economics 85, 755 - 786.
Chaplinsky, S. and D. Haushalter, 2003, Financing under Extreme Uncertainty: Evidence from
PIPEs, Darden Graduate School of Business Administration, University of Virginia working paper.
Chen, L., N. Jegadeesh, and R. Wermers, 2000, An Examination of Stockholdings and Trades of
Mutual Fund Managers, Journal of Financial and Quantitative Analysis 35, 343-368.
Chemmanur, T. and P. Fulghieri, 1997, Why Include Warrants in New Equity Issues? A Theory of
Unit IPOs, Journal of Financial and Quantitative Analysis 32, 1-24.
Chevalier, J., G. Ellison, 1999, Are Some Mutual Fund Managers Better than Others?, Journal of
Finance 54, 875-899.
101
Dai, N., 2007, Does Investor Identity Matter? An Empirical Examination of Investments by Venture Capital Funds and
Hedge Funds in PIPEs, Journal of Corporate Finance 13, 538-563.
Deja, R. and S. Waba, 1998, Propensity Score Matching Methods for Non-Experimental Studies,
NBER Working Paper, #6829.
Garner, J. and B. Marshall, 2005, Unit IPOs: What the Warrant Characteristics Reveal about the
Issuing Firm, Journal of Business, 78, 1837-1858.
Gomes A. and G. Phillips, 2005, Why Do Firms Issue Private and Public Equity, Convertibles and
Debt?
Gottesman, A. and M. Morey, 2006, Manager Education and Mutual Fund Performance, Journal of
Empirical Finance 13, 145-182.
Harlow, W. and K. Brown, 2007, The Right Answer to the Wrong Question: Identifying Superior
Active Portfolio Management, Journal of Investment Management 4, 1-26.
Hertzel, M., M. Lemmon, J. Linck, and L. Rees, 2002, Long-Run Performance Following Private
Placements of Equity, Journal of Finance 57, 2595-2617.
Hertzel, M. and R. L. Smith, 1993, Market Discounts and Shareholder Gains for Placing Equity
Privately, Journal of Finance 48, 459-485.
Hillion, P., and T. Vermaelen, 2004, Death Spiral Convertibles, Journal of Financial Economics 71,
381-415.
Hoitash, U. and R. Hoitash, K. Johnstone, 2009, Internal Control Material Weaknesses, CFO
Compensation, and the Moderating Effects of Corporate Governance, External Monitoring, and Firm Risk, Working
Paper SSRN: http://ssrn.com/abstract=1019359.
Holmstrom, B., 1979, Moral Hazard and Observability, Bell Journal of Economics 10, 74-91.
Krishnamurthy, S., P. Spindt, V. Subramaniam, and T. Woidke, 2004, Does Investor Identity Matter
in Equity Issues? Evidence from Private Placements. Journal of Financial Intermediation, forthcoming.
102
Lee, M., P. Lee, and S. Taylor, 2000, Unit Initial Public Offerings: Staged Equity or Signaling
Mechanism? Working Paper.
Leuz, C., A. Triantis, and T. Wang, 2006, Why Do Firms Go Dark? Causes and Economic
Consequences of Voluntary SEC Deregistrations, Working Paper No. RHS-06-045.
Mian, S., 2001, On the Choice and Replacement of Chief Financial Officers, Journal of Financial
Economics 60, 143-175.
Myers, S. and N. Majluf, 1984, Corporate Financing and Investment Decisions When Firms Have
Information that Investors Do not Have, Journal of Financial Economics 13, 187-222.
Ohlson, J., 1980, Financial Ratios and the Probablistic Prediction of Bankruptcy, Journal of
Accounting Research 18, 109-131.
Rajan, R., 1992, Insiders and Outsiders: The Choice between Informed and Arm's-Length Debt,
Journal of Finance 47, 1367-1400.
Schultz, P. Unit Initial Public Offerings: A Form of Staged Financing, Journal of Financial
Economics, 34, 199-229.
Wermers, R., 2000, Mutual Fund Performance: An Empirical Decomposition into Stock-Picking
Talent, Style, Transaction Costs, and Expenses, Journal of Finance 55, 1655-1695.
Wruck, K., 1989, Equity Ownership Concentration and Firm Value: Evidence from Private Equity
Financings, Journal of Financial Economics
103
doc_695043436.docx
A floating interest rate, also known as a variable or adjustable rate, refers to any type of debt instrument, such as a loan, bond, mortgage, or credit, that does not have a fixed rate of interest over the life of the instrument.
CASE STUDY ON RECENT INNOVATIONS IN
FINANCE: MICROFINANCING AND FLOATING
RATE CONVERTIBLES
ABSTRACT
The first essay provides theory concerning the risk-taking incentives of microfinance
borrowers in varying cases: individual liability, group liability without social sanctions, and group
liability with social sanctions. The results provide insight into how a community's social capital and a
country's credit rights interact to induce recipients of microfinance programs to take risk. Consistent with
recent anecdotal evidence that suggests a "dark side" to microfinance, the results show that communal ties
among joint liability borrowing groups may not lead to higher repayment rates and may have worse
welfare effects on the recipients by making the poorest group members unwilling to take the risks
necessary to grow a business.
The second essay considers floating rate convertibles (FRCs). FRCs are a category of
PIPE securities that receive negative associations in both the academic and professional literature. This
study sheds light on the managerial relationship to the decision to issue FRCs and to the variation in
market response to these issues. One main result of the study identifies influence of the CFO relative to
the CEO as significant in the decision to issue FRCs and in the market's immediate reaction to the
issuance. Another main result is that FRC issuing firms with CFOs without prior public equity
issuance experience have significantly negative long run abnormal returns, whereas FRC issuing firms
with experienced CFOs do not
Table of Contents
Dedication..................................................... ii
Acknowledgements................................................ iii Table of
Contents................................................ iv
Essay 1: A Theory on the Risk-Taking Behavior of Microfinance Borrowers, Creditor
Rights, and Social Capital..................................................... 1
I. Introduction..............................2 II. Literature
Review...........................4 III. Individual
Lending.............................8 IV. Joint Liability
Lending........................11 V. Welfare Comparisons and
General Equilibrium Extension.........25 VI. Empirical
Hypotheses..........................28 VII.
Conclusion..............................28
Appendix.................................29
References...............................58
Essay 2: Issuances of Floating Rate Convertible Securities and Financial Manager
Characteristics..................................61
I. Introduction................................................... 62 II.
Literature..............................63 III. Data
Sample..............................69 IV.
Hypotheses...............................81 V.
Results...............................84 VI.
Conclusion...............................89
Appendix.................................91
References................................101
iv
Essay 1: A Theory on the Risk-Taking Behavior of Microfinance Borrowers, Creditor Rights,
and Social Capital
Abstract:
This paper provides theory concerning the risk-taking incentives of microfinance borrowers in varying
cases: individual liability, group liability without social sanctions, and group liability with social sanctions. The results
provide insight into how a community's social capital and a country's credit rights interact to induce recipients of
microfinance programs to take risk. Consistent with recent anecdotal evidence that suggests a "dark side" to microfinance,
the results show that communal ties among joint liability borrowing groups do not lead to higher repayment rates and does
have worse welfare effects on the recipients by making the poorest group members unwilling to take the risks necessary to
grow a business. This paper contributes to the literature on contract design and on financial development and growth,
extending them into the realm of finance for the poorest and least able to access formal financial markets.
1
I. Introduction
Microfinance is the popular economic development program aimed at the world's poorest entrepreneurs
in many developing countries. It normally consists of lending very small amounts, as little as $75, to
invest in self-owned enterprises, in order to provide more opportunities beyond wage labor.
Microfinance is intriguing because it provides financial services to a very large segment of the world's population, who
otherwise would borrow in the informal financial market from friends, family, and moneylenders. Furthermore, prior to its
first implementation by economics professor Mohammed Yunus in Bangladesh in the 1970s, very poor borrowers in
developing countries were not expected to be willing or at least able to repay unsecured business loans made by lenders
from outside their communities.
It is commonly believed that microfinance programs succeed in leading to high repayment rates because
of the strong social ties that exist among their clients. "Social capital" is considered an alternative to physical capital in
the developmental economics literature. While borrowers may not have physical collateral to secure a loan, they do live
in tight knit cultures where the social repercussions from defaulting on a loan could be as costly as losing material
possessions. The first microfinance programs were thought to tap into this social capital by giving loans to groups, where
each member of the group was liable for all the others' share of the loan (Van Bastalaer, 2000).
Because all members of the borrowing group are liable for each other, they have incentives to punish
noncontributing members and to encourage each other to succeed. They also have incentives to screen potential members
into the group as well as verify that each group member exerts effort in the projects so that they can repay their shares of
the loans. Harnessing social capital is particularly important in countries where lenders have little recourse to deal with
borrowers that are able to repay their loans but choose not to do so. This inability is often due to poor institutions like
creditor rights. (La Porta, et.
al., 1998)
Even if borrowers own land, they may not be able to use it as collateral because of laws that restrict the
use of land as collateral or make the ability to seize collateral very costly or impossible. Also, record of title may be
unavailable due to poor record keeping.
Despite the theoretical support for the value of social capital in providing financial services to
developing nations' poor entrepreneurs, studies of microfinance programs have also yielded some puzzling results. One
is the success of microloans made on an individual liability basis.
If individual liability loans have similar repayment rates as joint liability loans, what does this say
about either the need for social capital or the necessity of the joint liability contract in harnessing it? The other puzzle
deals with the problem of loans not being put to productive use. That is, many of microborrowers seem to hold onto their
loans or consume them rather than invest them in a business.
2
This is one finding that does not have much theoretical explanation. Intuitively, the reasons given is that the members of
the group receiving the loan are too afraid of the repercussions from failure from their peers that they prefer to use the
loan to smooth their consumption and payoff the loan with income from pre-existing sources such as a wage earning job
or even a business that they did not invest the loan in.
In this paper, I develop a theoretical model on the relationship between the repayment strategies of
microfinance borrowers and the types of projects they invest their microloans in. Besley and Coate (1995) model how
group lending mitigates the moral hazard problem of borrowers being unable to credibly contract with lenders to repay
their loans even when they are able. Besley and Coate address the question of how the lending schemes affect the
incentives of borrowers to repay their loans, but they assume that the borrowers' projects are exogenously given. This
paper extends their model by endogenizing project selection under various lending schemes. By modeling the choice of
what project a borrower will choose, I provide a theory to explain the relationship between financial contract design, a
country's creditor protections, investment choice, entrepreneurship, and poverty.
I present conditions for microfinance programs resulting in optimal project selection. I show that
certain entrepreneurs who are jointly liable for their loans will not take the necessary business risks that microfinance is
supposed to induce them to take while not improving the repayment rate over individual liability. This result is driven
by strong social ties that exist among borrowing group members, which is usually thought of as a positive effect on social
welfare.
The conclusions of this study contribute to the theory of microborrower behavior in two main ways. The
first theoretical contribution of this study is an extension of Besley and Coate's (1995) comparison of the repayment
behavior of microborrowers under varying assumptions of liability and existence of social sanctions. I show that
borrowers under joint liability without the threat of social sanctions are more likely to repay their loans than when under
the threat of social sanctions. By introducing project and peer selection into the model, I get this result because the threat
of social sanctions stifles risk sharing and encourages free-riding among borrowers with low upside potential to their
project options.
The second theoretical contribution is the identification of people who choose particular types of
contracts where there is an option between individual and joint liability contracts. If there are no social sanctions, then
people with high upside potential to their projects take the joint liability contract because the benefits of risk sharing
outweigh the costs of free-riding. People with low upside project potential, however, take the individual liability contracts
because they do not share any risk under the joint liability contract by selecting safe projects. The introduction of social
sanctions to the model, however, effectively eliminates the use of the joint liability contract by every microborrower.
3
Ultimately, studying this question is important because microfinance institutions (MFIs), governments, non-governmental
organizations (NGOs), and microfinance investors want to know what model of microfinance to follow. Should a MFI
offer joint liability contracts to borrowers who are able to sanction each other if one borrower does not contribute to
repayment? Should a social planner subsidize MFIs
that offer joint liability contracts or those that offer individual liability contracts?
The remainder of the paper is organized as follows. Section two reviews the relevant literature on
microfinance and positions this paper in the existing literature. Section three presents a model of individual liability
lending. Section four presents a model of joint liability lending without and with social sanctions. Section five discusses
the results of the model. Section six concludes with a summary and direction for further research.
II. Literature Review
The literature on group lending argues for social capital to impact the likelihood of repayment through various channels.
The most frequently cited categorization of models explaining how social capital impacts group repayment rates is by
Ghatak and Guinnane (1999). The first is in a superior screening ability of peers over delegated monitors because most
groups select which borrowers can join them (Ghatak and Guinnane 1999). The second is a superior monitoring ability of
peers to control ex ante moral hazard (Stiglitz 1990). The third is a superior auditing ability of peers to limit verification
costs. The fourth is a superior enforcement mechanism through imposition of social sanctions should a borrower default
to control ex post moral hazard (Besley and Coate 1995). As opposed to comparing the effects of each of these four
assumptions on what kind of problem social capital addresses in group lending, I compare the impact of various forms of
social capital on risk taking and repayment of borrowers.
Besley and Coate (1995) provide a model to predict how the group liability aspect of microfinance affects
the repayment behavior of borrowers when repayment of loans is not enforceable. They compare the model's predictions
of repayment rates among three lending systems: individual lending, group lending without social sanctions, and group
lending with social sanctions. The individual lending system describes the traditional arrangement in which the bank
lends to an individual who is solely liable. The group lending system without social sanctions describes an arrangement in
which the bank lends to a group of borrowers who divide the loan among themselves and invest their shares in their own
enterprises, which are independent from one another. In this system, the group as a whole is liable, each member of the
group decides whether or not she contributes her share to the group's repayment, and the group cannot penalize
noncontributing members. The group lending system with social sanctions describes the same arrangement, except the
group can exert peer pressure on the members to contribute to the repayment of the loan.
4
Their main conclusions are twofold. First, the impact of group lending on repayment rates over that of individual lending
are affected by two countervailing forces: risk-sharing and free-riding. On one hand, if one member of the group cannot
contribute her share of the loan because of poor project returns, the other group members can cover her share. Theref ore,
group lending may improve loan repayment rates through sharing risk. On the other hand, a borrower who would have
repaid her loan under individual lending might take advantage of the group's incentives to cover her share of the loan.
This free-riding incentive leads to a coordination failure whereby the group as a whole will default even though certain
individuals would have repaid their own shares if they were individually liable. Without social sanctions, it is unclear as
to which effect dominates. The second conclusion is that the free-rider effect can be lessened if the group is able to use
social sanctions to pressure the members if they stand to lose a significant amount of social capital. They show that if
social sanctions are great enough, then group lending does dominate individual lending in regard to the repayment rate.
An extension of Besley and Coate's model is Che (2002)'s dynamic model of repayment behavior in a
repeated game. Che does not include social sanctions other than exclusion from participation in future borrowing
opportunities. He shows that while the static model has ambiguous results concerning group lending's repayment rates
due to the free-rider problem, the dynamic model shows that group lending weakly dominates individual lending due to
the cost of exclusion from future opportunities.
In addition to extending their model to a dynamic setting, among the recommendations that Besley and
Coate make for further research is to model the effect of group lending on the type of project that each member chooses.
As this paper will show, Besley and Coate's model can be adapted to find interesting implications for selection of project
risk by different lending schemes. In addition, I extend their model to include self selection of borrowers into groups. The
outcome of endogenizing project and group member selection is domination of the risk sharing effect without social
sanctions and suboptimal risk taking with social sanctions.
In addition to Besley and Coate (1995)'s focus on the enforcement problem, other theories focus on the
problem of (i) screening out risky borrowers, (ii) monitoring borrower effort ("ex ante" moral hazard), and (iii) verification
of project outcomes. Ghatak (1999) and Ghatak and Guinnane (1999) provide the most cited model of how joint liability
microloans induce risky borrowers to select other risky borrowers to form a group and safe borrowers select other safe
borrowers. By this assortive matching process, lenders are more able to identify risky borrowers from safe borrowers "by
the company they keep". Other theories such as van Tassell (1999) and Laffont and N'Guessan (2000) also conclude that
borrowers will match with borrowers of similar riskiness. Guttman (2006, 2007), however, offers a model that predicts
the opposite: Borrowers of opposite risk types will match with each other. Guttman makes this prediction because he
assumes that borrowers can make side payments to each other to
5
attract group members. While both risky and safe borrowers prefer to have safe peers, the value of a safe peer to a risky
borrower is greater than to another safe borrower. Intuitively, this greater value comes from a diversification benefit of
matching a risky borrower with a safe one. This later conclusion is also supported by my model as well as experimental
evidence (Gine, Jakiela, Karlan, and Morduch, 2010).
Stiglitz (1990) finds group lending's advantage in improving repayment rates come from the peer
monitoring effect, thus limiting the ex ante moral hazard of how group members will use their loans and exert effort. He
focuses on group members being able to observe the effort each applies to her projects and to write enforceable contracts
among themselves.
Ghatak and Guinnane (1999) present a model where verification of states is costly in the spirit of
Townsend (1979), demonstrating that joint liability contracts induce truth telling by borrowers concerning their projects'
payoffs by delegating the auditing function to group members. Therefore, the lender only has to verify the state when the
entire group defaults, thus reducing the number of cases the bank has to incur auditing costs.
While most research deals with comparing available microfinance contracts' impact on welfare, some
offers new kinds of contracts that are currently not observed. Bubna and Chowdhry (2009), for example, offer a new
institution they call, "microfinance franchising" in which a single MFI offers a lending franchise to local capitalists who
compete with a single moneylender. Their model is currently being experimented in Samoa. Ayi (2007) suggests that
MFIs offer a "microequity" contract whereby the MFI has an equity stake in the microenterprises, similar to venture
capitalists.
A similarly titled contract is also suggested by Pretes (2002). The newest stage in the microfinance
movement is that of "microinsurance," which offers insurance contracts to the same people that
microloans are intended for (Morduch, 2004; McGuinness and Tounytsky, 2006; Leftley and Mapfumo).
While all the theory on microfinance loan performance assumes that social capital is pivotal, the way in
which social capital functions differs among the theories. The empirical research also varies in its conclusions regarding
social capital's effect on loan repayment and on borrower welfare.
How aspects of borrowers' relationship to each other and their culture add to or subtract from social
capital is complicated. Studies on the type of social ties show that certain aspects of relationships among borrowers in a
group differ in their effect on borrower behavior. Hermes and Lensink (2007) survey of empirical studies on social
capital's relationship to loan repayment identifies several characteristics of social capital that actually have negative
effects on loan repayment: family membership in group, distance between members, strong social ties, group
homogeneity, relatedness,
6
sharing within group, and high level of joint liability. They also find that some of these characteristics also have positive
effects on loan repayment in other studies, indicating that the relationship between social capital and loan repayment is
actually ambiguous. Similarly another survey concludes, "The results of available empirical studies are contradictory
with respect to virtually all potential determinants of repayment performance" (Guttman, 2006).
Studies show that the joint liability contract is not the only means of harnessing social capital. De
Aghion and Morduch (1998) find that microfinance borrowers in transition economies who borrow on an individual
liability basis still have incentives to repay their loans because of social stigma over default. Also, an individual liability
loan can be marketed to, purchased by, and collected in groups, thereby lowering transactions costs, having a good
information source, fostering education and training, and increasing individuals' comfort with banks. I n a randomized
controlled field experiment, Karlan and Gine (2010) show that repayment rates of borrowers of a Philippine MFI with
branches that are randomly converted from offering offer joint liability to loans to individual liability loans do not differ
from unconverted branches. They find that social capital plays an important role in influencing repayment after branches
are converted to offering individual liability loans.
Studies on the impact of microfinance programs on repayment and on poverty reduction are not
uniformly supportive. However, they find that the impact of microfinance programs is not uniform across borrowers.
Hulme and Mosley (1996) and Morduch (1998) find microfinance programs are less effective in increasing income
among those below the poverty line. In Hulme and Mosley (1996), the researchers suggest that microfinance programs'
impact decreases with falling income because the borrowers below the poverty line take less risks, invest less in
technology, and use their loans to protect their subsistence. In some cases, the loans lower income among the poorest of the
poor (Khawari 2004). Many microfinance borrowers take loans to reduce variation in consumption and not to increase
expected income.
Pretes (2002) criticizes the use of microfinance for certain borrowers. He cites cases where very poor
borrowers become worse-off because of business failure. They attempt to repay their portion of the loan by borrowing
from moneylenders, selling their household assets or food, or leaving their home to find wage labor. Taking these drastic
steps as opposed to just defaulting as would be the case in a country with developed bankruptcy laws may be due to "the
darker side of collective peer pressure." Social sanctions may be so strong that the borrower may be in fear of becoming
an outcast. Also, since most microfinance borrowers are women in countries where they are under a high degree of male
control, the husbands of the borrowers sometimes take the money away, leaving the wife to struggle to find a way to repay
the loan. So, the informal institutions of the community and the home may not allow for microfinance to succeed in
empowering borrowers to take the appropriate risks as an entrepreneur.
7
Pretes postulates that the high repayment rates of microfinance programs are misleading as to their effectiveness in
reaching the very poor, as "financial benefits disproportionately accrue to the middle
poor and do not reach the very poor."
This paper contributes to the literature on microfinance by combining Besley and Coate's model with
other models of microfinance to consider how the group lending system affects peer and project choices. Allowing for
projects of varying levels of risk is an important assumption due to the empirical evidence that project selection matters.
The model's results provide a theoretical explanation for the finding that group lending can have the opposite effect on
investment decisions than is intended. I show how the nature of social sanctions that are levied by the group play a role in
this result.
III. Individual Lending
Consider an entrepreneur's payoff function when liability is individual. In the first period, t = 0, the
borrower receives a loan of l at an interest rate, i. At t = 1, the borrower chooses whether to invest in a safe or risky
project. At t = 2, the investment returns are realized and the borrower chooses whether to repay the loan. Assume that
without a loan, the entrepreneur would not be able to invest in any project. All loans are for the same amount and require a
repayment of r. r = l(1+i). The project returns a random
variable,u . If the risky project is taken, then the outcome has either a high or low payoff (u
L
oru
H
) with ~
equal probability. If the safe project is taken, then the outcome has a payoff ofu
S
with probability of 1.
u
S
= (u
L
+u
H
)/2. After the payoffs are realized, the entrepreneur has the choice of repaying or not repaying the loan. Partial repayment is not a
possibility. If she repays the loan, she has a net payoff of
u
~
- r.
The assumption of either full repayment or total default is made by Besley and Coate (1995). It is a
realistic assumption because these are very small loans, where the amount to be voluntarily paid can be thought of as
discrete. Alternatively, the consequence to defaulting can be thought of as discrete, where any kind of default disqualifies
the borrower from borrowing in the future. Therefore, she would not have any motivation to make a partial repayment.
Also, the model can be adapted to there being
partial repayment in that the borrower faces a potential penalty for strategically defaulting. The
penalty may include collection of a portion of her project's payoffs when she defaults.
If she does not repay the loan, then she incurs costs of various forms. First, the lender may penalize her
for defaulting through refusing to make future loans or sharing this information with other
potential lenders (a credit bureau). Second, the lender may retrieve a portion of what is owed by
litigation. Third, the borrower may face loss of reputation in her community. Fourth, the borrower
8
may expend resources to hide her profits from the lender and her community. Fifth, the borrower
may inflict guilt on herself for not repaying the loan when she has the means to repay it.
Because I am considering lending in a developing country where institutions that allow for contract
enforcement are lacking, the bank is limited in how much it can penalize the borrower. The costs of default
may differ from those in a developed country. First, the loan officer may not be able to credibly commit to
refusing to make another loan and there may not be a credit bureau if there are other lenders in the
market. Second, bringing the case to court may be too costly to the lender for the amount that would be
recovered, and the laws governing collection of debts in developing countries often favor the debtors over
the creditors. Third, though the borrower is individually liable, she may be concerned with what her
community thinks of her if she does not repay a debt. Because the borrowers under consideration typically
live in tight knit communities, the knowledge and opinions of others concerning one's own affairs could be
quite significant. On the other hand, her community may have greater solidarity with her rather than the
bank, thereby causing it to be understanding of a defaulting borrower's non-repayment. Fourth, the borrower
may very easily hide the amount of her project's payoffs from observation if she lives in a remote village.
Fifth, whether borrowers' average conscience in developing countries differs from other borrowers is
unknown.
Assume that the penalty for default is increasing in her project's payoffs. Assume there is a fixed
cost and a variable cost that is increasing in project payoffs. For simplicity, I assume that penalty
function is an affine function in project payoffs. Let of represent the fixed cost and o e (0,1)
~
represent the variable cost per unit of project payoff. The penalty function is defined as p(u )=of +
~
ou . Assume for now that of =0 so that all of the penalty is linear in project payoffs. Making this
assumption will not change the analysis as long as of is not greater than r. I assume that penalties
increase in project payoffs for the following reasons. First, limitation on obtaining future financing
limits the ability of the successful but defaulting entrepreneur from being able to fully utilize her present
project's payoffs. That is, a portion of the value of project payoffs may be in the ability to invest them in
future projects. If additional external financing is also necessary for this future investment and there are
increasing returns to scale, then the value of these payoffs will be less. Second, the greater the payoff, the
greater is the amount that may be retrieved through litigation. Third, the community would probably think
worse of a defaulting borrower the more able she is of repayment. Fourth, the more project payoffs there
are, the more there is to hide; the more there is to hide, the more costly it is to hide. Fifth, the defaulting
borrower would probably think worse of herself the more she is able to repay.
9
The entrepreneur will repay only when the payoff from doing so is greater than not doing so, which
is when u
~
> r/o. See Proof 1.
Therefore, the individual's payoff function is P * (u
~
) = max(u
~
÷ r, (1 ÷ o )u
~
) from implementing the
optimal strategy. Figure 1 graphs the individual lending optimal strategy payoff function.
P(u
~
)
r/ o
u
~
Figure 1. Individual borrower's net payoff as a function of her project's payoffs
(u
~
), assuming penalty parameter (o) and amount owed (r).
Note that this payoff function is weakly convex in u
~
:
Proposition 1: An individual liability borrower's optimal repayment strategy (s*)
implies the utility (U
I
) of the individual liability loan to be convex in her project's
~
payoffs (u ). See Appendix for proof.
As can be seen in Figure 1, if the entrepreneur were to select between the safe and risky projects,
then she would certainly choose the risky project if u
H
> r/o> u
L
:
Proposition 2: An individual liability borrower's optimal investment decision (p*) is
to take the risky project (R) rather than the safe project (S). See Appendix for proof.
10
However, if u S> r/o, then the lender would prefer that the safe project be chosen, in which case its
expected repayment would be r, rather than the risky project, whose expected repayment is r/2:
Proposition 3: The expected repayment rate of an individual liability loan is 50
percent. See Appendix for proof.
Typically, in developed countries' markets for loans where there are significant information
asymmetries, credit rationing is understood to occur due to the moral hazard at the investment choice
stage induced by setting interest rates high enough to compensate for the prior riskiness of the borrower
(Stiglitz and Weis 1981). Because this agent's action in this moral hazard problem occurs before the project
payoffs are realized, this type of moral hazard is referred to as ex ante moral
hazard. This is demonstrated in Proof 2, where the borrower would choose to invest in the risky project that
has only a 50 percent chance of her repaying the loan instead of the safe project, which the bank would
prefer her to take. The bank is limited in raising r to compensate for this problem because the critical point,
r/o, for the borrower to choose to repay the loan increases in r. However, if o is higher, then the expectation of repayment is
higher because the critical point for the repayment decision, r/o, is lower and the fraction of payoffs recoverable in default is
higher. Therefore, the bank will limit its losses through controlling the quantity of credit supplied rather than in
price.
In undeveloped economies with weak institutions, another type of moral hazard is introduced:
whether or not to default when the borrower is in fact capable of repaying the loan (Besley and Coate 1995).
This other type of moral hazard is referred to as either ex post moral hazard, strategic default, or
unenforceability. Ifo is higher, then the expectation of repayment is higher because the critical point for the repayment
decision, r/o, is lower and the fraction of payoffs recoverable in default is higher. This can be seen in the proof for
Proposition 1.
IV. Joint Liability Lending
Next, consider an entrepreneur's payoff function when the loan is made to a group. I model group lending
in a two-player (like Besley and Coate), three-period (t = {0,1,2}) game in which both borrowers choose
whether or not to contribute to the repayment of their loan. The two borrowers are identified as "Borrower j"
where j = 1 or 2. In the first period, t = 0, the two-member group forms and receives a loan. The loan
agreement stipulates that the group must pay a total of 2r in principal and interest in the last period, t = 2.
(Under individual lending, each borrower would have to pay r.)
At t = 1, each borrower invests her share of the loan in a project that yields a random variable, u
~
j .
She chooses between two investments, safe or risky.
11
At t = 2, each group member decides whether or not to contribute to the repayment of the loan.
Following the assumption of Besley and Coate (1995), the group as a whole can either default on the entire
loan or repay the entire amount. If each member does contribute, then borrower j has a payoff
of u
~
j -r. If one contributes nothing, then the other borrower can either decide to also contribute
nothing, thereby allowing the group to default, or to cover both members' share and repay the entire
loan. If the group defaults, then, the borrowers are penalized by the same amount as under
~
individual liability (p(u
j
)). The penalties are increasing in u
~
j . Therefore, if the group defaults, then
each member gets a payoff of u
~
j (1-o).
IV. A. Joint Liability Lending without Social Sanctions
It has already been recognized that simply making borrowers jointly liable for each other's loan does not
have strictly positive net effects on borrower repayment (Besley and Coate 1995). Furthermore, it has been
shown empirically in the Philippines that joint liability is not the only method by which social capital can
impact the probability of repayment by groups (Karlan, et. al.). Therefore, the first kind of group lending that
I consider is one where the group members are jointly liable for each other but cannot impose any kind of
penalty for non-contribution to the repayment of the loan. I call this type of harnessing of social capital as
joint liability lending without social sanctions. By making this distinction, I can separate out the effects of
joint liability itself from other factors on the enforcement mechanism.
Optimal Strategies at t=2
The two group members have to choose a strategy of either contribute (C) or not contribute (NC).
Let sj e {C,NC} denote the strategy played by borrower j. Each borrower's net payoff depends on the
strategy played by the peer. Denote borrower 1's net payoff as U
J
1
(s,u
~
) ÷ U
J
1
[(s
1
, s
2
),(u
~
1,u
~
2 )] .
Their payoffs from each possible strategy are denoted in Figure 2.
Borrower 2 Strategy (s2)
Borrower 1
Strategy (s1)
C
NC
C
u
~
1 ÷ r , u
2
÷ r
u
~
1 , u
~
2 ÷
2r
NC
u
~
1 ÷ 2r , u
~
2
u
~
1(1 ÷ o ) , u
~
2 (1 ÷ o
)
12
Figure 2. Joint liability without social sanctions net payoffs
U
J
1
(s,u
~
) , U
J
2
(s,u
~
) under the four possible strategy combinations
of both borrowers in a group.
Project returns, u
~
1 and u
~
2 , determine the payoffs of each combination of strategies. Therefore, the
~ ~
Nash equilibria vary
by the realizations of u and u . The optimal strategies, then, proven in
1 2
Lemmas 1.1 - 1.4 in the Appendix. Figure 3 presents these optimal strategies in a table.
u
~
2
u
~
2 < r/o r/ os u
~
2 < 2r/o
u
~
2 > 2r / o
u
~
1
< r/o
(NC, NC) (NC, NC) (NC, C)
u
~
1
r/ os u
~
1
< 2r/o
(NC, NC) (C , C ) (NC, C)
u
~
1
> 2r/o
(C, NC) (C , N C ) (C, NC), (NC, C)
Figure 3. Nash equilibrium strategies (s1*, s2*) for various project under joint liability without social
sanctions.
The impact of joint liability without any social sanctions on bank repayment appears ambiguous. On
one hand, there is a positive effect of joint liability on borrower repayment. Suppose the bank makes two
loans on an individual liability basis to the two members in this group. If one's project payoffs were less
than r/o but the other's were greater than 2r/o, then only one loan would have been repaid. However, under joint liability,
the borrower with the project with the higher payoffs will pay for her peer's loan. The bank then has both
loan repaid under joint liability. This positive effect on the probability of repayment of loans is the risk-
sharing effect.
Consider now a different scenario: Borrower 1 has payoffs greater than r/o but less than 2r/o, and
Borrower 2 has payoffs less than r/o. Under individual liability one loan is repaid. Under joint liability, neither loan is
repaid because Borrower 1 will not cover for Borrower 2. This negative effect on the probability of
repayment of loans is the free-riding effect.
A borrower's expected payoff function under optimal strategies, then is more complicated than under
individual liability because her net payoffs and optimal strategies depend on the payoffs of the other
borrower's project and her strategy. The dominance of the risk sharing effect or the free riding effect
13
depends on the project payoff. If her peer contributes to the repayment of the loan, Borrower 1, in
~
contrast to the individual liability case, keeps more of her project's payoffs in low payoff states (u
1
<
r/o) due to the risk-sharing effect, has the same net payoff in intermediate payoffs (r/os u
~
1 <
2r/o),
~
and keeps all of her project's payoffs in high payoffs (u
1
> 2r/o) due to the free-riding effect.
However, if her peer does not contribute to the repayment of the loan, Borrower 1 keeps the same
amount of the loan for low payoffs (u1< r/o), has lower net payoffs under intermediate payoffs (r/os
u
~
1 < 2r/o) due to the coordination failure induced by free-riding, and also has lower net payoffs
under high payoffs states due to being exploited by free-riding. These payoffs are represented in
Figure 4.
U
J
1
[(s
1
*, s
2
),
(u ,u
~
)] ~
1 2
s
2
= C
r/o
2r/o
s
2
= NC
u
~
1
Figure 4. Optimal strategy net payoff functions of Borrower 1. The solid lines
represent P*1(u1|C), the optimal strategy net payoff if Borrower 2 contributes to repayment. The
dashed line represents P*1(u1|NC), the optimal strategy net payoff if Borrower 2 does not
contribute to repayment.
The uncertainty surrounding what a borrower's payoffs does not only stem from the payoffs of her
peer's project, but also on which equilibrium strategy they play. When both projects have high
levels of payoffs, they play either (C, NC) or (NC, C). This surprising result that one borrower allows the
other to free-ride on her comes from neither borrower credibly being able to commit to the group to default
This leaves a question of which equilibrium strategy will be played when project payoffs are in these
ranges. If the game were moved from static to dynamic or a convention were applied to
14
it, then the Pareto improving strategy may be played more often. For the purposes of a borrower's
prior beliefs about which equilibrium strategy will be played, I will assume that the probability of 0.5
for each strategy when there are two equilibrium strategies.
3
Optimal Strategies at t=1
Each borrower will uses her expectations of her own project's payoffs and her peers at t=1 when she
chooses what kind of project in which to invest. Assume that there are two projects to choose from, as
under the individual liability case. For simplicity, add the assumption that the safe investment's payoffs are
strictly less than 2r/o. Does Borrower 1 choose the risky or the safe investment if Borrower 2 invests in the safe
investment? What does Borrower 1 choose if Borrower 2 invests in the risky investment.
To address the first question, assume that Borrower 2 invests in the safe investment. This means
that Borrower 2's project yields u
S
, where 2r/o> u
S
> r/o. If Borrower 1 invests in the safe project,
then her project's payoff is in the same range, and they would play (C, C). Borrower 1's expected net
payoff, then, is u
S
- r. If Borrower 1 were to invest in the risky project, then the group will default if
her project paysu
L
, or it will repay the loan with her paying all of it if her project paysu
H
. Her
expected payoff is u
S
-ou
L
/2 - r, which is less than u
S
- r, the net payoff from taking the safe project.
She, therefore, prefers to take the safe project:
Proposition 4: Under a joint liability contract, if one borrower invests in the safe
project, then the other borrower will also invest in the safe project (S) rather than the risky
project (R). See Appendix for the proof.
Next, consider what Borrower 1's net payoffs would be under the two investment options if Borrower
2 takes the risky project. There are now five possible net payoffs if she takes the risky project shown in
Figure 5.
3
One possible rule of the game is for each borrower to flip a coin when they come
together at t=2 to determine who plays her strategy first. If both have payoffs greater
than 2r/ , then the loser of the coin flip goes second and therefore covers the entire
loan.
15
u
~
1
,u
~
2
Possibilities
NE Strategies
(NC, C)
~
P
1
*(u
1
|s
2
)
u
1
H
Probability
0.125
u
1
H
,
u
~
2
u
1
H
,u
2
H
(C, NC)
(C, NC)
u
1
H
- 2r
u
1
H
- 2r
0.125
0.25
u
~
1
u
1
H
, u
2
L
u
1
L
,u
2
H (NC, C) u
1
L 0.25
u
1
L
, u
~
2
u
1
L
,u
2
L (NC, NC) u
1
L
(1 -o) 0.25
Figure 5. Optimal strategy net payoffs to Borrower 1 when both borrowers take the risky project.
The expected payoff for Borrower 1 in taking the risky project when Borrower 2 does
the same isu
S
-ou
L
/4 - 3r/4. If Borrower 1 takes the safe project when Borrower 2 takes the risky, then
Borrower 1 has possible payoffs shown in Figure 6 instead. Her expected net payoff from
taking the risky project would beu
S
-ou
L
/2.
u
~
1
u
~
1
,u
~
2
u
1S
,u
~
2
Possibilities
u
1
S
,
u
1
S
,
u
2H
u
2L
NE Strategies
(NC, C)
(NC, NC)
~
P
1
*(u
1
|s
2
)
u
1
S
u
1
S
(1-o)
Probability
0.5
0.5
Figure 6. Optimal strategy net payoffs to Borrower 1 when Borrower 2 takes
the risky project and Borrower 1 takes the safe project.
16
Her payoff from choosing the safe project over the risky project isu
S
-ou
S
/2 - [u
S
-ou
L
/4 - 3r/4] =
3r/4 -ou
L
/4. Comparison of the expected utilities yields the following result:
Proposition 5: Under a joint liability contract, if one borrower invests in the risky project, then the
other borrower will invest in the risky project only if her projects' high state payoff is greater than
3r
.
Otherwise, she will invest in the safe project. See Appendix for the proof.
o
Each borrower, therefore, knows that if she takes the safe project, then the other borrower will take
the safe project, also. If she takes the risky project, then the other borrower will take the safe project only
ifu
H
s 3r/o and will take the risky project only ifu
H
> 3r/o. The borrowers investing in different
projects, however, is not an equilibrium strategy because if Borrower 2 first chooses a risky investment
and Borrower 1 responds by choosing a safe investment becauseu
H
s 3r/o, then Borrower 2 would reverse her
investment decision to the safe investment as shown in Proof 4. Therefore, ifu
H
s 3r/o, then both borrowers choose
to make the safe investments; and ifu
H
> 3r/o, then both
borrowers choose to make the risky investments. In other words, if the mean payoffs of the
investments are higher, then the borrowers will choose risky strategies.
Optimal strategies at t=0
If the payoffs of the possible projects differ between the group members, then borrowers withu
H
s3r/o, who
always prefer the safe project select other borrowers who would prefer the safe project, too. Likewise, the borrowers withu
H
>
3r/o select other borrowers with the same possible high payoff state. Therefore, when there are no social sanctions on
noncontributing group members, there is
assortive matching of borrowers, consistent with Ghatak and Guinnane (1999) and Ghatak (1999):
Corollary 1: Borrowers with high possible project payoffs (uH > 3r
)
select each
a
other to take joint liability loans and invest in risky projects. Other borrowers
(uH < 3r
)
select each other and invest in safe projects. See Appendix for the proof.
a
Optimal group project selection under joint liability without social sanctions is peer dependent. That
is, even if the project opportunity sets between the two borrowers differ, each borrower's strategy is
dependent on whether at least one borrower always prefers to take the safe investment.
Does the lender choose to make this joint liability loan rather than two individual liability loans? If
u
H
s 3r/o, then both borrowers play "safe", which implies a repayment probability of 1. Ifu
H
> 3r/o,
17
then both borrowers play risky, which implies a repayment rate of (2r * 0.75 + 0 * 0.25)/(2r) = 0.75.
These repayment rates are improvements over the repayment rate of making two individually liable
loans: (2r * 0.25 + r * 0.5 + 0 * 0.25)/2r = 0.5 because the individually liable borrower will always
choose the risky investment. Since the lender has a higher probability of being repaid under joint liability,
borrowers benefit from reduced credit rationing and reduced interest rates. This is shown
formally in the proof for the following proposition in the Appendix:
Proposition 6: The expected repayment rate of a joint liability loan is between 75
and 100 percent.
Presuming the same availability and terms of credit, does the borrower choose to accept a joint
liability loan over an individual liability loan? Ifu
H
s 3r/o, then Borrower 1 has an expected payoff of u
S
- r if she has a joint
liability loan because both take the safe projects. This is less than the expected payoff from taking an individual liability
loan. See Proof 2. Ifu
H
> 3r/o, then Borrower 1 will have an expected payoff ofu
S
-ou
L
/4 - 3r/4. This net expected payoff
is also less than that under individual liability. The difference in net expected payoffs of the individual liability
loan over
the joint liability loan is [u
S
-ou
L
/2 - r/2] - [u
S
-ou
L
/4 - 3r/4] = (r -ou
L
)/4 > 0 sinceu
L
< r/o.
Therefore, if equal in terms and accessibility, microentrepreneurs would prefer the individual liability loan.
See Proposition 12 in the Appendix with proof.
The insight that the joint liability contract without social sanctions induces the borrowers to take
safe investments when the payoff possibilities are lower implies that the risk sharing effect does dominate
the free riding effect. This result contradicts Besley and Coate (1995), who argue that the ability of
borrowers to level punishments on one another is a necessary element to guarantee superior repayment
rates for joint liability loans. I find this different result because I relax their assumption of project choice
being exogenous.
IV. B. Joint Lending with Social Sanctions
Now, I allow group members to penalize each other if one does not contribute to the repayment of the
loan. Following Besley and Coate (1995), social sanctions are a function of payoffs of the
noncontributing borrower, and not on observation of effort or project selection. Social sanctions can be in
the form of loss of reputation in the community or inability to participate in future loans. "Social sanctions"
can also be seen as the group member's internal sense of obligation or guilt or for not contributing to the
repayment of the loan even if the community completely forgives her. With an outside, impersonal
institution, she may not have such guilt for defaulting. Social sanctions' sensitivity to realized returns are
assumed to be more punitive than the bank penalty functions; this
18
belief is the basis for arguments that the cultural realities of these borrowers can induce them to
repay their loans better than what a lender outside the community has at its disposal. The structure of the
social sanctions function, s(uj), is similar to that of the bank penalty function:
s(u
~
j ) = s
f +
o
u
~
j , where sf> af and
|
s|s 1. 1/| represents the increased degree by which the
contributing borrower can penalize the noncontributing borrower over that of the bank. A larger|
implies relatively greater leniency by the group.
Optimal Strategies at t=2
A group member's payoffs decrease by s(u
~
j ) only when the other group member repays the loan and
the member being sanctioned does not contribute. If the loan is not repaid (i.e., neither borrower
contributes), then they do not sanction each other, but they are both penalized by the bank. Continue to
assume that af = 0, and also assume for now that sf = 0. The borrowers' payoffs to each
pair strategies are given in Figure 7.
Borrower 2 Strategy (s2)
Borrower 1
C
C NC
u
~
-r, u
~
-r
1 2
u
~
-2r, u
~
(1-o/|)
Strategy (s1)
NC
u
~
1
(1-o/|),
u
~
2 -
2r
1
u
~
1
2
(1-o),
u
~
2 (1-
o)
~
Figure 7. Joint liability with social sanctions net payoffs (U
JS
1
(s,u ) ,
~
P2(u
2
|s1)) under the four possible strategy combinations of both
borrowers in a group.
Project returns, u
~
1 and u
~
2 , determine the payoffs of each combination of strategies as in the
previous considered case. Therefore, the Nash equilibria vary by the realizations of u
~
1 and u
~
2 .
The
optimal strategies are in Lemmas 3.1 - 3.4 in the Appendix with proofs and summarized in a table in
Figure 8. The optimal strategies when borrowers can sanction each other are less likely to be
dominated by free-riding. When both borrowers have payoffs in excess of|r/o, then the entire group
repays the loan. Furthermore, since|s 1, the lender is more likely to be repaid than in the
individual liability case, where the critical point for individual repayment is r/o>|r/o. However,
joint liability with social sanctions still suffers from the problem of a borrower with medium level
payoffs not contributing her portion when her peer's project has very low payoffs.
19
u
~
2
u
~
2 <|r/o |r/os u
~
2
< 2r/o u
~
2 > 2r / o
u
~
1
<|r/o
(NC, NC) (NC, NC) (NC, C)
u
~
1
|r/ os u
~
1
< 2r/o
(NC, NC) (C, C) (C, C)
u
~
1 > 2r/o (C, NC) (C, C) (C, C)
Figure 8. Nash equilibrium strategies (s1*, s2*) for various project payoffs under joint liability with
social sanctions.
Figure 9 graphs the payoffs of these optimal strategies. The new payoff functions with social
sanctions differ from those without. The joint liability loan without social sanctions does have
payoffs that exceed that of the social sanctions case and the individual liability case due to the reduction
in the free-riding effect. Borrowers are more likely to both contribute to the repayment of the loan for high
realizations of project payoffs. However, there are no net payoffs under joint liability with social sanctions
that exceed that of the individual liability case.
P*
1
(u
~
1 |s
2
)
s
2
= C
|r/o
r/o
2r/o
s
2
= NC
u
~
1
Figure 9. Optimal strategy net payoff functions of Borrower 1 under joint
liability with social sanctions. The solid line represents P*1(u
~
1 |C), the optimal
strategy net payoff if Borrower 2 contributes to repayment. The dashed line
represents P*1(u~
1
|NC), the optimal strategy net payoff if Borrower 2 does not
contribute to repayment.
20
Optimal Strategies at t=1
As in the other joint liability case, each borrower will use her expectations of her own project's payoffs and
her peers' at t =1 when choosing what kind of project in which to invest. Assume that there are two projects
to choose from as before and that that the safe investment's payoffs are strictly less than 2r/o. Does Borrower
1 choose the risky or the safe investment if Borrower 2 invests in the safe investment? What does Borrower 1
choose if Borrower 2 invests in the risky
investment?
First, consider what possible net payoffs Borrower 1 faces if Borrower 2 chooses the safe investment.
If Borrower 1 chooses the safe investment also, then they would play (C, C), and Borrower 1 has a net
payoff ofu
S
- r for sure. If Borrower 1 chooses the risky investment, then she has a net payoff of
u
H
- r if the project paysu
H
, and she has a net payoff of eitheru
L
- r oru
L
(1-o/|), depending on
whetheru
L
is greater than or less than|r/o, respectively. Ifu
L
>|r/o, then her expected net payoff is
(u
H
- r)/2 + (u
L
- r)/2 =u
S
- r, making her indifferent between taking the safe project and the risky
project. In either case, both she and her peer contribute to repayment. If u
L
<|r/o, then her
expected net payoff is (u
H
- r)/2 +u
L
(1-o/|)/2 =u
S
- ou
L
/(2|) - r/2. In this case, she will choose the
risky investment:
Proposition 7: If borrower 2 invests in the safe project, then borrower 1 will invest
in the risky project. See Appendix for the proof.
Next, consider what possible net payoffs Borrower 1 faces if Borrower 2 chooses the risky
investment. If Borrower 1 also chooses the risky investment, she faces four possible net payoffs. Figure 10
shows the net payoffs for that case. In this case, there are possibilities that Borrower 1 will either depend
on her peer to cover her portion of the loan or she will cover her peer's portion of the loan. The expected net
payoff from taking the risky investment when Borrower 2 also takes the
o (1÷ 1 )u
1L
+ 3r
risky investment is EU
JS
R1 R = u
1S
÷ |
4
. If Borrower 1 takes the safe project, however, then
she has possible net payoffs shown in Figure 11. The expected net payoff from taking the safe
EU
S
JS
1
= u
1S ÷
ou
1
+ r
.
S
investment when Borrower 2 takes the risky investment is
R
2
21
The difference in taking the risky project over taking the safe project when Borrower 2 takes the
EU
JS
R
1
÷ EU
S
JS
1
= 2ou
1
÷ (| + 2)r
,
which is greater than zero only if S
risky project is
R R
4
u
1S
> (| + 2)r . Therefore, Borrower 1 chooses the risky project when Borrower 2 chooses a risky
2o
project if
u
1S
> (| + 2)r
2o
and the safe project otherwise:
Proposition 8:
If
u
1L <
|r
and
o
borrower 2 invests in the risky project, then
borrower 1 will take the risky project only if
u
1S
>
(
2 + | )r
2o
. Otherwise, borrower 1
will take the safe project. See Appendix for the proof.
Project 1 Payoff NE Strategies P
1
*(u
1
|s
2
) Probability
u
1
H
,u
2
H (C, C) u
1
H
- r 0.25
u
1
H
,u
2
u
1
H
,u
2
L (C, NC) u
1
H
- 2r 0.25
u
1
u
1
L
,u
2
H (NC, C) u
1
L
(1-o/|) 0.25
u
1
L
,u
2
u
1
L
,u
2
L (NC, NC) u
1
L
(1 -o) 0.25
Figure 10. Optimal strategy net payoffs to Borrower 1 when both borrowers take the risky
project, there are social sanctions, andu
L
s|r/o.
22
Project 2 Payoff NE Strategies P
1
*(u
1
|s
2
) Probability
u
1
S
, u
2H
(C, C) u
1
S
- r 0.5
u
2
u
1
S
, u
2L
(NC, NC) u
1
S
(1-o) 0.5
Figure 11. Optimal strategy net payoffs to Borrower 1 when Borrower 2 takes the risky
project, Borrower 1 takes the safe project, there are social sanctions, andu
L
s|r/o.
Optimal Strategies at t=0
The following lemmas identify which peers each type of borrower prefers according to her expected
project payoff:
Lemma 4.1: For the borrower who will always choose the risky project, i.e. those
who have project expected payoffs greater than
(| + 2)r
,
it is preferable for her to
2o
find a peer who will choose the safe project. See Appendix for the proof.
Lemma 4.2: For the borrower who will invest in the safe project when her peer
invests in the risky project, it is preferable for her to find a peer with low expected
payoffs, i.e. between
r
and
(| + 2)r
.
See Appendix for the proof.
o 2o
By consequence of these preferences, borrowers with similar project expected payoffs will match
together, as stated in the following proposition:
Proposition 9: Borrowers will match with other borrowers with the same expected
project payoffs where there is a possibility of social sanctions. See Appendix for the proof.
23
As a consequence, there are is separation in investment strategies according to the groups' projects'
expected payoffs:
Proposition 10: The only investment strategies that will be played are P1=P2=R for
groups with high expected project payoffs and P1=P2 for groups with low
expected project payoffs. See Appendix for the proof.
Would we see groups who play opposite investment strategies? Guttman (2006, 2007) show that if
group members can select each other and they can make side payments to each other, then the risky borrowers will
match with safe borrowers. My model is consistent with this finding because groups of borrowers with low expected
payoffs play opposite investment strategies. However, my model does not identify which borrower would invest in
which type of project, though the possibility of side payments could resolve such a question as long as each borrower
has an equal probability of offering the side payment and receiving it.
In the above scenario, the repayment rate is 0.5:
Proposition 11: The expected repayment rate of a joint liability loan with the
possibility of social sanctions is between 50 percent and 75 percent. See Appendix for the proof.
Therefore, the repayment rate of joint liability contracts with social sanctions is greater than that of
the individual liability contract. However, if all borrowers have average project prospects that are sufficiently low, then
the repayment rate is no greater than the individual liability's. See the proof for
Proposition 11. Therefore, we see a case where a joint liability contract might be inferior to an
individual liability contract because both the lender will not be any better off, and the borrower will be worse off.
Note that the conditions for the mixed investment strategy of borrowers, for whomu1Ss (2 +|)r/(2o),
implies that the more stringent the social sanctions (lower|), the less likely it is that the mixed
investment strategy will occur because the minimum threshold for taking the (Risky, Risky) strategy
is lowered. Therefore, though joint liability with social sanctions leads to lower repayment rates
than a joint liability without them, stronger social sanctions do make it less likely that the strategies of (Risky, Safe)
will be played.
24
V. Welfare Comparisons and General Equilibrium Extension
The ultimate question for microfinance institutions, governments, and NGOs is what type of contract
maximizes welfare? The microfinance contract that maximizes societal welfare is one that increases the expected net
payoff to one borrower without decreasing the expected repayment to the lender.4 Recent experience with individual
liability contracts supports my contention below that the
individual liability contract is often superior to the joint liability contract. It has been found in
practice as well as in controlled experiments that MFIs that switched from joint liability to individual liability loans
did not experience a decrease in loan repayments.
Social capital, therefore, has a "dark side" of inhibiting some borrowers from taking more risk, which
does not improve repayment rates because of the enforceability problem that exists in the countries these loans are made
in. These results provide some counter predictions to the usual beliefs that group lending programs improve repayment
rates because the group is able to penalize its members by taking away some of their social capital. Furthermore, these
results give a theoretical explanation for anecdotal evidence of the negative welfare effects of microfinance programs on
very
poor borrowers (Khawari 2004, Hulme and Mosley 1996, and Pretes 2002). These borrowers in
particular seem to pick relatively overly safe projects. The empirical evidence shows it is the very poor for whom
microfinance does not work as well as intended. The results presented here show group lending produces the same
choice by borrowers facing project opportunity sets with a low expected outcome. If borrowers' incomes are positively
correlated with the project opportunity sets available to them, then these results may explain the disparate impact of
microfinance programs. Interestingly, social sanctions may work too well by making borrowers too scared to take on an
optimal amount of risk.
V.A. Only One Contract Is Offered
Inspection of Figures 1, 4, and 9 reveals that for the same loan availability and terms, borrowers' expected net payoffs
are higher under individual liability contracts for all projects than under either joint liability cases.
Proposition 12: In terms of borrower expected utility, the individual liability
contract weakly dominates the joint liability contracts assuming the same
4
An additional criterion for maximum societal welfare is the degree that business activity has a positive externality. If the
microentrepreneurs invest loans by purchasing capital and hiring labor within their community and the risky projects require
greater investment than the safe project, then this externality effect would also suggest that more risk taking by, Ceteris Paribus, would
increase welfare. I show later that if such an externality exists, then my conclusions are even stronger.
25
principal and interest across contracts (i.e., same r) and only one contract is offered.
See Appendix for the proof.
However, I also show that the repayment rate under joint liability without social sanctions is higher
by 25 to 50 percent than the repayment rate under individual liability. This increase in repayment probability is due to
the risk sharing benefit when risky projects are taken or to the incentive to take safe projects when the payoffs to the
risky investment are not very high. Because the lender can expect to be repaid with a higher probability, it will make
loans available to more borrowers by rationing credit less and/or reducing the interest rate.
If borrowers are able to sanction each other, however, there actually may be no effect on the
repayment rate over the individual liability contract if the sanctions are not sufficiently strong and average payoffs are
not sufficiently high. This surprising conclusion is due to the lack of risk-sharing benefit by groups forming whereby
only one borrower takes the risky project. This is worse than if the borrowers both chose to take the risky project
because if only one borrower takes the risky
project and the project paysuL, then there will be no chance of the borrower taking the safe project
being able to cover for her. The costs of being punished by one's peer are what drive this result.
The question has been raised in microfinance circles of whether the joint liability feature of group
lending is what harnesses the social capital that leads to high repayment rates. I also demonstrate how joint liability
contracts can induce greater repayment without peers being able to sanction each
other. The mechanism by which joint liability works to increase loan repayment is not through
inducing peers to punish each other, but rather through borrowers sharing risk or through cooperatively avoiding risk.
If peers can punish each other, however, joint liability does not necessarily work better than individual liability loans.
The dominance of the individual liability contract with the constraint that the lender can only offer
one contract is demonstrated in the proof of the following proposition, which is in the Appendix:
V.B. A Choice of Contract Is Offered
The microfinance industry has developed to the point that microborrowers have options of taking an
individual liability or joint liability contract. In this section, I allow the amount due, r, to vary
across contract types, which is dependent on the types of borrowers who may separate into either type of contract. It
also influences which borrower will take a particular contract type.
First, I compare the individual liability loan to the joint liability loan without the possibility of social
26
sanctions. The following proposition is obtained:
Proposition 14: If given the choice between an individual liability and joint liability
contract with the possibility of social sanctions, borrowers with
u
iH
< 3r
o
choose the
individual liability contract and those with
u
iH
> 3r
o
choose the joint liability contract.
The total principal and interest due on the individual liability contract is 150% of that of
the joint liability contract. See Appendix for the proof.
Proposition 14 implies that borrowers with better prospects (u
iH
>
3r
)
prefer the joint liability contract
o
because the interest rate is lower because the lender knows that peers share risk with them.
Borrowers with lesser prospects (u
iH
<
3r
)
prefer the individual liability contract because the cost
o
imposed by the lender is less than the cost of taking the safe project.
By offering the individual liability contract in addition to the joint liability contract, the borrowers
who otherwise would have invested in the safe projects under joint liability now invest in the individual liability
contract. Therefore, everyone invests in the risky project.
Next, I compare the individual liability loan to the joint liability loan where there is a possibility of
social sanctions. The following lemma and proposition are obtained:
Lemma 5: If given the choice between an individual and joint liability contract
where there will be social sanctions, borrowers for whom p*=(R,R) will choose the
individual liability contract. See Appendix for the proof.
Proposition 15: If given the choice between an individual liability and joint liability
loan where there is a possibility of social sanctions, no one will take the joint liabi lity
contract. See Appendix for the proof.
Therefore, the dominance of the individual liability contract is maintained where there would be
social sanctions under a joint liability contract under the general equilibrium assumptions and assumption A8 that
expected project payoffs are always greater than r/o. If this assumption is relaxed, however, then the lower interest rate
induced by the possibility of higher repayment under
27
joint liability would cause more people to take these loans who otherwise would not borrow at all.
VI. Empirical Hypotheses
The conclusions of this study imply some empirical hypotheses using data on the MFI level. Using
this level of data is newer in microfinance research because these data have only become available within the last
decade. One seminal paper is Cull, Demirguc-Kunt, and Morduch (2009), which studies the financial and customer
demographic ratios of MFIs as reported by www.mixmarket.org. They study such key questions in microfinance as
what is the potential tradeoff between MFI sustainability and reach to the poorest borrowers. Cull, et. al. (2009) find
that contract design substantially impacts MFI profitability, loan repayment, and costs. They find that MFIs that make
individual liability loans experience increases in portfolio quality and profitability when they raise interest rates.
Whereas they compare contract types of individual versus group based lending, the theory presented here predicts
different results among group based lenders based on the social ties of their customers. The difference arises from
varying strengths and types of informal institutions that impact how group members respond to non-paying peers.
The data used by Cull, et. al. (2009) could be augmented with measures of formal and informal
institutions within in each of the MFIs' markets. If these measures can be obtained and MFIs are identified by their mix
of making individual and joint liability contracts, then repayment rates across the three kinds of lending discussed here
could be compared.
This study also has implications for testing using micro level data, which is what the majority of
studies have used. If a measure of the project opportunities available to microborrowers can be collected, then there are
several more testable hypotheses. Microfinance borrowers with higher (lower) possible project payoffs are expected to
be more like likely to find peers with similarly higher (lower) possible project payoffs. Microfinance borrowers with
higher and more varied possible project payoffs are expected to be more likely to match with peers with lower and less
varied possible project payoffs. Joint liability borrowers are expected to switch to riskier projects if they switch to
receiving individual liability loans, especially if they have project opportunities with low upside payoffs.
The empirical testing of this model entails data collection challenges. First, measures of forbearance
by culture would require conducting surveys among every identifiable culture served by the MFIs
being studied. Second, measuring project possibilities directly also would require reliance on
surveying microborrowers.
VII. Conclusion
Microfinance has popularly been touted as a program that succeeds in improving its borrowers'
28
incomes by overcoming the moral hazard problem inherent in individual liability loans in countries with poor
institutions. This paper contributes to theory as to why microfinance does not always work as intended. Furthermore, it
shows that both formal and informal institutions matter, which could lead to certain policy prescriptions according to a
country's institutions.
Given that the whole purpose of microfinance is to promote entrepreneurship, which is inherently a
risky project, and if the borrower's prospects are low, then she will not use the funds to make a business grow. Rather,
she will use it for some other purpose, such as income smoothing. She may even forgo her entrepreneurial pursuits in
order to be certain to pay back the loan: There have been stories of borrowers taking on jobs in cities just to repay their
portions of loans rather than working on their own businesses. If the real need of these individuals is insurance, then
they may benefit more by microinsurance programs instead of microlending ones.
On the one hand, the presence of social sanctions in joint liability contracts may inhibit
entrepreneurial activity among people who would otherwise take business loans as individuals. On the other hand, joint
liability contracts can be made to groups with social capital but low payoff project opportunities would otherwise not
borrow money at all.
There are several testable empirical hypotheses implied by this model that may be tested. First,
MFIs that make individual liability loans should have lower repayment rates than those that make joint liability loans if
the group members are unlikely to punish one another for non-contribution. Second, MFIs that make individual liability
loans should have the same repayment rates as those that make joint liability loans where group members are likely to
punish one another for non- contribution and project payoffs are sufficiently high. To test these first two predictions,
one would regress repayment rate at the MFI level on the composition of individual liability contracts to total
contracts, the average measure of social ties among borrowers' communities, and the average measure of potential
project payoffs to the borrowers. Tests using this regression would be most powerful with a sample of MFIs that serve
identifiably specific types of borrowers by culture and economic status. Third, microfinance borrowers with higher
(lower) possible project payoffs will be
more likely to find peers with similarly higher (lower) possible project payoffs. Fourth,
microfinance borrowers with higher and more varied possible project payoffs will be more likely to match with peers
with lower and less varied possible project payoffs. Fifth, joint liability loan borrowers will be more likely to switch to
riskier projects if they switch to receiving individual liability loans. These third, fourth, and fifth predictions would be
tested using individual borrower data, which would depend more heavily on specialized surveys than the methodology
for testing the first two predictions.
29
Appendix
Assumptions:
1. Ordering of Payoff Possibilities for Borrower i:
0 s u
i L
< u
iS
< u
i
H (A1)
2. Payoff Probabilities Conditional on Investing in the Risky Project:
pr(u
~
i = u
iL
| p
i
= R) = pr(u
~
i = u
iL
| p
i
= R) = 1
2
3. Payoff Probability Conditional on Investing in the Safe Project:
pr(u
~
i = u
iS
| p
i
= S) = 1
4. Equivalence of Expected Payoffs of Both Projects:
(A2)
(A3)
E(u
~
i | p
i
= R)
=
u
i
+ u
i
= u
i
S
L
2
H
(A4)
5. Bank Penalty Parameter Bounds:
0 <o s1
6. Social Sanctions Parameter Bounds:
o s | s1
7. Low State Payoff Bounds:
(A5)
(A6)
0 s u
iL <
|r < r
(A7)
o o
8. Expected Payoff Bounds for Both Projects:
r < u
S
< 2r i
(A8)
o o
9. High State Payoff Bounds:
30
2r < uH
o i
(A9)
Proposition 1: An individual liability borrower's optimal repayment strategy (s*) implies the
~
utility (U
I
) of the individual liability loan to be convex in her project's payoffs (u ).
Proof:
~~
U
I
[s(u ),u ] is the utility of individual liability borrower from playing s e (C, NC) and realizing
outcome u
~
.
U
I
(C, u
~
) = u
~
- r (1)
U
I
(NC, u
~
) = u
~
(1-o) (2)
If
u
~
>
r
o
, then U
I
(C,
u
~
) - U
I
(NC, u
~
) > 0.
Therefore, s*(u ~
>
r
)
= C.
o
(3)
If
u
~
<
r
o
, then U
I
(C,
u
~
) - U
I
(NC, u
~
) < 0.
Therefore, s*(u ~
<
r
)
= NC.
o
(4)
Therefore, individual liability borrower's utility under her optimal repayment strategy is as follows
(combining (1) - (4)):
u
~
(1 ÷ o ) if u
~
< r
U
I
[s *
(u
~
),u
~
] =
o
~
(5)
u ÷ r
if u
~
> r
o
Since 0 < o< 1, (5) is a convex function.
Proposition 2: An individual liability borrower's optimal investment decision (p*) is to take the
risky project (R) rather than the safe project (S).
31
Proof:
E[U
Ip
(s*,u
~
)] is the expected utility of individual liability borrower from take project pe (R, S)
according to playing her optimal repayment strategy (s*).
The expected utilities from investing in the risky (R) and safe (S) project (p):
If p = R:
E[U
IR
(s*,u
~
)] = 1 U
I
(s*,u
L
) + 1 U
I
(s*,u
H
)
2 2
=
1
u
L
(1 ÷ o ) + 1 (u
H
÷ r) (6)
2 2
Recognizing that
1
u
L
+
1
u
H
= u
S
and suppressing the arguments of U
I
, (6) reduces to:
2 2
R
EU
IR
= u
S ÷
ou + r L
2
If p = S:
E[U
IS
(s*,u
~
)] = U
I
(s*,u
S
)
EU
IS
= u
S
÷ r
(7)
(8)
(9)
Subtracting (9) from (7) yields the difference in expected utility between the two investment
strategies:
EU
IR
÷ EU
IS
= r ÷ou
2
Since, by assumption,
L
u
L
< r
o
:
(10)
r ÷o( r )
EU
IR
÷ EU
SI
>
2
o
=
0
(11)
Therefore, the expected utility to the individual liability borrower of investing in the risky project
exceeds that of investing in the safe project, i.e., p* = R.
32
Proposition 3: The expected repayment rate of an individual liability loan is 50 percent.
Proof:
V
I
(s*,u
~
) is the ex post value to the lender from making an individual liability loan conditional on
the project payoffs and the borrower's optimal repayment strategy.
From Proposition 1:
0 if u
~
< r
V
I
(s*,u
~
) =
o
r if u
~
> r
(i)
o
EV
PI
(s*,u
~
) is the expected value to the lender of making an individual liability loan conditional on
the borrower taking investment P.
EV
P
I
*
(s*,u
~
) = EV
RI
(s*,u
~
) by Proposition 2.
EV
P
I
*
(s*,u
~
) = 1 ? 0 + 1 ? r = r
(i i)
2 2 2
Therefore, the expected repayment rate is EV
P
I
*
(s*,u
~
) = 1
=
50%.
r 2
Lemma 1.1: The unique Nash Equilibrium strategy (s*) for both borrowers under a joint liability
contract is to not contribute (NC) to the loan repayment when both projects' payoffs are less than
2r
.
a
Proof:
33
U
J
1
(s,u
~
) is the value to borrower 1 under a joint liability contract from playing s = (s1, s2), where s1
e (C1, NC1) , s2 e (C2, NC2), and u
~
= (u
~
,u
~
) .
U
J
1
[(NC, NC),u
~
] = u
~
1(1 ÷ o )
U
J
1
[(C, NC),u
~
] = u
~
1 ÷ 2r
1 2
(12)
(13)
Subtracting (13) from (12):
U
J
1
[(NC, NC),u
~
] ÷ U
J
1
[(C, NC),u
~
] > 0 since u
~
1 < 2r
.
The same
o
applies to
U
J
2
(s,u
~
) . Therefore, s* = (NC, NC) when u
~
1 < 2r
and
u
~
2 < 2r
.
o o
Lemma 1.2: The unique Nash Equilibrium strategy (s*) under joint liability contract is for the
borrower with the project payoffs greater than or equal to
2r
to
pay for both loans and the borrower
a
with project payoff less than
Proof:
2r
to
not contribute.
a
Let
u
~
1 > 2r
o
and
u
2
< 2r
.
o
From (12) and (13):
U
J
1
[(C, NC),u
~
] ÷ U
J
1
[(NC, NC),u
~
] = ou
~
1 ÷ 2r (14)
(14) is greater than zero since
u
~
1 > 2r
.
o
Therefore, borrower would not deviate from playing C.
Applying (12) and (13) to borrower 2
U
J
2
[(C, NC),u
~
] ÷ U
J
1
[(C,C),u
~
] = r (15)
(15) is greater than zero. Therefore, borrower 2 will not deviate from playing NC.
34
Therefore, s*=(C, NC) when
u
~
1 > 2r
o
and
u
~
2 < 2r
o
, and by the same reasoning, s*=(NC, C) when
u
~
1 < 2r
o
and
u
~
2 > 2r
.
o
Lemma 1.3: The unique Nash Equilibrium strategy (s*) under a joint liability contract is for both
r
borrowers to contribute (C,C) to repayment of the loan if both projects' payoffs are greater than
a
and less than
Proof:
2r
.
a
Let
r s u
~
< 2r
and
r s u
~
< 2r
.
o
1
o o
2
o
U
J
1
[(C,C),u
~
] = u
~
1 ÷ r (16)
U
J
1
[(NC,C),u
~
] = u
~
1 (17)
Subtracting (17) from (16):
U
J
1
[(NC, NC),u
~
] ÷ U
J
1
[(C, NC),u
~
] = ÷r (18)
Since (18) is less than zero, borrower 1 would like to deviate from playing C to NC if borrower 2 does
not respond by deviating. Borrower 1 anticipates how borrower 2 will respond by evaluating the
following:
U
J
2
[(NC, NC),u
~
] ÷ U
J
2
[(NC,C),u
~
] = 2r ÷ ou
~
2 (19)
Because
r s u
~
< 2r
,
the right hand side of (19) is strictly greater than zero: r > 2r ÷ ou
~
> 0 .
o
2
o
2
Therefore, borrower 2 would respond to borrower 1 deviating by also deviating. Because both
borrowers know that deviating from C to NC induces the peer to doing likewise, neither borrower
35
will deviate from playing C if the marginal value to each borrower from the group playing (C,C) over
(NC,NC) is positive:
From (12) and (16):
U
J
1
[(C,C),u
~
] ÷ U
J
1
[(NC, NC),u
~
] = ou
~
÷ r (20)
(20) is weakly greater than zero since
r s u
~
< 2r
.
Therefore, neither borrower will deviate from
o
1
o
C, and s*=(C,C) when
r s u
~
< 2r
and
r s u
~
< 2r
.
o
1
o o
2
o
Lemma 1.4: The only Nash Equilibrium strategies (s*) under a joint liability contract is for only one
borrower to contribute to repayment of the loan for both [(C,NC) or (NC,C)] if both projects' payoffs
are greater than or equal to
Proof:
From (12) and (13):
2r
.
a
U
J
1
[(C, NC),u
~
] ÷ U
J
1
[(NC, NC),u
~
] = ou
~
2 ÷ 2r (21)
Since
u
~
1 > 2r
,
the right hand side of (21) is greater than or equal to zero.
o
Therefore, borrower 1 will
not deviate from playing C.
From (13) and (16):
U
J
2
[(C, NC),u
~
] ÷ U
J
2
[(C, C),u
~
] = r (22)
Since the right hand side of (22) is positive, borrower 2 will not deviate from playing NC.
However, the same logic applies to show that it is also a Nash Equilibrium for borrower 1 to not
contribute and borrower 2 to contribute. Therefore, s* = [(C, NC), (NC,C)].
36
Proposition 4: Under a joint liability contract, if one borrower invests in the safe project, then the
other borrower will also invest in the safe project (S) rather than the risky project (R).
Proof:
E[U
J
11P
2
(s*,u
~
)] is the expected value to borrower 1 from borrower 1 taking project P1 and borrower 2 P
taking project P2 according to both players' optimal contribution strategies (s*) and payoff
~
possibilities (u ).
If P1 = R and P2 = S, then by Lemma 1.1 and Lemma 1.2:
s * (u
1L
,u
2S
) = (NC, NC) (23)
s * (u
1H
,u
2S
) = (C, NC) (24)
Therefore,
EU
J
1
= 1 U
J
1
[s*,(u
1L
,u
2S
)] + 1 U
J
1
[s*,(u
1H
,u
2S
)]
RS
2 2
=
1
u
1L
(1 ÷ o ) + 1 (u
1H
÷ 2r)
2 2
=u ÷ S
1
ou
1
L ÷ r
2
(25)
If P1 = R and P2 = S, then by Lemma 1.3:
s * (u
1S
,u
2S
) = (C,C) (26)
Therefore,
EU
S
JS
1
= U
J
1
[s*,(u
1S
,u
2S
)] = u
1S
÷ r (27)
To compare the marginal value to borrower 1 from taking the safe project over the risky project when
borrower 2 takes the safe project, subtract (25) from (27):
37
EU
S
JS
1
÷ EU
J
1
= u
1S
÷ r ÷ (u
1S
÷
ou
1
÷ r)
=
ou
1
RS
2
L
2
L
(28)
Because the right hand side of (28) is greater than or equal to zero, borrower 1 will invest in the safe
project if borrower 2 invests in the safe project. I.e., P
1
*
= S if P
2
= S .
Proposition 5: Under a joint liability contract, if one borrower invests in the risky project, then the
other borrower will invest in the risky project only if her projects' high state payoff is greater than
3r
.
Otherwise, she will invest in the safe project.
o
Proof:
If P1=P2=R, by Lemmas 1.1, 1.2, and 1.4:
s * (u
1L
,u
2L
) = (NC, NC) (29)
s * (u
1L
,u
2H
) = (NC,C) (30)
s * (u
1H
,u
2L
) = (C, NC) (31)
s * (u
1H
,u
2H
) = [(C, NC), (NC,C)] (32)
Therefore, the expected value to borrower 1 from taking the risky project when borrower 2 does
likewise is
EU
J
1
= 1 U
J
1
[s*,(u
1L
,u
2L
)] + 1 U
J
1
[s*,(u
1L
,u
2H
)] + 1 U
J
1
[s*,(u
1H
,u
2L
)]
RR
4 4 4
+ 1 {1 U
J
1
[(C, NC), (u
H
,u
H
)] + 1 U
J
1
[(NC, C), (u
H
,u
H
)]}
42
1 2
2
1 2
=
1
u
1L
(1 ÷ o )
+
1
u
1L +
1 (u
1H
÷ 2r)
+
1 [1 (u
1H
÷ 2r)
+
1
u
1H
]
4 4 4 42 2
38
= u
1S
÷
ou
1
+ 3r L
4
If P1=S and P2=R, by Lemmas 1.1 and 1.2:
s * (u
1S
,u
2L
) = (NC, NC)
s * (u
1S
,u
2H
) = (NC,C)
(32)
(33)
(34)
Therefore, the expected value to borrower 1 from taking the safe project when borrower 2 takes the
risky project is
EU
S
JR
1
= 1 U
J
1
[s*,(u
1S
,u
2L
)] + 1 U
J
1
[s*,(u
1S
,u
2H
)]
2 2
=
1
u
1S
(1 ÷ o ) +
1
u
1
S
2
= u
1S
÷
ou
1
S
2
2
(35)
To compare the marginal value to borrower 1 from taking the risky project over the safe project when
borrower 2 takes the risky project, subtract (35) from (32):
EU
R
J
1
÷ EU
S
JR
1
= u
1S
÷
ou
1
+ 3r ÷ (u
1S
÷
ou
1
) = a(2u
1
÷ u
1
) ÷ 3r
R
u
1S
=
u
1
+ u
1
L
4 2
S S
4
L
(36)
Using
L
2
H
in (36):
EU
J
1
÷ EU
S
JR
1
= ou
1
÷ 3r H
RR
4
(37)
39
The right hand side of equation (37) is greater than zero if u
1H
> 3r
and
less than zero if u
1H
< 3r
.
Therefore, P1* = R if P2 = R and
u
1H
> 3r
4
4
, and P1* = S if P2 = R and
u
1H
< 3r
4
.
4
Lemma 2: The expected value to one borrower of a joint liability contract when both borrowers
invest in the risky project exceeds the expected value when they both invest in the safe project.
Proof:
Subtract (27) from (32):
EU
J
1
÷ EU
S
JS
1
= r ÷ ou
1
L
RR
4
(38)
Because u
1L
< r , the right hand side of (38) is greater than zero. Therefore, both borrowers prefer
a
to choose the same investment strategies of P1=P2=S or P1=P2=R.
Corollary 1: Borrowers with high possible project payoffs (uH >
3
r
)
select each other to take joint
a
liability loans and invest in risky projects. Other borrowers (uH <
3
r
)
select each other and invest
a
in safe projects.
Proof:
Following Proposition 4 there is only a possibility of P1=P2=S or P1=P2=R because it is not possible for one
borrower to invest in the safe project and the other to invest in the risky project. Lemma 2 shows that both
borrowers would prefer that they both invest in the risky project. From Proposition 5, it is only the borrowers
who have high state payoffs under the risky investment who can commit to investing in the risky project
when her peer does the same. Therefore, borrowers who can commit to not deviating from investing in the
risky project when they both agree to do so would prefer to match with each other. The borrowers who
cannot make such a commitment will be forced to match with other such borrowers. Therefore, the
borrowers with the very high state payoff possibilities will
40
match with each other and invest in risky projects; the borrowers with the moderately high state
payoff possibilities will match with each other and invest in safe projects.
Proposition 6: The expected repayment rate of a joint liability loan is between 75 and 100 percent.
Proof:
Let ¢ e [0,1] be the fraction of borrowers with u
H
> 3r
.
a
EV
P
J
1
P
3
(s*,u
~
) is the expected value per borrower to the lender from making a joint liability loan to a
group conditional on borrowers' investment choices (P1 and P2).
By equation (26):
EV
S
J
S
(s*,u
~
) = 2r / 2 = r (39)
By equations (29) - (32):
EV
R
J
R
(s*,u
~
)
=
1
?
0
+
3
?
2r
=
3r / 2
=
3r
(40)
4 4 2 4
EV
J
is the expected value to the lender per borrower from making a joint liability loan to a group
without knowing what projects are available to them.
EV
J
= ¢EV
S
J
S
+ (1 ÷¢ )EV
R
J
R
= ¢r
+
3 (1 ÷¢ )r
4
= (3 +¢ )r
4
(41)
Therefore, the expected repayment rate is = (3 +¢ )r / r = 3 +¢
,
which is bounded between ¾ and
1 (75% and 100%) because
0 s ¢ s 1.
4 4
41
Lemma 3.1: The unique Nash Equilibrium strategy (s*) for both borrowers under a joint liability
contract with a possibility of social sanctions is to not contribute to the loan repayment when both
projects' payoffs are less than
Proof:
|r
.
o
JS ÷ joint liability contract with a possibility of social sanctions
U
JS
1
[(NC, NC)],u
~
] = u
~
1(1 ÷ o )
U
JS
1
[(C, NC)],u
~
] = u
~
1 ÷ 2r
To compare the value of not contributing over contributing, subtract (43) from (42):
U
JS
1
[(NC, NC)],u
~
] ÷ U
JS
1
[(C, NC)],u
~
] = 2r ÷ ou
~
1
(42)
(43)
(44)
Applying this case, i.e.,
u
~
1
<
|r
,
to equation (44):
o
U
JS
1
[(NC, NC)],u
~
] ÷U
JS
1
[(C, NC)],u
~
] > 2r ÷ |r > 0
The same logic applies to the borrower 2. Therefore, s*=(NC,NC) when
u
1 <
|r
o
and
u
2 <
|r
.
o
Lemma 3.2: The unique Nash Equilibrium strategy (s*) under a joint liability contract and
possibility of social sanctions is for the borrower with project payoffs greater than 2r
to
contribute
o
to the repayment of the loan for both borrowers and for the borrower with project payoffs less than
|r
to not contribute.
o
Proof:
Assume that
u
~
`
1
> 2r
o
and
u
~
`
2
<
|r
o
.
42
To compare the value of contributing over not contributing for borrower 1, subtract (42) from (43):
U
JS
1
[(C, NC)],u
~
] ÷U
JS
1
[(NC, NC)],u
~
] = ou
~
`
1
÷ 2r (45)
Applying this case, i.e.,
u
~
1 > 2r
,
o
to equation (45):
U
JS
1
[(C, NC)],u
~
] ÷ U
JS
1
[(NC, NC)],u
~
> 0 .
Therefore, borrower 1 will not deviate from contributing for both.
U
JS
2
[(C, NC)],u
~
] = u
~
2 (1
÷
o
)
|
U
JS
2
[(C, C)],u
~
] = u
~
2 ÷ r
(46)
(47)
To compare the value of not contributing over contributing for borrower 2, subtract (47) from (46:
U
JS 2
[(C, NC)],u ~] ÷ U
JS
2
[(C,C)],u
~
] = r
÷
ou
~
2
|
(48)
Apply this case, i.e.
u
~
2
<
|r
,
o
to (48):
U
JS
2
[(C, NC)],u
~
] ÷ U
JS
2
[(C,C)],u
~
] > 0 .
Therefore,
borrower 2 will not deviate from not contributing.
So, when
u
~
`
1
>
2
r
o
and
u
~
`
2
<
|r
,
s* = (C, NC) , and vice versa.
o
Lemma 3.3: The unique Nash Equilibrium strategy (s*) under a joint liability contract with
possibility of social sanctions is for both borrowers to contribute to the repayment of the loan if one's
project pays off
2r
or
more and the other pays off |r
or
more.
Proof:
o o
Let
u
~
1 >
2
r
o
and
u
~
2
>
|r
o
.
43
Apply (43) and (47) to borrower 1:
JS1
U
[(C,C),u ~] ÷ U
JS
1
[(NC,C),u
~
] = u
~
÷ r ÷ (u
~
÷ o
) =
ou
~
1 ÷ r
1 1
| |
(49)
Applying this case, i.e.
u
~
1 > 2r
o
to (49):
U
JS
1
[(C,C),u
~
] ÷ U
JS
1
[(NC,C),u
~
] > 0 . Therefore, borrower
1 will not deviate from contributing to repayment.
Apply (49) to borrower 2:
JS 2
U
[(C,C),u ~] ÷ U
JS
2
[(C, NC),u
~
] = u
~
÷ r ÷ (u
~
÷ o
) =
ou
~
2 ÷ r
2 2
| |
(50)
Applying this case, i.e.,
u
~
2
>
|r
o
t o (50):
U
JS
2
[(C,C),u
~
] ÷ U
JS
2
[(C, NC),u
~
] > 0 .
Therefore,
borrower 2 will not deviate from contributing to repayment.
So, when one project pays
2r
or
more and the other project pays |r
or
more, s*=(C,C).
o o
Lemma 3.4: The unique Nash Equilibrium strategy (s*) under a joint liability contract with the
possibility of social sanctions is for neither borrower to contribute to the repayment of the loan if one
|r
project payoffs less than
Proof:
o
and the other pays off less than
2r
.
o
Let
|r < u
~
s 2r
o
1
o
and
u
~
2
<
|r
.
o
Recall (44):
U
JS
1
[(NC, NC)],u
~
] ÷ U
JS
1
[(C, NC)],u
~
] = 2r ÷ ou
~
1
44
Applying
this
case,
i.e.,
|r < u
~
s 2r 1
to
the
above:
o
(2 ÷ | )r > U
JS
1
[(NC, NC)],u
~
] ÷ U
JS
1
[(C, NC)],u
~
] > 0 .
o
Apply
the
above
to
borrower
2
and
this
case,
i.e.
u
~
2
<
|r
:
o
U
JS
2
[(NC, NC)],u
~
] ÷ U
JS
2
[(C, NC)],u
~
] > 0 .
Therefore, neither borrower deviates from this strategy. So, s* = (NC,NC).
Proposition 7: If
in the risky project.
Proof:
u
1L <
|r
o
and borrower 2 invests in the safe project, then borrower 1 will invest
If P1=R and P2=S, then by Lemmas 3.4 and 3.3:
s * (u
1L
,u
2S
) = (NC, NC) (51)
s * (u
1H
,u
2S
) = (C,C) (52)
Therefore, the expected value to borrower 1 from taking the risky project when her peer takes the
safe project is:
EU
JS 1
= 1 [u
1L
(1 ÷ o )] + 1 (u
1H
÷ r )
(53)
RS
2 2
S u
1L
+ u
1
H
Use
u= 1 2
in (53):
EU
JS
S
1
= u
1S
÷
ou
1
+ r L
R
2
(54)
45
If P1= P2=S, then by Lemma 3.3:
s * (u
1S
,u
2S
) = (C,C) (55)
Therefore, the expected value to borrower 1 from taking the safe project when her peer does likewise
is:
EU
S
JS
1
= u
1S
÷ r S
(56)
To compare the marginal value to borrower 1 from taking the risky project over the safe project when
borrower 2 takes the safe project, subtract (56) from (54):
EU
JS
S
1
÷ EU
S
JS
1
= u
1S
÷
ou
1
+ r ÷ (u
1S
÷ r) = r ÷ u
1
R S L
2
2
L
(57)
Apply this case, i.e.,
u
1L <
|r
,
to (57): EU
JS
S
1
÷ EU
S
JS
1
> r(1 ÷ | ) > 0 .
o
R S
2
Therefore, borrower 1 will
invest in the risky project when her peer invests in the safe project and her risky project's low state
payoff is less than
Proposition 8: If
|r
.
o
u
1L <
|r
and
o
borrower 2 invests in the risky project, then borrower 1 will take
the risky project only if
Proof:
u
1S
> (2 + | )r
2o
. Otherwise, borrower 1 will take the safe project.
If P1=P2=R, then by Lemmas 3.1, 3.2, and 3.3:
s * (u
1L
,u
2L
) = (NC, NC) (58)
s * (u
1L
,u
2H
) = (NC,C) (59)
46
s * (u
1H
,u
2L
) = (C, NC) (60)
s * (u
1H
,u
2H
) = (C,C) (61)
Therefore, the expected value to borrower 1 from taking the risky project when her peer does the
same is:
EU
JS
R
1
= 1 U
JS
1
[s*,(u
1L
,u
2L
)] + 1 U
JS
1
[s*,(u
1L
,u
2H
)] + 1 U
JS
1
[s*,(u
1H
,u
2L
)]
R
4 4 4
+ 1 U
JS
1
[s*,(u
H
,u
H
)]
4
1 2
=
1
u
1L
(1 ÷ o ) +
1
u
1L
(1
÷
o
)
+ 1 (u
1H
÷ 2r) + 1 (u
1H
÷ r) (62)
4 4 | 4 4
S u
1L
+ u
1
H
Use u = 1 2
in (62) and reduce:
EU
R
JSR1
= u
1S
÷
o (1 ÷ 1 )u
1L
+ 3r
|
4
(63)
If P1=S and P2=R, then by Lemmas 3.4 and 3.3:
s * (u
1S
,u
2L
) = (NC, NC) (64)
s * (u
1S
,u
2H
) = (C,C) (65)
Therefore, the expected value to borrower 1 from taking the safe project when her peer takes the
risky project is:
EU
S
JS
1
= 1 U
JS
1
[s*,(u
1S
,u
2L
)] + 1 U
JS
1
[s*,(u
1S
,u
2H
)]
R
2 2
47
=
1
u
1S
(1 ÷ o ) + 1 (u
1S
÷ r)
2
= u
1S
÷
ou
1
+ r S
2
2
(66)
To compare the marginal value of borrower 1 taking the risky project over the safe project when her
peer takes the risky project, subtract (66) from (62):
EU
JS
R
1
÷ EU
S
JS
1
= 2ou
1
÷ (| + 2)r S
R R
4
(67)
A necessary and sufficient condition for the right hand side of (67) to be greater than or equal to zero
is determined by setting (67) greater than or equal to zero, which yields:
u
1S
> (| + 2)r
2o
Therefore, if borrower 1's projects' mean expected payoff exceeds
(68)
(| + 2)r
,
then she invests in the
2o
risky project when her peer does also. If her mean expected payoff is between r
and
(| + 2)r
,
o 2o
then she invests in the safe project.
Lemma 4.1: For the borrower who will always choose the risky project, it is preferable for her to
find a peer who will choose the safe project.
Proof:
If
u
1S
> (| + 2)r
2o
, then borrower 1 will invest in the risky project when her peer invests in the risky
project (Proposition 8) or if her peer invests in the safe project (Proposition 7). The relative value of
finding a peer who will invest in the safe risky project when she invests in the risky project over one
who will invest in the safe project is determined by subtracting (54) from (62):
48
o (1 ÷ 1 )u
1L
+ 3r o
| ÷1 u
1L
÷ r
S
|
S
ou
1
+ r =
|
S
EU
JS
R
1
÷ EU
JS
S1
R R
= u
1
÷
4
÷ u
1
÷
2
4
<0
(69)
Since (69) is negative, a borrower with an expected payoff on her projects greater than
values having a peer who will invest in the safe project.
( | + 2) r
2o
Lemma 4.2: For the borrower who will invest in the safe project when her peer invests in the risky
project, it is preferable for her to find a peer with low expected payoffs, i.e. between
( | + 2) r
.
2o
Proof:
r
and
o
Proposition 7 states that if one borrower invests in the safe project, then the other will invest in the risky
project. Proposition 8 states that if the expected project payoffs are low, then a borrower will maintain
investment in the safe project when her peer invests in the risky project. If both borrowers
have low expected project payoffs, then their optimal investment strategy, P* = (R, S) or
P* = (S, R) . There is no a priori reason for either borrower to expect that she will be the one to play
the risky investment strategy. Therefore, both will expect to play either strategy equally.
The expected value for a borrower with low expected payoffs (borrower 1) from having a peer with
similarly low expected payoffs is derived from (54) and (66):
1 EU
JS
1
+ 1 EU
JS
1
= u
S
÷ 2r + o (u
1L
+ u
1S
)
2
RS
2
SR
1
4
(70)
If the peer has high expected payoffs, however, the borrower with low expected payoffs is assured to
always invest in the safe project because she cannot credibly commit to the risky strategy and the high
expected payoff borrower will always invest in the risky project (Proposition 8).
The expected value for a borrower with low expected payoffs (borrower 1) from having a peer with
high expected payoffs is given in (64).
49
The marginal value to the low expected payoff borrower from having a peer with similarly a low
expected payoff project is the difference between (64) and (71):
(1 EU
JS
S
1
+ 1 EU
S
JS
1
) ÷ EU
S
JS
1
= o (u
1
÷ u
1
) > 0
2
R
2
R R S
4
L
(71)
Proposition 9: Borrowers will match with other borrowers with the same expected project payoffs
where there is a possibility of social sanctions.
Proof:
Borrowers with low expected project payoffs prefer to invest in the risky project (Lemma 4.2).
These borrowers cannot credibly commit to invest in the risky project with a peer with high project
payoffs because the peer would always invest in the risky project (Proposition 8).
Therefore, the only chance the low expected project payoffs borrower has to play the risky strategy is
to match with a borrower with low expected project payoffs, too (Proposition 7).
Proposition 10: The only investment strategies that will be played are P1=P2=R for groups with
high expected project payoffs and P1=P2 for groups with low expected project payoffs.
Proof:
This follows from Propositions 7, 8, and 9.
Proposition 11: The expected repayment rate of a joint liability loan with the possibility of social
sanctions is between 50 percent and 75 percent.
Proof:
Let
| e[0,1]
be the fraction of borrowers with
r < u
S
< (| + 2)r
.
o 2o
50
EV
P
J1SP
2
(s*,u
~
) is the expected value per borrower to the lender from making a joint liability loan to a
group conditional on borrowers' investment choices (P1 and P2).
EV
R
JR
S
(s*,u
~
) = (1 ? 0 + 3 ? 2r) / 2 = 3r
(72)
4 4 4
EV
R
JS
S
(s*,u
~
) = EV
S
JR
S
(s*,u
~
) = (1 ? 0 + 1 ? 2r) / 2 = r (73)
2 2 2
The expected value to the lender is the weighted average of (72) and (73) by the distribution of all
borrowers' expected project payoffs:
EV
JS
= (1 ÷ |)EV
R
JR
S
+ |EV
R
JS
S
= (3 ÷ |)r
4
(74)
The expected repayment rate is (74) divided by r and is bounded according to the value of |:
1 s 3 ÷|
s
3
(75)
2 4 4
Proposition 12: In terms of borrower expected utility, the individual liability contract weakly
dominates the joint liability contracts assuming the same principal and interest across contracts (i.e., same r).
Proof:
Case 1: Borrowers with low expected payoffs
From Proposition 2, Corollary 1, and Proposition 10 for borrower 1 with
u
1H
< 3r
:
EU
Ip
* = EU
IR
,
o
EU
Jp
1
= EU
S
JS
1
, and EU
J
pS*
1
e{EU
JS
S
1
, EU
S
JS
1
}.
*
R R
51
The value of the individual liability contract over the joint liability contract without the possibility of
social sanctions is given by subtracting (27) from (7):
EU
Ip
* ÷ EU
Jp
1
= r ÷ou
1
> 0 L
*
2
(76)
The value of the individual liability contract over the joint liability contract with the possibility of
social sanctions is given by subtracting (64) from (7) or (71) from (7):
EU
Ip
* ÷ EU
R
JSS
1
= 0 (77)
EU
Ip
* ÷ EU
S
JS
1
= o (u
1S
÷ u
1L
) > 0 R
Case 2: Borrowers with high expected payoffs
From Proposition 2, Corollary 1, and Proposition 10 for borrower 1 with
(78)
u
1H
> 3r
:
EU
Ip
* = EU
IR
,
o
EU
Jp
1
= EU
R
J
1
, and EU
J
pS*
1
= EU
R
JSR
1
.
*
R
The value of the individual liability contract over the joint liability contract without the possibility of
social sanctions is given by subtracting (32) from (7):
EU
Ip
* ÷ EU
Jp
1* = r ÷ ou
1
> 0 L
4
(79)
The value of the individual liability contract over the joint liability contract with the possibility of
social sanctions is given by subtracting (63) from (7):
r
÷
ou
1
L
EU
Ip
* ÷ EU
J
pS*
1
=
4
|
>
0
(80)
Proposition 14: If given the choice between an individual liability and joint liability contract
without the possibility of social sanctions, borrowers with
u
iH
< 3r
o
choose the individual liability
52
contract and those with
u
iH
> 3r
choose
the joint liability contract.
o
The total principal and interest
due on the individual liability contract is 150% of that of the joint liability contract.
Proof:
Let rI be the amount due for the individual contract and rJ be the amount due for the joint liability contract
without the possibility of social sanctions.
From Propositions 3 and 6:
EV
p
I
*
= r
I
2
EV
p
J
*
= (3 +¢ )r
J
4
(81)
(82)
If the lender were to choose rI relative to rJ such that the expected values to the lender is equal, then
set (81) equal to (82), which yields:
r
I
= (3 +¢ )r
J
2
(83)
The borrower will choose the individual liability loan over the joint liability loan if
(EU
Ip
* | r
I
) ÷ (EU
Jp
1* | r
J
) > 0 . Otherwise, she will choose the joint liability loan.
Case 1: p*=(R,R)
From Propositions 2 and 5:
(EU
Ip
* | r
I
) ÷ (EU
Jp
1* | r
J
) =
u
1S
÷
ou
1
+ r
I
L
÷
u
S
÷
ou
1L
+ 3r
J
Substitute rI with expression (83) in (84):
2
1
4
(84)
(EU
Ip
* | r
I
) ÷ (EU
Jp
1* | r
J
) =
÷
ou
1
+ (3 + 2¢ )r
J
< 0 L
(85)
4
53
Therefore, borrowers for whom p* = (R, R) (those with u
iH
> 3r
)
will always select the joint liability
o
contract.
Case 2: p*=(S,S)
From Propositions 2 and 4:
(EU
Ip
* | r
I
) ÷ (EU
Jp
1* | r
J
) =
u
1S ÷
ou
1
+ r
I
L
÷ u
S
÷r
2
(
1
J
)
(86)
Substitute rI with expression (83) in (86):
(EU
Ip
* | r
I
) ÷ (EU
Jp
1* | r
J
) = (7 +¢ )r
J
÷ 2ou
1
L
4
(87)
Use the assumption that
u
1L
< r
J
o
to show that the right hand side of (87) is greater than zero:
(7 +¢ )r
J
÷ 2ou
1
L > (7 +¢ )r
J
÷ 2r
J
= (5 +¢ )r
J
> 0 (88)
4 4 4
Therefore, borrowers for whom p*=(S, S) (those with
liability contract.
u
iH
< 3r
),
o
will always choose the individual
Since the borrowers will separate in this way, the lender believes that ¢ = 0. Therefore,
substituting in (83):
r
I
* = 3r
J
*
,
where rI* and rJ* are the equilibrium amounts due for both the
2
individual and joint liability contracts.
54
The lender values the joint liability contract at the general equilibrium amount due, rJ*, at (82)
evaluated with
¢ = 0 : EV
p
J
*
= 3r
J
*
.
4
The repayment rate in general equilibrium, therefore, is
EV
p
J
*
= 3r
J
* r = 3 = 75% .
4 4
Comment 1: Borrowers with better prospects (u
i
H
> 3r
)
prefer the joint liability contract because
o
the interest rate is lower because the lender knows that their peers share risk with them. Borrowers
with lesser prospects (u
i
H
< 3r
)
prefer the individual liability contract because the cost imposed by
o
the lender is less than the cost of taking the safe project.
Comment 2: Everyone invests in the risky project.
Lemma 5: If given the choice between an individual and joint liability contract where there will be
social sanctions, a borrowers for whom p*=(R,R) will choose the individual liability contract.
Proof:
Let rJS be the amount due under a joint liability contract with the possibility of social sanctions.
From Propositions 3 and 11:
EV
p
I
*
= r
I
2
EV
p
J*
S
= (3 ÷ |)r
JS
4
(89)
(90)
If the lender were to choose rI relative to rJS such that the expected values to the lender is equal,
then set (89) equal to (90), which yields:
r
I
= (3 ÷ |)r
JS
2
(91)
55
The borrower will choose the individual liability loan over the joint liability loan if
(EU
Ip
*) ÷ (EU
J
pS*
1
) > 0 . Otherwise, she will choose the joint liability loan.
If p*=(R,R), then:
1L
(EU
Ip
*) ÷ (EU
J
pS*
1
) =
u
1S ÷
ou
1
+ r
I
L
1
÷ u
S
÷ o (1
+
|
)
u
1
+ 3r
JS
(92)
2
2
Substitute rI with expression (91) in (92):
(EU
Ip
* | r
I
) ÷ (EU
R
JSS
1
| r
p
*
)
=
ou
1
(1 ÷ | ) + 4||r
JS
> 0 L
4|
Therefore, if p*=(R,R), then the borrowers prefer the individual liability contract.
(93)
Proposition 15: If given the choice between an individual liability and joint liability loan where
there is a possibility of social sanctions, no one will take the joint liability contract.
Proof:
If anyone does take the joint liability contract, it would be the borrowers with lower expected project payoffs
because Lemma 5 shows that the high expected project payoff borrowers will definitely choose the
individual liability contract. These lower expected project payoff borrowers are those who would play (R,S) or
(S,R) investment strategies. Therefore, if the joint liability contract is taken by
at least one group, | = 1.
Compare the expected values from taking the individual liability loan to the joint liability loan with
the possibility of social sanctions by subtracting (70) from (10) and substituting rI with expression
(91):
EU
Ip
* ÷
1
EU
JS
S
1
+ 1
EU
JS
1
= o (u
1
÷ u
1
) ÷ (1 ÷ |)r
JS
2
R
2
S
R
S
L
4
(
9
4
)
56
As explained above, if this contract is accepted by anyone, it is by the lower expected payoff
borrowers. Therefore, evaluate (94) with | = 1:
EU
Ip
* ÷ 1 EU
R
JSS
1
+ 1 EU
S
JS
1
=
o (u
1
÷ u
1
) > 0
2
2
R
S
4
L
(95)
Therefore, the borrowers with lower expected project payoffs would choose the individual liability
loan.
57
References
Ahlin, Christian and Townsend, Robert M., "Using Repayment Data to Test Across Models of Joint
Liability Lending" . Economic Journal, Vol. 117, No. 517, pp. F11-F51, February 2007 Available at
SSRN: http://ssrn.com/abstract=978564 or DOI: 10.1111/j.1468-0297.2007.02014.x
Armendariz de Aghion, Beatriz and Jonathan Morduch. 2000. "Microfinance Beyond Group Lending,"
Economics of Transition, Vol. 8 (2), 401-420.
Ayayi, Ayi. 2009. "Microfinance: Debt or Equity? What are the Implications for Social Welfare?" Global
Economy and Finance Journal, Vol. 2 (2), 64-80.
Besley, Timothy and Stephen Coate. 1995. "Group Lending, Repayment Incentives, and Social
Collateral," Journal of Developmental Economics, 46, 1-18.
Bubna, Amit and Bhagwan Chowdhry. 2009. "Franchising Microfinance," Review of Finance, Vol. 14
(3), 451-476.
Casssar, Alessandra, Luke Crowley, and Bruce Wydick. 2007. "The Effect of Social Capital on Group
Loan Repayment: Evidence from Field Experiments," The Economic Journal, Vol. 117 (February), F85- F106.
Che, Yeon-Koo. 2002. "Joint Liability and Peer Monitoring Under Group Lending," Contributions to
Theoretical Economics, 2, Article 3.
Conning, Jonathan. 2005. "Monitoring by Delegates or by Peers? Joint Liability Loans under Moral
Hazard," (June 2005). Unpublished working paper.
Cull, Robert, Asli Demirguc-Kunt, and Jonathan Morduch. 2006. "Financial Performance and Outreach:
A Global Analysis of Lending Microbanks," World Bank Policy Research Working Paper 3827, February 2006.
Cull, Robert, Asli Demirguc-Kunt, and Jonathan Morduch. 2009. "Microfinance Meets the Market,"
Journal of Economic Perspectives, Vol 23 (1), 167-192.
De Mel, Suresh, David McKenzie, and Christopher Woodruff. 2008. "Returns to Capital in
Microenterprises: Evidence from a Field Experiment," The Quarterly Journal of Economics, Vol. 123 (4), 1329-1372.
Ghatak, Maitreesh. 1999. "Screening by the Company You Keep: Joint Liability Lendings and the Peer
Selection Effect," Economic Journal.
58
Ghatak, Maitreesh and Timothy W. Guinnane. 1999. "The Economics of Lending with Joint Liability: Theory and
Practice," Journal of Development Economics, Vol. 60 (1999), 195-228.
Gine, Xavier and Dean Karlan. 2007. "Group versus Individual Liability: A Field Experiment in the
Phillipines" Center f or Global Development Working Paper Number 111,
http://ssrn.com/abstract981390.
Gine, Xavier and Dean Karlan. 2010. "Group Versus Individual Liability: Long Term Evidence from
Phillipine Microcredit Lending Groups," Yale Economics Working Paper No. 61; Yale University Economic Growth
Discussion Paper No. 970, http://ssrn.com/abstract=1407614.
Gine, Xavier, Pamela Jakiela, Dean Karlan, and Jonathan Morduch. 2010. "Microfinance Games,"
American Economic Journal, Forthcoming.
Gomez, Rafael and Eric Santor. 2003. "Do Peer Group Members Outperform Individual Borrowers? A
Test of Peer Group Lending Using Canadian Micro-Credit Data," Bank of Canada Working Paper 2003- 33.
Guttman, Joel M. 2006. "Repayment Performance in Group Lending Programs: A Survey," Networks
Financial Institute at Indiana State University Working Paper, 2006-WP-01, April 2006.
Guttman, Joel M. 2007. "Assortive Matching, Adverse Selection, and Group Lending," Journal of
Developmental Economics, Vol. 87 (1), 51-56.
Hermes, Niels and Robert Lensink. 2007. "The Empirics of Microfinance: What Do We Know?" The
Economic Journal, Vol. 117 (February), F1-F10.
Hulme, David and Paul Mosley. 1996. Finance against Poverty, Volume 1, Routlege Publications,
London.
Karlan, Dean S. and Jonathan Zinman. 2007. "Credit Elasticities in Less-Developed Economies:
Implications for Microfinance" (July 2007). (r&r American Economic Review)
Karlan, Dean S. and Jonathan Zinman. 2009. "Observing Unobservables: Identifying Information
Asymmetries with a Consumer Credit Field Experiment" Econometrica, Vol. 77 (6), 1993-2008.
Karlan, Dean and Jonathan Zinman. 2009. "Expanding Credit Access: Using Randomized Supply
Decisions to Estimate the Impacts," The Review of Financial Studies, Vol. 23 (6), 2560-2590.
Khawari, Aliya. 2004. "Microfinance: Does It Hold Its Promises? A Survey of Recent Literature,"
Hamburg Institute of International Economics Discussion Paper 276.
59
LaPorta, R., Lopez-de-Silanex, F., Shleifer, A., and Vishey, R. (1998), Law and Finance, 106 Journal of Political Economy,
1113 - 1115.
Leftley, Richard and Shadreck Mapfumo. "Effective Micro-Insurance Programs to Reduce
Vulnerability," Opportunity International Network
McGuinness, Elizabeth and Volodymyr Tounytsky. 2006. "The Demand for Microinsurance in
Pakistan." March 2006. Microfinance Opportunities
McLean, Kate. 2010. "Capitalizing on Women's Social Capital? Women-Targeted Microfinance in
Bolivia," Development and Change, Vol. 41 (3), 495-515.
Mersland, Roy. 2005. "The Agenda and Relevance of Recent Research in Microfinance." January 2005.
Unpublished working paper, Adger University College, Norway, http://ssrn.com/abstract=977138
Mersland, Roy and R. Øystein Strøm. 2009. "Performance and Governance in Microfinance
Institutions," Journal of Banking and Finance, Vol. 33 (4), 662-669.
"Morduch, Jonathan. 1998. "Does Microfinance Really Help the Poor? New Evidence from Flagship
Programs in Bangladesh," Research Program in Development Studies, Princeton University, Research Program in
Development Studies Working Paper 211.
Paxton, Julia, Douglas Graham, and Cameron Thraen. 2000. "Modeling Group Loan Repayment
Behavior: New Insights from Burkina Faso," Economic Development and Cultural Change, 48, 639-655.
Pellegrina, Lucia Dalla. 2006. "Microfinance and Investment: A Comparison Between Group Lending,
Bank Lending and Informal Credit" (December 15, 2006). Available at SSRN:
http://ssrn.com/abstract=953231
Pretes, Michael. 2002. "Microequity and Microfinance," World Development, 30, 1341-1353.
Stiglitz, Joseph E. 1990. "Peer Monitoring and Credit Markets," 4, World Bank Economic Review, 341-
366.
Townsend, R., 1979. "Optimal Contracts and Competitive Markets with Costly State Verification,"
Journal of Economic Theory, 21, 265-293.
Van Bastaleaer, Thierry. 2000. "Imperfect Information, Social Capital, and the Poor's Access to Credit,"
Center for Institutional Reform and the Informal Sector, University of Maryland Working Paper No. 234.
60
Essay 2: Issuances of Floating Rate Convertible Securities and Financial Manager
Characteristics
Abstract
Floating rate convertibles (FRCs) are a category of PIPE securities that receive negative associations in both the
academic and professional literature, earning monikers such as "death spirals" because of significant negative returns
to equity of firms subsequent to issuing them. This study sheds light on the managerial relationship to the decision to
issue FRCs and to the variation in market response to these issues. One main result of the study identifies influence of
the CFO relative to the CEO as significant in the decision to issue FRCs and in the market's immediate reaction to the
issuance. Another main result is that FRC issuing firms with CFOs without prior public equity issuance experience
have significantly negative long run abnormal returns, whereas FRC issuing firms with experienced CFOs do not.
Overall, I find support for the faulty contract design hypothesis for the firms with less experienced CFOs and a new
hypothesis consistent with optimal security design (OSD) for the firms with more experienced CFOs.
61
I. Introduction
Since 1995, a growing number of public firms have raised funds through issues of private investment
in public equity (PIPEs). PIPEs are common stock, preferred stock, convertible debt, convertible preferred stock,
stock warrants, and equity lines, which are sold to investors privately by public firms. While private issuances of stock
took place prior to 1995, that was the year in which the term, PIPE, came into use. PIPEs are typically more than
solely straight equity deals, but those that include asymmetric payoff features such as that of a warrant.
PIPEs typically have been seen as solutions to an information asymmetry problem that firms,
particularly small growth ones, face when needing to raise external funds publicly. Through the private negotiations
preceding a private issue, information can be conveyed to the investor at less cost than in a public issue. Nevertheless,
some private placement advisors are critical of a significant number of PIPE deals, contending that these deals' terms
are harmful to the issuing firms.
Given the theoretical benefit of PIPE issues in solving the information asymmetry problem and the
fact that firms choose to enter into deals that have been deemed harmful, why question the net benefit of any PIPE to
the firm? Investors in PIPEs that contain the arguably adverse terms for the firm are primarily hedge funds, which
make their profit on arbitrage opportunities and not on the long run performance of the firms. According to these
funds' detractors, they trick PIPE issuers to accept "bad" deals because issuing firm managers are unsophisticated or
have personal connections to the funds. The criticism says that the principal-agency conflict between shareholders
and management cause firms to issue PIPEs with "bad" terms.
A type of PIPE that has come under particular scrutiny is the floating rate convertible (FRC), which
is either a preferred stock or a bond that is convertible into common stock at a price determined by the future stock
price at the time of conversion. Chaplinsky and Haushalter (2003), Hillion and Vermaelen (2004), and others have
studied the firms issuing FRCs. They and finance practitioners have found a negative relationship between the
issuance of a floating rate convertible (FRC) and firm performance.
Existing evidence supports two theories for this relationship. The first, the faulty contract design
(FCD) hypothesis, states that firms mistakenly issue FRCs because they do not understand how they will impact long
term performance. The second theory, the financing of last resort (FLR) hypothesis, states that firms rationally issue
FRCs because they face severe information asymmetry problems and are unable to find financing elsewhere. This
second theory states that these firms' poor performances are simply anticipated by the FRCs themselves. Hindsight
indicates that certain FRCs
62
are not good financing options. Why then are they issued in the first place? How could have one predicted a poor
financing decision? And by implication, how can one predict poor financing decisions by firms now? I model the FRC
issue choice as a function of CFO characteristics. This approach yields interesting answers to the question of whether
FRC issues precipitate or simply anticipate poor future performance. It also addresses an interesting question of
whether the degree and type of sophistication of financial managers add value to the firm.
I show that the financial sophistication does indeed affect the FRC issuance decision and affects how
the market reacts to the issue. Firms where the CFOs are more highly compensated relative to their CEOs are more
likely to issue FRCs. Among FRC issuing firms, those with higher CFO-to-CEO compensation have lower abnormal
returns around the issuance announcements. In the long-term, FRC issuing firms whose CFOs do not have experience
accessing the public equity market have significantly negative stock returns. These findings support the FCD
hypothesis where the CFO has less experience. In addition, they support an alternative hypothesis of optimal security
design (OSD) where the CFO does have experience.
This paper is organized as follows. Section 2 reviews the literature on PIPEs and impact of financial
management on firm actions and performance. Section 3 discusses the sources of the data, methodology for
identifying control firms, and summary statistics. Section 4 presents hypotheses, including an introduction of the
optimal security design (OSD) hypothesis to the FRC literature. Section 5 analyzes the results of testing the
hypothesis. Section 6 concludes.
2. Literature
This study brings together two strands of financial research. The first is on the motivation and impact of PIPE
issues, particularly structured PIPE issues. The existing literature has come to support two major hypotheses for why
a firm would issue a floating rate convertible security (FRC): the faulty contract design hypothesis and financing of last
resort hypothesis. I join this research area to another strand, which is how CFOs' incentives and abilities along with
directors' financial ties affect financing actions of firms.
A. Private Investment in Public Equity (PIPEs)
How does a firm benefit from privately placing equity when it is already a registered public firm? In
a Myers and Majluf (1984) framework, a private issue could be a less costly way of conveying information to the
market while raising new equity. At the heart of Myers and Majluf's model is information asymmetry between a firm
and potential outside investors. In bad states, the firm knows that its stock is overvalued. In this case, the firm has the
reason to issue new equity because
63
it is overvalued. The market, then, would know whether the bad state of the world has been realized, and firm's stock
price would decline. If outside financing is necessary to invest in a positive net present value project, then some such
projects would not be taken. How, then, can this suboptimal investment policy be avoided? If information on the true
value of the assets and growth opportunities can be conveyed, then an issue of equity would not be a negative signal.
However, transmission of this information to the public is costly. For example, it may entail giving away trade secrets.
Theoretically, the certification benefit of PIPE issues is found in Hertzel and Smith (1993), which is
an extension of Myers and Majluf (1984)'s model of information asymmetry and security selection. Hertzel and Smith
(1993) suggest that private placements of equity can serve as a less costly way to convey information to the market and
thus allow for a more optimal investment decision rule in the Myers and Majluf framework. By issuing stock privately,
a firm can convey information through the negotiations with the investors, raise the capital it needs, and signal to the
market that it is issuing stock because of good growth prospects. Hertzel and Smith postulate and test that firms sell
their privately placed stock at a discount to pay for the due diligence costs.
The certification benefit of PIPEs is supported by the observation that the market reacts positively to
the announcement of PIPE issues even though they are typically issued at a discount to market price. Hertzel and
Smith (1993) strictly consider placements of common stock, and Chaplinsky and Haushalter (2003) consider all PIPEs
issued between 1995 and 2000. Consistent with Hertzel and Smith, Chaplinksy and Haushalter find positive
announcement returns for PIPEs. They also find that PIPEs with features that protect investors against declines in
stock prices have lower announcement returns than those with such protections. Examples of securities with
downside price protections are floating rate convertible debt and floating rate convertible preferred stock (FRCs).
They infer that the market interprets the terms as signals of the quality of the firms' prospects, because they indicate
that the issuing firms agree to the investors' concern that the firms' stock prices will decline in the future. The
convertible security holder may convert at a low stock price, forcing the firm to give more shares of stock than it
would at the time of the security's issuance. Terms with less downside protection indicates that private investors have
greater confidence in the firm's future performance. In other words, the structured nature of the PIPE signals that the
investor is unable to certify the value of the firm.
Studies on PIPE issuing firm performance that control for identity of the investor supports the
notion that there is a certification effect. Dai (2007) directly examines how VC's and hedge funds differ in their
relationship to PIPE issuing firms, finding that VCs tend to have seats on the issuing firms' boards of directors and to
have longer investment horizons than hedge funds. She finds
64
support for the argument that VCs provide certification to the PIPE issuers in which they invest. Similarly,
Krishnamurthy, Spindt, Subramaniam, and Woidtke (2004) find that the negative long run abnormal returns of firms
after issuing PIPEs are confined to the sample of nondistressed firms that issue to nonaffiliated investors. Distressed
firms, on the other hand, show a positive announcement effect and no long-run underperformance regardless of
whether the investors are affiliated or unaffiliated. This provides support to the hypothesis that the identity of the
investor provides a certification benefit.
Consistent with other studies (Hertzel and Smith, 1993; Dai, 2007; Krishnamurthy, et. al., 2004; and
Barclay, et. al. 2003), Brophy, et. al. (2006)¸ do not find support for a monitoring benefit in PIPE issues. The
monitoring hypothesis states that since PIPEs are issued to a block holder, the free- riding problem of equity
issuance to disperse shareholders is circumvented. The argument for a monitoring benefit is set forth by Wruck
(1989), who postulates that private placements enhance value by targeting a small number of investors rather than a
large number typical in a public issue. By concentrating ownership, investors are capable of monitoring the firm
more carefully. The discount that private investors usually receive, therefore, compensates for monitoring costs
according to Wruck's ownership concentration hypothesis. Hertzel and Smith, however, find that the ownership
concentration variable does not have statistical significance in a regression of abnormal returns in the presence of
certification variables. Barclay, Holderness, and Sheehan (2003) find that investors in PIPEs typically are passive, and
therefore they actually serve to entrench management, which they argue is why PIPE issues are typically followed by
long run negative abnormal returns.
Despite the support for a certification effect motivating the issuance of PIPEs at discounts, PIPE
issuers also tend to experience negative long run abnormal returns following the issue. This is documented by
Hertzel, et. al. (2002) and Chaplinsky and Haushalter (2003). Why then are PIPEs
issued?
One explanation is faulty contract design (FCD) hypothesis, which states that firms issue FRCs
because their managements do not recognize that the securities induce short selling in their firms' stock; the investors
of these FRC buy the shares in order to profit from anticipated manipulation of the stock price to the detriment of the
existing share holders. Even among nonstructured PIPE issues, Hertzel, et. al. (2002) find a negative post-event
performance, despite a positive announcement effect for private placements of equity. They interpret this result as
indicating that the issuing firm management is overly optimistic concerning growth opportunities. They speculate
that, rather than covering due diligence or monitoring costs, the discounts on privately placed stock indicate that the
investors in these issues are better informed concerning the firms' true lower value.
65
They do recognize such a story is inconsistent with an efficient market because, while privately issued stock are sold
at a discount, the market prices of existing stock are higher when the deals are announced.
Another explanation is the financing of last resort (FLR) hypothesis, which states that the issuing
firm needs external financing, and despite the costliness of the issuance, it will do so out of desperation. Indeed,
Chaplinsky and Haushalter (2003) find that over 80 percent of PIPE issuing firms have negative operating income
prior to an issue. They say that companies using PIPEs "appear to be highly distressed and have a high probability of
failure regardless of the actions taken by management. Therefore, it is difficult to judge the success of these contracts
based solely on the issuer's post issue performance." (Chaplinsky and Haushalter's study of PIPEs is restricted to the
1990s. Since 2000, however, PIPE financing has become more common among healthier firms.)
B. Structured PIPEs
If a non-structured PIPE provides a certification benefit, and the market reacts negatively to the issue of a structured
PIPE, then why would a firm opt to issue a structured PIPE? The development of hypotheses concerning relationship
between FRCs and their long run return and operating performance of their issuers is most comprehensively set forth
in Hillion and Vermalaen (2004). They test the faulty contract design (FCD) hypothesis, and financing of last resort
(FLR) hypothesis, as well as a third, the undervaluation hypothesis.
The undervaluation hypothesis states that firms issue FRCs because their price is undervalued.
Therefore, they issue a convertible security that converts at a future market price that the firm believes will match its
true value.
Using a dataset of all FRC issues from December 1994 - July 1998, Hillion and Vermaelen reject the
undervaluation hypothesis and find support for both the faulty contract design (FCD) and financing of last resort
(FLR) hypotheses. They reject the undervaluation hypothesis because firms that issue FRCs typically experience
significant negative abnormal returns in the years following issuance. Their support for the FCD hypothesis follows
from supporting several empirical predictions of this hypothesis. While negative abnormal returns would be expected
under both the FCD hypothesis and the FLR hypothesis, the reasons are different. One unique prediction of the FCD
hypothesis, for which the authors finds support, is that the conversion discount is negatively correlated with the
abnormal returns of the issuing firm and with the size of the issue. A unique prediction of the FLR hypothesis is that
the negative abnormal returns of an FRC issuing firm is accompanied by negative
66
abnormal operating performance because the issuance of the FRC is interpreted as a signal of future problems for the
firm. The authors find that this prediction holds also.
Chaplinsky and Haushalter (2003) consider all PIPE issues from 1995 - 2000. This study not only
considers more deals than Hillion and Vermaelen, but also provides further evidence of long run abnormal
performance of FRC issuers by comparing them to firms that issue PIPEs that do not provide the downside price
protection to the holders of the securities, such as straight equity and fixed rate convertible securities.
Brophy, Ouimet, and Sialm (2006) support the FLR hypothesis despite finding a relationship
between negative abnormal returns and the PIPE investor being more likely to be involved in short selling.
Specifically, they argue that firms issue FRCs as a last resort when they issue to hedge funds. They find that the long
run underperformance of traditional PIPEs is confined to the sample of firms issuing structured PIPEs to hedge funds.
This finding is consistent with the FCD hypothesis because hedge funds are not long term investors, but often engage
in significant amounts of short selling and operate by taking advantage of arbitrages (Dai, 2007). Nevertheless, the
authors do not accept the FCD hypothesis because the amount of short positions on firms issuing structured PIPEs to
hedge funds does not vary significantly from the short positions of other deals. Therefore, they argue that these
negative abnormal returns are due to these firms having the severest asymmetric information and agency problems,
risks that only hedge funds are able to hedge against. They also find that PIPE issuing firms that were backed by a
venture capitalist before IPOing do not experience the long run underperformance of other PIPE issuers. Their
interpretation of this finding is that VCs provide certification of these firms and therefore do not have the adverse
selection problems of firms without VC backing.
In summary of the literature on PIPEs, the motivation for the issuance is concluded to be despair
(FLR hypothesis) and/or ignorance (FCD hypothesis). A puzzling finding in the empirical literature, however, is that
there are long run negative abnormal returns to firms making private placements. Explanations for the puzzle have
been based on a lack firm management sophistication (ignorance; faulty contract design) or the market's slowness to
recognize that the underlying conditions of the issuer (despair; financing of last resort). Support has been provided
for both explanations. Observation of the level of sophistication and incentives of the issuing firm's financial
management would serve to disentangle the faulty contract design hypothesis from the financing of last resort
hypothesis. The literature has yet to incorporate such a control and therefore has not fully answered the question of
how much the security choice affects the value of the firm rather than merely signaling its existing value.
67
C. Financial Managers
Despite Hillion and Vermaelen (2004)'s observation that the amount of FRC's issuance has declined at the time of its
writing, Singh (2005) notes that there is still a significant amount of FRCs being issued. This observation either can be
indicative of the financing of last resort motivation functioning or a continued slowness of financial managers to
recognize the faulty design features of FRCs. According to Placement Tracker, there were 121 FRC deals in 2004.
While less than a third of issues made in 2000, this is still a significant amount of deals, amounting to $628 million
($3.2 billion in 2000). Though, Singh does find that more recently issued FRCs have terms that reduce the problems
that these FRCs have had in the past. Examples of terms that can control the price manipulation of the stock of FRC
issuers are (1) direct restriction on the investor from short selling, (2) floors on the conversion price, (3) restriction on
the number of shares convertible at one time, (4) reduced discounts relative to the reference price, and (5)
investigating whether the investor has engaged in price manipulation in the past. The change in these terms, therefore,
would support the view that these contracts were faulty and have merited being altered. However, not all issues
involve these
controls. Singh concludes,
"toxic convertibles represent a financial innovation that, through an
iterative and - unfortunately, for any investors - costly process, has improved its
design but is still used by the same types of firms as in the past..The rationale for use
of toxics appears to be driven by despair, ignorance, or both. However, with improvements
in contract design, as outlined earlier, smaller firms in need of capital that cannot
access the market for traditional securities are better
positioned to consider these securities"
The existence of research that cautions against the use of FRCs raises the question of why some
firms issue them and why others avoid them. The implication of the FCD hypothesis is that the financial
management of the issuers do not understand them. While there is a large literature on the impact of the chief
executive officer on the decisions and performance of the firm, there is an emerging literature on the impact of the
chief financial officer. CFOs do influence the performance of firms as evidenced by their removal following poor
performance in Mian (2001). The disciplinary removal of CFOs is robust to decision to retain or remove the CEO.
Chava and Purnanandam (2007) show that the incentives of the CFO, not the CEO, affect the firm's choice of floating
rate versus fixed rate straight debt. Brettel et. al. (2008) test Hackbarth (2004)'s model of firm leverage and CFO.
They find that firms with "overconfident" CFOs tend to have higher leverage.
While the above papers consider the interaction among firm performance, CFO actions, and CFO
incentives, the only area of research regarding direct measures of financial manager skill that the
68
author is aware of is in the mutual fund literature. Numerous studies document that mutual fund managers differ in
their stock picking skills (Wermers, 2000; Chen, et. al., 2001; Baker, et. al., 2005; and Harlow and Brown, 2007).
In a study of active fund managers, MBA school quality is positively and significantly related to fund
performance over 2000-2003 (Gottesman and Morey, 2006). This effect is particularly strong among
the top business schools, as ranked by Business Week. Other studies find a significance of the
undergraduate program from which the manager graduated (Chevalier and Ellison, 1999). Furthermore, other
certifications or degrees such as a CFA, other masters degrees, or Ph.Ds does not appear to correlate with fund
performance.
There also has been recent research regarding the financial connection and expertise of the board of
directors. Mitchell and Walker (2008) find that firms that are large and less likely to be in financial distress are more
likely to have commercial bankers on their boards. Also, firms that have higher leverage, less market value of equity,
and high investment / low Tobin's q or low investment / high Tobin's q are more likely to have bankers as directors.
Becker-Blease and Grein (2008) argue that
the advisory role of the board of directors needs to be considered in addition to the monitoring role.
III. Data Sample
A. Sources
All FRC deals made in 2001 and 2002 are considered. Deal data come from PrivateRaise. Stock
return and delisting information for each issuer come from CRSP, and financial statement data from Compustat.
Analyst coverage data come from I/B/E/S. CFO, CEO, and director data are hand collected primarily from Edgar
filings (10K's, 10Q's, and proxy statements mostly). Lexis Nexis and online business press articles are used where
Edgar and the companies web sites do not provide the information.
B. Pre-Issue Firm Characteristics
1. FRC Issuers
I identify the CFO characteristics of 61 firms issuing FRCs in 2001 and 2002. I am able to identify the CFO
characteristics of 43 firms that issued FRCs over the same period. Issuing firms tend have small size, high growth
opportunities, low leverage, high investment, operating losses, and high propensity to be in bankruptcy (median
Ohlson score of 0.88). See Table 1.
69
Table 1
Pre-Issue Firm Financial Characteristics
This table displays summary statistics of characteristics of FRC issuers and three groups of control firms, fixed price PIPE issuers, SEO issuers,
and match non-equity issuers measured in the year prior to the firms making these issues. (In the case of match non-issuers, FRC issuance
dates are imputed to them.) COVERED is an indicator that equals 1 if at least one stock analyst made a forecast for the firm. CFVOL is the
cash flow volatility, defined as the standard deviation of operating income up to twenty fiscal quarters before the announcement date. OSCORE is the probability of
becoming financially distressed defined by Ohlson (1980). FIRMVALUE is log of the market value of equity plus book values of preferred and total debt. TOBINQ is the
market value of the firm divided by the book value of the assets. LEVERAGE is the the long term debt divided by the book value of assets. INVESTMENT is the total of
R&D and advertising divided by the lagged property, plant, and equipment. PROFITABILITY is the operating cash flow before depreciation divided by lagged assets. FRC
issuers' variables' means are
significantly different from the fixed price PIPE issuers, SEO issuers, non-equity-issuers if denoted by an a, b, or c, respectively, at the 5% level.
Panel A: FRC Issuers
Variable
Statistic COVERED CFVO OSCORE FIRM- TOBINQ LEVERA INVEST- PROFIT-
L VALUE GE MENT ABILITY
N 43 43 43 43 43 43 43 43
Median 1.000 5.793 0.879 4.410 0.054 0.087 0.780 -0.192
Mean 0.581b 20.193 0.753 b 4.726b 0.180ab 0.470 2.563 -0.346bc
Standard Error 0.076 7.507 0.042 0.225 0.051 0.212 0.656 0.093
Panel B: Fixed Price PIPE Issuers
Variable
Statistic COVERED CFVOL OSCORE FIRM- TOBINQ LEVERA INVEST- PROFIT-
VALUE GE MENT ABILITY
N 40 40 40 40 40 40 40 40
Median 1.000 4.431 0.794 4.440 0.100 0.002 0.905 -0.139
Mean 0.600 10.784 0.715 4.245 0.135 0.100 2.942 -0.620
Standard Error 0.078 3.385 0.047 0.209 0.023 0.037 0.845 0.320
Panel C: SEO Issuers
Variable
Statistic COVERED CFVOL OSCORE FIRM- TOBINQ LEVERA INVEST- PROFIT-
VALUE GE MENT ABILITY
N 46 46 46 46 46 46 46 46
Median 1.000 18.352 0.530 6.792 0.015 0.555 0.076 0.150
Mean 0.934 25.127 0.529 6.693 0.030 0.562 1.787 -0.060
Standard Error 0.036 3.747 0.039 0.150 0.006 0.084 0.618 0.097
Panel D: Non-Issuers
Variable
Statistic COVERED CFVOL OSCORE FIRM- TOBINQ LEVERA INVEST- PROFIT-
VALUE GE MENT ABILITY
N 38 38 38 38 38 38 38 38
Median 1.000 14.222 0.870 5.047 0.035 0.022 0.183 -0.043
Mean 0.736 23.562 0.696 5.108 0.072 0.202 1.730 -0.057
Standard Error 0.072 4.221 0.053 0.308 0.018 0.043 0.624 0.043
The sample of FRC issuers tend to be in the information technology or pharmaceutical industries.
Using the 48 industry Fama-French industry definitions, the most represented industries among the 43 FRC issuing
firms are "Business Services" (ten), "Drugs" (five), "Medical Equipment" (four), and "Telecommunications" (four).
Only half of the firms have at least one analyst covering them in the year prior to issue. Therefore, I
mostly measure the degree of information asymmetry by using a dummy variable that equals one if the firm has at
least one analyst covering it in the prior year and zero otherwise. Among the 19
issuers that do have analyst coverage, the median earnings surprise is 30.80 percent. Am o n g
70
analysts covering the same firm, the median standard deviation in estimates of earnings is 16.76 percent. See Table 2.
Table 2
Pre-Issue Analyst Coverage Characteristics
This table displays summary statistics of analyst forecasts of FRC issuers and the three control group firms (fixed price PIPE
issuers, SEO issuers, and match non-equity-issuing firms). The mean surprise is the absolute percent difference in actual earnings from
forecasted earnings over the year. Dispersion is the standard deviation of forecasts among analysts. Maximum # of Analyst Coverage is the
maximum number of analysts covering the firm at the same quarter. FRC issuers' variables' means are significantly different from the fixed
price PIPE issuers, SEO issuers, non-equity-issuers if denoted by an a, b, or c, respectively, at the 5% level.
Panel A: FRC Issuers
Variables
Statistics
N
Median
Mean
Standard Error
Mean Surprise
19
0.308
0.396
0.065
Dispersion
19
0.005
10.103
9.472
Maximum # Analyst
Coverage
22
4.500
5.727b
0.947
Panel B: Fixed Price PIPE Issuers
Variables
Statistics
N
Median
Mean
Standard Error
Mean Surprise
21
0.302
0.709
0.234
Dispersion
20
0.059
1.041
0.895
Maximum # Analyst
Coverage
21
4.000
6.761
1.305
Panel C: SEO Issuers
Statistics
N
Median
Mean
Standard Error
Mean Surprise
45
0.082
0.291
0.107
Variables
Dispersion
45
0.049
0.109
0.032
Maximum # Analyst
Coverage
45
9.000
10.933
1.328
Panel D: Matched Non-Equity-Issuers
Variables
Statistics
N
Median
Mean
Standard Error
Mean Surprise
21
0.230
0.353
0.073
Dispersion
18
0.047
0.090
0.030
Maximum # Analyst
Coverage
21
5.000
10.761
3.110
These firm characteristics are consistent with prior studies of FRC issues and with the theory that
firms with severe information asymmetries and risk can use privately placed equity to find financing at a low enough
cost. Furthermore, the structured nature of FRCs indicates that investors have concerns that the values of the firms'
stock will drop. The apparent decline in number of FRC deals from 2001 to 2002 may also reflect the disproportionate
reluctance of investing in such firms as the stock market continued to cool.
71
2. Control Firm Identification
Finding the benchmark for these firms is a challenging task that some previous studies deal with.
Hillion and Vermaelen (2004) compare issuing firms to matched nonissuing firms using a propensity to issue matching
algorithm wherein they create a probit model for propensity to issue a FRC
(Dehejia and Wahba, 1998). They match every issuer with a nonissuer that has the closest propensity of
issuing a FRC.
Their probit model uses operating return on assets, profit margin, ROA, operating income / sales,
(capitalization expenditures + R&D) / assets, and market to book value of equity. I use the same or similar predictors
plus a measure for financial distress. I include the Ohlson financial distress variable ("O-score", or OSCORE)
(Ohlson, 1980) because I am considering a very special segment of stocks that are characterized by being in extreme
financial distress (Chaplinsky and Haushalter, 2003). These firms are likely to "go dark," i.e. cease trading on NYSE
or Nasdaq, which may occur because their stock price falls below the minimum levels allowed by the exchanges
(Leuz, Triantis, and Wang, 2006). Furthermore, the financing of last resort hypothesis states that it is precisely the
firms that are in financial distress that would issue a FRC.
In addition to OSCORE, I borrow variables used by Gomes and Phillips (2005) in their study of
public versus private security issuance choice. To measure risk, they use cash flow volatility
(CFVOL), defined as the standard deviation of operating income before depreciation (Compustat item data13) up to
twenty fiscal quarters before the announcement date. To measure information asymmetry, they use the mean earnings
surprise and dispersion of analyst earnings estimates discussed previously. Because half the sample of firms issuing
FRCs do not having any analyst coverage, I cannot use these variables without losing hal f the sample of firms,
leaving only 19. Therefore, I use a dummy variable, COVERED, that equals one if the firm is covered by at least one
analyst in at least one quarter prior to the issue, and zero otherwise. In addition to measures for risk and information
asymmetry, Gomes and Phillips (2005) use log of firm value, Tobin's q, leverage, investment in R&D, and
profitability as controls. The log of firm value (FIRMVALUE) is defined by the market value of equity plus book
values of preferred and total debt (in Compustat:
data24*data25 + data9 + data34 + data39). Tobin's q (TOBINQ) is defined by the market value of
the firm divided by the book value of the assets (exp(FIRMVALUE) / data6). The debt to asset ratio (LEVERAGE) is
defined by the long term debt divided by the book value of assets (data9t / data6t-1). Research and development
(INVESTMENT) is defined by the total of R&D and advertising divided
72
by the lagged property, plant, and equipment (data45t+data46t)/data8t-1. PROFITABILITY is defined
by the operating cash flow before depreciation divided lagged assets (data13t/data6t-1). 1
I regress the qualitative variable that equals one if a firm issues a PIPE of any kind and zero if it
does not on these variables with clustered standard errors by Fama-French 48 industry classification. The
regression model has an explanatory power of 9.86 percent. Consistent with previous research, firms that issue PIPEs
are more likely to be in financial distress and high R&D expenditure (all significant at the 5% or greater levels.) See
Table 3.
Table 3
Propensity to Issue a PIPE Logit Regression
Below are logit model estimates of the likelihood that a firm issues a PIPE. The standard errors are clustered by the Fama-
French 48 industry classification. COVERED is an indicator that equals 1 if at least one stock analyst made a forecast for the firm. CFVOL is
the cash flow volatility, defined as the standard deviation of operating income up to twenty fiscal quarters before the announcement date.
OSCORE is the probability of becoming financially distressed defined by Ohlson (1980). FIRMVALUE is log of the market value of equity
plus book values of preferred and total debt. TOBINQ is the market value of the firm divided by the book value of the assets. LEVERAGE is
the the long term debt divided by the book value of assets. INVESTMENT is the total of R&D and advertising divided by the lagged
property, plant, and equipment. PROFITABILITY is the operating cash flow before depreciation divided lagged assets. Estimates are
significant at the 10%, 5%, and 1% levels if denoted with *, **, or *** respectively.
Variable
COVERED
CFVOL
OSCORE
FIRMVALUE
TOBINQ
LEVERAGE
INVESTMENT
PROFITABILITY
CONSTANT
N
PSEUDO R-SQUARED
Coefficient Estimate
0.697***
-0.004
2.544***
0.094*
-0.017**
-0.106*
0.006***
-0.027*
-5.056***
10509
0.098
Robust Standard Error
0.132
0.002
0.206
0.051
0.008
0.061
0.002
0.014
0.302
Firms that are covered by at least one stock analyst are actually more likely to issue PIPEs. If
COVERED represents information asymmetry, then such a result is not expected if the regression were conditional on
issuing some type of equity security. However, the regression is based on all firms in the Compustat universe in 2001
and 2002. Therefore, firms that are not covered by an analyst at all, ceteris paribus, are less likely to issue a PIPE
(and in all probability, a SEO as well). If COVERED represents the costs to issuing equity publicly, then the fact that
PIPE issuers are likely to be covered by an analyst already may indicate that private costs borne by the managers of
the firm may be taken into consideration when making the issuance decision. Variables that are weakly significant in
the propensity equation are FIRMVALUE (positive sign), LEVERAGE (negative sign), and PROFITABILITY
(negative sign).
1 Gomes and Phillips (2005) use an alternative measure for financial distress, ALTMAN, a dummy variable that equals one if
the firm's Altman z-score is less than 1.81 (Altman, 2000). I also compute a propensity model using ALTMAN, and do not find qualitatively
different results. I choose the OSCORE because it produces a 1% higher pseudo-R2 in the logit regression.
73
Using the propensity scores computed on the universe of non-structured PIPE issuers, defined by
PIPEs that do not have downward price protections ("Fixed PIPEs"), SEO issuers ("SEOs"), and non- equity matched
issuers ("Non-Issuers"), I match the firms from each control group that have the closest propensity score
(PROPENSITY) to each FRC issuer.
Because the Non-Issuer control group has no issue date, for the purpose of the event study, I imput e
an event date to each Non-Issuer equivalent to the announcement date of a FRC issue of the firm with the closest
propensity score.
C. Deal Characteristics
1. FRC Deals
The median FRC deal is for $4.0 million and 12.1 percent of the issuing firm's market capitalization.
Table 4
PIPE Deal Characteristics
This table enumerates the characteristics of FRC deals and control set of fixed price PIPE deals.
Fixed Fixed
FRC PIPE FRC PIPE
Security Type Conversion Restriction
Common Stock 6 25 No 19 11
Convertible Debt 22 9 Unknown 14 25
Convertible Preferred 15 6 Yes 10 4
Stock
Conversion Type Selling Restriction
Fixed 0 40 No 27 26
Reset 22 0 Unknown 14 12
Variable 23 0 Yes 2 2
Warrants Included Hedge Restriction
No 15 23 No 23 25
Yes 28 17 Unknown 13 13
Yes 7 2
Antidilution Clause Forced Conversion
No 15 19 No 14 7
Unknown 12 11 Unknown 15 27
Yes 16 10 Yes 14 6
Hard Floor Price Investor Purchase
Rights
No 21 40 No 20 22
Unknown 7 0 Unknown 14 12
74
Yes 15 0 Yes 9 6
Soft Floor Price Mandatory Registration
No 33 40 No 6 8Unknown 7 0
Unknown 4 10 Yes 3 0 Yes 33
22
There is a wide range of deal sizes from $1.0 million to $1.5 billion, and percentages of market
capitalization from 1.7 percent to 108.8 percent. The median conversion price is calculated at 100.0 percent of the
reference stock price, with the lowest conversion price being 50 percent and the highest being 242 percent. These
deal characteristics are consistent with previous research finding that these deals are made at discounts and of
significant sizes. See Table 5.
Of the 43 deals, a significant number have at least one term that provides some type of limit on how
much investors can profit from stock price declines. See Table 4. Eighteen deals impose floors on the conversion
prices, and 10 deals restrict the amount of converting at one time. Two deals restrict short selling, and seven restrict
hedging by the investors, thus limiting their ability and incentive to exert selling pressure on the issuers' stocks. Ten
of the deals have forced conversion provisions, which would force the investors to convert their securities under
certain circumstances; this provision enables the firm to take advantage of a rising stock price by selling shares at the
future high market price.
Sixteen deals have anti-dilution clauses, which provide more shares to the investor if the firm issues
more shares in the future, so as to protect the investor from diluted value of shares. Thirty-three of the deals include
mandatory registration rights of the stock underlying the FRCs, making the investors able to sell the shares that they
convert. Nine of the deals have investor purchase rights, which gives the investor in the FRC the right of first refusal
when the firm attempts to issue future securities; thus, firms that issue these rights would be limited in its future
financing options.
Twenty-eight deals include warrants, which tend to have high exercise prices; median exercise price
is at a premium of 16 percent over the market price at issuance (See Table 5). Inclusion of warrants at premiums may
offset the interest the investors have in stock price declines. These warrants represent sizable stakes in the warrant
issuers, ranging from 4.4 percent to 150.0 percent of outstanding shares with a median of 33.3 percent.
75
2. Fixed-Price PIPE Deals
The deal sizes of the fixed PIPE issues of the control group are similar to those of the FRC issues.
The median deal size is $6.5 million and 16.0 percent of market capitalization. The median PIPE price is 97.0 percent
of the market stock price.
Fixed price PIPEs do not have the features that pressure the stock prices down as do FRCs.
Therefore, the net benefits of including specific terms differ between the two types of issuances. Fixed price PIPE
issuers are less likely to include conversion restrictions (4 out of 40 deals) and less likely to include hedge restrictions
(2 out of 40), indicating that some issuers who choose to issue FRCs over fixed price PIPEs negotiate terms that are
particularly important to moderating the impact of the FRC. However, other terms are equally present in the two types
of deals: short-selling restrictions and forced conversion clauses.
The fraction of deals with pro-investor terms is higher among FRC issues than fixed price PIPE
issues. Fixed price PIPE deals are less likely to have anti-dilution clauses (10 out of 40 deals) and less likely to have
registration rights (22 out of 40 deals). This pattern is contrary to that of the greater prevalence of pro-issuer rights
among FRC deals than fixed PIPE deals. Perhaps the investors in FRCs demand that they be protected from the
higher likelihood that the firm will need to raise more capital in the future via anti -dilution clauses and from the
likelihood that the investor will be unable to sell its shares if it waits too long to seek to have them registered.
Fixed PIPE issuers are also less likely to include warrants. Only 17 out of the 40 fixed price PIPE
deals include warrants. The lower rate of inclusion of warrants among fixed price PIPE issuers may indicate the lesser
need to provide rewards to the investor for upward movements in the stock price. The fixed price PIPE deals' median
warrant premium, 6.0 percent, and median warrant coverage amount, 50.0 percent, are comparable to the FRC issues.
3. SEO Deals
The size of SEO deals in comparison to the FRC deals is greater in absolute dollars but similar in percent or market
capitalization. The median SEO amount is $114.5 million, and the median deal amount as percent of market
capitalization is 16.8 percent. The median SEO price is 95.9 percent of the market stock price.
76
Table 5
Deal Characteristics
The pricing and quantity of deals among the FRC, fixed price PIPE, and SEO issuers are given below. The deal amount /
market cap is the amount raised divided by the market value of the stock at the time of issuance. The premium is the
percentage of the reference stock price. The warrant premium is in term of the stock price at the time of issuance. The warrant coverage is
the number of shares underlying the warrants as a percent of outstanding shares. FRC issuers' variables' means are significantly different
from the fixed price PIPE issuers, SEO issuers, non-equity-issuers if denoted by an a, b, or c, respectively, at the 5% level.
Panel A: FRCs
Statistics Deal Amount Deal Amount / Premium (%) Warrant Warrant
($000) Market Cap (%) Premium (%) Coverage (%)
N 43 43 43 28 27
Median 4.000 12.100 100.000 115.500 33.300
Mean 45.518b 21.041 103.504 117.642 49.644
Standard Error 34.694 4.035 5.554 8.740 7.447
Panel B: Fixed Price PIPEs
Statistics Deal Amount Deal Amount / Premium (%) Warrant Warrant
($000) Market Cap (%) Premium (%) Coverage (%)
N 40 40 40 17 17
Median 6.500 16.000 97.000 106.000 50.000
Mean 10.047 21.707 99.017 112.970 64.635
Standard Error 1.611 4.136 5.610 5.603 11.384
Panel C: SEOs
Statistics Deal Amount Deal Amount / Premium (%)
($000) Market Cap (%)
N 46 44 44
Median 114.550 16.760 95.893
Mean 171.258 36.238 101.490
Standard Error 33.836 11.037 4.294
D. CFO and Other Management Characteristics
CFO Characteristics. The median age of the CFOs of FRC issuers is 43.5 years. The median amount
that CFOs own of the firm is 1.0 percent. Fifty-one percent of CFOs have experience as a CFO, treasurer, or
comptroller of another public firm. However, only 27 percent were financial officers at public firms when those firms
issued either IPOs or SEOs. Among the more limited number of FRC issuers for which I could find educational data,
42 percent have CFOs who have MBAs. CFOs of FRC issuers are less likely to have been at the firm when it went
public than the control firms. See Table 6.
In addition to the CFO variables in Table 6, the tenure of the CFO at her firm is collected. Only two
of the CFOs of FRC issuers were at their firms when they made their initial public offerings. However eight fixed
price PIPE issuers, seven SEO issuers, and four non-equity issuers have CFOs at the time of issue who also were
CFOs at the time of IPO. See the Appendix for the biographical information provided by firms on their CFOs,
organized by CFO public equity experience.
77
Boards of Directors. The median board size of FRC issuers is six. The median percent of board members who are
also employees of the firm ("insiders") is 33.33. The median percent of board members who are financial
professionals is 16.67. Financial professional is defined as being an employee of a financial institution, such as an
investment bank or hedge fund, or being a financial officer, such as a CFO, of another firm. The median of the mean
age of the firms' directors is 54.8 years. The median ownership of the firm by all directors and managers is 18.1
percent. See Table 6 for summary statistics on all four groups of firms' managerial characteristics.
Table 6
Management Characteristics
The descriptive statistics of the CEOs, CFOs, and boards of directors of the FRC issuers and control firms are shown the
following panels. These values are collected in the period prior to the security issuances, or imputed issuances in the case of
the matched non-equity-issuers. CFO / CEO Comp is the fraction of CFO cash compensation to CEO cash compensation. % Board Insiders is
the percent of board members who are also managers of the same firms. % Board Financial Experts is the percent of board members who are
either employed by a financial institution or is a CFO. FRC issuers' variables' means are significantly different from the fixed price PIPE
issuers, SEO issuers, non-equity-issuers if denoted by an a, b, or c, respectively, at the 5% level.
Panel A: FRC Issuers
Standard
Variable N Median Mean Error
CFO with Prior Public Firm Experience 53 1 0.547b 0.069
CFO with Public Equity Offering Experience 54 0 0.203 b 0.055 Total Director and
Management Ownership 51 18.100 24.306 b 2.571
CFO/CEO Comp 44 0.622 0.692 b 0.051 %
Board Insiders 51 0.333 0.390 ab 0.041 %
Board Financial Experts 51 0.166 0.217 0.032
Avg. Board Age 51 54.80000 53.649 0.891
CFO with MBA 57 0.000 0.421a 0.065
CFO with Unknown Education 57 0.000 0.175 abc 0.050
CFO Ownership 47 0.010 0.020 b 0.005
CEO Ownership 51 4.4000 9.447 ac 1.939
CEO Compensation 50 325072 611264 236765
CFO Compensation 45 199615 229416 b 19358
CFO Age 36 43.500 42.250 b 1.257
Board Size 38 6.000 6.289 0.330
Panel B: Fixed Price PIPE Issuers
Standard
Variable N Median Mean Error
CFO with Prior Public Firm Experience 40 0.000 0.425 0.079
CFO with Public Equity Offering Experience 40 0.000 0.325 0.075
Total Director and Management Ownership 42 13.910 21.130 2.906
CFO/CEO Comp 41 0.606 1.978 1.323 %
Board Insiders 43 0.250 0.294 0.024 %
Board Financial Experts 43 0.222 0.271 0.030
Avg. Board Age 43 54.500 53.820 0.732
CFO with MBA 47 0.000 0.234 0.062
CFO with Unknown Education 47 0.000 0.489 0.073
CFO Ownership 42 0.010 0.012 0.003
CEO Ownership 42 0.0272 0.0626 0.016
CEO Compensation 43 310000 365279 49104
CFO Compensation 42 200053 210114 17327
CFO Age 34 42.500 43.882 1.452
Board Size 36 6.000 6.222 0.314
78
Panel C: SEO Issuers
Standard
Variable N Median Mean Error
CFO with Prior Public Firm Experience 41 1.000 0.804 0.062
CFO with Public Equity Offering Experience 41 1.000 0.951 0.034
Total Director and Management Ownership 40 7.850 16.052 3.024
CFO/CEO Comp 41 0.409 0.502 0.041 %
Board Insiders 39 0.285 0.290 0.024 %
Board Financial Experts 39 0.200 0.215 0.025
Avg. Board Age 38 55.160 55.832 0.949
CFO with MBA 47 0.000 0.361 0.070
CFO with Unknown Education 47 0.000 0.468 0.073
CFO Ownership 40 1.000 0.985 0.027
CEO Ownership 40 2.045 5.352 1.617
CEO Compensation 41 806077 1033665 120987
CFO Compensation 41 357180 389269 27236
CFO Age 44 45.000 45.750 1.112
Board Size 36 7.000 7.194 0.313
Panel D: Non-Issuers
Standard
Variable N Median Mean Error
CFO with Prior Public Firm Experience 39 0.000 0.435 0.080
CFO with Public Equity Offering Experience 39 0.000 0.282 0.073
Total Director and Management Ownership 39 21.400 25.801 3.421
CFO/CEO Comp 38 0.586 0.660 0.074 %
Board Insiders 39 0.333 0.366 0.039 %
Board Financial Experts 39 0.250 0.270 0.031
Avg. Board Age 39 53.571 52.761 0.992
CFO with MBA 47 0.000 0.319 0.068
CFO with Unknown Education 47 1.000 0.510 0.073
CFO Ownership 39 0.010 0.013 0.001
CEO Ownership 40 0.023 0.071 0.014
CEO Compensation 39 399250 547564 69821
CFO Compensation 39 206153 296634 52068
CFO Age 32 45.000 44.750 1.448
Board Size 31 7.000 7.129 0.421
PIPE Investor Identity. Investors in FRCs are unlikely to be firm managers. Only two FRC issues
are purchased by management, whereas eight fixed price PIPE issues were purchased by management. FRC and
fixed price PIPE issues are made to investors who hold seats on the issuers' boards at a rate less than ten percent. Four
FRC issuers and six fixed price PIPE issuers sold PIPEs to institutions with board seats.
E. Announcement Cumulative Abnormal Returns
A Fama-French three-factor plus momentum model was built to calculate abnormal returns. The CARs are also
robust to alternative market models: the capital asset pricing model and the Fama- French three factor model. The
coefficients for each model are determined using a 250 market day window prior to three months prior before the
announcement dates. Table 7 presents the cumulative abnormal returns (CARs) over an event windows covering five
days prior through five days after the announcement ([-5,+5]).
79
Table 7
Cumulative Abnormal Returns by Firm Group
Cumulative abnormal returns are calculated five days prior through five days after the issuance announcement (imputed
announcement in the case of the matched non-equity-issuers). The abnormal return is computed by using a four factor
market model: the excess market, small minus big portfolio, high minus low market to book portfolio, and the winners minus losers
momentum portfolio. The coefficients on each of these are calculated over the 250 market day period a month prior to
the events for each firm. Estimates are significant at the 10%, 5%, and 1% levels if denoted with *, **, or *** respectively.
Freque Standard
Firm Type ncy Minimum Median Maximum Mean Error
FRC Issuer 41 -0.387 -0.040 0.682 -0.011 0.035
Fixed Price PIPE Issuer 34 -0.421 0.004 2.339 0.152* 0.083
SEO Issuer 35 -0.188 0.024 0.889 0.063 0.030
Non-Equity Issuer 31 -0.301 0.016 0.473 0.028 0.030
Total 141 -0.421 0.013 2.339 0.055** 0.024
F. Post-Issue Firm Characteristics
In the year of the FRC issuance, the only financial ratios presented in Table I that significantly change is financial
distress (OSCORE) and the value of the firm (FIRMVALUE). The mean probability of financial distress increases
13.01 percent and the mean firms' value decreases by $0.50 million, both with significances greater than the one
percent level. From the year of issue to the following year, neither the level of financial distress nor firm value change.
However, from the year of the issue to the following year, Tobin's q increases by eight percent and R&D drops by 64
percent, both at the five percent level of significance. These patterns are also robust to industry adjustments. These
changes in operating performance over the period of a year prior to the year of issue are only found in the group of
FRC issuers. SEO issuers and non-equity issuers experience significant drops in the probability of financial distress
(SEO: -10.1 percent; Non-Issuers: -12.2 percent). SEO issuers' value significantly increases by 23.2 percent, and their
leverage significantly decreases by 16.4 percent. Fixed price PIPE issuers and non-equity-issuers decrease their levels
of R&D investment (Fixed PIPE: -127.6 percent; Non-Issuers: -51.2 percent). Among the control groups, there is little
change from the year of issue to the following year. Fixed price PIPE issuers are more likely to be in financial distress
by a mean of 10.5 percent. Non-equity issuers' profits increase by a mean of 6.7 percent.
While the level of R&D investment does not decrease significantly from the year prior to and year of
issue among FRC issuers, the fixed PIPE issuers and non-equity-issuers have significant decreases in R&D. However,
in the year following issue, only FRC issuers significantly reduce R&D investment. This finding is consistent with
Hillion and Vermaelen (2004)'s, who interpret the relative initial run up in R&D and subsequent decline in R&D as a
pre-issue confidence by issuing management in future returns to their investment.
80
Of the 43 FRC issuers, ten are delisted and one is acquired within one year following the issuance. However, only two
of the 40 fixed-price-PIPE issuers are delisted and three are acquired in one year following the announcement. None of
the 46 SEO issuers are delisted and one is acquired. The only control group that has a similar number of delistings to
the FRC issuer group is 38 non-equity issuers, of which seven are delisted.
The similarity between FRCs and non-issuers in terms of delisting rates indicates that the issuance
of a FRC does not increase its chances of remaining in business (or, is predictive of being able to stay in business). On
the other hand, fixed PIPE issuers and SEO issuers have a greater chance of remaining in business and being
acquired than FRC issuers and non-equity issuers. This pattern does not support the FLR hypothesis.
IV. Hypotheses
Hypotheses concerning the rationale for the existence of FRCs are presented in Hillion and
Vermaelen (2004): the undervaluation, faulty contract design, and financing of last resort
hypotheses. Two of these, the FCD and FLR hypotheses are not rejected by the authors, and they are two explanations
that are commonly accepted today for the existence of FRCs. Either of the two
hypotheses place FRCs in a negative light. In the case of the FCD hypothesis, the FRC is an
instrument of predation by unscrupulous investors who take advantage of firms whose management is either ignorant
or in collusion with the investors. In the case of the FLR hypothesis, the FRC is the only source of continued financing
for firms that would otherwise have to cease operations.
While the latter hypothesis offers a rational explanation for the existence of FRCs, it does not offer a
more positive explanation than that they are cheap enough for investors so that they will be willing to purchase them.
Why, however, is the floating conversion feature included rather than a more deeply discounted fixed conversion
price? The FLR hypothesis does not provide an answer.
Additional explanations appeal to fundamental concepts in corporate finance. FRCs may exist to
resolve problems that reduce firm value, which may be explained by an "optimal security design" (OSD) hypothesis.
Two possible examples of the OSD hypothesis are (i) the tradeoff between the
debt tax shield and financial distress and (ii) the problem of debt overhang:
(i)
Trade-Off: A firm determines an optimal leverage by maximizing the net benefit of the
debt tax shield minus bankruptcy costs. The costs of bankruptcy are higher with the likelihood of
default. If the market value of the stock falls after the issuance of debt and equity securities, the tax
shield benefits decrease due to lower likelihood of having any
81
positive pre-tax income before interest payments for the coming years and the bankruptcy costs
increase due to higher likelihood of bankruptcy. The FRC causes the investors to change the debt-equity
ratio in this scenario without the firm having to negotiate with bondholders.
(ii) Debt Overhang: A firm issues preferred stock or debt. After issuance, the real option
value of its growth prospects decrease. This is reflected in a declining stock price. If the capital structure
remains the same, the management, acting in the interest of the common equity holders, does not take
positive NPV projects because of the overhang
from debt or required dividends (Myers 1977). To avoid the renegotiation costs of
exchanging the preferred stock or debt for common stock, the firm issues a FRC so that when the real
option value goes down, the FRC is automatically converted into common shares because it is in the
interest of the FRC investors.
The hypotheses that this study tests are the FCD, FLR, and OSD hypotheses. The FLR hypothesis
and OSD hypotheses are similar in that they provide a rationale for FRCs that does not depend on behavioral
arguments. They differ in that the FLR hypothesis basically says that FRCs are issued because they are cheap for the
investor, but the OSD hypothesis provides rationales specifically for
the floating conversion price feature. Because this study's purview is empirical and the OSD
hypothesis is only introduced here, I only test the OSD hypothesis in the broad sense. Testing the specific
manifestations of the OSD hypothesis is reserved for future research.
All firms do not necessarily issue FRCs for the same cause. Some firms may be unwitting victims of
exploitive investors who intend to manipulate the stock price downward in order to expropriate a larger portion of the
firm's equity. Many firms could be fully aware of the costs of FRC issuance and make the decision to issue the FRC
wisely. Therefore, the empirical tests of this study do not seek to
accept one hypothesis as true for all firms and reject other hypotheses. Rather, the tests are
designed to test whether certain characteristics of firms' management can be linked to the various explanations for
FRC issuances.
The group of FRC issuing firms is compared to the three control groups: fixed price PIPE issuers,
SEO issuers, and no- equity issuing matched firms. Abnormal announcement returns are based on the Fama French
three-factor plus momentum model and are computed using the 250 market day window prior to the month of
issuance. The control groups are determined by using a score measuring the propensity to issue a PIPE. The non-
equity-issuing matched firms are imputed with
82
event dates the same as the FRC issuing firm with the closest propensity scores. Managerial
characteristics of FRC issuers relative to the control firms are related to the following three areas: propensity to issue
FRCs, stock market FRC announcement reaction, and long-term stock performance of FRC issuers.
In the first area, multinomial logit regression of the issuance type among the control groups is run
on managerial characteristics. Significant coefficients on sophistication and/or incentives provide
support to the FCD hypothesis and lack of support to the FLR and OSD hypotheses. If less
sophistication, greater conflicts of interest, and less monitoring of financial managers correlate with greater likelihood
of issuing a FRC, then the FCD hypothesis is supported. The FLR and OSD hypotheses would not necessarily be
refuted because it would be possible for certain FRC issuers to issue FRCs that benefit their firms if they have CFOs
who have above average sophistication than the entire sample of FRC issuers.
In the second area, the stock market cumulative abnormal return is regressed on the managerial
characteristics. If the market reacts negatively to firms issuing FRCs with unsophisticated or
conflicted management, then the FCD hypothesis may be true for those firms. In addition, I would interpret such a
relationship to support the OSD hypothesis because the market only would be reacting negatively when the FRC is
issued to the determinant of existing shareholders. I would not interpret such a finding to support the FLR hypothesis,
however, because if the market differentiates FRC issues by managerial characteristics, then it should also be able to
interpret the issuance as a negative signal immediately upon announcement rather than over a long period of ti me
after issuance.
In the third area, calendar time alphas of a portfolio long on experienced and short on inexperienced
FRC issuers is computed. If this portfolio is positive, then this supports the FCD and OSD
hypotheses. I would conclude that a non-equity-value maximizing choice was made where the firm management is
less experienced. On the other hand, among the firms where the firm management is experienced, I would conclude
that the FRC was issued to maximize existing shareholder value if
no long run abnormal returns are detected among these same firms. I would reject the FLR
hypothesis if only the FRC issuers with inexperienced CFOs have negative long run returns because the FLR
hypothesis predicts that all FRC issuers experience negative long run returns.
83
V. Results
A. Relationship of Managerial Characteristics on Issuance Choice
Unconditional Mean Comparison of Firm Managements. If FRCs are issued because they are faulty contracts rather
than financings of last resort, then the firms that issue FRCs are less financially savvy than similar firms. Or, their
management has conflicts of interest whereby they have interests with the FRC investors. I show that firms with CFOs
who have had experience placing equity
publicly and those with boards that are less composed of insiders are less likely to issue FRCs.
The unconditional means between the two groups show that 85.85 percent of the non-FRC issuing
firms have CFOs who are the CFO, treasurer, or comptroller over a firm when it makes either an IPO or SEO. On the
other hand, only 27.91 percent of the FRC issuing firms had CFOs with such experience. This is statistically
significant at the five percent level. (Compare Panel A with the other
panels of Table 6.) This difference in groups that are otherwise just as likely to issue a PIPE
suggests that the CFOs' experience issuing public equity predicts which kind of security is issued. This result supports
the argument underlying the FCD hypothesis that firms issue FRCs because their management is not comfortable with
alternative ways to issue equity.
The unconditional means of the two groups show that the average percent of insiders on the board of
directors is 29.84 percent among non-FRC issuing firms. (See Table 8.) On the other hand, the average insider
percentage is 40.33 percent among FRC issuers. The difference is statistically significant at the five percent level.
This difference in groups suggests that the quality of board monitoring affects the decision to issue FRCs. This result
also supports the belief that faulty contract design is an influential reason for firms to issue FRCs because if FRC
issuance were to maximize the value of existing equity in the firm, then firms whose boards are more independent
from management would permit the issuances of FRCs.
Other variables do not display statistically significant differences between the two groups. Whether
the CFO was previously a financial officer of a different public firm or a non-public firm does not make any
difference between the FRC issuers and the control firms. Officer and director ownership of the firm is not different
between the two groups. The percent of directors who are financial experts, i.e. those who are officers in a financial
institution or are CFOs themselves, is not different between the two groups either.
The data of one variable, MBA, is frequently less available than the others in the firms' filings.
Among the control group, 60.32 percent of firms' CFOs did have MBAs, and among FRC issuers, 44.44% had MBAs.
However, this difference is not statistically significant. Because this field has
84
more missing values, the regressions in the next section use a dummy variable that equals one when the education of
the CFO is unknown and zero otherwise, and the MBA variable equals one if the CFO is known to have an MBA and
equal to zero if otherwise.
Regression. A multinomial regression is run on the managerial characteristics and the PIPE
issuance propensity score. The dependent variable takes four values: 0 if the firm is a matched non- equity issuer, 1 if
a FRC issuer, 2 if a fixed price PIPE issuer, and 3 if a SEO issuer. The propensity score is included as a regressor as an
additional control beyond the fact that the firms are already similarly matched. The results of the regression ar e robust
to excluding the propensity score. I find that firms with CFOs with prior public equity experience are less likely to
issue FRCs at the ten percent level. The regression does not support the unconditional means finding that more board
insiders are positively correlated with FRC issuance in the presence of other management variables. I do find that the
ratio of CFO to CEO pay is positively correlated with PIPE issuance over SEO or no equity issuance at the five
percent significance level. This relationship may be saying that, all else equal, a firm where the CFO is more on par
with the CEO in influence is more likely to issue PIPEs.
Table 8
Issuance Choice: Multinomial Logit Regression
A multinomial regression model is presented. The left hand side takes four possible: FRC issuer, fixed price PIPE issuer,
SEO issuer, and match non-equity issuer. The non-issuer is the excluded class. The coefficients with standard errors in
parentheses are given below. Statistical significance at varying levels is denoted with a * (10%), ** (5%), and *** (1%).
Issuer Type
Variable
CFO with Prior Public Firm Experience
CFO with Public Equity Offering Experience
Total Director and Management Ownership
CFO/CEO Comp
% Board Insiders
% Board Financial Experts
Avg. Board Age
CFO with MBA
CFO with Unknown Education
CFO Ownership
FRC Issuer
1.253*
(0.745)
-1.451*
(0.864)
4.23E-05
(0.022)
3.707**
(1.599)
1.183
(1.597)
-1.768
(1.752)
0.032
(0.059)
-0.273
(0.817)
-2.117
(0.919)
Fixed Price PIPE
Issuer
-0.634
(0.704)
0.156
(0.758)
-0.016
(0.020)
3.420**
(1.555)
-2.984
(2.173)
0.881
(1.636)
0.077
(0.060)
-0.893
(0.879)
-0.687
(0.798)
-0.161
SEO Issuer
-0.138
(1.270)
5.249***
(1.550)
-0.188**
(0.075)
-3.253*
(1.807)
0.0662
(2.056)
2.562
(3.245)
-0.023
(0.089)
-0.143
(1.471)
-1.170
(1.615) -
0.942
85
CEO Ownership
CFO Compensation
Propensity to Issue PIPE
Intercept
Number of Observations
Pseudo R-squared
-0.071
(0.198)
0.040
(0.037)
-4.49E-06*
(2.41E-06)
-4.335
(8.496)
-2.125
(3.478)
112
0.419
(0.224)
0.022
(0.040)
-8.07E-06***
(3.08E-06)
-4.826
(8.173)
-1.935
(3.604)
(1.756)
0.188**
(1.756)
2.66E-06
(2.10E-06)
-8.718
(10.950)
1.521
(5.463)
B. Stock Market Reaction to FRC Announcement by Managerial Characteristics
The variation in the stock market reactions to FRC issuances is explained partly by management characteristics. Four
separate regressions are run, one for each group of firms. See Table 9. Among FRC issuers, the only characteristic at
the five percent level of significance in CARs around the announcement is the ratio of CFO compensation to CEO
compensation. The relationship between abnormal returns and this variable is negative, suggesting that the market
reacts negatively to FRC issuers by firms in which the CFO has more clout relative to the CEO.
The only other variable displaying significance in the FRC announcement reaction regression is the
unknown education dummy variable, having a positive sign. The interpretation for this result is unclear. Unavailable
educational data may be a proxy for either less reporting quality, which would
not support the argument that managerial quality impacts fi rm value. Alternatively, citing
educational background may be a substitute for citing experience, in which case the significantly positive sign on
unknown education would support the argument that managerial quality does impact firm value.
Number of board insiders appears as a significant variable in the SEO and Non-Issuer regressions,
but not the FRC issuer regression. This may indicate that corporate governance is a concern when SEOs are issued or
no issuance is made by firms similar to the issuing firms.
Apart from the CFO to CEO compensation ratio, I do not see market reaction being sensitive to
managerial characteristics in a clearly interpretable way. The market may be concerned in
particular by a FRC issuance announcement when the CFO is more highly compensated. This
relationship supports the FCD hypothesis where the CFO has less oversight because she holds a higher rank in the
firm.
86
Table 9
Regressions: 11 day CAR on Management Characteristics by Firm Type
Each column represents a separate regression of the CAR [-5,+5] around issuance announcements (or imputed
announcements) on the managerial variables. Cumulative abnormal returns are calculated five days prior through five days
after the issuance announcement (imputed announcement in the case of the matched non-equity-issuers). The abnormal return is computed
by using a four factor market model: the excess market, small minus big portfolio, high minus low market to book portfolio, and the winners
minus losers momentum portfolio. The coefficients on each of these are calculated over the 250 market day period a month prior to the
events for each firm. The managerial values are collected in the period prior to the security issuances, or imputed issuances in the case of the
matched non-equity-issuers. CFO / CEO Comp is the fraction of CFO cash compensation to CEO cash compensation. % Board Insiders is
the percent of board members who are also managers of the same firms. % Board Financial Experts is the percent of board membe rs who
are either employed by a financial institution or is a CFO. The coefficients with standard errors in parentheses are given below. Statistical
significance at varying levels is denoted with a * (10%), ** (5%), and *** (1%). FRC issuers' variables' means are significantlydifferent from
the fixed price PIPE issuers, SEO issuers, non-equity-issuers if denoted by an a, b, or c, respectively, at the 5% level.
Fixed Price
Variable
CFO with Prior Public Firm Experience
CFO with Public Equity Offering
Experience
Total Director and Management
Ownership
CFO/CEO Comp
% Board Insiders
% Board Financial Experts
Avg. Board Age
CFO with MBA
CFO with Unknown Education
CFO Ownership
CEO Ownership
CFO Compensation
Propensity to Issue PIPE
Intercept
FRC Issuer
-0.048
0.105
0.019
0.156
0.005
0.003
-0.396**c
0.179
-0.104 c
0.146
0.271
0.221
0.014
0.016
-0.008
0.132
0.364** ac
0.139
-0.090
0.057
0.005
0.005
3.44E-7
3.79E-7
2.159
2.162
-0.926 c
0.999
PIPE Issuer
0.027
0.416
-0.066
0.408
-4.69E-4
0.009
-0.611
0.857
0.100
0.975
0.120
0.848
0.001
0.028
-0.194
0.353
-0.256
0.273
-0.106
0.290
-0.001
0.015
2.53E-7
1.80E-7
4.299
3.540
-0.283
1.628
SEO Issuer
0.0839487
0.0526844
-0.1092252
0.1125912
0.0079937
0.0041605
-0.0588477
0.1246221
-0.3762595**
0.1701241
-0.0543524
0.1641585
-0.0050111
0.0053303
0.0768565
0.0938519
0.011357
0.1030467
0.3157607
0.2763186
-0.0056735
0.0045315
-1.18E-7
1.19E-7
1.204**
0.580
0.074
0.354
Non-Issuers
-0.1047058
0.0829212
-0.0378850
0.0892258
-0.0040279
0.0031122
0.0409362
0.1454364
-0.7817884**
0.2892039
-0.0922416
0.2475943
-0.0143507**
0.0068937
0.1083244
0.1307238
0.1142139
0.1187252
0.0272749
0.0456118
0.0154634**
0.0064923
1.73E-7
1.46E-7
2.255**
0.919
0.719
0.473
C. Long-Run Stock Performance
87
Four portfolios are created of firms that either issued each for each group of firms. The monthly return of these firms
is regressed on the Fama French plus momentum factors over 2001-2003. Each group's monthly portfolio is composed
of firms who issue (or are imputed with an issue) in the twelve months prior to the particular month. The non-issuers
have an issuance date imputed to them according to the issuance dates of the FRCs to which the non-issuer have the
closest propensity
scores. The significances of the estimates of alpha are evaluated by the mean return of 1,000
regressions using random samples of firms in the same size and book-to-market deciles as the FRC issuers (Mitchell
and Stafford, 2001). The portfolio returns of firm group p in month t are regressed
on a constant, the Fama French factors, and the momentum portfolio:
r
p
,
t
= o
p
+ |
p
,MKT r
MKT
+ |
p
,SMBr
SMB
+ |
p
,HMLr
HML
+ |
p
,MKT r
MKT
+ |
p
,UMDr
UMD
+ c
p
,t
Table 10
Calendar Time Portfolio Alphas
Each row corresponds to a separate regression that predicts the return of the row's portfolio description. The monthly returns
are regressed on the Fama French plus momentum factors over 2001-2003. Each group's monthly portfolio is composed of firms who i ssue
(or are imputed with an issue) in the twelve months prior to the particular month. The non-issuers have an issuance date imputed to them
according to the issuance dates of the FRCs to which the non-issuer have the closest propensity scores. The significances of the estimates of
alpha are evaluated by the mean return of 1,000 regressions using random samples of firms in the same size and book-to-market deciles as the
FRC issuers (Mitchell and Stafford, 2001).
Portfolio Alpha MKTRF SMB HML UMD R2
FRC Issuers -0.024 1.763*** 0.295 0.730 -0.267 0.54
(0.018) (0.495) (0.533) (0.570) (-0.400)
FRC Issuers - Fixed -0.013 -0.732 -1.135 0.763 -0.621 0.13
PIPE Issuers (0.025) (0.679) (0.731) (0.781) (0.548)
FRC Issuers - SEO -0.017 1.012 -0.658 0.919 -0.312 0.22
Issuers (0.022) (0.613) (0.660) (0.705) (0.495)
FRC Issuers - -0.050* 0.788 -0.651 0.624 -0.109 0.10
Nonissuers (0.025) (0.688) (0.741) (0.792) (0.555)
Inexperienced FRC -0.045** 1.441** 0.222 0.760 -0.291 0.42
Issuers (0.020) (0.538) (0.580) (0.620) (0.435)
Experienced FRC 0.097** 1.452 0.484 -0.824 0.480 0.12
Issuers - Inexperienced (0.040) (1.086) (1.169) (1.250) (0.876)
FRC Issuers
Experienced Fixed PIPE 0.008 -0.881 -0.265 0.467 -0.055 0.08
Issuers - Inexperienced (0.038) (1.016) (1.094) (1.169) (0.820)
Fixed PIPE Issuers
Experienced SEO 0.016 0.258 0.061 -0.675* 0.130 0.16
Issuers - Inexperienced (0.126) (0.339) (0.365) (0.390) (0.573)
SEO Issuers
Experienced Nonissuers -0.030 -0.765 1.207 -1.553 0.996 0.15
- Inexperienced (0.041) (1.106) (1.191) (1.273) (0.893)
Nonissuers
88
The regression of all FRC issuers shows a negative monthly alpha of -2.48 percent, which is
economically a large amount, but weak statistically at only the ten percent level. The weaker
significance contrasts with Hillion and Vermaelen (2004)'s stronger statistical finding of negative future abnormal
returns. Two differences between this study and theirs are the sample periods and the methodologies. First, Hillion
and Veramaelen (2004) use FRCs issued during the 1990s, and I use FRCs issued over 2001-2002. As the market had
more experience with FRCs, the firms may have become more careful with the terms to which they agreed, thus
reducing the likelihood that their stocks would enter "death spirals." Second, Hillion and Vermaelen (2004) use buy
and hold abnormal returns instead of computing calendar time alphas. This difference may be symptomatic of the
problems with the buy and hold return approach cited by Barber and Lyon (1997, 1999).
A portfolio with a long position in FRC issuers with CFOs who have prior experience issuing stock
publically and a short position in FRC issuers without CFOs with this experience show a more significant statistical
result. This "experienced minus inexperienced" FRC portfolio has a positive monthly alpha of 9.79 percent at the five
percent level of significance (Mitchell and Stafford t- statistic of 2.53). This striking difference among FRC issuing
firms provides support to the FCD hypothesis. The significantly greater returns FRC issuers with experienced CFOs
does not indicate these firms are struggling as the FLR hypothesis predicts. Rather, the non-negative returns to these
firms support the OSD hypothesis, that the FRC terms are appropriate for the firm and are optimal for existing
shareholders.
The same analysis performed on the three groups of control firms provides robust support to the
above findings. The portfolios long in FRC issues and short in either the fixed PIPE or SEO issuers both show
insignificant alphas, which does not support the FCD and FLR hypotheses, and does support the OSD hypothesis.
The portfolio long in FRC issuers and short in matched non-issuers, however, does show a negative alpha close to the
five percent significance level, which does not support the FLR hypothesis because the firms that do not obtain
additional equity financing perform better overall.
Portfolios long in experienced and short in inexperienced CFOs per control group display no
significant alphas. Therefore, the significant alpha of the FRC experienced minus inexperienced portfolio is all the
more compelling.
VI. Conclusion
This research adds to the body of knowledge on the motives for FRC issues by characterizing the types of issuers at
the managerial level. FRC issuing firms that have CFOs without prior experience
89
in making a public offering of equity have significantly poorer stock returns than FRC issuing firms
with CFOs who have prior experience making public offerings. The difference in stock returns
suggests that the experience of the CFO indicates which FRC deals will not maximize shareholder value. While some
firms may be taken advantage of when issuing FRCs, the FRC contract may be a rational security that is appropriate
for some firms by resolving the problems posed by classical corporate finance such as the trade-off between the debt
tax shield and bankruptcy costs and the overhang of debt on investment decisions.
I also find an interesting relationship between CFO compensation relative to the CEO and both the
decision to issue a FRC and the market's reaction to issuance announcements. Firms with higher CFO to CEO
compensation ratios are more likely to issue FRCs. Among FRC issue announcements, the market reacts negatively
when CFOs are more highly compensated. This finding merits further study on the relationship between CEO and
CFO with regard to financing decisions.
The OSD hypothesis is introduced to explain FRCs. Further research could formalize it and allow
for testing of how FRCs resolve issues that reduce firm value.
90
Appendix
FRC Issuers with CFOs with prior public issue experience
COMPUTER MOTION INC - GORDON L. ROGERS
GORDON L. ROGERS joined the Company as Vice President / Chief Financial Officer in March
2000. From 1999 to 2000, Mr. Rogers served as Vice President of Finance at ViroLogic, Inc. a medical biotechnology
company. Previously, he spent five years at Nellcor Puritan Bennett, Inc., one of the world's largest medical device
manufacturers, most recently as Controller for Worldwide Field Operations.
CRAY INC - KENNETH L. JOHNSON
Kenneth W. Johnson serves as Vice President - Legal, General Counsel and Secretary and has held those positions
since joining us in September 1997. From September 1997 to December 2001 he also served as our Vice President -
Finance and Chief Financial Officer. Prior to joining us, Mr. Johnson practiced law in Seattle for twenty years with
Stoel Rives LLP and predecessor firms, where his practice emphasized corporate finance. Mr. Johnson received an
A.B. degree from Stanford University and a J.D. degree from Columbia University Law School.
EXELIXIS - JOHN Y. S.ATO
Glen Y. Sato has served as the Company's Chief Financial Officer, Vice President of Legal Affairs and Secretary
since November 1999. From April 1999 to November 1999, Mr. Sato served as Vice President, Legal and General
Counsel for Protein Design Labs, Inc., a biotechnology company, where he previously served as the Associate General
Counsel and Director of Corporate Planning from July 1993 to April 1999. Mr. Sato holds a B.A. from Wesleyan
University and a J.D. and M.B.A. from the University of California, Los Angeles.
KEY3MEDIA GROUP INC - PETER B. KNEPPER
Peter B. Knepper was hired by Ziff-Davis in March 2000 to be our Executive Vice President and
Chief Financial Officer. From 1998 to March 2000, he was a private investor and consultant providing strategic
planning and financial management services. Mr. Knepper was previously Senior Vice President and Chief Financial
Officer of Ticketmaster Group, Inc., a position he held for more than ten years, from 1988 to 1998.
NEKTAR THERAPEUTICS - BRIGID A. MAKES
Brigid A. Makes has served as Vice President of Finance and Administration and Chief Financial Officer since June
1999. Ms. Makes has also served as Assistant Secretary since January 2001. From 1998 until joining Inhale, Ms. Makes
served as Vice President, Chief Financial Officer and Treasurer for Oravax, Inc., a life sciences company. From 1992
to 1998, Ms. Makes served in various management positions for Haemonetics Corporation, a developer of automated
blood processing systems, including, from 1995 to 1998, Vice President Finance, Chief Financial Officer and
Treasurer. Prior to Haemonetics Corporation, Ms. Makes held a number of financial management positions at Lotus
Development Corp. (now International Business Machines) and General Electric Co. Ms. Makes holds a Bachelor of
Commerce degree from McGill University in Finance and
International Business and an MBA from Bentley College.
RENTECH INC - JAMES P. SAMUELS
Mr. James P. Samuels, age 55, has served as Vice President-Finance, Treasurer and Chief Financial Officer of
Rentech since May 1, 1996. He has executive experience in general corporate management, finance, sales and
marketing, information technologies, and consulting for both large companies and
91
development stage businesses. From December 1995 through April 1998, he provided consulting
services in finance and securities law compliance to Telepad Corporation, Herndon, Virginia, a company engaged in
systems solutions for field force computing. From 1991 through August 1995, Mr. Samuels served as chief financial
officer, vice president-finance, treasurer and director of Top Source, Inc., Palm Beach Gardens, Florida, a
development stage company engaged in developing and commercializing state-of-the-art technologies for the
transportation, industrial and petrochemical markets. During that empl oyment, he also served during 1994 and 1995
as president of a subsidiary of Top Source, Inc. From 1989 to 1991, he was vice president and general manager of the
Automotive group of BML Corporation, Mississauga, Ontario, a privately-held company engaged in auto rentals, car
leasing, and automotive insurance. From 1983 through 1989, Mr. Samuels was employed by Purolator Products
Corporation, a large manufacturer and distributor of automotive parts. He was president of the Mississauga, Ontario
branch from 1985 to 1989; a director of marketing from 1984 to 1985; and director of business development and
planning during 1983 for the Rahway, New Jersey filter division headquarters of Purolator Products Corporation. From
1975 to 1983, he was employed by Bendix Automotive Group, Southfield, Michigan, a manufacturer of automotive
filters, electronics and brakes. He served in various capacities, including group director for management consulting
services on the corporate staff, director of market research and planning, manager of financial analysis and planning,
and plant controller at its Fram Autolite division. From 1973 to 1974, he was employed by Bowmar Ali, Inc., Acton,
Massachusetts, in various marketing and financial positions, and in 1974 he was managing director of its division in
Wiesbaden, Germany. He received a Bachelor's degree in Business Administration from Lowell Technological
Institute in 1970, and a Master of Business Administration degree in 1972 from Suffolk University, Boston,
Massachusetts. He completed an executive program in strategic market management through Harvard University in
Switzerland in 1984.
STAR TELECOMMUNICATIONS INC - KELLY D. ENOS
KELLY D. ENOS has served as our Chief Financial Officer since December 1996 and as Treasurer and Assistant
Secretary since April 1997. Prior to that time, Ms. Enos was an independent consultant in the merchant banking
field from February 1996 to November 1996 and a Vice President of Fortune Financial, a merchant banking firm,
from April 1995 to January 1996. Ms. Enos served as a Vice President of Oppenheimer & Co., Inc., an investment
bank, from July 1994 to March 1995 and a Vice President of Sutro & Co., an investment bank, from January 1991 to
June 1994.
TARGETTED GENETICS CORP - TODD E. SIMPSON
Todd E. Simpson has served as vice president, finance and administration, chief financial officer, treasurer and
secretary of Targeted Genetics since October 2001. From January 1996 to October 2001, Mr. Simpson served as vice
president, finance and administration and chief financial officer of Aastrom Biosciences, Inc., a public life science
company focused on the development of cell-based therapeutics. From August 1995 to December 1995, he served as
treasurer of Integra LifeSciences Corporation, a public biotechnology company, which acquired Telios
Pharmaceuticals, Inc. in August 1995. From 1992 until its acquisition by Integra, he served as vice president of
finance and chief financial officer of Telios and in various other finance-related positions. From 1983 to 1992, Mr.
Simpson practiced public accounting with the firm of Ernst & Young LLP. Mr. Simpson is a certified public
accountant. He received his B.S. in accounting and computer science from Oregon State University.
TIVO INC - DAVID COURTNEY
David Courtney was appointed by our Board to serve as a director in May 2002. Mr. Courtney joined TiVo in March
1999 as Vice President and Chief Financial Officer and in March 2000 was named Senior Vice President for Finance
and Administration. Mr. Courtney is currently Chief Financial Officer and Executive Vice President, Worldwide
Operations and Administration, serving in this capacity since October 2001. From May 1995 to July 1998, Mr.
Courtney served as a Managing
92
Director at J.P. Morgan, an investment banking firm, where he was responsible for building and
expanding the firm's high technology investment banking business in the United States. From 1986 to 1995, Mr.
Courtney was a member of the high technology investment banking group at Goldman, Sachs & Co., most recently
serving as Vice President. Mr. Courtney currently serves as a director of KQED Television, a non-profit affiliate of
the Public Broadcasting System in San Francisco, California. Mr. Courtney holds a B.A. degree in Economics from
Dartmouth College and an M.B.A. degree from Stanford University.
VIASOURCE COMMUNICATIONS INC. DOUGLAS J. BETLACH
DOUGLAS J. BETLACH has been our Executive Vice President, Chief Financial Officer, Treasurer and Secretary
since June 1999. Prior to joining Viasource, Mr. Betlach was Vice President, Chief Financial Officer and Treasurer of
Dycom Industries, Inc., a nationwide provider of engineering, construction and maintenance services to
telecommunications operators.
V-ONE CORP - MARGARET E. GRAYSON
MARGARET E. GRAYSON (54) was elected President and CEO in November 2000. She had served as the Company's
Senior Vice President and Chief Financial Officer since May 1999. Ms. Grayson was elected to the Board of
Directors in August 1999. Prior to joining V-ONE Corporation, Ms. Grayson served as Vice President of Finance and
Administration and Chief Financial Officer for SPACEHAB, Inc. (Nasdaq: SPAB) from September 1994 to October
1998. Immediately prior to joining SPAB, Ms. Grayson served as Chief Financial Officer for CD Radio, Inc. in
Washington, D.C., an early entrant in the satellite radio mobile communications market. Previously, Ms. Grayson
served as a senior executive and consultant to high technology start-up companies. Ms. Grayson is on the Board of
Directors of Ronbotics Corporation and the Advisory Board of Celsion Corporation. Ms. Grayson holds an M.B.A.
from the University of South Florida and a B.S. in Accounting from the State University of New York at Buffalo.
FRC Issuers with CFOs without public issue experience
ADEPT TECHNOLOGY - MICHAEL W. OVERBY
Michael W. Overby has served as Adept's Vice President of Finance and Chief Financial Officer since
March 2000. From December 1999 to March 2000, Mr. Overby held the position of Corporate Controller at Adept.
Prior to joining Adept, Mr. Overby was the financial executive for DG Systems, a leading provider of digital
distribution services to the broadcast advertising industry. From 1996 to 1998 he was Corporate Controller and
Director of Information Systems at Inprise Corporation, formerly Borland, a public software company. Mr. Overby
holds a B.S. in Business Administration from California Polytechnic State University.
ALKERMES INC. - JAMES M. FRATES
Mr. Frates has been Vice President, Chief Financial Officer and Treasurer of Alkermes since July 1998. From June
1996 to July 1998, he was employed at Robertson, Stephens & Company, most recently as a Vice President in
Investment Banking. Prior to that time he was employed at Robertson, Stephens & Company and at Morgan Stanley
& Co. In June 1996, he obtained his M.B.A. from Harvard University.
ALLIANCE PHARMACEUTICALS - TIM T. HART
TIM T. HART, C.P.A. Mr. Hart, who is 44, was appointed Vice President in May 1999 and Chief Financial Officer in
August 1998. He joined the Company in 1991 as Controller and has also served as Treasurer since 1994. Prior to
joining Alliance in 1991, he was employed in various financial management positions at Cubic Corporation for over
eight years. He was also employed by Ernst &Whinney in San Diego, California as a C.P.A.
93
ANTEX BIOLOGICS - GREGORY C. ZAKARIAN
GREGORY C. ZAKARIAN, CPA, age 52, has served as Vice President, Finance and Administration, Chief Financial
Officer and Treasurer of the Company since September 1992. He has served as Secretary of the Company since
November 1993, and as Assistant Secretary of the Company from September 1992 until October 1993. Prior to
September 1992, Mr. Zakarian was a partner with an international CPA firm.
APPIANT TECHNOLOGIES - DOUGLAS S. ZORN
DOUGLAS S. ZORN. Mr. Zorn has been our Chairman of the Board, Chief Executive Officer and
President since May 2000. Mr. Zorn served as Executive Vice President, Chief Financial Officer, Secretary and a
Director of the Company since our incorporation in October 1996 until May 2000. Mr. Zorn served as Executive Vice
President, Secretary and Treasurer, and Chief Financial and Operating Officer of BioFactors, Inc. from December
1993 until February 1997 and as a Director from June 1994 until February 1997.
CECO ENVIRONMENTAL CORP - MARSHALL J. MORRIS
Marshall J. Morris became the Chief Financial Officer of the Company on January 26, 2000. From
1996 to 1999 Mr. Morris was Treasurer of Calgon Carbon Corporation which stock trades on the New York Stock
Exchange and which is a worldwide producer of specialty chemicals and supplier of pollution control technologies
and services with annual sales of approximately $300 million. From 1995 to 1996 he served as a consultant with
respect to business management and strategic planning. From 1989 through 1995 Mr. Morris also served as the
Treasurer of Trico Products Corporation, an international manufacturer and distributor of original equipment
automative parts with annual sales of approximately $350 million.
CEL-SCI CORP - GEERT KERSTEN
Geert R. Kersten, Esq. Mr. Kersten was Director of Corporate and Investment Relations for the Company between
February 1987 and October 1987. In October of 1987, he was appointed Vice President of Operations. In December
1988, Mr. Kersten was appointed Director of the Company. Mr. Kersten also became the Company's Treasurer in
1989. In May 1992, Mr. Kersten was appointed Chief Operating Officer and in February 1995, Mr. Kersten became
the Company's Chief Executive Officer. In previous years, Mr. Kersten worked as a financial analyst with Source
Capital, Ltd., an investment advising firm in McLean, Virginia. Mr. Kersten is a stepson of Maximilian de Clara, who
is the President and a Director of the Company. Mr. Kersten at tended George Washington University in
Washington, D.C. where he earned a B.A. in Accounting and an M.B.A. with emphasis on International Finance. He
also attended law school at American University in Washington, D.C. where he received a Juris Doctor degree.
CHAMPION ENTERPRISES - ANTHONY C. CLEBURG
In 2000 Mr. Cleberg joined Champion from Washington Group International ("Washington Group"),
a publicly-held engineering and construction firm, where for the previous three years he was the Executive Vice
President and Chief Financial Officer. On May 14, 2001, subsequent to Mr. Cleberg's departure from Washington
Group, it filed a voluntary petition for bankruptcy under Chapter 11 of the U.S. Bankruptcy Code. On January 25,
2002, Washington Group completed its Plan of Reorganization and emerged from Chapter 11 bankruptcy protection.
Previous to Washington Group, Mr. Cleberg worked for Honeywell Inc. for 15 years in various senior financial
positions, leaving as
Corporate Vice President, Business Development.
CHELL GROUP - DON PAGNUTTI
Don Pagnutti was appointed our Vice President, Finance on September 19, 2000. Mr. Pagnutti has been our Chief
Financial Officer since September 1998, and was our Executive Vice President and Chief Operating Officer from
September 1997 to September 2000. From 1996 to 1997, he worked for Sullivan Entertainment Inc., as Executive Vice
President and Chief Financial Officer. From 1980 to
94
1996, he worked for Telemedia Communications Ltd., a large Canadian media company as Vice
President, Radio. Mr. Pagnutti is a Chartered Accountant and has a Masters Degree in Business Administration and a
Bachelor of Commerce Degree from the University of Toronto.
CLEAN HARBORS INC - ROGER A. KOENECKE
Roger A. Koenecke joined the Company as Senior Vice President and Chief Financial Officer in 1998.
From 1982 through 1997, Mr. Koenecke held a variety of management positions, including Senior Vice President and
Chief Financial Officer, with Millbrook Distribution Services, Inc. and its predecessor corporations, which are
engaged in the distribution of health and beauty care, general merchandise, and specialty food products. Prior to that,
he was an Audit Manager with Price Waterhouse & Co., an international accounting firm. Mr. Koenecke holds a BS
in Chemistry and MBA from the University of Wisconsin.
CYBERCARE - PAUL PERSHES
PAUL C. PERSHES (age 57) Class II, has served as a director since August 1996 and as our president since May
1997. In March 2001, Mr. Pershes assumed responsibilities as acting chief financial officer. Before joining us, Mr.
Pershes founded and served as an officer of Weinberg, Pershes & Company, P.A., an accounting firm, from July
1994 to May 1997. Before founding Weinberg, Pershes & Company, Mr. Pershes was a senior partner of the
international accounting firm Laventhol and Horvath for 18 years.
DALEEN TECHNOLOGIES - STEVEN M. WAGMAN
STEPHEN M. WAGMAN, 40, has served as chief financial officer of Daleen since June 2000 and has
served as an executive vice president of corporate development and secretary since June 1999. Mr. Wagman has over
12 years of finance, business and legal experience with high-growth software companies. Before joining Daleen, Mr.
Wagman served in various capacities with PowerCerv Corporation, an enterprise resource planning software
company, including Chief financial officer, treasurer, senior vice president of administration, general counsel and
secretary.
DATATEC SYSTEMS INC - ISAAC J. GAON
Isaac J. Gaon, Chairman of the Board since December 1997 and Director since 1992, has served as the Chief
Executive Officer since October 1994. He served as Chief Financial Officer from April 1992 until October 1994.
From September 1987 to December 1991, Mr. Gaon, a chartered accountant, served as President and Chief
Executive Officer of Toronto-based NRG, Inc., (a subsidiary of Gestetner International) an office equipment
supplier, and in several senior management roles within Gestetner Canada and Gestetner USA.
95
DIGITAL RECORDERS - LAWRENCE A. TAYLOR
Lawrence A. Taylor, age 55, has 12 years' experience in the transit industry, as well as extensive
knowledge and experience in auditing, merger and acquisition reporting, analysis and financial information-
technology systems. He has been the Company's secretary, chief financial officer and vice president since May 1998.
From March 1997 to June 1999, Mr. Taylor was a partner in the Dallas office of Tatum CFO, LLP, a professional
partnership of career CFOs. From March 1995 to August 1996, he was senior vice president of Precept Business
Products, Inc., a privately held holding company in Dallas that distributed business forms, construction and on-
demand courier services. From May 1991 to December 1994, he was vice president and group controller of Dallas -
based Mark IV Industries' Transportation Products Group, which included nine companies, subsidiaries and
operating units serving transit and transportation markets worldwide. Prior to 1991, he served in various financial
managerial capacities in the food processing, commercial construction and oil field supply industries, as well as other
manufacturing environments. A 1970 graduate of Wayne State University in Detroit, Mich., Mr. Taylor earned a
B.S. degree in Accounting. A Certified Public Accountant, he is a member of the Texas Society of CPAs and its
Dallas Chapter, the American Institute of CPAs, and Financial Executive International.
DYNTECK INC - JAMES A LINESCH
Since August 14, 2000, Mr. Linesch has served as the Chief Financial and Chief Accounting Officer,
Executive Vice President and Secretary, and since February 1997 Director, of TekInsight. Previously, Mr. Linesch
was the President, Chief Executive Officer and Chief Financial Officer of CompuMed, a public computer company
involved with computer assisted diagnosis of medical conditions, which he joined in April 1996 as Vice President and
Chief Financial Officer. Mr. Linesch served as a Vice President, Chief Financial Officer of the Company from August
1991 to April 1996. From May 1998 to August 1991, Mr. Linesch served as the Chief Financial Officer of Science
Dynamics Corp., a corporation involved in the development of computer Software. Mr. Linesch holds a CPA
certification in the State of California, where he practiced with Price Waterhouse from 1981 to 1984.
ECHOSTAR COMMUNICATIONS CORP - MICHAEL R. MCDONNEL
Mr. McDonnell joined EchoStar in August 2000 as Chief Financial Officer. Mr. McDonnell is responsible for all
accounting and finance functions of the Company. Prior to joining EchoStar, Mr. McDonnell was a Partner with
PricewaterhouseCoopers LLP, serving on engagements for companies in the technology and information
communications industries.
ELECTRIC CITY CORP - JEFFERY R. MISTARZ
Jeffrey R. Mistarz has been our chief financial officer since January 2000 and our treasurer since October 2000. From
January 1994 until joining us, Mr. Mistarz served as chief financial officer for Nucon Corporation, a privately held
manufacturer of material handling products and systems, responsible for all areas of finance and accounting,
managing capital and shareholder relations. Prior to joining Nucon, Mr. Mistarz was with First Chicago Corporation
(now Bank One Corporation) for 12 years where he held several positions in corporate lending, investment banking
and credit
strategy.
ELECTROGLAS INC - THOMAS E. BRUNTON
Thomas E. Brunton was appointed Vice President — Finance, Chief Financial Officer, Treasurer and Secretary of the
Company in November 2000. Prior to joining the Company, Mr. Brunton was Chief Financial Officer of Centigram
Communications from March 1998 to July 2000. He joined Centigram in March 1991 as Controller and also served as
Treasurer. Prior to his service at Centigram, he had financial management responsibilities at 3Com, Sun Microsystems,
and IBM/ Rolm.
96
EUROTECH LTD - JOHN W. DOWIE
JON W. DOWIE, 54, IS OUR VICE PRESIDENT, TREASURER AND CHIEF FINANCIAL OFFICER under an
employment agreement that expired February 6, 2001. He joined the Company in February 2000 after serving as Vice
President, Finance, and CFO for Research Planning, Inc., from September 1997. Prior to that, he served as Controller
for Automation Research Systems Ltd. from August 1992. He is a Certified Public Accountant and a Certified
Government Financial Manager. He holds a B.S. in Accounting and an MBA from Murray State University, and is a
Doctor of Business Administration candidate in Information Systems, Finance, and Marketing at Mississippi State
University.
GLOBAL TECHNOVATIONS - DAVID NATAN
David Natan - was appointed a director of the Company on April 16, 1998 in order to fill a vacancy. Currently, Mr.
Natan, a CPA, has been Vice President and Chief Financial Officer of the Company since June 1995 and Secretary
from August 1997. Mr. Natan previously served on the Company's Board from June 1995 to January 1997. Mr. Natan
brings nearly 20 years of management and analytical experience to his responsibilities. Prior to joining the Company,
from November 1992 through June 1995, Mr. Natan was Chief Financial Officer of MBf USA, Inc., which is a
Nasdaq listed subsidiary of MBf Holdings Berhad, a multi-national conglomerate. From August 1987 through
October 1992, Mr. Natan was Treasurer and Controller for Jewelmasters, Inc., an AMEX listed company.
HORIZON MEDICAL PRODUCTS INC - JULIE F. LANCASTER
Julie F. Lancaster has served as the Vice President — Finance since January 2001. Ms. Lancaster joined the
Company in 1994 as Assistant Controller and served in that capacity until 1996. From August 1996 through August
2000, Ms. Lancaster served as Controller of the Company. From August 2000 to January 2001, Ms. Lancaster served
as Director of Financial Reporting and Planning for the Company.
HYPERTENSION DIAGNOSTICS - JAMES S. MURPHY
James S. Murphy Mr. Murphy joined us as Vice President of Finance and Chief Financial Officer during May 1996.
In March 2000, his title was changed to Senior Vice President, Finance and Administration and Chief Financial
Officer. Mr. Murphy was Controller of Gaming Corporation of America from December 1992 through November
1995. From 1978 to 1988, he was a tax partner with Fox, McCue and Murphy, a certified public accounting firm
located in Eden Prairie, Minnesota. From 1970 to 1978, Mr. Murphy was employed by Ernst & Ernst (currently
named Ernst &Young LLP) with both audit (six years) and tax (two years) experience. Mr. Murphy is a member of
the American Institute of Certified Public Accountants as well as the Minnesota Society of CPAs. He holds a Bachelor
of Science degree from Saint John's University in Collegeville, Minnesota (1966) and a Master of Business
Administration degree (M.B.A.) from the University of Minnesota (1968).
INTERNATIONAL FIBERCOM INC - GREGORY B. HILL
Mr. Hill served as our Controller from September 1999 to March 2000 and became our Vice President -Finance in
April 2000. From June 1998 until June 1999 he was employed by All Star Telecom, an infrastructure development
subsidiary that we acquired in April 1999, where he served as chief financial officer and controller. From June to
September 1999, he served as Regional Controller of our Infrastructure Development Group. Mr. Hill is a certified
public accountant and served in the Technology Industry Group of Price Waterhouse providing audit, transaction
support, and business advisory services to technology companies from January 1992 through June 1998. He received
his bachelor of science in business administration from California State University Sacramento.
97
MEDWAVE INC - MARK T. BAKKO
MARK T. BAKKO is the Chief Financial Officer of the Company. He has served in this position since February 1996.
From 1984 to 1996, Mr. Bakko was with Deloitte & Touche LLP with his most recent position being a senior manager.
Mr. Bakko has been a Certified Public Accountant since 1985 in the State of Minnesota. Mr. Bakko holds a Masters of
Business Taxation and B.S.B.A. degree in Accounting from the University of Minnesota.
ONE VOICE TECHNOGOLOGIES - RAHOUL SHARAN
Rahoul Sharan holds a Bachelor of Commerce degree from the University of British Columbia and is a member of the
Institute of Chartered Accountants of British Columbia. Mr. Sharan was employed by Coopers & Lybrand (now
Pricewaterhouse Coopers) from 1984 to 1989. Since 1989, Mr. Sharan has been the President and a Director of KJN
Management Ltd., a private company that provides a broad range of administrative, management and financial
services to both private and public companies. Mr. Sharan has been a partner in S & P Group, a company that
specializes in investment financing for venture capital projects and real estate development and construction, since
1988. Mr. Sharan was also a Treasurer and Director.
ONSTREAM MEDIAN CORP - GAIL L. BABITT
GAIL BABITT, CPA. Ms. Babitt joined VDC as Chief Financial Officer in November 2000. From
1999 through October 2000 Ms. Babitt served as Vice President of Finance, North America and Corporate Controller
for TeleComputing ASA. TeleComputing ASA is a leading application service provider. From 1997 to 1999 Ms.
Babitt served as Manager-Transaction Services for Price Waterhouse Coopers LLP. During 1997 Ms. Babitt served
as Director of Finance for ToppTelecom, Inc. Topp Telecom is a prepaid cellular company based in Miami. From
1994 to 1997 Ms. Babitt worked in the audit group with Price Waterhouse Coopers LLP (formerly Price Waterhouse
LLP) and with Ernst & Young LLP from 1992 to 1994. Ms. Babitt has received a MBA from Boston University and
a B.S. from Nova Southeastern University.
P-COM INC - LEIGHTON J. STEPHENSON
Mr. Stephenson has served as Vice President and Chief Financial Officer since September 2000.
From 1993 to 2000 he served as Chief Financial Officer, Treasurer, and Secretary of Vallen Corporation, a Texas
company engaged in manufacturing and distribution of industrial safety products and services.
PENTON MEDIA INC. - JOSEPH A. NECASTRO
Joseph G. NeCastro, 44, Chief Financial Officer and Treasurer of Penton since June 1998. Before
joining Penton, Mr. NeCastro spent five years with Reader's Digest Association, Inc. Mr. NeCastro was Vice
President, Finance for Reader's Digest USA from 1995 until 1998 and Corporate Controller in 1994 and 1995.
RAMP CORP - GARY L. SMITH
Mr. Smith joined the Company as Executive Vice President and Chief Financial Officer in
December of 2000. From 1995 to 2000, Mr. Smith was with Provident Group, a financial advisory firm serving
companies operating in emerging market countries, where he was a principal. Previously, Mr. Smith was an
executive of American Express Bank, the international banking arm of the financial services conglomerate American
Express Corporation (NYSE: AXP), where he held various senior financial positions, most recently as Senior Director
and Commercial Banking Head, London Branch. He holds a BSc degree in Economics from the Wharton School and
an MSc in Accounting and Finance from the London School of Economics.
SCIENCE DYNAMICS CORP - ROBERT O'CONNOR
98
Robert O'Connor came to SciDyn from PricewaterhouseCoopers, L.L.P in Philadelphia, PA, where he
served as a manager of middle market advisory services. Mr. O'Connor brings with him a strong background in
corporate finance, including prior positions as Corporate Controller and Chief Financial Officer at three technology
companies. Mr. O'Connor received his MBA from Rutgers- Graduate School of Management, BS from Kean
University in Union, NJ, and he is a Certified Public Accountant.
STARBASE CORP - DOUGLAS S. NORMAN
Douglas S. Norman founded Starbase in September 1991. In February 2000, Mr. Norman was
appointed to serve as our Chief Financial Officer. From September 1997 to February 2000, Mr. Norman served as
our Chief Accounting Officer. In February 2002, Mr. Norman was elected Secretary. Mr. Norman has served as our
Assistant Secretary since February 1997 and Director of Finance from June 1996 to February 2000. Douglas S.
Norman is the son-in-law of William R. Stow III, a member of the Board or Directors. Mr. Norman holds a B.S. in
Business Administration from California State University and an M.B.A. from Loyola Marymount.
STORAGE COMPUTER CORP - PETER N. HOOD
Peter N. Hood, 60, has been the Company's Chief Financial Officer since May 16, 2000. Mr. Hood was previously
owner and Chief Executive Officer of Phoenix Custom Molders, Inc., a custom manufacturer of plastic parts from
1993 to 2000. Phoenix Custom Molders, Inc. filed Chapter 11 bankruptcy on September 4, 1996 in the U.S.
Bankruptcy Court for the District of New Hampshire. The Chapter 11 bankruptcy was entitled "In re Phoenix Custom
Molders, Inc." and docketed as Bk. 96-12443-MWV. Phoenix Custom Molders, Inc. emerged from Chapter 11
bankruptcy on August 17, 1997. He was also co-founder and Vice President of Phoenix Distributors, Inc., a business
involved in consolidating independent distributors of industrial gas and welding supplies from 1985 to 1993. From
1965 to 1985, he was with the accounting firm of Ernst & Young, becoming a partner in 1976.
He received his business degree from Northeastern University and is a certified public accountant.
TEAM COMMUNICATIONS GROUP - JAY J. SHAPIRO
Jay J. Shapiro became our President, Chief Operating Officer and acting Chief Financial Officer on March 16, 2001.
Mr. Shapiro will assist us in overseeing our corporate, financial and fiduciary activities worldwide. From 1993 to
2000, Mr. Shapiro, a certified public accountant, operated a private accounting and consulting practice specializing in
servicing the television industry. During such period, he served as a temporary corporate officer for several publicly
traded entertainment companies. Mr. Shapiro received his B.B.A. from the University of Wisconsin and a MBA (with
Distinction) in Accounting and Finance from Arizona State University Graduate School of Business Administration.
THINKPATH INC - KELLY HANKINSON
Kelly Hankinson has served as our Chief Financial Officer since May 1999 and as a Director since June 2000. Ms.
Hankinson served as our Controller from February 1994 to May 1999. Ms. Hankinson has a Masters Degree and a
Bachelors Degree from York University.
US PLASTIC LUMBER CORP - MICHAEL D. SCHMIDT
MICHAEL SCHMIDT is Treasurer and Vice President of Finance. Mr. Schmidt joined us in December 1997. Mr.
Schmidt has over 20 years of public and private accounting experience including
ten years in the environmental industry. Prior to joining us, Mr. Schmidt served as Chief
Financial Officer of Republic Environmental Systems, Inc., a publicly traded company and a leading environmental
service provider, headquartered in Blue Bell, Pennsylvania, a position he held for approximately ten years. Mr.
Schmidt has a B.S. degree in Business Administration from Rowan University and is a Certified Public Accountant in
the State of New Jersey.
VELOCITY EXPRESS CORP - MARK E. TIES
99
Mark E. Ties. Mr. Ties joined the Company in April 2000 as its Vice President of Finance. Mr. Ties is
also the Vice President of Finance for Velocity Express. Mr. Ties has more than 13 years of financial experience, of
which eight years have been at the executive level in a number of companies in varied industries. Since 1998 and prior
to joining the Company, Mr. Ties was a Manager and Senior Manager for Ernst & Young LLP in its entrepreneurial
services and mergers and acquisitions departments. From 1994 to 1998 Mr. Ties was the Chief Financial Officer of
Progressive Beauty Enterprises, Inc., a regional distribution company. Prior to 1994 Mr. Ties was the corporate
controller of MEI Salons, Inc. and prior to that he was a senior auditor for Coopers & Libran LLP. Mr. Ties is a
Certified Public Accountant.
WAVERIDER COMMUNICATIONS INC. - T. SCOTT WORTHINGTON
T. Scott Worthington has been a Vice President and the Company's chief financial officer since
January 1998. From 1988 to 1996, he worked at Dell Computer Corporation, in Canada, where he held numerous
positions including CFO of the Canadian subsidiary. From October 1996 to January 1998, he was a financial and
business consultant. Mr. Worthington is a Chartered Accountant.
100
References
Aier, J.K., J. Comprix, M.T. Gunlock, and D. Lee, 2005, The Financial Expertise of CFOs and
Accounting Restatements, Accounting Horizons, 19, 123-135.
Baker, M., L. Litov, J. Wachter, J., 2001, Can Mutual Fund Managers Pick Stocks?: Evidence from
Their Trades Prior to Earnings Announcements, Working Paper, New York University.
Barber, B. and J. Lyon, 1997, Detecting Long-Run Abnormal Stock Returns: The Empirical Power
and Specification of Test Statistics, Journal of Financial Economics 43, 341-372.
Barber, B. J. Lyon, and C. Tsai, 1999, Improved Methods for Tests of Long-Run Abnormal Stock
Returns, Journal of Finance 54, 165-201.
Barclay, M., C. Holderness, and D. Sheehan, 2003, Private Placements and Managerial
Entrenchment, William E. Simon Graduate School of Business, University of Rochester working paper.
Brophy D., Oiumet P., and Sialm C., 2006, Hedge Funds as Investors of Last Resort? Review of
Financial Studies
Chava, S. and A. Purnanandam, 2007, Determinants of the Fixed-to-Floating Rate Debt Structure of
Firms, Journal of Financial Economics 85, 755 - 786.
Chaplinsky, S. and D. Haushalter, 2003, Financing under Extreme Uncertainty: Evidence from
PIPEs, Darden Graduate School of Business Administration, University of Virginia working paper.
Chen, L., N. Jegadeesh, and R. Wermers, 2000, An Examination of Stockholdings and Trades of
Mutual Fund Managers, Journal of Financial and Quantitative Analysis 35, 343-368.
Chemmanur, T. and P. Fulghieri, 1997, Why Include Warrants in New Equity Issues? A Theory of
Unit IPOs, Journal of Financial and Quantitative Analysis 32, 1-24.
Chevalier, J., G. Ellison, 1999, Are Some Mutual Fund Managers Better than Others?, Journal of
Finance 54, 875-899.
101
Dai, N., 2007, Does Investor Identity Matter? An Empirical Examination of Investments by Venture Capital Funds and
Hedge Funds in PIPEs, Journal of Corporate Finance 13, 538-563.
Deja, R. and S. Waba, 1998, Propensity Score Matching Methods for Non-Experimental Studies,
NBER Working Paper, #6829.
Garner, J. and B. Marshall, 2005, Unit IPOs: What the Warrant Characteristics Reveal about the
Issuing Firm, Journal of Business, 78, 1837-1858.
Gomes A. and G. Phillips, 2005, Why Do Firms Issue Private and Public Equity, Convertibles and
Debt?
Gottesman, A. and M. Morey, 2006, Manager Education and Mutual Fund Performance, Journal of
Empirical Finance 13, 145-182.
Harlow, W. and K. Brown, 2007, The Right Answer to the Wrong Question: Identifying Superior
Active Portfolio Management, Journal of Investment Management 4, 1-26.
Hertzel, M., M. Lemmon, J. Linck, and L. Rees, 2002, Long-Run Performance Following Private
Placements of Equity, Journal of Finance 57, 2595-2617.
Hertzel, M. and R. L. Smith, 1993, Market Discounts and Shareholder Gains for Placing Equity
Privately, Journal of Finance 48, 459-485.
Hillion, P., and T. Vermaelen, 2004, Death Spiral Convertibles, Journal of Financial Economics 71,
381-415.
Hoitash, U. and R. Hoitash, K. Johnstone, 2009, Internal Control Material Weaknesses, CFO
Compensation, and the Moderating Effects of Corporate Governance, External Monitoring, and Firm Risk, Working
Paper SSRN: http://ssrn.com/abstract=1019359.
Holmstrom, B., 1979, Moral Hazard and Observability, Bell Journal of Economics 10, 74-91.
Krishnamurthy, S., P. Spindt, V. Subramaniam, and T. Woidke, 2004, Does Investor Identity Matter
in Equity Issues? Evidence from Private Placements. Journal of Financial Intermediation, forthcoming.
102
Lee, M., P. Lee, and S. Taylor, 2000, Unit Initial Public Offerings: Staged Equity or Signaling
Mechanism? Working Paper.
Leuz, C., A. Triantis, and T. Wang, 2006, Why Do Firms Go Dark? Causes and Economic
Consequences of Voluntary SEC Deregistrations, Working Paper No. RHS-06-045.
Mian, S., 2001, On the Choice and Replacement of Chief Financial Officers, Journal of Financial
Economics 60, 143-175.
Myers, S. and N. Majluf, 1984, Corporate Financing and Investment Decisions When Firms Have
Information that Investors Do not Have, Journal of Financial Economics 13, 187-222.
Ohlson, J., 1980, Financial Ratios and the Probablistic Prediction of Bankruptcy, Journal of
Accounting Research 18, 109-131.
Rajan, R., 1992, Insiders and Outsiders: The Choice between Informed and Arm's-Length Debt,
Journal of Finance 47, 1367-1400.
Schultz, P. Unit Initial Public Offerings: A Form of Staged Financing, Journal of Financial
Economics, 34, 199-229.
Wermers, R., 2000, Mutual Fund Performance: An Empirical Decomposition into Stock-Picking
Talent, Style, Transaction Costs, and Expenses, Journal of Finance 55, 1655-1695.
Wruck, K., 1989, Equity Ownership Concentration and Firm Value: Evidence from Private Equity
Financings, Journal of Financial Economics
103
doc_695043436.docx