FINANCIAL SYSTEM ARCHITECTURE AND THE

CO-EVOLUTION OF BANKS AND CAPITAL MARKETS

Fenghua Songand Anjan V. Thakor
December 19, 2008

Abstract
We study how financial system architecture evolves through the development of banks and financial
markets. The predominant existing view is that banks and markets compete, which often contradicts
actual patterns of development. We show that banks and markets exhibit three forms of interaction:
they compete, they complement each other, and they co-evolve. The co-evolution loop is generated by
two elements missing in previous analyses of financial system architecture: securitization and bank equity capital. As banks evolve via improvements in credit screening, they securitize higher-quality credits
in the capital market. This encourages greater investor participation and spurs capital market evolution.
And, if capital market evolution is spurred by exogenous shocks that cause more investors to participate
in the market, banks find it cheaper to raise equity capital to satisfy endogenously-arising risk-sensitive
capital requirements. This enables banks to serve previously-unserved high-risk borrowers, expanding
banking scope and spurring bank evolution. Numerous additional results and empirical predictions are
drawn out, and the implications of the analysis for bank governance and regulation are discussed.
Keywords: Financial System Architecture, Bank Evolution, Capital Market Evolution
JEL Classification: G10, G21

Acknowledgements: The helpful comments of Arnoud Boot, Jin Cao, Andrew Scott, and seminar participants at Federal

Reserve Bank of New York, International Monetary Fund, the 2008 Financial Intermediation Research Society conference
(Alaska), London School of Economics, Universitat Pompeu Fabra (Barcelona), and Washington University in St. Louis are
gratefully acknowledged.

Assistant Professor of Finance, Smeal College of Business, Pennsylvania State University. University Park, PA 16802.
Email: [email protected]

John E. Simon Professor of Finance, Olin Business School, Washington University in St. Louis. Campus Box 1133, One
Brookings Drive, St. Louis, MO 63130. Email: [email protected]

Introduction

A fundamental question in comparative financial systems concerns the most efficient way to organize the
transfer of capital from savers to investors. In particular, the question is whether the emphasis should be
on markets or on banks.1 This question is important because of its potential implications for aggregate
credit extension and growth in the real sector. Even though there is strong evidence that financial system
development positively affects growth, there is less consensus on whether the effect comes from bank or
market development, or even whether the specifics of how the financial system evolves matter for the real
sector.2 Thus, there is much we do not know. In particular, we have only begun to understand how
the architecture of the financial system the relative roles of banks and financial markets affects the
functioning of the financial system, and the effect of the financial system on the real sector. There are
numerous unanswered or partially answered questions. What is the relationship between borrower credit
attributes and the borrowers choice of financing source when banks themselves are financing partly from the
capital market, a competing financing source for borrowers? Is the emergence of one sector of the financial
system (either banking or financial markets) always at the expense of the other? In particular, how does
the development of banks affect the development of the capital market, and how does the development of
the capital market affect banks? Our objective in this paper is to address these questions.
The key theoretical findings on these issues in the literature can be summarized as follows. First,
market-based financial systems are not better than bank-based financial systems; they simply behave differently. For example, market-based systems provide better cross-sectional risk sharing, whereas bank-based
systems provide better intertemporal risk sharing (Allen and Gale (1997)). Market-based systems have an
advantage over bank-based systems in committing not to refinance unprofitable projects (Dewatripont and
Maskin (1995)), and markets may also provide managers valuable information through the feedback effect
of prices (e.g., Boot and Thakor (1997a), and Subrahmanyam and Titman (1999)). Bilateral financing,
common in bank-based systems, is better at protecting borrower proprietary information and at providing
R&D incentives for firms to undertake costly project search than multilateral financing that characterizes
market-based systems (Bhattacharya and Chiesa (1995), and Yosha (1995)). Market-based systems create
stronger financial innovation incentives (Boot and Thakor (1997b)), and are better at funding innovative
projects subject to diversity of opinion (Allen and Gale (1999)), but bank-based systems resolve assetsubstitution moral hazard more effectively (Boot and Thakor (1997a)). Second, the dominant view in the
1

Financial systems have been broadly classified as being bank-based or market-based, based on the share of banks and

other intermediaries in total financing provided by the financial system. A common example of a bank-based system is
Germany where banks emerged as the dominant financing source due to relatively few restrictions on their activities. The
most commonly-cited example of a market-based system is the U.S. where Glass-Steagall restrictions on banks were at least
partly responsible for banks achieving lesser dominance. An international comparison of financial system architecture appears
in Tadesse (2002).
2
Beck and Levine (2002), Demirg
uc-Kunt and Levine (2001), and Levine (2002) show that the positive impact of financial
system development on economic growth is unaffected by whether the evolution of the financial system is due to bank or
financial market development. Deidda and Fattouh (2008) find, however, that a change from a bank-dominated system to one
with both banks and markets can hurt economic growth.

literature is that, with some exceptions that will be discussed later, banks and markets compete, implying
that the development of one is at the expense of the other (e.g., Allen and Gale (1997, 1999), Boot and
Thakor (1997a), and Dewatripont and Maskin (1995)), an observation that seems buttressed by anecdotes
such as the shrinkage of depository institutions in the U.S. in the 1980s when (market-based) mutual funds
emerged.3
The result that banks and markets compete has potentially powerful policy implications, but does
not seem entirely consistent with the findings of several empirical studies.4 Demirg
uc-Kunt and Maksimovic (1996) find that stock market development engenders a higher debt-equity ratio for firms and thus
generates more business for banks in developing countries. Sylla (1998) provides a description of the complementarity between banks and capital markets in fostering the growth of the U.S. economy from 1790 to
1840, suggesting that it is important to take a broad view of financial development and pay attention to
the manifold ways in which components of a financial system, such as banks and securities markets, can
complement and reinforce one another.
We study how banks and markets affect each other and thus how financial system architecture affects
which borrowers are financed and the source of this financing by developing a model of their interaction
within an evolving financial system. In our model, the borrower chooses its financing source from the
following menu: (i) non-intermediated, direct capital market financing in which it borrows directly from
the capital market; (ii) securitization in which it lets the bank screen and certify its creditworthiness first
and then borrows from the capital market; and (iii) a relationship loan from the bank. There are two key
frictions that impede the borrowers ability to obtain financing. One is certification, a friction that arises
from the fact that the borrower pool consists of observationally identical but heterogeneous borrowers, some
creditworthy and some not. This creates the likelihood that even a creditworthy borrower may be denied
credit, and the more severe this friction the greater the likelihood of credit denial. The other friction is
financing, which arises from the dissipative costs of external financing, which include costs related to
the fact that those seeking financing and those providing financing may value differently the surplus from
the project being financed, leading to the cost of financing rising above the first best. We show that banks
are better at diminishing the certification friction, whereas banks and markets differ in terms of how they
resolve the financing friction. Exclusive bank or market finance does well at diminishing one friction, at the
expense of not diminishing the other. As long as both frictions are relevant, technological improvements
3

There is also a growing body of empirical research on financial system architecture as well as its impact on growth (e.g.,

Beck and Levine (2002), Deidda and Fattouh (2008), Levine (2002), Levine and Zervos (1998), and Tadesse (2002)).
4
In addition to the empirical studies, we can also see that the evolution of banks and capital markets in the United States,
United Kingdom, Germany and Japan during 1960 2003 shows complementarity between banks and markets most of the
time with occasional spurts of competition. This can be seen using data from The World Bank Group and defining bank
development as Bank Credit, which is the value of loans made by commercial banks and other deposit-taking banks to the
private sector divided by GDP (Levine and Zervos (1998)), defining Stock Market Size as the value of listed domestic shares
on domestic exchanges divided by GDP, and Bond Market Size as the ratio of the total amount of outstanding domestic debt
securities issued by private or public domestic entities to GDP (Beck, Demirg
uc-Kunt and Levine (2000)). In all four countries
over this time period, one observes Bank Credit, Stock Market Size and Bond Market Size growing together except over a few
short periods.

in either bank or market finance lead to borrowers shifting toward one source of financing and away from
the other. This is the standard result in the literature that banks and markets compete.
There are two ingredients in our analysis that enable us to go beyond this standard result and generate
numerous new results. One is securitization and the other is bank capital. With securitization, the
bank provides certification and the capital market provides financing. That is, securitization creates the
possibility of letting each sector of the financial system operate where it is best. Moreover, securitization
acts as a channel through which technological improvements in the banks certification technology not
only reduce the certification friction but are also transmitted to the financial market and lead to a better
resolution of the financing friction. Since the certification and financing frictions complement each other
in impeding the borrowers access to efficient funding, banks and markets are not in competition, but
complementary to each other.
Bank capital connects banks and markets in a different way. Capital market development reduces the
financing friction for the bank and lowers its cost of equity capital, which makes it privately optimal for
the bank to raise the additional capital needed to meet the higher capital requirements associated with
riskier loans that the bank may have otherwise chosen not to make. Thus, it is through bank capital that
capital market advances that lead to a more effective resolution of the financing friction end up being
transmitted to the banking sector, permitting the bank to more effectively resolve the certification friction
for some borrowers and expand its lending scope. That is, bank capital is the device by which capital
market advances benefit banks and even borrowers who take only bank loans.
In addition to the complementarity between banks and markets, our analysis also yields several additional results that speak to the questions raised earlier. First, borrowers with high creditworthiness opt
for non-intermediated, direct capital market financing; borrowers with intermediate creditworthiness raise
funds via bank securitization; borrowers with low creditworthiness take relationship loans; and borrowers
with extremely low creditworthiness are excluded from the credit market.5 Second, bank evolution due to
an improvement in the banks screening/certification technology expands the banks relationship lending
scope from below in that the bank now also lends to (previously unserved) riskier borrowers, expands the
banks securitization scope from below, and leads to capital market evolution by enhancing investor participation. Third, capital market evolution expands the banks lending scope from below, leading it to serve
more low-quality borrowers, and hence plants the seeds for bank evolution. That is, there exists a virtuous
circle in which banks and capital markets, even though they represent alternative and competing sources
of financing, also act as collaborators and co-evolve with each other. Numerous empirical predictions are
also extracted from the analysis.
All of these results are derived in a fairly general setting, but one in which the cost of capital market
financing, deposit insurance, and prudential capital requirements for banks as well as the costs associated
5

The result that the least risky borrowers go to the market, the riskier borrowers go to banks and the riskiest borrowers

are rationed is familiar; see, for example, Holmstrom and Tirole (1997). A key difference is that, unlike our model, the bank
does not itself raise funds from the capital market to finance itself. Another key difference is the absence of securitization in
the previous studies.

with these requirements are all taken as exogenous. After deriving our main results, we add more structure
to the model to endogenize these elements.
The essence of our analysis is that banks and markets exhibit three types of interaction: competition,
complementarity, and co-evolution. This three-dimensional interaction sets our paper apart from the
literature. In particular, our thesis that banks and capital markets complement and co-evolve is a departure
from the existing viewpoint that they compete and hence the growth of one is at the expense of the other.
For example, Allen and Gale (1997) show that while banks can provide more effective intertemporal
risk smoothing than markets, their effectiveness in doing so depends on the degree of competition from
the markets, with sufficiently strong competition resulting in disintermediation and impeded provision of
intertemporal risk smoothing by the bank. In Boot and Thakor (1997a), the capital market improves
firms real decisions through its information feedback from equilibrium prices of securities, while banks
are superior in resolving post-lending asset-substitution moral hazard. The choice between a bank-based
system and a market-based system is essentially a tradeoff between the improvement of real decisions from
the markets feedback function on the one hand and the attenuation of moral hazard by the bank on the
other. Thus, capital market evolution, as represented by increasingly efficient price feedback, diminishes
bank lending in that paper. Similarly, in Allen and Gale (1999), Bhattacharya and Chiesa (1995), and
Dewatripont and Maskin (1995), the borrowers choice is between financing from one source or the other,
so as borrowers show a preference for one financing source, they essentially shift away from the other
financing source.
Our analysis clarifies that banks and markets have different comparative advantages, and that they
are competing with each other only when they are viewed in isolation with no instruments that allow
banks and markets to specialize in their respective advantages not when they interact with each other.
Securitization provides one such vehicle, creating a benefit flow from banks to markets. Bank capital
provides another vehicle, creating a benefit flow from markets to banks.
On the complementarity between banks and markets, two previous contributions are related to our
work, although neither of them examines co-evolution. Allen and Gale (2000) note that intermediaries
may complement markets rather than substituting for them. In their analysis, intermediaries are able to
provide individuals insurance contracts against unforseen contingencies in obscure states that eliminate
the need for these individuals to acquire costly information about these states, thereby reducing their costs
of participating in markets. Unlike our analysis, however, bank equity capital and securitization are absent
in their analysis and there is not a feedback loop from banks to markets and another from markets to banks
such that both co-evolve as they do in our analysis. That is, their focus is entirely different. Holmstrom
and Tirole (1997) develop a model of financial intermediation in which firms as well as banks are capital
constrained. Firms with adequate (equity) capital can access the market directly, whereas those with less
capital borrow partly from banks and partly from the market (mixed financing). The bank needs capital
of its own to be induced to monitor the borrowers, which is in turn necessary to enable some borrowers
to obtain (indirect) market finance. One-way complementarity arises from the fact that the presence of
banks permits some borrowers to access the market, just as insurance intermediaries facilitate individual
4

market participation in Allen and Gale (2000). However, there are no benefit feedback loops of the sort we
have, so there is no examination of the co-evolution of banks and markets as in our analysis. Rather, their
focus is on the effects of reductions in different types of capital on investment, interest rates and forms of
financing.
The rest of the paper is structured as follows. Section 2 describes the basic model. Section 3 analyzes the
borrowers choice of funding sources, highlighting the competition dimension of bank-market interaction.
Section 4 generates our main results about the complementarity and co-evolution dimensions of bankmarket interaction. In Section 5 we put additional structure on the model to endogenize the cost of
capital market financing, deposit insurance and a regulatory capital requirement. The empirical predictions
emerging from the analysis of the model as well as the implications for the evolution of financial system
architecture and bank governance are discussed in Section 6. Section 7 concludes. All proofs are relegated
to the Appendix.

The Basic Model

In this section, we describe a simplified model that illustrates our main argument.

2.1

The Agents and Economic Environment

Consider a three-date (t = 0, 1, 2) economy with universal risk neutrality and a zero riskless interest rate.
There are five agents: the borrower, the bank, the depositors, the investors in the capital market, and the
regulator. The borrower may be either authentic or a crook. Both types of borrowers have access to the
same investment opportunity set in terms of projects. The project needs a $1 investment at t = 1, and
generates a cash flow of X > 1 for sure at t = 2. However, only an authentic borrower is interested in
investing in the project. A crook who raises financing for the project will abscond with the funds, leaving the
financier with nothing.6 The common prior knowledge at t = 0 is that with probability q [0, 1] a borrower
is authentic, and with probability 1 q a borrower is a crook. However, only the borrower itself knows
its true type. Thus, the informational problem faced by a financier here is adverse selection. The capital
in number. A subset of these investors, N in number,
market is comprised of finitely many investors, N
, will be participants in any particular security. While N
is exogenous, N will be endogenously
with N N
determined for every security financed in the capital market. Investors are atomistic and behave as price
takers. Each investor suffers a disutility, , if the borrower he has financed ends up defaulting, so this
disutility is experienced only when a crook is financed. We can think of this disutility as the cost the
investor suffers because of the cash flow shortfall he experiences when he does not receive repayment from
6

This can be either an issue of character or skill in developing the project or the cost of personal effort for the borrower

in implementing the project. That is, the crook may have a character flaw that makes absconding with the funds attractive
based on preferences, or may be unskilled in developing the project or may simply be too lazy, i.e., may perceive a personal
cost to develop the project that is too high.

the borrower on the market security he has purchased.7 We assume differs across investors, and is
distributed uniformly on support [0,
].8 We refer to those seeking financing as borrowers because they
are assumed to finance with debt contracts.9
The aggregate supply of deposit funding exceeds the maximum possible loan demand; the same is true
for the aggregate supply of funding from the capital market, either through securitization or through direct
market financing. Each borrower also has multiple a priori identical banks to choose from, although each
bank transacts in equilibrium with only one borrower.

2.2

Deposit Insurance and Regulatory Capital Requirement

The regulator determines the banks deposit insurance coverage and capital requirement at t = 0. We limit
the regulators deposit insurance coverage to either zero or full deposit insurance.10 Suppose the capital
requirement set by the regulator is E [0, 1]. Then, if the bank wants to lend, it needs to raise E in equity
from the capital market at t = 1 and borrow the remaining 1 E from depositors afterwards.11 In the
basic model, both deposit insurance and bank capital are treated as being exogenously given; they will be
endogenized in the complete model.

2.3

The Borrowers Choice of Financing Source

At t = 0, the borrower has three potential choices to finance its project: (i) borrow directly from the
capital market via non-intermediated debt financing; (ii) let the bank screen and (noisily) certify its type
first and then borrow from the capital market via bank securitization; and (iii) take a relationship loan
from the bank. With direct capital market access, the borrower completely bypasses the bank and hence
there is no screening certification provided to the borrower by the bank.
7

This assumption is reminiscent of Diamonds (1984) assumption of a non-pecuniary default penalty on the borrower that

defaults, except that here this penalty is suffered by the investor who purchases security from a crook. A simple way to
interpret this is to think of investors having their own personal borrowing, with each investors ability to repay personal debt
being predicated upon the repayment he receives on the borrowers securities he purchases in the market. Borrower default
can thus trigger default by the investor on his personal debt, with attendant default costs (as in Diamond (1984)) that vary in
the cross-section of investors. Alternatively, the inability to collect on the borrowers repayment obligation triggers a liquidity
problem for the investor, forcing him to sell personal assets at firesale prices to satisfy a liquidity need. These liquidity-related
costs will also typically vary cross-sectionally among investors.
8
The assumption of heterogenous disutility is not crucial to our main argument as long as there is some heterogeneity
among the investors in some (other) dimension that affects the cost of their providing financing to the borrower. Moreover,
the uniform distribution assumption for is made merely for algebraic simplicity.
9
Using equity would not change anything since there is no uncertainty about the project payoff here.
10
This is to simplify the analysis to focus on the main issues. Our main results remain qualitatively unchanged under an
assumption of partial deposit insurance.
11
The idea is that the bank must ensure that it is in compliance with regulatory capital requirements before it can lend.
Deposits are raised afterwards when the loan is actually financed. Whenever we refer to bank capital, we will mean the
banks equity capital.

With securitization, the bank screens the borrower first, and then decides whether to seek capital
market financing for the borrower at t = 1 based on the screening outcome. Because the entire funding
for the loan is provided by the market, there is no need for the bank to keep any capital against the loan.
We assume, however, that securitization involves the bank setting up a bankruptcy-remote special purpose
trust to which the loan is sold. This trust is set up at t = 1 after the bank knows the screening outcome.
The bank provides credit enhancement for the loan via collateral, which is available to investors in case
the loan defaults. This collateral is equal to a fraction, (0, 1), of the initial promised repayment of the
securitized debt to investors. The bank incurs a fixed cost, Z > 0, to set up a trust for securitization.
That is, the bank sets up a trust which sells the loan to capital market investors and collects $1 in
proceeds that get passed along to the bank, which then allows the bank to provide funding to the borrower.
The bank sets the borrowers repayment obligation as Rsec , but the trust promises investors a repayment
sec < Rsec . The investors recourse to the bank in the event of borrower default is R
sec . We assume
of R
that the bank surrenders control over the loan to the trust so that the securitization counts as a loan under
the rules of securitization accounting, and does not require the bank to keep any capital to support the
loan.12
With a relationship loan, the bank screens the borrower first, and then based on the screening outcome
decides whether to raise equity capital and deposits to fund the loan. Prior to screening, the bank posts
a loan interest rate it will charge if it decides to lend to the borrower. This is a precommitment by the
bank. A borrower that approaches the bank for a relationship loan precommits to accepting a loan offer at
that price. The role of the two-sided precommitment will be explained later. Deposit gathering is costly.
Think of it as the cost of setting up branches, employing tellers and so on. Although this cost has both
fixed and variable elements, we simplify by setting the fixed cost at zero and letting the variable cost be
> 0 per dollar of deposit. Since the borrower learns whether it is authentic or a crook before making its
financing choice, its choice of financing source can potentially convey information about its type.

2.4

The Banks Screening and Its Private Signal about the Borrowers Type

The bank specializes in a noisy but informative pre-lending screening technology that reveals the borrowers
type at t = 0. This screening occurs if the borrower approaches the bank for a relationship loan or
securitization. The screening yields a private signal s {sa , sc } to the bank, where sa is a good signal and
sc is a bad signal. Let
Pr(s = sa |authentic) = Pr(s = sc |crook) = p,

(1)

where p [1/2, 1] is the precision of bank screening. If the bank extends credit to the borrower only when
the screening signal s = sa , then p is simply the probability that an authentic borrower receives credit. We
treat p as being common knowledge and exogenously given for now; we will endogenize it later when we
12

FAS 140 is the accounting rule in the U.S. for whether a specific securitization structure qualifies as a loan sale. See

Greenbaum and Thakor (1987) for a discussion of securitization. We will see later, when we endogenize capital requirements,
that the securitization structure we use will not require any capital to be posted.

study the co-evolution of banks and capital markets. The cost to the bank of screening, when the precision
is p, is cp2 /2, where c > 0 is a constant. Each bank can screen only one borrower. Assuming that both the
authentic borrower and the crook choose to approach the bank for financing, the banks posterior beliefs
about the borrowers type after observing its private signal s are:
qp
q A [q, 1],
qp + [1 q][1 p]
q[1 p]
Pr(authentic|s = sc ) =
q C [0, q],
q[1 p] + [1 q]p

Pr(authentic|s = sa ) =

(2)
(3)

where Pr(crook|s) = 1 Pr(authentic|s) s {sa , sc }.

The bank then decides whether to accept the borrower (agree to extend a relationship loan or obtain
the financing via securitization) or reject it. As in Stiglitz and Weiss (1983), we assume that the banks
acceptance/rejection decision is public, so a rejected borrower will be unable to get credit anywhere else.13
The capital market does not possess such a screening technology, an assumption motivated by the existing
financial intermediary literature that banks are specialists in credit screening (e.g., Allen (1990), Boyd and
Prescott (1986), Coval and Thakor (2005), and Ramakrishnan and Thakor (1984)). In particular, the bank
is a specialist in processing soft information (e.g., Stein (2002)) about the borrowers character, which is
one of the five Cs of credit, and corresponds to the banks screening permitting it to noisily distinguish
crooks from authentic borrowers.14 In case the bank lends to a crook that is mistakenly identified as
authentic by the screening, we assume that the banks payoff is zero and depositors are paid off by the
deposit insurer.15

2.5

Market Structure and the Pricing of Securities

When the bank raises equity capital from the market, the equity contract stipulates that the banks initial
shareholders and the new investors (who purchase the equity issued by the bank) share what is left over
13

Bhattacharya and Thakor (1993) discuss how one can justify this assumption in a setting in which the banks rejection

decision conveys adverse information about the borrower, as it does here. This assumption simplifies the analysis, but is not
essential. For example, if a bank can soly noisily learn whether a borrower was previously rejected, it can adjust its posterior
belief accordingly. We will see later that the banks participation constraint will be binding in equilibrium given its prior
belief, p, about the borrowers type. Even noisy information that the borrower was rejected by another bank will lower the
banks belief that the borrower is authentic below p and it will wish to reject the borrower without screening because incurring
the screening cost will violate the banks participation constraint. As will be made clear later (Lemma 2), the banks public
acceptance/rejection decision acts as a credible mechanism by which the bank certifies the borrowers creditworthiness.
14
The capital market also has mechanisms with which to screen borrowers, such as bond ratings issued by credit rating
agencies. Moreover, public listing comes with significant information disclosure requirements that reveal information about
the borrower to investors, so the no-certification assumption in the public market should not be taken literally. Rather, it is a
statement about what happens with bank lending relative to direct market finance. In particular, the contemporary theory of
banking as well as the related empirical evidence strongly suggest that bank screening generates incremental payoff-relevant
information that goes beyond what is available from other sources in the capital market. The evidence provided by James
(1987) is particularly compelling. He finds that the announcement of a bank loan generates an abnormally positive stock price
reaction for the borrower, but an announcement of any other kind of external financing triggers an abnormally negative stock
price reaction.
15
Assuming that the bank too suffers a disutility from financing a crook does not qualitatively affect the analysis.

after the banks repayment to depositors is subtracted from the authentic borrowers loan repayment,
L, to the bank.16 The fraction of ownership in the bank sold to the new investors, 1 , is such that
the bank is able to raise the equity capital E that it needs to support the loan. The equity market is
competitive, so that 1 is determined to yield the marginal investor purchasing equity a competitive
expected return of zero. The banks initial shareholders obtain a share of the banks terminal payoff.
With multiple banks pursing each borrower, banks are Bertrand competitors for borrowers in the loan
market, so L is endogenously determined such that the bank earns zero profit in equilibrium.17 We also
assume a perfectly competitive capital market, which implies that the debt and equity contracts between
the capital market and those seeking financing (the borrower and the bank) are designed such that the
participation constraint for the marginal investor in the capital market is binding in equilibrium.18 The
deposit market is perfectly competitive as well (depositors are simply promised a competitive expected
return equal to zero, the riskless interest rate), implying that the expected repayment on a $1 deposit is
$1.

2.6

Summary of the Sequence of Events

At t = 0, the regulator sets the deposit insurance and capital requirement for the bank. At that time, the
borrower learns its type (i.e., whether it is authentic or a crook). The borrower then decides whether to
raise its financing directly from the capital market, or via securitization, or through a relationship loan
from a bank. If the borrower opts for either a relationship loan or securitization, it approaches a bank
and the bank conducts screening to determine the borrowers creditworthiness. The bank then makes its
acceptance/rejection decision.
At t = 1, with direct capital market financing, investors must decide whether to finance the borrower,
and they must do so without the benefit of bank screening that (noisily) sorts out crooks from authentic
borrowers. With securitization, bank screening nosily sorts out crooks at t = 0, so funding is provided
by investors if the bank screened the borrower affirmatively and accepted the borrower at t = 0. With a
relationship loan, lending will occur if the bank screened and accepted the borrower at t = 0. In that case,
the bank raises the equity capital to satisfy the regulatory capital requirement, E, on the loan, and then
borrows the remaining 1 E from depositors. With both securitization and relationship lending, the bank
has a choice to screen or not to screen. So incentives to screen must be provided.
At t = 2, if the borrower turns out to be authentic, its project payoff is realized and observed by all,
and financiers are paid off. If the borrower turns to be a crook, financiers are left with nothing. This
sequence of events is summarized in Figure 1.

[Figure 1 goes here]

16

Recall that the crook absconds with the funds.

That share of ownership covers the banks costs of screening, cp2 /2, and deposit gathering, [1 E].
18
We will explain later (Section 3.1.2) what we mean by a marginal investor, and show how his (binding) participation
17

constraint determines equilibrium investor participation in the capital market, and hence the cost of capital market financing.

The Analysis of the Basic Model: Choice of Funding Sources

In this section, we present a simple, reduced-form version of our model to succinctly convey the interactions
of the main forces that generate our key results. In this analysis, several elements of the model are taken
as exogenous in order to simplify. These elements are endogenized later in the complete model.
Assumption 1. Valuation Discount: For any borrower seeking financing from the capital market
through either direct market borrowing or securitization, the investors valuation of the expected debt repayment is a fraction (N ) (0, 1) of the borrowers valuation, where N is investor participation in that
security (non-intermediated debt or securitized debt) in the market. When the bank raises equity capital
from the market, the investors valuation of the banks terminal payoff shared between them and the bank
is also a fraction (N ) of the banks valuation, where N is investor participation in the banks equity in
the market. Moreover, 0 () > 0 and 00 () < 0.
The existence of a valuation discount means that capital market financing is costly not only because of
the friction arising from the fact that the borrower pool consists of crooks, but also those seeking financing
(the borrower and the bank) and those providing financing (investors) value differently the surplus from
the project being financed. While taken as an assumption for now, we endogenize this in Section 5 (see
Proposition 5) using a heterogeneous-priors setup in which a public signal about the borrowers project is
observed prior to the borrowers actual investment in the project. Due to heterogeneous priors, investors
and the borrower end up with possibly different posterior beliefs about the value of the project. Since
investors do not directly control project choice, the resulting possibility of disagreement over project value
endogenously generates a valuation discount of 1 (N ) on a $1 expected debt repayment or banks
terminal payoff (for bank equity). The discount 1 (N ) is a decreasing function of investor participation
in a given security (non-intermediated debt, securitized debt, or bank equity) in the market. The intuition
is that a capital market with greater investor participation (larger N ) in a given security has more depth,
thereby decreasing the valuation discount associated with the project. We endogenize this in the complete
model by showing that greater investor participation in the capital market results in lower disagreement
between investors and those seeking financing in equilibrium and hence a lower valuation discount due to
disagreement.19
19

The idea is as follows. Suppose there are N investors participating in a particular security in the capital market. In the

complete model, each investors likelihood to agree with the borrower about the value of the project, call it [0, 1], is an
independent random draw from some probability distribution. The capital market provides a mechanism whereby investors
with the highest valuation are able to bid for the security and, given investor risk neutrality, these investors are willing to
purchase all of the security at their valuations. That is, the security is purchased by investors with the highest among the
N investors; since s are random ex ante, the highest among N investors can be viewed ex ante as the N th order statistic
of . It is clear that the N th order statistic of (i.e., the expected likelihood of agreement between investors and the borrower
in terms of project valuation), and hence (N ), are increasing in N . A numerical example is useful for illustration. Suppose
is uniformly distributed on support [0, 1]. If N = 1, the expected agreement (1st order statistic) is simply 1/2. If N = 2,
R1
then the expected agreement (2nd order statistic) is 2 0 x2 dx = 2/3 > 1/2.

10

Assumption 2. Deposit Insurance and Capital Requirement: The regulator provides full deposit
insurance to the bank. The regulatory capital requirement is E [0, 1], which is decreasing in borrower
credit quality, i.e., E/q < 0.
The intuition for the assumption that E/q < 0 is as follows. Since the borrower should be charged
a lower loan interest rate when its credit quality is higher, it follows that the equilibrium loan repayment
is decreasing in borrower quality (i.e., L/q < 0). Since the banks asset-substitution moral hazard due
to deposit insurance is more severe when the loan repayment is higher (this will be formally shown in
the complete model), the regulatory capital requirement is also decreasing in borrower quality. While we
take both deposit insurance and bank capital as exogenously given for now, they will be endogenized in
the complete model, at which time we will prove that the optimal risk-sensitive capital requirement, E, is
strictly decreasing in borrower quality, q.

3.1

The Authentic Borrowers Payoffs from Various Funding Sources

Before analyzing the borrowers choice of funding source, we state a result about the sharing of the project
surplus between the bank, the borrower and the depositors/investors. We then examine the borrowers
payoffs from these various sources. Our focus is on an authentic borrowers. We shall assume for now that
a crook will make exactly the same financing choice as the authentic borrower. We will verify this formally
later as a feature of the equilibrium.
Lemma 1. When the loan market, capital market and deposit market are perfectly competitive in the
sense that the providers of finance act as Bertrand competitors in these markets, contracts are designed
in equilibrium to maximize the borrowers expected share of the project surplus subject to the participation
and incentive compatibility constraints of the financiers.
This result will be useful in the subsequent analysis to derive the properties of contracts and characterize
equilibrium surplus allocations. The intuition is that since all financiers are acting as Bertrand competitors
for the borrower, all forms of finance deposits, equity and bank loans are competitively priced to yield
financiers an expected return that they compute to be equal to the riskless rate (zero in our model).20
This result is in sharp contrast to Yanelle (1997), who builds upon Stahl (1988) to show that when
intermediaries compete for both loans and deposits, the competitive outcome involving the bank earning
zero expected profit need not obtain. The main reason for this difference can be seen as follows. Yanelle
(1997) studies intermediation using Diamonds (1984) model, in which there are increasing returns to scale
from intermediation; on the asset side the intermediary experiences increasing returns to scale because of
the reduction of duplicated monitoring as the intermediary grows in size, and on the liability side it is
20

Our notion of competition whereby contracts are designed to satisfy the participation constraints of investors, depositors

and the banks shareholders and maximize the borrowers surplus subject to these participation constraints plus incentive
compatibility constraints can also be found in various other papers (e.g., Besanko and Thakor (1987a, b) and Holmstrom and
Tirole (1997)).

11

because of the diversification benefits of size in reducing the risk of uninsured depositors. Intermediaries
thus have an incentive to corner either the deposit or the loan market to achieve a monopoly outcome.
By contrast, there are no such increasing returns to scale in our model. Each bank deals with only one
borrower there is no advantage in dealing with multiple borrowers and deposits are fully insured. In the
absence of increasing returns to scale, even two-sided Bertrand competition for loans and deposits yields
zero profit for the bank in equilibrium.

3.1.1

The Banks Acceptance/Rejection Decision

We start by analyzing the banks decision to accept or reject the borrower based on its screening at t = 0,
when the borrower approaches the bank for either a relationship loan or securitization. If the bank extends
a loan, then perfect competition in the loan market implies that, conditional on the information revealed
by the screening, the banks equilibrium loan pricing maximizes the authentic borrowers expected payoff
subject to the banks participation constraint. With securitization, the terms of credit for the borrower
are determined by the markets perception of the borrowers credit quality, which is affected by the banks
publicly-observable acceptance/rejection decision.
Lemma 2. In both securitization and relationship lending, the equilibrium must involve the bank accepting
the borrower if screening yields a good signal, s = sa , and rejecting the borrower if screening yields a bad
signal, s = sc .
If the bank makes its decision based on the screening outcome by accepting when s = sa and rejecting
when s = sc , then the bank can certify the borrowers creditworthiness to the market, which enables an
affirmatively-certified borrower to obtain better credit terms with securitization than would be available
absent the certification. Absent such certification, securitization will not be viable.21 Recall that investors
who purchase the securitized debt have recourse to the bank for a fraction of the securitized debt if the
borrower defaults. The key is that is set to be sufficiently high such that in equilibrium the bank finds
it not profitable to securitize a borrower without screening it first (otherwise, securitization will not be
viable as discussed before). Now, conditional on screening, if the bank were to also accept the borrower
when s = sc , the banks expected payment to investors under the recourse agreement would be so high
that the banks expected payoff would be negative.22 Thus, securitization with recourse ensures that the
banks acceptance/rejection decision is signal-contingent and therefore the certification provided by the
banks decision to securitize the loan is credible. The intuition for relationship borrowing is similar.
The upshot of this is that if a borrower with prior credit quality q is accepted by the bank for either
securitization or for a relationship loan at t = 0, it is certified by the bank to be authentic with probability
q A > q.
21

If such certification is not provided with securitized debt, then the credit terms that an authentic borrower obtains from

the capital market in securitization are the same as those from direct (non-intermediated) market borrowing. But there is a
securitization cost, Z, that is fully absorbed by the borrower. Thus, direct market financing strictly dominates securitization
if the latter involves no bank certification.
22
Note that the banks expected payoff in this case is even less than that from securitizing the borrower without screening.

12

3.1.2

The Authentic Borrowers Payoffs

We now compute the authentic borrowers net expected payoff at t = 0 associated with each financing
source, which will then help us to determine which source the borrower will prefer at t = 0. These
expected payoffs are computed prior to any bank screening of the borrower.
Direct Capital Market Access: We first analyze the equilibrium investor participation, Ndir , for any
borrower with prior credit quality q borrowing directly from the capital market via a debt contract. The
authentic borrower chooses the debt repayment obligation, Rdir , to maximize its expected payoff, denoted
as dir :
dir = X Rdir ,

(4)

subject to the marginal investors participation constraint:

(Ndir )[qRdir ] [1 q] = 1,

(5)

where is the marginal investors disutility of financing a crook, given by:

.
/
= Ndir /N

(6)

In (4), X Rdir is the net payoff to the authentic borrower. As for (5), note that the probability that a
borrower receiving direct market financing is authentic is q, in which case the investors are repaid. The
expected debt repayment is thus qRdir as valued by the borrower, but (Ndir )[qRdir ] < qRdir as valued by
the investors (see Assumption 1). The probability is 1 q that a crook will be funded, in which case the
investors suffer a disutility of . The expected payoff across those two states must equal 1, the financing
provided. To understand (6), note that only investors with disutility less than or equal to , defined in
(5), will lend to the borrower. The investor with disutility equal to is the marginal investor. From the
uniform distribution assumption for that disutility, we know that the fraction of investors with disutility
.
not exceeding is /
, which equals Ndir /N
Securitization: Next, consider a borrower with prior credit quality q whose bank loan is securitized.
With securitization, what we need to make sure of is that: (i) the bank will indeed screen the borrower,
and (ii) it will securitize only a borrower on which the screening outcome is s = sa . If this can be ensured,
then investors will be assured that with probability q A > q the borrower is authentic (see Lemma 2). The
equilibrium investor participation for securitization, denoted as Nsec , can be analyzed in the same way as
Ndir .
sec (the portion of the repayThe three choice variables, Rsec (the borrowers repayment obligation), R
ment passed along to investors), and (the fraction of the promised repayment for which investors have
recourse to the bank via collateral), are chosen to maximize the authentic borrowers expected payoff from
securitization, denoted as sec , which is the probability that bank screening reveals such a borrower to be
creditworthy, p, times the borrowers net payoff conditional on being funded, which is the project payoff,
X, minus the debt repayment to the bank, Rsec , i.e.,
sec = p[X Rsec ],
13

(7)

subject to the marginal investors participation constraint:

sec ] + [1 q A ][ R
sec ]} [1 q A ] = 1,
(Nsec ){[q A R

(8)

and the banks participation constraint (prior to screening):

sec ] [1 p][1 q][ R
sec ] {qp + [1 q][1 p]}[Z] [cp2 /2] = 0,
[pq][Rsec R

(9)

. As for (8),
where is the marginal investors disutility of financing a crook, given by /
= Nsec /N
sec ] is the investors valuation of their recourse to the cash collateral in case of default.
(Nsec )[1 q A ][ R
Note that (Nsec ) reflects the effect of bank screening on the market, since Nsec is influenced by the fact
that a securitized credit has been screened and certified first by the bank. To undersand (9), the banks
participation constraint, note that the bank only securitizes the borrower when s = sa (this will be verified
shortly). The banks valuation of its expected payoff is the probability that the borrower is authentic and
screening reveals it to be so, i.e., Pr(authentic) Pr(s = sa |authentic) = pq, times the banks net payoff
sec .23 The expected cost to the bank of providing recourse, is the probability that
in that case, Rsec R
the borrower is a crook but screening mistakenly yields a good signal, i.e., Pr(crook) Pr(s = sa |crook) =
sec . With probability Pr(s = sa ) = qp + [1 q][1 p] the bank sets up
[1 p][1 q], times the recourse, R
a trust, so the banks expected cost of setting up securitization is {qp + [1 q][1 p]}[Z]. Finally, cp2 /2 is
the banks screening cost.24
We also need to check the incentive compatibility (IC) constraints in (i) and (ii) above. Consider (i)
first. We need to ensure that the banks net payoff from screening and securitizing, given by (9), is no less
than that from: (a) not screening and not securitizing, and (b) securitizing without screening. Since the
banks payoff associated with (a) is zero, that constraint is obviously satisfied. As for (b), the IC constraint
is that the banks net payoff from securitizing without screening is non-positive (recall q is the prior belief
about borrower quality):
sec ] [1 q][ R
sec ] Z 0,
q[Rsec R

(10)

where we recognize that the bank will have to set up a securitization trust in order to securitize, whether
it screens or not prior to securitization. Since (10) is binding in equilibrium, we solve it to obtain:
=

sec ] Z
q[Rsec R
.
sec
[1 q]R

(11)

Now consider (ii) the bank should prefer to securitize only if s = sa . Securitizing after s = sa yields
a net payoff of zero, according to (9), so this will satisfy the participation constraint. To ensure that the
bank does not securitize when s = sc , we need:
sec ] [1 q C ][ R
sec ] Z [cp2 /2] cp2 /2,
q C [Rsec R
23

(12)

Note that the valuation discount, measured by 1 (Nsec ), only exists between investors in the capital market and the

borrower, but not between the bank and the borrower.

24
sec ] [1 q A ][ R
sec ] Z
Note that (9) is equivalent to: q A [Rsec R

14

cp2 /2
qp+[1q][1p]

= 0.

where the left-hand-side is the banks net payoff if it screens and securitizes a borrower for which s = sc ,
and the right-hand-side is the banks payoff if it screens and decides not to securitize. Solving this yields:

sec ] Z
q C [Rsec R
.
sec
[1 q C ]R

(13)

Since q > q C , we know that the given by (11) will satisfy (13). Thus, the equilibrium is given by (11).
Relationship Loan: Finally, consider an authentic borrower with prior credit quality q financing via a
relationship loan from the bank. Its expected payoff, denoted as loan , is:
loan = p[X L].

(14)

Given Lemma 1, the banks equilibrium choice of the loan repayment obligation, L, maximizes loan subject
to the banks own participation constraint (prior to screening):
[qp][]{L [1 E]} {qp + [1 q][1 p]}{ [1 E]} [cp2 /2] = 0,

(15)

and the marginal investors participation constraint in the capital market:

[1 ](Nloan )[q A ]{L [1 E]} [1 q A ] = E,

(16)

where Nloan is the equilibrium investor participation in the market providing equity capital to the bank,
.
and is the marginal investors disutility of financing a crook, given by /
= Nloan /N
These expressions can be understood as follows. The authentic borrowers expected payoff in (14) is
the probability, p, that such a borrower will receive credit (be affirmatively screened by the bank) times
the borrowers net payoff, which is the project payoff, X, minus the loan repayment, L. To understand
(15), note that if the loan is extended, the bank obtains a share of the terminal payoff, {L [1 E]}, and
the probability of loan repayment is the probability of extending the loan to an authentic borrower, so the
banks ex ante expected payoff prior to screening is [Pr(s = sa ) Pr(authentic|s = sa )][]{L [1 E]} =
[qp][]{L [1 E]}. The banks participation constraint (15) equates that expected payoff to the expected
cost of deposit gathering, [Pr(s = sa )]{ [1 E]} = {qp + [1 q][1 p]}{ [1 E]}, plus the cost of
screening, cp2 /2.25 The bank raises equity capital E to make its relationship loan. The marginal investors
participation constraint (16) equates the investors expected payoff from providing capital to E, the amount
of capital provided.26 The investors share of the banks expected terminal payoff is [1], and the expected
25

It is easy to verify, based on the similar argument as in the case of securitization, that in relationship lending the bank

will indeed screen and only lend to a borrower when screening yields s = sa . Note that (15) is equivalent to: [q A ]{L [1
2

cp /2
E]} [1 E] qp+[1q][1p]
= 0.
26
Note that in this analysis, it has been assumed that the cost of deposit gathering, [1E], and the cost of screening, cp2 /2,

are entirely borne by the bank but not shared by the investors who provide E. This is because of the investors valuation
discount of the banks expected terminal payoff. Take the cost of deposit gathering for example. Note that for every unit cost
of deposit gathering shared by the investors, from the banks perspective it needs to yield more than one unit of its terminal
payoff to the investors to compensate them for bearing the deposit gathering cost. To see this more concretely, note that it
can be derived from (15) and (16) that:

][Nloan ] + E
[1 q A ][
/N
cp

+
,
L = [1 E] 1 + A +
A
q
q (Nloan )
2q

15

terminal payoff itself is (Nloan )[q A ]{L [1 E]} as valued by the marginal investor, which is smaller than
the banks valuation, [q A ]{L [1 E]}.
Solving these three optimization problems in (4) (16), we have the following lemma:
Lemma 3. The equilibrium investor participation and the expected payoffs to an authentic borrower from
the three financing choices, non-intermediated debt, securitization, and relationship borrowing, are all in . For securiticreasing in borrower credit quality, q, and the number of investors in the capital market, N
zation and relationship borrowing, the equilibrium investor participation is also increasing in the precision
of bank screening, p; for each of these two financing choices, there exists a value of p that maximizes the
authentic borrowers expected payoff.
This lemma says the following. First, as borrower credit quality improves (larger q), the probability
of financing a crook decreases and hence more investors are willing to participate when the borrower opts
for direct market financing (lager Ndir ) or securitization (larger Nsec ), and when the bank raises equity
capital from the market in relationship lending (larger Nloan ). Second, for non-intermediated debt and
securitization, higher borrower credit quality not only leads to a lower debt repayment but also to a
lower cost of market borrowing because it elevates investor participation in the market; recall 0 (N ) > 0.
Thus, the authentic borrowers expected payoffs in direct market financing (dir ) and securitization (sec )
are both increasing in borrower quality. Turning to relationship lending, since the bank operates in a
competitive loan market, the equilibrium loan repayment only reflects the banks cost of providing a
relationship loan, part of which is the cost of raising equity capital from the market.27 Higher borrower
quality increases investor participation in bank equity in the capital market. This reduces the cost of
raising equity capital for the bank, thereby lowering the borrowers equilibrium loan repayment and in
turn increasing the authentic borrowers expected payoff from relationship borrowing (loan ). Third, a
) leads to greater investor participation in any security in
capital market with more investors (larger N
equilibrium, and hence a greater expected payoff for the authentic borrower regardless of its financing
choice. Finally, in securitization and relationship borrowing, the probability of financing a crook also
decreases when bank screening becomes more precise (larger p), which leads to the result that investor
participation increases with the precision of bank screening; the authentic borrowers expected payoff
consequently increases as well when p is low. However, as p further increases, the convex cost of screening
(cp2 /2), which is borne by the authentic borrower in equilibrium, becomes sufficiently high so that further
when the bank bears the entire cost of deposit gathering. Instead, if (0, 1) fraction of the deposit gathering cost is shared
by the investors, the banks and the marginal investors participation constraints become [q A ]{L0 [1 E]} [1 ] [1 E]
[cp2 /2] = 0 and [1 ](Nloan )[q A ]{L0 [1 E]} [1 q A ] [] [1 E] = E, respectively, where L0 is the loan repayment.
Straightforward calculations show that:

][Nloan ] + E + [1 E]
[1 q A ][
/N
[1 ]
cp
+
+
,
L0 = [1 E] 1 +
A
q
q A (Nloan )
2q
which is larger than L, since (Nloan ) < 1. That is, loan will be ceteris paribus smaller if the banks equity contract stipulates
the deposit gathering cost to be shared by the investors, which is suboptimal. The case for screening-cost sharing can be
analyzed in the same way.
27
The others are screening cost and the cost of deposit gathering.

16

increases in p cause the authentic borrowers payoff to decrease. Thus, there exists a payoff-maximizing
p (1/2, 1) for securitization and another for relationship borrowing.

3.2

The Authentic Borrowers Choice of Funding Source

In this section, we establish a proposition that characterizes the authentic borrowers choice of funding
source. It is useful to describe the intuition underlying this proposition before we present the formal details.
We begin at the lowest end of the borrower credit quality spectrum. As the authentic borrowers prior
credit quality declines, its loan repayment obligation L increases and its payoff from relationship borrowing
decreases (Lemma 3). Since L increases without bound as q decreases to 0, L becomes prohibitively high
for a sufficiently low q as no bank wishes to finance a borrower who is almost certainly a crook. Thus,
there exists a credit quality cutoff, call it ql > 0, below which borrowers cannot obtain bank financing.
This cutoff defines the banks lending scope, with a broader lending scope being associated with a lower
cutoff.
Now, for the lowest-quality authentic borrowers that qualify for credit (q ql ), a relationship loan
provides the highest benefit from bank screening. For such borrowers, going directly to the capital market
is relatively inefficient because the low q combined with the absence of bank screening means that investor
participation is quite low and the cost of market financing is very high. Bank financing with a relationship
loan, which consists of bank equity and deposits, has its costs too. One is the cost of deposit gathering,
[1 E], but this is relatively low for low-q borrowers because the banks capital requirement, E, is
relatively high for such borrowers. The other is the cost of equity capital the bank raises from the market
to support the loan, which the borrower must absorb in equilibrium. This cost, measured at the margin
by 1 (N ), arises due to the valuation discount with market financing investors value the borrowers
project lower than the bank does. Although this cost is also incurred with direct capital market access, it is
incurred over the entire loan amount ($1) with direct market finance rather than over only the bank capital
(E < 1) used to support the relationship loan. Thus, for the qualifying low-q borrowers, a relationship loan
dominates direct capital market access. These borrowers could, of course, choose securitization, whereby
they get the same benefit of bank screening that a relationship loan provides, but face different costs.
These costs include the fixed securitization cost Z, the banks cost of providing recourse to investors,
and the valuation discount on the capital market financing for the loan. For low-q borrowers, the cost of
recourse and valuation discount are high, so the sum of Z and the recourse and valuation discount with
securitization is also high. By contrast, because the banks capital requirement, E, for such borrowers is
high, the deposit-gathering cost with a relationship loan, [1 E], is low. The cost of equity associated
with the capital requirement is high for low-q borrowers, but this cost arises from the valuation discount,
and this discount applies to only the portion of the relationship loan funded by bank equity capital. By
contrast, the valuation discount applies to the entire loan with securitization. Thus, for low-q borrowers,
the total cost of a relationship loan is exceeded by the total cost of securitization. These borrowers therefore
prefer relationship loans.

17

As the borrowers credit quality increases further, the tradeoff changes as the banks capital requirement
against a relationship loan, E, declines and the deposit-gathering cost, [1 E], increases. The banks cost
of providing recourse with securitization for such a borrower as well as the valuation discount on capital
market funding both decrease and the fixed cost of securitization, Z, is unaffected. Thus, there exists a
quality cutoff, say qm , such that borrowers with prior credit quality q qm prefer securitization over a
relationship loan.
These intermediate-quality borrowers (q qm ) also compare their payoff from securitization to that
from direct capital market access. The benefit of securitization relative to direct market finance for such
borrowers inheres in the bank screening that accompanies securitization. The posterior belief about the
quality of a screened borrower receiving credit via securitization, q A , exceeds the prior belief, q, but this
posterior belief with direct market finance stays at q. Securitization thus leads to higher investor participation in the market and a lower valuation discount than direct market finance. Moreover, the certification
value of bank screening also enables the borrower to obtain better credit terms with securitization than
with direct market finance. Partially offsetting these benefits of securitization are the noise in bank screening that may cause an authentic borrower to be wrongly rejected by the bank, and the sum of the banks
securitization cost Z and the cost of recourse that must be absorbed in equilibrium by the borrower. The
noise in bank screening means that an authentic borrower is denied access to securitization with probability
1 p, which is the probability with which bank screening mistakenly identifies such a borrower as a crook.
Both Z and the cost of being erroneously denied credit do not vary with borrower quality, q, but the benefit
of securitization gets smaller as q increases, vanishing asymptotically as q 1. By contrast, the cost of
going directly to the market, as reflected in the valuation discount, is also getting smaller as q increases.28
Thus, among borrowers with prior credit quality q qm , there will be a cutoff, say qh > qm , such that
those with q < qh will prefer securitization and those with q qh will prefer direct market access. The
proposition given below formalizes this intuition.
and the marginal cost of deposit gathering
Proposition 1. Suppose the securitization cost Z < Z,
and are exogenous constants defined in the Appendix. An authentic borrower chooses
( , ), where Z,
its funding source in equilibrium as follows.
1. There exists a low credit-quality cutoff, ql > 0, such that an authentic borrower with q < ql cannot
obtain financing from the bank. Moreover, ql is decreasing in the number of investors in the capital
.
market, N
2. There exists a high credit-quality cutoff, qh > ql , such that an authentic borrower with q qh borrows
directly from the capital market and the one with q [ql , qh ) approaches the bank. The cutoff qh is
determined such that dir |q=qh = sec |qA =1 . Moreover, qh does not depend on p.
28

So is the cost of recourse with securitization, but that decline affects just a portion of the loan, whereas the effect on

market financing is greater because it affects the entire loan.

18

3. There exists a medium credit-quality cutoff, qm (ql , qh ), such that an authentic borrower with
q [ql , qm ) prefers a relationship loan and the one with q [qm , qh ) prefers securitization. Moreover,
qm is decreasing in p.
It is a unique universally divine sequential equilibrium (Banks and Sobel (1987)) for every crook within a
prior credit quality q cohort to choose the same financing source as the authentic borrower in that cohort.
This equilibrium is supported by the out-of-equilibrium belief that any borrower who makes a financing
source choice other than that described above is a crook with probability one.
Note that qh is determined such that the direct-market-financing payoff to an authentic borrower with
prior quality q = qh , dir |q=qh , is the same as the highest possible payoff that the borrower can get from
securitization when it is believed by the market to be authentic with probability one, sec |qA =1 . Define qh0
as the prior credit quality at which the borrower is indifferent between securitization and direct market
financing, i.e., sec = dir for q = qh0 , sec > dir for q < qh0 , and sec < dir for q > qh0 . It is clear that
qh0 < qh . The authentic borrowers with q (qh0 , qh ) would be better off with direct market financing than
with securitization (see Figure 2), and yet they borrow via securitization in equilibrium. This efficiency loss
arises from the universal divinity refinement of the sequential equilibrium. To understand this, suppose
the authentic borrowers with q (qh0 , qh ) chose direct market financing. Note that ceteris paribus a
crook strictly prefers direct market financing over securitization because bank screening associated with
securitization diminishes the likelihood of the crook obtaining funding. Thus, if a borrower with q (qh0 , qh )
defects from direct market financing to securitization, it would be understood by investors that the borrower
is more likely to be an authentic borrower rather than a crook, so by universal divinity the defector would
be perceived by the market as authentic with probability one. This would create an incentive for all
authentic borrowers with q (qh0 , qh ) to switch from direct market financing to securitization, so all nondefectors would be perceived as being crooks with probability one. This unravels the equilibrium, and no
borrower with q (qh0 , qh ) will be able to receive direct market financing. In a universally divine sequential
equilibrium then, authentic borrowers with q (qh0 , qh ) must choose securitization. It is clear that the
precision of bank screening, p, has no effect on qh .
As for the choice between securitization and relationship borrowing, more precise bank screening invites
greater investor participation in the market both for securitization and for the banks raising of equity to
support its loan. This lowers the costs of both bank equity and securitized debt. But the effect on
securitized debt exceeds that on bank equity. The reason is that more funding is raised from the market
with securitized debt than by the bank when it raises equity capital (E < 1). Hence, securitization becomes
more attractive as the precision of bank screening improves, i.e., qm /p < 0.
Authentic borrowers with prior credit quality q < ql are unable to obtain funding because of pro)
hibitively high loan repayment obligations with relationship borrowing. A larger number of investors (N
elevates investor participation in the market, thereby lowering the cost of bank equity. Due to the competitive structure of the loan market, this cost reduction is passed on in equilibrium to the authentic borrower

19

 allows more authentic borrowers with relatively low prior

receiving a relationship loan. Thus, a larger N
credit qualities to be able to finance their projects, thereby expanding the banks lending scope.
This proposition also states that in equilibrium the crooks mimic the authentic borrowers in their choice
of financing source. The reason is that any deviation from the choice of the authentic borrowers reveals the
crook noiselessly. This means that the equilibrium pools authentic borrowers and crooks, and financiers
learn nothing about a borrowers type by observing its financing choice.
These effects are graphically illustrated in Figure 2.

[Figure 2 goes here]

As our analysis thus far has revealed, there are two frictions that are complementary in the sense
that both impede the borrowers access to credit. One is a friction introduced by a lack of information
about borrower credit quality (certification friction), and the other is a friction that arises from the
dissipative costs of external financing, including the cost of the valuation discount (financing friction).
The certification friction manifests itself in the exclusion of creditworthy borrowers from credit, whereas the
financing friction manifests itself in a higher cost of capital for the borrower. Banks are better at resolving
the certification friction because of their superior screening technology. The financing friction is resolved
in different ways by banks and markets. Banks resolve it by funding themselves with insured deposits, but
do so by incurring a deposit-gathering cost. Markets resolve it by achieving a lower valuation discount via
greater investor participation.29 These different ways of resolving the financing friction generate benefits
that differ across borrowers based on borrower quality. The relative advantages of banks and capital
markets described here are linked to the roles ascribed to these financing sources in the literature, as we
show later when we endogenize our key assumptions.
We end this section with a comment on the role of the two-sided commitment. The bank needs to
precommit to an interest rate it will charge a borrower that it wishes to lend to because otherwise the
bank ends up with an ex post monopoly after having screened the borrower affirmatively. Thus, if some
banks make binding precommitments and others dont, borrowers will go to the banks that precommit.
This also corresponds to what we see in practice a borrower cannot be sure the bank will agree to lend,
but it knows the rate on the loan if the bank agrees to lend. Similarly, borrowers need to precommit that
if offered the loan at the posted price, they will take the loan. This is necessary because the bank earns an
ex post rent when s = sa and it does lend; this rent covers the ex post loss the bank suffers when s = sc
and it does not lend, thereby being unable to recoup its screening cost.30
29

Allen and Gale (1999) develop a model in which markets are superior to banks in aggregating the heterogeneous beliefs

of investors. This is similar in spirit to our setup.

30
This two-sided precommitment only simplifies the analysis and is unnecessary for the results. A competitive credit market
model in which there is no precommitment and borrowers choose how many banks to apply to and be screened simultaneously
is developed in Thakor and Callaway (1983). Our assumption here allows us to side-step the complications that arise in that
analysis, since these are quite peripheral to our paper.

20

The Analysis of the Basic Model: Co-evolution of Banks and Markets

and Financial System Architecture

In this section we analyze the implications of our previous analysis for the co-evolution of banks and capital
markets, followed by an examination of the implications of the analysis for financial system architecture.

4.1

Co-evolution of Banks and Capital Markets: Preliminary Remarks

We now examine how the evolution of the bank affects the capital market, and how the evolution of the
capital market affects the bank. That is, we are going back to t = 1 to solve for the banks choice of
screening precision, p. A preliminary result is useful for this analysis.
Lemma 4. At t = 1, the bank will choose a screening precision that maximizes the expected surplus of
the borrower at t = 0.
The intuition is as follows. At t = 0, banks are engaged in Bertrand competition and hence each bank
chooses credit contracts to maximize the expected surplus of the borrower subject to incentive compatibility
and participation constraints. Thus, all surplus goes to borrowers. Any bank that chooses a p at t = 1
that does not maximize expected borrower surplus at t = 0 will be unable to attract a borrower away from
a bank that chose the surplus-maximizing p at t = 1.31
Consider now a borrower with prior credit quality q, which is drawn from a uniform distribution over
support [0, 1].32 The banks problem at t = 1 is to choose p that maximizes borrower surplus given below:
Z

qm

ql

loan dq +

qh

qm

Z
sec dq +

qh

dir dq.

(17)

We now define bank evolution and capital market evolution.

Definition (Bank Evolution and Capital Market Evolution): Bank evolution is defined as a decrease in
the cost, c, of achieving precision in bank screening. We define capital market evolution as greater investor
participation in the capital market and hence a lower capital market financing cost for the borrower.
Our definitions of bank and capital market evolution are consistent with the traditional view of the
functions of a financial system. A primary function of a financial system is to acquire and process information. In our model, bank evolution is synonymous with investment in the banks screening technology
becoming more efficient (lower c), possibly due to advances in information technology and credit screening
31

Note that the bank is choosing p at t = 1 before it knows the borrowers q which becomes common knowledge only at

t = 0. Thus, p is not chosen for a specific q. However, assuming that the bank chooses p after observing q of the borrower it
faces will not qualitatively affect our results. Borrowers with q [qh , 1] will not approach the bank, so the bank will not invest
in screening precision if the borrower has such a q. Faced with a borrower with q [ql , qh ), the bank will choose p depending
on the q it faces. Capital market evolution will still lower the banks equity cost of capital and induce the bank to deal with
borrowers with lower values of q than before (see Proposition 3).
32
The uniform distribution assumption for q is made for mathematical simplicity.

21

models, so the bank is able to acquire and process information more effectively. Another function of a financial system is to mobilize savings by pooling capital from disparate individual investors and facilitating
trading. This corresponds to our definition of capital market evolution, since a capital market with greater
investor participation is able to perform this function with a lower financing cost.

4.2

The Effects of Bank Evolution on Capital Market Evolution

We now analyze the effects of bank evolution on capital market evolution and the borrowers financing
choice. Bank evolution has two effects. First, as shown in Lemma 3, bank evolution causes investor
participation in the capital market for relationship borrowing to increase, thereby lowering the banks cost
of equity to meet its capital requirement. Second, bank evolution also causes investor participation in
the capital market for securitization to increase. Thus, bank evolution plants the seeds for capital market
evolution by generating greater investor participation in the capital market. The following proposition
summarizes the effects of bank evolution on capital market evolution and the authentic borrowers financing
choice.
Proposition 2. Bank evolution has the following effects: (i) it expands the banks relationship lending
scope from below (ql decreases); (ii) it expands the banks securitization scope from below (qm decreases);
and (iii) it enhances investor participation in the capital market for both relationship borrowing and bank
securitization, thereby leading to capital market evolution.
This proposition can be understood as follows. First, bank evolution generates relationship-loan availability for authentic borrowers with low prior credit qualities who previously had no access to credit. That
is, bank evolution expands the banks lending scope at the lower end of the credit quality spectrum. Second, bank evolution not only broadens the scope for securitization (qm decreases), but also increases its
volume by increasing investor participation in the capital market for securitization. If we include both
securitization and the capital market for non-intermediated, direct borrowing in our view of markets,
then bank evolution invites greater investor participation in capital markets. Taken together, we note that
bank evolution can cause relationship banking to lose business to securitization at the top of the credit
quality spectrum (qm decreases), but gain market share at the bottom by extending its lending scope (ql
decreases). Hence, the overall effect of bank evolution on the banking sector itself is somewhat surprisingly
not unambiguous, but bank evolution is unambiguously beneficial to the capital market.
Relative to previous models of the borrowers choice between a bank loan and direct capital market
financing (e.g., Berlin and Mester (1992), Boot and Thakor (1997a), and Rajan (1992)), what our analysis
adds is securitization in which the bank facilitates the borrowers financing from the capital market through
an informative screening technology that enhances the creditworthiness of the securitized borrower pool and
thereby lowers the borrowers financing cost. However, the introduction of securitization in our model is not
merely adding an element to the borrowers financing choice menu for descriptive completeness. Rather,
securitization generates an important interaction between the bank and the capital market that profoundly
affects financial system architecture. It propagates banking advances to the capital market, permitting
22

capital market evolution to be driven by bank evolution. To see this clearly, it is useful to examine what
would happen if we excluded securitization, as has been typically done in analyses of financial system
architecture.
Corollary 1. Suppose there is no securitization. Then bank evolution expands the banks relationship
lending scope both from below and above, and the capital market loses borrowers to banks.
This result shows that when securitization, the conduit through which the benefits of bank evolution
flow through to the capital market, is excluded, we return to the standard result that banks and markets
compete. A technological improvement in the screening technology employed by banks leads to an increase
in the market share of banks at the expense of the capital market.
Of course, our analysis has ignored the role of non-bank financial intermediaries like credit rating agencies in this certification process. If such intermediaries were introduced in the model, then improvements
in the certification technologies of these non-bank financial intermediaries would enhance investor participation in the capital market independently of securitization and what banks do. However, a key difference
between banks and rating agencies and certification intermediaries is that banks commit their own equity
capital to the loan in their role as lenders, whereas rating agencies do not. Thus, banks have both financial
and reputational capital at stake (e.g., Boot, Greenbaum and Thakor (1993)), and rating agencies have
only reputational capital at stake.33

4.3

The Effects of Capital Market Evolution on Bank Evolution

We now analyze the effects of capital market evolution. Suppose there is some exogenous shock so that
. How will this affect banks?
more investors enter the capital market, thereby increasing N
Proposition 3. Capital market evolution expands the banks lending scope from below, and increases the
banks investment in the screening technology, thereby leading to bank evolution.
The intuition is as follows. Increased investor participation due to capital market evolution makes
equity cheaper for the bank and allows it to lend to borrowers with low qualities that were previously
denied credit (ql decreases). Thus, capital market evolution does not necessarily cause the banks business
to shrink as predicted by the existing models. Rather, in addition to the usual competitive effect, the
evolution of the capital market opens up segments of the credit market that were previously inaccessible
to the bank.34 Moreover, the marginal value of bank screening increases as the bank serves borrowers with
lower credit qualities, which consequently induces the bank to invest more in the screening technology.
This leads to a more precise bank screening technology and hence bank evolution.
33

This may be one reason why James (1987) finds that the borrowers average announcement effect for a bank loan is positive

and for capital market financing is negative.

34
This result depends critically on the assumption that bank capital requirements are increasing in credit risk, something
we will endogenize later.

23

In our model, there are potentially creditworthy borrowers being rationed by banks.35 That is, those
authentic borrowers with q < ql are not served by the financial system. Capital market evolution makes
bank equity capital cheaper, making it optimal for the bank to raise more equity capital. This additional
equity capital permits the bank to serve low-quality borrowers that were previously unserved, thereby
expanding the banks lending scope. The key role played by bank equity capital in our model is that it
enables the universe of the borrowers served by banks to grow larger with capital market evolution, so that
banks and markets need not compete in a static domain. This connects capital market evolution to bank
evolution, and allows market advances to positively affect banks. To the best of our knowledge, such a role
of bank equity capital has not been previously examined in the literature.
Numerous papers have examined bank capital. For example, Morrison and White (2005) study the
impact of bank capital requirements and regulatory auditing on financial crises. Gorton and Winton (2000)
examine the liquidity cost associated with bank equity capital. These papers are concerned primarily with
the more traditional roles assigned to bank capital, in contrast to our paper where it acts as a conduit for
the propagation of capital market advances to the banking sector. Of course, such a propagation would
not occur if we either excluded bank capital from the analysis or simply fixed the cost of bank capital
exogenously, as the corollary below shows.
Corollary 2. Suppose bank equity capital is exogenously fixed and the cost of this capital is also exogenously
fixed. Then capital market evolution causes the bank to lose some borrowers to the market.
Again, when we remove the channel by which the benefits of capital market evolution flow through to
banks, we get the standard result that a technological improvement in the market causes banks to lose
business to the market.
This part of our analysis highlights another key difference between our model and previous research on
financial system architecture. In previous studies, banks and markets interact in a fixed universe of those
seeking credit. In our analysis, capital market evolution lowers the cost of bank capital and increases the
set of creditworthy borrowers, so that banks and markets do not interact in a static domain. It is bank
capital that creates a benefit flow from markets to banks.

4.4

Co-evolution

We now show that not only do banks and capital markets complement each other, they also co-evolve.
We know from Proposition 3 that capital market evolution induces the bank to invest more in screening,
enabling the bank to increase the precision of its screening technology and consequently facilitating bank
evolution. Moreover, as bank screening becomes more precise, investor participation in the capital market
also increases (see Proposition 2), which consequently spurs capital market evolution. This co-evolution
dynamic is stated in our next result.
35

This is credit rationing in the sense of Stiglitz and Weiss (1983), since the bank is unwilling to grant credit to the borrower

even if the borrower offers to pay a higher price for that credit.

24

Proposition 4. Bank evolution spurs capital market evolution, and capital market evolution spurs bank
evolution. That is, banks and the capital market co-evolve with each other.
We saw earlier that instead of the usual result that the two are pure competitors, banks and markets
complement each other. More importantly, Proposition 4 shows that the complementarity extends to coevolution as there are circumstances in which there is a virtuous cycle in which each sector benefits from
the development of the other. This goes well beyond the one-way complementarity results in papers like
Holmstrom and Tirole (1997) that bank monitoring can improve capital market access for borrowers.
The intuition behind Proposition 4 is as follows. Bank evolution enhances the banks screening technology, which improves the quality of bank certification and facilitates resolution of the certification friction.
Capital market evolution invites greater investor participation in the market, which lowers the cost of capital market financing for the borrower and facilitates resolution of the financing friction. As we discussed
earlier, securitization provides a device through which the resolution of the certification friction facilitates
the resolution of the financing friction, thereby allowing bank evolution to benefit the capital market. Bank
capital is a device through which resolution of the financing friction enables the bank to expand its lending
scope, thereby helping resolve the certification friction for previously unserved borrowers and permitting
capital market evolution to benefit the banking sector.

The Complete Model

We now complete the model by endogenizing: (i) deposit insurance and bank equity capital; and (ii)
the valuation discount in market financing. We maintain the same setting for the agents and economic
environment as in the basic model, except that we now assume the borrowers project can be one of two
types: good (G) and bad (B). If the project is good, its payoff at t = 2 is X > 1 for sure. A bad project
always pays off zero at t = 2. The common prior belief at t = 0 is that the project is G with probability
(0, 1), and is B with probability 1 . We assume X < 1, i.e., the project has negative NPV a
priori. In what follows, we introduce a new element to the model: heterogenous prior beliefs. While this
additional structure is special, it should be noted that its main role is to endogenize Assumptions 1 and
2. In particular, heterogeneous prior beliefs allow us to simultaneously endogenize a valuation discount
((N )) that is increasing and concave in investor participation (N ) and justify complete deposit insurance
in a model without coordination failures. Alternatives to heterogeneous priors may deliver some of what
we need, but we have been unable to find an alternative that delivers all that we need to endogenize.
For example, heterogeneous transaction costs or risk aversion among investors may be able to generate a
valuation discount, but the important property that this discount is endogenously increasing and concave
in N would be lost. Moreover, transaction costs or risk aversion would not help us endogenize complete
deposit insurance without coordination failures.

25

5.1

Additional Model Structure: The Public Signal about Project Type and the
Potential for Disagreement Among the Agents

For any borrower with prior credit quality q, a public signal regarding the type of its project is observed
by all the agents at t = 1 just before deposits are raised. The signal is {G , B }, where G is a good
signal and B is a bad signal. Everybody sees the same signal, i.e., there is no disagreement regarding the
signal itself. Moreover, we assume that the common-knowledge prior probabilities are Pr( = G ) =
and Pr( = B ) = 1 for all projects regardless of the borrowers true type, since both the crook and
the authentic borrower have the same project access. That is, while the banks private signal s is about
the borrowers type, the public signal is about the type of the project that the borrower has.
Although all the agents see the same signal and have the same prior beliefs about the values (G or
B ) the signal will take, they have different priors about the precision of the signal. More specifically, the
signal precision, which we denote as , can take one of two values: the signal can either be precise (I) or
uninformative (U ). The probabilities of drawing I and U are [0, 1] and 1 , respectively. A precise
signal is viewed as perfect and causes the receiver of the signal to arrive at a posterior belief that puts
all of the probability weight on the value of the signal, and an uninformative signal has no incremental
information content, so it is disregarded and the posterior belief about the projects type stays at the prior
belief.
To see this concretely, consider the case in which the signal is G . When the prior belief about the
signal precision is I, the agents belief about the projects type is Pr(G| = G , = I) = 1; when the prior
belief about the signal precision is U , the agents belief about the projects type remains at its prior, i.e.,
Pr(G| = G , = U ) = . If the signal is B , it is clear that a precise signal leads the agent to believe that
the project is bad almost surely, and an uninformative signal does not change the agents prior belief about
the NPV of the project, which is again negative. Thus, a signal realization = B results in agreement
among all the agents at t = 1 that the project has negative NPV, regardless of prior beliefs about signal
precision.
The agents randomly draw prior beliefs about the precision of . We assume that the signal precision
drawn by an agent is privately observed by that agent and not verifiable by others. To focus on the main
issues, we assume the borrower, the bank and the regulator always agree with each other regarding the
precision of the signal, denoted as b .36 Even though there are possibly multiple depositors, we assume that
their prior beliefs about signal precision, denoted as d , are perfectly correlated, so that depositors act as a
monolithic group.37 We model potential divergence of prior beliefs between the borrower/bank/regulator
36

Dropping the assumption that there is no disagreement between the bank and the borrower will not qualitatively change

the analysis as long as we continue to assume that depositors may disagree with the bank. The key to the analysis is that
depositors may be unwilling to provide finance even when the bank finds the borrower creditworthy. Our assumption that the
bank and the regulator agree with each other helps to simplify the analysis of deposit insurance and capital requirement that
is presented later.
37
As we will show later, assuming heterogenous beliefs across depositors does not change our analysis.

26

on the one hand and depositors on the other hand regarding the signal precision via the following structure
of conditional probabilities:
Pr(d = I|b = I) = d [0, 1],

(18)

Pr(d = U |b = I) = 1 d .

(19)

Let i denote the investors prior belief about signal precision. We model potential divergence of prior
beliefs between the borrower/bank/regulator on the one hand and investors on the other hand regarding
the signal precision via the following structure of conditional probabilities:
Pr(i = I|b = I) = [0, 1],

(20)

Pr(i = U |b = I) = 1 .

(21)

Moreover, we introduce heterogeneity among investors by assuming that the value of varies in the crosssection of investors in the capital market, which will be made clear in Section 5.2.
From the standpoint of beliefs, we model depositors as homogeneous and investors as heterogeneous.
The reason for assuming homogeneous beliefs across depositors is that (insured) bank deposits represent a
single financial security that is likely to attract a homogeneous group of investors. By contrast, the capital
market offers a variety of risk-return tradeoffs and will attract a greater diversity of investors; we will say
more about this later. Coval and Thakor (2005) show how investors with different beliefs self-select and
invest in different securities.38 See also Allen and Gale (1988) for a related argument in a state-preference
framework.
The value of d () measures the degree of agreement between the borrower/bank/regulator and depositors (investors). The higher is d (), the greater is the agreement between the borrower/bank/regulator
and depositors (investors) in the sense that the higher is the probability that their prior beliefs about the
signal precision will coincide. A value of d = 1 ( = 1) indicates perfect agreement and a value of d = 0
( = 0) indicates perfect disagreement. The agreement parameter d () is affected by the attributes of
the borrowers project and/or its business characteristics. If a project involves a radically new product
or business design, there may be very little hard historical data to gauge the probability of the project
succeeding in the future. Project evaluation may thus have to be based largely on soft information that is
inherently subjective in nature (e.g., Stein (2002)), possibly causing d () to be low. By contrast, for a
project that is somewhat more familiar in the sense that similar projects have been tried in the past, there
may be a more balanced mix of hard historical data and soft information, so the value of d () may be
relatively high.
We assume that all agents have rational beliefs as defined by Kurz (1994a,b), who provides a theoretical foundation for heterogenous priors. Although Kurzs theory of rational beliefs has many aspects,
the two aspects most relevant for our analysis are that agents have different priors and that all these priors
are consistent with the data in the sense that none can be precluded by historical data. In a situation such
38

In that paper, financial intermediation arises endogenously as an institutional response to the beliefs irrationality of some

agents. We assume here, however, that all beliefs are rational, even though they are heterogenous.

27

as the setting we have for projects with a paucity of hard historical data and non-stationary distributions,
agents will not be able to uniquely derive the precision of the signal from historical data, so that many
different distributions of precision may be consistent with the data.39

5.2

Additional Model Structure: Investor Heterogeneity in the Capital Market

Each investors agreement parameter with the borrower, , is an independent random draw from a continuous probability distribution F (), with the associated density function f () and support [0, 1] q. The
equilibrium agreement parameter between the capital market and the borrowing agent is determined by
investor participation in the market. The capital market provides a mechanism whereby investors with
the highest valuation are able to bid for the security and, given investor risk neutrality, these investors
are willing to purchase all of the security at their valuations. As shown in Boot, Gopalan and Thakor
(2008), if there are N investors participating in the capital market in a given security for a borrower, the
market-clearing mechanism ensures that the security is purchased by investors with the highest agreement
parameter among the N investors, which is the N th order statistic of , denoted as max{1iN } i ,
where i is the agreement parameter between the ith investor and the borrowing agent regarding the
signal precision for the project. We call those investors maximal investors.40 Denote E(
) M , the
equilibrium agreement parameter in capital market financing for the borrower when there are N investors
participating in the market. The following result characterizes the relation between the equilibrium agreement parameter and investor participation in capital market financing.
Lemma 5. The equilibrium agreement parameter, M , is increasing in the equilibrium investor participation with capital market financing, N .
The intuition is that the N th order statistic of the agreement parameter is increasing in N , the number of
investors participating in the market.41 We will explicitly characterize M when we analyze the equilibrium
investor participation N for different modes of capital market financing. For analytical tractability, we
assume henceforth that follows a uniform distribution on support [0, 1].
39

Technically, what we are modeling is a setting in which the economic observables based on which agents form beliefs are

stable but not stationary (see Kurz (1994a,b)). In this case, the rational expectations hypothesis requires agents to have
information about underlying processes that cannot be derived from historical data, whereas the rational beliefs hypothesis
requires only that their beliefs be consistent with the data.
40
It is useful to distinguish between a maximal investor and a marginal investor. As explained earlier in the basic model
(Section 3.1.2), the marginal investor is the investor with the highest disutility of financing a crook (i.e., highest ) among
all the investors participating in any particular security. This investor determines investor participation in the security. A
maximal investor is the one with the highest valuation of that security (i.e., highest ) among all the participating investors.
If all investors have the same , the maximal investor is also the marginal investor. But this need not be so when s vary
across investors. In that case only the maximal investors will hold the security in equilibrium.
41
This has also been shown by Boot, Gopalan and Thakor (2008).

28

5.3

Discussion of the Model for the Borrower

As described above, we model each borrower as being distinct in terms of its credit quality q, and the
agreement parameter , which represents the extent to which capital market investors will agree with the
borrower regarding the value of the project. The credit-quality dimension reflects the usual post-lending
moral hazard on the part of the borrower, arising in our model from the possibility that the borrower is
a crook.42 The investor-agreement parameter , however, represents a departure from the usual commonpriors assumption in that we permit heterogeneous priors not only between the borrower and the investors,
but also among the investors themselves (recall the value of varies across investors).
Our choice of heterogenous priors for modeling disagreement is motivated by the fact that assessments
of the value of techological or product innovations are typically associated with a diversity of opinions, as
observed by Allen and Gale (1999). When something is new and unfamiliar, it is common for different agents
to have different beliefs about its future potential, and a paucity of historical data impedes convergence of
these beliefs (see Schumpeter (1934)). This stands in sharp contrast to investments in established industries
where there is an abundance of historical data drawn from stationary distributions of underlying economic
variables, and divergent beliefs can thus converge.
It is important to distinguish between disagreement and cash-flow risk. One may argue that innovations
are inherently riskier in a cash flow sense, so would it not suffice to model innovative projects as simply being
riskier than projects in established industries, rather than invoking heterogeneous priors and disagreement?
The answer is no. While innovative projects may involve high cash flow uncertainty, this kind of risk is an
inappropriate way to distinguish between innovative and established projects. For example, the U.S. credit
card business, which is well established, involves relatively high default risk, with annual default rates of
30% 40%. However, there is little disagreement over what these default rates are and how credit cards
should be priced. The key distinction between the old and the new stems not from cash flow risk but from
Schumpeters (1934) observation that ...the new is only the figment of our imagination, which means it
is difficult to bring hard data to bear on the problem of resolving differences of opinion.

5.4

Endogenizing Capital Market Valuation Discount, Deposit Insurance and Bank

Capital

We now use the additional model structure described above to endogenize the exogenous elements in
Assumptions 1 and 2 in Section 3.
42

Although we dont model legal systems formally, we can think of an economy dominated by borrowers with low q as

an economy with weak legal contract enforcement, while an economy populated with borrowers with high q as one with
strong legal contract enforcement. That is, the strength of contract enforcement in the legal system may affect the fractional
representation of crooks in the pool of those seeking financing.

29

5.4.1

Capital Market Valuation Discount

Consider an authentic borrower financing from the capital market via either direct borrowing or securitization. From the investors perspective, the borrower may invest in a project with negative NPV as
perceived by the investors but positive NPV as perceived by the borrower when they have different prior
beliefs about signal precision; this occurs when { = G , b = I, m = U }. That is why the investors
valuation of the expected debt repayment from the borrower is always lower than the borrowers valuation.
The case for bank capital is similar. The degree of such valuation discount is determined by the level
of agreement between the borrower and the maximal investor in the capital market, which in turn is
determined by investor participation in the market (see Lemma 5). The following proposition endogenizes
what was stated in Assumption 1.
Proposition 5. Suppose the equilibrium investor participation in the capital market is N for direct market
financing, securitization, or the bank raising equity capital. Then, the investors valuation of the expected
payoff made by the borrowing agent to them is a fraction (N ) of the borrowing agents valuation, where
(N ) is increasing and concave in N and is given by:
(N ) =

+N
(0, 1).
1+N

(22)

The intuition for this result is that an increase in the number of investors increases the equilibrium
agreement parameter (Lemma 5), and the consequently lower disagreement between the borrower and the
investors leads to a higher valuation by the investors.

5.4.2

Deposit Insurance

We now provide a rationale for deposit insurance within the context of our model. Suppose there is no
deposit insurance and the regulator sets no capital requirement for the bank (i.e., E = 0). Consider a
borrower with prior credit quality q choosing a relationship loan and suppose the bank finds the borrower
creditworthy and decides to finance it. The bank has to borrow the entire $1 from depositors at t = 1.
Without deposit insurance, depositors will only lend when they consider the project to be worth funding,
i.e., { = G , d = I}. Thus, depositors may withhold funding even though the bank and the regulator
believe the project is profitable. This occurs when { = G , b = I, d = U }, so the bank and the regulator
believe the project is worth funding whereas depositors believe the project is a bad bet. Conditional
on the borrower being authentic, this state occurs with probability [1 d ] > 0 as long as there is
some disagreement between the bank/regulator and depositors regarding the public signals precision. It
represents a perceived social welfare loss of q A [1 d ][X 1] to the regulator.43
43

This is because the regulator has the same prior belief of signal precision with the bank. This assumption is not crucial.

As long as there is some disagreement between the regulator and depositors, deposit unavailability at t = 1 always represents
a perceived social welfare loss to the regulator. Also, in a model with divergent beliefs, it is not possible to talk about
social welfare without determining who has the right beliefs. Thus, we refer to perceived social welfare. Note that the
assumption of homogeneous beliefs across depositors is also not critical: as long as there exist some depositors disagreeing

30

In the absence of deposit insurance, a capital requirement E (0, 1) cannot eliminate this perceived
social welfare loss to the regulator. The reason is that as long as the deposits are not fully protected by
deposit insurance, the amount of deposits needed for the project to be financed, 1 E, will be unavailable
whenever there is disagreement between the bank/regulator and depositors at t = 1.44 To eliminate this
perceived inefficiency, the regulator will provide full deposit insurance, which causes depositors to ignore
their potential disagreement with the bank/regulator regarding the project payoff and provide financing
whenever the bank raises deposits from them. This eliminates the possibility of a deposit shortage.45 We
thus have an endogenous justification for the full deposit insurance assumption stated in Assumption 2.

5.4.3

Bank Capital

Banks Asset-Substitution Moral Hazard: Assuming the regulator does not impose a bank capital
requirement, the introduction of deposit insurance generates an asset-substitution moral hazard problem
in that the bank may invest in a negative-NPV project due to the well-known deposit-insurance put option
effect. To see this, let us first write down the banks expected profit with a relationship loan when it faces
a capital requirement of E and invests only in a positive-NPV project:46

{qp + [1 q][1 p]} []{q A [L 1] + [1 + q A ][E]} [1 E] cp2 /2.

(23)

In equilibrium, this expected profit is zero. Now consider the banks project investment decision with full
deposit insurance and a zero capital requirement. In equilibrium, the regulator would want the bank to
not invest if the bank (hence also the regulator) perceives the project to have negative NPV. We will show,
however, that this equilibrium cannot occur without a capital requirement. Suppose the loan repayment
obligation is L > 1 in this conjectured equilibrium. In the state in which the signal is good but the banks
signal is uninformative, { = G , b = U }, the project is perceived to have negative NPV by both the
bank and the regulator, and should be rejected. However, since deposits are fully insured, depositors are
always willing to provide funds with the deposit repayment being $1. Thus, if the bank invests in this
with the bank, they will not provide financing, causing the deposit financing to be lower than $1, which again prevents the
bank from investing in the project and represents a perceived social welfare loss to the regulator.
44
The assumption of homogeneous beliefs across depositors simplifies but is not necessary for this result: as long as disagreement exists between the bank/regulator and some depositors, there will be perceived social welfare loss with some probability.
45
Note that in this setting with heterogeneous beliefs, providing deposit insurance does not represent a perceived cost
to the regulator conditional on the borrower being authentic, since the regulator (like the bank) believes that when the
authentic borrower invests in the project, it must be G (we will verify that this is true in equilibrium later) and depositors
get full repayment. That is, deposit insurance is a promise that represents real protection for depositors against crooks and
disagreement with the bank, and the only contingent liability it creates for the insurer is that the bank may unwittingly finance
a crook. Consequently, complete deposit insurance is a better strategy for the regulator than partial deposit insurance that
exposes the bank to a non-zero probability of not receiving deposit funding.
46
Note that from the banks perspective, conditional on screening yielding s = sa , the net terminal payoff shared between
the bank and investors is:
Pr( = G , b = I)[q A ]{L [1 E]} + [Pr( = B ) + Pr( = G , b = U )][E] = q A [L 1] + [1 + q A ][E].
{z
} |
{z
}
|
project invested

no project invested

Equation (23) is the banks ex ante expected payoff prior to screening.

31

negative-NPV project, with probability q A the authentic borrowers project will pay off X and it will
repay L to the bank, whereas with probability 1 q A the project will pay off zero (either because the
borrower is a crook with probability 1 q A , or with probability q A [1 ] the borrower is authentic but the
project turns out to be bad), leaving the bank with nothing, and the repayment to depositors ($1) in this
state is covered by deposit insurance. The banks expected profit from this investment is q A [L 1] > 0.
This breaks the conjectured equilibrium. That is, complete deposit insurance creates an asset-substitution
moral hazard problem.
Capital Requirement with Relationship Loan: We now endogenize E, the regulatory capital requirement that resolves the asset-substitution moral hazard associated with the banks overlending propensity
in the presence of complete deposit insurance.47 The time line here is as follows. The bank first screens the
borrower and then decides whether to accept or reject the borrower for a loan. If the loan is given to the
borrower, the bank raises E. After that, the bank receives the signal with precision b , based on which
the bank decides whether to accept or reject the project. If the project is rejected, the equity capital E is
shared between the bank and those investors who provided E. If the project is accepted and the borrower
is given a loan, then the bank proceeds to raise deposits and any surplus from the project is also shared
between the bank and those investors who provided E.48
The intuition behind why a capital requirement attenuates asset-substitution moral hazard is as follows.
Consider the banks investment decision for a negative-NPV project in the face of capital requirement
E > 0. If the bank rejects the project, E remains on the banks balance sheet and the bank retains
a share of that capital. If the bank invests in the project and it subsequently fails, the bank loses its
share of E, and the expected cost of losing capital increases with both the amount of capital the bank
is required to pledge against the loan and the default probability of the loan. Thus, a sufficiently high
capital requirement deters the bank from investing in a negative-NPV project. Another way to see this
is that the banks equilibrium expected profit is always zero. Without a capital requirement, the banks
shareholders can earn a positive expected profit out of equilibrium by lending to the borrower that has a
negative-NPV project. What a capital requirement of E does is that it makes the banks expected profit
from this out-of-equilibrium strategy negative. The banks cost of raising equity capital from the market
partially offsets this moral-hazard-attenuation benefit of capital, and introduces a tradeoff that determines
the equilibrium bank capital requirement.49 The following proposition endogenizes what was stated in
Assumption 2.
Proposition 6. The regulator provides full deposit insurance, and sets the banks capital requirement
E (0, 1),50 which is decreasing in the borrowers prior credit quality, q.
47
48

Overlending refers to the bank deliberately financing negative-NPV projects.

This setting is more complex than the one in the basic model. In Section 3.1.2 when we analyze relation loan, once the

bank accepts the borrower after its screening, E is raised and always invested. What is shared between the bank and those
investors who provided E is thus the surplus from the project investment. That is, E is left idle on the balance sheet.
49
Various other papers have shown how capital requirements facilitate prudential regulation by reducing the risk-taking
propensity of the bank. See, for example, Merton (1977), Morrison and White (2005), and Repullo (2004).
50
Its mathematical expression is in the Appendix.

32

The regulatory capital requirement is lower for higher-quality (higher q) borrower pools because the
asset-substitution moral hazard is less severe for such borrowers. The reason is that higher borrower quality
leads to a lower equilibrium loan repayment obligation and makes overlending less attractive for the bank,
thereby ameliorating asset-substitution moral hazard.
This result provides a new rationale for deposit insurance and bank capital regulation in the context
of heterogeneous agents. In our model, deposit insurance arises endogenously to eliminate a perceived
social welfare loss from the standpoint of the bank regulator. A bank capital requirement then emerges
as an endogenous response to the asset-substitution moral hazard induced by deposit insurance. What is
familiar about this rationale is that deposit insurance does indeed seek to protect depositors as in the usual
justification, but this protection is motivated by the regulators desire to ensure a dependable supply of
deposits for the bank and to preclude underinvestment in real projects due to divergent beliefs, rather than
to prevent bank runs due to coordination failures. We do not view this as a competing explanation for
deposit insurance, but rather as a complement to existing theories. It is somewhat similar to the rationale in
Morrison and White (2006) who show that, even without coordination failures, deposit insurance enhances
welfare when adverse selection is sufficiently severe.
Securitization: The analysis now also clarifies why no capital requirement is needed with securitization.
In the state in which the signal is good but the banks signal is uninformative, { = G , b = U }, the
project has negative NPV. Unlike the relationship loan case, however, the bank will be unable to securitize
the loan because investors will not purchase any claims against it. Thus, there is no asset-substitution
moral hazard.

Empirical Predictions and Governance and Regulation Implications

In this section, we first discuss the empirical predictions of our analysis, including those that have empirical support as well as those that remain to be tested, and then examine the governance and regulation
implications of the analysis.

6.1

Empirical Predictions

1. Our analysis implies that riskier firms prefer bank financing, while safer firms tap capital markets
(see Proposition 1). This is consistent with the existing literature (e.g., Bolton and Freixas (2000),
and Petersen and Rajan (1995)).
2. Recent empirical evidence on cross-country differences in economic performance based on financial
system architecture (i.e., the degree of bank orientation versus market orientation) indicates that economic performance seems unaffected by financial system architecture (e.g., Beck and Levine (2002),
and Levine (2002)). These findings cast doubt on the usefulness of the banks versus markets debate, and suggest that it is the overall ability of the financial system to ameliorate information and

33

transaction costs, not whether banks or markets provide these services (Beck and Levine (2002)),
that is of first-order importance. This view of financial system architecture is consistent with the
analysis in this paper. That is, since banks and markets co-evolve, a financial system with strength
in one sector will also display strength in the other (see Propositions 2, 3 and 4). Thus, financial
system classification as either bank-dominated or market-dominated based on banking scope permitted by regulators will not necessarily generate differences in economic performance across different
classifications.
3. A third implication of our analysis is that as the capital market develops, banks start to lend to
riskier borrowers, thereby expanding lending scope within a given financial system (see Proposition
3). Hence, economies with better-developed capital markets should have banks that lend to riskier
and smaller firms. That is, capital market development opens up previously inaccessible markets
for banks. We are not aware of any existing empirical evidence on this prediction, but believe it is
testable.
4. Securitization and bank equity capital play key roles in generating a co-evolution loop in our analysis.
Without these elements, we get pure competition between banks and markets (see Corollaries 1 and 2).
Thus, our analysis implies that competition was more characteristic of the interaction between banks
and the capital market prior to the advent of securitization, whereas complementarity describes this
interaction more effectively now. Moreover, there are many countries in which securitization is either
virtually non-existent or in its infancy. One should expect competition to dominate the interaction
between banks and markets in these countries. This prediction too remains to be tested.
5. The role played by banks will be diminished if non-bank financial intermediaries such as credit rating
agencies develop the screening technology needed for certification of borrower credit quality (see
the discussion following Corollary 1). Thus, our analysis suggests that in economies where credit
rating agencies play a bigger role, the impact of bank evolution on capital market evolution (via
securitization) should be weaker. This prediction awaits testing as well.
6. The valuation discount experienced by borrowers will decline and assets will rise in market value as
the capital market evolves (see Proposition 5). The co-evolution of banks and markets means that
this benefit will also be experienced as banks evolve.

6.2

Governance and Regulation Implications

Our analysis raises some issues that deserve further discussion within the context of bank governance and
regulation. The first issue is the relationship between capital market evolution and bank governance and
regulation. Capital market evolution leads banks to lend to riskier borrowers, and causes banks to enter
previously-untapped markets. This change in the banks asset portfolio may significantly alter the banks
payoff distribution, which suggests that the way the board of directors judges the banks CEO may have
to change since the boards payoff-dependent inferences about the CEOs ability should adapt to changes

34

in the banks asset payoff distribution. Moreover, as the banks payoff distribution is altered, so might
the incentives of the banks CEO to share information with the board, thereby affecting how corporate
governance functions (see, for example, Adams and Ferreira (2007), and Song and Thakor (2006)).
Bank regulation may also be affected by capital market evolution. One implication of our analysis is
that, as the capital market evolves, the prospect of raising bank capital requirements to deal with riskier
lending should be less unattractive to regulators since the cost of bank equity capital decreases with capital
market development. Moreover, our analysis shows that capital market development induces banks to lend
to riskier borrowers. So, to the extent that the determination of bank capital requirements is influenced by
asset portfolio risk as well as the cost of bank equity capital, an implication of our analysis is that capital
requirements ought to be dependent on the evolution of the banking sector itself. This seems to militate
against the adoption of uniform capital requirements across countries with different levels of development
of banks and capital markets.

Conclusion

We have developed the thesis that banks and capital markets exhibit three forms of interaction: competition, complementarity, and co-evolution. The key conditions for this three-dimensional interaction are
securitization and bank capital requirements. Securitization creates a vehicle by which bank evolution is
good for markets since the improved bank screening that accompanies bank evolution enhances the credit
quality of borrowers going to the capital market via securitization, thereby increasing capital market investor participation. And bank capital generates a mechanism by which the evolution of markets is good
for banks since such evolution reduces the cost of bank equity capital, incenting banks to hold more capital, thereby diminishing the rationing of potentially creditworthy relationship borrowers and increasing
bank lending scope. Besides providing a sharp departure from the existing theoretical notion of banks and
markets as competitors for a fixed pool of firms seeking financing, our analysis also generates a number
of testable predictions. A key insight of our analysis is that when banks and markets evolve, one cannot
think of the pool of borrowers as static; this pool endogenously evolves as well. This has implications for
bank governance and regulation.
Further research could go in various directions. One would be to empirically test the various new
predictions of our analysis. Another would be to formally introduce non-bank financial intermediaries like
credit rating agencies and examine the evolution of financial system architecture in a setting with richer
institutional detail.

35

Appendix
Define
y(x)
x

(x)
0 (x)

x, and denote y 1 as the inverse function of y;51

pk1
pk1 +[1p][1k1 ] ,

with k1 being defined such that 1 + Z =

N
]y 1
1+[1k1 ][
/N
[1k1 ]

N
k1 y 1 [1k
]

k2 , such that

N
]y 1
1+[1k2 ][
/N
[1k2 ]

N
k2 y 1 [1k
]

= H Z/p.

Parametric Restriction 1.

N
]y 1
1 + [1 x
][
/N
[1
x]

Z < Z p X

N
1
x
y
[1
x]

(A1)

( , ),

(A2)

Parametric Restriction 2.

where

x
Z + p

k2 X

]y 1 ( N )
1+[1
x][
/N
[1
x]

(y 1 (

N
[1
x]

]y 1 ( N E )+E
[1
x][
/N
[1
x]

))
p[1 E]

E
N
]y 1
[1k2 ][
/N
+E
[1k2 ]

E
N
y 1 [1k
]

1E

(y 1 (

E
N
[1
x]

))

x
,

k2 .

Proof of Lemma 1: For direct market financing, its clear that in equilibrium the borrowers repayment obligation
with a non-intermediated debt is stipulated such that the expected payoff to the lending investor just equals 1, the
financing provided: if the lenders participation constraint were not binding, the borrower would borrow from another
investor charging a lower debt repayment. The same argument applies to securitized debt. Consider relationship
lending. Note that financial intermediation here has a constant-returns-to-scale technology and each bank deals with
only one borrower. Now, it is clear that the participation constraints for depositors and investors in the capital
market (who provide equity capital to the bank) will be binding in equilibrium. If the bank were to obtain funds at
rates that slackened these participation constraints, the borrower would be better off going to a bank that procured
less expensive financing. Moreover, the bank loan repayment, L, will also be stipulated such that the banks expected
payoff just covers its cost (deposit gathering and screening): if this were not the case, the borrower would opt for
another bank charging a lower interest rate (lower L).

Proof of Lemma 2: With securitization, suppose the borrowers repayment obligation to the bank is Rsec , but
sec < Rsec to investors who purchase the securitized debt. That is, investors
the securitization trust passes only R
sec . In equilibrium, Rsec , R
sec and are set such that it is
recourse to the bank in the case of borrower default is R
incentive compatible for the bank to not securitize without screening (otherwise, securitization will not be viable),
sec ] [1 q][ R
sec ] Z 0. Now, conditional on screening (after the cost cp2 /2 is incurred), if s = sc
i.e., q[Rsec R
sec ] [1 q C ][ R
sec ] Z [cp2 /2],
and the bank were to accept the borrower, its expected payoff would be q C [Rsec R
whereas if the bank rejects the borrower, the bank cannot recoup its screening cost and its payoff is simply cp2 /2.
sec ] [1 q C ][ R
sec ] Z [cp2 /2] < cp2 /2, since q C < q. Thus, in equilibrium the bank will
Note that q C [Rsec R
reject the borrower if screening yields s = sc .
We now prove the case for relationship lending. We first claim that if there were no bank certification provided
to the borrower in relationship lending, relationship loan would be strictly dominated by direct market financing.
This is because: (i) the part of the loan raised from the capital market (E) involves the same cost as in direct
market borrowing if there were no bank certification associated with relationship lending, and (ii) the other part
of the loan raised from depositors (1 E) entails deposit-gathering cost. Thus, in order for relationship lending
to be viable, it must be accompanied by bank certification. Similar to the case for securitization, the equilibrium
loan repayment obligation (L) and the banks share of project net payoff () in relationship lending are set such
51

It is straightforward to check that y 0 > 0 and hence y is invertible.

36

that it is incentive compatible for the bank to not lend without screening, i.e., q{L [1 E]} [1 E] 0.
Now, conditional on screening, if s = sc and the bank were to accept the borrower, its expected payoff would be
q C {L [1 E]} [1 E] cp2 /2, which is smaller than the payoff if the bank rejects the borrower, cp2 /2, since
q C < q. These arguments prove the lemma.

Proof of Lemma 3: Direct Market Financing: Substituting Rdir and into the borrowers objective function, we
can rewrite the borrowers problem as:
max dir = X

{Ndir }

][Ndir ]
1 + [1 q][
/N
.
q(Ndir )

The equilibrium investor participation, Ndir , is given by the following first-order-condition (FOC):

(Ndir )
N
Ndir =
.
0
(Ndir )
[1 q]

(A3)

It is straightforward algebra to check that the second-order-condition (SOC) is satisfied. Note that the left-hand-side
(Ndir )00 (Ndir )
LHS
=

> 0 since 00 (Ndir )(Ndir ) < 0. The right-hand-side

(LHS) of (A3) is increasing in Ndir : N
[0 (Ndir )]2
dir
. Thus, we have Ndir /q > 0 and Ndir / N
> 0. The equilibrium debt
(RHS) of (A3) is increasing in q and N
repayment obligation is:
Rdir =

][Ndir ]
1 + [1 q][
/N
,
q(Ndir )

(A4)

=
where Ndir is given by (A3). Using the Envelope Theorem, we have dir /q = Rdir /q > 0, and dir / N
> 0.
Rdir / N
Securitization: Using straightforward algebra we can show from (8), (9) and (11) that:

][Nsec ]
cp2 /2
1 + [1 q A ][
/N
Z
cp
sec + Z + 1 q
Rsec = R
=
+ A+ .
A
A
q q qp + [1 q][1 p]
q (Nsec )
q
2q
Thus, the borrowers optimization problem is equivalent to:
][Nsec ]
1 + [1 q A ][
/N
.
A
q (Nsec )
{Nsec }
min

The equilibrium investor participation, Nsec , is given by the following FOC:

(Nsec )
N

N
=
.
sec
0 (Nsec )
[1 q A ]

(A5)

It can be verified that the SOC is satisfied. It is clear that the LHS of (A5) is increasing in Nsec , and the RHS of (A5)
. Thus, we have Nsec /q > 0, Nsec /p > 0 and Nsec / N
> 0. Moreover, for each fixed
is increasing in q, p and N
q, the RHS of (A5) is greater than the RHS of (A3). Thus, we have Nsec > Ndir , q. The proof of the comparative

statics of sec with respect to q and

to that fori direct market financing by using the Envelope Theorem.
h N is similar
][Nsec ]
1+[1q A ][
/N
cp2
Finally, note that sec = 2q + p X
[Z/q]{qp + [1 q][1 p]}, which is concave in p. Thus,
q A (Nsec )
there exists a p that maximizes sec .
Relationship Loan: From the banks and the marginal investors participation constraints, we have:

][Nloan ] + E
[1 q A ][
/N
cp

+ .
L = [1 E] 1 + A +
q
q A (Nloan )
2q

(A6)

The banks problem can be rewritten as:

min L.

{Nloan }

The equilibrium investor participation, Nloan , is given by the following (FOC):

(Nloan )
E
N

N
=
.
loan
0 (Nloan )
1 qA

(A7)

Again, it can be verified that the SOC is satisfied. The LHS of (A7) is increasing in Nloan , and the RHS of (A7)
. Thus, we have Nloan /q > 0, Nloan /p > 0 and Nloan / N
> 0. The proof of
is increasing in q, p and N

the comparative statics of loan with respect to q and N is similar to that for direct market financing by using the

37

Envelop Theorem. Finally, the existence of a payoff-maximizing p follows from the same argument as in the case for
securitization by observing that loan is concave in p.

Proof of Proposition 1: Bank Lending Scope: Note that L as q 0 (see (A6)). Thus, the existence of the
low cutoff, ql , is clear based on the discussion in the text. Note that ql is determined as loan |q=ql = 0. Thus,
= loan / N < 0, since loan / N
> 0 and loan /q > 0 (see Lemma 3).
ql / N
loan /q
Financing Choice: We first examine the authentic borrowers choice of funding source, assuming that the crook
chooses the same funding source as the authentic borrower with the same prior credit quality q; we will show later
that this indeed is a universally divine sequential equilibrium.
First, we analyze the cutoff qh . Define qh such that sec |qA =1 = dir |q=qh , i.e.,
X 1Z =X

][Ndir ]
1 + [1 qh ][
/N
,
qh (Ndir )

(A8)

N
dir )
where Ndir is given by y(Ndir ) (N
Ndir = [1q
0 (N
. The parametric assumption that Z is not too large (see
dir )
h ]
(A1)) guarantees that sec |q=qh > 0 and hence securitization is viable at q = qh . It is clear that dir |q=qh > sec |q=qh .
Note that (A8) is not a function of p. Thus, qh /p = 0.

Next, we analyze the cutoff qm . The existence of such a cutoff, based on the discussion in the text, can be
guaranteed by the parametric assumption in (A2) that is neither too large nor too small. More specifically, the
assumption that < guarantees that sec |q=ql < loan |q=ql = 0, and the assumption that > guarantees that
sec |q=qh > loan |q=qh . Combining these with the fact that both loan and sec are increasing and concave functions
of q establishes the existence of the cutoff, qm (ql , qh ). To prove the comparative statics of qm , note that qm is
determined by the following equation:
sec |q=qm loan |q=qm = 0.

(A9)

We have:
qm
loan /
=
< 0,

sec /q loan /q
since a higher only decreases loan but not sec , thereby making securitization a better funding choice than
relationship borrowing for the authentic borrower (i.e., qm decreases). Since loan / < 0, we must have sec /q
loan /q > 0, q. Then, we have:
qm
sec /p loan /p
=
loan /p sec /p < 0,
p
sec /q loan /q
where the last inequality can be proved as follows. Note that:

sec /q loan /q = sec /q A loan /q A q A /q > 0.

Thus, it must be true that sec /q A loan /q A > 0, since q A /q > 0. We have:

loan /p sec /p = loan /q A sec /q A q A /p < 0,

since q A /p > 0.
We now show that it is a universally divine sequential equilibrium (Banks and Sobel (1987)) that the crook will
choose the same funding source as the authentic borrower with the same prior credit quality. We begin by noting that
it is transparent that the equilibrium is sequential (Kreps and Wilson (1982)). Next, consider the choice between
securitization and direct market financing. For q [qm , qh ), the expected payoff from securitization to an authentic
borrower is sec , and the expected payoff from securitization to a crook is 1 p. Consider a defection to direct capital
market borrowing. Define Nd (a|direct market financing) as the set of investors who must participate in the capital
market with direct financing to make it strictly optimal for the authentic borrower to defect from securitization. Let
Nod (a|direct market financing) be the set of investors who must participate in the capital market with direct financing
o
as:
to leave the authentic borrower indifferent between direct market financing and securitization. Define Nda
o ,
sec = dir |N =Nda

and hence
o
Nod (a|direct market financing) = {Nda
},

and
o
Nd (a|direct market financing) = {N |N > Nda
}.

38

Note that a crooks expected payoff from direct market financing is 1, which is strictly greater than 1 p. A crook
only cares about the likelihood that it will get funded but not the credit terms (since it knows it will never repay
the loan). Hence, if we define Nd (c|direct market financing) and Nod (c|direct market financing) as the strict-defection
and indifference sets for the crook, we have:
Nod (c|direct market financing) = ,
and
Nd (c|direct market financing) = {N |N > 0}.
This means
Nod (a|direct market financing) Nd (a|direct market financing) Nd (c|direct market financing).
Thus, by universal divinity, investors must believe that
Pr(defector is crook|defection from securitization to direct market financing) = 1.
Given this belief, the authentic borrower nor the crook has an incentive to defect from securitization to direct market
financing, and hence this equilibrium is universally divine.
Now, consider a defection from direct market financing to securitization for borrowers with q [qh , 1). Note that
a crook is strictly worse off from that defection relative to not defecting since 1 p < 1. An authentic borrower is
also strictly worse off by defecting since dir is greater than the highest possible payoff from securitization; note that
dir > sec |qA =1 = X 1 Z, q [qh , 1). Thus, neither an authentic borrower nor a crook wants to defect. The
other cases can be proved in a similar way.

Proof of Lemma 4: First, note that for each given p, Bertrand competition at t = 0 ensures that all surplus at
t = 0 goes to the borrower regardless of its financing choice. Thus, a borrowers expected payoff at t = 1 (before
Rq
Rq
R1
it knows q) for each given p is given by qlm loan dq + qmh sec dq + qh dir dq. Denote the value of p that maximizes
that payoff as p . Second, if a bank does not choose p = p at t = 1, then it cannot attract any borrower at that
time. Thus, in equilibrium every bank will choose p = p .

Proof of Proposition 2: We know from the proof of Lemma 3 (see (A6)) that L/c > 0. Hence, we have
loan /c < 0, and consequently ql /c > 0. That is, bank evolution (lower c) causes the bank to expand its lending
scope from below (ql decreases). This proves (i). To show (ii) and (iii), first note that the banks optimal choice of
p at t = 1 increases as c decreases. Then, the claim in (ii) follows directly from the result that qm /p < 0 and
qh /p = 0 (see Proposition 1), and the claim in (iii) follows directly from the result in Lemma 3 that Nloan /p > 0
and Nsec /p > 0.

Proof of Corollary 1: If there is no securitization, the claim that bank evolution causes the capital market to
lose borrowers to the bank can be proved by observing that a lower c due to bank evolution increases the borrowers
expected payoff from relationship borrowing loan , but not from direct capital financing dir (note that dir /c = 0).
The claim that bank evolution expands the banks lending scope from below can be established in the same way as
in Proposition 2.

hR
i
Rq
R1
q

Proof of Proposition 3: Denote p argmax qlm loan dq + qmh sec dq + qh dir dq . Note that: (i) when N
increases, loan increases, and hence ql decreases (see Proposition 1), (ii) loan /p = [loan /q A ][q A /p]. Note
that loan /q A and q A /p are decreasing in q. This implies that loan /p is decreasing in q. Combining (i) and
.
(ii) yields p being increasing in N

Proof of Corollary 2: If the banks equity capital and its cost are exogenously fixed, then capital market evolution
) has no effect on the banks rasing of equity capital from the market. Thus, loan , and hence ql , will not
(larger N
> 0 (see Lemma 3). This will lead to
change with respect to capital market evolution. Also, note that sec / N

qm / N < 0, causing the bank to lose some borrowers to the market.

Proof of Proposition 4: This comes from combining the results in Propositions 2 and 3.

Proof of Lemma 5: It is clear that the distribution function of , denoted as F (

, N ), can be written as F (
, N ) =
QN
Pr(
) = Pr(1 , . . . , N ) = i=1 Pr(i ) = F ()N . Since F () [0, 1], it is clear that for any two
values of N , say N1 and N2 with N1 < N2 , F (
, N2 ) first-order stochastic dominates F (
, N1 ). This implies that M
is increasing in N , and hence proves the lemma.

Proof of Proposition 5: Consider direct capital market financing. When the state { = G , b = I, m = I}
occurs, i.e., both the borrower and investors perceive the project to be G, the debt repayment is Rdir as valued by
both the borrower and investors. However, when the state { = G , b = I, m = U } occurs, the borrower perceives

39

the project to be G whereas investors perceive it to be G with probability and B with probability 1 . In this
state, the debt repayment is Rdir as valued by the borrower, but is only Rdir as valued by investors. Thus,
Investors valuation of expected debt repayment
{M + [1 M ]}[Rdir ]
=
= M + [1 M ],
Borrowers valuation of expected debt repayment
Rdir

(A10)

where
Z

M = N

Z
N 1

F (x)

f (x)xdx = N

xN 1 xdx =

N
.
1+N

(A11)

Thus,
Investors valuation of expected debt repayment
+N
(N ) =
(0, 1).
Borrowers valuation of expected debt repayment
1+N

(A12)

The cases for securitization and bank equity can be proved in a similar way. It is clear that 0 () > 0 and 00 () < 0,
which is consistent with Assumption 1.

Proof of Proposition 6: Consider any borrower with prior quality q taking a relationship loan. Suppose the banks
asset-substitution moral hazard problem has been resolved, i.e., the bank only invests when { = G , b = I}. From
the banks perspectives, conditional on screening yielding s = sa , the net terminal payoff shared between the bank
and investors is:52
Pr( = G , b = I)[q A ]{L [1 E]} + [Pr( = B ) + Pr( = G , b = U )][E] = q A [L 1] + [1 + q A ][E].
The banks share of the net terminal payoff, , in equilibrium must be such that the banks participation constraint
is binding:53
{qp + [1 q][1 p]}[]{q A [L 1] + [1 + q A ][E]} {qp + [1 q][1 p]} [1 E] cp2 /2 = 0,
which yields:

[1 E] +
q A [L 1] + [1

= 1+

cp2 /2
qp+[1q][1p]
,
+ q A ][E]

cp2 /2
qp+[1q][1p]

(A13)

[ + + q A ][E]
q A

(A14)

If the bank invests in the negative-NPV project when the signal is good but uninformative, conditional on the
borrower being authentic, with probability the authentic borrower is able to repay L and the net terminal payoff
is {L [1 E]}, and with probability 1 the project turns out to be bad. Thus, the banks expected payoff from
cp2 /2

[ +][E]

qp+[1q][1p]
investing in this negative-NPV project is q A {L [1 E]} =
. To resolve the banks

asset-substitution moral hazard problem, the regulator needs to set the banks capital requirement high enough so
that its expected payoff from investing in the negative-NPV project is no more than its expected payoff from rejecting
cp2 /2

it, which is E, i.e.,

+ qp+[1q][1p] [ +][E]

E. This leads to:

cp2 /2
qp+[1q][1p]

+ [1 + ]

(A15)

Substituting (A13) into (A15) and treating (A15) as an equality yields:54

E=

[q A ][L 1]
.
1 q A

(A16)

52
Note that: (i) {L [1 E]} is the terminal payoff when the bank perceives the project to be worth funding, i.e.,
{ = G , b = I}, and makes a deposit repayment of 1 E after receiving L from the authentic borrower; and (ii) when
the bank does not invest, either because the signal is bad, i.e., { = B }, or the signal is good but uninformative, i.e.,
{ = G , b = U }, the equity capital raised at t = 0 is left intact and the terminal payoff is E.
53
This is the banks ex ante participation constraint before screening, where Pr(s = sa ) = qp + [1 q][1 p] is the probability
that the bank accepts the borrower.
54
The other solution is E = 1, which is dominated by the solution in (A16) because of the cost of raising bank capital. Also,
note that although the capital requirement in (A16) is designed to solve the banks asset-substitution moral hazard problem in
the state { = G , b = U }, it automatically solves the banks asset-substitution moral hazard problem in the state { = B }
as well, since the bank perceives that Pr(G| = B ) Pr(G| = G , b = U ) and hence its project-investment distortion is
less severe in the state { = B } than in the state { = G , b = U }.

40

We shall now determine the equilibrium loan repayment obligation L. From the investors perspective, the net
terminal payoff to be shared with the bank is:55

Pr( = G , b = I, m = I){L [1 E]}

A
q
+ [Pr( = B ) + Pr( = G , b = U )][E],
+ Pr( = G , b = I, m = U ){L [1 E]}
= q A {M + [1 M ]}[L 1] + {1 + q A {M + [1 M ]}}[E].
The ownership sharing, , is determined such that the participation constraint for the marginal investor is binding
in equilibrium, i.e.,

q A {M + [1 M ]}[L 1]
[1 ]
[1 q A ] = E.
(A17)
+{1 + q A {M + [1 M ]}}[E]
][Nloan ]. Combining (A13) and (A17), and substituting E with (A16), we have:
where = [
/N

][Nloan ]

[1q A ][
/N

1 q A 1+ +
1+M [1]
.
L=1+

q A
+ M [1]
1+

(A18)

1+M [1]

Substituting (A18) into (A16) yields the regulatory capital requirement:

E=

][Nloan ]
[1q A ][
/N

1+ +
1+M [1]
M [1]

1+ + 1+M [1]

(A19)

Note that the equilibrium capital market agreement parameter when there are Nloan investors participating in the
capital market is given by (see the proof of Proposition 5):
M =

Nloan
.
1 + Nloan

Substituting (A20) into (A18) and (A19) yields:

1 q A
3
Nloan + 2 Nloan + 1
L = 1+
,
q A
1
Nloan + 3

3
Nloan + 2 Nloan + 1
E =
,
1
Nloan + 3

(A20)

(A21)
(A22)

where
1
2
3

N
,

[1 q A ][1 + ]

1 + [1 ] N
1+
,

[1 q A ][1 + ]

.
[1 + ][1 + ]

What remains to be determined is the equilibrium investor participation Nloan for relationship lending. The bank
chooses L to maximize the authentic borrowers payoff from relationship borrowing, by minimizing L, given by (A21).
It is easy to show that L is a convex function of Nloan . Thus, the first-order-condition (FOC) for optimality yields
the solution for Nloan given by:
q
Nloan = 1 2 3 + 23 3 .
(A23)
Substituting (A23) into (A22) gives E. The claim that E/q < 0 can be proved using the Envelope Theorem.56
55
First, investors perceive the net terminal payoff to be {L [1 E]} when they agree with the bank that the authentic
borrowers project is worth funding, i.e., { = G , b = I, m = I}. Second, when the bank thinks the project is worth funding
but investors disagree, i.e., { = G , b = I, m = U }, the investment decision rests entirely with the bank, so investors perceive
that with probability the authentic borrower is able to repay L, and the net terminal payoff is {L [1 E]}, and with
probability 1 the authentic borrowers project defaults and nothing is left to share. Third, when the bank does not invest,
i.e., { = B } and { = G , b = U }, the equity capital raised is left intact and the net terminal payoff is E.
56
Note that the relationship between L and E is given by (A16) and hence the banks problem of minimizing L is equivalent
to minimizing E.

41

References
[1] Adams, Renee, and Daniel Ferreira, 2007, A Theory of Friendly Boards, Journal of Finance 62, 217-250.
[2] Allen, Franklin, 1990, The Market for Information and the Origin of Financial Intermediation, Journal of
Financial Intermediation 1, 3-30.
[3] Allen, Franklin, and Douglas Gale, 1988, Optimal Security Design, Review of Financial Studies 1, 229-263.
[4] Allen, Franklin, and Douglas Gale, 1997, Financial Markets, Intermediaries, and Intertemporal Smoothing,
Journal of Political Economy 105, 523-546.
[5] Allen, Franklin, and Douglas Gale, 1999, Diversity of Opinion and Financing of New Technologies, Journal
of Financial Intermediation 8, 68-89.
[6] Allen, Franklin, and Douglas Gale, 2000, Comparing Financial Systems, The MIT Press.
[7] Banks, Jeffrey, and Joel Sobel, 1987, Equilibrium Selection in Signaling Games, Econometrica 55, 647-661.
[8] Beck, Thorsten, Asli Demirg
uc-Kunt, and Ross Levine, 2000, A New Database on the Structure and Development of the Financial Sector, World Bank Economic Review 14, 597-605.
[9] Beck, Thorsten, and Ross Levine, 2002, Industry Growth and Capital Allocation: Does Having a Market- or
Bank-Based System Matter? Journal of Financial Economics 64, 147-180.
[10] Berlin, Mitchell, and Loretta Mester, 1992, Debt Covenants and Renegotiation, Journal of Financial Intermediation 2, 95-133.
[11] Besanko, David, and Anjan Thakor, 1987a, Competitive Equilibrium in the Credit Market under Asymmetric
Information, Journal of Economic Theory 42, 167-182.
[12] Besanko, David, and Anjan Thakor, 1987b, Collateral and Rationing: Sorting Equilibria in Monopolistic and
Competitive Credit Markets, International Economic Review 28, 671-689.
[13] Bhattacharya, Sudipto, and Gabriella Chiesa, 1995, Proprietary Information, Financial Intermediation, and
Research Incentives, Journal of Financial Intermediation 4, 328-357.
[14] Bhattacharya, Sudipto, and Anjan Thakor, 1993, Contemporary Banking Theory, Journal of Financial Intermediation 3, 2-50.
[15] Bolton, Patrick, and Xavier Freixas, 2000, Equity, Bonds, and Bank Debt: Capital Structure and Financial
Market Equilibrium under Asymmetric Information, Journal of Political Economy 108, 324-351.
[16] Boot, Arnoud, Radhakrishnan Gopalan, and Anjan Thakor, 2008, Market Liquidity, Investor Participation
and Managerial Autonomy: Why Do Firms Go Private? Journal of Finance 63, 2013-2059.
[17] Boot, Arnoud, Stuart Greenbaum, and Anjan Thakor, 1993, Reputation and Discretion in Financial Contracting, Americal Economic Review 83, 1165-1183.
[18] Boot, Arnoud, and Anjan Thakor, 1997a, Financial System Architecture, Review of Financial Studies 10,
693-733.
[19] Boot, Arnoud, and Anjan Thakor, 1997b, Banking Scope and Financial Innovation, Review of Financial
Studies 10, 1099-1131.
[20] Boyd, John, and Edward Prescott, 1986, Financial Intermediary-Coalitions, Journal of Economic Theory 38,
211-232.
[21] Coval, Joshua, and Anjan Thakor, 2005, Financial Intermediation as a Beliefs-Bridge between Optimists and
Pessimists, Journal of Financial Economics 75, 535-569.
[22] Deidda, Luca, and Bassam Fattouh, 2008, Banks, Financial Markets and Growth, Journal of Financial
Intermediation 17, 6-36.
[23] Demirg
uc-Kunt, Asli, and Ross Levine, 2001, Financial Structure and Economic Growth: A Cross-Country
Comparison of Banks, Markets, and Development, The MIT Press.
[24] Demirg
uc-Kunt, Asli, and Vojislav Maksimovic, 1996, Stock Market Development and Financing Choices of
Firms, World Bank Economic Review 10, 341-370.
[25] Dewatripont, Mathias, and Eric Maskin, 1995, Credit and Efficiency in Centralized and Decentralized
Economies, Review of Economic Studies 62, 541-555.
[26] Diamond, Douglas, 1984, Financial Intermediation and Delegated Monitoring, Review of Economic Studies
51, 393-414.

42

[27] Gorton, Gary, and Andrew Winton, 2000, Liquidity Provision, Bank Capital, and the Macroeconomy, working
paper, University of Minnesota.
[28] Greenbaum, Stuart, and Anjan Thakor, 2007, Contemporary Financial Intermediation, Elsevier.
[29] Holmstrom, Bengt, and Jean Tirole , 1997, Financial Intermediation, Loanable Funds, and the Real Sector,
Quarterly Journal of Economics 112, 663-691.
[30] James, Christopher, 1987, Some Evidence on the Uniqueness of Bank Loans, Journal of Financial Economics
19, 217-235.
[31] Kreps, David, and Robert Wilson, 1982, Sequential Equilibria, Econometrica 50, 863-894.
[32] Kurz, Mordecai, 1994a, On the Structure and Diversity of Rational Beliefs, Economic Theory 4, 877-900.
[33] Kurz, Mordecai, 1994b, On Rational Belief Equilibria, Economic Theory 4, 859-876.
[34] Levine, Ross, 2002, Bank-Based or Market-Based Financial Systems: Which Is Better? Journal of Financial
Intermediation 11, 398-428.
[35] Levine, Ross, and Sara Zervos, 1998, Stock Markets, Banks, and Economic Growth, American Economic
Review 88, 537-558.
[36] Merton, Robert, 1977, An Analytic Derivation of the Cost of Deposit Insurance and Loan Guarantees, Journal
of Banking and Financing 1, 3-11.
[37] Morrison, Alan, and Lucy White, 2005, Crises and Capital Requirements in Banking, American Economic
Review 95, 1548-1572.
[38] Morrison, Alan, and Lucy White, 2006, Is Deposit Insurance a Good Thing, and If So, Who Should Pay for
It? working paper, University of Oxford.
[39] Petersen, Mitchell, and Raghuram Rajan, 1995, The Effect of Credit Market Competition on Lending Relationships, Quarterly Journal of Economics 110, 407-443.
[40] Rajan, Raghuram, 1992, Insiders and Outsiders: The Choice between Informed and Arms-Length Debt,
Journal of Finance 47, 1367-1400.
[41] Ramakrishnan, Ram, and Anjan Thakor, 1984, Information Reliability and a Theory of Financial Intermediation, Review of Economic Studies 51, 415-432.
[42] Repullo, Rafael, 2004, Capital Requirements, Market Power, and Risk-Taking in Banking, Journal of Financial
Intermediation 13, 156-182.
[43] Song, Fenghua, and Anjan Thakor, 2006, Information Control, Career Concerns, and Corporate Governance,
Journal of Finance 61, 1845-1896.
[44] Stahl, Dale, 1988, Bertrand Competition for Inputs and Walrasian Outcomes, American Economic Review
78, 189-201.
[45] Stein, Jeremy, 2002, Information Production and Capital Allocation: Decentralized versus Hierarchical Firms,
Journal of Finance 57, 1891-1921.
[46] Stiglitz, Joseph, and Andrew Weiss, 1983, Incentive Effects of Terminations: Applications to the Credit and
Labor Markets, American Economic Review 73, 912-927.
[47] Schumpeter, Joseph, 1934, The Theory of Economic Development, Harvard University Press.
[48] Subrahmanyam, Avanidhar, and Sheridan Titman, 1999, The Going-Public Decision and the Development of
Financial Markets, Journal of Finance 54, 1045-1082.
[49] Sylla, Richard, 1998, U.S. Securities Markets and the Banking System, 1790-1840, Federal Reserve Bank of
St. Louis Review 80, 83-104.
[50] Tadesse, Solomon, 2002, Financial Architecture and Economic Performance: International Evidence, Journal
of Financial Intermediation 11, 429-454.
[51] Thakor, Anjan, and Richard Callaway, 1983, Costly Information Production Equilibria in the Bank Credit
Market with Applications to Credit Rationing, Journal of Financial and Quantitative Analysis 18, 229-256.
[52] Yanelle, Marie-Odile, 1997, Banking Competition and Market Efficiency, Review of Economic Studies 64,
215-239.
[53] Yosha, Oved, 1995, Information Disclosure Costs and the Choice of Financing Source, Journal of Financial
Intermediation 4, 3-20.

43

Figure 1: Sequence of Events

t 0

 The regulator sets the deposit insurance

and capital requirement for the bank.

 Borrowers type (i.e., whether it is

authentic or a crook) is realized and

revealed only privately to the borrower.
Each borrower has a project that needs a $1
investment at t 1.

 Each borrower decides whether to raise

funds via direct (non-intermediated)
capital market financing, securitization,
or a relationship loan from a bank.

 If the borrower decides on securitization,

the bank first screens the borrower. The
bank then agrees to attempt to raise funds
from the market at t 1 if it accepts the
borrower based on the screening
outcome. The securitization is with
limited recourse to the originating bank,
so the bank owes investors a fraction
of the promised repayment if the
borrower defaults.

 If the borrower decides on relationship

borrowing, the bank first screens the
borrower, and then decides whether to
accept or reject the borrower based on
the screening outcome.

t
1

t 2

 If the borrower raises funds from

 The borrowers type becomes public

the capital market directly, investor

participation in the borrowers debt
security in the capital market and
the equilibrium debt repayment
obligation for direct, non-intermediated
capital market financing are then
endogenously determined.

 If the borrower chose securitization

information. If the borrower is

authentic, the project cash flow is
realized and distributed among the
agents according to the stipulated
contract terms.

 If the borrower turns out to be a

crook, financiers are left with
nothing.

and was affirmatively screened

by the bank at t 0 , the bank seeks
funding for the borrower. Investor
participation in the borrowers
securitized debt in the capital market
and the equilibrium debt repayment
obligation for securitization are then
endogenously determined.

 If the borrower chose a relationship loan

and was affirmatively screened at t 0,
then the bank approaches the capital
market to raise equity to meet the capital
requirement E against the loan. Equilibrium
loan repayment obligation, investor
participation in the banks equity in the
capital market and the banks cost of equity
capital are all then endogenously determined.
The bank then borrows 1 E from depositors.

44

Figure 2: A Graph of Financial System Architecture: Different Sources of Financing

loan , sec and dir are the expected payoffs for the authentic borrower with prior credit quality q from relationship
borrowing, securitization and non-intermediated, direct capital market financing, respectively. The cutoff ql defines
the banks lending scope, qm is the cutoff between relationship borrowing and securitization, and qh is the cutoff
between securitization and non-intermediated, direct capital market financing.

The Authentic Borrower's Expected Payoff

!dir

!sec

!loan

q 0

q ql

q qm

q qh ' q q h

Borrower Credit Quality

45

q 1