Jaffee & Stiglitz, Credit Rationing

Chapter 16
CREDIT
RATIONING
DWIGHT JAFFEE
Princeton University
JOSEPH STIGLITZ
Stanford University
Contents
1.
2.
3.
Introduction
Basic notions
2.1.
Uncertainty
2.2.
Loan contracts
2.3.
The classification of borrowers
2.4.
The sources of loanable funds: Banks and bank deposits
2.5.
Definitions of credit rationing
2.6.
The earlier credit rationing literature: Brief review
Credit markets with imperfect information

3.1.
3.2.
4.
3.3.
Credit rationing with many groups
3.4.
Some comparative statics of credit rationing
3.5.
A consumer loan model
Extensions to the basic theory

4.1.
5.
Adverse selection effects

Adverse incentive effects
Enforcement problems
4.2.
Enforcement problems in multiple period models
4.3.
4.4.
Collateral, loan size, and non-price rationing

Issues of efficiency and welfare
Macroeconomic aspects of credit rationing

5.1.
5.2.
Market interest rates and the shadow price of credit

Cyclical changes in credit rationing
5.3.
The measurement of credit rationing
5.4.
5.5.
Fixed capital investment of business firms

Credit substitutes for bank loans
5.6.
5.7.
Housing investment
Consumption expenditures
5.8.
The role of credit for monetary policy
References
Handbook of Monetary Economics, Volume H, Edited by B.M. Friedman and F.H. Hahn
@~ Elsevier Science Publishers B.V., 1990
838
839
839
841
842
845
847
849
853
855
858
859
860
862
863
863
864
866
868
869
869
873
874
876
877
88O
882
883
885
838
D. Jaffee and J. Stiglitz
1. Introduction
Credit markets differ from standard markets (for chairs, tables, and pencils) in
two important respects. First, standard markets, which are the focus of classical
competitive theory, involve a number of agents who are buying and selling a
homogeneous commodity. Second, in standard markets, the delivery of a
commodity by a seller and payment for the commodity by a buyer occur
simultaneously.
In contrast, credit (in money or goods) received today by an individual or
firm is exchanged for a promise of repayment (in money or goods) in the
future. But one person's promise is not as good as a n o t h e r - promises are
frequently b r o k e n - and there may be no objective way to determine the
likelihood that the promise will be kept. 1 After all, for most entrepreneurial
investment, the project is sui generis. The need for credit is evidence of
change: those who control existing resources, or have claims on current wealth,
are not necessarily those best situated to use these resources. They thus
transfer control over their resources to others, in return for a promise.
The analysis of credit allocation can go astray in trying to apply the standard
supply and demand model, which, as well as it may work for the market for
chairs, is not totally appropriate for the market for promises. If credit markets
were like standard markets, then interest rates would be the "prices" that
equate the demand and supply for credit. However, an excess demand for
credit is c o m m o n - applications for credit are frequently not satisfied. As a
result, the demand for credit may exceed the supply at the market interest rate.
Credit markets deviate from the standard model because the interest rate
indicates only what the individual promises to repay, not what he will actually
repay (which means that the interest rate is not the only dimension of a credit
contract).
In the United States, a complicated, decentralized, and interrelated set of
financial markets, institutions, and instruments has evolved to provide credit.
Our discussion of credit rationing will focus on one set of these i n s t r u m e n t s loan c o n t r a c t s - where the promised repayments are fixed amounts. At the
other extreme, equity securities are promises to repay a given fraction of a
firm's profits. A spectrum of securities, including convertible bonds and
preferred shares, exists between loans and equity. Each of these securities
provides for the exchange of a current resource for a future promise. In our
~Moral hazard and adverse selection- special types of behavior we will describe l a t e r - m a y
affect the likelihood of loan repayment.
Ch. 16: Credit Rationing
839
discussion we shall uncover a n u m b e r of " p r o b l e m s " with the loan market.

While some of these problems are addressed by other instruments, these other
instruments have their own problems (although we will only be able to discuss
these further problems briefly in the space available).
The issue of the allocation of credit has profound implications at both the
micro and macro levels. At the micro level, in the absence of a credit market,
those with resources would have to invest the resources themselves, possibly
receiving a lower return than could be obtained by others. When credit is
allocated poorly, poor investment projects are undertaken, and the nation's
resources are squandered. Credit markets, of course, do exist, but they may
not function w e l l - or at least they may not function as would a standard
m a r k e t - in allocating credit. The special nature of credit markets is most
evident in the case of credit rationing, where borrowers are denied credit even
though they are willing to pay the market interest rate (or more), while
apparently similar borrowers do obtain credit.
At the macro level, changes in credit allocations are one likely source for the
fluctuations which have marked capitalist economies over the past two centuries. For example, the disruption of bank lending during the early 1930s may
have created, or at least greatly extended, the Great Depression of the 1930s.
The availability of credit may also be an effective instrument for monetary
policy. One example is that the Federal Reserve often provides liquidity when
the financial system is disrupted, such as after the stock market crash of 1987.
A n o t h e r example is that the Federal Reserve has used credit c r u n c h e s enforced credit r a t i o n i n g - to slow down an overheating economy.
2. Basic notions
Our discussion of credit rationing in Section 3 relies on certain features of loan

contracts and loan markets that make the standard demand and supply model
inapplicable and that give rise to credit rationing phenomena. This section
provides background about these special features.
2.1. Uncertainty
Differences between promised and actual repayments on loans are the result of
uncertainty concerning the borrower's ability (or willingness) to make the
repayments when they are due. This creates the risk of borrower default. Some
aspects of uncertainty may be treated with the standard model, as illustrated by
the capital asset pricing model or other models where there is a fixed and
840
known probability of default. 2 Therefore, default risk is a necessary condition,

but not a sufficient condition, for the standard model to be inapplicable.
Given that borrowers and lenders may have differential access to information concerning a project's risk, they may form different appraisals of the risk.
We refer to symmetric information as the case in which borrowers and lenders
have equal access to all available information. The opposite c a s e - which we
will call generically imperfect information - has many possibilities. Asymmetrical information, where the borrower knows the expected return and risk of his
project, whereas the lender knows only the expected return and risk of the
average project in the economy, is a particularly important case.
Uncertainty regarding consumer and (risky) government loans can be described with the same format used for firms, although, of course, the underlying sources of uncertainty are different. For individuals, future income prospects and demographic factors (children in college) illustrate the type of factors
that determine repayment prospects. For governments, the willingness to repay
may matter as much as the ability to repay; we discuss this special aspect of
loan markets further below.
The response of both lenders and borrowers to uncertainty is determined in
part by the extent of their risk aversion. Lenders may use credit rationing to
reduce default risk, so credit rationing is more likely when lenders have a
higher degree of risk aversion. However, we will assume that lenders are risk
neutral, since we want to investigate other factors that may be sufficient to
create credit rationing. To simplify matters, we will also generally assume that
borrowers are risk neutral. 3
The role of uncertainty is readily evident in loan markets where firms borrow
to finance risky investment projects. We will often use a simple, but instructive,
f r a m e w o r k - where each investment project has precisely two possible outc o m e s - t o illustrate the effects of uncertainty. We will denote the uncertain
outcomes as X, with X a the larger "good" outcome and X b the lower " b a d "
outcome ( X a > x b ) . The probabilities P" and p b indicate the likelihood of
each outcome (with P~ + pb = 1). The expected value X e ( = p a x a + pbxb) is
the central tendency of the outcomes. Several important properties of this
framework are worth highlighting.
(1) The economy may contain an array of projects with varying risks and
expected returns. The risk of each project can be measured by the difference
between X" and X b. The comparison of projects is further simplified by
ZIn particular, the standard model may be applicable to credit markets which issue and trade
securities that are free of default risk. For example, U.S. Treasury securities are free of default risk
because the taxation and money printing powers of the Federal government guarantee repayment.
However, other risks, such as inflation risk, apply even to Treasury securities.
3There may be less credit rationing when borrowers arc more risk averse, but only because
risk-averse borrowers may not even apply for risky loans.
841
assuming a mean preserving spread- that all projects (all possible (X a, X b}

combinations) have the same expected value X e [Rothschild and Stiglitz
(1971)].
(2) If each firm has one project, then risk is determined at the level of the
firm. If a firm has a portfolio of possible projects, then the evaluation of a
firm's risk requires the further identification of which project(s) it carries out.
(3) Although we use the two discrete outcomes model for its analytic
convenience, most of our results carry over when projects outcomes are
described by a continuous distribution. For such cases, the probabilities
associated with the continuous set of outcomes are given by a density function
f(X).
2.2. Loan contracts

Uncertainty is important in loan markets because it determines the likelihood
of default (a broken promise for repayment). Loan contracts specify the
amount borrowed B, and the interest rate r, so that the promised repayment is
the fixed amount (1 + r)B. As a basic case, we will consider a one-period
model in which the borrower repays the loan fully when the outcome X
exceeds the repayment amount (1 + r)B, and the borrower defaults when the
outcome X is less than (1 + r)B. We will also consider multi-period models in
which more complex loan contracts may be used and where further negotiations may occur after the project outcome is known.
Default on a loan contract creates a conflict between the borrower and the
lender. In past eras, lenders relied on debtor's prisons and physical punishment
to enforce loan contracts or to punish default. Now, in the case of default, at
least for the organized loan markets, lenders have to accept the principle of
limited liability- a lender's only recourse is to try to claim the assets of the
borrower. Specifically, when default occurs and there is only a single lender,
we assume that the lender receives the realized outcome X. We comment later
on the complications introduced by the costs of bankruptcy proceedings and by
the existence of more than one creditor.
Limited liability and the uncertainty of project outcomes jointly determine
the actual repayment received by the lender. There are three basic cases,
depending on the size of the contracted repayment (1 + r)B relative to the size
of the two possible outcomes. If the contracted repayment (1 + r)B is less than
the bad outcome X b, then the lender always receives the contracted repayment. If the contracted repayment (1 + r)B exceeds the good outcome X a,
then the borrower always defaults, and the expected repayment equals the
expected outcome of the project X e.
The main case of interest occurs when the loan repayment falls between the
842
two possible outcomes, that is when X b < (1 + r)B < X a. The expected repayment is then pa(1 + r)B + pbxb, reflecting repayment of the contracted
amount (1 + r)B when the good outcome occurs (with probability p a ) and
repayment of the available proceeds X b when the bad outcome occurs (with
probability pb).
Two fundamental features of loan contracts are illustrated by this result.
First, the expected repayment rises as the loan rate rises, since the promised
repayment (1 + r)B rises as the loan rate rises. Second, the expected repayment
falls as uncertainty rises, since the expected repayment is reduced by a lower
value for the bad outcome, but is unaffected by a higher value for the good
outcome [at most, the borrower pays back (1 + r)B]. Once a specific investment project is chosen and the loan rate is set, the borrower and lender
participate in a zero-sum game, with the borrower keeping that part of the
outcome not repaid to the lender. Consequently, other things the same,
lenders prefer safer projects and higher loan rates, while borrowers prefer just
the opposite.
Loan contracts also include t e r m s - called non-price t e r m s - which constrain
the borrower in order to reduce the likelihood of default. Collateral is among
the most important of these. Collateral consists of financial and tangible capital
assets that are pledged by the borrower to guarantee at least partial, if not
complete, loan repayment. If a loan defaults, then the collateral is used to
supplement the proceeds available to the lender (up to the repayment
amount). In the extreme case of a fully collateralized loan, the collateral value
is sufficient to cover any possible shortfall in loan repayment. In practice, even
highly collateralized loans do not generally warrant interest rates as low as the
rates on Treasury securities. We will look at some of the reasons for this,
including that the value of the collateral may be uncertain and that there may
be transactions costs associated with liquidating it. 4 As a result, collateral may
reduce, but generally does not eliminate, the risk of default.
2.3. The classification of borrowers

Given that the expected repayment depends on the risk character of a project,
it is natural for lenders to classify borrowers on the basis of their risk. Under
an efficient classification system, riskier borrowers are charged higher interest
rates to account for their higher likelihood of default. Banks, for example, use
a schedule of quoted interest rates, with the safest borrowers charged the
prime rate and riskier borrowers quoted a premium over the prime rate. A
premium above prime reflects a higher likelihood of default.
4Indeed, in the absence of these problems, the borrower could sell the collateral instead of
borrowing.
843
Borrower classification - based on risk screening - is a main function of the

banking system. After appraising the risk of a firm, a bank has a basis for
setting the loan interest rate and making the loan. Alternatively, another
lender can make the loan, with the bank selling its "screening" service in the
form of a guarantee against d e f a u l t : However, competitors may free ride on
screening information when it is publicly released. Also, the credibility of bank
screening is reduced to the extent that bank funds are not at risk with respect
to the outcome. Therefore, for these and other reasons, banks generally
integrate their screening and lending activities.
Since a borrower's classification determines the interest rate charged, the
efficiency of credit allocation in the economy is dependent on the accuracy of
the classification system [see Jaffee and Modigliani (1969) and Stiglitz and
Weiss (1981)]. With our decentralized system of banking, competition among
suppliers of credit works to eliminate systematic misclassifications. For example, if a firm is classified as excessively risky by one lender, competitive lenders
may offer a lower interest rate. However, owing to the intrinsically subjective
nature of risk evaluations and judgments, competition in the market for loans
is not likely to be as "perfect" as competition in the market, say, for chairs.
Even in insurance markets, where actuarial data are available to measure risk,
significant variations are sometimes observed in the premia charged for the
same risk. All the more so, competition in loan markets will not always
eliminate errors in borrower classifications.
Competition is also limited by "customer relationships" between lenders and
borrowers. These relationships tend to be exclusive (the borrower has a single
lender), both at a moment in time and over time. The market is, to use Okun's
terminology, a customer market, not an auction market. The following comments refer to some of the loan market features that create a need for exclusive
relationships between borrowers and lenders. The first three points refer to
exclusivity at a point in time and the last three to exclusivity over time.
(1) Since the costs of information for determining the likelihood of default
are predominantly sunk costs, there is an economy of scale in funneling all of a
firm's credit needs through one lender. For similar reasons, lenders tend to
specialize in making loans to particular industries. As a result, there is a
tendency toward natural m o n o p o l y - n o t in banking as a whole, but in the
supply of credit to a particular person, firm, or industry. 6
(2) Borrowers have an incentive to take actions that raise the probability of
default, as when a firm with several investment projects chooses the most risky
SFor example, the World Bank is starting to separate its screening and lending functions by
establishing an insurance firm to guarantee loans, thus enabling countries to raise funds directly in
world capital markets.
6Risk aversion and other arguments for diversification work in the opposite direction. Large
banks, however, can make large loans and still obtain the benefits of diversification. See Stiglitz
(1985).
844
one (because it provides the highest expected return). A lender can respond by
placing restrictive clauses in a loan contract, but the borrower's actions may
not be fully foreseen and costs may arise in later renegotiating such clauses.
Alternatively, lenders can influence the borrower's incentives, without directly
monitoring the borrower's actions, by controlling the size of the loan and
related terms of the loan contract. The returns from controlling the size of the
loan, however, may accrue to all of the firm's lenders, which leads to the next
point.
(3) Several important public good problems are associated with any classification of creditors. These problems arise when events occur which provide the
creditors, as a class, with some discretion over the actions of the firm. For
instance, many loan contracts contain a clause that states that the loan is in
default if any loan of the borrower is in default. This clause ensures that the
lender can maintain control of the collateral and can represent his interests in
the event of a reorganization of the borrowing firm. These public good
problems become paramount when bankruptcy occurs and the majority of the
creditors of any class must agree to any settlement. By the same token,
monitoring the firm's activities, to ensure that these adverse contingencies do
not arise, is a public good, the return on which accrues, at least to some extent,
to all creditors ]Stiglitz (1985)]. In brief, there is a natural incentive to
concentrate the sources of credit.
(4) Continuing relationships provide cost savings for lenders who make a
sequence of loans to the same borrower, since information obtained at one
date may also be used to assess risk at a later date. These cost savings are also
important for lines of credit, which are the basis for many business, credit card,
and consumer loans] In these cases, continuing customer relationships provide
the current lender a distinct advantage over competitors in assessing the
current riskiness of its customers.
(5) As a result of the information provided by a long-term customer
relationship, competitors may fear that winning away a customer means that
the previous lender learned of adverse developments. This illustrates clearly
why Akerlof's (1970) "lemons" principle is an important factor in loan
markets.
(6) Finally, the threat of termination of credit acts as an important incentive
device to encourage the firm to undertake safer actions this period [Stiglitz and
Weiss (1983)]. It is sensible that these "moral hazard" problems can be better
handled when there are exclusive relationships [see Arnott and Stiglitz (1985)].
Exclusivity arrangements limit, but do not eliminate, competition. When a
7Credit lines allow loans to be contracted on a forward basis. A credit line m a y serve an
insurance purpose: either protecting the borrower against an unexpected change in credit rating (if
the loan rate is fixed relative to the prime rate) or, less commonly, against a change in the level of
interest rates (if the specific level of the loan rate is fixed).
845
new borrower enters the market, for example, there will be competition to
become that firm's supplier of credit. Competition is limited, however, to the
extent that the costs of gathering information are sunk costs. So, given the
amount of ex ante competition, there will be less ex post competition. Hence, a
misclassification of a borrower's risk category may not be easily corrected
through competition.
2.4. The sources of loanable funds: Banks and bank deposits

Lenders obviously need funds to make loans, so the cost and availability of
loanable funds necessarily interacts with loan market activity. We will adopt
the descriptive shorthand of referring to lenders as "banks", and their sources
of loanable funds as safe "deposits". However, the following points indicate
how the analysis generally applies to all types of lenders.
(1) Bank lending is typical of most lending. Although bank loans are the
dominant form of credit in many countries, capital market instruments, such as
bonds, are also used for credit transactions. Capital markets and loan markets
share similar problems, and solve many of them in comparable ways, as the
following examples illustrate:
(a) The underwriting process for new risky bonds is intensive in gathering
information concerning the borower's credit worthiness, and the likelihood of default is reflected in the rate of return required by investors. 8
(b) Credit ratings are used to classify risk level (although the screening and
lending functions tend to be separated more in capital market transactions than in lending transactions).
(c) Restrictive covenants in corporate bond contracts are comparable to
non-price terms in loan contracts, both reflecting attempts by lenders to
control the extent of default risk. 9
(d) Only the low-risk bonds issued by the U.S. Treasury and A A A corporations trade in standard demand and supply markets.
(2) Bank deposits can be treated as safe securities. For simplicity, we treat
d e p o s i t s - the source of loanable f u n d s - as though they were free of default
risk. This is actually the case with respect to bank deposits insured by federal
deposit insurance. Even without such insurance, depositors generally receive
the fixed return promised by a bank because banks hold diversified portfolios
8The links between bank loan markets and the capital markets are further increased by the new
process of "securitization" in which banks package portfolios of their loans for sale in the capital
markets.
9Myers (1977) has emphasized the importance of restrictive corporate bond covenants for
understanding the financing behavior of corporations (including why the Modigliani-Miller
theorem may fail to hold).
846
of loans and their capital and loan loss reserves provide a further tier of
protection. 1
(3) The banking industry is competitive. This assumption applies to free
entry into the banking system as a whole, even though each bank is likely to
maintain exclusivity in its customer relationships. Until 1980, bank competition
in the United States was limited by Regulation Q deposit rate ceilings and by
various restrictions on bank charters, mergers, statewide branching, and
interstate banking. However, deregulation legislation and related developments have now removed most of the reasons for expecting bank markets to be
less (or more) competitive than say the market for chairs. Consequently, free
entry and zero expected profits are our maintained assumption, but imperfect
competition is considered where it may be relevant.
Competition causes expected bank profits to be zero, so banks pass through
their expected earnings on loans (net of operating costs) to depositors as
interest payments. Even though deposit rates equal expected loan returns, an
additional relationship is required to determine the common level for the two
rates. One possibility is to use the market-clearing condition of the Walrasian
demand and supply m o d e l - that the demand for loans (by borrowers) equals
the supply of deposits (by depositors)- to determine the equilibrium level of
the two rates.
Another possibility, developed below in Section 3, is that special features of
loan markets cause loan rates to be endogenously determined below the
Walrasian market-clearing level. To illustrate how this might work, without
going too far ahead, let us suppose that a binding usury ceiling exogenously
restricts loan rates below their Walrasian equilibrium level. 11 Competition will
force expected bank profits to be zero, so deposit rates must settle at the same
low level as the usury ceiling (default risk aside). As a result, there will be an
excess demand for loans - a simple example of credit rationing.
Imperfect competition in deposit and loan markets does not fundamentally
change the situation. If the special features of loans are ignored, then following
the standard model of monopoly, banks will maximize their profits by equating
the marginal cost of deposits with the marginal expected return on loans, while
satisfying all loan demand. On the other hand, if the special features of loans
cause loan rates to be quoted at a level lower than the monopoly level, then a
continuing state of excess demand may exist, just as in the competitive case.
~The recent experience of S and L's, however, suggests that in the absence of deposit insurance,
the reserves and capital m a y not suffice to protect depositors. The losses that depositors suffer as a
result of bank runs raise a different issue - bank liquidity - as discussed by D i a m o n d and Dybvig
(1983).
11Although the ecclesiastical sources of usury ceilings have presumably long passed, similar
ceilings still exist in m a n y states in the United States as part of c o n s u m e r protection laws for
unsecured c o n s u m e r and credit card loans.
847
Banks may also internally connect loans to deposits when the external
markets for these securities function imperfectly. Compensating balance requirements, for example, are a contractual a r r a n g e m e n t whereby a b o r r o w e r or
a holder of a line of credit is required to maintain a specified level of deposits
in the bank. This a r r a n g e m e n t can be interpreted as a "time-sharing" plan,
whereby a firm receives the right to borrow from time to time as part of its
compensation for maintaining deposits in the bank. Such arrangements m a y
allow banks to equilibrate the supply of deposits and the d e m a n d for loans
even when Regulation Q deposit ceilings and usury loan ceilings constrain the
m a r k e t levels of interest rates. Banks m a y also use these arrangements to
monitor a firm's financial condition by observing the patterns and trends in its
deposit balances.
In s u m m a r y , loan and deposit markets involve three different interest rates:
(1) the quoted loan rate determines the amount that borrowers promise to
repay; (2) the expected return on a loan is the quoted rate adjusted for the
lender's expected loss if default occurs; and (3) the deposit rate is the rate paid
to depositors to raise loanable funds. U n d e r competition, the deposit rate
determines the expected return a b a n k must earn on each loan. That is, a bank
determines the quoted rate for each loan so that it earns the required return;
or, if this is not possible, the bank will not m a k e the loan.
2.5. Definitions of credit rationing

Credit rationing is broadly defined as a situation in which there exists an excess
d e m a n d for loans because quoted loan rates are below the Walrasian marketclearing level. T h e r e are a n u m b e r of different types of credit rationing
depending on how excess d e m a n d is defined, on whether the excess d e m a n d is
t e m p o r a r y or continuing, and most importantly, on the factors that cause the
loan rate to be depressed. Given the n u m b e r of possible variations, definitions
of credit rationing have often been a source of confusion in the literature. For
this reason, we list here several c o m m o n definitions of credit rationing and rem a r k on the extent to which they are or are not relevant to the issues we address.
(1) Interest rate (or price) rationing. A borrower may receive a loan of a
smaller size than desired at a given loan rate. To obtain a larger loan, the
borrower has to pay a higher rate.
Remark. This describes standard price rationing and is not relevant to our
discussion of credit rationing. O f course, it is sensible that a borrower has to
pay a higher rate on a larger loan since the probability of default tends to be
higher on a larger loan. In contrast, it is striking that a large part of the
literature for m o n e t a r y economics and corporate finance assumes that there
848
exists a market in which people or firms can borrow as much as they like at a
fixed rate of interest [see Stigler (1967) and Stiglitz (1970)].
(2) Divergent views rationing. Some individuals cannot borrow at the interest rate they consider appropriate based on what they perceive to be their
probability of default.
Remark. Although lenders relative to borrowers may commonly have more
pessimistic appraisals of default risk, this is also not the focus of our analysis of
credit rationing. ~2 For example, if the Treasury bond rate is 8 percent, the fact
that most firms are "excluded" from the market for 8 percent loans simply
reflects the market's judgment that loans to these firms are risky. From this
perspective, a borrower's perception of his risk is mainly relevant as a factor
determining his demand for credit.
There is an analogous kind of rationing in labor markets: there may be an
excess supply of those willing to be president of General Motors at the going
market wage, but (at least in the neoclassic story) the reason that the president
of GM receives such a high wage is that he has " m o r e efficiency units": he is
equivalent to, say, 100 ordinary laborers. A person who could only supply 50
efficiency units (even working 24 hours a day) would thereby be excluded from
the market for G M president.
(3) Redlining. Given the risk classification, a lender will refuse to grant
credit to a borrower when the lender cannot obtain its required return at any
interest rate. Moreover, loans which are viable at one required rate of return
(as determined by the deposit rate) may no longer be viable when the required
return rises. (The term redlining originally referred to the cross-hatched maps
used by urban mortgage lenders to designate neighborhoods in which they
would not lend. Our use of the term does not imply any type of discriminatory
behavior.)
Remark. The possibility of redlining is easily illustrated in an extreme
example where the borrower's project has a single outcome X m. If the required
return is 6, then a loan in excess of B* = xm/(1 + 6) clearly cannot yield the
required return, so the firm will be rationed out of the market. The same result
can apply to the expected return on a loan when there are various possible
project outcomes.
In a similar fashion, a firm, which received a loan when the supply of
deposits was high and the deposit rate was low, may be rationed when the
supply of deposits shifts and the deposit rate rises. This happens because a
"promise" to pay a higher rate on a loan may not translate fully (or even at all)
into a higher expected return for the lender. For these firms the availability of
credit (the supply of deposits) - not the quoted loan rate - determines whether
12Divergent views o n l o a n r e p a y m e n t s are c o m m o n b e c a u s e t h e y refer to the fulfillment of a
p r o m i s e , not to an a c t u a r i a l e v e n t such as w h e t h e r o r n o t it will rain.
849
they can borrow. These firms would feel they are being rationed out of the
market. 13
(4) Pure credit rationing. T h e r e m a y be instances in which some individuals
obtain loans, while apparently identical individuals, who are willing to borrow
at precisely the same terms, do not.
Remark. This is the purest f o r m of credit rationing, and we show below how
it m a y arise when there is imperfect information. When it does arise, changes
in the availability of credit, not changes in the interest rate, may determine the
extent of borrowing.
We also show that redlining (type 3 credit rationing) may be nearly
indistinguishable from pure credit rationing (type 4) when lenders classify
borrowers into a large n u m b e r of groups so that each group has a small n u m b e r
of borrowers. In this case, borrowers in a redlined group m a y have nearly the
same features as the borrowers in a group that does obtain loans.
2.6. The earlier credit rationing literature: Brief review

The literature on credit rationing can be conveniently separated into two parts:
an earlier literature based on various loan m a r k e t imperfections; and a current
literature based on imperfect information. Imperfect information was suggested as a factor in loan markets by A k e r l o f (1970) and Rothschild and Stiglitz
(1971), and was first applied in a model of credit rationing by Jaffee and
Russell (1976). We briefly review the earlier literature in this subsection before
turning to the current imperfect information theories.
The earliest references to credit rationing pertain to usury ceilings [Smith
(1776)] and to the English banking and currency controversies of the
nineteenth century [Viner (1937)]. Modern references to credit rationing begin
with Keynes' discussion of an "unsatisfied fringe of b o r r o w e r s " in his Treatise
on Money (1930, I: pp. 212-213; II: pp. 364-367):
So far, however, as b a n k loans are concerned, lending does n o t - i n G r e a t
Britain at least - take place according to the principles of a perfect market.
T h e r e is apt to be an unsatisfied fringe of borrowers, the size of which can be
expanded or contracted, so that banks can influence the volume of investment by expanding or contracting the volume or their loans, without there
being necessarily any change in the level of bank-rate, in the demand13It has been debated whether redlining should be called credit rationing since, in the market's
judgment, redlined borrowers cannot provide the bank its required rate of return (in an expected
value sense). These borrowers are "rationed" out of the market in the same way that an individual
who cannot afford a $200 000 automobile is rationed out of the car market because the lender does
not believe that the auto loan will be repaid.
850
schedule of borrowers, or in the volume of lending otherwise than through

the banks. This p h e n o m e n o n is capable, when it exists, of having great
practical importance.
Keynes did not actively pursue this notion in his later work, but it became
part of the "availability doctrine", or "new view", of monetary control at the
Federal Reserve in the United States following World War II. The availability
doctrine was primarily macroeconomic in its outlook (we consider it later in
this regard), but it started a microeconomic literature concerned with the
theoretical underpinnings of credit rationing behavior. TM
2.6.1. The risk of default as the basis of credit rationing

Hodgman (1960) was among the first to consider the risk of default as the
reason a rational bank might not raise its quoted loan rate even though it faces
an excess demand for loans. Hodgman's model looked at a risk-neutral bank
that is making a one-period loan to a firm. The firm's investment project
provides possible outcomes X, bounded by k < X < K, with the probability
density function f[X]. The contracted repayment equals (1 + r)B, based on the
loan B and the lending rate r. If default occurs, the bank receives the available
proceeds X.
The bank is assumed to obtain its funds in a perfect deposit market at the
constant interest rate 6. For the single borrower, the bank's expected profits q~
are then written:
~ l +r)B
alp=
Xf[X]dX+
f(K
I+~)8(I+r)Bf[X]dX-(I+6)B.
(2.1)
The first term accounts for the bank's income X when there is default
( X < (1 + r)B), the second term for the income (1 + r)B when the loan is
repaid in full, and the third term for the bank's cost of funds (1 + 6)B. The
bank's decision variables are the quoted loan rate r and the size of the loan B.
This formulation can be used to derive the bank's loan offer curve for the
borrower, showing the size of the loan offered B for each loan rate r. Under
competition, the offer curve is simply the zero profit locus of contracts for
which the bank's expected profit q~ is zero. The loan offer curve, as illustrated
in Figure 16.1, has three basic properties:
(1) It is a horizontal line at the deposit rate 6 over the range of small loan
sizes where the loan is risk-free (B < k/(1 + 6)).
~4This literature has been surveyed in Lindbeck (1962), Jaffee (1971), Koskela (1976, 1979), and
Baltensperger (1978).
851
\
Loan rate r
Bank offer curve
r: iiiiiiiii
io eman O
Firm demand D2
k/(l+~)
I~*
Loan size B
Figure 16.1. A bank loan offer curve and alternative cases for the position of the borrowing firm's
d e m a n d curve.
(2) It has a positive slope over the range where the probability of default
rises with the size of the loan, as higher loan rates compensate the bank for the
higher likelihood of default on larger loans.
(3) There may be a maximum loan size B* beyond which the offer curve
becomes backward bending. Otherwise, were the loan rate r and the size of the
loan B to rise together over the full range, at some point the contracted
repayment (1 + r)B would equal the firm's best possible outcome K and
default would be certain.
Freimer and G o r d o n (1965) used the loan offer curve construction to
develop a redlining theory of credit rationing. Their point was that the size of
the loan demanded by an individual borrower could exceed the size of the loan
offered at any interest rate, as illustrated by demand curve D 1 in Figure 16.1.15
Hence, the size of the loan demanded might exceed the maximum size of the
loan the lender was willing to offer. The model, however, did not carefully
delineate between the information available to the borrower and to the lender,
so that the type of credit rationing was unclear. These issues are clarified by the
theories of credit rationing discussed in Section 3 that focus on imperfect
information.
Rationing is eliminated when the demand curve intersects the offer curve, as
illustrated by the demand curve D 2 in Figure 16.1. In this case, the loan
~SThis assumes the offer curve is the locus of contracts with zero expected profits. Freimer and
G o r d o n , in fact, a s s u m e the lender is a monopolist making all or nothing loan offers, but similar
considerations apply. In particular, along the loan offer curve, the probability of default approaches 1 as the loan rate approaches infinity.
852
d e m a n d e d and the loan offered are equal at the market-clearing rate r2, so
credit rationing - in terms of the size of loan - will occur only if the quoted rate
loan is maintained below r 2. Several alternative approaches were taken to
motivate why quoted loan rates might be maintained below the market-clearing
level. However, as we will now see, they are all based on exogenous institutional factors.
2.6.2. Borrower classification

The credit rationing theory of Jaffee and Modigliani (1969) used the fact that
banks classify borrowing firms into a small n u m b e r of groups based on
objective factors such as industry affiliation and firm size. Each bank determines a single interest rate for each group, even though the firms within a
group m a y be diverse with respect to risk and the amount of their loan
demand. Banks deal with the diversity by rationing those firms within the
group which have a loan d e m a n d that exceeds the loan offer.
The d e m a n d curve D 2 in Figure 16.1 can be interpreted as applying to a
particular firm in the group for which r 3 is the group loan rate. Jaffee and
Modigliani demonstrate that even when the group loan rate is selected to
maximize bank profits (over the group), some firms in the group are necessarily
in the situation illustrated. These firms, in other words, face credit rationing
because the bank recognizes that they have above average d e m a n d or above
average risk - so the size of the loan offered to these firms is smaller than the
size they demand. A firm might prefer to pay a higher interest rate to obtain a
larger loan, but this would conflict with the purpose of the classification
scheme, namely to simplify rate-setting with just one rate for each group.
2. 6.3. Customer relationships

In a related approach, H o d g m a n (1963) and Kane and Malkiel (1965) used
customer relationships to motivate why banks set loan rates below the marketclearing level of r 2. These models assumed that firms provide banks with
deposits on a continuing basis and in return the firms receive a priority status
for their loans. A priority status for loans is valuable precisely because banks
maintain loan rates below the market-clearing level. Consequently, the model
implies an equilibrium in which banks set loan rates below the market-clearing
level, and then give priority to borrowers with established bank relationships) 6
The main shortcoming of these earlier formulations of the customer relationship theory is that they fail to explain why bank competition for deposits and
loans does not cause deposit and loan rates to reach market-clearing levels.
16A similar mechanism may occur in labor markets when firms pay wages above the marketclearing level, and then use the threat of dismissal to discipline workers [see Shapiro and Stiglitz
(1984)].
853
Federal deposit rate ceilings used to provide one explanation for why deposit
rates might be constrained, but b a n k deregulation has now eliminated these
ceilings. In contrast, the theories of credit rationing discussed in the next
section show that imperfect information provides a consistent explanation of
why loan rates might be maintained below the market-clearing level and why
customer relationships remain so important.
As another variant of the customer relationship, Fried and Howitt (1980)
provided an implicit contract theory of credit rationing in which banks provide
firms with loan rate guarantees as part of risk-sharing arrangements. The
implicit contract notion, adopted f r o m labor economics, is a generalization of
the customer relationship. Fried and Howitt applied the concept to cases where
firms acquired protection against unexpected increases in their borrowing costs,
and they showed that what may a p p e a r as credit rationing can result as long as
it is costly to b r e a k contracts. The theory is subject, however, to the same
objections raised with regard to implicit contracts as a theory of u n e m p l o y m e n t
in labor markets. 17
In s u m m a r y , the theories reviewed in this section adopt the special characteristics of loan markets - such as the likelihood of default, the use of b o r r o w e r
classification schemes by banks, and the exclusive customer relationships that
arise between banks and b o r r o w e r s - to motivate why banks set loan rates
below the market-clearing level and then use credit rationing to balance the
d e m a n d and supply of credit. T h e r e is little doubt that these are important
features of loan markets and are part of the credit rationing process. H o w e v e r ,
the theories fail to explain the origins of the features as an integrated part of
their models. In contrast, we now turn to theories based on imperfect
information, which consistently explain both credit rationing and the special
characteristics of loan markets.
3. Credit markets with imperfect information 18
In this section we present the basic argument for why there m a y be credit
rationing in markets with imperfect information. In Figure 16.2 we have
17Implicit contracts do not provide a complete theory of rigid wage rates (or interest rates)
because the allocative role of the wage rate depends on the shadow price of labor for the marginal
worker, and this price may still be variable. See, for example, Stiglitz (1986).
18Large parts of this section are based on Stiglitz and Weiss (1981). The model with incentives
effects was independently analyzed by Keaton (1979). Recent surveys of the literature are
provided by Baltensperger and Devinney (1985) and Clemenz (1986).
The fact that markets in which there is imperfect information may be characterized by
non-market-clearing equilibria was noted in Stiglitz (1976a). In labor markets, models generating
these unemployment equilibria are referred to as efficiency wage models, as a result of the close
affinity of these models with those in which wages affected labor quality because of nutritional
effects [Leibenstein (1957), Mirrlees (1975a, 1975b), and Stiglitz (1976b)]. For recent surveys, see
Stiglitz (1987) and Yellen (1984).
854
Expected return
to the bank 'I,
Quoted interest rate r

Figure 16.2. T h e quoted interest rate r* maximizes the expected return to the bank.
plotted the expected return earned on a loan as a function of the quoted (or
promised) interest rate. Raising the rate of interest charged does not result in a
proportionate increase in the receipts of the lender, because the probability of
default may rise. Indeed, if the probability of default rises enough, the return
to the lender may actually decrease, as depicted in the figure. We denote the
quoted rate at which the lender's expected return is maximized as r*. No bank
will ever charge more than r*.
To see how credit rationing - a situation where there is an excess demand for
credit - may persist, recall the standard argument for market clearing. There it
is argued that if the demand for credit exceeds the supply of credit, lenders will
raise the rate of interest charged, which will increase the supply and decrease
the demand, until market clearing is restored. But at r*, no bank has an
incentive to raise its interest rate, because doing so only reduces the return it
receives. This is why the Walrasian (market-clearing) interest rate may not be
the equilibrium interest rate. Indeed, whenever the Walrasian interest rate
occurs at a level in excess of r*, the market equilibrium is characterized by
credit rationing: each bank would find it in its interest to lower the interest rate
to r*, even though the demand for credit exceeds the supply. 19
Therefore, to obtain credit rationing, it is only required that the expected
(certainty equivalent) return received by the lender does not increase monotonically with the rate of interest charged. 2 There are two basic reasons why the
relationship between the interest rate charged and expected receipts may not
be monotonic: adverse selection effects and adverse incentive effects. These
are now discussed in turn.
19With multiple Walrasian equilibria, if there is any Walrasian equilibrium with r in excess of r*,
then there exists a credit rationing equilibrium, though there m a y also exist another marketclearing equilibrium.
~That is, if there is ever a region of non-monotonicity, there is s o m e supply of credit (funds)
function such that the Walrasian equilibrium lies in the downward-sloping portion of the curve. In
this case, there will be credit rationing.
855
3.1. A d v e r s e selection effects

As the interest rate is increased, the mix of applicants changes adversely; safe
potential borrowers drop out of the market. This is the adverse selection effect.
Stiglitz and Weiss (1981) suggested several reasons why there might be an
adverse selection effect.
They couched their analysis in terms of a specific example, where the bank
(lender) had categorized potential borrowers by the m e a n return on their
investment projects. For simplicity, all investment projects were assumed to be
of the same size, and there was no collateral. 21 The return to the lender is
q~ = min{(1 + r), X / B } ,
where X is the return to the project, and B is the size of the loan. That is the
lender either receives the promised amount, or the firm goes into bankruptcy.
They ignored bankruptcy costs, and assumed that the returns in the event of
bankruptcy would be divided a m o n g the lenders. In Figure 16.3 we plot the
return received by the lender as a function of the return on the project, X. It is
a concave function. On the other hand, the return to the borrower is
7r = max{O, X - (1 + r ) B } .
Because of limited liability, the lowest return he obtains is zero (the return
function can be modified by collateral as we indicated earlier). This is a convex
function. The analysis then has three steps:
(1) Because of the convexity of the firm's profit function, borrowers whose
The return to
the bank q~
The return to the projectX

(l+r)B
Figure 16.3. The return to the bank is a concave function of the return on the project.
21We will later remove both of these assumptions.
856
return is riskier (in the mean-preserving sense) have a higher expected profit.
We denote the riskiness of a project (firm) by 0, with higher 0 representing
greater riskiness. If (risk-neutral) borrowers have to put up a fixed amount of
equity of their own, e 0, then a project will be undertaken if
E ~ > e0(1 + 6 ) .
Since the expected profit d e p e n d s on a project's risk, there exists a risk level
such that firms with 0 > 0 undertake the investment, while others do not. We
plot ETr - (e0(1 + 6)) as a function of 0 in Figure 16.4.
(2) An increase in the quoted rate r reduces the expected profits for all
borrowers. This downward shift in the ETr curve means that the critical value
of 0 is higher, so fewer firms will demand loans. Moreover, it is the safest
firms - those with the lowest value of 0 - which drop out of the market. This is
the adverse selection effect.
(3) The negative effect on the lender's expected return can now easily be
seen. Because of the concavity of the lender's return function, the expected
return to the lender is smaller with higher 0; this follows immediately given our
assumption that all projects in the given category have the same expected
return. If the expected return to the borrower for high-risk projects is higher,
then the expected return to the lender must be lower. Thus, the total expected
r e t u r n - a v e r a g e d over all applications for l o a n s - m a y either increase or
decrease when the interest rate r is increased. There is a positive direct effect,
but a negative adverse selection effect, as the best risks (those with the lowest
0) drop out of the market.
Expected profits
of the firm
ETr-e (1+5)
Project risk @
Figure 16.4. Only firms with projects at risk level 0 or higher will carry out their projects.
857
The fact that the adverse selection effect can easily outweigh the direct effect
can be seen most easily in an example with just two groups in the population,
groups a and b, with each group having two possible outcomes. The unsuccessful outcome for both groups is a return of 0. The successful outcome for the
groups are X" and X b and the probabilities of success are pa and pb,
respectively. We further assume that
and
X a< X b
pa > p b
with
paxa = /obxb
so that the mean returns are equal but group a is safer and group b is riskier.
We also set the loan size B equal to 1. Thus, if the good group a represent a
proportion F of the total population, and all individuals apply for loans, the
mean gross return to the lender is
(l+r)/5,
where
/5=Fpa+(l_F)p
b,
while if only the high-risk individuals apply, the mean gross return is
(1 + r ) P b .
There is a critical interest rate, r*, at which the safer borrowers stop applying.
Thus, the return to the bank declines precipitously at r*, as illustrated in
Figure 16.5.
An adverse selection effect arises because a higher loan rate affects the safer
borrowers - who anticipate they will always repay the loan - more than it does
the riskier borrowers - who will recognize that the loan rate does not matter in
situations where they have to default on the loan. (At the extreme, an
individual who was fairly sure he would not pay back the loan fully would be
Expected return
to the bank q5
Both types
I pply fr
J
/
bOn]rYohiegrhrpk iy
Quoted loan rate r

Figure 16.5. The determination of the optimal loan rate.
858
almost indifferent to the promised rate of interest.) Thus, raising the loan rate r
adversely affects safer borrowers m o r e than riskier borrowers.
The argument that a higher loan rate creates an adverse selection effect m a y
b e c o m e even m o r e cogent in a competitive setting, in which loan applicants
apply to several different banks. If one b a n k then raises its quoted loan rate, it
knows that only those borrowers who have been refused loans at all other
banks will accept its loans. If the bank is not confident about its judgment of
the riskiness of the loan applicant, the fact that other banks have rejected the
given applicant conveys a lot of information. W h e n the given b a n k is quoting
precisely the same rate as all other banks, then the same information is not
conveyed when a borrowers accepts a loan. 22
3.2. A d v e r s e incentive effects

As the interest rate is increased, applicants undertake riskier projects. We call
this the adverse incentive effect. O f course, lenders could try to control directly
the actions of borrowers, but unfortunately, monitoring is costly and never
perfect. Thus, lenders must resort to indirect control mechanisms. That is,
lenders know that the behavior of a borrower is affected by the terms of the
contract, including the interest rate. This is the basis of a key result of Stiglitz
and Weiss that we will now d e m o n s t r a t e - that the risk-taking of borrowers
rises as the interest rate they are charged rises.
This is seen most simply in the case of a firm with two techniques, a and b,
which are described exactly as were the two projects in the previous subsection, the only difference being that they now reside within a single firm. The
net expected return to the firm from undertaking project i( = a or b) is
P i ( X i - (1 + r ) B ) , which declines with r.
We plot the return in Figure 16.6, where the slope of the expected return
function is - p i B . The return on the safer project a decreases m o r e with an
increase in the interest rate because the safer project is more likely to have to
pay the promised rate. The firm undertakes the project with the highest
expected return; from the figure, it is apparent that for r < r*, the firm
undertakes the safer project; for r > r*, the riskier one. [The gross return to
the bank if the firm undertakes project i is (1 + r ) P i - B, with the safer project
ZZThis argument is weaker, at least for small changes in interest rates charged, in the context of
the customer markets discussed in earlier sections. On the other hand, a bank which suddenly
charged a rate in excess of that charged by other banks would suffer a loss of reputation, and
accordingly might find it more difficult to recruit borrowers.
A bank may deal with this problem by conditioning the extension of its loan to the provision of
matching credit by some other bank. This is another illustration of the public good aspect of loan
markets.
859
Expected return
to the firm E~r
Project b (riskier): slope = - pbB
Project a (safer): slope = - paB

Quoted loan rate r
r
Figure 16.6. The safer project is chosen at loan rates below r*, the riskier project at loan rates
above r*.
undertaken if r < r*. Thus, there will be a precipitous decline in the bank's
return at r = r*, similar to that illustrated in Figure 16.5.]
3.3. Credit rationing with many groups

When there are many groups in the population (distinguished in whatever
manner), quoted rates will be set for each group to make the expected return
on loans in each group equal to the deposit rate. In competitive equilibrium, all
loans must have the same expected return, this rate representing the cost of
deposits. Thus, we can divide the groups into three types, as illustrated in
Figure 16.7.
Type 1 borrowers are completely denied credit- they are "redlined" (credit
Expected return
to the bank
Type 2 is fully served
6
.....
i is-margmal
Type 1 is redlined
.
Quoted loan rate r
r*
Figure 16.7. Credit rationing status of different borrower groups.
860
rationing definition 3) because there is no interest rate at which the lender's

expected return reaches the required return.
Type 2 borrowers are fully served because banks actively compete to make
them loans. That is, the expected return on their loans (which equals the
deposit rate) is less than the maximum expected return a bank could earn with
a higher quoted rate.
Type 3 borrowers are the marginal group. They are rationed in the strongest
sense of the term (credit rationing definition 4): some members of this group
received credit, while other, apparently identical members (in terms of observable characteristics), are denied credit. Banks obtain the required return from
members of this critical, marginal group by charging them the interest rate
which maximizes the expected return.
Reduced credit availability always has its first impact on the marginal group:
more of these borrowers become rationed, while the interest rate charged
borrowers in this group may not change at all (the expected return is already
maximized). Moreover, a sufficiently large reduction in credit availability will
be reflected as a change in the marginal group. In this case, the interest rate
will be adjusted, the old marginal group will be totally excluded from loans,
and the new marginal group will be partially excluded from loans.
Riley (1987) has suggested that the credit rationing of the marginal group is
significant only when the number of groups is small. His point is that, as the
number of groups increases, the number of borrowers in the marginal group
falls, so the extent of this credit rationing falls. In the limit, the marginal group
would consist of just one borrower.
Stiglitz and Weiss (1987b) have responded that redlining (credit rationing
definition 3) and pure credit rationing (credit rationing definition 4) may be
nearly indistinguishable when there is a continuum of groups. This happens
because there are groups just above the marginal group who are not rationed
and groups just below the marginal group who are redlined. The characteristics
of these two groups will converge when there is a continuum of groups.
Consequently, the situation is effectively one of pure credit rationing, namely
that among groups of (nearly) indistinguishable borrowers some are credit
rationed and some are not. The extent of this kind of "nearly" pure rationing
may not diminish as the number of types increases. 23
3.4. Some comparative statics of credit rationing

Changes in uncertainty have an ambiguous effect on the quoted loan rate. For
example, as the economy goes into a recession, it is plausible that the expected
Z3Stiglitz and Weiss (1987b) also point out that there are many natural extensions of their basic
model in which credit rationing occurs at several, or even all, contracts. An example of such a
model with collateral is discussed in Subsection 4.3 below.
861
return falls, given the quoted interest rate; but there is no a priori reason to
believe that the quoted interest rate at which the expected return is maximized
will either increase or decrease, as illustrated in Figure 16.8.
Consider first the incentives model. The curve relating a bank's expected
return to its quoted interest rate can be derived from a more fundamental
curve, which relates the probability of default to the quoted interest rate. It is
more convenient to depict the bankruptcy probability as a function of 1/1 + r;
for the slope of this curve is just P / ( 1 / 1 + r) = P(1 + r), the expected return
Hence, the bank's optimal interest rate is that interest rate at which the slope
of a line from the origin to the curve is maximized. An increase in uncertainty
(associated with a move into a recession) may be thought of as shifting this
curve upwards; at each r there is a higher probability of bankruptcy. But there
is no easy way of seeing whether the interest rate at which the slope is
maximized is likely to increase or decrease.
We illustrate this with the case where there are only two types of activities,
the safer (a) and the riskier (b). The critical interest rate, where the expected
return to the borrower is the same if he undertakes either project a or b, is
1 + r* = [P"X a - pbxb]/[Pa
--
pb].
(3.1)
If as the economy goes into a recession, the probability of success of both safer
and riskier projects is reduced proportionately, then there is no change in the
quoted rate of interest, in spite of the reduction in the expected return to
loans All of the adjustments must occur then through loan availability. On the
other hand, if, as the economy goes into a recession, risky projects have their
success probabilities reduced more than proportionately, then the quoted rate
of interest will increase in a recession, 24 in spite of the fact that the demand for
Expected return
to the bank ap
/
Case t
i
~
Initial return function
Case a
rb *
r*
ra *
Quoted loan rate r
Figure 16.8. In a recession, the expected return function shifts down, but the optimal loan rate
m a y increase (case a) or decrease (case b).
74
- D e f i n e z = P " / P . b T h e n we can rewrite equation (3.1) as l + r ~ = [ r X " - X
ferentiating with respect to ~-, we obtain d r * / d r = - [ X " - X b ] / [ ' c - 112> 0.
] / [ ~ ' - 1 ] . Dif-
862
loans has decreased: it is easier to discourage borrowers from undertaking

riskier loans when the threat of bankruptcy is greater. 25
Similar results obtain when adverse selection effects are the reason for credit
rationing. Again, consider the two-group case. The critical interest rate at
which the safe borrowers drop out may increase or decrease in a recession. The
greater uncertainty associated with any project means that, at a fixed expected
return for the project, the expected return to the firm has increased; but the
expected return to any project is also likely to decrease in a recession.
3.5. A consumer loan model

Jaffee and Russell (1976) developed a model of credit rationing that applies
many of these same concepts to a consumer loan market. The basic structure is
a simple two-period model in which each consumer receives current income YI
and future income Y2, and has current consumption C 1 and future consumption
C 2. A loan B is taken out to increase current consumption and is repaid with
interest (at the rate r) from the income received during the future period. The
resulting consumption pattern is then
CI= YI + B ,
(72 = Y2 - (1 + r ) B .
The basic model assumes that all consumers have the same income and
consumption preferences. Consequently, all consumers have the same loan
demand, which is assumed to be negatively related to the interest rate.
A "moral hazard" is introduced by the fact that consumer i will default on
his loan if his cost of default Z i is less than the loan repayment (1 + r)B. The
cost of default is interpreted broadly to include the lost access to further credit,
the social stigma, and similar factors. The cost of default is assumed to vary
over the population: some consumers have high Z i values, so they tend to
repay their loans (they are "honest" borrowers); while other consumers have
lower Zi values, so they may intend to default. Banks, however, cannot
identify the consumers who are likely to default. In fact, consumers who intend
to default will act exactly as their honest counterparts so as not to identify their
true character.
Loans are made through a competitive banking industry with access to
deposits at the constant rate 6. If a bank knows that F percent of its borrowers
25A11 of these results extend to the more general model, described below, where there is
collateral, and multiple forms of loan contracts offered by firms [Stiglitz and Weiss (1987a)].
Ch. 16: CreditRationing
863
will repay their loans, then its profit function is
= / ' ( 1 + r)B - (1 + 6 ) B .
Given that competition forces bank profits to be zero, the equilibrium value for
the loan rate r* will be such that
1 + r* = ( 1 + 6 ) / F .
This shows that the loan rate factor (1 + r) exceeds the cost of funds factor
(1 + 6) by an a m o u n t 1/F that compensates the b a n k for the borrowers that
default. Loans are m a d e in the size d e m a n d e d by honest borrowers at the loan
rate r*. This is a " p o o l i n g " contract since all borrowers are offered the same,
single contract.
The honest borrowers have an incentive to create a separate loan pool
because they are subsidizing the dishonest borrowers by the amount that the
contract rate r* exceeds the cost of deposits 6. H o w e v e r , the dishonest
borrowers always mimic the actions of the honest borrowers. Therefore, the
best the honest borrowers can do is to adopt a pooling contract that has a
smaller size than their loan d e m a n d at r*. Even though this contract still pools
honest and dishonest borrowers, fewer borrowers default, so the loan rate is
lower. It is easily d e m o n s t r a t e d that the honest borrowers are better off in this
26
case.
4. Extensions of the basic theory

In the preceding section we explained why an increase in the interest rate
might have adverse selection and incentive effects, so that lenders might
choose not to increase the interest rate charged, even when there is an excess
d e m a n d for credit. The basic theory has been extended in a n u m b e r of
different directions to respond to criticism of the simpler versions of the theory
and to apply the basic theory in different institutional settings.
4.1. Enforcement problems

In some important contexts - international lending - legal recourse in the event
of a default is limited. In other contexts, legal recourse simply has very high
26If the market allows only a single contract, then this pooling contract represents an equilibrium. Otherwise, the concept of equilibrium depends on whether an informed or uninformed agent
has to act first.
864
costs. In these situations, a debt c o n t r a c t - a promise to pay in the f u t u r e - is

enforced only if the borrower faces costs of default that exceed the benefits of
default. E a t o n and Gersovitz (1981) have focused on the fact that borrowers
lose their reputation for repaying if they default, z7 Access to the capital m a r k e t
has a value, say M, because borrowing can be used to smooth consumption; in
the absence of borrowing and lending an individual (or country) must adopt
m o r e expensive ways to smooth consumption. The benefit of repudiating a
debt is D when the debt has a value of D (assuming the debt is a perpetuity
which promises to pay $1 a year, and the interest rate is r and is expected to
remain there). Thus, repudiation will occur if D > M. Since lenders know this,
they will restrict loans to M, i.e. there will be credit rationing. If there is
uncertainty about the future rate of interest or the value of M, then the initial
level of lending will be set so that there is some probability of default. A n
increase in the degree of variability should lead to a reduction in the m a x i m u m
amount that lenders will be willing to lend. 28
In other contexts, a default simply results in a period during which the
borrower must search out a new source of lending. Then the penalty for default
is, of course, much smaller [Allen (1983)].
4.2. Enforcement problems in multiple period models

While the meaning of default is clear in simple two-period models, it is not
clear in multi-period models. That is, the lender may have an incentive to
postpone a default by extending additional credit to the borrower. Thus, for a
default to occur there must in effect be complicity between the lender and the
borrower: the lender must refuse to continue to extend further credit on terms
which the borrower finds acceptable. Indeed, given that real costs are associated with default (the costs of bankruptcy or corporate reorganizations), there
should be mutually agreeable arrangements that avoid default [Eaton, Gersovitz, and Stiglitz (1986)]. On this basis, it is remarkable that the n u m b e r of
defaults is so high.
Two explanations suggest themselves. First, an effective threat of cutting off
credit may have important incentive effects on borrower behavior, causing
borrowers to undertake less risky projects, thus reducing the likelihood that
27They construct a simple model in which the by now familiar perfect equilibrium entails the firm
repaying the loan.
ZSThis analysis implies that loans to less developed countries (LDCs) may not be viable (in the
absence of trade or similar sanctions). In principle, an LDC should borrow funds when the
marginal productivity of its capital is high, and repay the funds to the developed economy later.
But the LDC may not have incentive at the future date to repay its loan: access to future credit, by
assumption, is no longer important. If lenders anticipate this, they may not lend initially.
865
the economic circumstances will occur under which default is contemplated

[Stiglitz and Weiss (1983)]. But how can the threat of cutting off credit be
made convincing? And even if one bank cuts off credit, why would not other
banks provide the credit? There are two possible explanations. Stiglitz and
Weiss provide a model in which banks earn profits on their initial loans; they
take a loss on the continuation of the loan contract, but they do so because the
contractual arrangement with the borrower allows them to cut off credit when
the default circumstances occur. As a result, when credit has been terminated
by one lender, no other lender would be willing to lend to the individual.
Seniority provisions in debt contracts, as well as the information-theoretic
concerns discussed earlier in this chapter, reinforce the conclusion that there
may be high costs imposed on the borrower by the termination of his
relationship with his lender.
Second, reputation models provide a different set of explanations. Reputations affect firm behavior in two ways. First, they provide the rationale for the
enforcement of implicit contracts, such as the continuation of the customer
relationship between a bank and a borrower. Secondly, they provide an
explanation for why, given that there is imperfect information, a bank may
sometimes force bankruptcy on a borrower who is technically in default, even
though a renegotiated agreement might appear to be a Pareto improvement.
The bank knows that if customers come to believe that such renegotiations will
occur, they will undertake riskier actions, making it necessary to undertake
these renegotiations, and thus the banks' expected profits will be lowered.29
[Eaton (1986) has addressed the question of how bank reputations can be
maintained in circumstances in which the owner-managers of banks are finitely
lived. He argues that an asset is created which can be transferred to the next
generation. The firm's current owner has an incentive to maintain this asset,
the firm's reputation.]
However, the possibility of mutually agreeable renegotiations which avoid
reorganization/bankruptcy costs must be taken into account. For as Stiglitz
(1972) pointed out, there is always a conflict of interest in the presence of
bankruptcy between the interests of debtors and of shareholders. So long as
control remains vested in the shareholders, they may take actions which,
29The Eaton-Gersovitz model is effectively a reputation model in which a borrower repays his
loan to protect his reputation. Indeed, to our knowledge their paper provided one of the first
theoretical models of reputation (in which maintaining a reputation constituted a perfect equilibrium). Subsequent more formal developments in reputation models have surprisingly failed to cite
their pioneering work.
Shapiro (1983) has provided an alternative theory of reputations in which reputations are based
on simple extrapolations of past behavior. In the context of highly complex environments, in which
the characteristics of the agents vary greatly and are hard to ascertain and there may not be
common knowledge concerning the "game" which the agents are supposed to be playing, Shapiro's
behavioristic model may provide more insights into the nature of the reputation mechanism than
the game theoretic models.
866
though increasing the value of their shares, decrease the value of debt. The
amount of this activity may increase the "closer" the firm is to bankruptcy; this
explains loan contract provisions which enable control of the firm to switch to
debtors when events occur that significantly increase the probability of the firm
being unable to meet its debt obligations, even though it currently is meeting
them. (Eastern Airlines went technically into default when it failed to obtain
wage concessions from its unions.)
4.3. Collateral, loan size, and non-price rationing

A continuing criticism leveled against credit rationing theories is that it may be
possible to adjust other terms of the loan contract to eliminate the credit
rationing. Two contract terms in p a r t i c u l a r - collateral and loan s i z e - deserve
attention.
4.3.1. Collateral
Several articles have suggested that credit rationing disappears when a bank
can set collateral requirements and interest rates simultaneously. The argument
is that the bank could offer a set of self-selecting contracts [Rothschild and
Stiglitz (1971)] that would fully reveal the risk character of each borrower; in
this case, credit rationing would not occur. For example, if the market
consisted of just two types of b o r r o w e r s - good risks and bad risks - then the
good risks might choose a contract with a lower interest rate and higher
collateral requirement (because it is unlikely they will default and lose their
collateral), while the bad risks might choose a contract with a higher interest
rate and a lower collateral requirement (because there is a higher probability
they will default and lose their collateral). As a result, if the only informational
imperfection concerned which borrowers were good risks and which bad, then
the bank could design contracts that reveal this information, and credit
rationing would not occur.
The fact that banks never have perfect information concerning the characteristics of their borrowers suggests what is wrong (or irrelevant) about this
argument: the conclusion holds only if individuals differ in just one dimension
(say wealth), so that a simple set of contracts can completely separate and
identify the different groups. 3 but if the groups differ in two dimensions (risk
aversion and wealth), then a perfect separation cannot be made with {interest
3Even then there may exist problems when there are a large number of groups. Under
monopoly, there may be partial pooling [see Mirrlees (1971), Stiglitz (1977), and a large
subsequent literature analyzing conditions under which the relevant functions are differentiable],
while under competition, there never exists an equilibrium.
867
rate, collateral} contracts. Of course, if individuals differ in just two dimensions, then it might still be possible to find a more complicated contract that
would perfectly identify the different groups. But so long as the dimensionality of the space of borrower characteristics is larger than the dimensionality of the space of contracts, it seems unlikely that perfect information can be obtained. 31
Stiglitz and Weiss (1981) showed that while increasing collateral requirements have a positive incentive effect, they could have a negative selection
effect. They argued that even if all individuals in society had the same utility
functions, wealthier individuals will, in general, be willing to take greater risks
(based on decreasing absolute risk aversion). Moreover, among those with
large amounts of wealth, there is likely to be a larger proportion of risk-takers:
individuals who gambled, and by chance won. Thus, as a result of such adverse
selection effects, it may not be desirable to require collateral to the point
where credit rationing is eliminated.
Stiglitz and Weiss (1986, 1987b) have also examined a model in which
adverse selection and incentive effects are both present, and in which interest
rates and collateral requ,.'rements are both used. They find that the equilibrium
may take on different forms:
(1) There may be a pooling equilibrium with credit rationing in which both
types of borrowers (high-risk and low-risk) adopt the same contract; collateral
is not increased, since doing so would have adverse selection effects; interest
rates are not increased, since doing so would have adverse incentive effects.
(2) Alternatively, equilibrium may be characterized by multiple contracts,
with credit rationing on each one. Some rich individuals who undertake risky
projects may have to accept low collateral, high interest rate contracts because
they do not receive high collateral, low interest rate contracts. The fraction of
those of each type at the low collateral contract is determined endogenously in
such a way to ensure that the expected return on a low collateral contract
exactly equals the expected return on a high collateral contract.
Since a loan application has no cost in the model, the extent of credit
rationing on different contracts does not serve as a screening device. However,
the role of rationing as a screening device can be examined using straightforward adaptations of similar models in the context of labor markets [Nalebuff
and Stiglitz (1983)].
3aSome of the papers that attempted to show that collateral requirements can eliminate credit
rationing did so by developing examples in which credit rationing does not occur. However, Stiglitz
and Weiss (1981) had pointed out earlier that credit rationing would occur only if the adverse
selection effect outweighed the direct effect of an increase in the interest rate. T h u s , creating an
example to show that credit rationing does not occur is an exercise directed at an irrelevant
question; they would have to show that credit rationing does not arise in any "plausible" model.
See, for instance, Bestor (1985).
868
4.3.2. Smaller loans

It has also been suggested that banks can substitute smaller loans for credit
rationing. To the extent that there are projects of fixed (or optimal) investment
scale, reducing loan size simply forces borrowers to put up more of their own
equity, and the effects of this are analogous to those just described for
collateral .32
But even when there are no non-concavities in the production technology,
the nature of the credit relationship may cause banks to avoid making loans
that are "too small". Loan size is an attribute of the loan contract which affects
borrower behavior; by reducing loan size, borrowers may take actions which
actually reduce the expected return to the lender (when appropriate account of
further "forced loans" are taken into account) [see Stiglitz and Weiss (1981)].
For example, in a multi-period context, borrowers always are in a position to
return to the lender to ask for further funds; a refusal to lend further funds
may then jeopardize the amounts already lent o u t . 33
4.4. Issues of efficiency and welfare

There are several basic results regarding the welfare aspects of equilibrium
with credit rationing:
(a) A market equilibrium with credit rationing (as in other cases in which
markets are characterized by imperfect information and incomplete markets) is
not Pareto efficient in general, even when account is taken of the costs of
information .34
(b) Pareto efficiency may well entail credit rationing (in other words, market
clearing is not, in general, an attribute of efficient markets in the presence of
imperfect information).
(c) When credit rationing occurs in markets, there are systematic biases
against undertaking projects which maximize expected returns.
Although our analysis has focused on competitive markets, informational
costs are likely to result in imperfect competition among banks (particularly ex
post, after the customer relations have been initiated). Credit rationing may
(will) characterize imperfectly competitive lending markets as well as competitive lending markets.
32Indeed, in some simple versions of the credit rationing model, increasing equity requirements
and increasing collateral requirements are equivalent.
33This is the point we have emphasized earlier in our discussion of lending with enforcement
problems. It was originallynoted in an important paper by Hellwig (1977) and in Stiglitz and Weiss
(1981).
34They are, to use the fashionable phrase, not constrained Pareto efficient. See Greenwald and
Stiglitz (1986a) and Stiglitz and Weiss (1981, 1983).
869
5. Macroeconomic aspects of credit rationing

The m a c r o e c o n o m i c ramifications of credit rationing have been a focus of
attention at least since the d e v e l o p m e n t of the Availability Doctrine at the end
of World War II. A t that time, there was pessimism regarding the effectiveness
of m o n e t a r y policy for two reasons:
(a) real expenditures were thought to be relatively unresponsive to interest
rates; and
(b) the Federal Reserve was under pressure to maintain stable (and low)
interest rates. 35
In this context, R o o s a (1951) and others at the Federal Reserve developed
the Availability Doctrine, according to which m o n e t a r y policy could still affect
real expenditures through a credit rationing channel. The Availability Doctrine
relied on three steps.
(1) Federal Reserve open m a r k e t sales of Treasury securities would cause
banks to reallocate their portfolios f r o m loans to Treasury securities. 36
(2) Banks would tend to reduce the quantity of loans through credit
rationing, not by raising loan interest rates.
(3) As credit rationing rose, rationed firms would face a rising shadow price
for credit, causing their investment activity to fall, even though m a r k e t interest
rates were quite stable. 37
We have seen in the previous sections how a theory of credit rationing has
developed that is consistent with the first two steps. We will now look at
research that has focused on step 3 - the macroeconomic effects of credit
rationing, particularly with regard to the fixed-capital investment of firms and
the housing investment and consumption expenditures of households.
5.1. Market interest rates and the shadow price o f credit

The connection between credit rationing and capital investment is illustrated in
Figure 16.9. The e c o n o m y is initially at point E 0, where the quantity of loans
Q0 is supplied by the banks (based on their supply curve So) at the loan interest
rate r 0. T h e r e is credit rationing here because the d e m a n d for loans (based on
35An important record of these views is provided in testimony provided by economists and the
Federal Reserve before the Joint Economic Committee (1952), the Patman Hearings.
36Lindbeck's (1962) survey of the Availability Doctrine discusses a variety of factors that cause
this shift.
37Modigliani (1963) particularly emphasized the shadow price of credit as the mechanism that
transmitted the impact of credit rationing from the loan markets to the real markets. The idea that
monetary policy was asymmetrical in its effect - you cannot push on a string - also was emphasized
in the Availability Doctrine.
870
Loan
rate r
T]...... ~.......
i
Loan
TOiiilrl
Q2
Q1 Q0
Loanquantity
Figure 16.9. A leftward shift in the loan supply curve raises the shadow price of loans.
d e m a n d curve Do) exceeds Q0- The shadow price of credit, r0, is determined
on the d e m a n d curve D O by the loan quantity Q0.
Now suppose that Federal Reserve action causes the supply curve of loans to
shift to the left, to curve S 1. This establishes a new equilibrium at a point such
as El, with a lower loan quantity Q1, a higher loan interest rate rl, and a
higher shadow price of credit r~. Moreover, the less the loan rate rises, the
m o r e the shadow price of credit rises. For example, if the loan rate were to
remain unchanged at the level r 0 - which the credit rationing theory in Section
3 indicates is quite possible - then the loan quantity falls to Q2 and the shadow
price of credit rises to r 2.
The actual pattern of real interest rates in the U.S. e c o n o m y since 1952 is
shown in Figure 16.10. The real interest rate is measured by the 6 to 9 m o n t h
commercial p a p e r rate minus the inflation rate for the Consumer Price Index.
Three distinct periods are evident.
(1) F r o m 1952 to the early 1970s the real interest remained in a narrow band
centered on 2 percent. (William McChesney Martin was Chairman of the Fed
for most of this period.)
(2) During most of the 1970s the real interest was negative, usually about
- 1 percent. (Arthur Burns was Chairman of the Fed.)
(3) During most of the 1980s the real interest rate was positive, sometimes
above 5 percent. (Paul Volcker was Chairman of the Fed.)
In the absence of credit rationing, the (before-tax) cost of capital to firms is
determined by the real interest rate. The cost of capital for a new capital
871
2-
-1
-2
-3
1962
1~
la6a
1~1
1~4
1~s7
197o
1973
1976
1979
1~2
l~s
Figure 16.10. The real interest rate: 1952-87.

investment that is initiated each y e a r and that provides a return of an equal
a m o u n t o f o u t p u t at the end o f each o f the following 5 years is illustrated in
Figure 16.11. 38 T h e cost of capital s m o o t h s out the y e a r - t o - y e a r fluctuations in
the real interest rate, so that over the 20-year period ending in the early 1970s,
the cost of capital was virtually constant at 2 percent. A f t e r that, the cost of
capital first fell and then rose in a s m o o t h fashion.
T h e variations in the cost of capital can be contrasted with the variations in
real, fixed, non-residential capital investment that o c c u r r e d over the same
period o f time, as illustrated in Figure 16.12. This figure shows that the annual
p e r c e n t a g e c h a n g e in the a m o u n t of real investment was often in the range
b e t w e e n - 2 and + 1 0 percent. T h e a m o u n t of variation in capital investment
appears to be a b o u t the same in the period prior to the early 1970s and the
period after that time.
T h e implication of Figures 16.11 and 16.12 is that variations in the cost of
capital a p p e a r p r i m a facie to be implausible as a m a j o r source of the o b s e r v e d
fluctuations in capital investment. 39 T h a t is, a t h e o r y of capital investment
c a n n o t be based on a constant - which the cost of capital roughly r e p r e s e n t e d
3SFor example, the capital investment in 1952 provides 1 unit of output at the end of each year
from 1952 to 1956. The cost of capital is the constant annual rate that creates the same present
value for this investment as the actual real rates shown in Figure 16.10.
39The point would be made even stronger by using an after-tax cost of capital to the extent that
the tax rates applicable to capital investment are reduced when capital investment is low.
872

5-YF-.AR
COST
OF
CAPITAL
4-
3-
-1
--2
1962
1955
1968
'
1961
1964
1967
1970
197.3
1976
1979
i
19e2
Figure 16.11. The cost of capital on 5-year investments.
CHNdOE
18
16
14
12
.~'
-,
_,
-6
-8
-10
-12
1962
w
1965
i
1968
1
1961
i
1964
l
1967
i
1970
i
1973
i
1976
i
1979
19e2
19e5
Figure 16.12. Annual percentage change in fixed non-residential investment.
873
throughout the 1950s and 1960s. Other factors, including credit rationing, must
therefore account for the observed range of variation in capital investment. We
will now evaluate credit rationing as a factor that may affect capital investment.
5.2. Cyclical changes in credit rationing
In our earlier discussion we analyzed the comparative statics of the credit

rationing model, with a particular focus on how interest rates wo~uld be affected
by changes in mean returns, uncertainty, and the availability of credit. The
analysis was, however, static, and a full cyclical analysis requires a more
explicitly dynamic model, yet to be formulated. Here, we offer three observations.
We began by discussing why the conventional "auction-market" analysis for
chairs and tables is inappropriate for the market for loans. An analogy may be
useful. Some might find it appealing for American universities to auction the
space in each incoming class to the highest bidders. After all, the individuals
are presumably well informed concerning their own abilities- they know how
hard they had to work to obtain the A - and might be the best judges of the
expected value added of, say, a Harvard education. Also, the losses created by
bad judgments would be borne by the individuals who over (or under) bid.
Standard efficiency arguments would thus seem to call for such auctions.4
However, none of the major private universities uses an auction system.
Presumably it is not believed that individuals are good judges of their own
ability. One reason might be that an auction system would have a systematic
bias; individuals who are over-confident (individuals that many of us as
teachers find particularly objectionable to instruct) would be over-represented.
Rational expectations arguments might well counter that, were these individuals to live the 9 lives of a cat, but with memory of only the return to learning,
not the learning itself, crossing over from life to life, they would not make
these mistakes. However, modern Western ideology has cast skepticism on
views of repeated reincarnations, which makes a theory of resource allocation
based on that premise particularly problematical. Accordingly, it is not surprising that non-price rationing systems have persisted at all of the major private
universities, in spite of the eminence of the economists on their faculty and the
ready availability of advice on how to improve the efficiency of resource
allocations.
Similar arguments hold for credit markets, with particular force in changing
environments. We argued earlier that credit is used when there is a discrepancy
between the availability of current resources and judgments concerning the
4Problems caused by lack of resources could presumablybe rectifiedby loan programs.
874
returns on potential uses of resources. But are individual judgments to be

trusted? Assume that interest rates were the only dimension of loan contracts
that individuals cared about. In undertaking projects, individuals compare the
expected returns with the expected interest payments. An individual who,
when confronted with a high current (short-term) interest rate, believes that
future interest rates are going to go down, may be willing to undertake a
project; while another individual, with an equally good project, who believes
that future interest rates are going to remain the same, will not. The bank does
not want to select among alternative borrowers on the basis of their projections
of future interest rates; it wishes to discriminate on the basis of the quality of
the project. But willingness to borrow (at any interest rate) may be affected as
much by the borrowers' optimism concerning future interest rates, prices, and
wages as it is by the "real" characteristics of the project. The price system
would work no better in allocating capital resources than it would in allocating
scarce places in our major private universities.
The second observation is that the response of real expenditures to credit
rationing can follow a variety of dynamic patterns.
(1) Expenditures may respond with variable lags depending on the initial
liquidity of the spending units and on their willingness to use this liquidity.
(2) The anticipation of future credit rationing may have current effects, even
when there is no credit rationing at present. Thus, the impact of credit
rationing cannot be assessed just by looking at those periods in which there is
direct evidence for its presence.
(3) There will be further multiplier effects of the initial changes in spending
induced by credit rationing.
The third observation is that the loan interest rate is not necessarily a good
measure of monetary tightness. For example, as we showed in the discussion of
the comparative statics of credit rationing in Subsection 3.4, the quoted loan
rate and the expected return on loans may even change in opposite directions.
5.3. The measurement o f credit rationing
A major pitfall in evaluating the effect of credit rationing on capital investment

is that the amount of credit rationing in the economy at a given time is not
readily measured. In principle, the amount of credit rationing could be
measured as the demand for credit minus the supply of credit. However, we
generally only observe the quantity of credit that is transacted (or that is
outstanding), not the amount that is demanded or supplied. This is illustrated
in Figure 16.13 for the interest rate r 0, at which the quantity that is transacted
might possibly correspond to the supply curve (at Q~), to the demand curve (at
Qd), or to any quantity in between, depending on how the difference between
875
Loan rate r
Loan supply
r0,.
Loan demand DO
()s
Qd
Loan quantity
Figure 16.13. The observed loan quantity is the minimum of loan demand and loan supply.
demand and supply is allocated. Whatever the specific case, if demand and
supply are not directly observed, then the amount of credit rationing is not
directly observed.
Nevertheless, Fair and Jaffee (1972) developed a basic technique for estimating the demand and supply curves for a market that may be in disequilibrium.
They assumed that transactions are voluntary, in the sense that the quantity
transacted Qt must be the minimum of demand and supply:
Qt = min{ Qa, Qs}.
In this case, only the darkened portions of the demand and supply curves in
Figure 16.13 would actually be observed. But it is still possible to obtain
maximum likelihood estimates of the demand and supply curves, and therefore
of the amount of credit rationing. Indeed, a large literature now exists for
estimating markets in disequilibrium, including applications to loan markets. 4a
There have also been attempts to measure credit rationing with proxy
variables and survey techniques: Jaffee and Modigliani (1969) developed a
proxy measure of bank credit rationing based on the percentage of banks loan
that are granted to low-risk firms; and the Federal Reserve carries out a
quarterly survey of the terms of bank lending that refers to availability of credit
[see Harris (1974) and Jaffee (1971)]. Although disequilibrium estimation,
proxy variables, and surveys all generally confirm that banks use non-price
rationing, these techniques do not directly provide evidence of the impact of
41The theory of disequilibrium estimation is developed in Maddala (1983) and Goldfeld and
Quandt (1975). Specificapplications to loan markets are discussed by Bowden (1978), Laffont and
Garcia (1977), and Sealey (1979).
876
credit rationing on real expenditures. We will now consider the evidence

regarding different classes of investment and consumption spending.
5.4. Fixed capital investment of business firms

The neoclassical theory of capital investment focuses on changes in the cost of
capital as a key variable determining changes in investment demand [Hall and
Jorgenson (1967), Eisner and Nadiri (1968), Bischoff (1971), and Clark
(1979)]. Consistent with the Modigliani-Miller theory, the theory assumes that
firms can borrow whatever funds they need to carry out their desired capital
investment. In contrast, if credit rationing limits a firm's access to external
finance, then the firm's liquidity and internally generated cash flow may
determine how much of its investment demand is realized as actual capital
investment [Meyer and Kuh (1957)]. Furthermore, credit rationing would have
a larger impact on firms that have a higher ratio of investment demand to
internally generated cash fow.
The relationship of capital investment to internal cash flow is the focus of
recent research by Fazzari, Hubbard and Petersen (F-H-P) (1988). They
extend a q theory of investment to include imperfections in loan and equity
markets that may cause firms to be credit constrained. Their empirical analysis
compares the investment behavior of classes of firms that vary in the extent of
their need for, and access to, external finance, based on panel data (from the
Value Line database) for individual manufacturing firms.
The F-H-P analysis shows that the investment expenditures of certain firms
display "excess sensitivity" to variations in their cash flow. In response to lower
interest rates or other factors that raise investment demand, the increase in the
investment expenditures of these firms is closely tied to the amount of their
internally generated cash flow. By implication, these firms do not have
substantial and continuing access to external financing.
F-H-P identify the firms that are hampered by inadequate access to external
finance as firms that retain most of their earnings- they pay little or no
dividends. Although the earnings retention rate is an endogenous variable for
firms, and therefore it is not an ideal basis for classifying them, the F-H-P data
indicate that these firms are also distinctive as being fast growing, relatively
small, and highly profitable.
All of the firms studied by F-H-P are relatively large- even their group of
small firms had an average capital stock of $347 million in 1984 (in 1982
prices). This means that about 80 percent of all manufacturing firms in the
United State are smaller than the F-H-P group of small firms. Since the
sensitivity to internal cashflow is presumably greater for smaller firms, this
reinforces the F-H-P conclusion that the investment expenditures of many firms
may be limited by the amount of their cash flow.
877
On the other hand, the firms studied by F-H-P also represent about 85
percent of the total assets held by manufacturing firms. This highly skewed
distribution of firms by size suggests that the contribution of smaller firms to
aggregate investment may be limited. That is, even though credit rationing may
have a significant impact on the investment spending of smaller firms, these
firms might account for only a small part of the fluctuations in aggregate
investment. Additionally, if the large firms that dominate the statistics for
aggregate investment are rationed by banks, they might still have access to
non-bank capital markets for equity and debt or to other sources of funds. We
will now consider some of the reasons that raising funds in the capital markets
may not substitute for bank credit.
5.5. Credit substitutes for bank loans

For credit rationing to be effective in reducing investment, firms must be
unable to raise capital in other ways, or the (marginal) cost of raising capital in
other ways must be higher than the cost of borrowing. Equity capital represents an obvious alternative to bank loans for this purpose. Also, trade
credit - loans from one firm to another - may provide firms with an alternative
source of working capital. We will now look at these in turn.
5.5.1. Equity capital as a substitute for bank credit

For (the manager of) a firm, equity has the distinct advantage that since there
is no fixed commitment to repay, there is no bankruptcy risk. Equity has,
however, three problems:
(a) Monitoring. Equities represent promises to pay shareholders a given
fraction of the firm's profits. But it may be difficult for shareholders to monitor
what the profits are. Or the firm's managers may divert a large fraction of the
firm's returns to their own uses. Alternative control mechanisms (voting,
takeovers) have only limited efficacy, partially because of free-rider problems
and the costliness of information. (The benefits associated with monitoring
accrue to all shareholders.)
(b) Signalling (adverse selection effects). Issuing equities may provide an
adverse signal to the market. If firms were risk neutral (and there were no costs
of bankruptcy), the owners of the firm would only be willing to sell a share of a
firm whose actuarial value is say, $X, for a price which equalled or exceeded
$X: so there would not exist a market for equities. 42 With bankruptcy
costs/manager risk aversion, there are gains from the issue of equities, but
42This is an application of the standard Akerlof (1970) lemons model to equities.
878
there will be a significant signalling effect, with only the worst firms (within any
category of firms) issuing equity. 43"44
(c) Agency (adverse incentive effects). The costs of monitoring managers'
activities imply that managers may have a wide discretion. This discretion is
limited by a large supply of outstanding debt in two ways. First, if managers
wish to avoid bankruptcy, then the necessity of meeting these fixed obligations
alters their actions. Second, and perhaps more importantly, banks (lenders)
monitor the actions of the firm directly, paying particular attention to actions
which adversely impact the likelihood that the loan will be repaid. They have
an effective t h r e a t - cutting off credit and demanding the repayment of outstanding l o a n s - not available to those who supply equity funds. This means
that effective control of managers is exercised as much by lenders as it is by
shareholders .45
As a result of these factors, firms may be reluctant to raise capital by issuing
new equities, even (or particularly) when they are credit constrained. Greenwald and Stiglitz (1986a, 1986b) refer to this as an equity constrained firm;
although firms have the option of issuing equity, the cost of doing so is so high
that they act as if they cannot do so. (Clearly, for large enough perturbations
to their environment, the constraint will no longer be binding.)
For many purposes the whole set of financial constraints is relevant; that is,
as a result of informational asymmetries, funds inside the firm are less
expensive than outside funds, whether obtained in the form of equity or in the
form of debt. Raising funds provides a signal, and outsiders' judgments, at
least in some cases, are likely to be more pessimistic concerning the firm's
prospects than insiders' judgments (or, in any case, less confident). Thus, what
the outsider expects to get may be less than the insider thinks he has promised
to give. 46 (Even in rational expectations models with risk neutrality, where
those who pay more than the market expects them to pay are exactly offset by
those who pay less, the informational asymmetries have real effects not only on
43This is demonstrated in the theoretical models of Stiglitz (1982), Greenwald, Stiglitz and Weiss
(1984) and Myers and Majluf (1984). Empirical confirmation is provided by Asquith and Mullins
(1986).
A n important aspect of the issue of equities is that they tend not to be issued continuously, but
rather in discrete amounts. With few exceptions, firms that enter the equity market do so only
infrequently. A dynamic model of equity issue which explains this is provided by Gale and Stiglitz
(forthcoming).
44Earlier models focused on the incentive of the original owners of the firm to sell their shares.
The willingness to do so had an adverse selection effect [Stiglitz (1974, 1982) and Leland and Pyle
(1977)]. Ross (1977) has argued that a firm's willingness to issue debt, and thus to incur
bankruptcy costs, has a positive signalling effect. In each of these instances, one needs to ask
whether there are alternative ways in which the given information can be credibly disclosed without
the costs (bankruptcy, lack of diversification, etc.) associated with the choice of financial structure.
45This is an old view. For a more modern rendition of the argument, explaining why the
free-rider problems which limit the effectiveness of shareholder control do not apply with equal
force to lenders, see Stiglitz (1985).
46This point was emphasized in Stiglitz (1972).
879
the allocation of capital, but also on the magnitude of investment; the increase
in investment by those who underpay for credit may be less than the decrease
in investment by those who overpay.)
5.5.2. Trade credit
Trade c r e d i t - loans between firms that are used to finance the purchase of
materials and goods in p r o c e s s - represents a way of redistributing credit
among firms, which could also serve as a substitute for bank loans [Meltzer
(1960), Jaffee (1971), and Duca (1986)]. The idea is that large firms, which are
not rationed by banks (or which use debt and equity capital markets), can
make loans to smaller firms that cannot obtain bank loans. Trade credit is a
major element in the U.S. credit system: trade credit outstanding is almost as
large as bank loans to business firms and as corporate bonds. 47
The price charged for trade credit by a lending firm is generally very high,
usually translating to an annual interest rate in excess of 24 percent. 48 This
implies that firms that borrow through trade credit have faced credit rationing
with respect to bank loans. In addition, most lending firms rarely change their
trade credit interest rate, so trade credit outstanding is basically determined by
the demand for the credit. (Of course, firms with very low credit ratings may
be denied access to trade credit.)
The existence of this trade credit system raises two basic questions:
(1) Why do lending firms extend trade credit, given that the borrowing firms
are likely to have been denied credit by banks?
(2) Why are lending firms willing to extend more trade credit when bank
credit rationing rises?
Earlier research tried to answer these questions by analyzing trade credit as a
sales expense (like advertising) of firms operating in imperfectly competitive
markets [Nadiri (1969) and Schwartz and Whitcomb (1980)]. However, a more
interesting and complete answer is provided by analyzing trade credit with the
same model we used in Section 3 to analyze the lending decisions of banks [see
Duca (1986)].
The key point is that trade credit allows for an efficient diffusion of
information in the form of risk-sharing arrangements that eliminates elements
of moral hazard. The idea is that firms which are selling goods to other firms
are likely to have better information about the profit prospects of the buyers
than would a bank lending to the same buyers. Trade credit can thus serve as a
risk-sharing arrangement between buying and selling firms.
47Accordingto the Flow of Funds accounts, the outstanding debts of non-financialbusiness firms
at the end of 1987 included trade debt of $546 billion, bank loans of $618 billion, and corporate
bonds of $790.
48For example, a firm may offer a 2 percent discount from its list price if payment is made
immediately (within 5 days); otherwise the list price must be paid within 30 days.
880
The result is that trade credit contracts and bank loan contracts can
represent two parts of a separating equilibrium: all borrowers first apply for
bank loans, and those who do not receive bank loans (due to redlining or to
pure credit rationing) then apply for trade credit. Firms that extend trade
credit will thus anticipate that their borrowers will have above average credit
risk, and therefore they would charge an above average interest rate. Moreover, since the trade credit interest rate mainly represents a premium for the
high risk of default, it would not be highly sensitive to the market interest rate.
5.6. Housing investment

Between World War II and 1980, housing investment was often the most
volatile component of aggregate expenditures. During most business cycles,
housing investment fell during the recession and rose during the boom by more
than 50 percent relative to its average level. At the same time, nominal
mortgage interest rates tended to be more stable than comparable bond
interest rates. Under these circumstances, credit rationing was considered to be
a major source of the observed fluctuations in housing investment. Both the
risk of default and government regulations were possible sources of this credit
rationing.
49
5. 6.1. The risk of default and mortgage credit

Most mortgage lenders try to control the risk of default on mortgages by using
two key ratios as their lending criteria.
(1) The equity ratio equals the borrower's equity investment in a real estate
property divided by the appraised value of the property Typical equity ratios
range from 5 to 25 percent Lenders charge lower interest rates on loans with
higher equity ratios because there is a greater likelihood that the housing
collateral will cover the loan amount should the borrower default
(2) The income ratio equals the carrying cost of the property (mortgage
interest, property taxes, etc.) divided by the borrower's income. Typical
income ratios range from 25 to 35 percent. To satisfy the income ratio
criterion, a borrower may have to buy a smaller house or make a larger
downpayment. Lenders use an income ratio to supplement the equity ratio
because real estate values can fluctuate and the transactions costs of selling a
property are substantial
Credit rationing in the mortgage market is a good example of a situation
where the redlining form of rationing merges into pure credit rationing
49jaffee and Rosen (1979) review the earlier housing investment cycles and the research that
pointed to credit rationing as a source of these cycles.
881
Specifically, young, first-time h o m e b u y e r s often do not have the equity or the

income to satisfy even the m i n i m u m requirements for the equity and income
ratios. Although this appears to be redlining, it occurs because lenders adopt a
small n u m b e r of equity and income ratio categories. If transaction and
information costs were lower, then lenders would adopt a larger n u m b e r of
categories, and the rationing would take the form of pure credit rationing, not
redlining.S
5.6.2. G o v e r n m e n t regulations a n d m o r t g a g e credit
Until 1980, two sets of g o v e r n m e n t r e g u l a t i o n s - usury ceilings on mortgage

interest rates and Regulation Q ceilings on bank deposit r a t e s - created an
additional source of credit rationing in the mortgage market. The Regulation Q
c e i l i n g s - which set limits on the deposit rates of commercial banks and thrift
institutions - created disintermediation - an outflow of deposit funds whenever m a r k e t interest rates rose above the ceilings. This h a p p e n e d with
some frequency during the 1960s and 1970s, with the result that the total
amount of lending by banks and thrifts had to be reduced. Credit rationing
arises under these circumstances to the extent that lending rates are constrained by the considerations of adverse selection and moral hazard discussed
in Section 3.
T h e r e were also reasons to expect disintermediation to have a particularly
strong impact on the mortgage market, one aspect of which could be credit
rationing.
(1) A b o u t half of all mortgage lending in the United States is carried out by
thrift institutions - savings and loan associations and savings banks. By regulation, these institutions raised most of their funds through deposits that were
subject to Regulation Q ceilings and invested most of these funds in mortgages.
As a result, an outflow of deposits due to disintermediation had a particularly
sharp impact on mortgage lending.
(2) Although large commercial banks had access to sources of funds that
were not affected by Regulation Q ceilings, they did not expand their share of
the mortgage m a r k e t significantly. One factor was that most mortgage loans
were fixed-rate, long-term contracts, so if these loans were made on the basis
of short-term deposits, a loss would result if interest rates rose. Most large
banks tried to balance the maturities of their assets and liabilities rather
closely, and this limited the amount of their mortgage lending (wisely so, as
5The amount of mortgage market credit rationing is also offset by the government's FHA and
VA mortgage programs. Evidence of credit rationing based on default risk is provided in recent
studies by Rosenthal, Duca and Gabriel (1987) and Duca and Rosenthal (1988). The studies
indicate that borrowers use more costly or cumbersome VA and FHA mortgages when they are
credit constrained in conventional mortgage markets.
882
evidenced by the huge losses suffered by thrift institutions when interest rates
soared in the early 1980s).
(2) Many states had l a w s - usury ceilings- which set maximum limits on
mortgage interest rates. When binding usury ceilings kept mortgage interest
rates below the level of rates on other loans or securities, lenders simply
stopped making mortgage loans. Furthermore, state legislatures were usually
slow to raise the usury ceilings, so the amount of credit rationing tended to rise
as market interest rates rose during the 1960s and 1970s. 51
The situation changed significantly beginning in 1980 when Congressional
actions (1) removed the state usury ceilings and (2) phased out all Regulation
Q ceilings over a 5-year period. These actions basically eliminated disintermediation and usury ceilings as a source of mortgage market credit rationing. 52
Other related changes in the mortgage market during the 1980s have also
tended to reduce the amount of credit rationing:
(a) The facilities for secondary market trading of seasoned mortgages have
improved, allowing lenders to separate the activity of making mortgage
loans from the activity of holding mortgage loans.
(b) A variety of mortgage pass-through securities (such as GNMAs) and
mortgage-backed securities now allow portfolios of mortgages to be
packaged for investors in a form comparable to other capital market
securities.
(c) Adjustable-rate mortgages have been developed to allow investors with
short-term funds to hold mortgages without facing a severe maturity
mismatch.
(d) Non-banking firms have entered the mortgage m a r k e t - for example,
General Motors Acceptance Corporation ( G M A C ) is now one of the
largest mortgage lenders in the United States.
5. 7. Consumption expenditures
It has been long debated in macroeconomics whether consumption depends on
current i n c o m e - the simple Keynesian t h e o r y - or on "wealth" (or the discounted value of future income) - the permanent income and life cycle theories
[Tobin and Dolde (1971)]. The existence of credit r a t i o n i n g - called liquidity
constraints in the consumption literature - is the factor that distinguishes these
5~Asmarket interest rates rose, banks would first ration more risky borrowers (those with lower
equity ratios and therefore higher loan rates). However, there were cases in which the usury
ceilings were below even the interest rates that lenders required on very safe mortgage loans.
52Congress allowed the states a period in which they could reinstate their usury ceilings by
passing new legislation, but no state took this action. Comparable usury ceilings still exist in many
states on consumer loans and credit card loans, but the states now tend to change these ceilings
more promptly when market interest rates change.
883
theories. The p e r m a n e n t income/life cycle ( P I / L C ) theory assumes that

consumers face no borrowing constraints and adopt optimal consumption plans
based on their current and future income, implying that current income will
have only a small effect on current consumption. In contrast, if consumers face
borrowing constraints, then consumption expenditures would show "excess
sensitivity" with respect to current income (relative to the P I / L C theories).
Empirical testing of these theories has b e e n especially active in recent years,
following the lead of Hall (1978) who introduced tests based on the Euler
equations (the first-order conditions for the optimal consumption plan). M a n y
i s s u e s - c o n c e r n i n g specification, econometric methods, and data s o u r c e s arise in carrying out these tests. Nevertheless, most studies reject the P I / L C
theories - for example, it is estimated that from 15 to 50 percent of consumption expenditures are determined by current income rather than by p e r m a n e n t
income. 53 Since it seems that most consumers do face borrowing limits, the
failure of the P I / L C theories is often attributed to liquidity constraints. 54
Liquidity constraints have important implications for m a c r o e c o n o m i c policy.
For one thing, changes in average tax rates affect consumption spending when
there are liquidity constraints, in contrast to the traditional case where
marginal tax rates matter. This is relevant because recent tax law changes have
focused on marginal rates. For another thing, liquidity constraints may cause
Ricardian Equivalence not to hold.
The impact of liquidity constraints on consumption expenditures was vividly
illustrated in March 1980 when the Federal Reserve (on the urgent request of
President Carter) instituted emergency credit controls that required banks and
lenders to limit the a m o u n t of new loans, especially to consumers (including
credit card loans). As a result, during the second quarter of 1980, consumption
expenditures experienced one of the largest quarterly declines on record.
Indeed, the effect was so powerful that the Fed rapidly reversed its policy,
dismantling the controls by the s u m m e r of 1980. 55
5.8. The role of credit for monetary policy

Credit has been considered to be an important transmission channel for the
effects of m o n e t a r y policy by various economists over most of this century.
53Recent studies that focus on liquidity constraints include Flavin (1981), Hall and Mishkin
(1982), Hayashi (1985a, 1985b), Hubbard and Judd (1986) and Campbell and Mankiw (1987).
See, especially, Hayashx (1985a, 1985b). It is instrucUve that one of the few firm confirmations
of the PI/LC theory-Bernanke (1984)- is based on the demand for automobiles, where credit
rationing should not be important because autos provide an exceptionally liquid form of collateral.
55For an account of this episode, see Greider (1987), a controversial analysis of Federal Reserve
operations, especially for the period beginning in 1979 when Paul Volcker became Chairman of the
Board of Governors.
54
'
884
Among others, this "credit" school of monetary policy includes Hawtrey

(1919), Roosa (1951), Gurley and Shaw (1960), and Friedman (1981). Their
shared view is that variations in the availability of credit can have large effects
on real economic activity, both on the aggregate level and on its distribution
among sectors and even individual projects. In addition, most would argue that
monetary policy should be carried out, or at least measured and evaluated,
with regard to its effect on the availability of credit.
The "credit" school is often compared with the "money" or Monetarist
school. The two schools differ most fundamentally over the modus operandi of
monetary policy- whether it is based on the amount of the money supply, or
on the availability of credit. This basic conceptual difference leads to different
views regarding the implementation of monetary policy: how it should be
measured, when and how it should be used, and what it might accomplish..
However, it is important to recognize that within the structure of the U.S.
banking and financial system, the monetary aggregates of the respective
schools- the money supply versus the credit level- are likely to be highly
collinear. That is, a Federal Reserve policy that increases the reserves of the
banking system is likely to raise the money supply and the credit level in
roughly the same proportion. This occurs because bank assets (consisting
mainly of loans and other credit instruments) equal bank liabilities (consisting
mainly of deposits that are part of the money supply).
The credit school has had a resurgence in recent years as the result of a
number of developments:
(1) As developed in our discussion, there now exists a rigorous theory of
credit rationing based on the risk of default when there is imperfect information. This theory explains both why credit availability is likely to be important
and why the institutions (banks) and the instruments (loan contracts) are
structured as they are. (The Monetarist school, in contrast, assumes basically
that capital markets are perfect - and also that government policy can control
money substitutes.)
(2) Evidence is accumulating that large changes in the availability of credit
certainly create correspondingly large changes in the level of economic activity:
Bernanke (1983) points to the breakdown of the economy's credit facilities,
and not the decline in money, as the monetary source of the Great
Depression.
Wojnilower (1980) argues that the availability of credit, not the cost of
money, has been the source of most large changes in the level of real
economic activity since World War II.
Nakamura (1984) discusses the key role of banks in determining whether to
force a borrowing firm into bankruptcy.
The Federal Reserve created a deep recession during 1981, yet the growth
rate of the money supply fell very little.
(3) Deregulation and other innovations in the financial system in recent
885
years have created large and unpredictable changes in the demand for money.
A key result is that since 1987, the Federal Reserve has not used the M1
money supply as an operating target for monetary policy. At the same time,
Friedman (1981) has shown that a broad measure of credit stands at least on
par with money as an instrument that (a) the Federal Reserve can control and
(b) forecasts future movements in nominal GNP.
References
Akerlof, G. (1970) 'The market for lemons: Qualitative uncertainty and the market mechanism',
Quarterly Journal of Economics, 84: 488-500.
Allen, F. (1983) 'Credit rationing and payment incentives', Review of Economic Studies, 50:
639 -646.
Arnott, R. and J. Stiglitz (1985) 'Labor turnover, wage structure, and moral hazard', Journal of
Labor Economics, 3: 434-462.
Asquith, P. and D. Mullins (1986) 'Equity issues and offering dilution', Journal of Financial
Economics, 15: 61-89.
Baltensperger, E. (1978) 'Credit rationing: Issues and question', Journal of Money, Credit and
Banking, 10: 170-183.
Baltensperger, E. and T. Devinney (1985) 'Credit rationing theory: A survey and synthesis',
Zeitschrift Fur Die Gesamte Staatswissenschaft, 141: 475-502.
Bernanke, B. (1983) 'Nonmonetary effects of the financial collapse in the propagation of the great
depression', American Economic Review, 73: 257-276.
Bernanke, B. (1984) 'Permanent income, liquidity, and expenditure on automobiles: Evidence
from panel data', Quarterly Journal of Economics, 99: 587-616.
Bernanke, B. and M. Gertler (1987), 'Financial fragility and economic performance', Princeton
University, mimeo.
Bestor, H. (1985), 'Screening versus rationing in credit markets with imperfect information',
American Economic Review, 75: 850-855.
Bischoff, C. (1971) 'Business investment in the 1970s: A comparison of models', Brookings Papers
on Economic Activity, 1971-1: 13-58.
Blinder, A. (1987) 'Credit rationing and effective supply failure', Economic Journal, 97: 327-352.
Blinder, A. and J. Stiglitz (1983) 'Money, credit constraints, and economic activity', American
Economic Review, 73: 297-302.
Bowden, R. (1978) The econometrics of disequilibrium. Amsterdam: North-Holland.
Calomiris, C., R. Hubbard and J. Stock (1986) 'The farm debt crisis and public policy', Brookings
Papers on Economic Activity, 1986-2: 441-479.
Campbell, J. and G. Mankiw (1987) 'Permanent income, current income, and consumption',
NBER Working Paper 2436.
Clark, P. (1979) 'Investment in the 1970s: Theory, performance, and prediction', Brookings Papers
on Economic Activity, 1979-1: 73-113.
Clemenz, G. (1986) Credit markets with asymmetric information, Lecture Notes in Economics and
Mathematical Systems 272. Berlin: Springer-Verlag.
Diamond, D. and P. Dybvig (1983) 'Bank runs, deposit insurance, and liquidity', Journal of
Political Economy, 91: 401-419.
Duca, J. (1986) 'Trade credit and credit rationing: A theoretical model', Research Papers in
Banking and Financial Economics, Board of Governors of the Federal Reserve System.
Duca, J. and S. Rosenthal (1988) 'Mortgage rationing in the post-disintermediation era: Does
FHA make a difference?', Economic Activity Working Paper 83, Board of Governors of the
Federal Reserve System.
Eaton, J. (1986) 'Lending with costly enforcement of repayment and potential fraud', Journal of
Banking and Finance, 10: 281-293.
Eaton, J. and M. Gersovitz (1981) 'Debt and potential repudiations', Review of Economic Studies,
48: 289-309.
886
Eaton, J., M. Gersovitz and J. Stiglitz (1986) 'The pure theory of country risk', European
Eisner, R. and M. Nadiri (1968) 'Investment behavior and neoclassical theory', Review of
Economics and Statistics, 50: 369-382.
Fair, R. and D. Jaffee (1972) 'Methods of estimations for markets in disequilibrium', Econometrica, 40: 497-514.
Fazzari, S., R. Hubbard and B. Petersen (1988) 'Financing constraints and corporate investment',
Brookings Papers on Economic Activity, 1988-2: 141-195.
Flavin, M. (1981) 'The adjustment of consumption of changing expectations about future income',
Journal of Political Economy, 89: 974-1009.
Freimer, M. and M. Gordon (1965) 'Why bankers ration credit', Quarterly Journal of Economics,
79: 397-410.
Fried, J. and P. Howitt (1980) 'Credit rationing and implicit contract theory', Journal of Money,
Credit and Banking, 12: 471-487.
Friedman, B. (1981) 'The roles of money and credit in macroeconomic analysis', Working Paper
831, National Bureau of Economic Research.
Gale, I, and J.E. Stiglitz (forthcoming) 'The informational content of initial public offering',
Journal of Finance.
Goldfeld, S. and R. Quandt (1975) 'Estimation in a disequilibrium model and the value of
information', Journal of Econometrics, 3: 325-348.
Greenwald, B. and J. Stiglitz (1986a) 'Money, imperfect information, and economic fluctuations',
Symposium on Monetary Theory, The Institute of Economics, Academia Sinica, Taipei,
Taiwan.
Greenwald, B. and J. Stiglitz (1986b) 'Externalities in economies with imperfect information and
incomplete markets', Quarterly Journal of Economics, 101: 229-264.
Greenwald, B., J. Stiglitz and A. Weiss (1984) 'Informational imperfections in the capital market
and macroeconomic fluctuations', American Economic Review, 74: 194-199.
Greider, W. (1987) Secrets of the temple. New York: Simon and Schuster.
Gurley, J. and E. Shaw (1960) Money in a theory of finance. Washington: The Brookings
Institution.
Guttentag, J. (1960) 'Credit availability, interest rates, and monetary policy', Southern Economic
Journal, 26: 219-228.
Hall, R. (1978) 'Stochastic implications of the life cycle-permanent income hypothesis: Theory
and evidence', Journal of Political Economy, 86: 971-988.
Hall, R. and D. Jorgenson (1967) 'Tax policy and investment behavior', American Economic
Review, 74: 194-199.
Hall, R. and F. Mishkin (1982) 'The sensitivity of consumption to transitory income: Estimates
from panel data on households', Econometrica, 50: 461-482.
Harris, D. (1974) 'Credit rationing at commercial banks', Journal of Money, Credit and Banking,
6: 227-240.
Hawtrey, R. (1919) Currency and credit. New York: Longmans, Green & Co.
Hayashi, F. (1985a) 'The effect of liquidity constraints on consumption: A cross-sectional analysis',
Quarterly Journal of Economics, 100: 183-206.
Hayashi, F. (1985b) 'Tests for liquidity constraints: A critical survey', NBER Working Paper 1720.
Hellwig, M. (1977) 'A model of borrowing and lending with bankruptcy', Econometrica, 45:
1879-1906.
Hodgman, D. (1960) 'Credit risk and credit rationing', Quarterly Journal of Economics, 74:
258-278.
Hodgman, D. (1963) Commercial bank loan and investment policy. Champaign-Urbana: Bureau of
Economic and Business Research, University of Illinois.
Hubbard, R. and K. Judd (1986) 'Liquidity constraints, fiscal policy, and consumption', Brookings
Jaffee, D. (1971) Credit rationing and the commercial loan market. New York: Wiley.
Jaffee, D. and F. Modigliani (1969) 'A theory and test of credit rationing', American Economic
Review, 59: 850-872.
Jaffe, D, and K. Rosen (1979) 'Mortgage credit availability and residential construction activity',
Brookings Papers on Economic Activity, 1979-2: 333-386.
887
Jaffee, D. and T. Russell (1976) "Imperfect information, uncertainty, and credit rationing',
Quarterly Journal of Economics, 90:651-666.
Kane, E. and B. Malkiel (1965) 'Bank portfolio allocation, deposit variability and the availability
doctrine', Quarterly Journal of Economics, 79: 113-134.
Keaton, W. (1979) Equilibrium credit rationing. New York: Garland Publishing Company.
Keynes, J.M. (1930) A treatise on money. London.
King, M. (1986) 'Capital market imperfections and the consumption function', Scandinavian
Journal of Economics, 88: 59-80.
Koskela, E. (1976) A study of bank behavior and credit rationing. Helsinki: Academiae Scientarum
Fennicae.
Koskela, E. (1979) 'On the theory of rationing equilibrium with special reference to credit
markets: A survey', Zeitschrift fur Nationalokonomie, 39: 63-82.
Laffont, J. and R. Garcia (1977) "Disequilibrium econometrics for business loans', Econometrica,
45: 1187-1204.
Leibenstein, H. (1957) Economic backwardness and economic growth. New York: Wiley.
Leland, H. and D. Pyle (1977) 'Informational asymmetries, financial structure and financial
intermediation', Journal of Finance, 32: 371-387.
Lindbeck, A. (1962) The 'new" theory of credit control in the United States, 2nd edn. Stockholm
Economic Studies.
Maddala, G.S. (1983) Limited-dependent and qualitative variables in econometrics. Cambridge:
Cambridge University Press.
Meltzer, A. (1960), 'Mercantile credit, monetary policy, and the size of firms', Review of
Economics and Statistics, 42: 429-437.
Meyer, J. and E. Kuh (1957) The investment decision. Cambridge, Mass.: Harvard University
Press.
Mirrlees, J. (1971) 'An exploration of optimum income taxation', Review of Economic Studies, 38:
175 -208.
Mirrlees, J. (1975a) 'The theory of moral hazard and observable behavior: Part 1', Working Paper,
Nuffield College.
Mirrlees, J. (1975b) 'A pure theory of underdeveloped economies', in: L.A. Reynolds, ed.,
Agriculture in development theory. New Haven: Yale University Press, pp. 84-106.
Modigliani, F. (1963) 'The monetary mechanism and its interaction with real phenomena: A
review of recent developments', Review of Economics and Statistics, 45: 79-107.
Myers, S. (1977) 'Determinants of corporate borrowing', Journal of Financial Economics, 5:
147-175.
Myers, S. and N. Majluf (1984) 'Corporate financing decisions when firms have investment
information that investors do not have', Journal of Financial Economics, 13: 187-220.
Nakamura, L. (1984) 'Bankruptcy and the informational problems of commercial bank lending',
Financial Research Center Memorandum 54, Princeton University.
Nadiri, M. (1969) 'The determinants of trade credit in the U.S. total manufacturing sector',
Econometrica, 37: 408-423.
Nalebuff, B. and J. Stiglitz (1983) 'Equilibrium unemployment as a worker selection device',
Princeton University mimeo.
Patman Hearings (1952) 'Hearings before the Subcommittee on General Credit Control and Debt
Management of the Joint Committee on the Economic Report', 82nd Congress, 2nd Session.
Washington: Government Printing Office.
Riley, J. (1987) 'Credit rationing: A further remark', American Economic Review, 77: 224-227.
Roosa, R. (1951) 'Interest rates and the central bank', in: Money, trade and economic growth:
Essays in honor of John H. Williams. New York: The Macmillan Company, pp. 270-295.
Rosenthal, S., J. Duca and S. Gabriel (1987) 'Credit rationing and the demand for owner-occupied
housing', Economic Activity Working Paper 79, Board of Governors of the Federal Reserve.
Ross, S. (1977) 'The determination of financial structure: The incentive signalling approach', Bell
Journal of Economics, 8: 23-40.
Rothschild, M. and J. Stiglitz (1971) 'Increasing risk: I. A definition', Journal of Economic
Theory, 2: 225-243.
Schwartz, R. and D. Whitcomb (1980) 'The trade credit decision', in: J. Bicksler, ed., Handbook
of financial economics. Amsterdam: North-Holland.
888
Sealey, C. (1979) ~Credit rationing in the commercial loan market: Estimates of a structural model
under conditions of disequilibrium', Journal of Finance, 34: 689-702.
Shapiro, C. (1983) ~Premiums for high quality products as returns to reputation', Quarterly Journal
of Economics, 98: 659-679.
Shapiro, C. and J. Stiglitz (1984) 'Equilibrium unemployment as a worker discipline device',
Smith, A. (1776) The wealth of nations, Modern Library Edition. New York: Random House.
Smith, B. (1983) 'Limited information, credit rationing, and optimal government lending policy',
Stigler, G. (1967) 'Imperfections in the capital market', Journal of Political Economy, 85: 287-292.
Stiglitz, J. (1970) 'A consumption oriented theory of the demand for financial assets and the term
structure of interest rates', Review of Economic Studies, 37: 321-351.
Stiglitz, J. (1972), 'Some aspects of the pure theory of corporate finance: Bankruptcies and
take-overs', Bell Journal of Economics and Management Science, 3: 458-482.
Stiglitz, J. (1974) 'Incentives and risk sharing in sharecropping', Review of Economic Studies, 41:
219-255.
Stiglitz, J. (1976a) 'Prices and queues as screening devices in competitive markets', IMSS Technical
Report 212, Stanford University.
Stiglitz, J. (1976b) 'The efficiency wage hypothesis, surplus labour and the distribution of income
in LDCs', Oxford Economic Papers, 28: 185-207.
Stiglitz, J. (1977) 'Monopoly non-linear pricing and imperfect information: The insurance market',
Review of Economic Studies, 44: 407-430.
Stiglitz, J. (1982) 'Ownership, control and efficient markets: Some paradoxes in the theory of
capital markets', in: K. Boyer and W. Shepherd, eds., Economic regulation: Essays in honor of
James R. Nelson. Ann Arbor: Michigan State Press, pp. 311-341.
Stiglitz, J. (1985) 'Credit markets and the control of capital', Journal of Money, Credit and
Banking, 17: 133-152.
Stiglitz, J. (1986) 'Theories of wage rigidities', in: J.L. Butkiewicz, K.J. Kotford and J.B. Miller,
eds., Keynes' economic legacy. New York: Praeger, pp. 153-206.
Stiglitz, J. (1987) 'The causes and consequences of the dependence of quality on price', Journal of
Economic Literature, 25: 1-48.
Stiglitz, J. and A. Weiss (1981) 'Credit rationing in markets with imperfect information', American
Stiglitz, J. and A. Weiss (1983) 'Incentive effects of termination: Applications to the credit and
labor markets', American Economic Review, 73: 912-927.
Stiglitz, J. and A. Weiss (1986) 'Credit rationing and collateral', in: J. Edwards, J. Franks, C.
Mayer and S. Schaefer, eds., Recent developments in corporate finance. Cambridge: Cambridge
University Press, pp. 101-135.
Stiglitz, J. and A. Weiss (1987a) 'Macro-economic equilibrium and credit rationing', Working
Paper 2164, National Bureau of Economic Research.
Stiglitz, J. and A. Weiss (1987b) 'Credit rationing with many borrowers', American Economic
Review, 77: 228-231.
Tobin, J. and W. Dolde (1971) 'Wealth, liquidity, and consumption', Federal Reserve Bank of
Boston, Conference Series No 5.
Viner, J. (1937) Studies in the theory of international trade. New York: Harper and Brothers.
Wette, H. (1938) 'Collateral in credit rationing in markets with imperfect information: Note',
Wojnilower, A. (1980) 'The central role of credit crunches in recent financial history', Brookings
Yellen, J. (1984) 'Efficiency wage model of unemployment', American Economic Review, 74:
200-205.

Jaffee & Stiglitz, Credit Rationing

Uploaded by

Copyright:

Available Formats

Jaffee & Stiglitz, Credit Rationing

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Jaffee & Stiglitz, Credit Rationing

Uploaded by

Copyright:

Available Formats

Chapter 16

The classification of borrowers

The sources of loanable funds: Banks and bank deposits

Definitions of credit rationing

The earlier credit rationing literature: Brief review

Credit markets with imperfect information

Credit rationing with many groups

Some comparative statics of credit rationing

A consumer loan model

Extensions to the basic theory

Adverse selection effects

Enforcement problems in multiple period models

Collateral, loan size, and non-price rationing

Macroeconomic aspects of credit rationing

Market interest rates and the shadow price of credit

The measurement of credit rationing

Fixed capital investment of business firms

The role of credit for monetary policy

D. Jaffee and J. Stiglitz

Ch. 16: Credit Rationing

discussion we shall uncover a n u m b e r of " p r o b l e m s " with the loan market.

Our discussion of credit rationing in Section 3 relies on certain features of loan

D. Jaffee and J. Stiglitz

known probability of default. 2 Therefore, default risk is a necessary condition,

Ch. 16: Credit Rationing

assuming a mean preserving spread- that all projects (all possible (X a, X b}

2.2. Loan contracts

D. Jaffee and J. Stiglitz

2.3. The classification of borrowers

Ch. 16: Credit Rationing

Borrower classification - based on risk screening - is a main function of the

D. Jaffee and J. Stiglitz

Ch. 16: Credit Rationing

2.4. The sources of loanable funds: Banks and bank deposits

D. Jaffee and J. Stiglitz

Ch. 16: Credit Rationing

2.5. Definitions of credit rationing

D. Jaffee and J. Stiglitz

Ch. 16: Credit Rationing

2.6. The earlier credit rationing literature: Brief review

D. Jaffee and J. Stiglitz

schedule of borrowers, or in the volume of lending otherwise than through

2.6.1. The risk of default as the basis of credit rationing

Ch. 16: Credit Rationing

Bank offer curve

D. Jaffee and J. Stiglitz

2.6.2. Borrower classification

2. 6.3. Customer relationships

Ch. 16: Credit Rationing

3. Credit markets with imperfect information 18

D. Jaffee and J. Stiglitz

Quoted interest rate r

Ch. 16: Credit Rationing

3.1. A d v e r s e selection effects

The return to the projectX

21We will later remove both of these assumptions.

D. Jaffee and J. Stiglitz

Ch. 16: Credit Rationing

Quoted loan rate r