Credit Risk and Credit Rationing

Donald R. Hodgman

The Quarterly Journal of Economics, Vol. 74, No. 2. (May, 1960), pp. 258-278.

Stable URL:
http://links.jstor.org/sici?sici=0033-5533%28196005%2974%3A2%3C258%3ACRACR%3E2.0.CO%3B2-X

The Quarterly Journal of Economics is currently published by The MIT Press.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained
prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in
the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/journals/mitpress.html.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic
journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,
and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take
advantage of advances in technology. For more information regarding JSTOR, please contact [email protected].

http://www.jstor.org
Wed Dec 19 15:51:49 2007
 CREDIT RISK AND CREDIT RATIONING*

I. Approaches to credit rationing, 258. - 11. The influence of credit risk

on loan payoff, 259. - 111. Implications for lender behavior and borrower access
to credit, 267. - IV. The central bank's influence, 275.

I. APPROACHES TO CREDITRATIONING
Credit rationing is a much debated phenomenon. Although
practical borrowers and lenders long have regarded credit rationing
as a fact of experience, economists of a more analytical persuasion
have been reluctant to accept it a t face value because of their diffi-
culty in providing a theoretical explanatioi~for the phenomenon which
is consistent with the tenets of rational economic behavior. Why
should lenders allocate credit by non-price means and thus deny
themselves the advantage of higher interest income? Explanations
which meet this difficulty are those which attribute credit rationing
to sticky interest rates associated with oligopoly in credit markets
and to ceilings on interest charges imposed by usury laws. But most
corporate borrowers are exempt from the protection of the usury
statutes, and sticky rates of interest usually rise without too much
delay as credit conditions tighten. These considerations have led
to the apparently prevailing view that credit rationing as a general
phenomenon, while it does occur, is primarily a temporary expedient
which lenders resort to only until they have made up their minds to
raise the interest rate to a new equilibrium level following an increase
in borrower demand or restrictive action by the monetary authority.
This view denies to credit rationing any significant influence on credit
availability other than in the fairly short run.'
Recently there have been attempts to free credit rationing from
*I am deeply indebted to Professor Colin R. Blyth of the Mathematics
Department, University of Illinois for developing my intuitive surmises into the
explicit mathematical reasoning of Section I. Professor Marvin Frankel read the
manuscript and prowlded helpful criticism. Professor Thomas A. Yancey gave
generously of his t'ime in several prolonged discussions of probability measure-
ments. I t is a pleasure to acknowledge the support of The Merrill Foundation
for the Advancement of Financial Knowledge for the larger study from which this
article derives.
1. For an explicit statement of this view see Monetary Policy and Manage-
ment of the Public Debt: Hearings before the Subcommittee on General Credit Control
and ~ k b Management,
t Joint ~ o m n i i t t e eon the Economic Report, 82d Congress,
2d Session, Statement of Paul Samuelson, esp. pp. 697-98.
 CREDIT R I S K AND CREDIT RATIONING 259

its dependence upon interest ceilings, whether voluntary or imposed,

by suggesting that credit rationing is an aspect of lender attitude
toward risk. T o date, so far as I am aware, these attempts have not
been able to cope with one particular difficulty: the principle that
when other factors such as credit rating or maturity are less favorable,
borrowers may succeed in obtaining the credit they wish by paying a
higher rate of interest to compensate lenders for the less favorable
risk features of the loan or investmenL2 Once this camel's nose is
under the tent, interest appears to regain much of the authority that
the concept of credit rationing seeks to deny to it. If a determined
borrower can always obtain funds by agreeing to a sufficiently high
rate of interest, he can be denied credit only by quoting him a rate
which he regards as prohibitive. This, however, is not credit rationing
in the specific sense but traditional interest rationing.
I n the analysis which follows I have sought to explore further the
implications of risk for the terms upon which money is loaned. M y
purpose is to provide a more general explanation for credit rationing
which does not rely upon oligopolistic market structure or legal
maxima t o the interest rate, which is consistent with rational behavior
along the lines of economic self-interest, and which is permanent
rather than temporary in its effect for so long as the general credit
situation which occasions it lasts. The implications of the analysis
extend to any credit transaction. However, the institutional influ-
ences which have conditioned my thinking are those of c.ommercia1
banks, so there may be some special assumptions imbedded in the
approach of which I am not aware. The reader may wish to remind
himself of this occasionally.

The attractiveness of a loan or investment to a lender depends

upon various factors: the interest rate, risk, possible benefits of a long
term customer relationship and so on. In this section we focus our
attention upon the interest rate and risk and their expression in a
probability distribution of possible payoffs which is logically implied
by the yield-risk features of any loan or investmenL3 We shall see
2. Fcr an otherwise suggestive treatment ot credit rationing as an aspect of
lenders' attltude toward risk which fails exactly a t this stage in the argument see
Ira 0. Scott, "The Availability Doctrine: Theoretical Underpinnings," The Review
of Economzc Studies, X X V (Oct. 1957). The dependence of Scott's solution on
interest ceilings is stated in fn. 4, p. 46.
3. Note that all features of a loan including customer relationship can ulti-
mately be reduced to the dimensions of a paycff distribution in an appropriate
dynamic analysis. I have not attempted this in the present static analysis.
However, this consideration signifies that the logzc of rational lender behavior is
260 QUARTERLY JOURNAL OF ECONOMICS

that the yield and risk aspects of an investment are systematically

interrelated and can be varied relative to each other by the borrower
only within specific limits. In particular, given the credit rating of
the borrower, risk of loss is an increasing function of the size of the
loan while the expected value of possible gains and losses (as a
measure of yield under risky conditions) is an increasing function of
the amount the borrower promises to repay (inclusive of interest).
However, the influence of the borrower's credit rating sets an upper
limit to the expected value of the payoff while no such limit applies
to the expected loss. These conclusions follow logically from some
fairly simple and intuitively reasonable stipulations concerning
borrower capacity to pay.
The purpose of this section is to establish the logical basis for
these conclusions. Their implications for lender behavior and for
borrower access to credit are explored in Sections I1 and 111. A
reader who is prepared to accept these conclusions as intuitively
convincing can save time and effort by skipping a t this point to
Section 11. The balance of the present section is for skeptics.
Suppose that a lender has applied the techniques of credit
analysis to the financial affairs of a potential borrower and has
summed up his (the lender's) confidence in the borrower's capacity
to pay various dollar amounts by assigning to each potential payment,
(y), a number between zero and one to represent the probability
which the lender associates with the event that the borrower would
pay exactly that amount if he promised to do so. This information
on the probability of the borrower's capacity to pay various hypo-
thetical sums can be summarized in the form of a mathematical
function. Let +(y) be the probability that the borrower will be able
to pay the amount y. Some very general restrictions on the nature
of C#J can be suggested. Since it is certain that the borrower can pay
nothing if he makes this promise, +(0) = 1. The consideration that
the financial resources of a specific borrower are not unlimited implies
+(y) = 0 for y sufficiently large. Moreover, it seerns reasonable to
assume that +(y) is nonincreasing (the probability of repayment
should not increase with the size of the amount promised) and con-
tinuous as well as differentiable for most y. Further, for a range of
small y values, $(y) may be close to 1. There are infinitely many
functions having these properties. The general appearance of such
functions is illustrated in Figure I .
more broadly encompassed in the present discussion than might appear to be the
case in view of the omission of customer relations, maturities and other dynamic
considerations.
 CREDIT R I S K A N D CREDIT RATIONING 26 1

For any y on Figure I we can read off the probability that the
lender attaches to the event that the borrower would pay exactly
that y should he promise to do SO.^ Let s be the amount that the
borrower promises to pay. The lender's attention will be confined
exclusively to s only if $ ( s ) = 1. For +(s) < 1 the actual payoff

FIGURE I

may be less than s. Assume that the borrower pays s if he is able

and otherwise pays the greatest amount which he can. Let Y be the
amount the borrower actually pays. This is a random variable. The
cumulative distribution function of Y is:

4. The function, 9, is assumed to be independent of the size of the borrower's

promise to pay and of the size of the loan granted. There must be many instances
where this is reasonable, although there may be other situations for which one or
both of these assumptions would have to be dropped.
262 QUARTERLY JOURNAL OF ECONOMICS

The appearance of this distribution function is illustrated in Figure 11.

The probability P ( Y 6 y) is nondecreasing in y, and a t y = s jumps
to 1 if it has not already reached 1 (since the borrower will never pay
more than s).
If +(y) is almost everywhere differentiable for 0 < y < s, the

-
FIGURE I1

probability distribution of Y can be given by the density -$'(y) on

0 < y < s, together with the non-zero probability on s:
b
P(a < Y < b) = 1- $'(y)dy,O 5 a < b s
a

P(Y = s) = +(s).
The appearance of this density is shown in Figure 111. Thus the
probability that Y lies between any two limits less than s and equal
 CREDIT R I S K A N D CREDIT RATIONING 263

to or greater than zero is shown by the area under the curve, while
the probability that Y equals s is simply +(s).
To understand the reason for our next step we must pause to
consider what aspects of the payoff function associated with a borrow-
er's promise to pay a particular sum the lender may regard as relevant.
One way to summarize the payoff function is to calculate its expected

FIGURE I11
S

value. There is, however, considerable cogency to the view often

expressed in the literature concerned with the analysis of risky choice,
that the lender (investor, gambler) faced with such a payoff distribu-
tion may not be indifferent to the dispersion of the distribution
around its expected value. Frequently some such measure of dis-
persion as the standard deviation or variance of the payoff distribu-
tion has been utilized to illustrate the influence of dispersion on the
utility of risky choices. Either of these measures is acceptable.
However, to emphasize the asymmetrical influence of payoffs which
imply gains versus those which imply losses, I prefer to use a measure
264 QUARTERLY JOURNAL OF ECONOMICS

of the expected value of probable losses to be defined below. Accord-

ingly, we shall proceed on the assumption that the aspects of the
payoff distribution which the lender regards as relevant are the
expected value of the entire distribution and the expected value of
the probable losses. The use of these measures is intended to be
suggestive rather than definitive.

The expected value of the payoff distribution depends upon the

4 function (illustrated in Figure I) and s, the promise to pay made
by the borrower. The expected value of Y is
S

E Y = sP(Y = s) + [ - y+'(y)dy
0
S

= j$(y)dy using integration by parts.

0
 CREDIT R I S K A N D CREDIT RATIONING 265

The derivative of the expected value with respect to s is

Accordingly, for a $ function similar to that in Figure I, the expected

value of the payoff increases with s a t a decreasing rate and approaches
a limiting value. This functional relationship between E Y and s is
illustrated in Figure IV.
EZ

FIGURE V

A loss occurs when the payoff is less than the amount of the loan
given the borrower. If the amount loaned is a and the amount paid
is Y, the loss Z is

The expected value of probable losses depends upon the borrower's

estimated capacity to pay (the function), his promise to pay (s),
266 QUARTERLY JOURNAL OF ECONOMICS

and the size of the loan (a) made by the lender. Thus,

FIGURE VI

Normally the lender will not make loans when a > s, but it is useful
to have EZ defined for that possibility to facilitate certain diagram-
matic treatment below. It may be useful to point out that an
increase in s after s = a does not affect EZ although it does increase
E Y until +(y) = 0. This is apparent from the fact that the payoff
function depends only on the + function and s. Thus, for any
 CREDIT R I S K A N D C R E D I T R A T I O N I N G 267

specific a, say a,, EZ declines as s increases from zero to a and there-

after remains constant for further increases in s as in Figure V.

111. IMPLICATIONS FOR LENDERB E H ~ ~ V I AND

OR
BORROWERACCESSTO CREDIT
We turn now to the implications of our analysis for lender
behavior and borrower access to credit. From the relationships dis-
cussed thus far it is possible to determine the various combinations
of expected value of payoff ( E Y ) and expected loss ( E Z ) which con-
front a lender who considers making various size loans (a) in response
to various promises to pay (s) on the part of a specific borrower.
The derivation of this information is described in Figure VI. The
northeast quadrant is identical to Figure I V except that EY for each
EY
s has been replaced by - for each s and a. Since EY increases with s
a
EY
until @(y) = 0, -increases similarly with s for each value of a. The
a
southeast quadrant of Figure V I is identical with Figure V except
EZ
that EZ has been replaced by - and possibilities for an additional
a
loan (a,) have been i n ~ l u d e d .Dividing
~ E Y and EZ by the relevant a
simply serves to emphasize the relationship of expected value and
expected loss to the dollar amount of the loan. These are essential
relationships when we come to consider the ability of the borrower
to meet competitive market terms. The northwest quadrant of
EY
Figure V I shows the aohievable combinations of - and for differ-
a a
ent loans (a) and promises to pay (s) for a borrower whose credit
rating, (@ function), has been determined. The southwest quadrant
EZ
is simply a means to transfer the - values from the lower half of the
a
vertical axis to the left half of the horizontal axis.
T o have this information on our borrower's capacity to denland
in a form more convenient for purposes of relating it to the terms upon
which the lender offers credit we shall now transfer it to Figure V I I
which is one of the key figures in our analysis. In Figure V I I combi-
EY EZ
nations of - and - which a particular borrower can attain by vary-
a a
EZ
5. The ratio - for any a is 1 when s is zero.
268 QUARTERLY JOURNAL OF ECONOMICS

ing his promise to pay (s) are shown as the locus of points along the
various loan curves labeled aj (where j = 0 . . . n). Variations in s
are implied but not explicitly shown in Figure VII. However, the
EY EZ
relevant value for s can be ascertained by inserting the -, -, co-or-

FIGURE VII

dinates of any point on Figure VII into Figure VI and reading off the
resulting required values for s. The curve labeled (s = a) which
slopes downward to the right is the locus of minimum expected losses
a t associated minimum values of E Y for loans of different amounts.
If there is any risk (+(y) < I), the expected value a t (s = a) is
 CREDIT R I S K AND CREDIT RATIONING 269

EY EY
always less than ( s = - < 1. Moreover, - for (s = a )
a ) , that is
a a
decreases as ( s = a) and associated EZ increase. Therefore, the
EZ EY
locus of minimum - and associated - values falls further below
a a
EY EY
the line - = 1 as ( s = a ) increases. The mazimum - attainable
a a

-
s -a
a

FIGURE VIII

dEY
by the prospective borrower for any a occurs when - becomes zero.
ds
EY
This maximum ratio for - (shown by MM' in Figure VII) also
a
declines as a increases.
Our next step is to introduce the borrower's demand schedule
which expresses his willingness as distinguished from his capacity to
meet the terms which the lender may require. The borrower's demand
schedule appropriate to Figure VII is easily derived from a conven-
a70 QUARTERLY JOURNAL OF ECONOMICS

tional demand schedule for loanable funds.6 Suppose our borrower's

conventional demand schedule is that shown by DD' in Figure VIII.
(The reader is asked to ignore the supply schedules in Figure VIII
for the present.) This schedule relates the size of the loan (a) which
the borrower will want to the contract rate of interest (r) charged.
s-a
B u t r = -. Therefore, any pair of (r, a) from the demand sched-
a
ule of Figure VIII implies a specific value for s, the borrower's promise
EY
to pay. Given specific values for s and a, specific values for - and
a
are likewise determined. Thus, the borrower's conventional
a
EY EZ
demand schedule is readily translated into terms of - and - and
a a
can be represented on Figure VII. In making the translation one
restriction must be observed. No matter how high the value for
T=-
s - a to which the borrower will agree, the upper limit to the
a
EY
corresponding value for - is determined by the borrower's credit
a
rating (the q5 function). This limit is represented in Figure VII by
the curve MM'. Therefore, the borrower's effective demand cannot
rise above MM'. The demand schedule of Figure VIII translated
into terms of Figure VII is shown by the dotted curve DD'. Although
the borrower might well wish to take larger loans a t combinationsbf
EZ
=and -lying below and to the left of the lower end of the demand
a a
schedule in Figure VII, the lender will never lend a t such terms
(unless he is an avid gambler with little regard for odds), since they
EY EZ
are always inferior to holding cash for which - = 1 and - = 0.'
a a
The terms upon which the lender is willing to lend are shown by
the supply curves (labeled SS,:; j = 0 . . . n) in Figure VII. Only
one such SS; curve is in effect a t any one time. We shall now examine
the factors which determine the shape and position of these S S ~ c! urves
as well as which curve is operational a t a particular time. These
factors include the subjective preferences of the lender for yield versus
6. We ahall not explore the analytical determinants of borrower demand
under risk and uncertainty, although this would improve the symmetry of the
discussion.
7. This statement ignores price changes in the markets for goods and
services.
 CREDIT RISK AND CREDIT RATIONING 27 1

risk of loss, the terms upon which he can exchange risk of loss for
yield as established by market conditions (i.e., borrower demand and
competing supply), his total supply of investable funds, and his
liability position which must be related to his asset position before
the implications of his asset position for yield and risk can be deter-
mined. The influence of a lender's liability position has been neglected
in the formal literature of investment choice. This is not the occasion
to attempt to repair this neglect. Perhaps it will be intuitively obvi-
ous to the reader that two lenders with different liability structures
may hold identical collections of assets and yet face quite different
yield-risk situations. The simplest way to handle this problem in the
present context is to incorporate the effect of the lender's liability
structure in his indifference curves. This means, of course, that two
lenders with identical subjective attitudes toward risk and yield are
likely to have quite different preference patterns for assets. (Differ-
ences in liabilities are probably a far more important cause of differ-
ences in institutional asset preferences than the subjective attitudes
of institutional managers.) Once we have iilcorporslted the influence
of liability position in the indifference curves, we are entitled to con-
centrate on the available asset characteristics and combinations of
assets.
The indifference curves (labeled 11, 12,etc.) in Figure I X repre-
sent our lender's preference pattern for yield versus risk in his total
asset collection as conditioned by his subjective preferences and his
liability position. Note that Figure I X is intended to represent the
risk-yield implications of the lender's total asset position in contra-
distinction to the marginal changes in that position caused by a single
borrower as reflected in Figure VII. (Note also, that the alternative
risk-yield combinations of Figure I X do not represent the ultimate
risk-yield position of the lender. For this, the influence of the liability
position must be made explicit.) The influence of market conditions
and the volume of total investable resources available to the lender
are represented by what we may call "opportunity curves'' labeled
01, 0 2 , etc. Only one of the opportunity curves is in effect a t any one
time. For given market conditions an opportunity curve can be
derived by considering the terms upon which the lender can extend
credit to the various borrowers who seek accomm~dation.~ On the

8. There is a problem of interdependence among payoff functions which I

propose to ignore. I t can be dealt with formally by replacing EZ by the variance
of the payoff distribution (for mathematical convenience) for individual borrowers,
introducing covariance measures between different distributions, and following
methods outlined by Harry Markowitz in "Portfolio Selection," The Journal of
Finance, March 1952.
272 QUARTERLY JOURNAL OF ECONOMICS

conventional assumption that borrowers' demand curves are down-

ward sloping and their credit ratings subject to deterioration as the
amounts they borrow increase, the slope of the opportunity curves
must diminish as they extend further to the right. This simply
reflects the fact that the lender must accept less farorable terms to
expand his volume of loans. Moreover, in a competitive market no
borrower will pay better than the worst terms.

CEY

CEZ
FIGURE IX

Given the opportunity line (borrower demand) the positions

available to the lender depend upon the volume of investable resources
in his possession. This magnitude is not shown explicitly in Figure IX,
but its effect is indicated by the solid portion of the opportunity lines.
The lender has sufficient investable resources to go out a given oppor-
tunity line to the point where the line becomes da.shed (- - -).
At that point the lender is fully invested (that is, cash is zero) and is
 CREDIT RISK AND CREDIT RATIONING 273

powerless to extend additional credit. The optimum risk-yield combi-

nation for the lender is shown diagrammatically as t,hehighest indiffer-
ence curve attainable (by intersection or tangency) by traveling along
the solid portion of the existing opportunity line. The optimum
point in Figure I X will normally be a tangency rather than an inter-
section. This follows from the consideration that a lender will usually
retain some cash for his own transactions and precautionary require-
ments. Therefore, his fully invested position (cash = zero) will
have to be further out on the relevant opportunity line than his
optimum position. But this implies a tangency optimum. However,
an intersection optimum is possible if the lender is restrained from
reducing his (as he regards it) surplus cash because of a minimum
cash requirement imposed by law or regulatory authority.
In a competitive market our borrower must pay the going rate
for funds. Let A (where A = Zaj and j = 1 . . . n) represent total
earning assets of the lender. Then Eya t (or in the neighborhood of)
ZEZ
EY
the optimum point in Figure I X are the - terms which our borrower
EZ
must meet to qualify for a loan.
The loans and terms available to the particular borrower to
which Figure VII applies are shown by the relevant S S ~curve on that
diagram. Let us suppose that the tangency optimum in Figure I X
establishes SS; as the operational supply curve in Figure VII.9 Then
our borrower is able to qualify for any loan between zero and al.
Given supply curve SS; our borrower cannot qualify to borrow more
than al because his credit rating (6function) does not warrant a
larger loan a t market terms. Assuming s 2 a, the loan requested
carries with it an irreducible risk of loss indicated b s the location with
EZ
respect to the -axis of the vertical portion of the respective a curve.
a
EY
The maximum - or E Y which the borrower can generate to offset
a
that irreducible risk also is a function of the #I function (and s). Thus,
the area of effective borrower demand is bounded by the MM' curve,
9. If the optimum in Figure IX occurs a t an intersection rather than a
tangency (due to an imposed minimum cash requirement) the relevantSS1jcurve
EY
in Figure VII will intersect the vertical axis a t a point where - > 1 since the
a
marginal utility of available additional loans wi!l exceed the marginal utility of
cash to the lender.
274 QUARTERLY JOURNAL OF ECONOMICS

the relevant SS' curve, and the vertical axis. Clearly, as changes in
the optimum point on Figure I X cause the operational SS' curve of
Figure VII to rotate (or shift) northwest, our borrower's ability to
qualify for loans is steadily and inescapably decreased. In particular,
the maximum loan for which a particular borrower can qualify declines
as the minimum qualifications in terms of E Y / a and E Z / a increase.
If the borrower's willingness to demand falls short of his capacity, the
loan agreed upon is shown by the intersection of SS' and DD' in
Figure VII.
I t is instructive to look a t the analogue of the supply and demand
curves of Figure VII in conventional supply and demand terms.
Since we already have the demand equivalent in Figure VIII we need
only translate the supply curves of Figure VII into their equivalent
expression in Figure VIII. This is readily done by translating the
and a values of points located on a supply curve of Figure VII
a
into their interest rate and a equivalent for the particular borrower.
EY
Given a and -, the s required of the borrower can be read from the
a
s - a
northeast quadrant of Figure VI thus determining r = -. The
a
conventional supply curves analogous to the two-dimensional supply
curves SS~,SS; and S S ~ in Figure VII are identically labeled in
Figure VIII. It must be emphasized that these conventional supply
curves of Figure VIII are specific to one particular borrower (or, a t
least, one particular 4 function). We may note that supply curve S S ~
in Figure VIII is interest-elastic both above and below its inter-
section with DD'; the borrower could obtain additional funds if he
were willing to increase his promise to pay, (s). At a,, however, SS;
becomes totally interest inelastic. The borrower's capacity to borrow
has reached its limit. Should SS; replace SS~,our borrower would
find himself borrowing a t absolute capacity a t a,. He would be
powerless to borrow more, since his credit rating would prevent him
from increasing E Y to compensate the lender for the unavoidable
increase in E Z which would accompany an increase in a. A further
shift to s S ~in Figure VIII would further reduce the maximum loan
available to the borrower from a, to a,. The usual interpretat,ion of
the intersection of DD' and S S ~leads to the erroneous conclusion
that the contract rate of interest for a, is r,. In fact, however, the
contract rate for a, is r,. There are two reasons for this. The borrower
will receive a, either by promising rl or r, and, hence, will promise the
 CREDIT RISK AND CREDIT RATIONING 275

EY
lesser, r,. The lender, thinking in terms of - and for a,, will
a a
regard these values as identical for rates r , and m and will be indifferent
between the two contract rates of interest.
Whenever the borrower's demand curve intersects with a vertical
portion of the relevant supply curve, the particular borrower will be
unable to obtain additional borrowed funds by promising to pay
additional interest. Moreover, as the supply curve shifts to the left
and upward, the borrower will encounter progressively more stringent
restrictions on the supply of funds which he will be unable to over-
come by offering to pay more interest. However, another borrower
with a superior credit rating (4 function) may continue to be able to
borrow as much as he wishes and even may not be required to pay
much additional interest to meet the higher qualifications imposed
by the lender. The borrower with the inferior credit rating will
regard "availability" as more important than interest in determining
his access to borrowed funds.
I t is this combination of circumstances which has come to be
known as "credit rationing." Clearly, however, the "rationing"
involved here is quite a different phenomenon from that associated
with a regulated ceiling price. Two points of difference may be empha-
sized. First, the lender is not denying himself (or being denied) an
advantage (higher interest) which he normally seeks, but is behaving
rationally in the face of risk. Second, the supply of loanable funds
to some borrowers remains interest-elastic even after becoming com-
pletely inelastic to others. In this connection we may note that the
lender's indifference curves in Figure I X as well as the supply curves
in Figure VII retain their elasticity even after credit is rationed.
Indeed, since there is always the "unsatisfied fringe of borrowers,"
to make credit rationing depend upon interest inelastic supply without
distinguishing among different borrowers would make our theory
contradict the empirical evidence that there are always some borrow-
ers who can obtain additional accommodation a t higher rates of
interest.
IV. THECENTRAL BANK'SINFLUENCE
There are numerous topics in monetary theory which can usefully
be explored from the vantage point we have now reached. In the
interests of brevity I shall confine my remarks to establishing that
central bank policy can alter the location of the optimum in Figure I X
and, hence, the terms available to the individual borrower in Figure
VII or VIII. Then I shall comment on the implications of this
276 QUARTERLY JOURNAL OF ECONOMICS

process for the ability of the central bank to restrict credit without
"excessive" increases in the effective rate of interest on government
bonds.
Suppose the lender, a commercial bank, happens to be a t an
optimum point PI in Figure IX. Now imagine that the central bank
raises reserve requirements. This has no initial effect on the market
value of the bank's total assets (its wealth position), but it does
require. the bank to sell some earning assets to increase its cash
(including reserve balances). Three effects follow: (1) the volume of
earning assets which the bank can hold is reduced : it must move back
along the opportunity curve toward the origin; (2) to the extent that
the sale of bank assets to replenish its cash position reduces asset
prices, interest rates are raised and a new opportunity curve (say O2)
is created with slope higher than the original opportunity curve;
(3) to the extent that interest rates rise and asset prices fall the market
value of the bank's total resources falls; since the value of its total
liabilities (principally deposits) remains virtually unchanged, it is
now in a more risky position than before. In Figure I X such an
increase in risk has to be shown by an increase in the slopes of the
indifference curves such as represented by I;. All these effects rein-
force each other to produce a new optimum point in Figure I X (say
EY
a t Pz) for which the ratio -per dollar loaned or invested is higher than
EZ
before. This is equivalent to moving to a higher supply curve in
Figures VII and VIII and thus to requiring higher terms from the
borrower.
A similar chain of consequences follows central bank sale of
government securities in the open market. If our bank is typical,
this action will both reduce its reserve balances and its deposit liabili-
ties. The net effect will be to reduce its cash position and hence its
liquidity by a greater proportion than its liabilities and hence to
increase risk. This will steepen the slope of the indifference curves in
Figure I X a t the same time that it reduces total investable resources.
Moreover, the sale of government securities by the central bank will
have raked interest rates and depressed asset values. This raises the
slope of the opportunity line and reinforces the decline in investable
(indeed, total) resources in the hands of the bank. Once again the
EY
ratio - per dollar loaned or invested will be higher a t the new opti-
EZ
mum than a t the old and terms to borrowers will have risen.
 CREDIT RISK A N D CREDIT RATIONING 277

What insight does this analysis permit into the ability of the
central bank to restrict credit expansion without, or with only a
modest, rise in interest rates on government securities? Clearly, any
restrictive action by the central bank, ceteris paribus, must produce
some rise in interest rates on government securities, although the rise
may be almost imperceptible if conditions are favorable. The extent
of the rise in interest which must accompany a given degree of credit
restriction will depend upon the character and composition of aggre-
gate borrower demand. The larger the proportion of marginal borrow-
ers in the aggregate demand (those who do not wish or are unable to
raise when terms increase) the more effectively will credit be
EZ
restricted for a given rise in interest on government securities. The
larger the proportion of determined prime borrowers the greater the
reliance that will ha.ve to be placed on interest rationing as opposed
to credit rationing to restrict credit. Of course, the higher interest
rates go, the more numerous are the borrowers who become marginal
in terms of credit rationing (risk considerations) regardless of their
willingness to commit themselves to higher interest charges. Ulti-
mately every borrower is limited by his capacity rather than his
willingness. Accordingly the higher interest rates go, the more perva-
sive will become the influence of credit rationing. More and more
borrowers will find that '(availability" rather than interest determines
their access to credit. Indirectly this will affect even the prime
borrowers, since the markets they sell in are dependent on effective
demand and this may come to be strongly influenced by the restric-
tions that credit rationing places on their customers.l
The rise in interest rates that accompanies credit restriction (or
increase in borrowers' demand with money supply constant) can,
of course, be restrained by the presence of interest ceilings, either
those which are due to the oligopolistic structure of credit markets or
those due to usury laws. The presence of such ceilings will accentuate
the non-price rationing of credit. However, the credit rationing
occasioned by such ceilings lasts only so long as the ceilings are main-
tained. (Usury ceilings may be presumed to last longer than oligopo-
listic ceilings, but most corporate borrowers are exempt from usury

I. Note that if the central bank tightens credit with member bank reserves
constant, for example by buying short term government securities and selling
long terms, it may exercise its influence directly on the velocity rather than the
volume of the money supply, since it may impede the transfer of idle to active
balances while leaving money supply unchanged.
278 QUARTERLY JOURNAL OF ECONOMICS

ceilings.) The credit rationing which stems from borrowers' credit

ratings, however, is permanent for so long as the general credit situa-
tion which gives rise to it lasts. I t is this more general phenomenon
of credit rationing, long neglected in monetary theory, which has
been the central issue in this article.