A Theory of the Banking Firm

Author(s): Michael A. Klein

Source: Journal of Money, Credit and Banking, Vol. 3, No. 2, Part 1 (May, 1971), pp. 205-218
Published by: Ohio State University Press
Stable URL: http://www.jstor.org/stable/1991279 .
Accessed: 14/10/2013 00:51

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

Ohio State University Press is collaborating with JSTOR to digitize, preserve and extend access to Journal of
Money, Credit and Banking.

http://www.jstor.org

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
 MICHAEL A. KTjFjIN

A Theoryof the BankingFirm

I. INTRODUCTION*

IN SPITE OFTHEIMPORTANCE of commercialbankingboth

as a major financialintermediaryand as an importantlink in the monetary
transmissionprocess,thereis little consensusas to whatconstitutesa workable
and productivetheory of the bankingfirm. Neoclassicalmicroeconomicanal-
ysis is rarelyinvoked to explain bank behavior,primarilybecausethere is so
little agreementeven as concernsfundamentalconcepts.lFor example,do stock
or flow variablesmeasurethe relevantconcepts of bank output and input?If
neitherinput nor output can be appropriatelydefinedit becomespresumptu-
ous to speakof a productionfunctionrelatingthe two.
In the face of conceptualdiflicultiesin drawingthe analogybetweena bank
and the typical firm of neoclassicalanalysis,most treatmentsof the bank at
the microlevelhave concentratedon one specificproblem;the allocationof a
bank's funds among competing stocks of assets.2That is, a bank has been
treated,not primarilyas a firm but as a rationalinvestorin an environment
characterizedby risk or uncertainty.The neoc]assicalanalysisof the firm has
yieldedto portfoliotheory.
If there is relatively little microanalysisof the banking firm, there is a
plethora of literaturerelating the impact of bank market structureon per-
* Assistant Professor,Departmentof Economics, Indiana Universityon leave to the Fed-
eral Deposit InsuranceCorporation.The author would like to thank the Corporationfor its
researchsupport.Policy implicationsderivedfrom the analysisare those of the author and are
not to be attnbuted to the F.D.I.C.
1Two exceptionsare John Karaken [8] and F. W. Bell and N. B. Mulrphy[1].
2 Perhapsthe best exampleis RichardPorter [11].

A. KLEINis assistant
MICHAEL professorof economicsat theIndianaUniversity,Bloom-
ington.

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
206 : MONEY,CREDIT AND BANKING

formarlce.3 Whateverthe meritsof such studiesit does seem prematureto ask

how bank behavioris affectedby variationsin marketstructurewhen we have
no theory to describe that behavior under any specific set of assumptions
about the externaland competitiveenvironmentin which the bank operates.
Such an environment,in turn, is largely determinedby bank regulation.It
follows that, in order to come to some tentativeevaluationof the impact of
such regulation,a theory of the bankingfirm is required.
The purposeof this paper is to set forth such a theory in a mannerwhich
specificallyallows for the role of marketstructureand competitionwithin the
structuralrelationswhichthe bank confronts.Thus, the model shouldprovide
an analyticalframeworkfor appliedresearch.The next section sets forth the
generalnatureof the model. In SectionIII, the detailedstructureof the model
is laid out and a solution for the bank decisionvariablesis found. SectionIV
appliesthe precedinganalysisto the problemof interestrateregulation.Finally,
the last sectionsummarizesthe study and presentssome suggestionsfor future
research.

Il. SOME
GENERAL
CONSIDERATIONS

Any model of the bankingfirm must, due to the complexityof the institu-
tion, be relativelyabstract.The nature and degree of the abstractionwill be
determinedby the author'sconceptionof what is particularlyimportantabout
a commercialbank. Unfortunately,argumentsover such mattersoften seemto
approachthe metaphysicalwith the result that little of value emerges.Never-
theless, it is importantat the outset to set forth the generalway in which this
paper will view a commercialbank.
A bank is, firstly,a subset of financialintermediariesin general.That is, it
securesfundsfrom surplusspendingunits and transmitsthem to deficitspend-
ing units. Although the specific assets purchasedby banks may differ from
those of other intermediaries,this is of secondaryimportance.Banks are dis-
tinguishedfrom otherintermediariesbecausethe formercan attractone source
of funds, demanddeposits,withoutthe paymentof explicitinterest.This is so
becausedemanddepositsare a generallyacceptablemediumof exchangewhich
is superiorto currency(whichalso bearsno explicityield) in a wide varietyof
transactions.
The issuanceof demanddeposits means that banks are the administrators
of the nation'spaymentsmechanism.Such administrationconstitutesa service
providedby the bankingsystem to the non-bank public. Two points should
be noted. First, scarce resourcesare utilized in the provision of this service.
That is, there is a social cost to the utilizationof the paymentsmechanism.
3See, for example,T. G. Flechsig [5] and FranklinR. Edwards [3] and the literaturecited

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
 MICHAELA. KLEIN : 207

Secondly,banks must determinethe minimumpricenecessaryto inducethem

to providethat service.It follows that an economictheoryof the bankingfirm
mustexplainthe processwhichdeterminesthe pricechargedfor these services.
Specifically,is there a relationshipbetweensuch pricingpolicy and the rate of
interestofferedby bankson a stock of demanddeposits?If so, what is the im-
pact of the legal prohibitionof explicit interest on demand deposits on the
price of this uniquecommercialbankingservice?Subsequentanalysiswill at-
tempt to elucidatetlae answersto these importantquestions.
Funds securedby banksin the form of time deposits,demanddeposits,and
ownershipclaims are investedin a wide varietyof earningassets, the revenue
from which constitutesthe main source of bank income. These assets differ
widely in terms of their expectedreturns,marketand/or defaultrisk charac-
teristics,liquidity,and so forth. One characteristicdistinguishingassets which
is rarelyrecognizedwill here play a centraltole. We refer to the elasticityof
asset supplyto the individualbank. Assets such as governmentsecuritiescan
be said to be in perfectlyelastic supply to the individualbank since the ex-
pected returnand risk characteristicsof such securitiesare unaffectedby in-
dividualbank decisionsas to whetheror how much of these securitiesare to
be purchased.Privatesecurities(loans), on the other hand, are in imperfectly
elastic supply to the individualbank. Ceterisparibus,if a bank wishes to in-
creaseits loan/asset ratio it must accepta reductionin the marginalreturnon
loans. The author has arguedelsewherethat failureto recognizethis distinc-
tion constitutesa majorweaknessin traditionalportfoliotheoreticmodels and
is responsiblefor the almostnonexistentuse of such modelsin appliedbanking
research[9].
The model to be set out in the next section seeks to explain the following:
(1) the equilibriumscale of the bank, (2) the compositionof the bank's asset
portfolio, (3) the compositionof the bank's liabilities,(4) the rate of interest
on bank loans, (5) the yield the bank oSers on its time and demanddeposit
accounts.

III. THE MODEL

A. The Basic Setup

The bank is assumedto have a preferenceorderingover P, the rate of re-
turn conequity, whichcan be representedby a utilityfunctionlinearin P. His
decisionrule is to maximizeexpectedutilityor, equivalently,the rate of return
on equity.4It is necessaryat the outsetto providea generalformulationof that
4 A quadraticutility function would have been more general.However, the increasedcom-
plexity of the algebrawas felt to outweighthe benefitsof increasedgenerality.Michael Klein
[9] deals with the problemof imperfectasset elasticitiesutilizinga quadraticutility furlction

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
208 : MONEY,CREDIT,AND BANKING

rule.The bankhas two primarysourcesof funds;the equityoriginallyinvested

in the firm, denoted as W, and borrowedfunds securedthroughthe issuance
of varioustypes of deposits,denotedas B. Assumethe bank issues m types of
deposits Bs (i = 1 *s m) at rates of interestdenoted as R. It follows that
Si BS= B. Let a, denote the proportionof total funds F obtainedthrough
the issuance of the ith deposit type. Then,

F= W+E,B, (1)

But

B,= ot,F (2)

Therefore,

F = W + FE, oes or W-F[1- E cri] (3)

Fundssecuredfrom equityand the issuanceof depositsare allocatedamong

n asset classes. Let Xs denote the proportionof total funds allocatedto the
ith assettype (j = 1 ***n) and let Ej denotethe expectedrate of returnon that
asset. By the balancesheet constraint,

EX>= 1 (4)

The expectedrateof returnon total funds,EF, is givenby

EF= EX1ES- Ecl!,Rs (5)

Finally, the expectedrate of returnon equity, Ew, is given by equation (6).

Ew= EF EXjEj-E
j *
°t2.R2. (6)
1-St, l-Eti

We now turn to an explicitstatementof the returnson bank assets and the

costs of bank borrowing.

B. The Returnon BankAssets

The asset universeconfrontingthe bankeris assumedto consist of cash, a
homogeneousgovernmentsecurity(a consol), and private securities(loans).

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
 MICHABLA. KLBIN : 209

Reserverequirementsand other restrictionson asset choice are ignored. Let

us begin with privatesecurities.s
We assumethat the bank confronts a demandcurve for loans which is a
functionof the contractrate of interestr and a vector of exogenous variables
which influencethe state of loan demandconfrontedby a particularbank. It
is furtherassumedthat borrowersare viewedas a homogeneousgroup by the
bankandthat all noninterestloan termsarefixedandidenticalto all borrowers.
Let XLdenote the proportionof funds allocatedto privatesecurities.Then,

r = f(XL), J (XL)< Q (8)

Equation(8) is the analogueof a demandcurvefor loans.

Unless the bank assigns a zero probabilityto the events of partialor com-
plete defaulton the privatesecurity,the expectedreturnon loans ELmust be
less than the contractrate of interestsince the latter representsthe maximum
returnthe bank can receive.Generally,

EL< r if aL > ° (9)

whereaLis a measureof defaultrisk, the standarddeviationof the probability

distributionof loan payments.Since borrowersare assumedto be identical
and since the noninterestterms of the loan are assumedexogenous and the
same for all loans, we have aL = aL. That is, defaultrisk is exogenousto the
bank.6From (8) and the above discussionwe get

EL = *(XL), h (XL)< O (10)

Unlike privatesecurities,governmentsecuritiesare free of defaultrisk and
are in perfectlyelastic supplyto the individualbank. Such assets constitutea
secondaryreservewhichcan be liquidatedrapidlyshouldan unexpecteddeposit
loss exhaust t}e bank's cash holdings. Under such circumstancesthe resale
price of the securitiescan be viewed as a random variablewhich, in turn,
means that the holding period rate of return,denoted as , is also a random
variablewith densityfunction+(g). The expectedrate of returnon government
securitiesE,, is, therefore

Es= J g¢(g)dg (11)

-1

The bank decision variableis denoted as Xa, the proportionof government

securitiesto total assets.
5A more detailedanalysisis given in [91.
6 This is unlikelyto be true. However,we are abstractingfrom problemscomected with loan
risk appraisal.

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
2 IO : MONEY,CREDIT,AND BANKING

Finally, we turn to the bank's cash holdings. To this point, random ele-
mentshave enteredthe analysisonly with respectto the rate of returnon bank
earningassets.The explicityield of cash, however,is nonrandomand equal to
zero. Nevertheless,cash does yield an implicitreturn.An increasein cash hold-
ings reducesthe likelihood of a cash deficiencyand, if there is some penalty
cost for such a deficiency,the reductionin the expectedloss obtainedfrom the
holdingof cash can be viewedas the implicityield of this asset.7
Thus, it is necessaryat this point to introduceexplicitlyanotherelement of
random variation in the externalenvironmentconfrontingthe bank. In the
next subsectionit will be shownhow the bank determinesthe pricesit will pay
for various types of deposits and how these prices, in conjunctionwith the
depositsupplyfunctionsthe bankconfronts,determinethe scale and composi-
tion of the bank's deposit liabilities.Such supply functions denote, for each
price,the expectedvalue of the depositliabilitiesthe bank will assume.How-
ever, the transactionsof the bank's depositorsset up a flow of reservesinto
and out of tlle bank in question.The possibilitythat, for a period of time a
net outflow of deposit funds will occur, cannot be neglected.
At any given point in time, a bank is receivingreservesfrom both public
flows of currencyand drafts drawnon other banks. Its disbursementsfollow
a similarpattern.Net disbursementsare definedas disbursementsminus re-
ceipts and can be viewedas a randomvariable.Let z denotenet disbursements
as a fraction of total funds and assumethat z has density k(z). The bank is
presumedto incur a penalty if its cash holdings are insufficientto meet its
disbursements.8 Viewedin this manner,cash is held for precautionaryreasons;
it is an asset held in orderto meet a liability which is stated in fixed dollars
and whose time of repaymentis unknown.9
Let the penaltycost per dollarof cash deficiencybe denotedas n. If the lbank
held no cash it would expecta loss equalto
c

n J Zk(z)dz ( 12)

where c is the largestnet disbursementto which the bank assigns a nonzero

probability.If the bank holds cash as a proportionof total funds equalto XTX
the expectedloss on cash managementis
rc rc

nJ
Xr
(z-Xr)k(z)
dz < nJ
°
zk(z)dz (13)

7 The analysisthat follows is indebtedto the work of George R. Morrison [10].

8 A bank will liquidate governmentsecuritiesor borrow dependingon the costs of each
method of adjustment.For simplicity,we assume that the marginalcosts of adjustmentare
identicalfor both methods.
9 See EdwardL. Whalen [131.

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
 MICHAEL A. KLES : 2I I

For simplicityassumethat k(z) is rectangularand equalto l/(c-b) whereb

is the lowest conceivabledepositloss (highestpossiblegain). Then

nt (z Xr)k(Z)
dz = n[(( r_ b) ] (14)

Equation (14) representsthe expectedloss, expressedas a weightedrate of

return,from the bank's cash managementpolicy.

C. The Deposit Supply Functions: General Formulation

The banks is assumedto issue demand deposits, Dl, and time deposits,
D2 . Two characteristicsdistinguishingthese liabilitieswill be relevantto our
later analysis.First, demanddepositsare a media of exchange,and depositor
transactionsutilizingthis mediaimpose a cost on the issuingbank. Secondly,
law precludesthe paymentof explicitintereston a stock of demanddeposits,
but does not precludesuch intereston time deposits.At this point, however,
it will be advantageousto place these distinctionsin the backgroundand pro-
ceed, temporarily,with a more generalformulation.Then, in the next section,
we can utilize the distinguishingstructuraland regulatorycharacteristicsof
demand deposits in an analysis of the impact on bank behaviorof the pro-
hibitionof intereston these deposits.
We content ourselveswith a very generalformulationand assumethat the
supplies of time and demand deposits to the individualbank are increasing
functions of the yields, implicit and explicit, which the bank offers on these
accounts.l°Specifically,

D1= Dl(Rl), Dl'(Rl) > O (15)

D2= D2(R2), D2'(R2) > O (16)

Further,define

= F (17)

- ( 18)

That is, cz denotesthe proportionof total funds obtainedthroughthe issuance

of the ith deposittype.

10They are, of course, affectedby other variables.This will be discussedin Section IV.

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
2 I2 : MONEY,CREDIT,AND BANKING

D. The Solutionof the Model

the preceding
Substituting structural provided
relationsintothe framework
by equations(1)-(6), we have

Ew= [1 _ _ ] [XLh(XI,)+ X,,E,,_ n[2( * _ c))] _ aIR1-cg2R2 ( 19)

Now

1- - = W= 1 + ( W ( ) (20)

Similarly,

1-oel-a2 W (21)

and

1-ctl-Ct2 W (22)

the aboveinto (19)yields

Substituting

Ew = [1 + 1( ) W ( ] [XLh(X) + XeEg-n 2(c-b) ] (23)

-W [RlDl(Rl) + R2D2(R2)]

(23)is to be maximizedsubjectto E Xj = 1.
The methodof undetermined multipliersyieldsthe followingfirst order
conditionsfor a profitmaximum,wherer is the Lagrangean multiplier:

dEw= D1 (R1) Ea- W(RlDl(R1) + Dl(Rl)) = ° (24)

Ea-
dEw= D2WR2) W(R2D2
(R2) + D2(R2)) = ° (25)

dEw -1 + Dl(Rl) + D2(R2)] (X,.h'(Xz,)+ h(X))-r = o (26)

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
 MICHAEL A. KLEIN : 2 I3

dJEw= [1 + Dl(Rl) + D2(R2)] E -r = O (27)

dEw = [1 + Dl(Rl) + D2(R2)] [-n (Xr-c)] _ r = ° (28)

aEw = XL+ XR+ Xr-1 = ° (29)

where

Ea= XL^(XL) + XaEa-n [(Xr-c) ]

The solution for the bank decision variablesis relativelyeasy. From (26)-
(28)we have

XLA(XL) + ^(XL ) = = -n
Esw [Xr_ b ] ( 30)

Now the expressionon the left is simplythe marginalreturnon loans. Thus,

XLis chosen at the point at whichthe marginalreturnon loans is equalto the
average(and marginal)expectedreturnon governmentsecurities.
A similarresultholds for the bank'scash holdings.Let us call the reduction
iD the expectedcost imposed by a cash deficiencyfrom holding Xr of assets
as reserves,the total expectedreturnon cash holdings,Er. That is,

Er= n | O zk(z)dz-n | Xr(z-Xr)k(z) dz (31)

Performingthe indicatedoperations,

Er= 2( _rb) (2c-Xr) (32)

The marginalreturnfrom an incrementin cash holdingsis given by equation

(33).

dX, c-b ( r) ( )

If we multiplyboth n and (c-Xr) by minus unity, it is seen that the expreso

sion is identicalwith the right hand side of (30). Thus, cash is held until its

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
2 I4 : MONEY,
CREDIT,AND BANKING

marginal (implicit) return is equal to the expected return on government

securities.This completesthe solutionfor the assetselectiondecisionvariables.
We turnnow to the ratesof interestthe bank will offeron its deposits.From
(24) we get

R1 = Ea_ D1,((RR)9 (34)

and from (25) we get

R - E _ D2(R2) (35)

This completesthe solutionfor the bank decisionvariables.

IV. APPLICATION
AND INTERPRETATION
OF THEMODEL

A detailedexaminationof the content of currentregulatorypolicy and its

impact on bank market structureand behavior is beyond the scope of the
presentpaper.Nevertheless,the precedingmodelhas a numberof implications
respectingthe desirabilityand likely impact of diverse forms of regulatory
policy. In addition,such an analysishas implicationsfor futureresearchin the
bankingareaand for this reasonit is importantto make a preliminaryattempt
at providinga frameworkfor the implementationof such analysis.

A. The Reg?>lationof Interest Rates on Demand Deposits

Interestrate regulationis probablythe single most conspicuousfacet of the

limitationson competitivebehaviorimposedby the regulatoryauthorities.All
federallyinsured commercialbanks are prohibitedfrom paying explicit in-
tereston demanddepositsand are subjectto restrictionson the maximumrate
payable on time and savings deposits. For memberbanks, these restrictions
are imposedby the FederalReserve.Nonmemberfederallyinsuredbanks are
subjectto identicalrestrictionsimposedby the F.D.I.C. The above model has
a number of implicationsfor such regulation.We shall concentrateon the
zero interestceiling for demanddeposits since the generalizationto other in-
terestceilingsis straightforward.
One of the initialjustificationsfor interestrate regulationwas that competi-
tion for deposits between banks would lead to 'unsound'portfolio policies.
Driven by higherinterestrates on its sourcesof funds, a bank was presumed
to seek out high yield (and high risk) uses of funds. George Benston [2],
among others, has found little empiricalevidenceto substantiatethis claim.
An appropriatequestionat this point is whetheror not such behaviorshould

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
 MICHAELA. KLEIN : 2 I5

be expected on a priori grounds. The answer is given by the optimization

condition (30) for the bank asset selection variables.Neither the cost ot de-
posits nor the parametersof the deposit supply functions appearin the op-
timizationconditionand therefore,cannot aSect asset selection.llOn the other
hand,the portfolioyield Eadoes aSectthe interestratebanksare willingto pay
for depositsas is seen by examinationof (34) and (35).
Turningto the eSects of a zero interestceilingon demanddeposits,equation
(34) providesus with the implicityield whicha profit-maximizing bank would
provide in order to induce depositors to hold its demand liabilities. It now
behooves us to definethe yield on demanddepositswith some care. If banks
are prohibitedby law from paying a positive price directlyfor a productive
input,competitioninsuresthat individualbankswill inducedepositorsthrough
other forms of price concessions.Thus, depositorsmay be given preferential
priceor queuingtreatmenton loans, or they may be providedwith 'free" an-
cillaryservices,etc.l2Still anotherpossibilityis to reducethe chargea bank im-
poses on administeringthe transactionsmechanismbelowthe cost of providing
these services.An interestingproblemconcernsthe extent to which these dif-
ferent methodsof adjustmentare utilizedby individualbanks in orderto at-
tract and keep demanddepositaccounts.In whatfollows we shall assumethat
the only outlet for such price competitionis in the setting of servicecharges
on demanddeposit accountactivity.
Transactionsservices provided by banks utilize scarce labor and capital
resources.Let us denotebank outputin this activityas A, whereA is the num-
ber of transactionsper account per time period.l3Furtherassume that A is
relatedto the inputsof capitaland laboraccordingto a Cobb-Douglasproduc-
tion function exhibitingconstant returnsto scale. Then, if the prices of the
servicesof capitaland labor are assumedexogenousto the bank and invariant
to the level of bank utilization,the cost of providingthese servicescan be ex-
pressedby an equation such as
C= eA, e> O (36)
where C denotes the total cost per account per time period of providingthe
servicesof the paymentmechanism.
Furtherwe specifythat the total servicechargeper accountper time period,
S, bears a linear and proportionalrelationshipto the degree of account ac-
tivity. Thus,
S= dA, d2° (37)
whered is the basic bank decisionvariable,the servicechargeper transaction.
11If the parametersof the net disbursementsfunction are affectedby the demand deposit-
time depositmix, this conclusionwould have to be modified.I am indebtedto AlfredBroaddus
for this point.
12 Donald Hodgman [6] and E. Kane and B. Malkiel [7] take up this issue.
13 A is assumedto be unaffectedby bank servicechargepolicy.

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
2 I6 : MONEY,CREDIT,AND BANKING

The dollarvalue of demanddepositsaccountscan be expressedas a product

of the numberof accounts,M, and the averagesize of an account, N. If ac-
tivityper accountis identicalacrossaccounts,we can defineR1as

R1 = (e d)AM= (e-d)A (38)

That is, the implicityield which a bank offerson demanddeposit accountsis

definedas its operatingloss per time period in providingpaymentsservices
expressedas a fractionof total demanddeposits.
Equation (34) determinesthe yield on demanddeposits and (38) provides
the level of d which is necessaryin orderto obtain that yield at an individual
bank. From (34) it can be seen that

R > 0 if E > Dl(Rl)

Therefore,

d < e if E > Dl(Rl)

Even if R1 is the same for two differentbanks, the servicecharge,d, may be

differentdue to variationsin A, N, and e acrossbanks.
In this connectionit should be noted that the practiceof viewingthe ratio
of bank servicechargesto the stock of demanddepositsas a negativerate of
return on demand deposit holdings is conceptuallyincorrect.Much use has
been made of this statisticby Edgar Feige [4] and others. The model shows
that, in fact, banksmay inducedepositorsto hold demanddepositsby provid-
ing a positive,albeitimplicitreturnon these holdings.A roughmeasureof this
rate of returncan be found by takingthe differencebetweena bank'scosts of
administeringthe paymentsmechanismand bank servicechargerevenueand
then dividingthis differenceby the stock of demanddeposits.
As an illustration,we utilizedata from the FederalReserveFunctionalCost
AnalysisProgramfor 769 banks with total depositsless than $50 million. In
1967, the averagebank had approximately$8 million in regularcheckingac-
count funds.Thesefundsgeneratedtransactionscosts of $177,000whileservice
charge income was only $53,000. Thus, the implicit rate of return on these
depositswas approximately1.6 percent.We concludethat the prohibitionof
intereston demanddepositsis at least partiallyoffset in the above manner.

B. Role of Structureand Competitionin the Model

Accordingto the precedinganalysis,the offeringrates on bank depositsare
functions of the profitabilityof bank lending and the parametersof the de-

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
 MICHAELA. KLEIN : 2 I 7

posit supplyfunctions.These parameterscould be expectedto differcross-sec-

tionally for two reasons. First, those economic variables(such as per capita
income) which affect the demandfor financialassets differcross-sectionally.
Secondly, marketstructureand the degreeof bank competitionexhibitrather
large degreesof cross-sectionvariation. These three types of variables,then,
play an integralpart in the analysisof bank offeringrates on deposits.
The fact that external economic and marketstructurevariablesare an in-
tegral part of the model means that the model's empiricalimplicationsare
rather extensive. Two examples will be cited. Let us first look at the asset
selectionprocess. Sincethe modelmakesallowancesfor imperfectassetelastici-
ties, differencesin loan demandacrossbanks will lead to differentasset selec-
tion choices. Traditionalportfolio theory is silent on this problem.l4Since
market structureand competition can be expected to affect the shape and
positionof the h function,these variablesare also relevantto the assetselection
process.
On the liability side, if the D1 and D2 functionswere identical,the implicit
rate on demanddeposits would be set equal to the explicit rate on time de-
positsby an individualbank.In this connectionthe followingproblememerges:
demanddepositsappearto be considerablymore profitablefor banksthan are
time deposits. As pointed out earlier,FunctionalCost analysisdata puts the
averagecost of demand deposits for the bank sample discussedat approxi-
mately 1.6 percent.The correspondingfigurefor time deposits is in excess of
4.3 percent.The obvious explanationthat time deposits carry an explicit in-
terestexpensewhereasdemanddepositsdo not is clearlyinadequate.If demand
depositsaremoreprofitablethan time deposits,why don't individualbankscut
servicechargesfurtherin order to capturethe accountsof other banks?
A plausiblehypothesiswould startfrom differencesin the competitiveforces
a bank confrontsin securingthe two types of accounts.In a recent study for
the FederalReserveBoard,BernardShull [12]concludesthat nonlocalcompe-
tition for time and savingsdepositsforces banksin isolatedone and two bank
towns to raise their offeringrates on these types of deposits. Specifically,his
findingsseemto indicatethat local marketstructureis of lesseningimportance
in understandingthe functioningof time and savingsdepositmarkets.
It seemsreasonableto supposethat sincedemanddepositsareused primarily
for transactions,the proximityof the depositorto the bank is of prime im-
portance.Competitionby banks withinthe local area may lead the depositor
to substituteone bankfor the other,but it is unlikelythat nonlocalcompetition
can have a similareffect.This is not to say that thereis no compensationwhich
can induce a depositorto shift his checkingaccountto a nonlocal bank, only
that within the relevant range (rememberingthat explicit payments are il-
legal) such substitutionis likely to be minor. This is an area in which future
researchcould be highlybeneficial.
l4seeKlein [9]

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions
2 I8 : MONEY,CREDIT,AND BANKING

F. Concluding Comments
Most of the questions raised in this paper cannot be resolved by a priors
theorizing;basically,they are empiricalissues. It has been the purposeof this
paperto demonstratethat the developmentof a simplemicroeconomicmodel
of the bankingfirm is an importantfirst step both in discerningwhat are the
problemsof interestin appliedresearchand in suggestingplausibleand testa-
ble hypothesesconnectedwith them.
The neoclassicalanalysisof the firmfirst developedmdividualbehavioruw-
der a specifiedset of assumptionsabout the externaland competitiveenviron-
ment withinwhichthe firm operated.Only then did it ask how firm behavior
is aSectedby variousin these conditions.The literatureon bankingappearsto
be attemptingto answerthe second problemwithout dealingwith the first.

LITERATURE
CITED

1. BELLL, W. and NEIL B. MURPHY.Cost in CommercialBanking: A

FRBDERICK
QuantitativeAnalysis of Bank Behaviorand its Relation to Bank ReguZation.
ResearchReportNo. 41 to the FederalReserveBank of Boston.
2. BENSTON,GEORGEJ. ''IrlterestPayments on Demand Deposits and Bank In-
vestmentBehavior."Journalof Political Economy, 72 (October, 1964), 431-
49.
3. EDWARDS,FRANELIN R. Concentrationand Competitionin CommercialBank-
ing: A StatisticalStudy. ResearchReportNo. 26 to the FederalReserveBank
of Boston.
4. FEIGE,EDWARDL. The Demand For Liquid Assets: A TemporalCross-Section
Analysis. EnglewoodCliffs:PrenticeHall, 1964.
5. FLECHSIG, lkEoDoRs G. BankingMarketStructureand Performancein Metro-
politanAreas. Boardof Governorsof the Federal ReserveSystem, 1964.
6. HODGMAN,DONALD.CommercialBank Loan and Investment Policy. Cham-
paign:Universityof Illinois Press, 1963.
7. KANE,EDWARDJ. and BURTONMAT.KIP.X-. "Bank Portfolio Allocation, Deposit
Variability,and the AvailabilityDoctrine." QuarterlyJournalof Economics,
74 (February,1965), 113-35.
8. KARAKA,JOHNH. "CommercialBanks and the Supply of Money: A Market
Determined Demand Deposit Rate." Federal Reserve Bulletin, 53 (October,
1967), 169>1712.
9. KLEIN,MICHAEL A. "ImperfectAsset Elasticity and Portfolio Theory,"Ameri-
can Economic Review (in press).
10. MORRISON, GEORGER. Liquidity Preferences of CommericalBanks. Chicago:
Universityof ChicagoPress, 1966.
11. PORTER, RICHARD C. "A Model of Bank Portfolio Selection." Yale Economic
Essays, 1 (1961), 323-59.
12. SHULL,BERNARD.'Nonlocal Competition for Time Deposits in Isolated One
and Two Bank Towns."Stafl Economic StudiesNo. 40. Board of Governors
of the FederalReserveSystem.
13. WHALPN,EDWARDL. "A Rationalization of the Precautionary Demand for
Cash."QuarterlyJournalof Economics, 80 (May, 1966), 31S2S.

This content downloaded from 189.100.215.218 on Mon, 14 Oct 2013 00:51:15 AM

All use subject to JSTOR Terms and Conditions