Public Disclosure Authorized

10095

Monetary Policy
Instrulments
Public Disclosure Authorized

for Developing
CCountries
Public Disclosure Authorized

¢. ~~~0

edited by
Gerard Caprio, Jr.
Patrick Honohan
Public Disclosure Authorized

AWorld
BankSymposiwn-

A WorldBan Sympsiu
 lZ i - ;

MonetaryPolicyInstruments
for DevelopingCountries

editedby
Gerard Caprio,Jr.
Patrick Honohan

A World Bank Symposium

The WorldBank
Washington,D.C.
© 1991 The International Bank for Reconstruction
and Development / THE WORLDBANK
1818 H Street, N.W., Washington, D.C. 20433, U.S.A.

All rights reserved

Manufactured in the United States of America
First printing October 1991

The findings, interpretations, and conclusions expressed in this study are entirely those of
the authors and should not be attributed in any manner to the World Bank, to its affiliated
organizations, or to members of its Board of Executive Directors or the countries they rep-
resent.

Because of the informality of this series and to make the publication available with the least
possible delay, the manuscript has not been edited as fully as would be the case with a more
formal document, and the World Bank accepts no responsibility for errors.

The material in this publication is copyrighted. Requests for permission to reproduce por-
tions of it should be sent to Director, Publications Department, at the address shown in the
copyright notice above. The World Bank encourages dissemination of its work and will
normally give permission promptly and, when the reproduction is for noncommercial pur-
poses, without asking a fee. Permission to photocopy portions for classroom use is not re-
quired, although notification of such use having been made will be appreciated.

The complete backlist of publications from the World Bank is shown in the annual Index
of Publications, which contains an alphabetical title list and indexes of subjects, authors,
and countries and regions. The latest edition is available free of charge from the Publica-
tions Sales Unit at the address in the copyright notice or from Publications, World Bank,
66, avenue d'1ena, 75116 Paris, France.

Gerard Caprio, Jr., is a senior financial economist, Financial Policy and Systems Division,
Country Economics Department, World Bank. Patrick Honohan, a research professor at the
Economic and Social Research Institute, Dublin, was at the time of writing a senior econ-
omist in the division.

Library of Congress Cataloging-in-Publication Data

Monetary policy instruments for developing countries / edited by

Gerard Caprio, Jr., Patrick Honohan.
p. cm. - (WorldBank symposium)
Papers from a conference sponsored by the Country Economics
Department of the World Bank
Includes bibliographical references.
ISBN0-8213-1969-8
1. Monetary policy-Developing Countries-Congresses. I. Caprio,
Gerard. II. Honohan, Patrick. III. International Bank for
Reconstruction and Development. Country Economics Dept.
IV.Series.
HG1496.M635 1991
332.4'91724-dc2O 91-38837
CIP
a

Foreword

Rapidstructural change and widespreadadoption of fi-
nancial sector reforms in developing countries have
placedpressure on traditional instruments of monetary
control. It is widelyacceptedthat, if the necessarymac-
roeconomiccontrol can be maintained,a move to an in-
direct, market-oriented system of monetary policy
instruments will help the financial sector perform in a
sounder and more efficient manner, resulting in the
maximum contribution to economicdevelopment.
It is not long since many industrial countries adopted
sweepingadjustments of their systemsof monetarycon-
trol, dismantling bank-by-bankcredit ceilingsand sec-
toral credit allocations. A number of middle-income
countries have also progressed along similar paths.
Manyother countries are now lookingat the experience
of these pioneersto see what lessonscan be learned.
With these developmentsin mind, the FinancialPol-
icy and SystemsDivisionof the WorldBankorganizeda
seminar in May,1990,which brought together experts
from industrial and middle income countries, together
with some of the Bank's own financialsector specialists
and those of the International MonetaryFund, to discuss
the lessonsof recent experiencewith indirect methods of
monetary control. The seminar formed part of the
Bank's ongoing effort to evolve and disseminate best
practice in various aspects of financialsector reform. It
is hopedthat this volume,which reports the editedpro-
ceedingsof the seminar,will be of valueto policymakers
and students of developingcountries.
In keepingwith the informal character of the semi-
nar, the viewsexpressedshould be regarded as personal
ones: they do not necessarilyreflectthe viewsof any in-
stitution.
MillardF.Long
Chief,FinancialPolicyand SystemsDivision
TheWorldBank

Acknowledgments
The editorsacknowledgethe contributionof a number of peopleto the productionof this volume.RobertKatt pro-
vided invaluableassistancein editing the transcribedportion, proofing,and laying out the finishedvolume.Karin
Waeltidiligentlyassisted in the transcription processitself,while Wilai Pitayatonakarnand MeganPomeroyorga-
nizedthe word processingfor later drafts. Regardingthe seminar itself,YahayaDokaand Susan Sebastianassisted
with organizationalissues,while MillardLongand AlanGelb contributedthe inspirationand impetus for both the
seminarand this volume.WealsogratefullyacknowledgeJ.B.Zuluand DouglasScott of the InternationalMonetary
Fund's Central BankingDepartmentfor providingthe assistanceof their unit, and severalanonymousrefereesfor
helpful comments.Nonetheless,the opinionsexpressedherein reflect only those of the authors, participants,and
the editors.
The WorldBankconference"MonetaryPolicyInstruments for DevelopingCountries"took place on May16-18,
1990,in Washington,D.C.,under the sponsorship of the CountryEconomicsDepartment.

iii
 Contents

Contributors and Participants ix

1. The Useof MarketInstruments for MonetaryPolicy 1

GerardCaprio,Jr. and Patrick Honohan

The Objectivesof MonetaryPolicy 1
Instrumentsfor SmoothingBank Liquidity 3
Copingwith Shocks 6
The Impactof GovernmentDeficitFinancing 7
ExternalFlowsand MonetaryPolicy 7
ConcludingCommentsand Outlineof the Volume 8

PARTI. NUTSANDBOLTS:PRACTICAL
ISSUESONMOVINGTO
INDIRECTIMPLEMENTATIONOFMONETARY POLICY

Introduction 13

2. Central Bank LiquidityManagementand the MoneyMarket 15

Paul Meek
Central BankMonetaryManagement 15
GeneralGuidelinesfor MoneyMarketDevelopment 22
The Caseof Indonesia 24

Discussion 29

3. The Useof MonetaryPolicyInstruments by DevelopingCountries 37

R. Barry Johnston
WhyIs There a Needto ReformMonetaryControlTechniques? 37
KeyElementsin a LiberalMonetaryControl Framework 38
Managingthe Reformin MonetaryControlTechniques 42

Discussion 44

v
4. BuildingFinancialInstitutions for a Market-BasedMonetaryPolicy 49

Steven Grenville
AnotherModelfor OpenMarketOperations 49
Issuesfor the AustralianModel 51
Useof OpenMarketOperationsto AffectMonetaryDemand 53
Conclusions 60

Discussion 61

PARTII. MONETARY
TARGETING
ANDCONTROL

Introduction 65

5. MonetaryTargeting:Lessonsfrom the U.S. Experience 67

DavidLindsey
1970through October1979 67
October1979Throughthe Fallof 1982 68
The Fallof 1982to October1987 70
October1987to the Present 73

Discussion 80

6. MonetaryTargets:EuropeanExperience 83

CharlesGoodhart
OnMonetaryTargets 84
The EuropeanMonetarySystem 87
Central BankIndependence 88
Comment:PolicyConstraintsin DevelopingCountries
DonaldJ. Mathieson 91

Discussion 93
An Elaboration 98

PARTIII. BUDGETDEFICITSANDMONETARY
POLICY

Introduction 103

7. Impact of GovernmentDeficitson MonetaryPolicy:The Caseof Italy 105

CesareCaranza

The Economicand FinancialPicture 105

A NewStrategyfor MonetaryPolicy 105
Resultsof the NewPolicy 106
The RemainingProblems 107
PotentialSolutions 107

vi
8. Impact of GovernmentDeficitson MonetaryPolicy:The Case of Chile 109

Juan AndresFontaine
Factorsin Chile'sFinancialLiberalization 109
ManagingMonetaryPolicyin Chile 110
Debt-EquitySwaps 112

9. MonetaryManagementwith High Inflation:The BrazilianExperience 115

CarlosAlberto Queiroz
Developmentof the MoneyMarketand MonetaryPolicy 115
TheYearof Super-HighInflation 116

Discussion 118

PARTIV.THE INTERACTION
OF EXCHANGE
RATEANDMONETARY
POLICY

Introduction 123

ExchangeRatePolicy
10. 125

Charles Freedman

Some Definitions 125

FixedVersusFlexibleRates 125
MonetaryPolicyand InterventionPolicy 127
Sterilizationof Changesin InternationalReserves 128

11. Interaction of ExchangeRatePolicyand MonetaryPolicy:

The Caseof Malaysia 131

Lin See Yan

Malaysia'sEconomicBackground 131
ExchangeRate Policyand MonetaryPolicy 132
Conclusion 134

Discussion 134

vii
 Contributorsand Participants

Contributors MichaelKlein
Country Department, Europe, the Middle East and
GerardCaprio,Jr. North Africa,The WorldBank
FinancialPolicyand SystemsDivision,TheWorldBank DavidE. Lindsey
Boardof Governorsof the FederalReserveSystem
CesareCaranza
ExecutiveDirector,The WorldBank Paul Meek
Paul MeekAssociates,formerlyFederalReserveBankof
CharlesFreedman NewYork
Bank of Canada
DonaldJ. Mathieson
Stanley Fischer ResearchDepartment,InternationalMonetaryFund
MIT,formerlyChiefEconomist,The WorldBank
Ann-MarieMeulendyke
Jean AndresFontaine FederalReserveBankof NewYork
Consultant,formerlyBancoCentral de Chile
CarlosAlbertoQueiroz
Peter Garber BancoCentraldo Brasil
BrownUniversity
DouglasScott
StevenGrenville Central Banking Department, International Monetary
ReserveBankof Australia Fund

Charles Goodhart RezaVaez-Zadeh

London Schoolof Economics Central Banking Department, International Monetary
Fund
Patrick Honohan
Economicand SocialResearchInstitute (Dublin),for- Lin See Yan
merly Financial Policy and Systems Division, The BankNegaraMalaysia
WorldBank

H. BarryJohnston
Central Banking Department, International Monetary
Fund

ix
Participants Sikander Rahim
S. Ramachandran
The WorldBank RobertoRocha
Andre Ryba
Jean-JacquesDeschamps SilviaSagari
AlexanderFleming Fernando Saldanha
TomGlaessner KhalidSiraj
Kristin Hallberg AlfredoThorne
MagdiIskander John Todd
RoyA. Karaoglan MoinaVarkie
AshokKhanna MichelWormser
MichaelKlein
KatherineMarshall IMF
HerminiaMartinez
BarbaraMierau-Klein IldefonsoGuajardo
FernandoMontesNegret Jian-Hai Lin
Paul Murgatroyd MelhemMelhem
AlbertoMusalem BobM. Traa
Joseph Pernia TarisaWatanagase
Paul Popiel LorenaZamalloa

x
 The Use of Market Instruments for Monetary Policy
GerardCaprio,Jr.and Patrick Honohan
For a century or more it has been acceptedthat inappro-
priate decisionsas to the price and availabilityof credit,
or as to the expansionof monetaryaggregates,couldhave
damaging effects on price and output stability in the
economyas a whole. Just how such monetary policyac-
tions have their effectand what is the best way to imple-
ment policyhas, however,remainedcontroversial.In the
last few years, there have been considerable changes in
the way in whichmany industrial countries approachthe
formulation of monetary policy. The changes have ac-
companied rapid developmentin the sophisticationand
depth of financialmarketsand havebeen both a response
to this developmentand a catalyst for it. In developing
countries, both the evolution of financial markets and
growing disenchantment with directed credit programs
and with bank-by-bankcredit ceilings have led to in-
creasedinterest in at least examiningand possiblymov-
ing to indirect methods of implementing monetary
policy.
Asa result of the WorldBank'sincreasedinvolvement
in financial sector development issues, especially
through the growing number of structural and sectoral
adjustment loans with a financialcomponent,it has been
asked to provide assistancein this area. Giventhat the
mechanisms by which monetary policy is implemented
can have important implicationsfor the long- term de-
velopment of the financial sector (as explainedbelow),
the WorldBank has respondedto these requests in con-
junction with the International MonetaryFund (IMF),
whose interest in assuring effectivetools for monetary
policyis clear.Indeed,the importanceof this topic led to
a sessionat the conferenceon collaborationbetweenthe
WorldBank and the IMF.
This paper providesan overviewof the policyissues in
this area that face developingcountries in the light of in-
dustrial country experienceduring the last couple of de-
cades. It drawson discussionsat the seminarand, while
not pretending to be a thorough summary, captures
1
some ofthe key issuesraised.The next section discusses
the objectivesof monetary policy and how these have
evolvedin recent years. There followsan account of the
differentpolicy instruments availableto the monetary
authorities. How these instruments have been used to
copewith the main shocks affectingmonetary policy-
those relatedto government deficitfinancingand to ex-
ternal flows-are the subject ofthe penultimatesection.
Thepaper endswith a briefoverviewofthe main messag-
es of the papersand discussionsreported in this volume.
The Objectivesof MonetaryPolicy
Some basic relationships are fairly reliably estab-
lished for many countries and can be taken as a basis for
a more completeunderstandingof howthe financialsec-
tor works in a market-orientedeconomy.For example,
overa protractedperiod,rapid growth in the moneysup-
plywilllead to sustainedinflation;an increasein the cost
of credit or a reduction in its availabilitywill dampen
economicactivity' and will also tend to slowthe under-
lying rate of inflationdespite representingan additional
cost to industry.
On the other hand, there is not a clear one-for-onere-
lationshipbetweenmonetarygrowthand inflationin the
short run. Domesticprices can be sticky;exchangerates
can overshoot their equilibrium levels followinga dis-
turbance and mayevenveerawayfrom equilibriumfora
while. Innovationsin the financial sector may alter the
equilibriumrelation betweenmoney,prices, and output
in ways that are hard to predict and may lead to an in-
crease in the volatility of this relationship (Goodhart
1989;Lindseyand Wallich1989).
Acceleratinginflationin many industrial countries in
the 1960sand 1970scaused authorities to reviewtheir
approach to monetary policy.It became widelybelieved
that the common use of interest rates as operating tar-
gets for monetarypolicyhad contributedto inflation,as
2 Gerard Caprio, Jr., and Patrick Honohan
politicalpressures had combinedwith policyinertia to
slowthe responseof monetary policyto rising prices. In
the absence, at least from the early 1970s,of fixed ex-
change rate anchors, excessivemonetary expansionhad
been tolerated.To correct this state of affairs,severalin-
dustrial countries began to rely more heavily on the
quantity of money as an indicator of monetary condi-
tions or even as an intermediate target. It was felt that
the moneysupplywouldproveto be a good leadingindi-
cator of inflationaryconditions;keepingit under control
would stabilizeinflation.Thedegree to which the money
supply was targeted, to the exclusionof other factors,
varied from country to country.In particular,a degreeof
interest rate stabilitywasalso maintainedin most coun-
tries. However,for some two and a half years (1979 to
1982)the U.S.authorities allowedinterest rates to move
rather widely,in an attempt to keep on course for their
monetarytargets.
Just as the monetaryaggregatesbecamemore closely
targeted, they began to perform less reliablyas a predic-
tor of inflationaryconditionsor of the state of aggregate
demand, as noted by Goodhart (1989) and others. In
some part, this contrary behaviorreflectedliberalization
of financial markets, which had been occurring around
the same time (Broaddus 1985;
195; Judd
Jdd and
the ametime(Braddu nd Scadding
Saddng 1982;
182,
de Vries1986).To some extent, it also occurred because
near substitutes for the targeted aggregateswere uti-
Iizedby the markets, once the authorities attempted to
exert contractionarypressure. For example,in the Unit-
ed States, controls on
n M1 helped spur
ed Sttes,contols
MIhelpe spu the growth of
thegrowh of
money market funds and led to a varietyof transactions
serviceslinked to componentsof the broader moneyag-
gregates.
Bythe end of the 1980s,most governmentshad for-
mally or informallyabandonedthe narrow focuson tar-
geting monetary aggregatesin favorof a more eclectic
approach that allowedthem to include a number of dif-
ferent indicators of the state of the economyas a guide
to policy.They had alsoinstituted a more flexibleregime
of monetary instruments, which allowedthem to influ-
ence monetary conditionsmore quicklythan in the past
and with a graduated pressure. At the same time many
governments began to allow a much greater degree of
competition in the financialsystem;the use of more in-
direct means of monetary control in this more competi-
tive environment helpedto ensure that monetarypolicy
measureswere not as easilyevadedbydisintermediation
as they had been in the 1970s.
Analternative approachto the focus on monetary ag-
gregates as the main intermediateobjectiveof monetary
policy has been the use of ceilingson aggregate credit
expansion.The ceilingswere usuallyensured by distrib-
uting sub-ceilings on a bank-by-bankbasis. This ap-
proach was widely used in European countries in the
1970s and still forms the basis of monetary policy in
many developingcountries (especiallyin Africa).Howev-
er, the definition of the institutions to be included in
such credit ceilings gaverise to many opportunities for
avoidancethrough disintermediationand the develop-
ment of near-bankand parallel credit markets, which
greatly reduced the effectivenessof these instruments.
Bank-by-bankceilingsalso distorted competitionby pe-
nalizing more dynamicinstitutions and discouragedre-
source mobilization; once a bank reaches its credit
ceiling,it has no incentiveto competefor additionalre-
sources, regardless of the profitabilityof its clients' in-
vestment opportunities (Johnston and Brekk 1989;
Cottarelli et al. 1986). Varioussuggestions have been
made for attempting to restore competition by permit-
ting the trading of credit quotasbetweenbanks or by es-
tablishing an automatic link between each bank's
current deposit mobilizationand the allocation to it of
future credit ceilings. However,these theoretical sug-
gestionshavenever beensystematicallyapplied,as many
governmentspreferred to replace the bank-by-bankap-
proach altogether.
proch atether.
Directed
tangled withcredit programs haveof
the implementation regularly
monetary policy.en-
become In
ticuledw iethe developingcontry pol
particular,a variety of developingcountries and several
industrial countries havehad monetarypolicysubverted
byattempts to exemptpriority-sectorcredit from overall
2Widespreadrecognitionofthe
or bank-by-bank
drawbacks ceilings.
of directedcredit schemeshas led to a decline
dr opular it willingto red these
in their popularity;oncewillingto reconsiderthese pro-
grams, authorities havebecomemore willingto examine
monetary policyinstruments as well.
Countriesthat move away from bank-by-bankcredit
ceilings to a more indirect means of monetary control
may find that they cannot achievepreset objectivesfor
moneyor credit aggregateswith the same degree of ac-
curacy.The classicexampleof this problem is the expe-
rience of the United Kingdomin the 1970s,which saw
considerableexpansionin monetary depth when credit
ceilings were removed. Neither credit nor money dis-
playeda stable relationshipwith nominal GNPor infla-
tion in those circumstances, and a single-minded
pursuit of a rigid target for the aggregateswould have
been costly (Goodhart1989).Developingcountries that
moveto indirect monetary control are also likelyto ex-
perience less certainty when using monetary aggregates
as intermediate objectives.It may be less difficult to
achieve intermediateobjectivesthat are elements of the
central bank balancesheet, but the reliabilityof these as
a meansof influencingaggregatedemandor inflationre-
mains relativelyunexploredfor developingcountries.

Instruments for Smoothing Bank Liquidity
Banks settle their debts with one another and meet
cash withdrawalseither by drawing on their credit ac-
counts at the central bank or by using currency in their
vaults and tills. Theseassets, representingbank liquidity
or the reserves of the banking system, are liabilitiesof
the central bank. In caseof need, banks can also sell liq-
uid assets or borrowfrom each other or from the central
bank. Overthe longer run, banks can repaytheir short-
term borrowingthrough the proceedsof maturing loans
to customers or by mobilizingadditional deposits.The
speed and easewith whichsuch actions can be taken de-
pends on the sophisticationof the banking system.
Whenthe banking systemas a wholeis short of liquid
funds,there is a generalizedupwardpressure on interest
rates and a tendencyfor bank credit to be expensiveand
scarce. Such pressures can be easedby central bank ac-
tion to provide liquidity to the system; an important
function of the central bank is to ensure that seasonal
and random influenceson liquidityconditionsare offset,
so that they do not result in correspondingvariationsin
interest rates. On the other hand, the central bank can
alsotake the initiativeto ease or tighten liquiditycondi-
tions dependingon how it perceivesa need to stimulate
or to restrain aggregateexpenditure(Binhadiand Meek
1989).
In the past, many central banks providedsemi-auto-
matic borrowing facilities to banks at posted interest
rates (knownas the discountrate, Bankrate, or Lombard
rate), which were varied infrequentlyand which effec-
tively placed a ceiling on short-term interest rates.
Where banks were customarilyborrowersfrom the cen-
tral bank, its postedrate also provideda floor for short-
term interest rates, as surplus banks could lendto deficit
banks at or near the postedrate. Becauseof the political
sensitivityof this key rate, upwardadjustmentswere of-
ten madetoo late and in steps that were too small.
Asa result, most central banks in industrial countries
have movedawayfrom automatic lending facilitiesand
nowmanage interest rates in a more flexiblemanner,us-
ing a variety of new instruments. Each country has
adapted its system of liquidity control in accordance
with localconditions,with the result that a great variety
of arrangements are in use, though each tends to satisfy
certain keyrequirements.First, there is no longeran au-
tomatic availabilityof borrowingfrom the central bank
at posted rates; accordingly,short-term interest rates
tend to be more flexiblethan in the past. Second,most
systemsstill have an upper and lower bufferto prevent
undue interest rate gyrationsin case the day-to-dayin-
struments for influencing interest rates are unable to
copewith a big surge in the demandfor liquidityor in
The Use of Market Instrumentsfor Monetary Policy 3
the availabilityof liquidfunds. Third, the fluctuationsof
interest rates provide much informationto the central
bank to help it gauge market conditions. 3 Fourth, for
open economiesoperating a peggedexchange rate sys-
tem, the regime often providesfor an automatic stabiliz-
ing response of short-term interest rates to foreign
exchangeflows.
Liquiditymanagement in most industrial countries
takes placetodayin a much more developedand compet-
itive money market than was the case only a fewyears
ago. Whileother policyand technologicalchanges have
alsoplayeda part in developingmoneymarkets, it can be
saidthat the more flexibletechniquesof monetary man-
agement have required and encouragedthese develop-
ments. The economyhas been well served by a deeper
moneymarket,which allowsgrowingnumbersof corpo-
rate borrowersdirectaccessto short-termfundswithout
havingto pay for bank intermediationcosts. The deeper
money market has alsoensured a more competitiveand
probably less costly system of bank intermediation.
Firmsthat are allowedto hold certificatesof depositsor
commercialpaper bearing market rates of interest will
not settle for below-marketrates on their deposits.De-
veloping countries can also benefit from such reforms
(and somealreadyare) as they,too, movetowardsrefined
techniques of monetary management. In particular,
deeper moneymarkets allow banks to economizeon li-
quidity holdings; their absence may partly explain the
predominanceof assertions that banks in the poorest
countries are exceedinglyliquid (Caprioand Honohan
1990).
The most striking illustration of howapparently dif-
ferent regimesof monetary control actuallyachievethe
same result is the relation of posted rates to market
rates. In some countries (for example,Franceand Cana-
da) the central bank's posted (discount)rate is normally
keptwell abovemarket rates. In others (for example,the
UnitedStates, Japan,and the Netherlands)it is normally
belowmarket rates (Kneeshawand Vanden Bergh1989).
The resolutionof this apparent paradoxlies in the differ-
ent manner in which lendingat the discountwindowis
practicedin the two groups of countries.Where the dis-
count rate is below the market rate, its availabilityto
borrowers is administrativelyrestricted. In these sys-
tems it is normal on any given day to see some banks
borrowingat this rate. But most avoidor limit the need
to borrowby holding a margin of excessreserves,even
though the opportunitycostof these reserves,represent-
ed by the interbank lending rate, is abovethe discount
rate. In this situation, the discount rate providesa firm
lowerbound to interest rates; it is alsoa somewhatflex-
ible upper bufferin that the banking system will be ad-
mitted to the central bank discount window if the
4 Gerard Caprio, Jr., and Patrick Honohan
interbank rate movestoo far abovethe discount rate. In
countries where the discount rate is kept well above
market rates (clearly indicating its penalty nature), it
providesan upper bound to interest rates, but is not
heavily used by banks because of the availability of
cheaper market alternatives.A central bank operatinga
higher-than-marketdiscount rate policy usually has a
wayofstepping in quicklyto providea depositfacility(or
to offer short-term bills on itself at a givenyield), in or-
der to placea floorbelowinterest rates if theyare quickly
fallingtoo far. Thus, when Germany moved its posted
central bank lendingrate from just belowto well above
market interest rates in the mid-1980s,it also began to
conduct the bulk of its money market interventions
through mechanisms that did not involvethe posted
lending rates. In particular, it began to use what are
known as "reversedtransactions" (explainedbelow)in-
stead of lending.
Some industrial countries havenot achievedthis de-
gree of flexibilityin their interest rate management.For
example,Belgianbanks hold large quantities of govern-
ment bills,whoseyieldrepresents the cost of short-term
funds in Belgiumand is essentially determined by the
authorities' funding decisions.In Sweden,the banks are
subject to an automatic graduated scale of central bank
lending rates and a central bank depositfacility.In con-
trast, the central bank lending rate in the United King-
dom is used more as a fine-tuning mechanism and is
managedon a day-to-daybasis.
Betweenthe floor and ceilingbuffersprovidedin the
manner just outlined, the central bank usuallyhas a va-
riety of other instruments to influence bank liquidity
conditions. Many of these involvequantity rather than
price (interest rate) decisionsby the authorities. For ex-
ample, it is normal for centralbanks to makeprojections
as to the liquidity needs of the system over a period of
one month or more and to makeinterventionsto add or
drain the amount of liquiditythat seems,on average,ap-
propriate for that period.Predictableshorter-term fluc-
tuations can similarly be met with instruments of the
appropriatematurity.It willbe desirableto dealwith un-
expectedpressures using differentinstruments tailored
to the particular circumstances.Use of one instrument
rather than another can come to signal the authorities'
intentions, so that action taken to offsetsome expected
fluctuation is not misunderstoodas a change of policy,
thereby avoiding unintended speculative pressures in
the market.
Amongthe instruments used are reversed transac-
tions4 in domestic securities and foreign exchange,
transfersof government deposits,and outright purchas-
es and sales of domesticsecuritiesand foreignexchange
(Kneeshawand van der Bergh 1989;Meulendyke1990).
Asregards the last category,outright open market oper-
ations in domestic securities are most appropriate to
meet trend needs for liquidity changes. However,they
are not used extensivelyoutsidethe UnitedStates,partly
becausemost other countries do not haveas rich a mar-
ket of first-rate short-term securities. Lowerquality se-
curities are unlikelyto be usefulas collateralbecauseof
their credit risk, while long-termsecurities are more ex-
posed to interest rate variations. Indeed, this is one of
the principaladvantagesof reversedoperations:by buy-
ing a government securityand contracting to resell it at
an agreed price, the authorities retain the initiative in
terms of amount,maturity,and timing, regardlessof the
terms of the underlying security. 5 Reversedoperations
can be implementedwithout much effecton the priceof
the underlyingsecurity,and they have most of the char-
acteristics of a secured loan without having to be made
at the postedrate. Theshortage of first-classcollateralin
developingcountries has made reversedtransactionses-
peciallyuseful;although the central bank can decide to
makeanyasset eligiblefor reposon a bilateral basiswith
selectedinstitutions, acceptanceby a wide range of mar-
ket participantswill be crucial to promoteactive trading
in reversedtransactions.
For speedyaction, central banks often turn to foreign
exchange(forex)swaps,which are reversedtransactions
in foreignexchangemarkets (Kubarych,1978).The size
of the foreign exchange market also means that large
volumesofliquiditycan be addedor drainedin this man-
ner. It is the main intervention technique used by the
Swissauthorities. However,this approach is not without
drawbacks.First, there are usually only a handful of
large participants in the foreign exchange market, so
forexswapsare not goodforprovidingor withdrawingli-
quidity on a broad base from the whole bankingsystem.
Second,provision of liquiditythrough forexswaps (i.e.,
buying foreignexchangespot with a contract to resell it
later) at a time of speculationagainst the domesticcur-
rency is very riskyfor the authorities, as they cannot be
sure there willbe no devaluation.Theauthorities in sev-
eral countries, industrial as wellas developing,havelost
large gamblesof this type.
Shifting government deposits between commercial
banks and the central bank is another tool bywhich the
authorities can influenceliquidity conditions, as noted
by CharlesFreedmanbelow.Arise in commercialbanks'
share of government deposits increases bank liquidity
just as do expansionaryopen market operations.Howev-
er, unless there are only a fewbanks (asin Canada,where
the instrument has been most used-see Shearer,
Chant,and Bond1984),this instrument suffersfrom the
selectivitymentionedfor forexswaps.There can be tech-
nical difficultiesas well: the procedureneeds to be gov-

erned by a framework agreement on the distribution
among banks, the remuneration of the deposits,and the
security that the government receives for its deposits.
Shifting government depositsneither requires nor fur-
thers the developmentof moneymarkets. This is at the
same time a practicaladvantageand a reason for not re-
lying on them indefinitely.Nevertheless,the manage-
ment of governmentdepositsneedsto be givenattention
in the context of monetarymanagement,as the distribu-
tion and size of these deposits will influence liquidity
conditions whether or not they are consciouslybeing
used as an instrument of policy.
The move from a system of monetary management
where the central bank providesliquidityon a bilateral
bank-by-bankbasis,as with the discountwindow,to one
where the focus is on managing the overall quantity of
liquidity,requires the developmentof markets in which
the banks and possiblyother participants can compete
for their liquidityneeds.This point is stressedespecially
by Paul Meekin the next paper.Once again, there is no
unique model for successfuldevelopmentof such mar-
kets; various institutional arrangements are possible.
However,participants in the wholesalemarket for very
short-term funds require trustworthy counterparties
and an instrument, or collateral,of the highest quality.
For these reasons, several countries conduct their li-
quidity management in a restricted market to which
only selectedinstitutions, for example,the banks or ap-
proved moneybrokers,are admitted.
Although there is alwaysthe risk that, by restricting
the number of participants, the authorities will not
achieve as competitive a market as might otherwise
emerge,a restricted market of this type can be a useful
firststep towardsthe evolutionof a widermoneymarket.
In Thailand,for instance, the commercialbanks make
bids and offersfor repurchase transactions in long-term
government paper. The transactions are for a few pre-
definedmaturities and the central bank,acting as a bro-
ker and market-maker receives the bids, arranges
matches, and may intervene on its own account. An al-
ternative approach in which the central bank also deals
with a restricted market is the treasury bill auction in
the Philippines. Fewer than twenty (nonbank) autho-
rized dealers participate, but the auction providesthe
main instrument of interest rate policy in that country,
as the authorities decidehowmany ofthe bidsto accept.
In some countries it is not possibleto identifya suffi-
cient number of appropriately credit-worthycounter-
parties, as where the bankingsystem has beenweakened
bywidespreadloan lossesor where the financialsector is
too small or concentrated. Elsewhere,there may be a
shortage of good qualitycollateral,as wherethe govern-
ment doesnot borrowdomesticallyor where there have
The Use of Market Instruments for Monetary Policy 5
been problems of government domestic payments ar-
rears. In these cases action to correct underlying defi-
ciencies, for example by rehabilitating the banking
system,encouraging new entrants, and clearing up do-
mestic paymentsarrears, is desirableanyway.It should
be tackled before attempting market-oriented reforms
that cannot be supported by the current state of the fi-
nancialsystem.In severaleconomies,such as Indonesia
and Taiwan,the central bank has issued its own short-
term paper to get around the problem of a deficiencyof
collateral;this solution is being activelyconsidered in
Botswana.
One advantage to using specialized money-market
brokersor market-makers,insteadof banks, is that they
have the incentive and a clear objectiveto widen the
number of counterpartieswith whom they deal, a point
emphasizedclearly by Meek.Even if it is possible for
these brokers to fund their lendingfrom banks, a wider
clientelewilltend to lowertheir fundingcosts.It alsoal-
lowsthem more flexibilityin managingtheir portfolioby
sellingpaper outright to nonbankclients. In seekingout
this clientele,they will broaden the scope of the money
market, providebetter cash management opportunities
for large companiesand other wealth-holders,and im-
prove the access of quality borrowers to short-term fi-
nance at the lowestpossibleinterest rates. On the other
hand, it may be more efficient, especiallyin smaller
countries, for the central bank to deal directly with
banks without going through the additionalintermedi-
ank withou going ou the additionalinterm -
aryter ofbrokers.Manyoftheoadvantagesofhavingbro-
kers maybe lost if they are wholly-ownedand controlled
by the banks.
If central bank managementof bank liquidityis to be
effectivein influencingmonetaryconditionsgenerally,it
is essentialthat banks do meet their cash obligations,in-
cluding the maintenance of any reserve requirements.
Unauthorizedoverdraftsthat arise from the clearing of
checks must be automaticallypenalizedby the central
bank with sufficient severity to make such overdrafts
quite exceptional.However,it is not strictly necessaryto
have formal reserverequirements to achieve monetary
control,as demonstratedby their absencein Canadaand
the United Kingdom.Banks will necessarilyhold some
reserveson a voluntarybasisto meet a bunchingof with-
drawalsor loan requests from important clients (Caprio
and Honohan 1990; Simpson 1987; Harrington 1987).
Nonetheless,many countries do have reserve require-
ments, and there is the view,noted by DavidLindsey,
that these requirements provide a fulcrum on which
monetary policy can operate with a more reliably pre-
dictable impact on monetary aggregates (Lindseyand
Wallich1989;Lindseypaper in this volume).
6 Gerard Caprio, Jr., and Patrick Honohan
An additional merit of reserve requirements is that
they can be adjusted to produce desired changes in li-
quidity. They can be raised in times of unexpectedly
strong capital inflowsto effecta broad-basedmopping-
up of excessbank liquidity,thereby sterilizing the do-
mestic impactof the inflows.Apartfrom such occasional
circumstances,however,reserverequirementsare often
consideredtoo blunt an instrument to be used for vary-
ing liquidityconditions. Even slight variations in them
can produce large changes in the amount of deposits
consistent with reserves.Furthermore, as money mar-
kets become more sophisticated,the scope widens for
avoidanceof reserverequirementsthrough the substitu-
tion of non-reservableinstruments. Indeed,this motive
was also prominent in the rise in severalcountries of
money market accounts. These accounts were popular
becausethey evadedreserve requirements and thus al-
lowedthe paymentof higher interest rates.
Reserverequirements were originallyintroduced in
many countries as a prudential measure to ensure that
banks would have sufficient liquidity on hand to meet
unexpecteddeposit withdrawals.For that reason, they
were expressedas a fraction of deposits.For monetary
control purposes,there is no reasonwhy they could not
be expressedas a proportion of credit,especiallyif credit
is considered a more relevant intermediate objectiveof
monetary policy.This has been done in various coun-
tries. It is also possibleto havea progressivescheduleof
reserve requirements-for example,through marginal
reserve requirements on extensionsof credit as overall
credit targets are approached-as a more flexiblealter-
native to bank-by-bankcredit ceilings. In most coun-
tries, where reserverequirementsare based on deposits,
the requirements are definedby referenceto a previous
level of deposits.This allowseasymonitoring, though it
may introduce some delay in the impact of monetary
policyon the aggregates.This "lagged-reserveaccount-
ing" also makes it easier for banks-especially those
with an extensivebranch network or a cumbersome re-
porting system-to know on a timely basis their re-
quired levelof reserves.Whateverthe base or purposeof
reserverequirements, the requirement to maintain the
reservesshould allow the flexibilityof permitting aver-
aging overthe maintenance period,to avoidunnecessary
Figure 1-1.SimplifiedCentralBank BalanceSheet
ASSETS
Net ForeignAssets
DomesticCredit
Claimson government,net
Claimson banks
Claimson nonfinancialfirms (if any)
day-to-dayfluctuations in liquidity conditions. Coun-
tries that have ignored this point haveexperiencedlarge
swings in interest rates as banks have had to scramble
continuallyfor reserves.
Theassets eligibleto satisfyreserverequirements are
usuallydefinedto include specifiedclassesof depositat
the central bank and possiblyspecial government bills.
Some countries also include vault cash. Most countries
do not include instruments that bear a full market rate
of interest, and therefore the requirement represents a
distorting tax on financial intermediation.Admittedly,
the central bank does providevaluable services free of
charge to the banking system,but these servicesare not
proportionalto the tax base.When inflationand nominal
market-clearing interest rates combine with high re-
serve requirements,the tax can be very severe.Accord-
ingly it is best to limit the size of reserve requirements
and to link the rate of remuneration on required re-
serves to the market rate (evenif for revenuepurposesit
is kept some distancebelowmarket).
Copingwith Shocks
For the most part, the assets that make up bank re-
serves-usually currency and deposits at the central
bank-are liabilitiesofthe central bank (seeFigure 1-1).
Non-bankholdingsof currency are a relativelypredict-
able quantity. (Exceptin centrally planned economies,
they alsoare alwaysprovidedfullyto meet demand;con-
trol over the moneysupplydoes not mean squeezingthe
availabilityof hand-to-hand currency.)To influencethe
quantity of bank liquidity,the central bank must there-
forehaveregardto the remaining elementsof its balance
sheet. Adjustments in these elements usually reflect,
through the balance sheet constraint, a simultaneous
adjustment in bank liquidity.The other main items are
central bank lending to banks, holdingsof government
obligations,and foreignexchange reserves.(For exam-
ple, when a bank purchases foreign exchange from the
central bank for a customer, the central bank's foreign
exchangeholdingsfall, as does the bank's depositat the
central bank.) Byremovingthe banks' automatic access
to borrowingfrom the central bank at or belowmarket
rates, it is possibleto bring the first of these items under
LIABILITIES
ReserveMoney
Currencyissued
Bank deposits
Other non-governmentdeposits(if any)
Capital,reservesetc.

control.The other two items, which maypresentgreater
difficulties,are treated belowin turn.
The Impact of GovernmentDeficitFinancing
Some countries haveestablisheda firm separationof
monetaryand fiscalpolicyin that they prohibitany lend-
ing from the centralbank to the government.Lessrigor-
ous rules in other countries establish quantitativelimits
on such lending.Where lack of fiscalrestraint presents
serious risks to monetary stability,such rules can have
merit, despitetheir arbitrary nature. In most countries,
however,the central bank has to copewith providingan
environment in which the financing of the government
deficitis achievedwhile maintaining as much monetary
and generaleconomicstabilityas possible.
In Italy,enormous governmentdeficitshave been fi-
nanced from domestic savings,while inflation has de-
clined. The central bank has withdrawn from direct
support of the governmentbond market.6 This has been
achieved by maintaining a level of real interest rates
which, though high, especiallyin recent years, can only
be describedas realisticin the context of the large defi-
cits (Padoa-Schioppa1987).As Cesare Caranzaexplains
below, the authorities have progressivelydismantled
regulations that had been designed to provide captive
markets for governmentpaper (such as obligatorymini-
mum holdings by contractual savings institutions and
controls on holdings of foreign currency assets). Now
holders of government paper do so voluntarily. They
continue to be willing to absorb more, whereas those
who were penalizedby the regulations of the past have
been slowto acquirevoluntaryholdings.The withdrawal
of the central bank has reduced the rate of creation of
bank liquidity and has slowed the growth in nominal
money.These changes haveallowedthe currency to re-
tain its position in the European MonetarySystem ex-
change rate arrangementsand have reduced the rate of
domesticprice inflation,
Where a government has not been able to survive
without extensivemonetary financing,it has provedim-
possible to restrain inflation. In Brazil, the extremely
high inflationof 1989left the central bank no option but
to confine its objectivesto maintaining a functioning
system of financial intermediation, again by ensuring
that interest rates were at a realisticlevel.Makingno at-
tempt (at the level of monetary policy)to restrain the
nominal expansionin monetaryaggregates,the author-
ities instead focusedon indicators of inflationaryexpec-
tations, including the parallel markets in foreign
exchangeand gold.Theyadjusted interest rates on a dai-
ly basis to achieve positive ex ante real interest rates.
This tactic allowed the financial sector to continue to
The Use of Market Instrumentsfor Monetary Policy 7
function for some time, despite the difficultiesof high
and unpredictableinflation (withmonthlyrates of infla-
tion rising as high as 80 percent or more).
In Chilealso-though inflationhas been much lower
in recent years-monetary policy is operated through
objectivesforthe real interest rate. In effect,this interest
rate is transmitted to the market through the central
bank's lendingwindowand its depositfacilityfor banks.
The preciseapproach is somewhatdifferent;it relies on
the daily monetary correction factor (about one-thirti-
eth of the previous month's calculated inflation rate),
whichis commonlyused as a basisfor indexedcontracts
in Chile. Indexationof monetary instruments remains
controversial,however.Manyfeel that, by loweringthe
perceivedcosts of inflation,indexationreduces the will-
ingness of politicalauthorities to take the necessaryfis-
cal steps to halt it. On the other hand, the extensionof
indexationto household mortgage debt (which has not
been successfullyimplementedin all of the countriesat-
tempting it) can create an important lobbyagainst infla-
tion.
External Flowsand MonetaryPolicy
Conceptually,a freely floatingexchangerate concep-
tually means that there are no flowsinto or out of the
central bank's foreign exchange reserves. In the polar
oppositecase of a rigidlyfixedexchangerate, the central
bank has no freedomto control its foreign exchangere-
serves,as it has agreed to acceptor provideforeigncur-
res, asaIt has funds on demand.Beten cur-
renraaest dometi fudsfon
d emand.
B etween
these
two extremesare a variety of differentapproachesto ex-
change rate management, which generate a variety of
implicationsfor the operation of monetary policy (Argy
1982).
If a country has adopteda free float, it cannot rely on
exchangerate policyto giveit a stable price level.Mone-
tary policy-particularly limiting the nominal expan-
sion of bank liquidity-will be the key to controlling
inflation. In these circumstances,a stable interest rate
maybe the enemy of price stability,particularlyif infla-
tionary expectations build up and result in more and
more borrowingfrom the central bank at the-now too
low-stable interest rate.
An active interest rate policy in a floating exchange
rate regime has a doubleeffecton demandconditionsin
the economy.By acting to raise interest rates, the au-
thorities increase costs and lowerdomesticdemand di-
rectly. The resultant capital inflowsserve to appreciate
the currency, thereby also reducing external demand.
Thus, the number of sectors exposedto the first-round
impactof monetary policyis increased,and the negative
8 Gerard Caprio, Jr., and Patrick Honohan
impact on any one sector, such as housing, may be di-
minished.
A fixed exchange rate, by contrast, should ensure a
good degree of price stability,depending,of course, on
what currency or basket has been adoptedfor the peg.
But the exchangerate is fixedonly as long as the author-
ities havesufficientreservesor foreignborrowingcapac-
ity to defendit. An inappropriatelyhigh levelof domestic
liquidity expansion (resulting, for example, from the
central bank financing much of a high budget deficit),
willresult in a steadydrain of reservesand, eventually,a
forcedabandonmentof the peg, perhaps precipitatedby
a well-judgedattack on the currency by speculators.
A fixed exchangerate can also be vulnerable even if
monetary policyis not systematicallymisalignedin this
way. For example, spontaneous short-term capital in-
flows,if not sterilizedby offsettingaction by the central
bank, may relax domestic credit conditions so far that
wagesand costs generallyget out of line. This weakens
the competitiveposition of the economyand results in
an unsustainabledrain on reservesover the longer run,
especiallyif the capital flowsare reversed.7 Prompt ac-
tion by the central bank to sterilize some of the inflow
may be necessaryto prevent this cycle,which has been
observedin severaldevelopingcountriesas wellas in the
smaller developedcountries. There can also be sponta-
neous speculativeattacks on a fixedexchangerate-al-
though the onus must be on the authorities to proveto
their ownsatisfactionthat the flowsare spontaneousand
do not reflect an underlying weaknessin the economy.
Fewcountries havesufficientresources to ride out such
unwarrantedspeculativeattacks without adjusting poli-
cy.The normal action is to allow the drain of domestic
liquidity,which such an attack produces, to raise inter-
est rates-dramatically if necessary.This can make un-
successfulspeculationso costlythat the attack is quickly
chokedoff.
However,the impact of the balance of payments on
monetaryconditionsneednot be destabilizing.Indeed it
has long been observedthat a fixedexchangerate coun-
try whose competitiveposition deteriorates will experi-
ence a current account deficit which, if not sterilized,
will tighten domestic liquidity conditions enough to
lower domestic demand, divert the production effort
into exports,and improvecompetitiveness.
ConcludingComments and Outline of the
Volume
The remainder of the volume is organized in four
parts. The first part, entitled "Nuts and Bolts," contains
contributions on the day-to-daypracticalitiesof operat-
ing monetary policy under a market-based system. A
mixture of theory and practical experiencefrom coun-
tries as different (albeit neighboring) as Indonesiaand
Australiaillustratesjust what is meant by market-based
monetary policy.The need for a paralleldevelopmentof
securitiesmarkets, as policyinstruments evolve,is high-
lighted, together with some practicalsuggestionsabout
howto encouragethis.
The second part, "MonetaryTargetingand Control,"
stands back from operational detailsto look at the mac-
roeconomicobjectivestowardswhichmonetary policyis
directed.The pronounced change in industrial country
attitudes to monetary targets is stressed,and the limita-
tions of the old approach for developingcountries are
noted.
Thetwo main factorstending to knock monetarypol-
icy offcourse are the exigenciesof governmentdeficitfi-
nance and the management of international balance of
payments flows and the exchange rate. Parts Three
("BudgetDeficitsand MonetaryPolicy")and Four ("The
Interaction of ExchangeRateand MonetaryPolicy")look
at these factors in turn, with casestudies from develop-
ing and industrial countries.
The papersand discussionsreported in this volume
contain manydetailedrecommendationsforthe applica-
tion of monetarypolicyinstruments to developingcoun-
tries. A number of key messages may be highlighted
here.
* Indirect methods of monetary control require a
certain degree of interest rate volatility. As
stressed by Paul Meek,the challenge for the au-
thorities is to act promptly and vigorouslyeven
whenunpalatableinterest rate increasesare need-
ed. Theymust also ensure that their institutional
set-up makesprovisionfor them to lowerinterest
rates, when that is appropriate.
* The successfulcentralbank will listen to the mar-
ket as well as guiding it. Freed from the con-
straints of quantitative controls, financial
marketswill often signal inflationarypressuresor
an impending slump, thereby assisting the for-
mulation of monetaryplans and policy. 8
* The stabilityof moneydemandrelationshipscan-
not be relied upon in developingcountries, partly
because there is not sufficient historical experi-
ence with liberal markets on which to base fore-
casts. David Lindsey, Charles Goodhart, and
StevenGrenvilleall noted the difficultieswith in-
termediate targets even in industrial countries
with more experienceand the result that mone-
tary targeting no longerholds the position it once
did.Accordingly,it maybe necessaryfor develop-
ing countries to concentratedirectly on targeting
foreign exchangereservesrather than on choos-

ing an intermediate target such as the quantity of
money.9Barry Johnston advocates this approach,
though he observes that a tight monetary policy
can result in private capital inflows rather than
reducing the current account deficit.
* As noted by Johnston, there may be one-off port-
folio adjustments following financial market lib-
eralization and the removal of quantitative
controls. The authorities must be prepared for
these adjustments, often involving a greater ex-
pansion of credit demand than of deposit supply.
The inflationary consequences may be arneliorat-
ed by appropriate interest rate levels or, depend-
ing on circumstances, capital inflows,
* Optimal functioning of indirect methods requires
the development of new markets; these in turn
enhance the efficiency of financial markets and
their ability to serve the economy, including its
investment needs. As argued by Meek, the objec-
tives of financial market development and mone-
tary policy thus go hand in hand. Cesare Caranza
shows how this complementarity became crucial
to the success of Italian policy in the 1980s. A re-
lated passage is Juan Andres Fontaine's concrete
account of how the development of debt-equity
swaps by the Chilean authorities contributed to
stabilization there.
* Also, it is useful to provide some hint of caution
about the risks for central bankers in being too
doctrinaire about the market-oriented policy in-
struments advocated in this volume. Several of
the papers that follow give illustrations of practice
deviating trom the pure market approach. With-
out necessarily endorsing every one of these, we
cannot disagree with the general point that emer-
gencies can arise that do require old-fashioned
non-market-clearing solutions."'
Finally, shifting away from direct means of control-
ling monetary policy is by no means universal in its ap-
peal. Direct controls are simple to operate. They seem to
offera sure handle on overall credit or money growth. As
noted by several observers, moving away from direct
controls often involves a fundamental reorientation of
central bankers and government officials, not only in re-
gard to directed credit but also concerning the financing
of government debt. However, monetary officials in a va-
riety of countries have found that there is no fool-proof
method to guarantee the achievement of any overall
monetary target. Bank-by-bank credit ceilings suffer
from the same limitation; eventually, nonbanks arise to
escape credit limits, and banks have every incentive to
evade controls. Moreover, such ceilings limit competi-
tion and, by choking off innovation and prompting ex-
FH,' I1L' 9 Maltket Iwirumlentl lo,, Monciarv Polit v 9
cessive holdings of liquidity, can curtail growth both in
the financial sector and in the rest of the economy. Al-
though not all countries are in a position to apply imme-
diately the experience already gained by industrial
countries in operating indirect methods of monetary
control, it can be expected that more and more monetary
authorities will soon begin to follow the lead of several
Asian countries, in particular.
Notes
1. TIhat is, for an economy not experiencing financial
repression. If, on the other hand, interest rates have
been repressed to artificially low rates, and credit ra-
tioned arbitrarily, an increase in interest rates may lead
to greater availability of credit and an improvement in
long-term growth prospects.
2. In France, a program dubbed the "encadrement du
credit' entailed a target for the net asset growth of each
financial institution, with penalties imposed that were a
(geometrically rising) function of the extent of the trans-
gression (Mourgues 1988 and Joint Economic Commit-
tee 1981.) tlntil 1979, however, credit for energy, exports
and "social" housing was exempt. As expected, credit to
these sectors boomed and led to a loss of monetary con-
trol. Coffee exporting countries in Central and West Af-
rica were among those who experienced similar
difficulties in the late 198)s.
3. For market movements to have meaningful infor-
mation content, the market must be fairly competitive.
If there is only a handful of institutions in the market,
they could collude to send misleading signals to the
monetary authorities.
4. These are often known as '"repos" or "swaps."A re-
versedltransaction involves the sale of a security and an
agreement to repurchase it later at a fixed price. It makes
liquid funds available to the seller for the agreed dura-
tion of the transaction. The relation between the current
price and the agreed future price establishes an implicit
interest rate for what, for many purposes, can be
thought of as a secured loan.
5. Thus the central bank can do a one-day or one-
week repo with a 3(0-yearbond as an underlying security.
In Tunisia, the government has in effect securitized part
of the banks' loan portfolio by allowing some loans, pre-
sumably to high-quality risks, to serve as the underlying
basket of securities for repos. This operation also could
be viewed as a rediscounting of a basket of loans, com-
pared with the usual rediscounting of specific, or indi-
vidual, loans.
6. Although the overall public sector borrowing re-
quirement has remained above 10 percent of GDP since
the niid-1970s, in recent years the non-interest compo-
10 GerardCaprio, Jr., and Patrick Honohan
nent has declined to 2-3 percent of GDP.Indeed, the
pressure of a rising interest bill, associatedwith the need
to finance government debt at market rates, has in-
creased the pressure on the government to reduce the
non-interest deficit.
7. Short-term capital inflowscan be spurred by a va-
riety of factors,such as a rise in the price ofa natural re-
source, as in the case of oil or natural gas exporting
countries, or a significant liberalizationof the domestic
economy,as in the SouthernConeeconomiesduringthe
1970s.
8. Steven Grenville (and indirectly, Paul Meek)de-
scribe how the central bank benefitsfrom allowingthe
market scopeto move interest rates. Discussingthe link
with exchange rate policy,Charles Freedmangoes fur-
ther in arguing that the authorities can do little to fight
against fundamentalmarket forces.
9. Authoritiesin large economiesshow signs of mov-
ing toward an eclectic approachwhere ultimate targets
of inflation and growth are emphasized.Officials in
smaller economiestend to focus more (or exclusively)
on the exchangerate or foreignexchangereserves.
10.Don Mathiesonmakes this point about responses
to financial system failures,while CarlosQueiroz illus-
trates the knife-edgealong which the Brazilian mone-
tary authorities have walked in a difficult policy
environment. Lin See Yandescribesthe (controversial)
use of targeted interest subsidiesin responseto a foreign
exchangecrisis.
References
Argy,Victor,1982."Exchange-RateManagementin The-
ory and Practice," Princeton University,Princeton
Studies in International Finance, No.50.
Bernanke, Ben, 1989. "MonetaryPolicy Transmission:
Through Moneyor Credit,"Federal ReserveBank of
PhiladelphiaBusinessReview,January/February.
Binhadiand Paul Meek,1989."ImplementingMonetary
Policy,"forthcoming in a volume by the Australian
NationalUniversity.
Broaddus, Alfred, 1985. "Financial Innovation in the
United States - Background, Current Status, and
Prospects," Federal Reserve Bank of Richmond
WorkingoPapera85-2.
Caprio,Gerardand PatrickHonohan,1990."Whatis Ex-
cess Liquidity,"processed,WorldBank.
Cottarelli,Carlo, GiampaoloGalli,PaoloMaruloReedtz
and Pittaluga, 1986. "MonetaryPolicythrough Ceil-
ings on Bank Lending,"EconomicPolicy,October.
de Vries,Rimmer,1986."GlobalCapitalMarkets:Issues
and Implications,"in WallenbergPapers, volume 1,
no. 4, October.
Goodhart,Charles, 1989."TheConductof MonetaryPol-
icy,"The EconomicJournal,June.
_ , 1984. Monetary Theory and Practice: The UK
experience,London:Macmillan.
Harrington, R. 1987. Asset Liability and Management,
Organizationfor Economic Cooperationand Devel-
opment, Paris,occasionalpaper.
Johnston,R. Barry,and OddPer Brekk, 1989."Monetary
Control Proceduresand FinancialReform:Approach-
es, Issues, and Recent Experiences in Developing
Countries,"International MonetaryFund, WP/89/48,
June.
Joint Economic Committee,1981.MonetaryPolicy,Se-
lectiveCreditPolicy,and IndustrialPolicyin France,
Britain, West Germany,and Sweden, Astaff studyfor
the Joint EconomicCommitteeof the Congressofthe
UnitedStates,June.
Judd, John P. and John L. Scadding,1982."The Search
for a Stable MoneyDemand Function," Journal of
EconomicLiterature, September.
Kneeshaw,John T.and P.Vanden Bergh, 1989."Changes
in Central BankMoneyMarketOperatingProcedures
in the 1980s,"BISEconomicPapers,No.23, January.
Kubarych,RogerM. 1978. Foreign Exchange Markets
in the United States, Federal Reserve Bank of New
York.
Lindsey,DavidE. and Henry C.Wallich,1989."Monetary
Policy,"in TheNew Palgrave,edited byJohn Eatwell,
MurrayMilgate,Peter Newman.NewYork,NY:W.W.
Norton.
Meulendyke,Ann-Marie,1990.U.S.MonetaryPolicyand
Financial Markets, Federal Reserve Bank of New
York.
Mourgues,Michellede, 1988. LaMonnaie:System Fin-
ancieret TheorieMonetaire.Paris: Economica.
Padoa-Schioppa,Tommaso,1987."ReshapingMonetary
Policy,"in Macroeconomicsand Finance:Essays in
Honor of Franco Modigliani, Rudiger Dornbusch,
Stanley Fischer,and John Bossons, eds. Cambridge,
Mass:MITPress.
Roubini,Nouriel, 1988."Offsetand Sterilization under
Fixed Exchange Rates with an Optimizing Central
Bank,"NBERWorkingPaper no. 2777,November.
Simpson,Thomas, 1987.Money,Banking, andEconom-
ic Analysis, third edition. Englewood Cliffs, N.J.:
Prentice-Hall.
Shearer,R.A.,J.F.Chant, and D.E.Bond, 1984.The Eco-
nomics ofrtheCaadian FinancialSystem, pp 444-52.
 PART I

Nuts and Bolts: PracticalIssues on

Moving to Indirect Implementation of
Monetary Policy
 Introduction
The move from direct to indirect monetary controls was
accomplishedfirst in industrial countries with smoothly
functioningmoneyand capital markets. Policymakers in
developingcountries are starting without these advan-
tages. This sectiondeals with the basic practicalrequire-
ments for moving to indirect controls. The key is
ensuring that the monetary authorities have the tools
necessaryto smooth bank liquidity.Central bank inter-
vention in the markets for treasury bills and similar in-
struments are usuallythe mediumthrough which policy
is effected.
Severalspeakersstressedthe fact that movingto indi-
rect monetary managementand the developmentof effi-
cient securities marketswere closelylinked. Neithercan
be wholly successfulwithout the other. In country after
country, this developmenthas required shifting govern-
ment bond sales from captivebanks to a wider market.
The process has been quite recent even in industrial
countries. (StevenGrenvilledescribesthe Australiancase
in this part, and the Italian and Chilean cases are de-
scribed in Part III.)
Though developmentof the money markets will not
occur overnight, Paul Meek and Steven Grenvilleboth
emphasizedthat the way monetary policyis conducted
can do much to either stimulate or retard their develop-
ment. Deepermoney markets will not only facilitate a
smooth implementationof monetary policybut also im-
provethe efficiencyof the financialsystem.' Issuing gov-
ernment paper too frequentlyor providingliberal access
to the discount windowstifles the developmentof inter-
bank lending.Similarly,if the authorities eliminatevari-
ability in interest rates, they will prevent the emergence
of brokers and dealers, who could facilitate the absorp-
tion of interest rate risk in the private sector. As Barry
Johnston also argued,all of this requires a significantre-
orientation in the thinking of the monetaryauthorities.2
Sufficientattention has to be paid to so-called"back-
officesupport" in the central bank, which in fact should
13
be viewedas "front-line"tasks,beforethe authorities can
rely on open-markettypeoperationsas a channel forim-
plementing policy.For example,policymakerswillben-
efit from being able to forecast seasonal patterns in
credit demand. When a central bank is able to forecast
flowsaffectingbank liquidity,it is in a better position to
achieve desired trend in interest rates. Meekremarks
that this is not strictly necessaryfor short-term liquidity
managementpurposes,and there are differencesof opin-
ion on how much effort should be devotedto the short-
term forecasts.However,for the medium-termformula-
tion of monetary policy objectives,forecastingof flows
does become a more central part of the whole exercise,
as pointed out by BarryJohnston. Centralbanks aiming
for greater reliance on indirect methods of implement-
ing monetarypolicyshould makean earlystart on build-
ing these skills.
The prosand cons of reserverequirements,including
precise methods of calculation and enforcement,were
debated.Althoughreserverequirementscan aid the cen-
tral bank in its effortto determinebanks' liquidityneeds,
they can easilybecomea tool for taxingthe financialsec-
tor and reducing the competitiveposition of the banks.
Reserverequirementsare not necessaryfor implement-
ing monetary policy-this point resurfacedduring the
presentationby DavidLindseyin Part II-but they can
helpstrengthen the central bank'shand in times ofmon-
etaryrestraint bymakingthe banks cometo it for liquid-
ity. On the other hand, reserverequirements haveoften
been used as a form of taxation, reducing the effective-
nessof the bankingsystem.Such distortionscan be min-
imized by remunerating banks for the reservesthey are
required to hold.
Another controversial issue was the proper role of
banks in developmentfinance.Mr.Johnston expounded
one view in this old debate, arguing that banks should
not providelong-term credit,whichwas more appropri-
ately the role of a capital market. Others have argued
14 Introduction to Part I
that one of the roles of banks is to engage in the term
transformationof risk, and that by pooling risks and ju-
diciouslymixingdebt maturity,banks can safelypartici-
pate in term lending. For instance, one leading
economic historian has attributed a slowdownin U.K.
growth during the latter part of the nineteenth century
to the Britishbanks' withdrawalfrom term lending; in
contrast, the increasedwillingnessof German banks to
make term loans is seen as having boosted industrial
growth there (Kennedy1987).
Dr. Grenville'spresentation drew on the recent Aus-
tralian experienceof a rapid transformationin the finan-
cial sector and in the conduct of monetary policy.He
underlined the importance of admitting non-financial
institutions into the money market as a means of in-
creasing competition, a potentially important point in
many small developingcountrieswhere there are only a
handful of banks. He also stressedthat the presenceof a
captive market for government securities can subvert
the developmentof a market. For instance,accounting
conventionscan inhibit institutions (reluctant to recog-
nize the capital loss)from selling,beforematurity, low-
interest bonds they have been forced to acquire. This
point is raised later in the contributionsby CesareCar-
anzaand Juan AndresFontaine.The simultaneousdevel-
opment of an active government securities market and
new monetary policy instruments, which are comple-
mentaryprocesses,requires close coordinationbetween
the monetaryauthorities and those chargedwith public
debt management. If debt management is allowedto
take precedence,this can slowthe effectivetransition to
market-basedmonetary control. (The Indonesian case,
discussedby Meek,providesan examplewhere market-
based methods were developedusing central bank and
private securities rather than those issued by the Gov-
ernment.)
The speakersin this session had considerablepracti-
cal experiencein money market affairs.Paul Meekwas
vice president and monetary adviserin the open market
area ofthe FederalReserveBankof NewYorkfrom 1973
to 1985.Since then he has servedas a consultant to cen-
tral banks, including those of Indonesia,Malaysia,Paki-
stan, China, Portugal and Singapore. Steven Grenville
has had a distinguishedcareer at the ReserveBank of
Australia,while Barry Johnston worked at the Bank of
England and the U.K.Treasurybefore joining the Cen-
tral BankingDepartmentof the International Monetary
Fund.
Notes
1. The greater the liquidityof the moneymarket, the
less needof banks to hold excessiveamounts of liquidity.
Grenvilleprovidesevidenceof a significantdeclineof ex-
cess liquidity in Australia,as the money market deep-
ened.
2. Speakers did not, however, advocate special tax
privilegesto promote money or capital market develop-
ment.
Reference
Kennedy,WilliamP., 1987.Industrial Structure, Capital
Markets, and the Origins of British Economic De-
cline, CambridgeUniversityPress, Cambridge, En-
gland.

CentralBank LiquidityManagementand the MoneyMarket
Paul Meek
Central banks havea major role in developingand imple-
menting monetary policy, conditioned by the political
and legal system within which they operate. They must
be concernedwith two aspects of monetary policy: eco-
nomic management and liquidity management. Eco-
nomic management involves promoting sustainable
economicgrowth over the longer term by keepingmon-
etary and credit expansion in step with an economy's
non-inflationaryoutput potential. It requires operating
within the constraints imposedby foreign exchangere-
serves, external debt, and the availability of foreign
grants and credits. Liquiditymanagement, in contrast,
has a shorter time horizon. It involvesassuring that the
banking system can provideflexiblyin the short-run for
the cash and paymentneeds of the society.
Central banks have other important functions that
bear on their monetary policy responsibilities.In most
countries they are required to assure the soundnessand
safety of the payments system, regulating and supervis-
ing the operations of the commercial banks, and fre-
quently other financialinstitutions as well.An adequate
supervisorysystem and an effectivelegal system may be
prerequisitesto the desirability,or evenfeasibility,of de-
veloping financial markets as a means of fostering eco-
nomic efficiency.
Assuming that the preconditions do exist, central
banks can help money markets developby the way they
carry out their liquidity management function. The
present paper takes up first the issues involvedin such
management. It then presents some general guidelines
fordevelopingan efficientmoneymarket. Finally,it gives
an in-depth look at Indonesia'sexperiencein recent years
in movingfrom a credit control systemof monetarycon-
trol to one involvingopen market operations in a fledg-
ling money market.
15
2
Central Bank MonetaryManagement
In developingcountries the monetary authorities
have tended to use dirigiste approaches to economic
management. They have often chosen to set interest
rates and exchangerates administrativelyand to impose
credit control on individualbanks as a means of influ-
encing the balance between the supplyand demand for
real resources. The credit control approach has been
used frequentlyin Asia,often in conjunctionwith a na-
tional planning exercise-in India, Indonesiaand Paki-
stan, for example. Oftentimes the authorities also
prescribe institutional holdingsof government and oth-
er securities and attempt to direct credit to favoredeco-
nomic sectors. Administrative control over credit
expansionoften workswell for a time in the formal fi-
nancialsystem,to moderateinflationarypressureswhile
fostering development. In time, however, it reduces
competitionwithin the banking system, favorsexisting
lines of business at the expenseof new enterprises,and
results in high intermediationspreads.Governmenten-
terprises and the governmentitselfare apt to take an in-
creasing share of credit flows, sapping growth. These
adverse consequenceshave sparkedwidespreadinterest
in liberalizingfinancialsystemsto spur competitionthat
will encouragesavingsand improveresourceallocation.
To central banks, the encouragement of money and
securities markets offers one means of enhancing effi-
ciencyby multiplyingthe channels through which sav-
ings and credit can flow. Such markets also enable
monetary policy to operate flexiblyto foster national
macroeconomicgoals, while leavingthe distribution of
domesticsavingsand investment to competitiveforces.
Theycan alsoassist the central governmentin financing
its owncash requirements in a non-inflationarymanner.
16 Paul Meet'k
Howevera financial system is organized,the central
bank has to findways of providingliquidityroutinelyso
that banks can meet varying short-term demands for
cash and credit without impairingthe reservesthey are
required to keepwith it. Banksdrawdowntheir working
balances at the central bank wheneverthere is a large
public demandfor currency or credit to pay taxes,meet
payrolls,or paybills. If the central bank does not offset
the resultant drain on bank reserves, the banks are
forcedto scramblefor funds, drivingup interest rates or
repatriating funds from abroad. Similarly,if it does not
mop up the surplus reservesproducedby a return flow
of currency after major holidays,short-term interest
rates will fall and there may be an outflow of funds
abroad.Thecentral bank can rely on its discountwindow
to handle the fluctuationsof bank reservesin relation to
requirements-setting terms on which the individual
banks can deal directly with it. Or it can initiate open
market operationsto offsetsuch changes, leaving it to
the interbank market to distribute the reserveswithin
the banking system. A sizable share of reservechanges
involve uncontrolled movements in other elements of
the central bank's own balance sheet. The treasury
makes large net disbursements from its accounts with
the central bank at some seasons, adding to reserves,
while at other times its balancesswellwith tax receipts,
havingthe oppositeeffect.Whenthe central bank inter-
venes in the foreignexchangemarket to maintain conti-
nuity, its purchases and sales impact bank reserve
positionsand the domesticmoney market. It must also
take account of swings in the net credit to banks that
may result from the central bank's arrangements for
check collectionand payment.It has to balance the col-
lectivereserve position of the banking system if it is to
keep domestic short-term interest rates reasonably
steadyfrom dayto day.
Oncea central bank movesfrom credit control by ad-
ministrativefiat to the use of financialmarkets, its eco-
nomic and liquidity management commitments are in
constant tension. The authorities cannot become ob-
sessedwith holding short-term rates steadyifthey are to
managethe economyproperly.Theymust be preparedto
change short-term interest rates promptly as their per-
ceptions of the economicoutlook change.Holdingsuch
rates steady for extended periods can easily result in
moneyand creditgrowth overa longertime horizon that
is either excessiveor deficient. Those central bankers
who haverelied on credit ceilingsas their principaltool
for economic management may find it hard at first to
change quicklyenough the terms on which they make
reservesavailable.Operating in financial markets with
flexibleinterest rates involvesa basic change in orienta-
tion. The improvementin savingand resourceallocation
that financialmarkets can bring providesstrong justifi-
cation formovingahead with their development.But the
credit demandsof the governmentand private business
can get out of hand unless the monetaryauthorities pos-
sess the authority and the will to make interest rate
changes that enforcemarket rationing.
Approachesto LiquidityManagement
Central banks in developingcountries usuallybegin
operatingin a financialenvironment dominatedby com-
mercialbanks.Theseexpandtheir branchesand services
as the economy's financial requirements increase with
development. In time, finance companies, merchant
banks, discounthouses, stock exchanges,and insurance
companiescan develop,dependingon the pace of busi-
ness developmentand the enablingactions of the fiscal
and monetary authorities. In this environment the cen-
tral bank usuallypermits individual commercialbanks
to replenishtheir reservesby borrowingfrom the central
bank, discountingapprovedpaperwith it at postedrates,
or selling it foreign exchange.Commercialbanks with
excessreservesmaybe permitted as well to buy foreign
exchange or government securities directly from the
central bank. The central bank itself respondspassively
to the demandsof the commercialbanks, balancingthe
system as a whole by a series of bilateral transactions
with individualinstitutions.
The interest and exchangerates at which the central
bank deals with the commercialbanks dependupon its
approach to economic management, including the ex-
tent to whichit permits domesticresidents to switchbe-
tween domesticand foreigncurrency. If credit controls
are the principleinstrument of monetary policyand ac-
cessto foreignexchangeis controlled,the rates at which
the authorities discount or sell paper bilaterallymay be
arbitrary and remain fixedfor long periods.In Pakistan,
for example,the State Bankdealswith excessliquidityby
selling 91-day treasury bills on tap at a 6 percent dis-
count rate. Then when a bank needs reserves,it redis-
counts the bills before maturity at 6.05 percent. The
State Bank's rates of issue and discount bear little rela-
tion to effectivereturns on bank depositsor loans,but its
bill operations keep the rates on daily interbank loans
steadyat around 6 to 6.5 percent. Such routine accom-
modation inhibits the developmentof liquiditymanage-
ment skills that are needed in a modern economy.
Whena centralbank phases out credit controls,it has
to adapt its liquiditymanagementtechniquesin order to
influence money and credit growth through interest
rates. Oftenthe firststep is to developan array of rates at
which it is preparedto make loans or to sell short-term
assets. It might, for example,establish lendingtranches
 Central Bank Liquidity Management and the Money Market 17

that involve a step-up in interest rates as individual spect to the calculationperiod.Commercialbanks make
banks increasetheir borrowing.The central bank could every effort to meet their requirements since deficien-
still rely on balancing the system in the short run cies usuallysubject bank managementto embarrassing
through bilateral transactions with individual banks. inquiriesfrom the central bank as wellas the assessment
For the longer term, it would haveto adjust the levelof ofa penalty.In a fewcountries,the monetary authorities
rates or the structure of its discount facilitywhenmoney operate with zero or minimal reserverequirements, ex-
and credit growth appearedto be departing significantly erting their leverage on the working balances of the
from a desiredpath. banks. In the UnitedKingdom,for example,the Bank of
Amore important steptoward market-orientedpolicy England has an understanding with the major banks
procedures takes place when the central bank assumes that their workingbalanceswith it will each day remain
responsibilityfor evening out swings in bank reserves at or abovea certain level.
relative to demand on its own initiative, rather than Theperiodsselectedforreservecalculationand main-
waitingpassivelyforindividualbanks to cometo it. Once tenance affect the manner in which both the central
it begins to supply or absorb liquiditythrough market bank and the commercialbanks go about their business.
intervention, the discount window plays an important, Use of a single day in calculating requirements-per-
but subordinate,safetyvalverole. Operationallythe cen- haps the end of a monthly or semi-monthly period-
tral bank can supplyreservesthrough the market when- givescommercialbanks an incentiveto minimize depos-
ever the short-term interbank rate rises abovea certain its on the day in question. The reserves required may
point and withdraw them wheneverthe rate fallsbelow change significantlyfrom one period to the next, since
some lower point. The distribution of reserves among depositson the basedate can vary a great deal depending
different banks is left to the interbank and other asset on the day ofthe week or time of the month. Usinga dai-
markets. The discount window serves as a backstop in ly averagecalculationfor the baseperiod has distinct ad-
case reservesremain short after the central bank oper- vantages. There is less scope for commercial bank
ates. churning to reduce requirementsthrough their ownac-
tion, when each day'sdepositsenter the reservecalcula-
The Banking System 's Demand for Reserves tion. The averaging process also tends to dampen swings
To maintain reasonably steady short-term interest .. in required reservelevelsbetweenperiods.
The length of the reservemaintenance period influ-
rates from dayto day, the central bank seeksto provide The le aden of reserve
sni avail-
bankswith sufficientreservesto meet any mandatoryre- ences the scaleand frequencyof swmgsIn reserveavail-
serve requirements plus whatever excess reserves or ability that the central bank seeks to offsetand hence,
working balances the paymentsystem needs. What this the behavior of the money market. If the commercial
implies for central bank operationsdependson the de banks must meet their requirementsof minimumwork-
positscommercialbanks maintain with it in practice, as ing balancesevery day, the central bank faces each day
determinedby their workingneedsas well as the regula- the formidabletask of countering the net effecton re-
tions governing reserve requirements and transfers of serves of all uncontrolled changes in its balance sheet.
funds and securities. The Bankof England,for example,must operate dailyin
the moneymarket to balance that day'sreservechanges,
Reserverequirements with end-of-dayloans to discount houses availableto
meet residual needs. The overnight interbank rate
In most countries, the central bank specifiesthe re- movesduring each dayas individualbank reserveposi-
serves that commercialbanks must hold, usually calcu- tions change.One cannot makeup tomorrowfor today's
lated as a fraction of selected categories of customer shortfall.
deposits with each bank. Some central banks use uni- Malaysia'sexperienceprovidesanother example.Until
form requirements across depositcategories to equalize recently Bank NegaraMalaysiaspecifiedthat required
the implicit tax involved,while others specify lower re- reserves be maintained on a semi-monthlybasis, but
quirements on time and savings deposits than on de- held them frozen throughout the maintenance period.
mand deposits.Theprocessof liquiditymanagementcan Accordingly,the separateworking balancesof the banks
be adjusted to whateversystem is in place. Typically,re- bore the full brunt daily of all changes in the central
quired reservesare calculatedon the basisof depositson bank's balance sheet. Since Bank Negara at the time
one day or on the averageof deposits during a specified madecredit availableat day'send onlyat a markup over
reservecalculationperiod.The banks must keep the re- the highest interbank rate of the day, the swings in the
serves during a reserve maintenance period, which is rate were quite large.Assoonas it allowedbanksto meet
usuallylaggedfrom a few days to severalweeks with re- their reserverequirements on a dailyaveragebasis and
18 Paul Meek
modifiedthe rate chargedon residualloans to the bank-
ing system, the scale of its daily market operations
dropped sharply and the daily interbank rate steadied
within a narrow range.
Manycentral banks havechosen to allowcommercial
banks to meet their requirements on a dailyaverageba-
sis over the maintenanceperiodas a wayof reducing the
scale of reserveimbalancesand of dampeningswingsin
the interbank rate. Some imposelimits on howlow end-
of-daybalancescan be in relation to requirements,to re-
duce the likelihoodthat the banks as a group will accu-
mulate large deficiencies during the period. Large
cumulative deficitsor surpluses can exert considerable
upwardor downwardpressure, respectively,on the daily
interbank rate when banks attempt to square their posi-
tions as the end of the period approaches.To moderate
such pressures, some central banks permit reserve ex-
cesses or deficits to be carried forward, within limits,
into the next reservemaintenance period, as is done in
the UnitedStates.
The length of the lag betweenthe reservecalculation
periodand the reservemaintenanceperiodaffectsthe re-
serve management problem facing both commercial
banks and the central bank. If both the central bank and
the commercialbanks havedeposit data in hand before
the reserve maintenance period begins, then both will
know in advancethe reserve targets at which they are
shooting. If the two periods overlap,both must operate
on the basis of projections,which are likelyto have siz-
able margins of error. It helps when both havea good fix
on requirementsin time to take correctiveactionbefore
the maintenanceperiodends.Withsuch informationthe
central bank can make reasonableestimatesof the total
reservesthat the banking systemwilldemandduringthe
maintenanceperiodto coverrequirements and allowfor
the excessreservesnecessaryto lubricate the payments
system.
Advanceknowledgemay be desirablefrom an opera-
tional standpoint,but it is not critical. The central bank
still has to estimate the uncontrolled factors affecting
the supplyof reserves.Variationsin currency in circula-
tion, treasury balancesat the central bank and other fac-
tors maywell be considerablylarger than the variations
in required reserves.The central bank will alwaysface a
margin of uncertainty about the demand for reserves
that will impactthe moneymarket.
The Demandfor Excess Reserves
Whateverthe levelof requirements,banks customar-
ily need a certain amount of excessreservesto keep the
system operatingsmoothly.In managingtheir positions,
banks have to copewith large in and out transfers initi-
ated by their customers.Not surprisingly,when the mu-
sic stops at the end of the day, there are usually
unanticipatedsurpluses that cannot be disposedof until
the nextday.Themore banks there are in the system,the
greater the volume of excessreserves that are likely to
accumulate during the reserveperiod.
In the UnitedStates, for example,many of the banks
and the other depositoryinstitutions that haveaccessto
the FederalReserve'sdiscount windowand wire transfer
facilities have very small reserve requirements. Their
need for working balanceswith the Federal Reserveex-
ceeds their requirements.As a result the aggregatevol-
ume of reserveson the central bank's booksat the end of
each day'stransfersof funds and securities is quite size-
able-almost US$1billion in 1989. In conducting its
open market operations, the Federal Reservemust add
an estimate for excessreservesto estimated required re-
servesin eachtwo weekreserveperiod,to producea pro-
jected demand for total reserves. If it wishes to raise
short-term interest rates for economicmanagementrea-
sons, the central bank does not fullymeet this demand,
forcing banks to increase their borrowingfrom it. If it
wishesto lowerrates, it can supplymore reservesthan it
expectswillbe demanded.
A banking system'sneed for excessreservesdepends
in part on whether the central bank's rules allow vault
cash to be counted toward the fulfillment of require-
ments. When vault cash can be counted, commercial
banks typically retain enough, distributed in their
branch banks, to meet local demands for cash while
seeking to minimize the expensivemovement of cash.
Some banks may then find that the residual require-
ments met by central bank balancesare less than they
need for working balances. The smaller these residual
balancesin relation to bank activity,the more likelythat
banks will hold excessdepositswith the central bank. On
the other hand, if vault cash cannot be counted toward
requirements, then commercialbanks will seek to min-
imize the cash held in their own vaults, withdrawing
cash as needed from the central bank and redepositing
excesscash with it. Both the central bank and the com-
mercialbanks willthen incur significantadditionalcosts
in transporting and handling the cash.
The central bank's own rules for permittingtransfers
of funds on its booksfrom one bank to another also in-
fluence the banking system's demand for excess re-
serves.Some centralbanks, Bank Indonesiafor one, will
honor transfer orders only if the transferring bank has
collected funds on deposit in its account. In conse-
quence, most Indonesian banks maintain deposits be-
yondtheir requirements in order to transfer funds early
in the day or to be able to meet adverse clearings.Bank
Indonesiain its open market operationshas to allow for

a sizable demand for excessreserves. Moreover,banks
typically pay a higher interbank rate in the morning
than after the clearing,usuallyone percentage point or
more.
In contrast, the FederalReserveSystemin the United
States required for many years only that banks havead-
equate balances at the end of the business day. To en-
couragean efficientfinancialsystem in an environment
with thousands of banks, it allowedbanks to transfer
funds or make paymentfor securities evenif such trans-
actionsresulted in temporary overdraftsin their Federal
Reserve accounts. One consequence was the develop-
ment of a very efficientinterbank (Federalfunds) mar-
ket. Individual banks could redress a loss of funds
experiencedin the normal courseof businessbyborrow-
ing funds overnightor for longerterms from banks with
surplus funds. Borrowingbanks in the major cities go
into overdraft automaticallyeach morning when they
wire funds to their lending banks to repay the previous
day's loans. They cover the overdraftwith incoming re-
ceipts or borrowingsthat are credited to their accounts
later in the day.This practiceof extendingintradaycred-
it to banks that are overdrawnexposesthe Federal Re-
serve to the risk that a bank fails before it restores its
balance at the central bank. Concernedabout its credit
exposure in a world financial system with enormous
transfers, the Federal Reservehas been developingnew
rules and proceduresfor monitoringand controlling its
exposurewithout adverselyimpacting the operation of
financial markets.
TheSupplyof Reserves
In managing liquidity,a central bank developsits own
proceduresfor supplyingor absorbingreserves.Asnoted
earlier, this may involvea passive response system in
which discount window rules allow depository institu-
tions to discount various kinds of paper or obtain ad-
vanceswhen confrontedwith a shortageof reserves.The
central bank mayalsostand readyto sell treasury or oth-
er paper at preset rates to bankswith excessfunds. Once
a central bank undertakes to manageliquidityactively,it
must itselfadd to, or reduce,reservesby buying or sell-
ing assets in the money market. The discount window
then becomes a safety valve rather than the principal
means of meeting liquidityneeds.
The Discount Window
The managementof liquiditytakes as giventhe other
credit activitiesof the central bank. Central banks often
serve as the channel for distributing subsidizedcredits
to favoredenterprisesor sectors of the economy.Or they
providetemporaryoverdraftfacilitiesto the central gov-
Central Bank Liquidity Management and the Money Market 19
ernment or take down a portion of new government
loans. These credit activities, which may loom quite
large in the central bank's balancesheet, are a source of
reserve changes, necessitating an equilibrating re-
sponse.The central bank must copewith changes in re-
serveavailabilityfrom whateversource, including those
that originatewithin the bank itself.
Many central banks have a formal discount window
that purports to be availableto commercialbanks that
experiencereserveshortages but which in fact goes un-
utilized because the discount rate is set well abovethe
cost of adjusting through other means. Banks in Paki-
stan can obtain neededreservesby discountingtreasury
bills at a rate wellbelowthe formal discount rate, while
the State Bank uses credit controls for economicman-
agement. In Indonesiaa few years ago, banks could sell
foreign exchange they held abroad to the central bank
for same-daycredit,whereasthe discount rate was usu-
ally three percentage points or so abovethe dailyinter-
bank rate. What matters is not the definition of the
formal discount windowbut the alternativesthe banks
face in making reserveadjustments on their own initia-
tive.
If central banks depend on influencingmoney and
credit growth through interest rates, they can still estab-
lish an array of rates at which they will deal with their
commercialbanks to adjust liquidity.Balancein the sys-
tem as a wholeis achievedthrough independentactions
of the banks themselvesin bilateraldeals with the cen-
tral bank.Rates in the interbank market for fundsreflect
the interaction of bank credit demandswith the central
bank's rate schedules.They,in turn, affectthe rates that
the banks quote on deposits and loans. Monetarycon-
trol's effectivenessdependson the readinessof the au-
thorities to move its administeredrates.
Thecritical step forwardin liquiditymanagementoc-
curs when the central bank undertakes to supplyor ab-
sorb reserves on its own initiative. It then addresses
reserveshortages or surpluses in the system as a whole,
leaving the distribution of reservesto the operation of
the moneymarket. Openmarket transactions are based
on competitivebids or offeringsby market participants,
not a preset central bank schedule,and the centralbank
determines the amount of securities purchased or sold.
The introduction of open market operations does not
eliminate the need for discount windowoperations. In-
dividualbanksstill needa lender of last resort when they
find themselvesshort at the end of the day,after all in-
terbanktransfershavetaken place.The discountwindow
providessuch a facilityto individualinstitutions but no
longer is the vehiclefor major adjustments in reserves.
The interaction betweenopen market operationsand
the discount window is central to economic manage-
20 Paul Meek
ment.Arrangementsdifferconsiderablyfrom country to
country. In the UnitedStates, with its thousands of de-
positoryinstitutions, accessto the windowis restrained
by internal rules that define acceptable borrowing in
terms of frequencyand institutional size. When open
market operations do not fully meet the banking sys-
tem's demandfor total reserves,the banks increasetheir
recourse to the discountwindow;the Federalfunds rate
also rises but remains moderate even though the dis-
count rate is belowthe Federalfunds rate. The spreadbe-
tween the two rates indicates the bite of policy
tightening. It increaseswhen the FederalReservesatis-
fiesreserveneeds more through rationedrecourseto the
discount window than through purchases of govern-
ment securities in the open market. The central bank
can probe toward monetary restraint or ease without
moving the discount rate. Changesin the discount rate
are used to reinforcechanges in market rates initiated
through open market operations, when the economic
situation callsforadvertisingpubliclythe shift ofpolicy.
A differentapproach is used by the Bank of Canada
whose banking system consists of a few banks with na-
tionwidebranching systems.The Bankof Canadasets its
discountrate each weekat a penalty markup overthe av-
erage issuing rate establishedon three-month treasury
bills in the weeklyauction. As long as the central bank
findsthe existinglevel of short-term market rates satis-
factory from an economic management standpoint, it
providesfully the reserve needs of the banking system,
so that there is no need for borrowingat the discount
window. However,if the economic situation calls for
higher rates, the Bank of Canadaholds backon supply-
ing reservesso that one or more banks are forcedto bor-
rowat the penalty rate. This generates upwardpressure
on all short-term rates. Once these have risen to the de-
sired level, the central bank resumes meeting the full
needs of the system so that there is no further need for
banks to borrow.
The Bank of Italy establishesa step function of dis-
count rates confronting its individualbanks. The more
often a bank borrowsfrom the central bank, the higher
the rate it pays. This approach involvesa measure of
market disciplineon the individualbank, rather than the
administrative guidance used by the U.S. Federal Re-
serve System.The central bank can initiate an increase
in interest rates by limiting the supplyof non-borrowed
reserves,forcing individual banks to borrow at higher
and higher rates. When rates haverisen to the appropri-
ate level,the Bankof Italycan resumesupplyingreserves
as neededand adjust its discountrate scheduleappropri-
ately.
OpenMarket Operationsand the Developmentof the
MoneyMarket
Centralbanks cannot rely on open market operations
foraeconomicpandliquidity managementrwithoutconsid-
erable developmentof the financial structure and mar-
kets. Yet central banks have a leading role to play in
these institutional developments.An excellentexample
ofthe steps that can be taken is providedby Malaysia.'At
an early stage after independencein 1957,the monetary
authorities encouragedthe formation of new national
commercialbanks and branch banking,which resulted
in a sharp declinein the dominant share previouslyheld
byforeignbanks. The firstdiscount house wasorganized
in the 1960s,and the number had grown to seven by
1986.The BankNegaraMalaysiaalso persuadedthe trea-
sury to begin issuing securities with different coupons
and maturities to meet the needs of a wide range of in-
vestors,includingthe EmployeesProvidentFund,which
took down 20-yearbonds.In 1973,the year in which the
Malaysianand Singapore currencies ceased to be inter-
changeable,the sale of treasury bills movedto an auc-
tion basisand the Malaysianringgit wasfloated.Bankers
acceptancesand negotiablecertificatesof depositwere
authorized in 1979. Finance companiesand merchant
bankswere establishedas well in the 1970s.
BankNegarafrom its beginningrequired the banks to
hold specifiedliquid assets in addition to required re-
serves, to assist in monetary control and to ensure that
the banks playeda direct role in financing the govern-
ment's developmentefforts. In 1986 the primary liquid
asset requirement stood at 10 percent of total deposits
and was met principallywith holdings of treasury bills
and depositswith the discount houses. Banks were re-
quired to hold an additional10 percent in secondaryas-
sets, consisting of commercialbills and other Malaysian
governmentsecurities.The issuanceof treasury billsand
other government securities was kept within the
amounts that banks and other financial institutions
were required to hold so that the interest rates on these
issues were usually below those that would have pre-
vailedin a free market. The discounthouses wereable to
paya higher rate on depositswith them than were avail-
able on treasury bills because they were permitted to
hold higher-yieldinggovernment securities with matu-
rities out to three years, which the banks could not
count as primary liquid assets. The interbank market
was quite activein maturities ranging from overnightto
one month, while transactionsin maturitiesout to three
months were common.
As noted earlier, Bank Negaramanaged liquidity by
posting rates at which it would buy bankersacceptances,
as wellas treasury securitiesmaturing out to 20 years.It
alsomight sell treasury bills,and on occasionother gov-

ernment securities, on the posted yield curve. At one
time it also placedgovernment depositswith the banks
for one month or longer in an effortto influencemarket
rates. The banking department participated activelyin
the foreign exchange market to maintain a degree of
continuity in the valueof the ringgit. It also occasionally
used swap transactions to relieve strains in the domestic
moneymarket and servedas lender of last resort to the
banks.
This passiveapproachsufficedto meet the system'sli-
quidity needs, stabilizingthe daily interbank rate once
bankswere allowedto meet their requirements on a dai-
ly average basis. However,the central bank's own will-
ingness to trade government securities bilaterally on a
posted yieldcurve left little opportunityfor a secondary
market to develop.The fact that government securities
carried below-marketyieldsmade them unattractiveto
anyonenot subjectto liquidasset requirements.The dis-
count houses consideredtheir function primarilyone of
maturitytransformation.Theyconcentratedon carrying
profitablya portfolio of treasury and short-term private
instruments with the inexpensivefunds generatedbythe
qualificationof their deposits as primary liquid assets.
They could, in fact, shift a part of their interest rate risk
to the central bank by selling to it from portfolio.The
housesdid playan active role in developingthe markets
for bankers acceptanceand certificatesof deposits,buy-
ing the securities and selling them under repurchase
agreements for shorter periods at rates that provideda
positiveinterest rate "carry."But the secondarymarkets
were thin even in these instruments.
In 1987BankNegarabegan the transition to a policy
of promotingan activesecondarymarket in government
securities and managingliquiditythrough market inter-
vention. Takingadvantage of a declinein interest rates,
the treasury increased its sales of treasury bills and
shorter maturities of Malaysiangovernment securities
to the point that these bore market yields.Bank Negara
began buyingand selling such issues only on its ownini-
tiative and at the rates at whichthey were trading in the
emergingsecondarymarket. In early 1989it designated
seven discount houses as exclusiveunderwriters of the
weeklytreasury bill auctions,with the requirement that
they make secondarymarketsto all comers. BankNega-
ra also reduced the primary liquid asset ratio to 5 per-
cent, to encourage them to depend more on market-
makingforprofitsand less on their privilegedpositionas
deposittakers. At the same time the Bankappointedthe
discount houses, seven merchant banks and four com-
mercial banks as primary dealers in Malaysiangovern-
ment securities. They were made responsible for
underwriting auctions of such issues and making sec-
ondary markets in them. Bank Negaraannounced that
Central Bank Liquidity Management and the Money Market 21
henceforth it would look to open market operations as
the primary means of managingdomesticliquidity.The
discount windowwas relegated to a subordinate role,
availableto individual institutions only under special
circumstances.
Forecasts and the Management of 7TeasuryBalances
A central bank embarkingon the use of open market
operationsto manage liquidityfindsit helpfulto develop
forecastsof the likelybehaviorof the major components
of its own balance sheet. In many countries, the inter-
locking patterns of treasury receipts and expenditures
subject bank reserves to sizable, and often erratic,
swings.The central bank also needs to compensatefor
predictableflowsof currencyinto and out of circulation
in connectionwith the country'spaymentpracticesand
seasonalpatterns. The central bank's ownnet extensions
of credit to state enterprises or favoredindustriesneedto
be included in the reserve outlook.Transactionsin for-
eign exchange are much less predictable,but their ef-
fectson bank reserves,as they occur, shouldbe available
to the managerof domesticopen market operations.Ma-
jor swingsstemming from changes in required reserves
and other factorsmaybe reasonablypredictable-for ex-
ample, the credit float stemming from the check collec-
tion process in the United States. Forecasts of the
prospectivemovements in reserves from such factors
can givethe central bank guidanceconcerningthe mag-
nitude and directionof its ownintervention.Whatis im-
portant to decision-makingis the disciplinethat such
forecastsimposerather than the size of forecasterrors.
A regular forecastingprocedure provides the manager
not only with information but also enableshim to ac-
quire a feel for the likelymargin of error in the forecasts
themselves.
Reserveforecastsare not, however,a prerequisitefor
effective liquidity management through open market
operations. The interbank market itself provides con-
tinuing information on the interplay between the de-
mand for reserves and their availabilityin the banking
system.Commercialbankers in charge of managingthe
reserve positions of their banks keep close watch on
where they stand in relation to their reserve require-
ments.Theybid for funds in the interbank market ifthey
are short and lend funds there ifa long positionis build-
ing up. In an active market, the interbank rate indicates
whether aggregatedemand is buildingup in relation to
supply. Since the object of liquidity management is to
maintain reasonablestabilityin the overnight interbank
rate, the central bank can intervenewhen the rate moves
aboveor belowthe range it desires.It can use repurchase
agreements to inject reserves on a temporary basis, to
22 Paul Meek
counter unusual strains, or the reverse of such opera-
tions if the money market becomes undesirablysloppy.
These short-term operations are an important supple-
ment to the outright purchases or sales of securities ad-
dressed to longer-lastingdevelopmentsin the demand
for total reserves.
The consciousshifting of treasury balancesbetween
the central bank and the commercialbanks can play a
useful role in reservemanagement outsidethe market.
In the United States, the treasury maintains balances
with most banking institutionsand makesdepositsin, or
calls on, such balancesas necessaryto evenout its own
balancesat a level agreed on with the Federal Reserve
System. This procedure helps cushion bank reserves
from the variabilityof tax collections,expenditures,and
securities sales. In Canada,the treasury has delegated
management of its balances to the Bank of Canada,
whichshifts balancesas its primarymeans of managing
liquidity.While this approach deals admirablywith li-
quidity in its system, it does bypassthe money market.
The central bank of a developingcountry can contribute
more effectivelyto financialmarket growth if it is itself
an activeuser of the market than if it developsa system
that bypassesthe market.
General Guidelinesfor MoneyMarket
Development
Nurturing an efficientmoneymarket is a many-sided
endeavor.Such a market requiresa legal,regulatory,and
supervisoryframeworkthat assures market participants
that contracts will be promptly fulfilled and that the
credit risksof buyingand sellingmarket instruments are
small and manageable.Officialoversight is needed to
keep tabs on the market's functioningand to discipline
those that engage in unsound business practices.There
must be negotiable instruments of high credit quality,
issuedunder lawsand rules that spellout the conditions
of issuance and the protections availableto the buyer.
The market requires financial intermediaries-either
dealersthat buy and sell from position at their ownrisk
or brokersthat charge a commissionto bring buyersand
sellerstogether.The central bank,moneymarket instru-
ments, and dealers all have essentialroles to play in de-
velopingan efficientmarket.
The CentralBank
The central bank is a natural leader of effortsto im-
prove a nation's financialsystem. It overseesand helps
organizethe paymentssystem neededto supplythe cash
and credit requirements of the economy.It supervises
the banks, and often other financialenterprises as well,
to maintain public confidencein the safety of deposits
and the integrity of financial transactions. The central
bank serves as fiscalagent for the ministryof finance-
banker to the government,servicingits depositaccounts
and issuingits securities.It is the operatingarm of mon-
etary policy in both economic and liquidity manage-
ment.
Thispaper has concentratedthus far on the contribu-
tion the central bank can make to market development
by using the markets to manageliquidityand the econ-
omy. But more than the central bank's own activityis
needed to energizea new market. Its transactionswillbe
only a small part of the total in a well-functioningmar-
ket. The central bank must take steps to enhance the
standing of the instruments availablein the market and
the dealerswho make the secondarymarket in them. A
good moneymarket rests on the participationof a wide
range of businessesand investorsseekingsafeoutlets for
their short-term funds. It is not enough for the central
bank to embrace rate flexibilityin its own operations if
the market remains limited to transactions between it
and the commercialbanks. The central bank must find
waysof encouragingdealersor brokers to seekout retail
customers in business and finance, so that the market
will developbreadth and depth.
Instruments
A moneymarket needs instruments to trade that are
of high quality,to attract the short-term funds of those
who prize safetyof principalmore than yield.Atreasury
bill or central bank certificatematuring in three months
or less provides an ideal instrument. Both are of the
highest quality, eliminating default risk except in the
most extraordinaryof circumstances.Either can anchor
the short-term market, providinga reference yield for
privatepaper.The central bankshould normallyconduct
the auctions of such paper as well as provide the book-
keepingand transfer facilitiesthat facilitate trading. It
can underscorethe qualityof the paperby discountingit
in its lending operations.When the central bank gradu-
ates to managing liquidity through the market rather
than through the discountwindow,it can take the lead
in buying and selling such open market paper.The cen-
tral bank must alsooverseethe credit qualityand market
practicesof the dealersor brokers that makethe second-
ary market.
In developinga treasury bill or central bank certifi-
cate market, it is important that the issues sold at auc-
tion mature at intervals of at least a week, as in the
United States. Dating 91-day paper to mature on each
business dayof the week as in the United Kingdomre-
duces the size of each maturity and inhibits trading in

the resulting 65 issues. Spacing issues at least a week
apart makesit easier to achievea tradablesize.It also en-
hances the role of dealers,since they willbe called on to
write a repurchase agreement to the specificdate their
customer needs cash or to repurchase an instrument
from him on that day. The length of the maturity auc-
tioned initiallycan reflect the existingstage of develop-
ment. In a rudimentary market, it can be as short as a
fewweeks.As the Indonesianexamplelater in the paper
makesclear, it is not evencritical that the amount to be
auctionedbe announcedin advance.Whatis necessaryis
that biddersbe free to bid at rates of their own choosing
and that the issuer accepts bids in sufficientvolume at
market rates to providea tradable supply.Onecan move
to standard-sizeauctions and extend the maturity to 13
weeks as the central bank, dealers, and other partici-
pants learn to use and trade the instrument effectively.
Central banks need also to work activelywith banks
and market makers in developingthe market for bank
certificates of deposit (CDs)and bankers acceptances.
Their supervision of the banks providesthe investing
public with assurance that such bank liabilitiesare of
high quality,eventhough not as risklessas treasury bills
or the central bank's own liabilities.The central bank
prescribesthe conditionsunder whichbanks can accept
privatedraftsto createacceptancesand deals itselfin the
market as a means of assuring itself that its regulations
are being followed.The central bank alsoallowsfinance
companiesand merchantbanks to issuetheir own short-
term paper.Here, too, it must providea degreeof super-
visionif the paper is to be acceptableto a wide range of
investors.
In time, oncepublic confidencein the moneymarket
builds up, it may become possible for businesses with
high credit standings to place their own paper with in-
vestors, either directly or through dealers. Such com-
mercial paper usually bears a higher interest rate than
bank-sponsoredpaper but still enablesthe issuerto raise
funds at a lower rate than through borrowing directly
from a bank or through bankers' acceptances.Use of
such an instrument dependson the develepmentof a so-
phisticatedgroup or investors,which are willingto bear
the risks ofbuying such paper to obtain the higher yields
they offer.
Brokersand Dealers
In building a good money market, the central bank
needs to encouragethe developmentof dealerswho will
make markets in short-term instruments. Such market
makingis essentialto enablebuyers of the paper to sell
it beforematurity,to raise cashwhenevertheyneed it. In
well-developedmarkets with many participants,brokers
Central Bank Liquidity Management and the Money Market 23
can performthe function of bringing buyers and sellers
together,charging one or both a commissionfor the ser-
vice as in the interbank and foreignexchangemarkets.
But developinga goodsecondarymarket for short-term
instruments usually requires an intermediarythat is a
dealer rather than a broker-one prepared to buy and
sell from position, embracing the risks and rewards of
ownership.Such a dealerhas a strong incentiveto build
customer participationin the market, for only that will
enable him to control his own risks.
A dealer in securities has three sources of earnings.
First, he can usuallyearn a positivereturn from financ-
inga positionin short-termsecurities dayto dayor week
to week at a lowerinterest rate than is being earned on
the securities themselves. Secondly,a dealer makes a
spread on the differencebetween his bid and askedpric-
es-the greater the activity,the more he makes.Finally,
the dealer can make a capital gain-or loss-from
changes in the valueof his position as interest rates rise
or fall.
In the early stages of developinga market, dealers
normallyfocuson the positive"carry"betweenthe rate
earned on securities held in position and the rate at
which they can be financed for shorter periods. The
spreadis often quite largebecauseof the premiumbanks
and others place on liquidityin the absenceof a func-
tioning secondarymarket. Dealerscan usuallyfinancea
position that is ten, twenty,or even fiftytimes their net
worth, since they need only supply a narrow margin of
collateral on such high quality,short-term paper. The
earnings potential on a leveragedposition is quite large
in a stable rate environment.The dealer is rewardedfor
taking the risk that interest rates will rise and that he
may have to renew his short-term financing at rates
higher than he is earning on his position.But the lever-
age of his positionexposesthe dealerto a substantial risk
that his entire equity can be wipedout bya rise in rates.
Accordingly,the dealerneeds to developcash customers
to whom he routinelysells moneymarket paperoutright
as an investmentof their liquidfunds.In case ofa rise in
rates, such customers experienceonly the opportunity
cost of foregoinghigher earnings for the remaining life
of the security.The dealer,however,wouldhaveto mark
down the value of his whole inventory and could be
forced to sell at a loss in order to maintain margins on
the securities pledgedto lenders.
Nurturing a group of dealers that make secondary
markets in short-term paper involvesmany challenges
to the centralbank. It must selectthe dealerswith whom
it willtrade, establishingcapitalrequirementsand devel-
oping a reporting system that permits it to assess the
risks that dealersare running in their day-to-dayopera-
tions. Thecentral bank willneed to see that the maturity
24 Paul Meek
of the paper it or the ministryoffinancesells to the deal-
ers doesnot overreachtheir abilityto managetheir ma-
turity risk. It should conduct its own operations on a
best-price basis that enforces competition among the
dealers.It mayprovidesome support to fledglingdealers
through special financing facilities, an underwriting
commission,or by givingthem a favoredrole in a set of
liquid asset requirements imposedon financialinstitu-
tions. Note, however,that such privilegescreate vested
interests and may be hard to withdrawas the market de-
velops.
There is alwaysan adversaryelement in the relation
between the central bank and the dealers.This flowsin
part from its supervisoryrole, but dealers are alsoapt to
want central bank protection from some of the business
risks that are properly theirs. They may tend to prefer
the passiverole of earning a positivecarry rather than
developinga customer market, looking to the central
bank to rescue them if it raises interest rates in pursuit
of its economicobjectives.The central bank has to prod
them into market-making by tying its recognition to
their outright transactions with customers. Only by
buildinga goodcustomer basecan dealersmanageinter-
est rate risk in bad times, in order to surviveand prosper
in good times. Whilesome specialaid may be necessary
in unusual circumstancesto buttress the market'sabili-
ty to continue making markets, the central bank must
avoid developinga market in which the "carry" profits
accrue to the dealers and the interest rate losses are
borne by the central bank.
The Case of Indonesia
Indonesiapresents an interesting exampleof a coun-
try that has recentlymovedfrom administrativecontrols
to flexiblemonetarymanagementthrough market oper-
ations.2 This shift has been part ofa broad governmental
effort to use market forcesto restructure the economy.
The authorities havesought to enhance financialcom-
petition and thereby promote a high level of domestic
savingsand a more market-orientedallocationofreal re-
sources.The motive force for this transformationeffort
came from the sharp fall of oil prices in the early 1980s.
At that time, the petroleum sector accountedfor 20 per-
cent of GNP,70 percent of exportsand two-thirdsof gov-
ernment revenues.The drop in oil prices cut deeplyinto
government receipts at a time when borrowingabroad
becamemore difficult,necessitatinga substantialcut in
development outlays. The government responded in
1983by devaluingthe rupiah by 38 percent against the
dollar to encouragenon-oil exports and by instituting a
newprogram of domestictaxation.Majorchangesin the
financialsystem were alsoset in motion.
MonetaryPolicy and the FinancialSystem
At the time of the 1983 reforms,Bank Indonesia re-
lied on a system of individualbank credit ceilingsas its
primary defense against excessive domestic money
growth.It alsoemployeda complexsystem of rediscount
credits to priority sectors (termed liquiditycredits) and
controlled the loan and deposit rates of the five state
banks, which accounted for three-quarters of commer-
cial bank deposits.Whenever reserve shortages devel-
oped,the central bank advancedliquiditycredits to the
banks.
Monetarypolicygoalswereset in the contextof an an-
nual government-wideplanning exerciselinkedto a five-
yearplan forthe economy,one in which state enterprises
played a major role. Bank Indonesia based allowable
credit growth for each fiscalyear on the money supply
growthbelievedconsistentwith anticipatedreal growth
and allowableinflation,after allowingfor the expected
behavior of the balance of payments.The central bank
workedout with the Ministryof Financecredit ceilings
for each of the large number of public enterprises. The
openness of the financial system, with capital move-
ments virtually free of controls and individualspermit-
ted to hold dollar balances domestically,counseled a
cautious approach to credit expansion.Fortunately,the
central bank enjoyedconsiderableautonomywithin gov-
ernment by virtue of the legal changes made in the
1960sto enableit to bring runawayinflationunder con-
trol.
The 1983reformsaimedat increasingthe flowofpub-
lic savings through the financial system and reducing
the extension of subsidizedcredit by the central bank.
The existing credit control system was abolished out-
right. Bank Indonesia was charged with developing
means of pursuing annual monetary growth targets
without interfering directly in the lendingand deposit-
taking activities of the commercial banks. The state
banks were freed from controls over their deposit and
lending rates, enablingthem to competemore effective-
ly with the private banks, which had grown rapidly in
previousyears by servingan active privatebusiness sec-
tor. Banks were also permitted to sell certificatesof de-
posit for the first time. However,the conceptof the state
banks as engines of developmentlingered on. They re-
tained their privilegedposition as channels for subsi-
dized liquidity credits and remained the only
institutions permittedto hold the depositsofstate enter-
prises.
In developingindirect means of monetary control,
Bank Indonesia confronted a money market that was
still in a formativestage. It includedan interbank mar-
ket and a small market for the paper issued by non-bank

financialinstitutions (NBFIs),which were authorizedto
engage in lending and underwriting operations. There
wasalsoa foreignexchangemarket, in whichBankIndo-
nesia balancedthe net demandor supplythat developed
in the bourse.Commercialbanksborrowedunsecured in
the interbank market, principallyfrom overnightto one
week,although there was some activityin maturities of
one month and longer.The state banks mainly supplied
funds while private domestic banks and foreign banks
were the chiefborrowers,reflectingin part restraints on
bank branching. All banks held substantial reserves
abroad and could convert them to rupiah on the same
day through bourse transactions with the central bank.
Developinga RevisedStrategy of MonetaryControl
After the reform, Bank Indonesia continued to pre-
pare annual objectivesfor monetary growth related to
the government'sfive-yearplan. It deviseda format for
tracking the monetarybaseweeklyin relation to a target
correspondingto the Bank's annual monetary growth
objectives.To foster indirect methods of monetary con-
trol, the central bank introduced in 1984a redesigned
discount window and its own central bank certificates
(SBIs-Sertifikat Bank Indonesia).It discontinuedpay-
ing interest to the banks on excessreserves,stimulating
activityin the interbankmarket. Byselling SBIs,matur-
ing in one and three months, Bank Indonesiaprovided
the banks with an investment outlet and could absorb
excessreserves. The SBIs were sold weeklyat auction,
but the central bank controlled, and seldom changed,
the rates it offered. Bank Indonesia named FICORIN-
VEST,an NBFI95 percentownedby it, to makea second-
ary market in the new instrument.
The central bank introducedtwo discount windows:
FacilityI to providecredit to meet liquidityneeds for up
to two weeks and Facility II to provide credit for two
months to encourage banks to make term loans. It set
the discount rates for use of the facilitieswell abovethe
rates on short-term interbank deposits to emphasize
that bankswere to raise funds from depositors,not from
the centralbank.Abank borrowingat either windowwas
not required to submit collateral other than its own
promissorynote.
These initial arrangements had one notable short-
coming.The central bank had no meansof addingto re-
serveson its own initiative.In practice, the banks proved
reluctant to use the discountwindowbecauseof its high
rates and also because of concern that such borrowing
would reflect adverselyon the bank's credit worthiness
and management. To meet liquidity needs, the banks,
which maintainedforeign exchangeabroad about equal
to Bank Indonesia'sown holdings, dependedon selling
Central Bank Liquidity Management and the Money Market 25
foreignexchangeto the central bank for same-daydeliv-
ery.A serious problem for both economicand liquidity
management arose in August-September1984 when
U.S. interest rates and the dollar's exchangevalue rose
sharply.Bank Indonesiaallowedthe rupiah to depreciate
at a somewhatfaster rate than previouslyagainstthe dol-
lar to spur the exportsector.This movetriggereda spec-
ulative outflowas banks and their customers sought to
hedge against the possibilityof another maxi-devalua-
tion. Respondingto the drain ofbank reserves,ratesrose
as high as 90 percent in the overnightinterbank market
as banks scrambled for funds. Bank Indonesiarelieved
the strain byopeninga specialcredit facilitythat enabled
the banks to borrowfor up to one year. This, together
with the return flowof fundspromptedbythe high rates,
restored liquidityand brought interest rates backdown.
Togiveit a tool for supplyingreservesin such circum-
stances, Bank Indonesiaestablishedregulationsin early
1985under which commercialbanks and finance com-
panies could issue trade bills and bankers' acceptances
(SBPUs-Surat BerhargaPasar Uang).The central bank
also empoweredFICORINVEST as its agent to discount
the new instrument at an announced scheduleof rates as
well as to discount outstanding SBIs,the central bank's
own certificates.Onlybank-endorsedSBPUswere made
eligiblefor discount at FICORINVEST or Bank Indone-
sia. BankIndonesiastood readyto discountpaper for Fl-
CORINVEST at an attractive rate to the extent it could
not financeits position elsewhere.The new facilitypro-
vided banks a means of obtaining reservesindirectlyat
lowerinterest rates than the officialdiscount rate.
Proceedingbytrial and error, BankIndonesialearned
to manage the liquidity of the banking system quite
smoothly,although it was somewhat handicapped by
lack of timely information on the current level of re-
serves and the reserve outlook. The commercialbanks
maintained reserverequirements, calculated on the ba-
sis of contemporaneousliabilities,during four periods
each month. Those with surplus reserves-ordinarily
the state banks-invested in SBIs in the weeklyauction
at rates set by the central bank. Banksthat found them-
selvesshort of reservescould discounttheir ownSBPUs
at FICORINVEST within the credit limits establishedby
Bank Indonesia.They could also discount SBIswith FI-
CORINVEST at more attractiverates than were available
at the central bank's discount window.Bank Indonesia
thus made a two-way market through FICORINVEST
that kept the overnightinterbank rate from moving too
widely.The banks avoidedthe discountwindow,prefer-
ring the cheaperalternativesavailableat FICORINVEST.
The market for SBIsand SBPUsbeganto develop.Fl-
CORINVEST wasable to financea portionof its portfolio
of SBIsand SBPUsthrough repurchaseagreementsaway
26 Paul Meek
from the central bank.It alsoresoldthe two instruments
on an outright basis to a limited extent. The two-way
market madeby FICORINVEST alsofacilitatedthe trans- ed, FICORINVEST's
fer of funds within the banking system.The state banks
were the principalbuyers of SBIs,while the privateand
foreign banks relied on discounting their own SBPUs
regularlyto meet part of their financingrequirements.
Problemsof Economicand LiquidityManagement
In pursuing its monetarygrowth objectives,BankIn-
donesia confrontedthe openness of its financialsystem
and the modest size of its exchange reserves.In 1985,
against the backgroundof declining U.S. interest rates
and a weakeningdollar,it was able to loweropen market
rates and its discount rate by about 3 percentage points
in order to stimulate domesticeconomicactivity.It was
also able to allow the rupiah to depreciateagainst the
dollar at a modestpace,to the same end. But public sen-
sitivity to the possibilityof a further maxi-devaluation
remained a conditioning influence.Despitethe adverse
economiceffectsof a large further declinein oil prices,
the central bank could not reduce SBI and SBPUrates
further during the 18 months followingAugust 1985.
Unableto provide further stimulus through monetary
policy,the authorities devaluedthe rupiah by 44 percent
in September1986.
The movecaught nearly everyoneby surprise, exacer-
bating anxieties of businesses and individuals accus-
tomed to holding dollar,as well as rupiah, assets. While
the sizeof the move should havedefusedfearsof further
devaluation, it did not. Large outflows of foreign ex-
change developedin late 1986.The monetaryauthorities
chose to maintain the existing rate structure for SBIs
and SBPUsin support of domestic economic growth.
Therewas heavydiscountingofSBPUsat FICORINVEST, it set in the auction. BankIndonesiaitselffoundthe new
sizablediscountingof SBIs,and moderateuse of the Fa-
cility 1 discount window at its penalty rate-all to fi-
nance the purchase of foreign exchange from Bank
Indonesia.The reservestrains generatedby the outflow
resultedin a rise of about 3.5 percentagepoints in short- SBIsto absorb liquidityand occasionallybought SBPUs
term interbank rates, a modest counter to the outflows
themselves.
Althoughthis tempest subsided,a much more serious
storm began to build up in April 1987.Bank Indonesia
initiallytried to maintain its lowinterest rate policy,but able to extract a substantial spread over the rate they
this soon provedimpossibleas a full-blownexchangecri-
sis emergedin Mayand June. Sharp increasesin the SBI
and SBPUrates proved insufficientto stem the tide. In
late June the Governmentdrainedreserveson a massive
scale by ordering four state institutions to withdraw Rp
900 billionfrom time depositsat the state banks and to
purchaseSBIs.The central bankraised its basicdiscount
rate in four steps from 18.5percent in Aprilto 30 percent
in earlyJuly. It alsosharply reduced,and then eliminat-
committed credit lines for the pur-
chase of SBPUs.Theovernightinterbank rate, whichhad
been in a 13 to 14 percent range in April,rose to a 35 to
40 percent range in earlyJuly.
These drastic measures succeeded in defendingthe
rupiah's exchangerate and generating a large reflowof
funds in the next few months. Theyalso led Bank Indo-
nesia to rethink its approach to reserve management.
Rigidlyfixingshort-term interest rates had contributed
to the crisisby inhibiting change.The central bank con-
cluded that the large credit lines outstanding to the
banks through FICORINVEST's discount facilitiesposed
an unacceptablerisk offinancingspeculativeoutflows.It
decidedthat henceforthit would deal in SBIsand SBPUs
only on its owninitiative,makingits ownjudgementson
the scale of operations and the rates at which it would
deal.
The refluxof reservesled to excessliquiditythat pro-
vided a favorableenvironment for introducing sales of
seven-daySBIsin the market at auction nearlyeveryday.
By October 1987, the central bank had negotiated a
smooth transition of the overnight interbank rate back
to a range of 10.5percent to 12.25percent. Giventhe ex-
cess liquidity,there were fewoccasionson which the au-
thorities bought SBPUscompetitivelyin the market, so
the salesof SBIsbecamethe main instrument forliquid-
ity management.The promising SBPUmarket dried up
without the involvementof the central bank, reducing
the adjustment options availableto the private banks
that normallyhad loan demandsin excessof their depos-
it generating capacity.The auction technique did, how-
ever,enable the central bank to signal the direction in
which it wishedto nudge interest rates bythe cutoffrate
rate flexibilitya great improvement over administered
rates.
Usefulas the auctions were, a number of problems
arose as the central bank regularlyauctioned seven-day
under repurchase agreements. The seven-day instru-
ment was too short forany secondarymarket to develop.
The state banks, which were the chief repositoriesof li-
quidity,bought onlywhat they needed.In fact,theywere
were receiving in overnight market by virtue of their
market power. Bank Indonesia could not effectively
bring pressure on the banks through sales becausethey
could alwayslet their existingholdingsmature over the
next few dayswithout replacement.Accordingly,it had
difficultyin February-March1988 when it sought to
nudgerates higher to forestallexchangerate speculation

when the cabinet changed. The state banks simply did
not bid to take up additionalSBIs at the higher rate lev-
els desired.
FurtherDevelopments
The appointment of a new cabinet and central bank
governorin April 1988led to an intensivereviewof the
financialsystem.In Octoberthe governmentproclaimed
a comprehensiveset of measures to make the financial
system more competitiveand to give Bank Indonesia
scope for more effectiveopen market operations. They
authorized the establishment of new national banks,
joint ventures with foreign banks, and rural credit
banks, subject to minimum capital requirements. They
permitted non-bank state and local government enter-
prises to hold up to half of their funds with private na-
tional banks, allowingthese banks to compete for such
deposits for the first time. The regulations established
legal limits on the credits that banks could extend to a
singleborrower,but expandedthe abilityof banksto pro-
vide foreign exchange services without prior approval.
Non-bank financial institutions were also allowed to
open a branch officein each of sevenmajor cities.
Bank Indonesia reduced reserve requirements from
15 percent to 2 percent of banks' third party liabilities
and extended the requirements to NBFIs.This change
helpedoffseta flat tax of 15percent just imposedon the
interest that the banks paidon time depositsand certifi-
cates of deposit. The combined effectwas to shift to di-
rect taxation for the benefit of the governmentfrom the
indirect tax of reserverequirements, which accrued to
the central bank. The banks were required initially to
buy three-month and six-month SBIs equal to 80 per-
cent of the cut in their requirements.
Bank Indonesia,operatingunder the authority of the
reform package,took steps to developfurther its open
market operations and encourage the growth of the
moneymarket. It introduceda two-daylag in the settle-
ment of foreign exchange transactions, cutting off the
accessto reserveson a same-daybasis.In January 1989,
it appointeda group of 15privatebanks and NBFIsto act
as underwriters of the SBI auctions and as market mak-
ers in the secondarymarket. The central bank required
all bids to be submitted through the market makers,pay-
ing them a small commissionon their successfulbids. It
also improvedthe mechanism for transferring owner-
ship of SBIs,so that the new dealers could re-sell them
to non-banksas well as banks, either outright or under
repurchase agreement.
The central bank began selling one-month SBIs in a
weeklyauction in January 1989,and introduceda three-
month maturity in May.At this stage it chose not to an-
Central Bank Liquidity Management and the Money Market 27
nounce in advancethe amount to be auctioned. This
proved wise because bidding by the state banks proved
erratic, depending on whether they were flush with
funds or not. When the Ministryof Financemade large
outpaymentsto government enterpriseaccounts,a state
bank receivingsuch funds would make large bids in the
auction at a narrow scale of rates, but once the funds
were spent, it might not bid at all.The 15dealerschosen
by the central bank were all small in size relativeto the
state banks, so they made only small bids in the auction
fortheir own account.Theyservedprimarilyas conduits
for the much larger bids of the state banks and were un-
able to make bids of any size in the secondarymarket.
The auctions themselvesproceeded quite smoothly
with Bank Indonesiavarying the amount sold in accor-
dance with the bids received.The private banks and the
NBFIssoon found that the new SBIs,and the compulso-
ry ones issuedthe previousOctober,couldbe sold under
short-term repurchaseagreements (repos)for a week or
longerat rates lowerthan those at which theycould bor-
rowunsecured in the interbank market. The dealersalso
financedprofitablytheir positions through short-term
repos. There was some selling by private banks to the
state banks of the longer SBIs issued after the October
package,but outright buying and selling of the newly
auctionedissueswas on a limited scale.
Bank Indonesiacontinued to operate almost dailyin
the market to maintain reasonablestability in the over-
night interbank rate. Each businessday,the manager of
operationswould consider the ease or tightness of the
money market in relation to his objectives.Takingac-
count of the limited informationavailableon the reserve
outlook, he would decidewhether to sell seven-daySBIs
the next day or make repos on SBPUs.If the central
bank's economicmanagementcalledfor loweringinter-
est rates, then he would not satisfythe demandfor long-
er-dated SBIs in the weeklyauction. The banks would
then have to disposeof their excessesin the interbank
market or by buying seven-day SBIs, which carried a
lower interest rate than the longer maturities. Coordi-
nating dailyoperationswith the weeklyauctions in this
way,in 1989Bank Indonesiawas able to maintain rea-
sonable stability in the overnight interbank rate in the
short run while loweringits levelprogressivelyover the
period.
Lessonsof the Experienceto Date
Bank Indonesiahas made notable progress in devel-
oping more flexibletechniques for managing liquidity
and pursuing its monetaryobjectives.The road has been
rather rocky,punctuated by periodicdisturbancesstem-
ming from devaluationfears and currency outflows.But
28 Paul Meek
the authorities have responded to each challengewith
institutional changes that have moved toward a more
market-oriented approach. They have begun to adjust
short-term interest rates flexiblyin accordancewith eco-
nomic and foreignexchange market developments.By
maintaininga steady,modestpace of depreciationof the
rupiah againstthe dollar,they seem alsoto havereduced
significantlyexpectationsof further maxi-devaluations.
This has permitted a reduction in the interest rate pre-
mium that previouslywas necessaryto keepfrom losing
exchangereserves.
The Indonesianexperienceindicatesthat it takestime
to move from a system of administered controls to a
more market-orientedone. Central bankers tend to see
credit controls as powerfultools yieldingpredictablere-
sults, and they enjoythe power such controls givethem.
Relinquishingcontrols means giving up the familiar-
setting limits on quantities and interest rates-for the
uncertainworld of influencingdepositgrowth bychang-
ing interest rates and affectingthe demand for deposits
and other financial assets. The administrativemindset
does not change overnight.Marketdevelopmentneeds
to proceedstep by step so that both the authorities and
financial institutions gain confidencein their ability to
function in the evolvingsystem.The central bank has to
keep close watch on developmentsyet avoid bailing out
poor managementdecisionsin the financialsector.
In Indonesiaboth the central bank and the financial
communityare still learning how to operate in a com-
petitivemarket environment. Bank Indonesiacan now
manage liquidity effectivelywith the present market.
But the secondarymarket for SBIsand SBPUslacks suf-
ficientcapitaland operationalexpertiseto allowthe cen-
tral bank to affect interest rates with the speed that
would be neededto copewith the kind of exchangeflows
experiencedin the past. There is more it can do to en-
couragea better secondarymarket.
Oneneed is to distinguishclearlythe weeklyauctions
of longer-datedSBIsfrom the dailysales of seven-dayor
shorter maturities.Todevelopthe secondarymarket it is
desirableto build up a regular and predictablesupplyof
three-month SBIs,leavingit to dealersto make second-
ary markets in the outstanding issues. Eliminatingthe
weeklyauction offour-weekSBIswould be helpful,since
selling that maturity now enables investors to obtain
them from the central bank without recourseto the sec-
ondary market. Pre-announcingthe amount to be sold
each week of the three-month issue would enable bid-
ders to know the total amount forwhich they must com-
pete.It should result in somevariationin the cutoffrate
from week to weekand also promote swappingof issues
alreadyoutstanding for the new issue.
Whereasthe weeklyauctions could be standardizedin
amount, BankIndonesia'sdailyoperationswill necessar-
ily vary in size, depending on the reserve situation.
There is no needto restrict the sales of short-datedSBIs
to a seven-daymaturity.There maybe timeswhen excess
liquidityis so great that BankIndonesiawould prefer to
sell SBIsmaturing in two or three daysin order to mop
up such liquiditywithout creating a reserveshortage in
the next reserve-averagingperiod.The central bank also
needs to use its regulationsand operationsto promote a
levelplayingfield for financial competition,countering
the tendencyof the state banks to exert market power.At
the same time, better coordination is needed with the
Ministryof Financeto developa mechanismfor reduc-
ing the size of the net flowsin and out of its central bank
account. BankIndonesia'sdailyoperationscouldthen be
keptwithin the developingcapacityof the market.
The banks and dealersstill relytoo much on auctions
conducted by Bank Indonesiarather than operating as
competitive,profit-orientedunits. Some big banks tend
to rely on the auctions to adjust their reservepositions
eventhough they could trade profitablyin the secondary
market. The dealers depend on Bank Indonesia's com-
missions and maturity transformation-financing long-
er paper with short-term funds-for their profitability.
Theyhavebeen content to earn the wide spreadbetween
the rates on longer-datedcentral bank paper and their fi-
nancingcosts without much concernfor the risk that in-
terest rates mayfluctuate.Theyare likelyto quit making
marketsat the first sign of trouble and turn to the cen-
tral bank for rescue.The dealersare not large enough to
accommodatethe potentialactivityof the state banks in
SBIs.But they could do much more than they havedone
thus far to make small markets and educate customers
about the gains to be earned by riding the yieldcurve.
These are not unusual teething problems for a new
market whoseparticipantsare learningto manageinter-
est rate risk. In the past, the dominant position of the
state banks has posed a problem becauseof the market
power they exert.Alreadythere are signs of keencompe-
tition in the market for the depositsof state enterprises
that may alleviatethis problem. The central bank itself
can reinforcemarket competitionthrough its ownoper-
ations. It must transact business only on a best-priceba-
sis and avoid supplying SBIs to state banks that have
missed in the auction. Bank Indonesia should require
dealersto sell SBIsoutright to non-bankcustomers and
stand ready to bid in the secondary market on a scale
commensurate with their capital position. It should
weed out those who only take the central bank's com-
missions for submitting the bids of others in the auc-
tions.

BankIndonesianeedsas well to developa strategyfor
improvingthe market in bank paper: the SBPU.Com-
pletelyphasingout centralbank supportfor that market
through FICORINVEST in 1987wasa mistake.TheBank
can help reinvigoratethe market by buying SBPUsperi-
odicallyfrom the dealers on an outright basis. Its SBI
marketing effort can also help if the volume of three-
month SBIsin bank portfolioscan be built up sufficient-
ly so that it alternates betweenpurchasesand sales in its
daily open-market operation. That would afford more
opportunitiesfor participatingin the SBPUmarket.
In the finalanalysis,it is the market participantswho
haveto developthe moneymarket. The central bank can
carry out its liquiditymanagement in waysthat use the
market. But the market'srationale has to rest on the ser-
vice it providesto financialinstitutions andthe economy
at large. The steepness of the yield curve suggests that
there are ample incentives for the money market to
makean importantcontributionto a more efficientallo-
Discussion
Question: If you were advisingthe government of a
developingcountry on setting up a market for short-
term government securities, what could influenceyour
choice as to whether it should be a market in treasury
bills or central bank bills?Does the fiscalposition of the
governmenthave anything to do with the choice?
Mr.Meek:I would think so, yes. In general, it is in the
central bank's interest to try to developmarket mecha-
nisms for raising treasury cash. Otherwise,the treasury
is knockingat the central bank's door.So, giventhe pol-
icy mandatesof the centralbank, it has a large self-inter-
est in trying to work with the treasury,to get the treasury
to issue securities as market instruments.A treasury bill
market is a goodplace to start.
Indonesiawasa specialcasebecauseit had no treasury
instruments. But I think you haveto look at each situa-
tion on its own.The important thing is to findsomething
that fitswith whateverstage of developmentexistsin the
country. For example,if a market has alreadystarted in
commercialbankers' acceptances,then the central bank
ought to participatein that market.
It may be of some interest that one of the purposesin
founding the U.S. Federal ReserveSystem-one of the
narrow self-interests of the commercialbanks-was to
authorize U.S.bankers' acceptances.They were tired of
financing all these things with bills drawn on London.
Theydid not havethe legalauthority to havebankers'ac-
Central Bank Liquidity Management and the Money Market 29
cation of financialresources.One should expectfurther
progress in Indonesia.
Notes
1. See Lin See Yan,"MonetaryControl and Financial
Reform:Malaysia'sExperience,"InternationalMonetary
Fund, Central Banking Seminar,November28 -Decem-
ber 9, 1988, and Bank Negara Malaysia,Money and
Banking in Malaysia,1989.
2. This section draws heavily on Binhadi and Paul
Meek,"Implementing MonetaryPolicy in Indonesia,"
November9, 1989,which is to appear in a forthcoming
volume publishedby the AustralianNationalUniversity.
See alsoV.Sundararajanand LazarosMolho,"Financial
Reformand MonetaryControl in Indonesia,"presented
at a conferencesponsored by the FederalReserveBank
of San Francisco,September23-25, 1987.
ceptancesin the UnitedStates. So, embeddedin the Fed-
eral ReserveActwas a commitment to try to developa
market for bankers' acceptances.The first governor of
the NewYork ReserveBank had been the president of
BankersTrust Company.One of the major activitiesof
the central bank in the 1920swas the effortto developa
NewYorkmarket for bankers'acceptances.This is a case
in point ofa centralbank setting out to developa market.
The treasury billmarket alsocame along in the 1920s,at
about the same time.
I do not know whythey chose to auction a 91-daybill
on one particular date and have it mature on a single
date. But that was highly beneficialto the development
of a treasury securities market becauseit gavedealers a
role in underwriting the issue and financing positions
with repurchaseagreements.I think that choicewas im-
portant in the market'sdevelopment.
In part, differencesbetweenthe U.S.and U.K.govern-
ment debt marketsspring from very differentdebt man-
agement philosophies. The tradition in the United
Kingdom has been to finance the government at the
long-term end of the market with coupon issues-
gilts-rather than with short-term instruments. The
United States pursued over the years-to its disadvan-
tage in some respects-a policyof debt marketing: "We
have somethingfor everybody.Wehave bill auctions ev-
ery week, and every month we have a two-yearissue, a
three-year issue. Wehavea five-yearissue,a ten-year is-
30 Paul Meek
sue, and so on." The U.S. Treasuryis unconcerned that
everythingis not financedon a long-term basis.And so
the U.S. marketable debt has a much shorter average
maturity than U.K.debt.
Mr. Grenville:I thought I would say a word for the
other side,just in caseanyonethings the U.K.systemhas
nothing at all in its favor.We havea system in Australia
like that in the UnitedKingdom.Thosewhowant to buy
treasury notes at tender can get them at any time in the
followingweek;the notes mature 90 days from date of
purchase.This approachhas the advantageof putting an
automatic shock absorberinto the system.It is left to the
market to work out on which days over the next week
there willbe excessliquidity.The market can then get rid
ofthis liquiditybytaking up treasury notes on those par-
ticular days. It is just one way of handing over to the
market a bit more of the decision-makingprocess and
smoothing out the liquiditychanges.Wedo havethat bit
of advantage.
Mr.Meek:Of course, that stems from the fact that the
Australiansystem, like that of the UnitedKingdom,re-
quires that the systembe balancedout at the end ofeach
day. If you have reserverequirements, averagingover a
period of time is obviouslymuch easier for all con-
cerned.
Question: You mentioned the benefits of having a
two-weekperiod for averaging reservesto meet reserve
requirements, rather than a one-weekperiod.Is that al-
waysthe case?Wouldit not dependon the stage offinan-
cialdevelopmentof that particular economy?
Mr.Meek:Oh sure, I think that can be tailored to the
individualcountry.Mysuggestionto Indonesia(ofa two-
week period)severalyears agowas made at a time when
they had contemporaneous reserve requirements. But
the actual state of reserveswas not known until many
weeks after the fact. So the central bank had no idea of
the banks' needs for reserves. The managers have no
guidanceas to the demandfor reservesto meet require-
ments.
In the caseof Indonesia,this was not a very important
or disturbing problem;they do not havethe kind of vol-
atilitythat occursin the U.S.banking system.It does not
really matter whether or not the reserverequirements
are contemporaneous. If you have lagged reserve re-
quirements, the important thing is that all the banks
know at the beginning of the reserve holding period
what their requirements are. And the central bank
knows their requirements. That knowledgeis of some
value.
Mr. Honohan: When you say it was useful to move
from a one-weekto a two-weekreservemaintenance pe-
riod, doyou mean that it helpedto lowerthe volatilityof
interest rates?
Mr.Meek:Byuseful,I meant the market movedinter-
est rates in an appropriate direction without central
bank intervention.In the U.S.system,for example,when
Wednesdaywas the settlement dayin a one-weekperiod
and Fridaycountedfor three daysof the seven,the room
to maneuveror bet on the future course of interest rates
was modest. The two-weekperiod gave the banks more
scopeforbetting on the future courseof interest rates. If
the FederalReservetrading deskhad a borrowingstarget
when a lot of bullish economicinformation came out,
desk managers would say the FederalReservewill have
to tighten up interest rates. So they would go long early
in the period to ensure their reserverequirements were
covered.The overall effectwasto firm up interest rates.
If there were a seriesof strong economicdata, the fed-
eral funds rate might firm up bya percentage point over
time, simplybecauseexpectationsabout the future had
changed. The Federal Reservewas in the enviableposi-
tion of being ableto sit back.Wewere not interveningon
the funds rate; we did not havenarrow limits on it. If the
economicdata continued to movethe same way,someof
the movetoward tighter moneywould alreadyhavetak-
en place.It helped overcomethe internal lag in the cen-
tral bank's decision-makingprocess.On the other hand,
if the economicdata moveddifferentlyand expectations
changed, the funds rate would come back down.There
wasa range of plus or minus one percentagepoint on the
funds rate that the market itselfwould movebyvirtue of
havinga two-weekperiod,whereasthe one-weekperiod
lackedthat.
Myobservationfrom one visit to the GermanBundes-
bank wasthat their one-month reserveperiodgavethem
even more of a market-ledeffecton interest rates.
Question:I want to followup on the last question.In
an inflationaryeconomy,wouldyou suggest introducing
a moneymarket at once or controlling inflationfirst be-
fore proceedingto the creation of the moneymarket?
Mr.Meek:I think that developinga money market, a
primaryand secondarymarket in treasury debt, is some-
thing that should be on the agenda of the central bank
all the time. Countrieshave had a long period of time,
twentyyearsor more, to developcredit controls and oth-
er means of financial control that we all now think are
not optimal.Overthe past ten years,the countries them-
selveshavelearnedthat this kind of system is inefficient,
leads to a misallocationof resources, and is not goodfor

long-term growth.There is a fashionin these things; ten
years agovery fewpersons had any interest in the finan-
cial markets in developingcountries. Now it is all the
rage.
Question:When credit controls are removed,some-
times the central bank tends to start changing the re-
serve requirements.What has been the result, especially
when reserverequirements were initiallyraised, of try-
ing to reduce them again?
Mr.Meek:I am sure other people in the room havea
better fix on that than I have. Myown personal view is
that it dependsverymuch on the developmentproblems
in the particular country. I agree with all of the argu-
ments for a "level playingfield"-low reserve require-
ments and such things-under idealcircumstances.But
in a country that has great difficultycollecting taxes,
then some level of reserve requirements doesn't seem
like a bad idea to me. Of course it cannot last for a long might introduce a liquidasset requirementstemporarily
periodof time.
To return to the exampleof Indonesia,they cut their
reserverequirements from 15 percent to 2 percent, but
then imposeda 15 percent tax on interest paid by com-
mercialbanks to depositors,which is reallyjust the same
thn. Th onl real difrec wa tha inta ,oh
thing.tral
banl realedifferene wars thatistrear gofthe worse than one country, which shall remain nameless,
central bank receiving the earnings, the treasury got
them.
I become a little concernedthat some advisers per-
haps over-emphasizethe importanceof the levelplaying
field,equalizingopportunities,and avoidingany incen-
tivesthat maylead onepart ofthe market to developfast-
er than other parts.When a developingcountry is still at
a fairlyearly stage of development,I think those may be
ideals and goals toward which the system should evolve
over time. But I do not know that they are the most
pressing immediateconcerns.
Question: Would you advocate varying reserve re-
quirements from time to time as a monetary instru-
ment?
Mr Meek:Well,that affectsthe wedgebetween the in-
terest rate paid on depositsand interest rates charged
borrowers.That may have some value. Toofrequently,
though, I think reserverequirements have been raised
and liquidasset requirementshavebeeninstalledprima-
rily to channel funds to the governmentfrom the private
sector. All of this redounds to the disadvantageof the
country's economic development. I'm not sure one
should encouragethe use of reserverequirements,but I
don't really havea position on it.
Central Bank Liquidity Management and the Money Market 31
Mr Honohan:If reserverequirementsare often used,
and are perceived,as a fiscal instrument and not as a
monetary instrument, would not that fact militate
against their introductionin a system where previously
there were none? Wouldit not be dangerousto blur the
distinction between fiscal objectives and fiscal instru-
ments on the one hand and monetary objectiveson the
other?
Mr. Meek:Yes,I think that risk exists. However,at
times circumstanceslead us to recommendsecond-best
or third-best measures.I had thought that I would never
advocateliquidasset requirements.Then I went to Chi-
na to advise them on debt management.The Chinese
had sold nine-year bonds, and then five-yearbonds,
three-year bonds, and two-yearbonds; 1990is the year
they all comedue. They haveno mechanismfor refund-
ing. And the banking system owns almost none of the
bonds. I suggested that as a temporizing measure they
to, in effect,providethe funds that they could use to pay
offthe existingbonds. I havebeen told they did not ac-
cept myadvice.Theyjust madethe state enterpriseshold
onto their bonds.
Mr.Honohan: I don't know whether that is better or
that organizes its bond sales by paying off all the old
bonds in Novemberand issuing new bonds in February.
Theyhaven't workedout how to bridgethe Novemberto
Februaryperiod.
Question:Do you havea rule of thumb for when the
central bank ought to exit from making the market? To
what indicators might it look when deciding to begin
getting out of setting prices?.
Mr.Meek:I think the whole movementtoward mar-
kets and toward abandoning credit ceilingsas a mone-
tary control is the basic issue. That issue need not be
finallydecidedwhen the market is started, but the cen-
tral bank must commit to moving in the direction of
markets.
Asan example,Pakistan hada credit control program
that involvedceilingson all the commercialbanks.They
had a simple rule for issuing some treasury securities.
Theyissued 91-daytreasurybills on any dayat 6 percent,
and they were prepared to repurchasethem at 6.05 per-
cent. The 6 percent was unrelated to the market rate,
whichwould havehad to be around 8 percent. Rather,it
performed the function of liquidity management per-
fectly well.Anyonewith excessfunds bought bills at 6
percent; if someone needed funds, he could sell bills
32 Paul Meek
back to the central bank at 6.05 percent. The banking
systemmanages the wholeliquiditysystemthrough this
kind ofmechanism,but it is not a secondarymarket or a
market in which treasury bills haveany significantrole
as a moneymarket instrument.
I think the decisionshould be madeon a case-by-case
basis.Evenso, I think there are more opportunitiesthan
may at first appear.I shall never forget advising some
yearsago in Portugal.It wassaidthat, becauseit wasjust
a small country,they couldnot aspireto open market op-
erations for a long time, ifever.AfterI was there about a
week, I pointed out that they were already doing open
market operations.Three times a weektheywere having
an auction of participationsin the central bank's portfo-
lio. Their explanationwasthat, becausethey had nation-
alized all the banks and had a credit ceiling,the banks
attracted far more deposits than they could utilize for
lending within the credit ceiling.The central bank was
sharing its higher-yieldingtreasurysecurities,so the na-
tionalized banks would not go bankrupt.To them, that
was not an open market operation,it was moppingup ex-
cess liquidityin the system. So there are many ways to
begin developinga market.
Question: What kinds of financial institutions are
able to trade in central bank certificatesin Indonesia?
Mr. Meek: Central bank certificates can now be
bought by anyone.They are bought initiallythrough the
dealers,who act as agents. I have encouragedthe Indo-
nesians to direct sales efforts towards the insurance
companiesand other financial institutionswho are nat-
ural residual holders of liquid balances in this form.
Dealers have begun by contracting repos (repurchase
agreements)with them for a week.The rate these insti-
tutions receiveis better than they can get at a bank. Asa
result, they have developeda fairly good repo market.
The central bank did developall the housekeepingfacil-
ities,and so on, to permit fairlyrapid transactionsin the
paper.
That is just what needsto be done;one needsto devel-
op a nonbank market for the central bank paper. Origi-
nally,holdersof central bank certificateswere limited to
the banks, but now anyonecan hold them.
Question:Is there also an interbank market in Indo-
nesia, and who are the main suppliersof funds to it?
Mr.Meek:Yes,indeed.The state banks,which hold 75
percent of deposits, typicallyare the ones that receive
large governmentdeposits.They do not know how long
they are goingto havethem, so these banks havea high
liquiditypreference.Byand large, their loan demand is
less than that of the private banks,which haveclose ties
to private enterprises. As a result, the borrowers in the
interbank market are largely the national private and
foreignbanks. One of the monetary reformswasto drop
the limitationon the amount a participantcould borrow
in the interbank market, because these banks had good
loan demand.
Although the state banks are suppliersto the inter-
bank market, the rates at which they are willingto sup-
ply indicate their managers have not been completely
converted to maximizing profits. For example, rather
than placing excessreservesavailablein the afternoonof
one day, if the rate is lower than they expect,they will
hold the reservesfor fear of loweringrates on moneyto
be placedthe next day. They end up holding excessre-
serves becausethey do not act to maximizeprofit.
I think this is a "teething" problem;these banks are
just learning how to function in competitive markets
and in a more competitivebanking system. One of the
healthy aspects is that the private banks havegenerated
a lot of competition for deposits.Whenwe visited there
last year,we found that this competition had produced
some unexpectedresults. Whereasthe central bank and
government were anxious to lower interest rates, the
state banks were afraid to lowertheir rates becausethey
would lose depositsto the private banks. As one would
expect,all this is getting workedout over time.
Question:Yousaid that the recipientswould receive
dollar deposits?Was there much currency substitution
in favorof the dollar?Was there capital outflowto Sin-
gapore?
Mr.Meek:I think the answer involvesthe difference
between wholesale and retail. That is, individualscan
hold time depositsthat are denominatedin dollars and
carry an interest rate that follows rates in Singapore,
though with some lag. Individualswith small holdings
do not havedirect accessto Singapore.There is not a lo-
cal dollar market that pays a wholesale rate, in some
sense, for dollar deposits.That market is only available
in Singapore.Businessmenor bankers who havelarger
holdingsand are worriedabout devaluationturn in their
rupiahs to the central bank for dollarsand placethe dol-
lars in Singapore.
Mr. Grenville:There were two ways by which enter-
prises couldbypassthe localbanker.If they could not get
liquiditywithin the credit ceiling, they could get it on
the local market at the market rate. That bypassedthe
commercialbanks, but they paid much more for it. Or
they could try to pick it up in Singapore,where they
could get credit denominatedin foreigncurrency.

Mr.Meek:Also,the central bank provided.swapfacili-
ties for foreigncurrency.But not everybodycould access
Singaporecredit.
Mr. Honohan: In some countries the authorities
maintained credit targets for the sake of the current ac-
count in the balance of payments.They want to restrain
overall expendituresto get the current account right. In
that case,this kind of leakageto Singaporeor some oth-
er readilyavailablemarket is a big problem.In other cas-
es, the authorities do not care about the current
account; theyjust want to ensurethat no officialfinanc-
ing is involved.If they can encourageprivate capital in-
flows, that is well and good. Private inflows do not
require officialfinancing of the balance of payments.
Then the moneyfrom Singaporeis treated as welcome.
However,since it might leavequicklyin some cases,it is
a shakier kind of financing.
Question:Youtalkedabout the other mechanismsfor
liquidity management. Could the central bank have
managedbank liquidityby withholdingor increasing li-
quiditycredits?
Mr. Meek:They used that approach at times in their
credit restraint program, which produced a liquidity
shortage. And in really tight times, they would make
such credits available.Basically,though, the timing of
priority credits is not under the central bank's control.
Theyoriginatein the commercialbanksand come to the
central bank. It is an uncontrolled elementto which the
central bank must respond.
Part of the problem with the steep yield curve is the
inflowand outflowof the treasury'sbalancesat the cen-
tral bank.Thesetend to be largerelativeto the size ofthe
market. So a question arises on whether to supplement
the open market operationwith a mechanismforplacing
treasury fundswith the banks. Malaysiafor example,did
this at one point, although it was not for the purpose I
am suggesting.The central bank was allocated certain
treasury funds that it could depositwith banks.
However,in Indonesia I have not yet detected any
great enthusiasm in the central bank for a mechanism
that would place treasury depositsawayfrom it. In lieu
of that, clearlya much better informationalsituation is
needed,one in which the treasury and the central bank
share intelligenceon big receipts or deposits.Some ef-
fort has been madein that direction,but it is still imper-
fect.
Most central bankerswith experiencein open market
operationswould agree in the utility of being able to
forecast the balance sheet. I suggested to the Indone-
sians fiveyears ago that it would be good to start fore-
Central Bank Liquidity Management and the Money Market 33
casting their balance sheet. There is a discipline this
brings: a knowledgeof seasonal demands.The central
bank people alreadyhave a visceralsense for that; they
know there are certain demandsfor cash, and so on.
And yet, although forecasting the future balance
sheet is useful and teaches those involveda gooddeal, it
is not necessaryfor effectiveconductof open market op-
erations. Indonesia has a complete and comprehensive
system even though the central bank does not forecast
the demand for reserves.The interbank rate itself tells
one whether the system is short or over-supplied.Every
afternoon around six o'clock, the DeputyGovernorde-
cideswhether or not to havean auction of seven-daycer-
tificates the next day. He knows what the foreign
exchangemarket has done and what central bank certif-
icates are maturing. He can guess whether the system
will be short or long the next dayand decidewhether or
not to do something. So, desirablethough it maybe, one
does not need projections of reserves to conduct open
market operations.
In the Indonesiancase,there wasa one-daylag. They
wouldannounce at the end ofone daythat theywere go-
ing to auction certificatesor makereposthe next day.So
they were not readingthe eventsof the day on which an
operation took place.
Mr.Grenville:In Australia,wherewe havegoodstatis-
tics and a goodrelation betweenthe central bankand the
government, we do this whole task. We do these daily
forecasts of liquidity,as I am sure they do here in the
UnitedStates, going out severalweeks ahead. That pro-
cess is very usefulfor our understandingof the market.
Weeven publish each morning at 9:30 a.m. the amount
by whichwe think the system willbe down or up. So we
stick our necks out. But when we get the bids in from
peoplewho want to do businesswith us, we may change
our minds on how much we want to do in the market
that day.Wecan makemistakes;the market sometimes
has a better feel forwhat is happeningwith liquidity,and
it lets us know.Just becausewe plan to do $A500million
in market operations,does not meanwewill actuallydo
$A500million. If you set it up right, the market can tell
you a lot.
Mr.Meek:Yes,in the Indonesiancasewith contempo-
raneous accounting,that is the bestguidancethe central
bank has. The informationthe banks have about their
positionis much more up to date than the central bank's
informationabout their position.
Question:With respect to the issue of credit control
and the market, it seemsthat the important question is
whether the officialinterest rate right now is more or
34 Paul Meek
less in line with the world interest rate. If it is, the au-
thorities would not havea big problem in trying to con-
trol credit.
Mr.Meek:In their situationwith modestexchangere-
serves and a completelyopen system, they cannot pur-
sue a differentinterest rate policy.Since I visited there,
the large premium for devaluationrisk has diminished
considerably.They are conducting their operations in
such way that peopleare now beginningto believethat
the gradualadjustment ofthe exchangerate is enough to
meet their needs and that there will not be another
maxi-devaluation.
The implicationsofthose maxi-devaluationsfor mon-
etary policyare interesting. They limited what the cen-
tral bank could do to lower interest rates, becausethey
resulted in a risk premiumwhich kept rates higher than
manypeople thought they should be.
Question:For some countries, foreigncurrency de-
posits are quite large.What was the size of the foreign-
denominated deposits in Indonesia at the time of the
maxi-devaluation?Theywere not that large,were they?
Mr.Meek:They were not huge, no. Theywere limited
mostlyto urban areas and reasonablysophisticatedpeo-
ple.
Mr. Honohan:I havea question about these devalua-
tions. A crucial, difficult issue comes up for a central
bank that is operating one of these unsteady exchange
rate policies,one that is supposedlyfixedbut is beingad-
justed at intervalsby these maxi-devaluations.Asyou de-
scribed the risk premium, it is more than could be
justifiedover long periods of devaluations.
This is a common phenomenon,but what should be
the central bank's response to a surge in this premium?
I accept that the central bank must livewith the premi-
um and cannot avoid responsibilityfor the effectof the
devaluation.The dilemma arises when lenders assume
that another maxi-devaluationwilloccur,probablysoon,
but the central bank knowsthere is no plan for a maxi-
devaluation.The situation is a bit like the crises of the
nineteenth century: what should central bank policybe
when there is a run on the overallsystem?The old solu-
tion was that the central bank should lend as much as
possible.
Mr.Meek:At high rates.
Mr.Honohan: Or, at least in practice,just to provide
all the liquiditythat is required. Now,that might be the
appropriate response for the hypothetical central bank
that is sure it will not devalueand has the means of pro-
vidingthe liquidity,that is, has accessto large reserves.
In the Europeanarrangement,whereunlimited reserves
are available,that could be the correct response for a
government that is sure where it is going: either the
country should devalueat once or it should lend in un-
limited amounts. In practice,this route is not often fol-
lowed;spikesin interest rates occur before devaluations
in the Europeansystem.
However,the optionof lendingin unlimited amounts
is not availableto Indonesia.Therefore,there appearsto
be a deficiencyin the arrangements for this kind of ex-
change rate system. If the country does not have the
credit facilitiesto back up its policy on exchange rate,
perhaps it should go to a much more flexibleexchange
rate system insteadof one with temporarypegs and ma-
jor devaluationjumps at intervals.
Mr. Meek:I think the latter is what the Indonesians
have undertaken to do. That approach has usually
washed out fairly quickly.For example,they made the
mistake in December 1986 of trying to ride the crisis
through. Theyhad no intention of devaluing,and so on,
but that just culminatedin a much bigger crisislater on.
Since that time, things have been reasonablywell be-
haved.
Mr.Grenville:The Indonesiansofferedto swap local
currency for foreigncurrency.This swap facilitywas not
open for everyone but open for those with heavy ex-
change rate exposure.That, in a sense, is making a bet
on whether there will be a devaluation.If the swapoffer
is used and no devaluationoccurs, then the central bank
is all right. If the devaluationdoes happen, the central
bank has taken major losses.
Mr. Honohan:Yes,that is the key point. Is the gov-
ernment sure it can maintain the rates? In the caseas de-
scribed,which mayhavebeen an oversimplification,the
central bank knew that it did not want to devalue and
had no intention of doing so. If it can staywith that po-
sition, the dice are loaded;it knowsthat it will win the
bet. If the bank'sactions indicatethat it knowsit willwin
the bet, when it offersthe bet nobodywill take it. How-
ever,to hedge the bet by sayingthat this week the inter-
est rate is 20 percent, next week it will be 23 percent,
revealsa lack of confidence.
Mr.Grenville:But when the bet is much bigger than
the foreign exchange reserves, the central bank gets
worried.They did not get to that stage, but they went to
the stage where foreign currency swaps were not far
short of their total foreignexchangereserves.

Mr. Honohan: Yes,and central banks lose these bets
because,as in the caseof the Philippines,they are forced
to take the bet. They probablyknewthey would lose it,
and they did loseit.
Mr.Meek:For a central bankerin a situation like that
in Indonesia,the best response is a quick defense. The
sharp rise in interest rates in the middle of 1987evoked
many complaints.Yet,it is not clear that the rise had any
lasting economicconsequences.In fact,I think it formed
the basis for the growing confidencethat there would
not be another maxi-devaluation.
Mr. Grenville:The two things go together. One very
goodmove the Indonesiansmadewasto price the use of
the currency swapfacilityin line with the interest rates.
The premium (or cost) of the swapwasbased on the dif-
ferencebetween domestic(i.e.,rupiah) and foreign(i.e.,
U.S. dollar) interest rates. As domestic interest rates
rose, the swap premium also rose. While the central
bank wasofferingthe bet, it wasalsomakingthe bet pro-
gressivelymore in its favor.If it wins the bet, it is in a
very goodposition.
Mr.Honohan:If a rise in short-term interest rates for
severalmonths is not costlyfor the economyand for in-
vestmentdecisionsand businessconfidence,then surely
one should go for the short-term rise. Still, I am not sure
it is without significant costs. Those costs could be a
heavy price to pay in situations where the country
should just devalueat once. In other cases,the country
willbe able to ride it out; if the centralbank is not wrong,
sooner or later the money has to come back. If it can
hold out, it will be all right.
Mr.Caprio:In general,the Central Bankhas to guard
against a common occupationalhazard, to wit, that of
not recognizingwhen a disequilibriumis fundamental.
At least a part of the "art of central banking" involves
making such difficultdecisions.
Central Bank Liquidity Management and the Money Market 35
Question:I think that betting against the market is a
dangerouspath to follow.If there is an expectationin the
market, there is little you can do about it. In 1982,the
Argentine central bank went into this rough exercise.
The result was that, as they say in Argentina,"all the
losseswere socialized."The privatesector wascomplete-
ly wiped out by the high cost of the debt in foreigncur-
rency.Allthe losseswere taken bythe central bank. So I
think it is very dangerousto try to intervenein a market,
for the central bank to take such a high risk.
In the case of Indonesia, I understand there were
some dollar-denominateddeposits,whichthe systemap-
parentlywasable to continueintermediatingdespitethe
expectationof devaluation.Perhapsthere were also dol-
lar-denominatedloans.When the banks receivea deposit
in dollars,what do they do?Do they alsolend in dollars?
That is one solution; another would be to index de-
positsor loans denominatedin dollars,or to indexbonds
to the dollar.Indexationis alwaysonewayto copewith a
devaluationrisk. Mexicohas followedthis coursefor the
last couple ofyears. Realdomesticinterest rates in Mex-
ico havebeen extremelyhigh. The government has be-
gun to issue indexed bonds; this allows the banking
system to index deposits and loans in dollars. But the
dollar interest rate domestically is still high, which
provesthe problem is not just the expectationof devalu-
ation but also some risk of default.There is no financial
engineeringthat can solvethis problem.
Mr.Honohan:I think you are sayingthat countries in
volatilesituations without significantforeign exchange
reservesreally cannot commit to a fixedexchange rate
without heavycosts.A fixedexchangerate policy,where
the possibilityof devaluationexists,is a risky policy to
follow.
Note
1. Editor Note:This point was emphasizedby Deputy
GovernorLin of BankNegaraMalaysia;see chapter 11.

The Use of Monetary Policy Instruments by Developing
Countries
R. Barry Johnston
Today'ssessionis entitled nuts and bolts, and this lecture
is on new instruments and techniques.BeforeI get down
to "nuts and bolts," I think it is worth reminding our-
selvesthat the typesof instruments and techniquesI will
describe are developedin pursuit of broader objectives
for economicgrowth and development.I stress this be-
cause the transformationto a more market-orientedfi-
nancial systemsis not necessarilyeasy to accomplish.It
often involvesa fundamentalreorientationof thinking in
the central bank as regardsits role in allocating financial
resources, the commercializationof state-ownedfinan-
cial institutions and in the government concerning the
paymentof market-relatedinterest rates on its debt.
The reform of monetary instruments should also be
viewedas only part of broader financial reforms which
aim at making the financial system more responsiveto
market forcesand more competitive.I would include in
this:
(1) Steps to liberalizeinterest rates
(2) A reduced reliance on direct controls to allocate
credit and the removal of other discriminatoryregula-
tions that inhibit competition between financialinstitu-
tions and fragmentthe financialsystem
(3) A reduction in the barriers to entry of new finan-
cialinstitutions and the exit and restructuring of existing
inefficientinstitutions
(4) The developmentof new financial markets and
new financialinstruments.
Financial reform is in its turn an integral part of
broader economicliberalizationthat involvesa shift from
allocating resources through directives, controls, and
subsidiestoward a greater role for market forces.
Why Is There a Need to ReformMonetary
Control Techniques?
The system of monetary control in most developing
countries has typicallyinvolvedsome ifnot all of the fol-
37
3
lowing elements: direct controls on interest rates (in-
cluding preferential rates for certain loan categories);
aggregateand individualbank credit ceilings;selective
credit controls and preferential central bank refinance
facilitiesto direct credit to priority sectors; and high re-
serveand liquidasset requirements,designedboth to ab-
sorb liquidityand to providegovernmentdeficitfinance.
The directcontrols havea number ofdrawbacks.They
often become ineffectiveover time as an instrument of
monetary control and inhibit efficient resource alloca-
tion. Creditceilingsand other directcontrols force insti-
tutions into portfolio positions that they would not
otherwise voluntarily accept. Hence, banks and their
customers have incentivesto avoidthe direct controls.
The implicittax imposedon commercialbanks,and thus
on their borrowersand depositors,by credit ceilings,in-
terest rate controls,and high (but low-or non-renumer-
ated) reserve and liquid asset ratio requirements,
encouragesthe emergenceof other, unregulated,finan-
cial intermediaries and instruments that compete with
the regulated ones. This weakensmonetary control by
eroding the effectivenessof the direct control and dis-
torts monetary indicators.
Monetarycontrol is also weakenedwhen credit ceil-
ings involvea buildup of excessliquidityat regulated in-
stitutions, as this discourages deposit-taking by
regulated institutions, and in turn inhibits savings mo-
bilization or causes disintermediation.Savingsmobili-
zation may not be promoted simply by mandating
positive real deposit rates at regulated institutions, as
they mayeither turn awaydepositorsor findwaysof pay-
ing lower effectivedeposit rates. It is generallynot pos-
sible to control both the cost and quantity of credit,
although in practice this is often attempted using direct
controls.
The efficiencyof credit allocationis alsoadverselyaf-
fected by direct controls. It is difficultto design proce-
dures that do not penalize more efficient institutions,
38 R. Barry Johnston
and the support of inefficientinstitutions may lead to
higher averagetransaction marginsand lendingspreads.
High liquid asset ratios create captivemarkets for gov-
ernment securities and an inappropriatepricingof cred-
it, including a substantial interest subsidy to
government.Withdirect controls, little attention is paid
to developingmoneyand capital markets,whichare cen-
tral to the efficientallocation of resources.
KeyElements in a LiberalMonetaryControl
Framework
Themain elementsof the reformof monetary control
instruments are
(1) The developmentof a frameworkfor making de-
cisionsabout interest ratesand the path of financialvari-
ables in a liberalsystem
(2) The introduction of new techniques for govern-
ment domesticdebt management
(3) Areformof central bank operationsand facilities
(4) Supportinginstitutional reforms.
The Monetary Control Framework
In a liberalsystem,the main instrument of monetary
control is the central bank's control overthe stock of re-
serve money (cash and balanceswith the central bank)
and hence over money market interest rates. Money
market rates, in turn, affectother lending and deposit
Table3-1. Componentsof the ReserveMoneyProgram and DiscretionaryPolicies
NondiscretionaryElements
Net purchaseor sales of foreignexchange
against local currency.
Changein currency in circulationoutside
the central bank.
Changesin holdingsof obligatoryreservesand
precautionarybalancesdue to changes in bank
depositsand balance sheet totals.
Net receiptsand paymentsthrough government
accountswith the central bank including
redemptionsand issuesof governmentsecurities.
Redemptionsof central bank bills and tap sales of
bills;maturing central bank loansand assistance,
and current receipts;and paymentsby the
central bank.
Automaticcentral bank credits to prioritysectors.
rates. Such a system doesnot rely on high or evencom-
pulsoryreserveor liquidasset requirements but only on
the central bank's ability to manage its own balance
sheet and to control the terms at which it is willing to
provideassistanceto cover reserveshortages.
The central element of the indirect approachto mon-
etary control is thereforethe central bank's control over
the items in its own balance sheet. Paul Meekhas de-
scribed what these items are. Table 3-1 breaks these
items down into nondiscretionaryfactors-those which
are beyondthe short-term control of the central bank-
and the discretionary operations of the central bank,
whichcan be used to manage the supplyof reservemon-
ey relative to the demand for it, to meet required and
precautionary reserve balances or the level of money
market interest rates.
The frameworkformanagementcan be formalizedby
developinga reservemoneyprogram that involvesfore-
castingthe main nondiscretionaryelements and setting
out the options for the discretionaryfactorsin order to
achieve policyobjectives,or to adjust the policy targets
in response to unanticipated movements in reserve
moneycomponents.The establishmentof such a frame-
work requires a blend of economicand statistical analy-
sis and operational experience and often involves
reorganization in the treasury and central bank. The
treasury needsto be able to monitor and forecastits own
cash position,while the central bank has to developthe
frameworkfor forecastingand coordinatinginformation
DiscretionaryElements and Policies
Policywith regardto the exchangerate and net
foreignexchangemarket intervention.
Obligatoryreserverequirements;penaltiesfor obligatory
reserveshortages;availabilityand interest rate on
borrowingfrom the Central Bank.
Governmentdepositand debt management.
Sales of central bank and treasury billsat auction;
policytowardsthe tap window;and short-term
moneymarket intervention.
Changesin rediscount policiesand discountwindowrates.

on the trend in other domestic and external variables
and developingthe overall implicationsfor the cash re-
serves of the financialsystem.
The establishmentof the monetary operatingframe-
work shouldbe one of the critical first steps in the adop-
tion of indirect instruments of monetary control. In
practice, however,the establishment of such a frame-
work is often givenlowerprioritythan, say,the more vis-
ible auctioning of treasury bills. Consequently,the
efficiencywith whichmonetarypolicyhas beenconduct-
ed with indirect instruments has sometimesbeen poor.
DomesticDebt Management
The starting point for developingmarket-basedmon-
etarycontrol in developingcountries has often beenpro-
cedures for the primary issues of treasury bills and/or
central bank bills. In some countries the catalyst has
been markets in longer-term government securities.
However,because commercialbanks are often already
required to hold treasury bills to meet liquid asset re-
quirements or have built up excess reserves with the
central bank, treasury and central bank bills are often a
useful first instrument in developingmarket-basedpro-
cedures.
Issues of central bank paper have been considered
usefulbecauseof the potentiallygreater freedomfor the
central bank to tailor issues of its own paper to achieve
monetary objectiveswithout direct interference from
governmentbudgetaryconsiderations.However,the net
budgetaryand monetaryimpactfrom using centralbank
rather than treasury paper is similar, and the choice is
largelyan operational decision.
The role of security issues as instruments of mone-
tary control can be illustrated using basic financialflow
and balance sheet equations. The basic approach is to
generate projectionsof (a)the supplyof cash reservesre-
sulting from transactions by the government and the
central bank and (b)the privatesector's demandfor cash
reservesresultingfrom the public'sdemandfor currency
and the banking system'sdemandfor required and pre-
cautionary cash reservebalances.The volumeof securi-
ties issued is used to manage the differencesbetween
these suppliesand demands.
A neutral monetary policy stance-i.e., one which
would leavemoneymarket rates unchanged-would in-
volveremovingan excesssupplyof cash reservesby sell-
ing a volumeof treasury billsto the private sector equal
to the differencebetween the forecastsupply of cash re-
servesand the estimate of demand (or relievinga fore-
cast reserveshortage by not rolling overmaturing bills).
Arestrictivepolicystance designedto raiseinterest rates
would involvethe creation of a reserveshortage by sell-
The Use of Monetary Policy Instruments by Developing Countries 39
ing a volumeofbills in excessofthe forecastreservesur-
plus. In this case, and assuming the forecasts are
accurate, banks would find themselvesshort of reserves
and interbank rateswouldtend to rise. If the sale of trea-
surybills reducedthe aggregatesupplyof reservesbelow
reserverequirements,bankswould haveto seek margin-
al accommodationfrom the central bank and the inter-
est rate at whichthe central bank providesthe assistance
would becomethe keydeterminant ofthe overalllevelof
interest rates. An expansionary stance would involve
sellingfewerbills than the forecastcash surplus. In this
casebankswould haveexcesscash reserves,which could
push interest rates down.Interbankrates wouldfall,and
there wouldbe repaymentsof central bank credit.In this
framework,the designof rediscount facilitiesis an inte-
gral elementof reserveand interest rate management.
Selling techniques for primary bill issues include
"free"auctions, fixedprice tenders, and tap sales at pre-
determined interest rates. Under free auctions, the cen-
tral bank announces the volume of bills to be sold and
asks for competitive bids. Bills are allocated starting
with the highest bid price and moving to the next high-
est, etc., until the tender is fullysold.Undera fixedprice
tender or with tap sales, the central bank sets the inter-
est rate on bills and acceptsthe volumedemand at that
rate. With a free auction, interest rates fully adjust to
achieve the quantitative objectivesfor the sale of bills
and hence forthe targeted stockof reserves,but this may
risk volatilityin interest rates from auction to auction.
On the other hand, with a fixedprice tender, the interest
rate is determined in the short run and quantitiesare al-
lowedto vary. The tender price can be varied from one
tender to the next to exercise the desired quantitative
control, but this procedure mayrisk a "bias to delay"in
adjusting interest rates.
In many countries that are introducing market-ori-
ented procedures,billsare auctioned, but no tender vol-
ume is announced in advance. The central bank
exercisesits discretionin acceptingor rejectingbids, de-
pending on its quantitative or interest rate objectives.
The decision not to announce the tender amount can
thus givethe central bank added flexibility,although at
the cost of reducing informationto market participants.
In the initial stages, the benefitsto the central bank of
the increased flexibilitycan help overcome concerns
about implementingnew monetary control procedures.
In some cases, bills are auctionedand tender volumes
announced in advance,and a permissiblerange is also
announced for bids that wouldbe accepted.
The introductionof treasury billauctions often needs
to be accompaniedwith a broader reformofthe domestic
debt management instruments and policy.For example,
the simultaneoussale of treasury billsat auction and on
40 R. Barry Johnston
tap at rates closeto those determinedin the auctions can
deflectdemandfrom the auction and bias upwardthe in-
terest rate bids at the auction.On the other hand, allow-
ing captiveinstitutions to bid at the auction can bias the
bidded interest rates downward.In either case,the auc-
tion bidswill not be representativeof market rates.
CentralBank Facilitiesand Operations
In support of the developmentof a more market-ori-
ented financialsystem, the central bank mayhave to re-
form its existingfacilitiesand introduce new operational
arrangements.
RefinanceFacilities
Inance
isamanycourcenta re freetial. refi
nance~~~~~~~~~~ isamjrsuc.frsremnyceto.Sc
refinanceis often automatic against eligiblebank loans
and carries below market interest rates. The central
bank often lacks the instruments to control the expan-
sion of the refinance in the short-run, which therefore
has the potential to undermine monetarycontrol, while
the preferential interest rates distort the allocation of
credit and the competitivepositionof differentfinancial
institutions. Bringing this preferentialrefinanceunder
control is often a matter of urgency.
Reformsof the refinancefacilitiescould include the
introduction of bank-specificcredit ceilingson the vol-
ume of short-term refinance and raising and unifying
the interest rates on this refinanceat the bank rate. The
bank rate should be raised and maintained above the
treasury bill rate. Such short-term refinance should
eventuallybe mergedwith the central bank's dailymon-
ey market operations.
As regards long-term project refinance,this could be
auctioned,as this would help determine a more appro-
priate pricing for long-term funds and providethe cen-
tral bank with closer control over its total volume.Over
time, as the capital market develops,the auction rates
and capitalmarket rateswouldconvergeand the auction
of refinancecould be phased out and replacedwith capi-
tal market transactions.
Proceduresfor Day-to-DayLiquidityManagement
Auctionsof primarybills to the general publicinvolve
administrativecosts and delays.Asa result, primarybill
issues have tended to occur no more frequently than
once a week,which also seems reasonablyoptimal tim-
ing for these issues. However,as the actual outcome for
reserves may deviate significantlyfrom the outcome
projectedby the authorities, there maybe a needfor sup-
plementary central bank operations to manage money
market liquiditybetween the issues. If not offset by the
central bank,the resulting reserveshortagesor surplus-
es could result in excessivevolatilityin interest ratesand
could complicatebanks' compliancewith legal reserve
requirements.
The aim of achieving interest rate flexibilitywhile
avoidingexcessivevolatilityoften callsfor day-to-dayin-
terventionby the central bank.The most common forms
are lending through the central bank rediscount win-
dow,outright sales and purchases, and repurchases of
securities.
The availabilityof usable collateral is often a critical
bottleneckin central bank day-to-dayliquiditymanage-
ment and the developmentof moneymarkets more gen-
erally.For example,in many countries,the use of lender-
of-last-resortfacilitiesis inhibited by the lack of unen-
cumberedcollateral.Banks mayhold large stocks of liq-
uid assets, but these are not usable as they are held to
meet liquidasset re uirements.To increasethe usabilit
q qs y
of liquidassets, the liquidasset requirements can be re-
placedwith norms that would specifythe averagehold-
ings of liquidassets over a sufficientlylong period so as
not to restrict banks' ability to utilize liquidassets at a
particularpoint in time. Alsothe central bank should en-
couragethe developmentof bankers acceptances (BAs).
The BAis a convenientinstrument that can be used to
collateralizeinterbank trading between institutions of
diversecredit standing and can be used by the central
bank to inject liquiditywhen there is a shortage of trea-
sury or central bank bills. Withan increasedavailability
of collateral,the rediscountwindowcan becomea useful
instrument for liquidity management by banks and
hence for the central bank to direct the level of interest
rates through its rediscountrate policy.
A rediscountwindowonly providesa facilityto elimi-
nate reserveshortages;there is a risk that the rediscount
rate wouldbe viewedas the indicativerate forall interest
rates, inhibiting their market determination. These
problems can be mitigated by providinggraduated ac-
cess to the rediscountfacilityat progressivelyhigher pe-
nal rates and special interest-bearingaccounts with the
central banks for excessreserves.
Theseproceduresare basedon preannouncedinterest
ratesand leavethe initiativeto obtain or disposeof bank
reserves entirelyto the commercialbanks.The alterna-
tive is for the central bank to take the initiativeas re-
gards the quantity of intervention and leave interest
rates to be market-determined. For example, when
shortages are indicated, the central bank could an-
nounce its willingness to buy bills up to a certain
amount, then ask forcompetitiveoffers;when surpluses
are indicated,the central bank can announce its willing-
ness to sell a certain volume of bills,then ask for com-

petitive bids. The central bank would then be able to
acceptthe bids and offersto meet its interventionand in-
terest rate objectives. By restricting the auctions of
money market intervention instruments to a few key
moneymarket participants,these auctions can be more
flexiblyimplemented than the general auctions of pri-
mary bills. Experience indicates that active secondary
marketsare most likelyto developwhenthe central bank
begins these daily discretionary operationsand ceases
posting intervention rates.
The developmentof dailyoperationsoften needsto be
supportedby reforms of reserverequirements.Penalties
for noncompliancewith reserverequirements mayhave
to be increasedto ensure that the reserverequirements
are a binding target for the central bank's day-to-day
money market operations. The lagging of reserve re-
quirements can help provide the authorities with firm
informationabout the levelof reservesdemandedby the
banking systems.Also,averagingthe periodfor compli-
ance with the requirements can introduce additional
flexibilityand reduceinterest rate volatilityand the need
for day-to-dayinterventionby the central bank.
One of the major sourcesof noise in the moneymar-
ket can be the clearing and settlement arrangements.
The money market providesa means of access to good
funds for the processing of payments and the mainte-
nance ofovernightrequired reservebalances.Totake op-
timal reserve positions, a bank requires up-to-date
informationon its cash positionand its customers' pay-
ment flowsand time to offsetthese, ifnecessary,through
the money markets. If banks do not knowtheir require-
ments for goodfunds until afterthe moneymarketshave
closed, the consequencesare thinner markets, greater
volatility in interest rates, increased risk for market
makers, and larger holdings of precautionary reserve
balancesby banks. The lack of informationcreates a se-
rious problem for the central bank, since the central
bank would not know the aggregateshortages or sur-
pluses that it wouldneed to offsetin its dailymoneymar-
ket operations.
One way of avoiding these problems is a system of
same-daysettlement for moneymarket transactionsand
next-daysettlement for all other paymentorders. In this
waybanks and the centralbank can know,usuallyin the
morning, the net paymentorders transacted the previ-
ous daythat haveto be settledtoday.Theycan then enter
the moneymarket to trade and balancetheir settlement
positions. Similarly,the central bank can establish a
timetablefor its ownmoneymarket intervention.
Repurchasetransactions havea number of advantag-
es for central banks in managing moneymarket liquidi-
ty. It is not necessary to have developed financial
marketsto undertake a repurchasetransaction.The cen-
The Use of Monetary Policy Instruments by Developing Countries 41
tral bank needs only to communicate to the dealers or
banks its desire to trade and the direction in which it
wishesto dealand to ask forinterest rate bids and offered
amounts. Repurchases are self-liquidatingand can be
targeted to temporary fluctuations in bank reserves.
They overcomethe frequent problem of a shortage of
specificmaturities and the large transaction costs in-
volvedin buying and selling securities outright to man-
age a temporary liquidityimbalance.Theyare therefore
a flexibleand cost effectivetool for managing highly
variablebank reserves.
SupportingArrangementsand Institution Building
Paul Meekhas alreadytouchedon these issues.In ad-
dition to financial institution building, let me briefly
mention a number of broader structural and policy is-
sues that should also be addressedas part of the reform
of monetary instruments.
The liberalizationof interest rates is often inhibited
bythe captivemarketsfor governmentsecuritiescreated
by high liquid asset requirements or other portfoliore-
strictions. Liquidasset ratios may therefore need to be
reduced or the range of assets eligibleto meet the re-
quirements expandedto include prime quality private
papers.This wouldbe desirableanyway,as a wayof stim-
ulating the market in BAsand other private papers.Pub-
lic sector financial institutions (and the central bank)
should havetheir securitiesallotted outsidethe auctions
to a dsrtin ateauctio rule T e iomguaran-.
teta aktbsdsse ol ecmeiie
IHence,measures to increase financial sector competi-
tion, including bank restructuring and the reduction in
barriers to entry that I mentioned at the beginning,may
have to eny the reformsa
haveto accompanythe reforms.
For interest rates to havethe appropriate impact on
resource allocation,borrowers have to be responsiveto
market prices. A high incidence of credit undiscipline
and loan delinquency,or arrangementswhereborrowers
or banks expectto have their loans coveredby the gov-
ernment, can render interest rates meaninglessin the
allocationof credit. The strengthening of bank supervi-
sory arrangements,bankruptcy laws,and the legal pro-
cedures for loan recovery may therefore have to
accompanythe financialreforms.
Asa result ofinterest rate liberalization,interest rates
on government debt may rise, with budgetaryimplica-
tions. Similarly,parastatalsand other borrowers,which
haveaccessto bank loansat preferentialrates, mayexpe-
rience an increase in borrowingcosts.As a result, cer-
tain projects may no longer be viable and productive
sector restructuring has to take place.
42 R. Barry Johnston
Managingthe Reformin MonetaryControl
Techniques
Let me finishthis talk by discussingsome of the com-
plex issues in the implementationof monetary control
that can accompanyfinancialreform-issues that weare
only now comingto gripswith.
Sequencingof Reforns
The successfultransition to more market-orientedfi-
nancial systems requires careful consideration of the
speedand sequencingofreform.It takes time to build up
expertise,to changeattitudes,and to establishnew insti-
tutional arrangements necessary to ensure effective
monetarycontrol and competitivemarket mechanisms.
This suggests a phasedapproach to reform.
Underthe phasedapproach,interest rates couldbe set
free in stages, for example,by gradually wideningthe
permissibleranges for deposit and lending rates or by
steadilyadjustingminimum depositrates and maximum
lending rates. The traditional instruments-reserve re-
quirementsand rediscountwindows-could be deployed
initiallyto manage bank reservesand therebythe credit-
creating capacity of the banking system,while phasing
out credit ceilingson banks.However,credit ceilingson
individualbanks could still inhibit competition;there is
a risk-evident in some countries-that the phasing of
reform is so slowthat little fundamentallychanges.
Figure 3.1 Dynamics of Credit and Deposit Growth with Financial Liberalization
A. PositiveRealInterestRates
Growvthof A
aggregates
in real | Credits
tenns
Deposits
Date offinancial
liberalization
Time
Deposit growth
minus credit
growth
If the initialfinancialsystemis highly regulated,with
each regulationsupporting the other, it may be difficult
to implementa gradual, effectivereform. In such cases,
a major one-stepchange in the operational system may
be unavoidable.This has been the policy followedin a
number of countries. However,rapid structural changes
carry a risk of a loss of monetary control and broader fi-
nancial sector instability and require careful manage-
ment.
A detailedexaminationof the sequencingof financial
reforms in five countries has providedsome important
conclusionsabout the dynamicsof the growth of depos-
its and credit followingfinancialliberalization.Theseare
illustrated in Figure 3-1. In the pre-reform period, de-
posit growthdeclinesbecauseof the negativereal depos-
it rates and the overall repressednature of the financial
system.Credit growth is maintainedthrough increasing
liquiditysupport from the central bank. Depositgrowth
that is lower than credit growth is associatedwith in-
creasing resource pressuresin the pre-reformperiod.
After financialliberalization,both credit and deposit
growthincrease.However,the responseof credit growth
is initially more rapid than depositgrowth. The decline
of depositsin the pre-reformperiodreflecteda voluntary
portfolio response to financial repression rather than
specificcontrols. In the post-reform period there is a
gradual portfolioadjustment by depositors to the new
liberal financial situation. Credit growth, in contrast,
was constrained by direct controls with an excess de-
B. NegativeReal InterestRates
Growvth
of Ai
aggregates
in real
terms Credits
D/s=
Date offnancial
'/liberalization
Time
Deposit growth
minus credit
growth

mand for credit. Once the direct controls are removed,
financialinstitutions respond by meeting the excessde-
mand for credit, and credit expandsrapidly.The initial
effect of the financial liberalizationis, therefore, to in-
crease the resource imbalances.
The subsequent developmentin deposit and credit
growth depends on the structure of real interest rates.
Whenreal interest ratesare positive,the growth of credit
slowsdown comparedwith the initialpost-reformcredit
boom.The growth of credit may remain higher than in
the pre-reform period becauseof the general increased
role of the financial sector in mobilizingand allocating
resources followingfinancialliberalization.The growth
ofdepositscontinuesto increase.This reflectsthe lagged
portfolioadjustment to the financialliberalizationmea-
sures, the developmentof newfinancialinstruments and
institutions followingthe liberalization,and the reduc-
tion in central bank liquiditysupportof financialinstitu-
tions, which in turn are forced to mobilizedeposits to
meet credit demand.Aftersome point the growth of de-
positsand credit converge,allowingfor balancedgrowth
with a higher levelof overallresource mobilizationthan
in the pre-reformperiod.
If real interest rates are negative in the post-reform
period, this encourages a more rapid growth of credit
and slowergrowth of deposits.The maintenance of neg-
ative real rates requires expansionarycentral bank poli-
cies that financethe faster growthof credit in relation to
deposits.As a result, the growth of deposits does not
catch up with the growth of credit and resource imbal-
ances remain. These maybe wider than in the pre-liber-
alizationperiod.
Even with the maintenance of positive real interest
rates, there is likely to be an initial post-liberalization
credit boom that can pose a threat to economicstability.
The managementofthe credit boomis a critical element
of successfulfinancialliberalization.For example,finan-
cial liberalizationmayhaveto be accompaniedby a ster-
ilization operation aimed at absorbing excessliquidity.
However,to the extent that the initial credit growth re-
flectsa one- time stock adjustment to a new equilibrium
position,an attempt to constrain it solelythrough inter-
est rates could result in veryhigh real interest rates. The
economy-wideimplicationsmaythereforehavetobeoff-
set by an accompanyingreduction in the fiscaldeficitor
a temporary increasein externalborrowing
Designof FinancialProgramsin the Post-Reforn
Period
The design of financialprograms mayhave to be re-
thought followingfinancialreform. Financialprogram-
The Use of Monetary Policy Instruments by Developing Countries 43
ming, at the level of the banking system, is likelyto be
made more difficultand responsesmore uncertain fol-
lowing the change to indirect monetary control proce-
dures.
First,the relationshipsbetweenmoneyand credit and
final objectivesthat existedduring the pre-reformphase
are likelyto be less reliable followingfinancial reform.
The introduction of new instruments may shift the in-
terest sensitivityof moneydemand.For example,a liber-
alizationof interest rates on bank depositsmay result in
broad money becoming less sensitiveto changes in the
general levelof interest rates. The removalof creditceil-
ings that constrain portfolio allocation can result in
portfolioshifts, including a demandto hold larger mon-
ey balances at any given level of interest rates and in-
come. Higher interest rates may lead initially to
distressed borrowing. Money multipliers may also
change with reform,for example,when there is reinter-
mediationto the bankingsystemleadingto reducedcash
holdingsby the non-bankpublic.
Second,the attainment of short-run quantitativetar-
gets for bank credit, consistentwith a desiredoutcome
for the balance of payments(establishedas part of the fi-
nancial program),wouldbe extremelydifficultwhen in-
terest rates or the cash reserveof the bankingsystemare
the principal monetary control instruments. This re-
flects the considerable uncertainty about the size and
even the sign of the short-run interest elasticityof the
demandfor bank credit.This is not to say that an overall
inflation or balance of payments outcome could not be
achieved using indirect monetary control instruments
but that precise financial programmingat the level of
the banking systemwould be difficult.
A practicablemonetary policyframeworkconsistent
with a market-oriented system of monetary control,
quantitative performance criteria, and explicit targets
for the balance of paymentsis likelyto be one definedat
the level of the central bank's balance sheet rather than
the banking system. Monetarycontrol would be exer-
cised through sales of securities to the private sector,
with the aim of achievingquantitativetargets forthe net
domestic assets of the central bank. This framework
wouldalso be consistentwith the use ofbroad or narrow
monetary aggregates as intermediate targets. The set-
ting of quantitativetargets on the central bank's balance
sheet also fits neatlywith the use of primaryauctions of
securities as the principal market-basedinstrument of
monetary control in underdevelopedfinancial markets.
These are areas and questions that are under study.
44 R. Barry Johnston
Discussion
Mr.Caprio:What role do you see for liquid asset ra-
tios? In some countries, the authorities view them as a
tool of monetary policy;in others they are viewedas a
tool for prudential concerns.
Mr. Johnston: Liquid assets should not be a tool of
monetary policy.They should be primarily a prudential
instrument. That is why I emphasizethe availabilityof
usable collateral.Unfortunately,when you set require-
ments the collateralbecomes unusable. Therefore,the
prudential buffer that the stock should provide is not
there. One mayhavebanks that meet a liquidityrequire-
ment of 35 percent of assets but have no collateralto
trade. So they cannot go into the money market; they
cannot go to the central bank. The central bank quite of-
ten ends up having to announce another window at
which it makesunsecured loans,becausethe banks can-
not use their liquidassets stock as collateral.
Quite often liquid asset requirements cannot be rap-
idly reducedwithout major effectson the government's
borrowing position because much of its deficit is fi-
nancedthrough liquidasset requirements on the banks.
As I suggested,the first step is to try to get these liquid
AsIsuggested,
first stepistotrytogettheselten
the
assets into a usable form, so they can serve the pruden-
tial role of providing liquidity to the banking system.
This should be the primary objective.For this reason,
the liquidasset requirementshould be stated as an aver-
age over one or two months, so the banks can use it on a
dailybasis,but re-establishthe needed ratio for the stat-
utory period.
In addition, many bank supervisorsemphasize that
the country should determine its minimum liquidasset
requirements based on strictly prudential guidelines.
Such guidelinesmight look at the maturity structure of
the assets and liabilities,rather than just mandating a
blanketratio of 35 percent on the wholebalancesheet. It
is better to analyze in finer detail the system'sneedsfor
liquidity.For example,demand deposits are not really
very liquid,in the sensethat they normallyare returned
to the system. On the other hand, one-month deposits
tend to be withdrawnwhenthey mature. The aim should
be to developrelativeweights and ratios which will pro-
vide an appropriateamount of liquidityfor the system
for prudential reasons.
Question:At the beginningof your presentation,you
mentioned reasons why one should or might want to
moveawayfrom direct credit controls. For the countries
where you have experience,why did they actuallytake
the decision?
Mr. Johnston: In some cases it was evidentthat ceil-
ings were having no influence.Basicallythey were just
repressing a number of regulated banks. Outside the
ceilings,you had the "gossamer"banks, whichwere do-
ing things similar to what the regulated banks did but
were not subject to the ceilings. In that sense, the au-
thorities had lost monetary control. Wehaveworkedon
one that was near that point.
For other countries,it wasvery much a consciousde-
cision.They wanted to liberalizetheir financialsystem;
theywantedto moveto market-orientedprocedures,and
this was part of that process. They realized they could
not have a liberal system if they had credit controls. I
think these havebeen the two main reasons.
Question:Has there been a caseof liberalizationthat
has been both rapid and stable?
Mr.Johnston: I am not aware of one, although my ex-
perienceis not extensive.On the one hand is the example
of Korea, where liberalizationhas been in progress for
years; they are still regulated,and they havenot had
instability. But I do not think that is rapid financial
transformation.The present experiencein Eastern Eu-
rope is still being playedout. I do not think they haveyet
experiencedthis problem,but it is premature to say what
is happeningthere. The circumstancesmay be a bit dif-
ferent in Eastern Europe, where the banking systems
were much more constrainedand the private sector was
not sufficientlydevelopedto take advantage of the re-
movalof ceilings.
Question:Whatabout the experiencein NewZealand
with rapid liberalization?
Mr. Caprio:They began liberalizationin 1984.New
Zealand did everything very much like the southern
cone countries. They openedup on the capitalflowside
and removed all the credit controls. Real interest rates
went up to about 15 percent, then came down to about
8-10 percent. In fact, New Zealand's experience more
closely resembles graph A in Figure 3-1. There was a
boom in credit for about five years, but it finallycame
back in line with deposits.
Question:In many cases in the developingcountries,
banks must rely on short-term deposits.They lack the

longer-term resources they need, which of course ex-
plains why term credit is not availablefor economic
growth. Besides interest rate regulation, is there any-
thing the central bank can do and should do to encour-
age saversto shift to longer-termsecurities?
Mr.Johnston:I am of the viewthat commercialbanks
should not be in the business of providing long-term
credit. Short-term financing institutions have short-
term liabilities.Mostof their assets are in trade and oth-
er credit. Thoseassets should not be directed into long-
er-term lending;that is the role of the capital markets.
The wayto get longer-termcredit is to developthe capi-
tal markets. Get bonds and equities into the system; let
the private sector take the responsibility.But do not put
the risk on the banks.
In these countries, the banks are already weakened.
To try to force them outside their preferentiallending
policies can be a risky strategy.Instead, the risk should
be dispersed more generally in the system, and that is
the role of the capital markets. To generate the capital
markets, change the refinance facilities, make interest
rates flexible,and reduce the preferential credits. Ulti-
mately, that will help to get long-term credit into the
economy.
Question: Operationally,should we be encouraging
the simultaneous developmentof the capital market
whenweare
wih
woring
the mnetarysector?Or doe
when we areworkngwththemonetarysector?Ordoes
Mr. Johnston: Definitelyit should be simultaneous.
The working paper that has been circulated specifically
recommends reform of the monetary instruments si
multaneously with developmentof the money market
and capital markets. These elements of the system are
mutually supporting;each is needed for the overall sys-
tem to progress.
I believethe emphasisshould be on the capital mar-
kets. Ultimately,that is where the domestic resources
should be mobilizedto financedevelopment.
* Question:I suppose,though, the emphasiswouldde-
pend on the level of institutional developmentin the
economyas a whole. For some countries, is it not unre-
alistic to expectthe capital marketsto provide much in
the way of resources in the near term? Few countries
havecapitalmarkets providingevenone quarter ofgross
investment needs.
Mr. Johnston: But we normally find that, when the
capital market is not providingany resources, it is be-
cause the central bank is providingthem or the develop-
The Use of Monetary Policy Instruments by Developing Countries 45
ment-financeinstitutions are getting highly subsidized
loans.As a result, no one else can makeany money pro-
vidinglong-term resources.There is no role for the cap-
ital market. Theyieldcurve maybe bendingdownwards,
so no one can intermediateat the long endbecauselong-
term borrowersare getting all this priority credit.
Question:So you are sayingthat Germanyand Japan
are wrong in dependingon the bankingsystem for long-
term financing.
Mr. Johnston: Well,that is a broader issue. Yes,that
raises the whole question of how one sets out the liber-
alizationprocess.There is a differentmodel.The one we
have discussedhere is the model of liberalizingthe fi-
nancialsystemso that it takes the decisions.The Korean
and Japanese model is different. There, funds for devel-
opment were directed through the banking system. In
Japan they havehad to reform becauseof the increasein
the public sector bonds being issued. I agree there is a
fundamentalquestion of what is the appropriatedevel-
opment model.
Question:We have been talking about monetary lib-
eralizationand the accompanyinginstability.Youmade
a point that I think is quite important. Atthe same time
that a moneymarket is developed,a frameworkfor deci-
ikig within tecentral a fmus aob dece-
sion-making within
oped. Isn't this the central
something bank
that must also
basically be devel-
cannot be
gotten from a text book.One must see it operating and
then start to developthe decision-makingcapability.In-
variably,while the learning is occurringthere will be in-
stability.There is no way to avoid this; it is a cost that
must be paid for the long-term gain of having a liberal-
ized market.
Participant:The people in that market have to learn
how to take the risks.
Mr.Johnston: Part of the role of the Fund and of the
World Bank is preciselyto help speed up the learning
process.The CentralBankingDepartmentprovidesresi-
dent expertsto centralbanks, to advisethem on precise-
ly these questions.
Question:Yes,but wouldn't you say that what can be
done is to reduce the periodof instability,yet there will
inevitablybe some instability?
Mr.Johnston: Yes,I think that is right. What I have
beenpresentingis the case for trying to at least be aware
that it will happen and designingthe instruments that
46 R. Barry Johnston
can minimizethe instability.That is wherewe should re-
ally try to work.I don't think we can avoidthe instability.
Mr.Caprio:If I understand the point of the last ques-
tion, it was directed more toward fashions.It is fashion-
able now to havea treasury bill market,just as it usedto
be fashionableto have a steel plant. But no attention is
paid to whether or not the central bank should have,for
example,a forecastingdepartment.
Mr.Johnston: Yes,that is true. Everyoneis running
out to trade treasury bills but not providingthe support
for that market. That is reallywhere there is a problem.
Question:Youmentioned the desirabilityof variable-
rate instruments. In the contextof a high-inflationenvi-
ronment when your market is in the process of emerg-
ing, what is your view of instruments indexedto some
kind of price index?
Mr.Johnston: We have varying views on this. It de-
pends on whether you are in the processof a fundamen-
tal adjustment program that is intended to bring
inflationdown.
In one country that was in that situation,we suggest-
ed not to index,eventhough theyhad inflationof 50 per-
cent a month. Theideawas not to index,so as to increase
the speedof adjustment. However,the central bank did
specifyin its auctions a nominal interest rate, which at
the time was about 20 percent, as an indicator to the
market. With annual
markt.inflation
ithannalrunning att perhaps
iflaionruning prhas 2,000
2000
percent, this of course raiseda credibilityissue that had
to be addressed.The adjustment program was designed
to address that issue, and so the decisionwas not to in-
dex.
More generally,if a country is not in the position of
fundamentallyrestructuring its financialsystem-is not
intending to announcean adjustment program-in fair-
ness the only thing to do is to index.The index needs to
be keptas current as possible;lagsin updatingthe index-
ation should be reducedto lessenthe inflation risks. In-
dexation itself does not necessarilyundermine market
orientation becauseyou can still sell index instruments
belowpar and allowthe market to determinethe real in-
terest rate. Youjust index the inflation component.
Question:Howcan you introduce a competitivemar-
ket for treasury bills in situations, typical in Africa,
wherethere is little competition in the banking systems
and wherethe banks form a club or cartel.
Mr.Johnston: In some countries, starting to bring in
market instruments has, in itself, weakenedthe cartel.
In particular, opening the treasury bill auctions to the
non-bank sector providesan alternativesourceof funds.
Thosewhootherwisewoulddepositfunds with the bank-
ing system can now buy treasury bills. Therefore, the
market can help to reduce the monopolisticposition in
banks. This is particularlytrue if the banks are not pay-
ing market rates on their depositswhenthe government
starts to issue treasury bills.
Sri Lanka, for example, had a fairly monopolistic
banking system.When the central bank started to mar-
ket treasury bills aggressively,the banks lost deposits.
Nowthere is a very liberal,free,and competitiveinterest
rate structure, at least on the wholesalemarkets. This
followedfrom the aggressivepolicy of the central bank
in issuing treasury bills. So use of a money market in
this way may actually be part of one's strategy for re-
formingthe system.
As this exampleshows,financial reform is a case-by-
caseprocess.There is no generalmodel. The IMF'sexpe-
rience in dealingwith a large number of countries indi-
cates how the instruments I have discussed can be
helpful. Evenso, everycountry needsto be examinedas
an individualcase.
Mr. Honohan:In that context,a large number of the
African countries don't yet satisfy the basic precondi-
tions. Theirwholestructure is at an earlier state.
Fordexample,ing isomeAficns ctrs aren
many deposit-takinginstitutions that are not allowing
the
can'tdepositors
pay, just to withdraw
come theirweek."
back next funds. They
In thesay, "We
United
cat pay , ifabanknext couldn thUge
States of the 1930s,ifa bank said it couldn't pay,a huge
crowd of depositorsformed and began hammering on
the doors. But in Africa,though the banks concernedare
not allowing withdrawal of more than perhaps $200 a
week per account, there are no lines, becausedepositors
know they will not get any money.The depositorsjust
leave,hoping that some daysomethingwill happen.
Thoughyou may say that there is an absenceof good
collateralin the market, and bankers'acceptancesmight
providea solution, realisticallymost banks will not buy
bankers' acceptances!They won't even trust an inter-
bank loan.
Mr.Johnston: In that case, the acceptancesyou first
use to get the system started are those of whomeveris a
goodprimarycredit risk. Then it is his credit rating that
you trade. So it is not the banker's acceptancethat you
trade. For example,you might get the oil companyto is-
sue the paper,if the central bank is lending to the com-
pany.That is how you can generate some paper to trade
in a money market.

Mr.Honohan:Nobodyhas yet taken you up on what I
thought was a pretty dramaticassertion.At least I under-
stood you to say that, "For the purpose of adjustment
programming, let's not bother with aggregate credit
anymore. If we want to go down the route of indirect
monetary control, we are not going to hit those credit
targets.Wecan hit narroweraggregates,i.e., elementsof
the central bank's balance sheet. The IMF should put
into its adjustment programs that, whilethere could be
intermediate targets such as broad money,the govern-
ment is not required to hit those targets."
I have two questions about this assertion. First, will
this work for a very unstable country,and get them out
of their mess? Currently,we talk about the balance of
paymentsgap and fillingthe gap. We tell the country it
has to fill a gap of US$50million, of which this amount
will come from the World Bank, this much more will
come from the IMF,and some more from somewhere
else.The monetaryprogram is all workedout on the ba-
sis of how much credit there will be. If, instead,you de-
cide not how much credit there will be but only the
amount of central bank money,you may end up with a
residualgap.
Second, when you spoke of an intermediate target,
you mentioned broad money but not credit. Do you
think credit should be droppedaltogether?
Mr.Johnston: Onthe first question,when the levelof
the central bank is programmed,the external target is
the net foreign assets of the central bank. That is a re-
serve position.Youcan do the programmingon the cen-
tral bank's balance sheet or on the banking system.
Indeedit has beendone on the banking systemfor many
countries. But the external target is different;it is not
the overall balance of paymentsbut the net foreignas-
sets of the central bank. So it doesn't create a problem.
One can still developexactly the same flow of funds
structure. The aggregates to be included differ from
those at the aggregatelevel,but it can be done. It is not
a big problem.
Mr. Honohan:So the program is designedin terms of
foreignassets;thereforeit is still the external financing
gap that is being addressed?
Mr. Johnston: Yes.In fact, the Western Hemisphere
Departmentof the IMFhas usuallydone their program-
ming on the central bank balance sheets. The AfricaDe-
partment has tended to use the bankingsystem.The Asia
Departmenthas used both, but more on the bankingsys-
tem. Therefore,we are not necessarilymoving into new
territory in terms of programs.Rather,the use of instru-
ments changesthe waywe think about programdesign.
The Use of Monetary Policy Instruments by Developing Countries 47
With respect to your question about credit,we once
did a survey just to find out about credit. We sampled
journal articles publishedover the last fiveor so years.
There were 450 articles on the demand for money and
one article on the demand for credit. I defy anybodyto
show me a stable demand for credit function. Charles
Freedman has supported the concept of credit as an in-
termediate aggregatebut has recently retracted on the
basis that demand-for-creditis no more stable than the
moneydemandfunctions.In short, ifwe could establish
appropriateways of estimatinga demandfor credit and
we thought they were reliable,then yes, it would be a
usefulinstrument. At the moment, I am not optimistic;
but we should not stop the search for that reason.
Mr. Grenville:The thing that forced Australian au-
thorities in the direction of targeting credit was not so
much the change in which sideof the balance sheet was
beingwatched as it was the breadth of the credit aggre-
gate. We wanted something broad because of all the
churning that was occurringwithin balance sheets dur-
ing the deregulationprocess. For example,when we re-
moved our reserve requirements, that fundamentally
changed the wayM3 operated.Just a fewyears earlier,a
re-regulated system had changed it in the other direc-
tion. To try to eliminate that churning within the bal-
ance sheet of the formalfinancialsystem,wewere forced
to somethingthat gavethe very broadestrepresentation
of what was happeningto the system.That is whysome
of us did not accept that a more stable relationshipcan-
not be gotten for credit than for money.
Mr.Johnston: WhenI workedfor Charles Goodhart,I
wasaskedto estimate the disequilibriummonetarymod-
el of the U.K.economy.The ideawas that we would de-
velop a model for the disequilibrium of monetary
aggregatesthrough the counterpart equations.Mydiffi-
culty with that approachwas whywould one attempt to
generate a model for money,which throughout most of
the literature is the more stable aggregate,from all these
unstable counterparts.The counterpartsdid not seem to
me to haveany basic forecastingrelation to the genera-
tion of money.In fact, the causalityruns the other way
around.
Where I think you and I part companyis on the idea
that the counterpartsare in some sensemore stable than
the monetary function itself. I think the counterparts,
betweenone another,offseta lot of the noise in any one
of them. The monetaryaggregate,overlonger periods,is
the most stable ofthe components.That is whyit is more
useful.
Still, havingsaid that, I cannot say that M3 has been
a stableaggregatein the UnitedKingdom.It has beenre-
48 R. Barry Johnston
defineda number of times. Andthey will continue to re-
define it as the financial system evolves, because
basicallyM3tries to measure a conceptof moneythat is
difficultto measure. Where do you cut off moneyfrom
other assets?That is the fundamentalproblem,and it is
not an easilyresolvedproblem.There are a lot of criteria
for defining money,but none is satisfactory.Goingto a
narroweraggregatehelps to some extent, but it still has
these problems.
Mr.Honohan:On the issueof moneydemandstability
in the developingcountries, from my own limitedexpe-
rience, for the variouscountries forwhich I neededto do
such an analysis,it wasdifficultto get well-behaved,sta-
ble-lookingdemandfor moneyfunctionsshowinginter-
est elasticity.Even in the franc zone, which is a rather
stable area, it wasdifficult.For the most of the rest of Af-
rica, too, and in China there was a lot of instability.
Mr.Johnston: The problem is that the data are not
availableto do adequatetesting. For example,GNPseries
are at best annual and lag by a fewyears.It may be pos-
sibleto get some other aggregates,but the inflation fig-
ures are not reliable.The interest rate series may not be
particularly representative. So the immediate, major
problemto be facedis what to estimate. There is not the
richnessof informationthat can be had in the industrial
countries, in which to search for stable aggregates.
Therefore, quite often one ends up with some rough
rules of thumb about where money is goingand, hope-
fully,an adequate reviewprocess.
This increasesthe importanceof havingthe indepen-
dent central bank "captain"in place.That person has to
exercisea lot of judgment about the movementsin the
monetary aggregatesand does not havethe support of,
for instance,stable demandequations,to concludethat
interest rates should or shouldnot be raised.Hemust be
able to ask questions such as, "Whathappened?Did one
of the large importers come in yesterdayand put this
large monetary perturbation into the system?"That is
whyI feelthat the judgmentalaspectsofmonetarypolicy
need to be well developedin the developingcountries.
Question:It seemsto be more difficultto get banks to
competeon the lending sidethan on the depositside.
Mr. Johnston: Certainlyit is true that if there is no
capitalmarket, there is no substitute for bankcredit, and
that is a problem. In Indonesia, reduction of exchange
controls led to an injection of competition,which freed
up the system.So you can bring in externalcompetition;
you can allowentry of some foreignbanks.
Also, the country probablyhas a lot of state-owned
banks, over whose decisionssome moral suasion could
be exerted.In one small country, one of the big banks is
state ownedbut wasa memberof the bankingcartel. The
authorities there complainedthere was no competition.
We said, "Canyou not generate competitionyourselfby
telling the state-ownedbank to leave the cartel and set
its own rates?"Oneneedsto look institutionallyat the is-
sues.
Mr. Honohan: Under the influence of our agencies,
the central bank in Ghanadecidedto deregulateinterest
rates. They said, "Weno longer regulate the bank's poli-
cy."Nowthe Ghanacommercialbank,which is ownedby
the government and has 50 percent of the market, sets
interest rates. Deregulationin this casewasjust a trans-
fer of control from the central bank to the commercial
bank.
Mr. Johnston: That agrees with my point about the
phasingof reforms.Asanother example,Sri Lankaintro-
duced a repurchase market, but nobodyever traded re-
purchase agreements because the process was so
cumbersome. One of the adjustment program condi-
tions had been to establisha repurchase market,but this
did not mean that it wasa functionalmarket.
These examplesshow why it is important to think
more fundamentallyabout what weactuallyare trying to
accomplishin these processes.It is not just a question of
throwing a lot of monetary instruments at the system
and hoping.
 BuildingFinancialInstitutions for a Market-Based
MonetaryPolicy
Steven Grenville
Mybrief is to talk about Australia,which I shall do, but
not in great detail. Australia is an interesting case be-
cause, in a period of ten years from about 1975to 1985,
we went from a system of extreme controls to a system
that is very deregulated.The current system has floating
exchangerates, and we operate monetarypolicythrough
open market operations.
Wedo,however,havesome idiosyncrasiesin Australia.
If I were to explainexactlyhow our open market opera-
tions worked,it would resemblethe story of the bumble-
bee-aeronautical engineers tell us that it cannot
possiblyfly.Yetit does, and our system in Australia is a
bit like the bumblebee.If someonehad presentedthe cur-
rent designbeforeit wasput into practice,wewould have
read the descriptionand said there is not the slightest
hope that such a system can work. Andyet, it worksper-
fectlywellfor us. At this conference,you have heard that
differentsystemssuit differentplaces.I think that is true;
our system is fine for us, but I would not wish to impose
it on anyoneelse. It is an accident of history as much as
anything.
Therefore,I shall try to distillfrom our experiencethe
things that might have relevance elsewhere.Since you
have heard not one but two papers with a similar ap-
proach, I do not want to go through the same details
again. I shall try instead to approachsome of the same is-
sues from a differentangle and hope to provokediscus-
sion as weproceed.
The storyof that ten yearsof transition might be sum-
marized as "whatwent wrong,and howwe tried to fixit."
We did not have the opportunity for intellectual consid-
eration of the question of sequencing.We respondedto
what was happeningin the market. For instance,we did
not floatthe exchangerate in 1983becausewefinallyhad
gotten around to readingMilton Friedmanand suddenly
recognizedthat we needed a floating exchange rate to
gain control of monetary policy.Rather,we floated be-
cause the old systembroke down.Wetried to fixit two or
49
three times. It became clear that we could not fix it, so
we floated.
There may be countries where sequencingis impor-
tant. Without doubt it is an interesting issue to think
about. But our experienceis that you do what is neces-
sary to evadethis steamroller comingtowardsyou. You
do not think much about exactlywhat to do exceptto get
out of the way.
One of the lessons from our deregulationis that the
process, once started, has a momentum that was not
foreseen.In history,when one looksback, certain criti-
cal eventsmaybe seen, such as the assassinationofArch-
duke Ferdinand setting off the First World War. In
Tolstoy'sWarand Peace,the critical eventoccurs when
Kutuzov decides to move towards Borodino. Tolstoy
makesthe point that, at the time, it doesn'tseem like an
important event.The decisionhavingbeen made, events
roll on, and one doesnot havemany choices.
Another Modelfor Open Market Operations
The previousspeakershavehad the chance to set out
the big picture. WhileI know I am supposedto staywith
the little picture, the casestudy, I cannot resist havinga
go at the big picture, too.Again,the lesson to take home
from all these versionsofthe big picture is that no single
viewof the world is exactlyright. Weall haveslightlydif-
ferent perspectives.Besides,some of our actions were
based on a viewof the way systemworks. So I shall take
a fewminutes to set out my point of view.
Figure4-1 showsthe central bank'sbalance sheet and
the commercial banks' balance sheet. The diagram is
routinelyused to teach the credit multiplier,which I am
sure you remember from Economics1. On the left is the
central bank's balance sheet. Suppose we conduct an
open market operation that adds a dollar to the bank re-
serves, as recorded on the liabilityside of the central
bank's accounts, and of course, that dollar also appears
50 StevenGrenville
Figure 4.1 Sample Balance Sheets for the Central Bank and Commercial Banks
CENTRAL BANK
ASSETS LIABILITIES
foreign exchange
bank reserves (+$J)
bonds
-loans to govt. currency
-open mkt. opns. (+$J)
on the asset side of the balancesheets of the commercial
banks. On the usualviewof the credit multiplierprocess,
there is a required reserve ratio, linking deposits and
bank reserves,whichis the fulcrum holdingthe balance
sheet together. Becausewe havethat extra dollar of de-
posits, loans can increase, which holds the whole bal-
ance sheet together. The critical fulcrum is the level of
bank reserves.
Of course, in this model the extra dollar goes into
bank reserves,and this allowsthe banks to expandtheir
balance sheets. To increase loans, banks must change
the interest rates they charge on loans. To fund these
loans, the banks bid for deposit funds. So there is a
change in the monetary aggregates, as bank deposits
rise. There is also a change in interest rates. According
to this model, then,* out of the open market operation
comesa change in monetaryaggregatesand a change in
interesta hatgesin monetaryaggregatesand a changein
interestrates.
From our experience,we do not see the system oper-
ating like that. Rather, in our view the world is much
simpler.The open market operationsaffect short-term
interest rates directly.They directly affectthe cash rate,
the short end of the yieldcurve.Andthat, in a sense, be-
comes the cornerstone for the whole of the yield curve.
The holder of a 90-daybank bill considers the cost of
holding that bill to maturity and what the central bank
will do to influencethe short-term interest rate. In our
model, therefore,open market operationsaffectinterest
rates directly,which affectthe costof funds in the com-
mercialbanks' balancesheets. This in turn affectsloans,
since there is an interest elasticityon the demand for
loans.
Overall,this view ends up in much the same place as
the first view of the system. The differencesbetween
them are not worth fightingover.Still,like manydoctri-
nal issues,we can debatethem at length. It ought to be
possible to resolvethese various views simply because
balancesheets must balance.Identitiesremain identical,
so all of these differencesin modelscan be reconciled.
But the differencesdo represent different perspectives
on the mechanismsunderlyingthe balance sheet.
COMMERCIAL BANKS
ASSETS LIABILITIES
loans
deposits
bank reserves
Within the old perspective,reserve requirements act
as the critical fulcrum of the system. In our model, the
fulcrum is the payment system,the check clearing sys-
tem. We leave that check clearing system either flush
with funds or short of funds. That is howwe change in-
terest rates. Wedo not change ratesby selling bondsand
findingthat the price of bondshas changed,so therefore
the interest rate, which is the inverseof the bonds' price,
has changed. Rather,when the banks get together every
night to clear the checks written on them, we make it
more or less difficultfor them to have enough reserves.
If wewant interest rates to go up, we keepthem short of
reserves.They becomeworried and do various things to
raise interest rates.
From my perspective, then, the required reserve ratio
~~~~~~~~~~~~~is
unimportant.
uimpourtant The system could
Teqsyseme cu operate
oa perfectly
ct, well
l well
without reserve requirements. In fact, Australiaessen-
tially has no required reserveratio. There are prudential
ratios, but they certainlyare not ratios that are changed
for purposesofmonetary policy.Overtime theyhaveac-
tuallydecreased,as other systemsof prudential controls
havebeenadded,such as capitaladequacyrequirements.
The important thing, though, is that reserve require-
ments are not the fulcrum of the monetary policy sys-
tem. The system would work perfectlywell with zero
reserverequirements.
When a country sets up a monetary policy system or
when someone offerssuggestions on such systems, the
keypoint to remember is how costlyand distorting those
high reserve requirements can be. In Australiaback in
the 1950s,the banks providedabout two-thirdsof finan-
cial intermediation.Bythe early 1980s,becausereserve
requirements had been placed on the banking system,
their role in intermediation had fallen to 40 percent.
Other participants had entered and taken the action,
largely because of the various constraints imposed on
the banks. When constraints are imposedon the banks,
the action moveselsewhere.The financialsystem moves
awayfrom the controls,and the central bank finds itself
controllingjust a small pieceof the system.
As further evidenceof how bad the distortion can be,
considerwhat occurred when Australiareduced the re-

serve requirement so that it became solelya prudential
requirement.AfterSeptember 1988when the reduction
occurred,the actionmovedbackto the banks.The effect
showedin the balancesheets. In the yearafter the reduc-
tion, M3 (a bank-basedmeasure of the monetary aggre-
gate) grew by 30 percent. Any unaware outsider-a
Martianversed in the economicsof MiltonFriedman-
who came into Australia in that period in late 1988and
early 1989wouldhavethought we hada hopelesslyloose
monetary policy.M3, the aggregate we had previously
targeted, was growingat 30 percent a year.In fact,Aus-
tralia had quite tight monetary policiesat that time. The
abnormal growth in M3was simplythe effectof this in-
termediation returning to the banks' balance sheet after
we sharply reduced the reserve requirements. As it re-
turned, things which hadn't been caught in M3 were
suddenlybeing captured.
Allthis churning within the system alsoexplainswhy
we havenot been successfulin targeting anyof the mon-
etary aggregatesin Australia.But my main point is that
things like reserve requirements can be very costly if
they are not properlyset up. At present, we do not want
much in the wayof reserverequirements at all. Wehave
a system that works perfectlywell with no reserve re-
quirement.
Speakingof potential heresies, I should add that we
could operatethe systemwithout bonds,too. Wehappen
to use bonds for our open market operation becauseit is
convenient.Still, it is not at all the textbookmodel.We
are not selling bondsto change the priceof bonds so that
the inverse of the price-the interest rate on bonds-
changes. Rather,we are interested in the other "leg" of
the bond transaction. When we sell a bond, we receive
cash. It is the cash "leg"in whichwe are interested, be-
cause we want to change the state of liquidity in the
banks' check-clearing system. What happens on the
bond side ofthe transaction is not important. Whatdoes
matter is the cash that was paidfor that bond when we
sell it, and this reduces banks' liquidity, leaving them
uncomfortablyshort of funds when they come to clear
their checkswith each other. Alternatively,when webuy
a bondwe put cash into the system.
I do not want to leavethe wrong impression.Bonds
are veryconvenientto use in open market operationsbe-
cause there is no credit risk. The government issues
them; they are homogeneous.But you could operate
with anything. Youcould sell the officefurniture, ifyou
had enough officefurniture and you could do the trans-
actions. If you are selling the officefurniture and getting
cash for it, that would work perfectlywell because it is
the cash leg of the transaction that matters. Of course,
sellingthe officefurniture would be very messy,it is ob-
viouslybetter to sell pieces of paper.Even ifthose pieces
Building Financial Institutionsfor a Market-Based Monetary Policy 51
are privatepaper, that is all right from the standpoint of
controllingliquidity.The real questionconcernswheth-
er the central bank wants to hold that private paper on
its balance sheet. Is the paper secure? Is it homoge-
neous? Those are the important issues. So neither re-
serve requirements nor bonds as such are important to
our open market operations.
Nowcomesthe part that will probablybe the greatest
heresy of all. With respect to operational targets, and I
emphasizeoperationaltargets, our concernis the inter-
est rate rather than the quantityof any aggregate.Wedo
the calculations to estimate whether the system is in
surplusor deficit-or, as wesay,"up"or "down."Thates-
timate is taken into account in the dailyopen market op-
erations. Ultimately,however, we are looking at the
interest rate in a very short-term, operationalsense.
Weall knowthe reasonswhy interest rates cannot be
used as a long-term goal.First, theyare normallynomi-
nal rates, and all sorts of problemsarisein relating them
back to real rates. Secondly,there is the problem of dy-
namic instabilitythat the text books discuss.If a nomi-
nal interest rate is being targeted and nothing else,the
system is unstable.If the system receivesa shockthat in-
creasesinflation,the real interest rate decreases.There-
fore, the economygoes even faster.That is the dynamic
instabilitythat results if a nominal interest rate is the
sole long-term target. So again, do not misunderstand
me; I am not suggestinginterest rates as a long-term ul-
timate target. Rather,in practiceone is forcedoperation-
ally to use interest rates as the operationalobjective.
Of course, in the conceptualframeworkI explained,a
monetarytarget is still possiblein principle.The balance
sheet (Figure 4-1) still balances. The funding of those
loans is the monetaryaggregate.Andso, by operating on
interest rates and therebyinfluencingloans,depositsare
affected.Sincedepositsare affected,the monetaryaggre-
gate is beingaffected.So,whateverone takes as the long-
run target, it can still be envisionedwithin the frame-
work presented in the diagram. Still, operationallythe
interest rate in the short term is what one must follow.
Issuesfor the Australian Model
WhileFigure4-2 is very simple,it encompasseswhat
the Australianauthorities are trying to do, at leastat the
conceptual level. The curve SS is the supply of bank li-
quidity,whichis what wethink the central bank can con-
trol. Through the central bank's actions, SS is movedto
the left or right. By doing that, we operate on the de-
mand for bank liquidityand change the interest rate. At
the conceptual level, that is what occurs, and it raises
three issues.Thefirst is how one actuallymovesSS, the
supply of liquidity.Since that is the purpose of perform-
52 Steven Grenville
Figure 4.2 Supply and Demandfor Bank Reserves
Interest SI
rate S system, largely things that originate on the central
- - - - -i- - - -hit
i......
............... . .. . . . .<
S SI Bank
Reserves
ing open market operations, we shall spend the most
time on it, in the followingsection. Thesecond issue-
on which I shall spend very little time, although it is
both difficultand interesting-is the slope(interest elas-
ticity) ofthe demandcurve.That slopegivesyou the abil-
ity to change interest rates.
In this simple conceptualmodel, ifthe demandcurve
had no slope, interest rates could not be changed by
changingthe supplyof liquidity.There is a raft of issues
concerned with ensuring that the demand for bank re-
serves is interest-elastic.Ensuring interest elasticityis
not always easy in a system like ours, where clearing
house requirements absolutelymust be met. The inter-
est elasticityof the demandforliquidityis not so clear.A
certain amount is required, but more than that is not
needed. It is an on-offdemandwith a hair trigger. That,
however,is a story for another day.
The third issue concerns the importance of a safety
valve. In practice, operating a system like this requires
that something like a shock absorberor a safetyvalvebe
built into it. The payment system, through which the
central bank performsits operations,is hit byall sorts of
shockseveryday.The clearanceof checksbetweenbanks
is very easy-they alwaysmatch exactly.For every per-
son who has written a check, for everydebit, there is a
credit somewhere in the system. If it were only the
banks' checksthat were being cleared,the systemwould
alwaysbalance exactlyat the end of each day. But the
paymentsystem is hit by other shocksfrom outside the
bank's balance sheet. Such things include the central
bank's transactions on the foreign exchange market.
Othersoriginatein the government,becausethe govern-
ment runs a budgetdeficitand governmentdebt imping-
es on the system. So the clearing house system is being
by a great many shocks each day.Those shocksare,
in effect,movingthe SScurve left and right and affecting
interest rates. Some are rather large shocks,in fact. The
systemshould havesome safetyvalves.
We can use this diagramto show,at least at the con-
ceptuallevel,howto put safetyvalvesinto the system.In
a systemas simple as the one portrayed,the safetyvalve
would simply be to release liquidity into the system
when the interest rate reachesa levelyou want. Normal-
ly, this would be something like a rediscount rate or
lender-of-last-resortinterest rate. If the central bank as-
sesses the supply of bank liquidity incorrectly,interest
rates go up but only until this safety valve level is
reached.Topreventrates rising abovethat level,the sys-
tem would supply as much liquidity as the system de-
mands, once interest rates get to that level.
There is one final point to be made about this model.
There is a tendency to want to smooth out all the little
bumps in this system.Asthe paymentsystem is buffeted
by all sorts of shocksand those SS curvesfor the supply
of liquidity move back and forth, one may desire to get
rid of the movement.One can, of course, get rid of many
of the bumps. The government has the power to set up
institutions that remove many of them. A case can be
made, though, for not getting rid of every bump in the
market.
Central banks fall along a spectrum. At one end are
those that want to control the rate of interest in the mar-
ket exactly,everysingle day.The Bank of England is at
that end ofthe spectrum.At the other end, some central
banks are preparedto leavea fair amount of latitude, on
a day-to-daybasis, to the system. The rates are left to
bump up and down from dayto day. Leavingthe market
a certain degree of latitude helps to depoliticizeinterest
rate determination. Central banks that rigidly control
everyshort-term interest rate can be askedby politicians
to lower them, or worse still, to attempt to lower long
term rates over which they have little or no control.
In addition to providingthis political heat shield, a
degree of flexibilityalso dampens the public'sexpecta-
tion that the central bank shouldexplaineverylittle fluc-
tuation in interest rates. In Australia,wefound this quite
useful. When the rate on interbank liquidityfunds fluc-
tuated, there were not a great many persons-journal-

ists and so on-ringing us up to ask if we were
tightening monetary policy.Mostpersons acceptedthat
there were day-by-dayfluctuations. By the time it be-
came apparent that we had tightened monetary policy,
the journalists were too embarrassed, having missed
when it actually occurred,to ring us up. It was a tech-
nique for softeningthat difficulttime when you want to
raise interest rates.
Use of Open Market Operationsto Affect
MonetaryDemand
I shall now return to the first of the issues raised
above:howto actuallygo about shiftingthe supplycurve
SS through open market operations. Three things are
needed to carry out open market operations:(1)the cen-
tral bank must havecontrol of its own balancesheet; (2)
the system needs a good,professional,money market;
and (3) the commercial banks must be unshackled. I
shallexpandon each ofthese requirements individually.
GainingControlof the CentralBank's Balance Sheet
Barry Johnston and Paul Meekboth alluded to the
first requirement, control over the central bank's bal-
ance sheet. I think we haveall said, in oneway or anoth-
er, that open market operations cannot be done if the
central bank does not have control of its balancesheet.
The simple model of a central bank balance sheet in
Figure4-1 can be used to illustrate the point. In the past
in Australia,we lost control over the balance sheet in
three places, basically.The first is marked as "Foreign
Exchange"in the figure.We had a fixedexchange rate.
From the point of view of the central bank, a fixed ex-
change rate is an obligationto deal at that rate with any-
one who comes to the central bank with currency to
exchange.At the end of the day,they come to the bank
with their foreignexchange,and it must give them do-
mestic currency. These transactions increase the other
side of the central bank's balance sheet. The two items
for Bank Reservesand Currency are, of course, reserve
money.So the transactionsincrease base money.
Thus, as long as the exchangerate is fixed,there is at
least the potential for the central bank to lose control
over its balance sheet. If the exchangerate is set in the
right place, so market participants do not want to deal
with you, that is fine. That is one way of maintaining
control over the balancesheet.A pure floatingrate is not
necessary,if the market clearing rate can be predicted
exactlyevery day and if the exchangerate is set at that
point. In Indonesia,which Paul Meekdiscussedearlier,
the rate is set everyday.In my viewof the world, that is
a fixedrate; it is a fixedrate that changeseveryday. But
Building Financial Institutionsfor a Market-Based Monetary Policy 53
at the end of the day,the centralbank is left with this ob-
ligation to dealwith all comers. This obligationworries
those of us who want the central bank to gain control
over its balance sheet. The fixedexchangerate entailsan
obligation;the central bank has no choice but to deal
with people who either want to exchange foreign ex-
change for localcurrency or vice versa.
Foreign currency, then, can become a window
through which a central bank may losecontrol over its
balance sheet.Anothersuch windowmayoccur through
the seconditem on the assetside ofFigure4-1. Although
this item is labeled"Bonds," it includes all means of
fundingthe governmentdeficit.Supposethat, when the
government runs a deficit, it funds that deficit not by
selling bondsto the public but by sellingbonds (or some
piece of paper like a bond) to the central bank. In the
process,reservemoney increases.
To close this second window,the central bank must
strike some sort of deal with the government on what
will be done with the budget deficit and how it is to be
funded. Bynow,Australia is fortunatelyin the position
where there is a verypure separationofgovernmentdebt
policyfrom monetarypolicy.This means,of course, that
the government has undertaken to fund deficitsby sell-
ing government bonds or other securities to the public
and willnot ask the central bank to fund it.
Not every country can afford that luxury, and Paul
Meekreminded us earlierthat it is not something every
central bank should absolutelydemand. It may be very
nice to operate in a system where the central bank need
never worry about the government'sdeficit messingup
its open market operations,becausethe deficitis always
fullyfunded by selling bonds.
Althoughthat is the idealsituation, in Australiaa sys-
tem that lacked that commitment until about 1982
workedreasonablywell. Theyhad a few differentmeth-
ods of selling bonds: these methods all had in common
that they never sold enough bonds to fund the govern-
ment deficit.This meant that the first part of the central
bank bond sales,which lookedlike open market opera-
tions, were reallydebt managementin disguise.Wesim-
ply could not leavethat liquidityfrom unfunded budget
deficits in the system, so the first requirement of our
open market operationswasto sell bondsto soakup that
liquidity.
That kind of systemworks all right, too, but it is a bit
messy.In our case, it meant that bonds were being sold
by the treasury to the public-a rather reluctant pub-
lic-at a certain price X.For the most part, the purchas-
ers were captivebondholders,who were in fact forcedto
buy our bonds. That necessarilyconstrained our open
market operations, because having sold government
debt to these poor captivesat price X, it was difficultto
54 Steven Grenville
sell it to the public in the form of open market opera-
tions at a more attractiveyield.Wediddo a bit ofthat but
it was difficult.
We are nowin a much better systemwith full separa-
tion of debtpolicyfrom monetarypolicy.But ifyou can't
get perfection at once, you work with whatever con-
straints you have.
The third window for loss of balance sheet control
does not showvery clearlyin Figure 4-1. It concernsthe
rediscount facility.Especiallywhen one is setting up a
market for government securities, the tendency is to
want to providerediscountfacilities.Onewayto provide
liquidity to that market is to tell potential purchasers
that if they buy these government securities and subse-
quentlywant to sell them back, they can do so, perhaps
at a differentprice.Yet,wheneverthe central bank has a
commitment of that nature, it has lost control over its
balnce
In Aushtratoa
Australia,weehadext fairly
fairlygent. ous rediscount,
generous rediscount, lend
lend-
er-of-last-resort,facilitieswhen we were trying to build
up a government securities market. Over time we have
made those facilitiesless and less attractive.In terms of
Figure 4-2, we have pushed the line for the rediscount
rate far abovewherewewant interest rates to be. In oth-
er words,we prefer to control Interest rates by our open
market operations,rather than by the public tappingthe
safety valve of the rediscount rate to get their extra
funds. If peoplewere consistently coming to our redis-
count windowfor significantsums of money,after sever-
al tose
dys o ntha,chrge f te opn maket
al days of that, those in charge of the open market
operationswouldhavetorunontotheirswords.Itwould
be regarded as a failure.
We would not want to do that becauseinterest rates
would once again be controlled.Wewould be back to a
form of the old system; the rediscountrate would in ef-
fect become an interest rate that we were imposingon
the system.Andone of the reasonsfor going to a system
of open market operationsis to get awayfrom setting an
interest rate. Wewant to maintain a system where the
market has a big hand in setting interest rates. So we do
not want that rediscountsafetyvalveto be operatingvery
often. It is there in casesomebodymakesa mistake,ifan
individual bank or financial institution needs it. But it
should be seen as a safetyvalveand not as a normal way
of setting interest rates.
Allthese things must be done to get control over the
central bank's balance sheet. Ideallythere should be a
floating exchange rate, which closes the Foreign Ex-
change window.Ideally debt policy is completelysepa-
rated from monetarypolicy,which closesthe windowof
loans to the government.And ideallythe rediscountfa-
cility is only used for trivial sums of money.All these
things are designedto do one thing: to give the central
bank control over its balance sheet.
FosteringA Strong and ProfessionalMoneyMarket
Thesecond major requirement foropen market oper-
ations is to havea market in which to operate.A country
may be able to get by with something pieced together
and held in place with bits of wire, so to speak.But the
central bank cannot really carry out good open market
operationsuntil there is what I would characterizeas a
professionalmoneymarket.
An interesting example came up in the course of a
previous floor discussion. Suppose the country is run-
ning a budget deficit and sold instruments to house-
holds, for instance, to finance that deficit.As a case in
point, Australiahad a very successfulhousehold instru-
ment, which was called an Australian Saving Bond.
When we set its interest rate at the right level, house-
holds-meaning small investors-bought these like
hot-cakes.Thatwas the waywefunded our budgetdeficit
when it started to blowout in the middle of the 1970s.It
wasa verysuccessfulformof fundingfora budgetdeficit.
Whynot make instruments like these AustralianSav-
ings Bondsthe basis of the central bank's open market
operations?The instruments are out there; why not buy
and sell those? The argument against doing so is that
this instrument went into households.In fact,it wasgiv-
en the special characteristic of a constant principalval-
e, bae weallowedithe o b redscont at any
ue, because we allowed them to be rediscounted at any
stage at their facevalue. For that reason, they would not
have made a good instrument for open market opera-
tions. But even if these bonds had not had that charac-
teristic, by using them as the basis of open market
operations we would have been imposingcapital gains
and losseson households.
Nowit is possibleto operate on that basis,but I think
it makes it difficult;it is a constraint on open market op-
erations. If you are planning an open market operation
that will imposea capital loss on households,then you
will probablyencounter constraints. It is much better to
be operating in a deep professionalmarket, where big
playersare operatingwho understand there are risks of
capital gains and losses,so they cannot complainwhen
you impose capital losses on them. That is the sort of
market required for open market operations.
One fortunateside effectof our long history of budget
deficitsin Australiawas that over the years we became
goodat sellinggovernmentsecurities.Wehad a well-de-
velopedmarket in governmentsecurities long beforewe
decidedto do any open market operations.The market
was not in fact deepenough to sell all the bonds needed
to fund the budget deficit,but it was a well-developed

market. It had been operating since 1960 or so; open
market operationsonly started about 20 years later. We
were lucky,then, in that we coulddo it in two stages. We
developed the government securities market in the
1960sand 1970s.Andthen in the early 1980s,westarted
to use that market for open market operations.
Perhaps ten years ago, the buzz words for a sound
market were "width,depth, and resilience."Whatshould
be done to encourage a market with those characteris-
tics? Let me start with somethings one should either not
do or do onlywith great reluctance.The first, to which I
have alreadyalluded, is to have captiveholders of gov-
ernment securities. Captiveholders are parties-per-
sons or institutions-who in one way or another are
forced to hold the government securities that you are
selling. Youmay put honey around this instrument to
make it attractive for them, but you get into trouble
whenyou rely on captiveholdersto take all your govern-
ment securities.
In Australiawe relied on the savingsbanks and the
life insurance companiesin particular.We gave the life
insurance companiesa tax break if they took a certain
proportion of government securities. The problem was
that once those securities were on their balance sheet,
they would not sell them again. We had sold the securi-
ties to them at the wrong price; if they sold the securi-
ties, they wouldactuallytake a loss in their accounts.So
they held them to maturity.Of coursethey actuallytook
a loss on day one when they bought the securities from
us. But they did not haveto crystalizethat loss as a cap-
ital lossat all.The waytheyavoidedhavingthe lossshow
up was to hold them to maturity.
Figure 4.3 Holders of Commonwealth Government Securities (End-June Figures)
i Other EM Otherprivate
Public bodies financial institutions
$B
50
40-
30
20 -4
10 . 20
1977 1981 1985 1989
Building Financial Institutionsfor a Market-Based Monetary Policy 55
If your intention is to developa secondarymarket in
government securities, then you do not want parties
holding the government securities to maturity.A great
temptation when trying to developthe government se-
curities market is to find a few readyvictims,a fewcap-
tives, and load them up with government securities.
That doesthe job ifyou just want to fund the budget def-
icit. It does not work if you want to use that market to
carry out open market operations.So we got rid of those
captiveholders.
Figure4-3 illustrateshowimportantthe captivehold-
ers were. In addition,the big holdersin those figuresare
the banks. Tosomeextentthey werecaptiveholdersalso.
For instance, we made the savings banks hold govern-
ment securities as their major asset. Of course, there
were also prudential reasonsfor requiring the banks to
hold government securities.So there wassome sense in
which banks should be made to hold governmentsecu-
rities. The relativeimportanceofthe prudential role and
the deficit funding role cannot be clearly separated.
However,the effectsare markedlydemonstratedby the
life insurance companies. As soon as we unshackled
them, they stopped holding government securities. As
everyoneelse did in the 1970s,Australia had negative
real interest rates. Weburned those captiveholdersbad-
ly at that time. Theywill not be easilyburned again, and
their balancesheets now reflectthat.
In summary,one must get past the stage of captive
holders as quicklyas possible.Open market operations
cannot be effectivein a market dominatedby captives.
Anotherthing not to do is to givetax breaks.There is
a great temptation to use tax breaks to encourage mar-
[= Life assuranceoffices = OtherBanks
E3 Autthori ed MM dealers _ Reserve Bank
%
100o
80-
60-
X/I~~~~~~~~~
-
1977 1981 1985 1989
56 Steven Grenville
kets, but there are disadvantagesin doing so. They seg-
ment the market betweenthose who can take advantage
of them and those who cannot. They distort secondary
market activitiesif preferentialtax treatment is not ac-
cordedto all securities.They are inequitablein terms of
income distribution.They are a hiddenindirect subsidy,
which does not appear on the balance sheet and whose
effectivenesstherefore cannot be assessed.
More legitimate methods are availableto foster a
market with width, depth, and resilience.Anythingthat
encourages secondary markets is a good idea. Some-
times it helps if the government,the authorities, are ac-
tive in trading in the market. But, as Paul Meekpointed
out earlier,this must be done carefully.Do not take all
the businessawayfrom those whose interest is to earn a
profit by trading in the market. If the authorities do all
the job, nothing maybe left for privatetraders.Abalance
must be struck.
In Australia,for instance,we deal in the government
bond market almost every day for reasons unrelated to
our open market operations.Eachday,more or less,two
separate operationsare done. The open market opera-
tion occursearly in the morning. Later,towardsmidday,
we do some dealingin the bond market. Nowthat's just
designedto help the bond market along. However,ifthe
central bank dominatesthat market, then that can hurt
the privatetraders whom you want to develop.
A rediscount facilityis another possibility.I have al-
ready presented the case against an over-activeredis-
count facilityas a threat to control of the central bank's
balancesheet.
Yetanother possibilityis derivativemarkets. In Aus-
tralia, the introductionof a ten-yearfutures contract for
bonds gavea great boost to the bond market.The central
bank had nothing to do with its introduction. It was a
private initiative that started because the market was
there. Still, it certainly has increasedthe liquidityof the
bond market.
The central bank may be able to encourageinstitu-
tions in the market to repackagegovernment paper.For
example,the holder ofa largeblock ofgovernmentsecu-
rities may hold that on one side of its balance sheet,
while repackagingit into smaller instruments to be sold
on the retail market. Youcan encouragethat.
Trivialas the point mayseem,it is nonethelessimpor-
tant to determine the right kind of securities and the
right length of the issues. Issues to be decided on in-
clude:should the instruments carry capitalrisk or be re-
deemableon demand at face value, be bearer bonds or
inscribed,what should be the maximumholding period,
shouldthere be dealersor just brokers,who are the mar-
ket makers, and what is the size, maturity and target
market for each issue. The authorities also must decide
how much informationshould be given to the public on
the central bank's portfolioand on estimates of govern-
ment funding requirement over the longer term. Lastly,
if securities are being sold by tender, the authorities
must decide:
* Howmuch informationis providedto tenderers
* Administrativedetailssuch as settlement period
* Safeguardsto ensure that bidders actuallyfulfill
their bids
* Who can participatein the tender
* Whether the tender is the solemethod of setting
bonds,or is there a limited"tap" as well
* Howthe authorities participate in the tender
* Whether there is a reserve price (i.e., maximum
acceptableyield)
* Whether to settle the tender at the differential
yieldsthat tenderers havebid or to settle it by the
"Dutch auction"system of a uniform price
* Whether non-competitivebids are accepted (i.e.,
bidsfor a limited quantitythat willbe allocatedat,
say,the averageprice of the competitivebids)
* Howto ensure that the tender is covered.(Theis-
sue here is whether to use a select group whith,
in return for some special concessions,ensures
that the tender is fullycovered.)
When we first started selling bonds in Australia,we
had an implicit understanding with the market that
bondswith the same characteristicswouldnot be issued
twice. If we put out 15-yearbonds due to mature on a
particular date with a particular yield,wewould not is-
sue bondswith exactlythose characteristicsagain. I sup-
pose the reasoning was that the market should know
exactlyhowmuch ofa securitywouldbe placed.Theydid
not want the risk that, after theyhad bought from one is-
sue of securities, the central bank would subsequently
floodthe market with securitieshaving exactlythe same
characteristics.
Atthe start, the market seemedto want a varietyof se-
curities.At one stage, there were over a hundred differ-
ent government securities issued with different yields
and differentmaturities. Somewere very popular; some
were not. Figure 4-4 showswhat the yield curve looked
like. The dotted line with all the spikes is a yieldcurve,
although no one would recognize it as such from the
textbook descriptionof a yield curve, which is supposed
to be smooth. That did not happen with instruments
such aswe had on issue,some ofwhichwere liquidwhile
others were not. Thosewith a ten-yearmaturity that had
the advantageof correspondingto the ten-year bond fu-
tures were very liquid. They are representedby the two
downwardspikes in the figure. Other issues, which in
fact had similar characteristics,neverthelesshad vastly
differentyields.

Figure 4.4 TreasuryBond Yield Curve
15.00
14.50
14.00 .- * .,2. ,.***,,g,,,+, ,.,
13.50 _ ' '' ' .',',,
13.00 _ ' , , 31-Mar-87
12.50
12.00 l l l l l l
Jun-90 Jun-91 Jun-92 Jun-93 Jun-94 Jun-95 Jun-96 Jun-97 Jun-98 Jun-99 Jun-00 Jun-01 Jun-02 Jun-03 Jun-04 Jun-OS
Bygivingthat kind of yieldcurve, the market was in
effecttelling us that it did not want the degreeof product
differentiationwe thought it wanted. Instead, it wanted
the product to be uniform. In fact, the liquidity was
greatestwhen most of the market was ten-year bonds.
Thatwaswhat the market wanted;the more ofthose that
were put out there, the better.That wasthe wayto get li-
quidity.
So, trivial as it maysee, one must look at what is hap-
pening in the market all the time. Whatsuits the market
at one stage may not suit it later.Asthe market develops,
its requirementsmight becomedifferent.When wereal-
ized what the market was telling us, sometime around
March 1987, we started consolidation operations. We
would buy backthe unpopular securities and sell more
of the "hot" (i.e., popular) ones, which we got the gov-
ernment to issue. We did more than a billion dollars
worth of these highlyprofitabledeals,just quietlydoing
some everyday.No one realizedwhat washappening;the
yield curve stayed unchanged.Graduallythe market re-
alizedwhat washappening,and the yieldcurve did what
the textbookssay it should. It became more or less a
straight line. The solid line in Figure4-4 shows the yield
curve several years later, for May4, 1990.In fact, the
market caught on to us much sooner than that and flat-
tened out the yieldcurve.
In summary,this exampleillustratesthat the actions
neededto encourageliquidityin the governmentsecuri-
ties market may be roundabout and may change over
time. The important thing is to stay in touch with the
Building Financial Institutionsfor a Market-Based Monetary Policy 57
/ >
f , , *
l ll l
market at all times. Read the market's signals,and take
advantageof the opportunitiesthat arise.
The critical issue for achieving liquidityin the sec-
ondarymarket is to get the pricingright. The priceof the
issuedinstrument must be a market price.Howdoes one
determine a market price?I shall be dogmatichere and
say a tender is necessary.
As an intermediatesystem beforewewent to the full
tender,we tried a tap system. Astock of governmentob-
ligationswas made availableat specifiedterms. If it sold
well, which indicatedthat it had been correctlypriced,
we kept it at that price. If the pricewas wrong in either
direction,that issuewould be closedoffand another put
in its place.Table4-1 lists all the stocksthat were issued
during the period of about 20 months when the tap sys-
tem was operating.
Wethought this tap system might work well because
it enabledus to take a stock offthe market if it was either
unpopular or too popular; that is, if wehad mispricedit.
In that period of 20 months, we issued 22 different
stocks. We worked hard to get issues on the market at
the right price. Yet, I must say that I believewe failed.
For example,although weintendedto sell $A500million
of stock 7, wesold $A12million. Stock16went the other
way.We intended to sell $A300and sold $A352million.
As the table shows, it was extremelydifficultto get the
price right, even though we were issuing new stock
wheneverwe thought we had gotten the previousprice
wrong. Even with all that effort,we could not make it
work.
58 StevenGrenville

Table 4-1 Tap Stock Sales

AMOUNT ($MILLION) TIMING
INDICATED SALES MATURITY OPENING CLOSING
1 500 237 Apr 82 30.4.80 27.6.80
2 250 263 Apr 85 30.4.80 9.5.80
3 400 281 May86 2.6.80 6.8.80
4 500 293 Nov82 7.7.82 5.8.80
5 250 59 June 90 7.7.80 10.12.80
6 500 145 Feb 83 21.8.80 6.2.81
7 500 12 Nov84 21.8.80 7.11.80
8 300 224 May82 14.11.80 21.11.80
9 300 20 Sept 82 26.11.80 6.3.81
10 150 77 Dec87 19.12.80 9.6.81
11 400 253 May83 9.2.81 8.5.81
12 400 47 July 84 9.3.81 9.7.81
13 500 383 Apr83 11.5.81 25.9.81
14 150 74 Aug88 10.6.81 6.11.81
15 500 539 June 84 10.7.81 23.10.81
16 300 352 June83 28.9.81 23.10.81
17 500 178 Apr84 28.10.81 5.4.82
18 500 544 June85 28.10.81 13.11.81
19 150 19 Apr88 9.11.81 5.4.82
20 400 89 Nov85 19.11.81 7.5.82
21 400 287 Feb 84 14.5.82 16.7.82
22 400 128 Aug85 14.5.82 16.7.82
*UntilTapStock 13excludessubscriptionsbythe RBAand LCIR.Thereafter,RBAsubscriptionsonlyareexcluded.

In 1982we went to a tender.That seems to me to be bondmarket. Within a yearafter wewent to a tender sys-
crucial in getting the price right. In turn, getting the tem, the market liquidity began increasing;it just grew
price right is crucial to ensuring an active and liquid like wildfire. If you get the pricing right a lot of other
market. Figure 4-5 showsaverage dailyturnover in the things will look after themselves.

Figure 4.5 Bond Market Average Daily Turnover Unshacklingthe BankingSystem

$ billion With respect to the requirement to unshackle the
400 - commercialbanks,most countries start with their banks
surrounded by constraints. If open market operations
350- are to be effective,all those shacklesmust come off.
The simple reason is that the commercialbanks are
300- the conduit, the core of the financialsystem. Whycarry
250- out open market operations,if theyare not being used to
affectthe banking sector,whichin turn passesthe effects
200- on to the rest of the economy?The ultimate aim of car-
rying out open market operations is not merely to
150-o change interest rates in the short term cash market but
to let those rate changes act as a basicbuilding blockof
100- the cost of funds, which a banking system or financial
system then passeson to its customers. Therefore,un-
50 - less the banks are freed from constraints on interest
rates and on their balance sheets, open market opera-
0y tions will not be effective.
75 177 1 79 81 83 85 87 89
years ended June
 Building Financial Institutionsfor a Market-Based Monetary Policy 59

It used to be that way in Australia;the banks' balance
sheets were tied so rigidly that they could not move
without comingto the central bank for permission.They
were not ableto managethe liabilitysideof their balance
sheet to match the asset side. Theywere passivedeposit
takers; they took whatever the customer brought to
them in depositsand loaned them out. Becausethe sys-
tem was very bumpy, the banks had to hold large re-
serves. Figure 4-6 characterizes their balance sheet
management.Until the early 1980s,they needed to hold
large excess reserves. Those reserves changed a great
deal during the year.Alltheir effortswent into managing
that intra-year liquidity flow.The whole of their effort
wasaimed merely at trying to figure out what was likely
to happenand how to respond.
Now that the banks have been unshackled,they can
enter the short-term money market to get whateverde-
posits they need. They operatewith much less excessli-
quidity. With that off their minds, they have time to
think about other things than how to maintain their in-
tra-year liquidity.They can get on with the job that
banks are supposedto be doing,which is to decidewho
the good customers are and lend out the money they
take in to customers whowill pay them back.
Other changesin the banking systemalsooccurred in
response to the removalof constraints. Figure4-7 com-
pares cash rates in 1983and 1985.Underthe old system,
cash rates bounced around a lot; they smoothed out
when the shackles came off. As Figure 4-7 shows, the
changesalso smoothedthe intra-yearpattern of interest
rate swings.
Figure 4.6 Liquid Assets in Excess of Minimum
Requirements, as a Percentage of
% Deposits %
10 10
4
8 8
6 5 quarter weighted 6
moving average
4 4
2 2
o o
1975 1977 1979 1981 1983 1985 1987 1989
We are now using our liquidityoperationsto smooth
out these rate variations. The point of doing so is not
simplyto have a flat line instead of a wobblyline. The
major benefit gainedfrom all this is that the attention of
persons in the financial sector could move away from
concernwith havingenough liquidityeach dayand turn
to the real businessof intermediation:the work ofbring-
ing in moneydepositsand lending it out productivelyto
borrowerswho will be able to repay it with interest.

Figure 4.7 Cash Rates 1983 and 1985, Weighted Average Paid by Aughorized Dealers on Clients Loans
Outstanding
1983 1985

20- 20 - 20
18- - 18 - - 18
16- 16- -16
14- 14 -- 14
12- - ~~~~~~~~~12
-- 12
10 - ~~~~~~~~~10 10

8- ~~~~~~~8- 8
6- , l,,, ,, ,, |e|lllrlllll|slulllllllllll6 6
4- ~~~~~~~~~4- 4

Auguist September October Auigust September October 2

60 Steven Grenville
Conclusions
Let me say that, while this is howwe did it in Austra-
lia, imitatingus might not be the best way.In particular,
open market operationsare not necessarilythe right way
to go for every situation. It was long ago that the old
preachersaidthere is a time for everything.I believethat
appliesas wellto open market operations.Asa country's
financial system develops,the central bank needs more
sophistication in its operations. In a sense, the two
things go hand in hand; if a very unsophisticatedsystem
of central bank operationsis left in placewhile the finan-
cial market is becomingmore sophisticated,the market
will roll over that system, leavingthe authorities unable
to achievetheir objectives.Increasedsophisticationnot
only createsthe needto changethe systemof operations,
it also creates the environmentin whichthe new system
can succeed.It is a matter of choosingthe right time, as-
certainingthe appropriatemoment for the change.
The progress in Australiawas, in my view,paced by
developmentsin the market. Amongthe critical events,
our financialintegration with the rest of the world dur-
ing the 1970s was extremely important. It forced us
downthis path. But oncestarted downthat path, we had
no choice.
I must say I do not think the current system answers
all of our problems.Thecurrent system has the potential
to let us control the monetaryaggregateswith a high de-
gree of precision.That is what we wanted;it is why we
put the systemin place.Yet,nowthat we havethe ability,
we are not sure we want to use it. We have much less
confidencenow that we can steer our system using an
intermediate goal. To be more specific,in Australiawe
haveabandonedthe idea,at least for the moment, of an
intermediatetarget.Weshoot straight for someultimate
targets.'
Therefore,evenwhenone has put one ofthese sophis-
ticated systemsin place,they don't solveall of the prob-
lems. Centralbankerswill still be left with a great many
difficultissueson howto run monetary policy.
Note
1. Editor's note: The interventions by Lindsey and
Goodhartin the next section elaborateon this point.

Discussion
Question:I have a question regarding the stabilityof
demand for reserves. Your graph makes it look easier
than it is in practice. Yousaidthat the supplyof bank re-
serves should be controlled, but in fact the demand for
money,the demandfor reserves,can be extremelyunsta-
ble. In Colombia,many of the shocksto the system,par-
ticularly the pressures coming from the Treasury,such
as withdrawing money by collecting taxes or abrupt
changes in expenditures,will shift this demand for re-
serves drastically.This introduces a tremendous insta-
bility in the interest rates. What has been your
experienceof the demandforreserves?Howstable has it
been during the process?
Mr. Grenville:The demand for reservesin our banks
is fairlywelldefinedand relativelystable.The big chang-
es come on the supplyside.In terms of Figure4-2, all the
action is on the supplyof reserves,which is hit by huge
shocks everyday.The success of our more recent policy
is that we havebecomebetter at offsettingthose shocks.
For example,government taxes used to fall due at one
time of the year.We have persuadedthe government to
spread those out during the year.We now have much
more information on what shocks are hitting the sys-
tem, so we have gotten better at capturing the changes
in supply.The results, as Figure 4-7 shows, have been
that interest rates, which were veryjumpy on a day-to-
day basisand showeda clear seasonaltrend, are now sta-
ble.
Nothing was done to the demand curve, which is in
fact quite stable. One reason it is so stable is that the
banks havefaith in our ability to maintain adequatere-
servesin the system.The short answerto your question,
then, is that the demand has not beena problem for us,
the supplywas what we needed to stabilize.
Question:You stressed that the operational focus is
the interest rate, not in a long-term sense but rather in
a short-term sense. Since your ultimate objectivesmay
be somethinglike the price levelor the levelof nominal
income, I suppose that your statement means that you
change the short-term target as circumstances vary.
Couldyou giveus some ideaof how you changeyour op-
erational targets as a function of the ultimate variables
that you want to achieve?Perhapsyou could get into the
intermediateaimsand the ultimate aimsofthe exercise.
Mr. Grenville:That would really require another two
hours, but I will answer briefly.From 1975to 1985we
searched for an intermediate target. In those days, we
Building Financial Institutionsfor a Market-Based Monetary Policy 61
had M3as a target. In the 1970s,we usuallymissed it but
not by much. Then in the 1980swhen the systemwasde-
regulated,we missedit by more and more. Wefelt it was
becoming less relevant because it focusedon the banks
while other financial institutions were increasinglyim-
portant. In January 1985,we abandonedtargeting of M3.
We got out of the wayjust in time; we had targeted 10
percent growth in M3 for that year,but it turned out to
be 18percent. Wewould havelost all our credibilityifwe
had been that far offour declaredtarget. Yet,and this is
the keypoint, monetarypolicyat that time was tight. No
one saidwehad missedour target becausemonetarypol-
icy was loose. All the churning on the balance sheets,
with depositsmovingout to thrifts and then back,led us
out to broader and broader aggregatesin the attempt to
nullify those intra-institutional changes. So we went to
broad money.But that was not quite broad enough be-
cause it excludedbank bills,whichwe found were being
used a gooddeal in intermediation.
That led us to the broadest of the aggregates,which
we call credit. It does not matter that it comesfrom the
other sideof the balancesheet, it can be liabilitiesjust as
wellas credits,so long as it capturesthe total amount of
intermediation.But it turned out that aggregatewas not
good,either.In the mid 1980sthe credit aggregategrew
at 20 to 25 percent a year for the fiveyears from 1984to
1989.At this time, inflation was decliningand nominal
incomeaggregateswere growing at around 12percent.
To make a long story short, we do not think there is
any intermediate target that will serve for us. Like all
goodcentral banks,we are concernedwith inflation and
price stability.So we are interested in forward-looking
indicatorsof inflation:howfast is the economyrunning;
howtight is the labor market; what is happeningto price
expectations?We try to read the last of these from the
shape of the yield curve.
Whenwe think those indicatorsare changing,we ad-
just interest rates because of our general view that the
directionof causal transmission is from interest rates to
economicactivity.Wehavedone extensiveGrangertests
but have not been able to find that connection directly.
Althoughwe cannot prove the causal direction, we ac-
cept the common sense view that, if interest rates are
raised enough, projects become unprofitable, and
growth of activitydecreases.
That does in fact happen, but the calibrationof the
transmission mechanism is extremelypoor. For exam-
ple, in Octoberof 1987we realizedthat monetary policy
needed to be eased. Starting in April 1988,we realized
the economy was running too fast. So we started to
62 Steven Grenville
tighten, and wetightened again and again. Interest rates
rose progressively.The treasury bill rate, for example,
climbed from 11 percent to 18 percent over a period of
six to nine months. At 18 percent, the real rate of inter-
est was around 10percent, whichwe reckonedwas a suf-
ficiently tight monetary policy.The economy still was
growingat 10percent in real terms; employmentwasris-
ing at 4 percent annually.So,although we think weknow
the transmission mechanism,the lagson it are extreme-
ly long and the calibration extremely poor. Eighteen
months after we first began tightening, the economy
slowed,and that should eventuallybring down the infla-
tion rate.
Question:Do capital flowsfrom outside affectthe lag
in the transmission mechanism?
Mr. Grenville:That should act as a second lever for
transmission.As interest rates rise, externalcapitalwants
to come into Australia,but it cannot comein becausewe
havea floatingexchangerate. On an adjusted basis,there
should be no movement.
Mr. Honohan: So the exchange rate increases,which
makesyour exports lesscompetitive,whichalsoslowsac-
tivity?
Mr.Grenville:Yes,so the transmissionmechanismnot
only adjusts interest rates for the activity but also may
leadto a higher exchangerate. The exchangerate went up
sharplyfrom $AO.57per U.S.dollar in the middle of 1988
to $AO.87severalyears later. Still, economicactivitydid
not begin to slowuntil eighteen months after the tight-
ening started.
 PART II

MONETARYTARGETINGAND CONTROL
 Introduction
Movingaway from bank-by-bankcredit ceilings raises
the questionof appropriateintermediatetargets for cen-
tral bank policy. The presentations and discussions in
this section focusedon the extensivelessons of the in-
dustrial countries with targeting issues and their rele- rates are the real operating target, then there is little
vance for developingcountries. DavidLindsey,Deputy
Directorof the Divisionof MonetaryAffairsat the Board
of Governorsof the FederalReserveSystem,provideda
comprehensive insider's view of the difficulties in
achievingmonetary control in a periodof rapid financial
change and of the Fed's search for stable money and
credit relationshipswith real activity.In Europeancoun-
tries, too, as Charles Goodhartreported, various mone-
tary aggregates,credit,interest rates, and exchangerates Theseforcesrecentlyhavebeen at work in the European
haveall been consideredas potential targets. (Goodhart,
who is now Norman SosnowProfessorof Lending and
Financingat the LondonSchoolof Economics,was for-
merly Adviserat the Bankof England.)But, regardlessof
the intermediate target, central bankers keep theirGoodhart
closely on the ultimate targets of inflation and output
and, even in formerly more monetarist central banks,
have movedto an eclecticview of monetary policy.An
important lesson of industrial country experienceis that
there is no guaranteed method for attaining monetary
policytargets (direct methods included).Officialsmust
avoid regarding econometricrelationshipswith uncon-
ditionalbeliefand insteadbe preparedto adjust to behav-
ioral changes associatedwith financialreform.
In many countries with a floatingexchangerate, in-
terest rates havebecomethe more important intermedi-
ate target, as they were before the rise of monetary
targeting. But the authorities are now more alert to the
long and variable lags of monetary policy. Professor
Goodhart explainshow lags probablycontributed to the
difficulty of targeting money aggregates in the early
1980s.
Questionsarose concerningthe use and designof re-
serve requirements. Severalcountries-such as Canada
and the United Kingdom-have operated with no re-
serve requirements.Lindseynoted that reserverequire-
ments help implement monetary policy when some
aggregatereservesmeasure is being targeted; if interest
monetary control function for required reserves.When
reserve requirements are employedwith an aggregate
reservestarget, it is helpful to keepthem uniform; oth-
erwise the money multiplier can be altered by shifts
among differentinstitutions.
In countries participating in fixed exchangerate ar-
rangements, national monetary policybecomes ineffec-
tive as capital mobility and asset substitutability rise.
MonetarySystem(EMS),with the dissolutionof barriers
to capital flowsand the rising substitution among EMS
currencies. Not unlike the experienceof the Southern
Cone countries in South America a decade ago,
described how,as the exchangerate peg has
become credible,members of the "European Southern
Cone" have been experiencing inflows of capital, deteri-
..
oratig current account positions, relatvely highInfla-
tion rates, and appreciatingreal exchangerates. In this
situation, it is importantthat governmentsbe ableto use
fiscal policy to restrain inflationary pressures, since
monetary policyhas been sacrificedto maintain the ex-
change rate peg.
Governments in several countries are recognizing
that monetary policy deliberationsshould be insulated
from political pressures and that therefore it is impor-
tant to increasethe independenceof centralbanks. Aso-
ciety can choose to peg its currency to one that is
managedby an independentand firmlyanti-inflationary
monetary authority, or to erect legalbarriers around its
owncentralbank.A discussionensued regardingthe dif-
ficultyin establishing a proper incentive structure for
central bankers; some feared that monetary incentives
65
66 Introduction to Part 11
favoringonly low inflation might bring about such a re-
sult at a high price.
WhileLindseyand Goodhart describemonetary poli-
cy practicein industrial countries,DonMathieson(Chief
of the FinancialStudiesDivisionin the ResearchDepart-
ment of the IMF)concentrates on developingcountries
in asking why there has been a traditional reliance on
credit controls and whytheyare coiningunder pressure.
Exogenousshocks tend to be relativelylarger for many
developingcountries than for industrial countries. The
former do not have the depth of financialmarkets to ab-
sorb and spreadout the effectof such shocks.It is partly
as a result of this that developingcountries rely on "fi-
nancing"large losses,causedby such economicshocks,
through the centralbank's balancesheet, and thus effec-
tively through the inflation tax and other forms of re-
pression ofthe financialsystem.Ratherthan allowcrises
to lead to greater instabilitythrough their effecton sky-
rocketing interest rates, the authorities in developing
countries have had recourse to direct controls. Mathie-
son adopts the pessimisticview that such recourse, in
times of crisis,willcontinue to be a featureof developing
country policy.
Dr.Mathiesonalsodescribesthe macroeconomicand
structural preconditions for successful transition to
market-basedmethods of monetarycontrol. One maybe
allowedthe optimisticopinion that careful attention in
advanceto ensuring that these preconditionsare satis-
fied (includingthe changes in net credit demandnoted
by Johnston in his paper) would minimize the need for
policyreversals.
Centralbanks can only hope to be independentwithin
a government, rather than independentof the govern-
ment. When shocksare sufficientlylarge,be it a stabili-
zationprogram in a LatinAmericancountry or a driveto
currency union in Europe,a central bank will ultimately
haveto makesome compromiseor see its independence
limited.The true art of central banking, then, concerns
the nature of the trade-offsbargained by the monetary
authorities. Goodhart noted that the importance of dis-
cretionary changes in policywill persist as long as gov-
ernments maintain their monopolyof the right to issue
money.

Monetary Targeting:Lessons from the U.S. Experience
DavidLindsey
I will attempt today the ambitioustask of weavinga con-
ceptual discussionof U.S.monetary control techniques
into a chronologicaldiscussionof four separate periods
in recent U.S.monetaryhistory: 1970 through October
1979,October 1979through the fall of 1982,the fall of
1982to the fall of 1987,and the fall of 1987to date.I will
address changes both in the tactics of operating proce-
dures and in the strategy toward intermediate targets
used by the Federal Reserve Open Market Committee
(FOMC)over the four periods. The outstanding issues
related to monetary targeting in the U.S. will, I hope,
emerge clearlyfrom the discussion.
1970 through October1979
Over the 1970s, the Federal Reserve gradually
strengthened its reliance on monetary aggregates.Step
by step, the monetaryaggregates-mainly Ml and M2-
supplanted interest rates as the primary intermediate
targets for monetary policy.In the early 1970s,the Fed-
eral Reserve began focusing internally on the growth
rates of monetary aggregatesas indicators of monetary
stimulus or restraint. The Federal Reserve started an-
nouncing publiclyits desired ranges for annual growth
rates for selected monetary aggregatesin response to a
joint resolution of Congresspassedin 1975.Later,Con-
gress legislativelymandatedthis practice.
The current provisionsof the FederalReserveAct, as
amendedbythe Full Employmentand BalancedGrowth
Actof 1978,require the Boardof Governorsof the Fed-
eral ReserveSystemto report to the Congresstwiceeach
year on the "objectivesand plans of the Boardof Gover-
nors and the FederalOpen MarketCommitteewith re-
spect to the ranges of growth or diminution of the
monetary and credit aggregates."This section of the Act
also states that "nothing in this Actshall be interpreted
to require that the objectivesand plans [for the mone-
tary and credit aggregates]... be achievedif the Board
67
5
of Governorsand the Federal Open MarketCommittee
determine that they cannot or should not be achieved
because of changing conditions ... provided ... the
Boardof Governorsshall include an explanationof the
reasons for any revisionsto or deviationsfrom such ob-
jectives and plans."
The Federal Reserve's record in attaining the an-
nounced annual ranges for Ml and M2 from 1976 to
1979was somewhatmixed.The rangesforeach calendar
year are shownby the lines in the top panelof Figure5-1
and the actual growth rates of Ml and M2are shownby
dots, using the definitionsof these aggregatesin force
during these years. Over each of these calendar years,
only one of the aggregatesgrew within its range, while
the other exceededits upper bound. In addition,Ml ex-
ceededits upper limit during the first three quarters of
1979.
For most of the decade,the FederalReserverelied on
the federal funds rate-the interest rate that banks
charge one another on overnight loans of reserves-as
its operatingtarget in attempting to attain its monetary
objectives.When moneygrowthwas fasterthan desired,
the FOMCraised its operating range for the funds rate,
shownby the solid bands in Figure5-2. The FOMClow-
ered the range when money growth was undesirably
weak.The trading desk at the Federal ReserveBank of
NewYorkaltered the supplyof nonborrowedreservesto
keepthe supplyof total reservesequalto the demandfor
total reservesat the desiredfunds rate. This kept the ac-
tual funds rate, shown by the crosses, generallywithin
the operating range.
Through these procedures, the Federal Reserve
sought to influencedirectly the quantity of money de-
manded by the public. It tried to select a level of the
funds rate that would make the public want to hold an
amount of moneyequal to the targeted value. Whenthe
trading desk raised the funds rate, other short-term in-
terest rates tended to rise in sympathy.Thepublic then
68 David Lindsey
Figure 5.1 Adopted Ranges and Actual Growth Rates for Ml and M2, 1976-1979
MI (currency and demand deposits)
Actual
X -
1976 1977
X Tairget*I
z ZX Rcitge
M2 (Ml plus savings and small time deposits at commercial banks)
1976 1977
found market instruments more attractive relative to
moneybalances,whichweresubject to interest-rate ceil-
ings or outright prohibitionson the paymentof interest,
The resulting transfer of funds from monetary to other
financialassets was reflectedin a reducedstock of mon-
ey. Over time as well,the higher interest rates and ac-
companying tighter credit conditions tended to damp
spendingand hence transactions demandsfor money.In
practice,when moneydemandstrengthened,the FOMC
did not alwaysalter the funds rate promptly enough or
sufficientlyto keep monetary aggregates consistently
within their ranges, evenover the longer run.
October 1979through the Fall of 1982
On October6, 1979,facingabove-targetmonetary ex-
pansion,worseninginflation,and their consequencesin
domestic and international financial markets, and dis-
satisfied with the old procedures, the Federal Reserve
switchedits operating target from the federalfunds rate
to nonborrowedreservesas a sign of its commitment to
longer-run restraint on moneygrowth. The Federal Re-
serve announced that it had switched procedures "to
supportthe objectiveof containing growthin the mone-
1978 1979
Ata
1978 1979
tary aggregates[by] .......placing greater emphasison the
supply of bank reserves and less emphasison confining
short-term fluctuationsin the federalfunds rate."
Holdingto a nonborrowedreservestarget path essen-
tially gives the short-run total reserves supply curve a
positiveslope, as in Figure 5-3. The stock of total re-
servesis on the horizontal axisand the federalfunds rate
is on the vertical axis. The supply curve is positively
slopedmainlybecausean increasein short-term interest
rates relativeto the discountrate induces depositoryin-
stitutions to increase their borrowing at the discount
window.With nonborrowedreservesfixed,total reserves
suppliedwillbe higher,and hence the stockof reservable
depositsthat can be supportedby the outstanding non-
borrowedreserveswill be larger.
The increase in borrowinginducedby a given rise in
the fundsrate has a limit, though, becausedepositoryin-
stitutions are expectednot to make repeatedor continu-
ous use of the discount facility for adjustment credit
under normal circumstances. Thus banks experience
some rising implicit cost to added use of the discount
windowas they anticipatemore administrativepressure
to adjust reservepositionsin some other way.Bankswill
increase their borrowinguntil the implicit cost on the

Figure 5.2 Short-run Tolerance Ranges and Actual Levels of Federal Funds Rate
16*
14
12-
10 -
averaigefederailfunds raite'
6 tolerantce ranige
41976 1 1977
last dollar of discount borrowings has risen by an
amount equal to the initial change in market interest
rates. In equilibrium, the marginal cost to a bank on
each of its managedliabilitiesis equalized,includingthe
sum of the explicitdiscount rate and the implicit mar-
ginal cost of discountwindowborrowing.
A nonborrowedreservesoperating target, therefore,
providesan automaticself-correctingmechanismacting
partially to resist divergencesof the money stock from
its targeted value that could be mimickedonly through
judgmental adjustments under a federal funds rate
guide. Under the nonborrowed reserves procedure, a
surge in nominal money demand relative to target-
owing to an increase in real income,a rise in the price
level, or a positiverandom disturbance-will cause re-
quired reservesto increase and shift the demand curve
for total reservesto the right. As banks increase bor-
rowedreservesto fillthe gap, the implicitcostof the bor-
rowingrises and they bid up the funds rate. Asthe funds
market movesup and to the right on the total reserves
supply curve, the funds rate rises. The surge in money
will be partially checked by the induced rise in short-
term interest rates.
MonetaryTargeting:Lessonsfromthe U.S.Experience 69
~~~~~~~~~~~~Actual
level of
'W~~~~~~~~~~~~Short-run
1 /978 1 1979
Figure 5-4 shows how this automatic mechanism
workedduring the three years after October 1979.The
two panels indicatemonthlylevelsofMl and M2relative
to the upperand lowerbounds of their annual ranges. A
strengthening of money relative to target raised re-
quired reserves.Since nonborrowedreserveswere held
to a fixedpath by the trading desk,reserve positionsof
depositoryinstitutions were automaticallytightened as
institutions in the aggregatewere forcedto borrowthe
extra reserves at the discount window. These adjust-
ments plus seasonal borrowings are shown in Figure
5-5.
Institutions were induced to borrow additional re-
serves at the windowonly after they had first attempted
to acquire the needed reservesfrom other institutions.
In doing so, they bid up the federalfunds rate relativeto
the discountrate. This responseof the funds rate may be
seen in Figure 5-6. Higher short-term interest rates in
turn encourageddepositoryinstitutions and the public
to make secondaryadjustments to balance sheets that
curtailed depositexpansionand helped to bring money
backto path. This processoccurred automaticallyin the
absence of additional deliberate actions by the Federal
 70 David Lindsey
Figure 5.3 Reserves Market
Federal
Fiends
Total
Reserves
Demand:s
Required
+ Excess( *) Total
Reserves
Supply:
Nb oBrrowed(d
) omize on conventional checking account balances sub-
Discouznt Broe\
Rate -- \ and the financial regulatory agencies to allow the cre-
Nonborrowed"
(*) This schedule shifts withchanges in the demandfor
reservable deposits or in the willingness of banks to
hold reserve sitrplitses.
(**) This schedule shifts with changes in open market operations,
in the discountrate, or in banks' willingnessor needto borrow rates than was true of demanddepositsor currency.After
Reserve. However,the Federal Reserve also frequently
opted to amplifythese automatic effects.For example,in
the case of a monetary overshoot,it might have lowered
the nonborrowedreserves target or raised the discount
rate.
From October 1979to the fall of 1982,divergencesof
money,especiallyMl, from long-run targets were closely
associatedwith changes in borrowed reserves and the
federalfunds rate in the same direction. Such variations
in short-term interest rates were part and parcel of the
process that returned money to longer-run objectives
over time in the faceof divergences.For those three years
the nonborrowedreservesoperating procedurespermit-
ted short-term interest rates to move over a much wider
range in the intermediate run than under the previous
procedures.
Overthose three years,the FederalReserveslowedthe
trend rate of moneygrowth,and while real economicac-
tivity stagnated and unemployment rose to a postwar
record,the rate of inflationessentiallywashalved.But as
monetarist critics noted, moneygrowthas well as inter-
est rates showed more intermediate-term variability.
What explains this behavior? That money and interest
rates movedin the same direction establishesa prima fa-
cie case that the demand curvesfor money and total re-
servesvaried over a wider range than their supplycurves
did. Moreover,there is evidencethat the FederalReserve
in fact confronted a more unpredictable demand for
moneyafter the inception of the new procedures.
At the start of 1981,negotiableorder ofwithdrawalac-
counts (NOWs)wereauthorized nationwide,boostingac-
tual Ml demand. Working in the other direction, cash
management techniques evidently spread to medium-
size corporations, and some demand depositors also
shifted into moneymarket mutual funds.As in the mid-
1970s,the high market interest rates that accompanied
high rates of inflation prompted moneyholders to econ-
ject to interest-rate ceilings,and promptedthe Congress
ation of newaccounts payinghigher returns.
By 1982,the growth of fixed-ceilingNOWshad evi-
dently raised the interest elasticityof Ml demand. The
sizabledeclinesin rates in the second half of 1982began
to engender a larger increasein desiredmoneyholdings
Reserves than suggested by conventional money demand func-
tions. Figures 5-7 and 5-8 show that declines in the
three-month bill rate relative to the NOWrate (Figure
5-7) narrowedgreatlythe opportunitycost of Ml (Figure
5-8). The NOWaccount component, and Ml as a whole,
had becomemore elasticwith respect to market interest
all, using hypotheticalfigures,a fall in short-term mar-
ket rates from 15 percent to 10 percent represents a de-
cline of one-third in the marginal opportunity cost of
holdinga demanddeposit(evenone payinga positiveav-
erage implicitreturn via servicesindependentof account
size). Bycontrast, the marginal opportunitycost of hold-
ing a NOWaccount payinga 5 percent explicitrate of re-
turn is reduced by one-half. For 1982 as a whole, Ml
growth substantiallyexceededits upper bound, while the
measured velocityof Ml, shown in Figure 5-8, posteda
nearly 6 percent decline, unprecedented in the postwar
period.In light of these circumstances,and with further
impending deregulationhavinguncertain effectson Ml,
Chairman Volckerannounced on October 9, 1982,that
for the time beingthe FOMCwasplacingreducedempha-
sis on Ml as a guide to policyrelativeto broader mone-
tary aggregates.
The Fall of 1982 to October 1987
The de-emphasisof Ml in the fall of 1982, together
with the more flexiblestrategy that followedregarding
the monetary aggregatesmore generally,was mirrored
by a shift in operating procedures.The nonborrowedre-
serves operating procedure,with its automaticityderiv-
ing mainly from variations in required reserves on
 MonetaryTargeting:Lessonsfrom the U.S.Experience 71

Figure 5.4 Ml and M2 Growth Relative to Annual Ranges, 1980-1982

Ml
$500

$450 -
Actual monthly
leve'l

X£n$400 _
$4009,' An Annual range

$350_
1979 1980 1981 1982 1983

M2
$2,200

. $2,000

.Z $1,800

Actual monthlv
level
$1,600 -
Annual range

$1,400I I, l I l l l I, l l
1979 1980 1981 1982 1983

transaction deposits, has to be modified if Ml is de- federalfunds rate operating target. Instead, the nonbor-
throned as the primaryintermediatemonetarytarget (or rowed reservestrategy and its automaticitygaveway to
if a more judgmental approachis takenin reactingto Ml a techniquethat allowedthe funds rate to be determined
movements). The Federal Reserve did not return to a by the market, through the targeting of discount win-
Figure 5.5 Discount Window Borrowing, Adjustment Plus Seasonal Borrowing,
1979-1982 (monthly)

3 -

0
1979 1980 1981 1982 1983
72 DavidLindsey

Figure 5.6 Federal Funds Rate and Discount Rate, 1979-1982 (monthly)

25
Fede2a5fiindsrate Discountratepluts
20 Federal-fundsrate surchargerate

10 ~~~~~~~~iscount
rat

5
1979 1980 1981 1982 1983

dowborrowingfrom one reservemaintenance periodto
the next, implemented by allowing a flexiblenonbor-
rowed reservespath. Over longer periods,"the degree of
restraint on reservepositions,"as reflectedin the levelof money markets generally,as well as to disturbances in
adjustment plus seasonal borrowings at the discount
window,wasaltered morejudgmentallyin responseboth
to movementsin monetary aggregatesrelativeto short-
term objectivesand to other economicand financialde-
velopments.
Theshift in operatingproceduresweakenedthe earli-
er correspondencebetween movementsin monetaryag-
gregatesrelativeto their ranges, on the onehand, andin
discount borrowingsand the federal funds rate, on the
other. As the top panel in Figure 5-9 shows, the funds
rate from the fall of 1982to late 1987displayeda much
smoother pattern than it did during the previousthree-
year period.
The use of borrowingas an operating guide afforded
the FOMC,through the desk, considerable influence
overconditionsin the federalfunds market and to a less-
Figure 5.7 Three-Month TreasuryBill Rate and NOWRate
er degree other money markets,while still enablingthe
funds rate to fluctuate in responseto changesin market
expectationsof policy actions and other forcesaffecting
the supplyand demandfor reserves.Therewas a system-
atic, if somewhat loose, associationbetweenthe spread
of the funds rate over the discount rate and the willing-
ness of institutions to draw on adjustment plus seasonal
borrowings.Greater discount borrowing in the aggre-
gate tended to be associatedwith a wider spread of the
funds rate over the discount rate-which encouraged
more institutions to rely on the windowmore frequently
and for larger amounts. This associationbetweenactual
borrowing and the spread remained discernible from
1982through 1987,as maybe seen in the lowertwo pan-
els of Figure 5-9.
Although a borrowing guide providedan anchoring
mechanism that generallyacted to confine unexpected
movementsof the funds rate, upon occasion the two-
week average funds rate varied significantlyfrom the

Percent

16-

14 -3 Month T-bill rate

12-

10 : \

8-

6- NOW rate ., . . .,
....
~~~~~~.
..................... ......
........ .. ........
....
. ... ..............

I I I I I I I I I I I
1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990

Figure 5.8 Ml Velocity and Opportunity Cost
Ratio Scale
7.5-
,' ys
7 u(Left
MI Opportunity Cost(*)
(Right Hand Scale)
1978 1979 1980 1981 1982 1983
(*) Two quaner moving average. Assumes zero return on demand deposits.
area expectedto be typicallyassociatedwith a given dis-
count rate and a given intended levelof borrowing.
Severalepisodesof more sustainedshiftsin the funds
rate-discount rate spread relative to borrowing can be
seen in the lower panels.In the summer of 1984,in the
aftermath of the troubles of Continental Illinois Bank,
the spread widened while borrowing stayed around
U.S.$1billion, as large institutions became more reluc-
tant to be seen borrowingat the window.Subsequently,
this reluctance disappearedand the relation returned
more to normal. Then, in early 1986, borrowing de-
clined,but the spreadfellby less than wouldbe expected.
This greater reluctance seemedpermanent, perhaps re-
flecting more caution in reserve management after the
Bank of NewYork'spublicizedcomputer breakdownon
November21, 1985.
October 1987to the Present
Afterthe stock market break on October19, 1987,ad-
justment borrowingdroppedstill more, relative to the
spread, in several steps. Recently in 1990,adjustment
borrowinghas averagedonly US$100million.The weak-
ness in adjustment borrowinghas been overlaidbywider
swings over the calendar year in seasonal borrowing.
The predictabilityin the relation betweenthe funds rate-
discount rate spreadand adjustment plus seasonalbor-
rowinghas all but disappearedin the last three years.
Under circumstances of such evident shifts in bor-
rowing behavior,the FOMChas chosen to alter its bor-
rowingoperating objectivemore flexiblyin the last two
Monetary Targeting:Lessons from the U.S. Experience 73
Logarithmic Scale
-18
Is
~~~~~~~MI
Velocitv 1
Hand Scale) - 12
*. 6
1984 1985 1986 1987 1988 1989 1990
and a halfyears to prevent unexpectedfunds rate move-
ments, rather than to allowthe funds rate to divergeper-
sistently from expectations.The FOMCmay not have
revertedall the wayback to the funds rate targeting pro-
cedure of the 1970s,but it has comeclose.
The desired funds rate has been adjusted fairly fre-
quentlysince late 1987,as was the borrowingobjective
from late 1982to the fall of 1987.But these adjustments
havebeen more judgmental than in the early 1980s,and
are made on the basisof a wide range of incoming data,
not just M2 and M3.The reason partly stems from the
swingsin GNPgrowth relativeto the growthof the broad
aggregates, that is, from variations in their velocity.
Movementsin M2velocityin the 1980sare shownin Fig-
ure 5-10;the three-month treasury bill rate is compared
with the averagerate on M2in Figure 5-11. M2has not
proven to be a reliable short-run indicator of GNP or
thus a reliableshort-run guide to policy.
Thechart suggeststhat this problemis attributable to
a sizableinterest sensitivityof M2demand,though it has
less sensitivity than Ml demand over annual periods.
Swingsin M2'sopportunitycost, in Figure5-10, are fair-
ly well mirrored over time by swings in V2. Thus, the
FOMChas been forced to consider the likely interest
rates associatedwith prospectiveM2 paths, and the ef-
fect of those rates on spending,production and inflation
pressures.
The article on "MonetaryPolicy"by Henry Wallich
and myselfnotes the problems in some circumstances
for macroeconomicstabilizationwhen the central bank
adheres rigidlyto a predeterminedtarget for the money
74 David Lindsey

Figure 5.9 Money Market Conditions

Federal Funds and Discount Rates and Borrowings

Federal Funds and Discount Rates Percentagepoints

20
Monthly Surcharge

* -~~~~~~~~~~~~~~~~~~~
_
*w |: n r\ FederalFunds Biweekly
* ,Thru 5/2/90
- 12

DiscountRtee ,

4
1980 1981 1982 1984 1984 1985 1986 1987 1988 1989 /990

Spread between Federal Funds and Discount Rates Percentagepoints

8

* ~ ~ ~ ~~2
0

1980 1981 1982 1984 1984 1985 1986 1987 1988 1989 1990

Adjustmentplus Seasonal Borrowings(*. Millionsof dollars

3200

-2400

-1600

~~~ -800

!~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
' ;, S>S
Di<r

1980 /981 1982 1984 1984 1985 1986 1987 1988 1989 1990
(*) Excludesspecialsituationborrowing.

Box 5.1 Recent FOMCStatements of
Objectives
Portions of the Long-run Paragraph Adopted
at the February 6-7, 1990, FOMCMeeting
anTheFederalOpenMarketCommitteeseeksmonetary
andfinancial thatwillfosterpricestability,
conditions
promotegrowthin outputon a sustainablebasis,and
contribute to an improved pattern of international
transactions....
Thebehaviorof the monetaryaggregates willcontinue
to beevaluatedin the lightofprogresstowardpricelevel
stability,movementsin their velocities,and develop-
mentsin the economyandfinancialmarkets.
OperationalParagraphAdopted at the
February 6-7, 1990, FOMCMeeting
In the implementation ofpolicyforthe immediatefu-
ture,the Committeeseeksto maintainthe existingde-
gree of pressure on reservepositions. Taiking
account of
progresstowardpricestability,the strengthofthe busi-
ness expansion,the behavior of the monetary aggre-
gates, and developments in foreign exchange and
domesticfinancialmarkets, slightlygreater reservere-
straint or slightly lesser reserverestraint would be ac-
ceptablein the intermeeting period.The contemplated
reserve conditions are expected to be consistent with
growth of M2 and M3 over the period from December
throughMarchat annualratesofabout7 and3.5per-
cent respectively.The Chairmanmay callfor Committee
consultation if it appears to the Managerfor Domestic
Operationsthat reserveconditionsduringthe periodbe-
forethe nextmeetingare likelyto be associatedwitha
federalfunds rate persistenlyoutside a range of 6 to 10
percent.
Votes for the paragraph on short-term policy imple-
mentation: Messrs.Greenspan,Corrigan, Angell,Boe-
hne, Johnson, Kelley,LaWare,and Stern. Votesagainst
this action: Messrs.Boykinand Hoskinsand Ms.Seger.
stock when its demand is relatively interest sensitive
(Lindsey and Wallich 1989). For example, suppose
spending had clearly weakened and the central bank has
responded by adding to reserve availability in the face of
a very interest-sensitive demand for the targeted mone-
tary aggregate. The resulting fall in interest rates has led
to a sizable overshoot of money from target. In this cir-
cumstance, it may turn out better for the economy if the
central bank accepts the full overrun of money above tar-
get. With a highly interest-sensitive demand for money,
only a small reduction in interest rates is implied
keeping money on target when spending turns down.
Monetary Targeting: Lessons from the U.S. Experience 75
This easingin financialconditionswillprovideonlylittle
offset to the weakness in economicactivity,absent an
upwardadjustment to the target for moneygrowth. The
FOMCthus has needed to be flexiblein approachingits
annual or quarterlyobjectivesfor M2 or M3. It comple-
ments monetaryindicators by monitoringdirect indica-
tions of real activityand price trends, and by examining
indirect evidenceprovidedby movementsin other finan-
cialvariablesin domesticfinancialand international ex-
change marketsas well.
Box 5-1 indicatesthe long run objectivesthe FOMC
has set for itself, in the most recent publiclyavailable
record of policyactions. The secondparagraphindicates
that monetary aggregateshavea policyrole, but are not
rigid guides. Their behavior is evaluatedin the light of
progress toward price stability,movementsin their ve-
locities,and developmentsin the economyand financial
markets.
The operational paragraph's second sentence makes
clearthat movementsin moneygrowth relativeto annu-
al rangesor quarterlygrowth paths specifiedin the third
sentence are not accorded top priority. Behavior of the
monetary aggregatesare third among the factors to be
Box 5.2
Underlyig Monetary Relatons
Lowercasevariablesare the logarithmsof the uppercase
variables.
QuantityEquation
(1) MxV = P x Q= GNP
M = money
V= incomevelocityof money
P GNPdeflator
Q = real GNP
Long-RunPrice Concept
(2) P* = M x V*
Q*
V*=long-run velocity
Q*= real potential GNP
IdentityBetweenPrice, Output, and VelocityGaps
(3) (p - p*) = (q* - q) + (v - v*)
price gap output gap velocitygap
by
76 DavidLindsey

Figure 5.10 M2 Velocityand Opportunity Cost

Ratio Scale
1.9 -
_ 15

1.8 - M2 OpportunityCost(*) -10

1 7 - / V < H a~~~~~h
Hnd Scale)

, . ~~~~~M2
Velocin \t^
16 | ~~~~~(Left
HandScale) \ 3
2
1.6-

1.5 -

1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990
(*) Twoquartermovingaverage.

considered in making adjustments to the operating
stance of policyduring the inter-meetingperiod.
As a consequenceof the more eclectic approach to
policyguides, monetary growth rangessince 1982were
made less confiningand, even so, have not alwaysbeen
attained. This is shown in Figure 5-12. The MI ranges
werewidenedand twicethe basewas shiftedto midyear,
reflectingits de-emphasisin policy.Nonethe-less, they
were missed in two of four years beforerangeswere dis-
Figure 5.11 Three-Month TreasuryBill Rate and Average Rate on M2
continued in 1987.The ranges for M2,M3,and domestic
nonfinancialdebt were widenedin 1988,with the domi-
nant considerationfor M2 being its significant interest
elasticity.M2 rangeswere attained in all the years since
1982 except 1987.But M3 ranges were attained only in
1985, 1986, and 1988. This year both aggregateswere
within their ranges through March. To be sure, M3 was
close to its lowerbound, and as Figure 5-12shows,M3is
estimated to have fallen just below its lower bound in

Percent

16-

14 - Three-Month T-bill rate

12-

10

6 M2 Ownrate(*) ........... '. .

I I, , , ,

1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990
(*) Rate on componentsweightedby sharesof componentsinM2.
 Monetary Targeting: Lessons from the U.S. Experience 77

Figure 5.12 Monetary and Debt Aggregates

Ml Billions of Dollars Debt Billions of Dollars
850 10,400

- Actual Level Actital Level

- - - Estimated Level 82 10,000

_ ' _ -800
800 //-6.5~~~~~~~~~~~~~6.%
__
r - / m/ ~~~~~~~~~~~~~~~~~-
9600
-775

- - ~~~~~~~~~~750

I I i I I I II II I I I I I I I I I I 725 I I I I I I I I I I I I I I I I I I I I 8800
O N D J FM AM J J A S O N D J F MA MJ O N D J FM AMJ J A S O N D J FM A M J
1989 1990 1989 1990
M2 Billions of Dollars M3 Billions of Dollars
400 4300

- Actiual Level 350 Actual Level 7.5%

. - Estimated Level / 7% . . . Estimated Level 4200
7% ~~300 65

250 - - 4100

- 200 / 2.5%

>~~~~~~~~~~~ 150 - ->4000

0 3900
-50

O NDJ FMAMJ J AS O ND J FMA MJ ONDJFMAMJJASONDJFMAMJ

1989 1990 1989 1990
Figure 5.13 M2 Velocity

1.80 -

1.75-
Mean (1955:QI - 1988:Ql)
1.70-

1.60-
t | V65/,X- --- ---
1.55

1955 1960 1965 1970 1975 1980 1985 1990

78 David Lindsey
April.The weaknessin M3since the spring of 1989owes
mainly to the downsizingof the thrift industryand the
lessened need of these institutions for managed liabili-
ties in M3.
Despitethe shorter-run loosenessbetween monetary
aggregatesand GNP,somejustificationfor continuingto
announce rangesfor the broader aggregatesand forgiv-
ing them weightas a longer-run guide to policy can be
advanced.That rationale involvesM2'ssuccessfulprop-
erties as a long-run indicator of price trends. This indi-
cator propertymaybe seen makinguse of the conceptof
M2 per unit of potential GNPand ofan equilibriumlevel
of M2velocity.Mydiscussiondraws on some work that
has been done byBoardstaff (Hallman,Porter,and Small
1991).
Box5-2 reviewsthe keyconcepts and relation-ships.
Equation 1 is the equationof exchangeMV= PQ,which
states that the stock of moneyMtimes its incomeveloc-
ity,V,equals the product of prices, P,and real output, Q.
If we consider a long-run situation where velocity may
be presumedto havesettleddownto an equilibriumlevel
V* and real output is at its potential level identifiedas
Q*, the quantity equation can be rearranged to deter-
mine the long-run priceleveltowardwhichactual prices
presumablyare headed,for any given levelof the money
stock. This long-run priceconcept,which is calledP*, is
definedformallyin Equation2. The equationstates that,
in the long run, priceswillbe proportionalto the money
stock per unit of potential real GNP,with the proportion-
ality constant givenby V*.
From the quantity equation for actual pricesand the
equationforP*, we can derivean identity for the gapbe-
tween actual and long-run price levels from the differ-
ence betweenequation (1)and equation(2). Specifically,
Equation3 in the bottom panel of the first exhibitstates
that for the logarithmsof the variables,the gapbetween
P and P* is equal to the sum of the output gap, the dif-
ference between potential real output Q* and current
real output, and the velocitygap, the differencebetween
the current value of velocityand its long-run value V*.
If, for example,prices were belowtheir long-run level,
this situation would be consistent with the economy
running abovecapacity,with velocitybelowits trend, or
both. Asa result, priceswould then be expectedto rise to
reach their equilibriumlevel.
Finally,the separate terms on the right hand side of
Equation 3 can be identifiedwith differentviewsof the
inflationprices. The output gap is commonlyassociated
with the expectations-augmentedPhillips curve, while
the velocitygap has more of a pure monetarist orienta-
tion.
To implement P* empirically,we need to select a
monetary aggregatewith a long-run velocity levelthat
can be readilydetermined.Basedon the long-run stabil-
ity ofM2velocity,V2,shownin Figure 5-13,it appears to
be much easierto specifya long-run velocityestimatefor
this aggregatethan for any other aggregate.WhileV2 is
sensitiveto the movementsin its opportunity cost, flex-
ibilityin M2depositrates tends to stabilizethese costs in
the long run. Evidently,V2 eventuallyreturns to a level
close to its historicalaverage,shown by the long dotted
line in this exhibit. In what follows,V*will be set equal
to its longer-run historical averagein the construction
of long-run prices.
The price level, as measured by the GNPdeflator,is
plotted with the value of P* in the top panel of Figure
5-14,which uses a ratio scale.The four-quarteraverage
rate of inflation is shown in the lowerpanel of this fig-
ure. Nine vertical lines are shown in both panels at the
point where the price levelcrossesthe curve for P*.
In general, this chart shows that when the long-run
pricemeasure is abovethe current deflator,inflationwill
accelerate,and, conversely,it will decelerate when the
long-runprice measuremovesbelowcurrent prices.The
chart also shows that the change in the inflation rate
usuallylags the change in the gap between P and P*. In
the latest episodein which P* rose abovethe deflatorin
early 1985,prices did not showmuch tendencyto accel-
erate on a four-quarteraveragebasisuntil 1988.In 1989,
P* fell to approximateequality with the current price
level, suggesting a roughly stable rate of inflation over
1989, which, in fact, transpired. I have simulated the
model over 1990 and 1991,assuming M2 growth over
both years at the midpoint of the announced range for
thls year: 5 percent. The model suggests a gradual de-
trend in inflation,to about 3.75 percent.
hining
This evidenceindicatesthat M2'sgrowth,when aver-
aged over extended periods, minus the accompanying
growth of potential real GNP,maynot divergetoo much
from the long-termaveragerate of inflation.According-
Ily,some role, though clearlynot an exclusiveone, espe-
cially overshort-term horizons, for M2in the makingof
monetary policyseemswarranted.
References
Lindsey,DavidE. and HenryC.Wallich,1989."Monetary
Policy,"The New Palgrave edited by John Eatwell,
MurrayMilgate,Peter Newman.NewYork,NY:W.W.
Norton.
Hallman, Richard D. Porter, and DavidH. Small, 1991.
"Price LevelTied to the M2 Stock of Moneyin the
Long Run" Forthcoming American Economic Re-
view,September.
 MonetaryTargeting:Lessonsfrom the U.S.Experience 79

Figure 5.14 Inflation Indicator Based on P(*)

Logarithmic scale

Current price level (P)

te0' , ~~~~~
100

equilibrium
price level
(P*)
> Lt 50

1960 1970 1980 1990

Percent

10

Inflation
6

I ~~~~~~~~~~~~2

I I -~~~~~~~~~~~~~~~~0
1960 1970 1980 1990
Inflation (bottom panel) is the percentage change in the implicit GNP deflator from four quarters earlier. For 1990. Q2
to 1919:Q4 P(*) is based on M2 growth of 5 percent over 1990 and 1991, and P is simulated using the price gap
model developed by Hallman, Porter and SmalL
80 David Lindsey
Discussion
Question:There is often a questionabout the levelsof
optimal reserverequirements. I think that severalyears
ago, some people argued for uniform reserve require-
ments when M2was itselfa target for monetary policy.I
believeit was arguedthat shiftsof the portfoliowouldbe
neutralizedor offsetthrough the system of uniform re-
serverequirements. If these assertions are true, whyar-
en't reserverequirements in the UnitedStates uniform?
Mr. Lindsey: Prior to the MonetaryControl Act of
1980, the Federal Reserve argued fairly strongly that
uniform reserverequirements on transactions balances
wouldbe helpfulin tightening the link between total re-
servesand MI. In part becauseofthese conceptualargu-
ments for uniform reserve requirements behind
transactions depositsand MI, the MonetaryControlAct,
as passed, embodiedan almost ideal structure for con-
trolling Ml through reserves.
There are some ironies here. In the late 1970s,Paul
Meekand I debated whether, to control Ml, one would
find not only uniform reserverequirements helpful but
alsocontemporaneousreserveaccounting.So shortly af-
ter the passageof the MonetaryControl Act,the Federal
ReserveBoard staff went to work and pushed through
the FederalReserveSystema procedureforcontempora-
neous, rather than lag, reserveaccounting. Asa result,
in the UnitedStateswe reallyhavea very goodsystemfor
controlling Ml through a reservesoperatingtarget. The
irony of the situation, of course, is that in 1982we aban-
doned Ml as an intermediate target and in 1987 even
stoppedgivingranges for MI. Butwe alsoabandonedus-
ing reservesas the operating instrument. We fell back
first to discount windowborrowingas the operating in-
strument and now,in effect,use the federalfunds rate.
So I would say to Paul Meek,looking backover these
years,that I mayhavewonthe battle but lost the war. In
reality,whenan aggregatereservesmeasure is not being
used as the central bank's operating instrument, I think
there is onlya very weakintellectualcasefor havingany
reserverequirements.As Charles Goodhartcan tell you
from experienceat the Bank of England, if an interest
rate is being used as the operating instrument, interest
rates can be controlled just fine without required re-
serves.
The only theoretical concessionI would make (and I
knowsome disagreeevenwith it) is that reserverequire-
ments do preventdepositoryinstitutions from payingan
offeringrate that is fullycomparableto the rate of return
on their assets less some service charge fee. It drivesa
wedgebetween a market rate, like a three-month bill
rate, and the offeringrate on the monetarydepositliabil-
ity of the institution. This means that, as market rates
rise, the wedgemay stayconstant as a percentage,but it
opens up a widergap in terms of basis points. Since this
tends to reduce the demand for the aggregate, it gives
the central bank the ability to control the growth of the
aggregateon the demandsidevia an interest rate handle.
As a hypothesis,suppose required reserves are com-
pletelyeliminated,so that institutions can payvirtually
the same amount on their MMDA(moneymarket depos-
it accounts, which pay market-determined interest
rates) that theyreturn on their assets-their holdingsof
treasury bills, for example. I would argue that in this
case the rate of return on moneywill movevery close to
the rate of return on market instruments. There will be
a kind of Keynesianliquiditytrap that would fillthe en-
tire moneydemandspaceon a chart of interest rates and
money demand. Then it would not matter where the
central bank set the short-term interest rate, in the ab-
stract at least,becauseofferingrates wouldfollow.Banks
could make up for their cost by charging explicitfees;
they could carry the same rates on MMDAas they return
on their assets. It seemsto me that the depositorswould
be more or less indifferent between the MMDAand a
treasury bill.The demandfor moneywouldbecomevery
unstableand the utility of moneyas an intermediatetar-
get would,I believe,be evenmore questionablethan it is
already.
Apartfrom this argument, which I admit has a hypo-
thetical tone, the casefor reserverequirements is not ex-
ceptionallystrong if there is a tendencyfor sluggishness
in the adjustment of offeringrates. We haveseen this in
the UnitedStates,in contrast to my hypotheticalcase.If
I were making decisions for a developing country, I
would think twice before imposing reserve require-
ments.The case forthem is not all that strong. I knowof
no central bank in the world that uses their total re-
serves or even their nonborrowedreservesor monetary
base as operating targets to control the monetary aggre-
gate. The possibleexceptionsare Switzerlandand Ger-
many,but neither do it all that rigorously.
Question:It seems that becauseuniform reserve re-
quirements haveoften been promoted by both the IMF
and the World Bank. I was never convinced that we
should have uniform reserve requirements in the first
placeand, in the second place,that they should be set so
high.

Mr.Lindsey:Well,uniform reserverequirements are
helpfulif you are trying to set a path in advancefor total
reserves.It meansthat anyswitchesof depositsfrom one
institution to another will not affectthe "moneymulti-
plier," that is, they will not affect the relationship be-
tween total reserves and money.Therefore, if you are
controlling the money supply by controlling total re-
serves, uniform reserve requirements are helpful. But
that is a big "if."Fewcentral banks are eagerto walk up
to the precipice of what they saw the U.S. Federal Re-
serve do between 1979 and 1982: letting market rates
move over a very wide range just to keep money under
short-term or intermediate-termcontrol.Also,a 15 per-
cent reserverequirement is a pretty big tax that, I would
argue, tends to be passed backto the depositor.If there
are other nondepositoryintermediaries,such as money
market funds, that can start offeringtransactions-type
or savings-typedepositsand do not have this reservere-
quirements "tax" imposedon them, they certainly have
quite an advantage.It's not a levelplayingfieldif depos-
itories are stuck with reserve requirements but other
playerswho can enter the field,such as moneymarket
mutual funds, do not havethose requirements.
Mr.Honohan:Reserverequirements of 15 percent are
even more penalwhen the country has double-digitin-
flationor even80-90 percent annual inflation,as occurs
in some of the countries affectedby these requirements.
The penaltybecomesreallysevere.
Mr.Lindsey:Yes,in terms of percentagepoints, that is
right.
Mr. Caprio:We will see this issue again on Fridayin
the case of Italy.It is discussedin the paper,which is in
the conference readings, by Padoa-Schioppa.It details
very clearlythe problemsthe Italians had with dramati-
callydifferentreserverequirement levelsand the result-
ing controllabilityproblems like those to which David
[Lindsey]has alluded.
Mr. Lindsey:Charles Goodhart has made the point
that this competitivedisadvantageof depositoryinstitu-
tions can occur in such a way that your measure of the
Ml moneyaggregateis compromisedby shifts of trans-
actions services out of the depositorysystem and into
componentsthat are not included in your transactions-
deposits.In a limited way,wesaw it in the UnitedStates
with NOW(negotiableorder of withdrawal)accounts.At
first these were not includedin Ml, but our fleet-footed
central bank soonaltered the definitionof Ml to include
them. So that's a potential issue that needs to be consid-
ered as well.
Monetary Targeting: Lessons from the U.S. Experience 81
As a conceptual matter, I would argue that if this is
what worriesyou about reserverequirements,then just
payinterest on them. Ifyou reallywant to control money
via total reserves(whichno one wants to do), then keep
the reserverequirements but pay interest on them. You
still run into the problem in my hypotheticalcase that
the own rate comes close to the market rate. The de-
mand for the aggregate may become highly unstable.
But at least you don't have the problem of the reserves
"tax."If you think there will be this problem with insta-
bility of demand anyway,there is no point in trying to
control the monetary aggregateclosely.Youmight just
as well take offthe reserverequirements.
Question:But in the 1950s or 1960s,deposits with
thrifts were not includedin M2,were they?
Mr. Lindsey:They were not then. Indeed, they were
not until 1980.At the same time that we put NOWac-
counts in Ml, we redefinedour conceptof M2to encom-
pass deposits at thrifts. This measure of M2 is the one
used in Figure 5-13. However,Figure 5-1 does not use
this definitionof M2 but the one in force at the time.
Question:Isn't there substantialevidencethat in the
1970speoplewere using depositsat thrift institutions at
that time as a substitute for bank accounts and that
these depositswere not then includedin M2?
Mr.Lindsey:Whenwe workedon that redefinitionof
M2,we did look at that evidence.Some of my own re-
search suggestedit wouldbe a mistaketo drawthe divid-
ing line of M2beforeyou got to the small time deposits,
then have another moneyaggregatethat includedsmall
time deposits.As interest rates vary up and down, we
haveseen a significanttendencyfor shiftsbetweensmall
time deposits on one hand and savingsaccounts, NOW
accounts, and money market funds on the other. When
M2is definedbroadlyenough to encompassall those de-
posits, the shiftsare internalizedin that aggregate.
We actually had considered the other definition be-
fore, but it was pretty clear there was not much differ-
ence, in the United States at least, between the ways
savers use their savingsaccounts at thrifts and those at
banks. Thrifts were just beginning to offer NOWac-
counts in NewEnglandat that time, so the decisionwas
made to look at all depository institutions. Although
there was a little nostalgiawithin the Reserve Bank of
San Franciscofor the old definitionof M2for a coupleof
years, that was the end of it. No one has really second-
guessedthat decision.
In evaluatingthe P* model and the constancyof the
M2velocityin Figure 5-13,one reaction might be, "Is it
82 David Lindsey
not remarkablethat the velocityof M2 is so stable over
time?" On the other hand, when we definedthe aggre-
gate in 1980,we certainly were awareof its movements
with respectto nominal incomeand interest rates at the
time. I think that, in a sense, we stackedthe cards in fa-
vor of a sensibleaggregatewhen wedefinedM2that way
in 1980.It has held up reasonablywell since.
I think one can exaggeratethe reliabilityone should
grant the P* modelas a mechanismthat explainsU.S.in-
flation,given the fact that M2wasvery carefullydefined,
earlyin the decadeof the 1980s,to be as sensiblean ag-
gregateas we couldcome upwith. The P* modelcertain-
ly would not work as wellfor other aggregates;we knew
that, whenwedefinedit that way.Still, I would argue the
evidencedoes suggest it made sense to put thrifts and
banks together as depositoryinstitutions.
Mr.Honohan:Yousay the FederalReservehas, since
1982,movedreallytowardsan eclecticapproachto mon-
etary management.Now it maybe that this has worked
well enough because conditions-at least at the infla-
tion-output side-have been fairlysteadyin the United
States since 1988. In short, it may not be that hard to
fine-tunethings that are pretty much goingall right.
However,in many of the countries with which we
deal,things are alreadyvery far wrongor are likelyto be-
comeveryfar wrong quickly.Wouldyou agreethat some
anchor might be needed if a lot of things are going
wrong?And if things started going badlywrong, would
you jump for M2or for some other kind of main anchor
in a storm where so many anchors havefailedto hold?
Mr. Lindsey:The Federal Reservejumped to Ml in
late 1979,and I think that servedus well.It providedpro-
tection from political pressures that otherwise would
havekeptinterest rates from rising to the levelnecessary
simplyto bring downaggregatedemandand, in time, in-
flation.Atthat time, the FederalReservecould essential-
ly say,as it could not today,"Giventhe seriousnesswith
whichweare pursuing the Ml target,weare not control-
ling short-term interest rates. They are free to move as
necessaryto bring the quantityof Ml demandedbackto-
ward its intermediatetarget."
I think that washelpfulfor the country.It wasa break;
it providedan anchor for a time. There are manywaysin
which it was not a perfect anchor. We certainly did not
come that close,month to month, to hitting the targets.
But when all is said and done, I think it served pretty
well,ifonly as a kind of excusefor doingwhat we needed
to do to get a handle on inflation.
WouldI movetoward M2today?That is not an entire-
ly hypotheticalquestion. It is true that things have gone
fairlywell in the 1980s.Wehave not come close to price
stability-although the projection from the P* model
with moneygrowth in the 5 percent range over the next
two years (seeFigure 5-14)would tend to bring the rate
of inflation down gradually.But that is not a foregone
conclusion, nor is it a foregoneconclusion that M2will
grow at only a 5 percent annual rate. We'vedone fairly
well;we haven't brought inflationdown,but we'veman-
aged to stabilize the economy reasonablywell through
the 1980s.Is there a guarantee that we'll continue to do
that? I don't think there is.
One can envisageall kinds of exogenousshocks,fail-
ures ofwill, unexpectedpressureson aggregatedemand,
or what haveyou, which could moveus back into esca-
lating inflation.I am not predictingthat, but it is possi-
ble. Andit is possiblethat in retrospect the FOMCwill
look back on its behavior over the 1980sand conclude
that, once again as in the 1970s,it movedtoo little too
late. "If only we had focusedon M2 during that period
more than we did, we might have prevented that from
happening."
If we everget to that situation, I think there would be
a case for going explicitlyto an anchor to try to get a
handle on the inflation process and bring things back.
Perhaps this should be the final point of my talk: There
is a case forkeepingyour eyeon long-termM2growth-
and forthat matter on longer-termnominalGNPgrowth
and other nominalvariables-just to makesure you are
not driftingup and awayfrom the objectiveof price sta-
bility rather than toward it.

Monetary Targets:European Experience
CharlesGoodhart
I want to start with a digressionpromptedbyDavidLind-
sey's excellent paper. In particular he addressed the
question ofwhat happenedin the UnitedStates in 1979-
1982.That will be a period in monetary history which
economichistorianswillrake over againandagain, rath-
er like 1929-1933in UnitedStates monetary history and
various occasionsin Britain, such as the suspension of
convertibilityin the Napoleonicwars. I thought that
David'spresentation of that period, though admirable
was still what one might describe as the Federal Re-
serve'snormal position.This positionholds that both in-
terest rates and the relativelyshort term movementsof
the monetary aggregates,notably Ml, were extraordi-
narily variable over that period because the demand
functionsfor the relevant monetaryaggregateswere also
extraordinarilyvariable.Just for the record, I think that
it needsto be saidthat at least two other sets of explana-
tions existfor that period.
One of these, which I will call the second set of expla-
nations, is the one favoredbythe monetaristshere in the
United States. They describe it as simplya control fail-
ure. The Federal Reservewas using the wrong tech-
niquesand the wrong instruments for monetarycontrol.
It had a laggedaccounting basis,it did not imposepen-
alties for borrowingat the discount window,and so on
Allan Meltzer,Robert Rasche, and others are identified
with this view.
Then there is the third sort of explanation,which
Charles Freedman from Canada and I have presented.
Under this view,what happened in 1979and 1982was
predier.thisview,predicthaed it; wo1979pret d exacy
predictable. We predicted it; we would predict exactly
similar eventsfor any similar country that tried to intro-
duc tiht,reltivly
duce tight, relatively hor-tem, control
short-term, ontol over
veraa mne^
mone-
tary aggregatein a situationwhere ultimatelythe effects
on the monetary on thggregtes
monearywere,
aggregates ere, and
nd hd
had to ome,
to come,
through interest rate changes.
The problemsimplyarises as a result of lagsin the de-
mand-for-moneyfunctions,particularlythe lags before
83
6
interest rate changesaffectthe demandfunctions.Weall
know that such lags exist. In Equation 1, I've simply
written an equation down with the demand for money
(M)as a function of nominalincomes f(Y,i),with interest
rates laggeda number of periods.
Md =f (y_ it, t it 3 ) (1)
tu=o( that the moneysuppl (1)
suppose that the moneysupplyrises but becauseyou
as policy-makerhave not yet observedthe rise, you are
holding interest ratesat their initial level.Aftera month
or so, you observethat the moneysupplyis rising rather
faster.In the first month,you reallydon't know whether
the trend is real, so you waitanother month. Themoney
supply goes up again. After two or three months, you
start to raise interest rates.
If there is a lag in the effectof interest rates on money
demand, the first month that you push up interest rates
very little will happen. So the followingmonth, money
supply growth looks even worse; you push up interest
rates again. If the lag requires sixmonths beforeinterest
rate increasesaffectthe demandfor money,you contin-
ue pushinginterest rates up in each of those months un-
til rates are very high. Giventhat the interest rate effect
on demand is significant but lagged,once it starts to
work it worksmore and more. It alsointroducesindirect
effectson income (Y),which starts going down.So a lot
of things start going in reverse.It is not difficultto show
that ifyou havelagsin the demand-for-moneyfunctions,
particularlywith respect to interest rates, the more you
try to maintain stable control of M on a monthly basis,
the more you run into what is known technicallyas in-
strument instability.Youend up in a total quagmire.In
ment ista hat happend in 1979q982. An
my
wouldview, that again
happen is what happened
in any in 1979-1982.
other country Andto
attempting it
do hame thin in try ot on attemting to
do
sis;the same thing.
it cannot Don'ttry to do it on a continuousba-
be done.
84 Charles Goodhart
On MonetaryTargets
That was the digression;now to my theme proper. I
am goingto ask you to rememberwhywe are concerned
with monetarytargets at all. I shallargue that amongthe
requisitesfor monetary targets are relativelystable rela-
tions between the monetary aggregateand nominal in-
comes in the short run and between the monetary
aggregateand prices in the longer run. I shall then indi-
cate that, in the European context,these relations have
not been sufficientlystable over recent years. For the
majority of countries involved,this has not mattered at
all for a number of years.The exceptionsare the Federal
Republicof Germanyand the United Kingdom,whose
philosophyand experience I shall discuss. In practice,
the others havenot been on a monetary aggregatetarget
at all. Instead, their nominal anchor has been the ex-
change rate mechanism:their membershipin the EMS
(EuropeanMonetarySystem).
What I most want to bring to your attention this af-
ternoon are the changingviewsabout the waythis exter-
nal nominal anchor, the exchangerate mechanism,has
been changing. This analysis is known as "the new
EMS."Then I shall turn to the issue ofwhat happensfor
those countries without this external nominal anchor,
either becausethey are the hegemoniccenter ofthe EMS
(Germany)or are outside the EMS (the United King-
dom). Since they lack this external anchor,what should
they do if they no longer can, or no longer choose to,
adopt a monetary target, whether becauseof the insta-
bility ofvelocityor another of the reasonswhymonetary
targets are no longer maintainedvirtuallyanywhere?
I shall also talk about a constitutional idea whose
time has come. This idea is to give the central bank
much more independence.In effect,the legislatingau-
thorities say, "Wedon't know exactlyhow one controls
inflation.But ifwe give the central bank the mandate to
do so, particularlyifwe givethem an incentivestructure
in terms of salariesand reappointmentsso they will try
to do so, we believethey will use their interest rate in-
struments to achieveit. Although nobodyknowsexactly
what is necessary, we believe it can be done." This, I
think, is the current state of thinkingabout central bank
independencein quite a number of countries.
If I have time, which I probablyshall not, I shall also
talk about the effectivenessand range of various central
bank policy instruments in conditionswhere exchange
controls havebeen removed.This is enormouslyimpor-
tant. The one term that I did not hear in all the previous
discussionof required reserveratios was "exchangecon-
trols." Youcannot imposediscretionaryburdens, which
required reserve ratios effectivelyare, on independent
countries unless those independent countries isolate
themselvesfrom the world financialsystem by using ex-
change controls.
First then, why did central bankers start with mone-
tary targetry? One considerationwith respect to central
bankers whichyou must never forget,and more so now
than ever,is that central bankershavea club. I know be-
cause I used to belong to it. They all talk to each other.
That means that they tend to behave rather like the
sheep on my farm;they all tend to move together.There
are fashions, and undoubtedlythere was a tremendous
fashionfor monetarytargets in the mid to late 1970sand
the early 1980s.That fashion has now changed;mone-
tary targets are out of fashion. Even though, as David
Lindseywassaying,the UnitedStateshas a perfectlyrea-
sonablemonetary target in M2,whyisn't the FederalRe-
serve using it? In part, it is becausethis is an ideawhich
is unfashionable.In economicsyou can see ideas going
around and around. There is a splendid quote, which I
got from Dennis Robertson,that intellectualsare rather
like hares;as with anyother formof hare coursing, ifyou
stand in one place they alwayscome backto you in due
time. So ifyou like monetary targets, don't worry,in due
time they will come back.
Whydidwe actuallyhave monetary targets?The first
point to note is that no one reallyis interested in M2 or
M3or Mx,as such.When NigelLawsonestablishedmon-
etarytargets, it wassaid that in the working men's clubs
in Newcastle,the talk was of nothing but Sterling M3.
But the ultimate reason for concern with monetary tar-
getry is that it is supposedto havea stable link with eco-
nomic events and developmentsthat affect economic
welfare,which does concern us. The supposedlink is to
either price levelor real incomes,via that dear old stan-
dard, the quantity identity (Equation2).
MV= PY (2)
=> PY (3)
If moneyvelocityis predictable-which is evenmore
important than its being stable-then a given M will
equate, in the short run, to a given nominal income. In
the medium to longer term, where one hopes to be able
to assume that real incomeswill revert to some equilib-
rium, the relation between money and price level im-
plies that monetary growth and the rate of inflation are
closely related. This relation is considerablystronger
than relation (3) that the central bankers had relied on
in earlierperiods,which wasjust to vary interest rates in
a way thought appropriate to bring P and Y back into
line.
The problem with reliance on the interest rate rela-
tion is that the effecton expenditures,and therefore on
the ultimate generalprice of output, is in fact influenced
by real interest rates rather than nominalrates. Whereas

nominal interest rates can be measuredexactly,expecta-
tions of price inflation are highly variable. Therefore,
one does not know real interest rates with any clarity,
and nominalinterest ratesare a misleadingindicator.On
the other hand, as long as velocity remains relatively
predictable,control of Mgivesyou a goodidea ofwhat to
expectof P and Y.
David Lindsey commented that if velocity, even
though predictable,is veryvariable becausethe interest
elasticityof the moneyaggregateis high, then you can't
run a monetary target; you will not be able to explainto
people at large the re-entry problem, which he de-
scribed,and problemsof that kind. I am not entirely per-
suaded by his reasoning.Certainly,such variabilitydoes
make monetary targetry more difficult.It would be nice
to havevelocitybe both predictableand stable, but pre-
dictablealonewould probablyget you most of the way.
Another advantage of using monetary aggregates
rather than interest rates is that a monetary target al-
lowsfor almost rule-basedadjustment. Interest rate ad-
justment, which is clearly discretionary,is subject to
politicalcontrol and politicalinfluence.There is also the
considerableconcern that the authorities would always
be tuning too late andvary interest rates too little if they
were allowedto use their discretion.Targetingmonetary
aggregatesis supposedto reduce the risk of that.
Yetanother factor is the use of monetary aggregates
as a leadingindicator.As I saidbefore,our interest is not
in Sterling M3or M2for its ownsake; we are concerned
with P and Y.If wecould actuallyobserveP and Y,imme-
diatelyand accurately,whyworry about M?M is simply
an indicatorwith noise-it certainlyhas a lot of noise. If
a rule can be introducedthat is tied to M,why not intro-
duce a rule that is tied directly to P? Take,for example,
the discussionsearlier about nominal anchors. Accord-
ing to this view of Mas just a noisyindicator,what some
of these countries should do is to collect data for an eco-
nomic series on wages.Onceyou have that, require the
central bank to raise interest rates by 1.5 times the in-
crease in wage earnings in the previous 6 months, or
somethinglike that.
Whatyou reallywant is to ensure that, everytime you
start getting an inflation, interest rates go up fast
enough and far enough. Nowif you can observeP and Y
accurately,whybother with M? This would be a form of
"Tobin"approachon nominal incomes.
There are severalcounter-argumentsto this noisy-in-
dicator case against monetary targets. First, it may be
that M, though noisy, is a leading indicator.The causal
chain linking money with nominal incomes and prices
maybe indirect.Ashas happenedin manycountries,you
can first get credit expansion,monetaryexpansion,asset
price inflation, and then on to commodityprice infla-
Monetary Targets: European Experience 85
tion. There can be quite a period of time betweenthe ef-
fect on M and the consequencesfor PY.If you see M
moving first, you may see PYmovinglater. This leading
indicatorargument, then, is an importantone. Whylook
at the noisyindicatorof what reallyconcernsyou,which
is inflation,rather than directlytargetingwhat concerns
you? Well,if Mis a leading indicator,and ifit is causal-
if the reason for the increase in P and Y is essentially
monetary-that is an argument for targeting M.
These arguments did exist in my own country and
many European countries. The receivedposition held
that V (velocity)was stable and predictable,that Mwas a
leading indicator, and that targeting M would lead to
more automatic, rule-driven,changes in interest rates,
rather than discretionary changes. When we compare
these expectationswith what actuallyhappened,I think
the main problem has been a declinein the stability of
the velocityfunction. Tosee why,let's considerthe indi-
vidual aggregates,starting with the UnitedKingdom.
The firstaggregateis Ml: narrowmoney. Acombina-
tion of high inflationand deregulation,togetherwith in-
creased competition, led banks to provide interest
bearing checkable-typedeposit accounts. These are ei-
ther going to be in Ml or close to Ml. Whenthis occurs,
the Ml aggregatejust goescrazy.That has certainly hap-
pened in the UnitedKingdom.In the last couple ofyears,
the clearing banks have introduced all kinds of check-
able sight deposits,interest bearing demanddeposits.As
a result, the non-interest-bearingcomponent of Ml has
collapsed,while the interest bearingcomponent(and to-
tal Ml) have risen out of sight. The structural changes
havebeen so dramaticthat Ml has becometotallymean-
ingless.
What has happenedto broad money, which was the
government'spreferred target initially?That is rather
more difficult. Remember that Sterling M3 regularly
grewmore slowlythan nominal incomes.That is, its ve-
locity was rising throughout the 1960s and up to the
time that a medium-termfinancialstrategywith a mon-
etarytarget wasadoptedin 1979.Eversince that strategy
was adopted, velocity has turned around. Sterling M3
has growna great deal faster than nominalincomes,and
velocityhas fallenvery sharply.
What brought about this dramatic and tremendous
shift in velocity,which, I might add, none ofus predicted
in advance?Well,weare not absolutelycertain,but there
are probablytwo or three causal factors. First, at the
same time that the Conservativegovernment went for
monetarytargetry,it introduceda considerabledegreeof
deregulation,particularlyderegulationof the banks.All
direct constraints on bank lendingwere removed,in part
becauseexchangecontrols were removedin 1979.Once
you get rid of exchangecontrols, if you havedirect con-
86 Charles Goodhart
trols on your ownbanks a large part of your banking in-
termediation simplymoves offshore.We could see that
happen.
One can never tell what the actual effectis of direct
controls and banking system regulation. But one factor
that was importantwasthe effecton bank strategy.After
deregulation, the banks changed the kind of lending
they did. Prior to 1979, banks did very little mortgage
lending in the UnitedKingdom;it was all left to the spe-
cialized building societies, which are our more stable,
profitable, and better-run equivalent of the American
Savings and Loans (S&Ls).(We do not have an S&L
problem forall kinds ofreasons;depositinsurance isjust
one of them.) In any case,the banks went into mortgage
lending,and the increase in the rate of lendingafter de-
regulationaltered dramaticallyas the banks shiftedtheir
strategicpolicy.So their lendingwent up; the rate of in-
crease in bank lending,after deregulation removedthe
supplementaryspecialdepositsystem in 1979,has aver-
aged wellover 20 percent per annum in the 1980s.
This was financedin large part by issuing more at-
tractive interest bearingdepositsof variouskinds. David
Lindsey'snightmare occurred;most of the moneysupply
was effectivelyofferedat market-related interest rates.
Furthermore, the increasedcompetitionmeant that the
margin, or the spreadbetweenlending rates and deposit
rates, went down.If you ask, "Whatis the cost of inter-
mediation?" the answer ought to be, "the spread." If
there is zero cost of intermediation and intermediation
providesany servicefor you, the amount of intermedia-
tion that you ought to undertake ought to be infinite. If
you get some particular liquidityfrom holdingdeposits,
while the cost of borrowingis the same as the costof de-
posits, you borrow an infinite amount and you have an
infinite amount on deposit.There is no charge and you
are totally liquid. It's a wonderfullife.
The lowerthe spread,the more intermediationpeople
willdo. In the UnitedKingdom,the total volume of bor-
rowing by persons and companieshas shot up, and the
total amount of depositsthey hold has shot up. The de-
gree of intermediationis inverselyrelated to the spread.
This is in large part where reserverequirements come
in. The reserve ratio requirements increase the cost of
intermediation.They are important in this respect.I am
not sayingthat it is particularlydesirable,simplyin pur-
suit of some monetary aggregate that may or may not
havea stable relationshipwith nominal incomes,to less-
en the efficiencyof the banking system in a developed
country, which needs proper financial intermediation.
But that is another story.
In large part then, the stability of a monetary aggre-
gate is likelyto be a function of the degree to which its
own interest rate is artificiallyor institutionally con-
strained. The one monetary aggregatethat does not pay
interest is MO,which is cash in the hands of the public,
(although it could in principle payinterest). MOwas the
monetary aggregateto which the British authorities re-
treated when all the other monetary aggregates lost
their stabilityand went haywire.BarryJohnston,who is
not here at the moment,wrote a paper forthe U.K.Trea-
sury showing that the demand for MOwas reasonably
stable and has remained so. This providesa fig leaf for
the Chancellorof the Exchequer;he can say that mone-
tary targets have not been completelyabandoned;when
they havebeen.
Whyin practicedoesMOnot matter? It doesn'tmatter
in practice because nobodybelievesthat it is a leading
indicatoror causal factor.Everybodyknowsthat anyone
can get currency on demand. It would be inconceivable
for the Bank of England not to allow anyone to obtain
cash when theytry to cash a check.Theconvertibilityre-
quirement on the banks must remain. This meansthat a
change in MOneither causes nor predicts a change in
nominal incomes or consumers' expenditure.It is sim-
plya coincidentindicator.It is not a bad one, although it
is a bit noisy;it is a coincident indicator of consumers'
expenditure.
However,we alreadyknow consumer's expenditures;
we get data promptly.MOdoes not tell us anythingwe do
not knowalready.Accordingly,insofaras the Chancellor
of the Exchequer,the Treasury,and, to a degree, the
Bank of England are concerned about consumers' ex-
penditure, theywill raiseinterest rates anyhow.But they
do not reallylook at MO;nobodylooksat MO.It is a pure
fig leaf.
Let us move on to Germany,where the concernsand
problems are different. Germanyhas not had so much
high inflation, has not had so much structural change.
Therefore,for a long time, there was far less disturbance
to the demand-for-moneyfunctions and the velocity of
moneythan in the UnitedKingdom.The Germansmain-
tained, and attached a great deal of importanceto, their
Central BankMoney(CBM)target. CBMwas simplythe
total liabilitiesof the central bank. But they never con-
sideredit as a simple monetarybase aggregate;they con-
sidered it as a weightedbroad monetary aggregate.
Neverthelessthe weight in currency was far higher that
the weight of the other constituent elements, which
were weightedby the reserveratio applicableat a partic-
ular point in history.
In recent years,the amount of deutsche mark curren-
cy, particularlythat held in Eastern Europefor savings
and other purposes,started to increase.For this and oth-
er reasons,about two or three yearsago the relationship
between currency, CBM,and what was happening to
German nominal incomes just went haywire.One of a

number ofproblemswith currencyis that it is frequently
held abroad in large quantities; this is true of the U.S.
dollar and the deutsche mark. In addition,a large pro-
portion of what is held is either held for nefariouspur-
poses or whenevera survey is done, nobodyknowswhy
it's held, or both. So the Germansmoved from a CBM
target to a broad monetaryaggregate.
That broad monetary aggregateis about to be deval-
ued even further becausethe German monetary union,
or GMUas it is now known,meansthat "Ostmark"(mark
of the German DemocraticRepublic)will be exchanged
for deutsche mark in some ratio. Nobodyhas the slight-
est idea of what the East Germansare going to do with
their new deutsche mark holdings. The complexityof
monetary changes will be of the same order as those
which I havebeen talking about in the UnitedKingdom.
The Deutsche Bundesbankwill not know how to apply
their monetary target to makesense of what is happen-
ing there over the next couple of years. So they will go
backto adjusting policyin responseto concernwith the
non-monetaryaggregates:inflationand a wholeseries of
other factors.
Finally let us turn to Italy and Spain, where the re-
movaofexcang
cotros in199 wil cusesuc
structural change that again the demand-for-money
functionswill go out the window.The marginal-indeed
the average-reserve ratios in Italyare on the order of 25
percent. Although some interest is paid on reserves,
there is something like a 10 percent differencebetween
the interest paidon reserverequirements and the inter-
est availablein money markets. Now 10 percent on 25
percent is in effecta 2.5 percent burden in addition to
the spread. I can tell you that bankers in London are
lookingforwardwith enormous enthusiasm to interme-
diating in lire at the moment that exchangecontrols are
lifted.
The Spanish have already started to dismantle their
reserve ratios. This has nothing to do with monetary
union; it has everythingto do with the removal of ex-
change controls in the run-up to the single Europeanfi-
nancialsystem. Onceexchangecontrols go, bank clients
will move to the cheapestpossibleintermediarycenter.
If there are significant differencesin the burdens on
banking systems,they will movevery rapidly.
In any case, Italy and Spain and all the peripheral
countries in the ERM(Europeanexchangerate mecha-
nism) havereallynot beenconcernedwith monetarytar-
getry over the past years. They have had an external
anchor in the form of maintenanceof the exchangerate
mechanism. The EMS,or ERM,has movedthrough at
least three stagessince it was introduced.Neverthink of
the EMSor the ERMas somethingthat is static; it is very
dynamicand changeable.
Monetary Targets: European Experience 87
The European MonetarySystem
The first stage of the EMSwas between 1979and the
time in 1983when Mitterrand switchedhorses and re-
versedthe whole thrust of French policy.In these first
fewyears of the EMS,there were frequent realignments
through which the nominal exchange rates were re-
alignedto accommodatethe wageand price adjustments
that had occurred since the last realignment, thereby
maintaining the stabilityof real exchangerates. In other
words, during its earlyyears the EMSdid not reallyhave
any long-term discipline. It was merely a short-term
pegging of nominal rates until they were realigned in
light of differinginflationrates.
The French change of strategy, which occurred in
March 1983,was the crucial occasion in the history of
the EMS.Thisbegan the periodofwhat is knownas Ger-
man hegemony. During this period, the Deutsche
Bundesbankset monetary policy,implementedthrough
monetary targetry,to try to achieve domesticprice sta-
bility in Germany,without significantor seriousconcern
for what was happening in the other EMS countries.
Havingchanged their strategy, the other countries were
primarily concerned with holding the nominal peg to
Germanyand trying to achieveconvergence.Initiallythe
private sector, havingobservedthe earlier realignments
to maintain the peg, did not believetheir monetary au-
thorities. Because the private sector did not believe
them, the nominalinterest rates in the peripheral coun-
tries had to rise to somethin like the inflationdifferen-
tgies a rise to balkethe feariof an
tial plus a risk premium to balance the fear of an
adjustment.
Since 1987,the situation has changed again. Private
agents in the various countries have increasinglycome
to believethat the peripheral governmentscan achieve
and maintain fixednominal parity.An influentialrecent
paper by Giavazziand Spaventa(1990)analyzesthe cur-
rent situation through the followingset of equations:
y = - Pr + lAs - 6u (4)
(p - pt) a(pt - pt-i) + fy (5)
s = e + p* - p (6)
rh =i - E(pt+l - Pt) (7)
XbE(pt+i-Pt) }(8)
ri = mi p
+ E(et+i- et)- E(pt+
1 - Pt
Equation4 expressesthe real output y in a particular
country as a function of the real interest rate r, the real
exchangerate s, and fiscalpolicyu. Equation5 is a par-
tially backwards-lookingdetermination of inflation,
while equation 6 definesthe real exchangerates, where
88 Charles Goodhart
the nominal exchangerate, e, is definedas the domestic
currency price of foreignexchange.
The last two equations are the truly important part
for the authors' analysisof peripheral economies.In par-
ticular, the peripheral economiesunder discussionare
the Europeansouthern cone economies,includingItaly,
Greece,Portugal, and Spain. One mayalso include Ire-
landand England as honorarymembersof Southern Eu-
rope. The authors dividea peripheral economyinto two
groups. Those in the householdgroup can borrow only
within the domesticeconomy.For them, the real inter-
est rate, as expressedin Equation7, is the local interest
rate i, less the expected rate of inflation. The second
group comprises firms, which can borrow either at
home or abroad. If they borrowat home, they face the
real interest rate, which again is the local domestic in-
terest rate less the expectedrate of inflation.If they bor-
row abroad,they borrowat the foreignrate i*, whichfor
this analysis, is the German rate, less two expectation
factors (Equation8). The first is the expectedexchange
rate depreciationover the next periodof borrowing.The
second is the expectedlocal inflationrate.
The authors' point is that after 1987,the expectation
has been that, for the foreseeablefuture, the govern-
ments of countries such as Italy, Spain, Portugal, and
Ireland will not devaluebut will maintain the peg. Be-
cause of their relativesuccessin doingso, they need not
devalueeven iftheir inflation is higher than Germany's.
Thus the expected rate of depreciation is much lower
than the expectedrate of inflation.Firms in Spain, Italy,
Portugal, and so on, finding a German interest rate a
great deal lowerthan the domesticrate and havinga low
expectationof a change in the exchange rate, have bor-
rowedabroad heavily,becausetheir effectivereal inter-
est rate is thereby substantially reduced. In addition,
foreign investors, for examplefrom Germany,will put
money into these countries because doing so offers a
considerable economicadvantage,as long as one does
not expectthe exchangerate of deutschemark forthe lo-
calcurrency to be reduced.Accordingly,the real interest
rate in Spain,Italy,and Ireland-and in the UnitedKing-
dom, if it were to enter the EMS-goes down with a
bang.'
The immediateeffecton these peripheral countries of
maintaining their EMSposition (or of entering the EMS
system) now becomes much more stimulatory than in
the period of periodicdevaluations.It producesa highly
effective expansionary monetary policy because local
firms are borrowingabroad and not expectinga devalu-
ation. The countries experience huge capital inflows.
This creates a tremendous problem for the Italians, the
Spanish,the Portuguese, and others in sterilizingthe in-
flows.Their foreign exchangereserves have risen enor-
mously. If they cannot sterilize the inflows, their
domesticnominal interest rate goes down, which rein-
forcesthe expansion.
In this situation, a peripheral economy's exchange
rate goes to the top of the EMS band. At the moment,
whose exchange rates are at the top of the band and
whoseare at the bottom?Thoseat the top of the band are
Spain,Portugal,and Italy.At the bottom are the Nether-
lands and Germany.Becausethese peripheral countries
are booming, there is no monetary discipline.A great
deal of pressure builds on the wage front. For example,
wage increasesare now a source of concern in Spain.
TheEMSdisciplineis beginningto disintegrateunder
these particular influences. If the discipline disinte-
grates, the real exchange rate of course starts to go
wrongand increasinglywrong.But if the relativesize of
the real interest rate effectis greater than the relative
size of the real exchange rate effect,the whole system
can becomeunstable.This may leadto very severeprob-
lems and turbulence. These effects can be offset and
dealtwith, but they should be offsetand dealt with byfis-
cal policy.In other words, if a peripheral country in the
EMS (or one entering the EMS as a more inflationary
member) finds that the credibilityaccorded to its ex-
changerate will leadto capital inflowsand monetaryeas-
ing, that country must introduce a much more
deflationaryfiscalpolicy,or things will go badlywrong.
In a sense, the messagethat I want to leavewith you
is that, as the EuropeanCommunityapproachesirrevo-
cablyfixedexchangeparities, there will increasinglybe
just one Europeanmonetary policy.This appliesboth to
the current EMSand the evenmore ambitious develop-
ment towards European MonetaryUnion, expressedin
the Delors Committeereport. At the limit, with a single
currency,there would be no national monetary policies
at all. The interrelationbetweenfiscalpoliciesand mon-
etary policiesis of the essence.When national monetary
policy becomes predicated upon the maintenance of a
fixed exchange rate, operating fiscal policywithin that
conjoined system, in an EMSor European system, will
become increasinglyimportant. Unfortunately,one of
the great difficultieswith economicanalysisof EMUhas
been that much of it has assumedfreelyflexibleand per-
fectly adjusting markets. There has not been nearly
enough analysisof howfiscalpolicyshould be coordinat-
ed with monetary policyin all this.
Central Bank Independence
Wehavenot yet addressedthe questionsof what mon-
etary policyought to be in the hegemonic center or of
what monetarypolicy should be in the UnitedKingdom
if it stays outsidethe EMS.We are in a world in which,

as I have said, monetary targetry is out of fashion. The
instabilityofthe velocityfunction has madeit effectively
unusable. If you are either the "center" country or too
big to link to somebodyelse'scurrency,or for some oth-
er reasonyou want to maintain national autonomy,what
do you do?This brings us to the idea of central bank in-
dependence,towardswhich I think a lot of countries are
moving.This idea has become increasinglyfashionable.
What does central bank independencemean? It does
not mean giving the central bankers the independence
to print money to paythemselveslarge salaries.The in-
dependencewillbe qualified.Aspart of the central bank's
constitutionalmandate, its overridingprioritywill be to
maintain price stability. This has occurred in New
Zealand and to a degree in Chile and probably other
countries. Tosome large extent it is intendedfor a "Eu-
roFed,"a Europeansystem of central banks. Indeed,the
idea may possiblyexpandto provide an incentive struc-
ture forcompensatingthe central bank governorand se-
nior officialslinked to that priority.
My suggestion is that the central bank governor
should be paid some lowlyregular salaryirrespectiveof
outcome, perhaps a salaryequivalentto that of a profes-
sor of economics.Asthe governorgetsincreasinglyclose
to the desiredrate of price increase over an appropriate
period,he or she receivesa bonus. In the UnitedStates,
I would recommendthis should rise, in total, to some-
thing on the order of a million dollars; in the United
Kingdomto somethinglike halfa millionpounds. So the
central bank govemor becomes a very rich person if
price stabilityis achieved.
There are arguments against my suggestion. If the
onlyincentivefor the centralbank governoris to achieve
price stability and the central bank does not care about
anything else, one can alwaysachieve price stabilityby
shovinginterest ratesup to 30 percent. If I willget a mil-
lion dollarsto achieveprice stabilityin the UnitedStates
in three yearsand nothing else matters to me, I willpush
interest ratesup to 30 percent todayto bring about a ma-
jor depressionover the next fewyears, then lowerinter-
est rates down gently and run away with my million
dollars after three years. So there are problems about
trying to set out the central bank objectivesin a waythat
does not actuallyencouragesuch behavior.
Another major argument against a simple incentive
structure concerns coordinationwith fiscalpolicy.The
Monetary Targets: European Experience 89
coordination argument is an important one, and must
be addressed.
The last topic I was goingto talk about was monetary
policyinstruments. Canwe introduce direct credit con-
trols in a worldwithout exchangecontrols?The answer
is probably"No."Dowe want to maintain exchangecon-
trols in our various countries? Not in the medium term
if you can get away from it. Concerning interest rates,
there is an interesting paper in the latest Bank of En-
gland QuarterlyBulletin (Bankof England 1990).It in-
dicates that, as a concomitant benefit of reducing
spreads and having a far greater gross increase in both
borrowing and deposits, interest rates in the United
Kingdomnow impingeon a far widercommunity.Many
of the English young are so overborrowedwith huge
mortgagesthat when interest rates go up, if they are to
pay their mortgage they cannot go on holiday,they can-
not buy consumerdurables, they can barely eat. So you
can understandwhypoliticalpopularityis affectedby in-
terest rates. When your personal sector is heavilybor-
rowed, it has an effect.
Finally there is monetary base control. Is monetary
base control anything other than an alternative way of
bringing about interest rate variability,which you could
bring about directly?Or is there some direct,supplyside
effect that it has, in addition to its effect on interest
rates? In the United Kingdomrecently,Gordon Pepper
has stirred some interest by arguing that there is a dif-
ference. Mostof the rest of us do not believethat there
is, but it is still an issue of some interest.
Note
1. Editors'note: Sterlingentered the ERMin October
1990 (five months after the conference),and interest
rates did initiallycomeunder some downwardpressure.
References
Ciavazzi, Francesco and Luigi Spaventa 1990. "The
'New' EMS,"Centre for Economic Policy Research
DiscussionPaper No. 369, London,January.
Bankof England 1990."The Interest RateTransmission
Mechanism in the United Kingdom."Bank of En-
gland QuarterlyBulletin, May.
 Comment:Policy Constraintsin Developing
Countries
Donaldi. Mathieson
I wasaskedto discussthe relevancefor developingcoun-
tries of the previousspeakers'points on the use of mar-
ket-based monetary instruments. I would like to touch
on four basic issues. First, why have developingcoun-
tries traditionally favored credit controls rather than
market-basedmonetary instruments?Secondly,whyare
these credit controls coming under increasingpressure
in developingcountries? Third, what are some of the
preconditions for a successful transition from credit
control to a market-basedmonetary instruments?Final-
ly,why is it unlikelythat developingcountries willcom-
pletelyabandon the use of credit controls and directed
credits.
In discussingthese issues,I want to maketwo caveats
clear. Developingcountries are a rather heterogeneous
lot, and I shall be considering only a middle-income,
middle-sizedeconomy.Mysecond caveatconcerns the
objectives of monetary policy.I shall assume that the
fundamentalobjectiveof monetary policyis price stabil-
ity. I don't care whether "price stability"is definedin
terms of a price index or a nominal aggregatesuch as
nominal GNP.I want to stress this objectivebecause I
don't think it has been pursued by most developing
countries. For most of them, price stability has been
only a second or third order objective.
Let me turn first to why developingcountries use
credit control systems.Althougha number ofdeveloping
countries,particularlyin Asia,are beginningto moveto-
ward market-basedmonetary policyinstruments, this is
still not a widespreadtrend. I see two reasons whymost
developingcountries have relied so heavily on direct
credit control programs throughout the past two or
three decades.One reasonstems from the desire to have
some control over macroeconomicdevelopments;the
second reflectsfiscalpolicyconcern.
Credit ceilingscan be appliedto lending by all finan-
cial institutions to the private sector or they can be ap-
plied to lending to both the private and government
91
sectors. Whatevertheir form, credit ceilings have been
perceivedin many developingcountries as a useful in-
strument for macroeconomiccontrol, regardless of the
exchangerate structure that is in place.This statement
should be qualifiedslightly,since it assumes that the
overallfiscalposition of the economyis not wildlyout of
control. If there is a large fiscaldeficit financedby the
provision of central bank credit to the government,
there is no hope forany control.Moreover,attempting to
maintain some sort of overall credit ceiling in such an
environment would mean a sharp shift in the distribu-
tion of credit from the private sector to the government,
with all the attendant effectson privateactivity.
Let's assume, therefore, that the authorities do have
some fiscalcontrol. If credit is not availablefrom foreign
markets (in contrast to the case that Charles Goodhart
wasconsidering),it is difficultforthe country'sresidents
to escape the constraints imposedby credit ceilings, at
least in the short run, even if there is a fixed exchange
rate. Domesticfirms, for example,would find it difficult
to quicklyadjust output in order to increase their ex-
ports to the rest of the worldand thereby acquireforeign
exchange.As a result, credit controls have often been
viewed as quite effective in restraining aggregate de-
mand.
Creditceilingshavealsobeen importantbecausethey
complement the credit allocation rules and self-con-
tainedtax-subsidyprogramsthat are oftena part of a sys-
tem of financial repression. The tax-subsidysystems
reflectthe ceilingson lendingand depositinterest rates
and the credit allocation rules that direct credit to fa-
voredborrowersor activities.The overallcredit ceilings
specifiesthe total levelof credit;whereasthe tax-subsidy
schemeallocatesthe credit across users.
Why have these monetary systems based on credit
ceilings been under pressure? There are two reasons.
First, as everyonehere is well aware,the effectsof finan-
cial repression on economic efficiencyhave become
92 Charles Goodhart
clear, and this has stimulated effortsto liberalizefinan-
cial systems. In such liberalized systems, the market
rather than the authorities determinesthe leveland dis-
tribution of credit. This has removed one of the pillars
supportingthe use of direct credit controls, which is to
ensure consistencyof the overall distribution of credit.
The second reason has been developmentof "pseudo"
capital mobilityin many developingcountries. I do not
meanby this that a country's residentscan borrowin the
international markets; most domesticborrowers do not
have access to these markets. Rather,it reflects the fact
that, over the previousdecade or two, the domesticresi-
dents of many developingcountries haveaccumulateda
large stock of externalassets. Thesecan be effectivelyre-
patriated through various means during a periodwhen
credit is severelyrestricted.No one, however,would say
that this is a goodwayto reversecapitalflight;it certain-
ly is not a way to get capital backon a permanent basis.
Even so, one often sees this phenomenon and, when it
occurs, it undermines the effectivenessof a credit ceil-
ing.
So from both the "fiscalpillar"side and the monetary
control side, the effectivenessof the credit control pro-
grams that have been so heavily used ueaDby developing
aveopn
countries is being erodedby structural changes.One re-
sponsehas been to move to market-basedmonetary pol-
icies as a way.to regain control (or totrysomething
iewsIashaiwaytoereghaibeenan cotrlem(r to tryt aswell)
But there are a number of requirements for makingthis
transition, including the freeing of interest rate and the
devlomen
o bradbasd ndliquid securities mar-
develNopmeteveryofnbroad-bsredllaillind to meetthese re-
kets. Not everycountry is reallywillingto meet
quirementormk
quirements or make the h rniin
transition atttithis stage.
tg.I In
addition,anyfiscalcontrol problemmust be addressedif
theo ranstio
sccee. isFom vrios stdie of i-
the transition is to succeed. From various studies of fi-
nancial liberalization,we know that a net loss of reve-
nues can result when the financialliberalizationoccurs.
These revenueshaveto be replaced.
Thirdly,I do not think enough attention has beengiv-
en to the requirement,whichCharlesGoodhartstressed,
for a nominal anchoring in the system. In developing
countries that are contemplating a move to market-
basedinstruments, one often hears a lot of discussionof
the technicalaspects of the move,such as how to intro-
duce marketsfor central bank securitiesinto the system.
These are important issues. However,the fundamental
problem of movingto the newsystem is the one that the
previousthree speakersaddressed:what is the nominal
anchor? Is there going to be an exchangerate anchor or
a monetary aggregate anchor? Or is another choicego-
ing to be made?This problem was never adequatelyad-
dressedunder the direct credit control system,which is
one reasonwhy many developingcountries havehad in-
flationarybiases.
A fourth requirement for successfultransitions con-
cerns the issue of whether there will be any stable rela-
tionships maintained among the authorities' policy
instruments, any intermediatetargets, and the nominal
income or price level targets they have selected. In the
UnitedKingdomand the UnitedStates, there have been
difficultieswith the stability of these relationshipsas a
result of what I would call, by developingcountry stan-
dards,second-orderstructural changes.
Finally,let me explainwhyI do not think direct credit
controls, or what I would call directedcredit, will disap-
pear from the set of tools used by central banks in devel-
oping countries. There is a fundamental structural
differencebetweenthe size of the shocks experiencedby
industrial and developing countries. While a major
terms-of-tradeshock or a move ofthree or four hundred
basispoints by international interest rates may be diffi-
cult enough to absorb for the industrial countries, they
can havefirst-ordereffectson almost all enterprises and
the entire financial sector in a developingcountry. In
such situations, central banks in developingcountries
can findthemselves facedwith a financialsystem that is
etnctve insolvent.all
effectivelymsolvent.
While the losses associatedwith these shocks could
be absorbedimmediatelyby the private sector,most de-
veloping countries prefer to absorb the loss over time
through the fiscalprocess.However,there is a real differ-
ence between howthis can be done in developingcoun-
tries as opposed to industrial countries. For example,
considerthe way in which the UnitedStates is handling
it.aig these
nre-La rss h crsi wa,i fet
and Loan and In effect, a
ts Savings
large financial shock crisis. Thehave
there crisisbeen
was,dead-weight
loses that havet tee syst eTish
osses that have to be absorbed into the system. This has
not been done by having the central bank purchase the
Savingand Loan'sbadassets;nor is a large proportionof
the losses being brought into the current fiscal ac-
counts. Instead, the bad assets are being taken into the
accounts of the ResolutionTrust Corporation,which is
financing itselfthrough the issuance of long-term gov-
ernment-guaranteedbonds. It is only as the interest and
principalpaymentson these bondsare madethat the full
losseswill be realizedover time.
Developingcountries don't have this option. They
don't have domestic bond markets where they can
spreadout the costs of a financialshock over time. They
cannot go to the external markets to borrow because,
when the shocks hit, their creditworthinessusually de-
teriorates. So how do they handle this shock?Basically,
they take it into the accounts of the central bank. It is as
ifthe UnitedStates were to combinethe balance sheets
of the FederalReserveand the ResolutionTrust Corpo-
 I Monetary Targets: European Experience
ration. If you look at the central bank's balance sheet
(with its assets and liabilitiesmarked to market) in a pe-
riod after major shocks,you will see negativenet worth,
which represents an implicit future tax liabilitythat is
associated with distributing the cost of the shock
through time.
One of the problems in developingcountries is that
the authorities havetended to imposemost of the taxes
that are needed to pay for erasingthis tax liabilityfrom
the central banks accounts on the holders of domestic
assets. This has often involved,especiallyin Latin Amer-
ica, either the inflationtax or taxes on financialwealth
and financialincome.The realizationthat holdersof do-
mestic claimsare likelyto payfor most of the cost asso-
ciated with large shocks to the economyhas become,I
wouldargue, part ofthe "receivedwisdom"that is passed
on from one generation to the next, at least in Latin
America.It is deeplyingrained in the public'sperception
of the risks of holding domesticassets, and I think it is
one of the principal reasonswhy capital flightfrom this
region has been so large.Moreover,evenwith goodpol-
icies and good financialreturns in normal periods,this
perception maypersist. It maybe hard to dispel this per-
ception until such time as another shock occursand the
government demonstrate that the holders of financial
assets will not be taxed (or taxed disproportionately)to
pay for it.
The difficultiesinvolvedin dealing with large finan-
cial shocksalso explainswhy directed credits and credit
ceilings are likely to remain a part of the "tool kit" of
Discussion
Question:Asyou said, there is a problem of coordinat-
ing monetary and fiscalpolicy.But there is also a prob-
lem of coordinating monetary policy and exchangerate
policy,or managementof exchangecontrols.How doyou
see that relation? Would you suggest that the central
bank manage both monetary policy and exchange rate
policy?Or should it be more like the Bundesbankin Ger-
many,where the responsibilitiesfor exchangerate policy
are not that clearlyin the hands of the central bank?
Mr.Goodhart:The answerto this question,in myview,
is certainly"yes."Youwould run into difficultyifyou had
a central bank that wasgiven independenceformonetary
policy while the exchangerate was determined by some-
policy, hlse.
Thereexchaebe aratnumber
wa de ined bysome-i.
bodyelse. There would number of points of conflict,
For one thing, ifthe central bank were given the priority
of achievingprice stability,it would be difficultif some-
body devalued the exchange rate against the central
93
central banks in most developingcountries. Since there
are no domestic bond markets or external markets in
which developing countries can obtain long-term fi-
nance in a crisis period, the only shock absorber they
have is the central bank. One part of the process of ab-
sorbing a shock will indeed be direct credits to specific
industriesand specificclasses of borrowersthat are pre-
sumed to be viableif they can get through the current
period.Whetherthe presumptionis true or not is anoth-
er issue.
To close, I would like to considerwhat typeof mone-
tary system most developingcountries are likelyto be
using in the future. Again,I must stress that I am gener-
alizing in an area where generalizationis hazardous at
best. I would saythat a two-tiersystem of instruments is
graduallyemerging in most developingcountries. One
tier will consist of the use of market-basedinstruments
to handle the day-to-dayoperationsof monetary policy.
These instruments may also be used to control expan-
sionary trends in the monetary base,although, as an al-
ternative, countries may very well choose to peg the
exchangerate. But there will continue to exista second
tier of crisismanagement tools focusedon direct credits
and institutional aids.
Whetherthis is goodor bad deservesa lot of analysis,
and I willnot attempt to reacha judgment on it. But I do
think that the institutional and structural differences
between industrial countries and the developingcoun-
tries will contribute to fundamental differencesin the
structures of their monetary policyregimes.
bank's wishes.For another thing, particularlywith flexi-
ble exchangerates, the movementof the exchangerate is
closelyrelated to the monetary policy.If there is to be a
pegged exchangerate, then the central bank has got to
vary monetary policyin order to maintain the peg.
In the NewZealandcase,the act saysthat the Federal
ReserveBankof NewZealandshall aim for pricestability,
with the assumption that the exchangerate will be flexi-
ble. If, on the other hand, the authorities introduce an
override to try to peg the exchange rate with another
country, then of course the central bank should-its re-
quirements must-shift to maintain that peg. Then it
cannot at the same time control the rate of inflation,be-
cause,as long as the peg is maintained,the rate of infla-
tion will dependon the inflation rate of the currency to
hc e eln spged hte.ti egdt
whlichNew Zealand IS pegged,whether Itis pegged to
94 Charles Goodhart
The same thing applies,for example,in Hong Kong,
which has its currency linked with the U.S. dollar.The
Hong Kong monetary authorities have no control over
the domesticrate of inflation.In a sense, then, if you are
going to givethe central bank independenceto achieve
price stability,you must effectivelyallow it a degree of
exchange rate flexibility.If you want an exchange rate
peg, then you cannot require the central bank to achieve
pricestability.Youeither havethe externalanchor or the
domestic nominal anchor. Central bank independence
reallyonly makes sensewhen you are looking for an in-
ternal anchor.If you havean externallink or an external
anchor, then the central bank's instruments are entirely
dedicatedto maintaining the external link.
Question:I confessI do not followthe EMSproblem.
Perhapsyou can helpme by explaininghowthis problem
would be different, and how the possible adjustment
mechanisms would be different, if it were occurring
within a country.For example,supposeit occurredwith-
in two very distinct regions of a country that had these
two different sets of characteristics. Could the same
kinds of problemsdevelop?Couldthe adjustment mech-
anisms within a national boundary perhaps differ from
those which exist across national boundaries?
Mr.Goodhart:It is in part a transition problem,in the
sensethat countries like Spain,Portugal,and to a degree
Italy,are entering a system to which they had not previ-
ouslybelonged.Within a single country,the rate of infla-
tion isn't all that much differentfrom region to region.
The wage levels can differ,but the price inflation rates
are not all that much differentwithin a country. When
the previouslyfloating currencies of two countries are
pegged,however,differentinflationrates will persist for
a while despite the peg. One will see capital inflows,
monetary expansion,wage inflationand a worseningof
the current account in the countrywith the higher infla-
tion rate.
This correspondsto a number of stylizedfacts about
the EMS at the moment. Which are the countries into
which there are large capitalflows?Whichare the coun-
tries which have had very large increases in reserves?
Whichare the countrieswhich are at the top of the band?
Which are the countries which have the greatest infla-
tion rate? Which are the countries whose current ac-
counts are worsening rapidly?Which are the countries
whosereal exchangerate is appreciatingwhile the nom-
inal exchangerate remains pegged?The answer to each
question is "the southern cone countries of Europe."
Mr.Honohan:It seemsvery like the euphoria that has
occurred in other countries, including Chile,at a time of
similar policychanges.I wonder,doesit havesomething
to do also with the publicity about 1992, which also
started to get goingaround that time?
Question:If I have it right, it seems that the problem
for southern European countries is basicallyone of the
real equilibrium exchange rate. The inflation coming
from the inflowof reservesis basicallya mechanismthat
puts upwardpressure on the exchangerate peg. If that is
so, then I don't see how a con-tractionaryfiscal policy
would help.
Mr. Goodhart: Regarding the southern European
cone, I think the concern is that the inflowsof money
will result in much less disciplineon wages than had
been hoped. So there will be less convergenceand, con-
sistent with a fixednominal exchange rate, the real ex-
change rate will appreciate.That leadsto a worseningof
imbalancesin the current account.It willtend to worsen
quite sharply,with a shift out of the tradeablegoodssec-
tor into the nontradeablegoods sector.Afiscalcontrac-
tion could help in two ways. It would put downward
pressure on interest rates and thereby slow capital in-
flows,and it would directly affectthe labor market and
slowwage inflation.
If correctivefiscalaction is not taken, the growingin-
ternal and external imbalanceswill ultimately provokea
revision, perhaps sharp, of expenditures,precipitating
fairlydrastic measures.
You can have large currency realignments occur
again, which defeatsthe purposesof those who desire to
bring the exchangerate system evercloserto irrevocable
parities. Or it will require a major change in policy.Ei-
ther of these will increasethe potential for a fairlydrastic
change or shift in the policyarrangementsat some stage
in the next couple of years.And they might have fairly
devastatingeffectson, for example,Stage 1 of the Delors
committee proposalsor the next stage of the European
monetary union.
Question:Generallyspeaking,in Latin Americathe
lowest rates of inflation occurred before 1920, when
there were no central banks. Is there anything one can
learn from this for the future?
Mr.Goodhart:Oh, yes, absolutelyindeed.The radical
ideawhose time has comeis central bank independence.
The radical idea whose time has not yet come, but may
come, is to get the government out of the monetary sys-
tem altogether.
The reallyradical libertarianssay that there has been
a closecorrelation betweenendemicinflationand the in-
volvement of the government, through central banks,

with monetary creation. They say the sole appropriate
function of the government is to define the monetary
unit-the pound, peso, peseta, or whatever-as being
the equivalentof a particular basket of commodities,of
real assets;or of some foreignasset, such as one escudo
is equal to one U.S. dollar. Then it should be left com-
pletelyto the privatebankingsector to maintain the cur-
rency's reputation by maintaining convertibility with
private bank notes, private deposits, and so on. There
should be no central bank, no lender of last resort. The
government defines the currency unit, but otherwise
gets out of the monetary creation system altogether.
That idea's time will come if central bank indepen-
denceturns out to be ineffectiveand wego on havingen-
demic serious inflation. But for the moment it's not
practicalpolitics. It is the idea of a fringe of libertarian
thinkers like GeorgeSelgin,KevinDowd,DavidGlasner,
and LawrenceWhite,whoseideasare fascinating,but for
the moment are not practicalpolitics. Central bank in-
dependenceis practicalpolitics.Gettingthe government
out of the monetary system isn't for this decade.
Mr.Honohan:If may use my privilegeas chairman to
makea glosson the point, I believeI haveobserveda sys-
tem very similar to this. It is a "real bills" system in op-
eration in Central Africa,in the Banque Des Etats De
L'AfriqueCentrale (BEAC)zone, where the central bank
has, until recently, more or less unconditionallyredis-
counted all billssatisfyingcertain arbitrarycriteria.This
means the private bankers decidehow much credit to
give out. The result is essentially "free banking." This
system fell afoul of the trap foreseenby the nineteenth-
century monetary theorists. Namely,it does not have a
nominal anchor; and there is a fallacyof compositionin
that the so-calledreal bills are not clearlyidentifiablein
advance.
Mr. Goodhart: But the libertarians' proposal is not
"real bills." It is the gold standard, but with the added
gimmick that gold is too variable as a commodityrela-
tive to general prices. So, instead of making gold the
standard, a generalbasket of goods is the standard.
Mr.Grenville:I agree that central banks could move
in the direction of greater independenceand then con-
centrate on price stability.A small country can choose
between various anchors, including the currency of a
larger country (as the smaller EMS members have
done). Speakingas a little country, as an Australian,we
have informallythought about the joys of anchoring to
another currency becausewe have seen the great power
of this "saferisland"to get us from high rates of inflation
back downto low rates ofinflationwith very little cost in
Monetary Targets: European Experience 95
terms of output. In other words, it would allow us to
jump down the long-run Phillipscurve.
With respect to what occurred when the new mem-
bers of the EMSjoined up, the point that was raised ear-
lier was about credibility.Suddenlythe new members
had credibilitythat theywould lose their inflationcosts.
We would love to have that. The reason countries like
Australiahesitate is that we think we are very different
from the countries that we mightjoin onto.
It has been suggested at times that we should hitch
ourselvesonto Japan. However,giventhat our compara-
tive advantagesare exactlythe opposite of Japan's, we
would find our exchangerates doing exactlythe wrong
thing for us. Theterms of trade wouldbe movingin a di-
rection unfavorablefor us but favorableforJapan.So the
adjustment, the softening effect,that we currently can
get by changingour exchangerate would not only be lost
but reversed.It is not becausethese ideas are a million
milesawayfrom the viewsof central bankersthat we do
not act on them. It is because, at least for the moment,
we think that on balancethis is not the wayto go. There
are softer waysof adjusting.
Of course, these considerationsreturn us to the old
idea of optimal currency areas. Optimal currency areas
are geographicregionsthat havea lot of similarity.I sup-
pose that the counter-argumentto that idea is Charles
Goodhart'sexampleof the UnitedStates,which has very
differentzones andyet managesto operatewith one cur-
rency. So there are a lot of issues here that don't quite
hold with theory.
Mr. Goodhart: One consideration that does surface
from time to time (and I would think it surfaces more
frequentlyamong WorldBankcountries)is the option of
removingthe central bank from the system as much as
possible.One can take the Panama route to this, which
was also consideredseriouslyby Israel. Youjust change
from havinga local currency to a currency in U.S. dol-
lars. In effect,then, you havelocal fiscalpolicy,but you
don't have monetary policy.
In a sense, that's what has been attempted in Hong
Kong. The problem is that, if you haven't gone too far
downthe hyperinflationroute, the only thing you can do
is to relate the currencybase strictly to the availabilityof
U.S.dollars.Even going too far downthe hyperinflation
path will usuallylead to being a dual currency economy,
in any case. There are some countries that could have
gone on a dollar base some time ago, and they would
have been much better off. But relating the currency
base strictly to U.S. dollar availability,and holding to it,
gets the central bank out of localmonetarypolicy.It gets
you onto a credible, reasonable, monetary basis and
leavesthe fiscalauthorities to do what they can.
96 Charles Goodhart
Mr.Honohan: Is it not difficult,though, to keep the
central authorities out of the system?I think of the case
of Liberia, which started with the Panama system and
then decidedto mint Liberiandollar coins.Nextit decid-
ed to mint Liberian five-dollarcoins. Verysoon there
were more Liberian coins than U.S. notes, in effectbe-
cause the authorities would not leavethe system alone.
Question:AlthoughI agreewith the independenceof
central banks and central bankers, I would like to raise a
question. Basically,someonehas got to set performance
criteria for the central bank-a job description-and
then assess performance against those criteria. Who is
going to do that and on what basis?
We can set price stability as one criterion, but it is
generallyrecognizedthat another of the central bank's
tasks is to maintain the stability of the banking system.
For example,if, in order to restrain inflation, interest
rates are raisedby 30 percentagepoints and banks fail as
a consequence,the central bank mayhave fought infla-
tion but has createdother problems.So wecomebank to
the issueof designinga role for the central bank and set-
ting its objectives.
thoe isues re ctualy bing
Mr.elloodhrt:
addressed.If you read the ReserveBankof NewZealand
Actor the various paragraphsin the report of the Delors
committee, or what will no doubt comeout of the Eco-
intergovernmental dicsin
.a.
Community'sinegvrmna
nomicnoiComnt' discussions at
the end of this year, those exactissues are currently be-
ing seriously addressed.There are legislativeacts that
includestatements of the objectivesofthe central bank.
Now,the relationshipbetween financialstability and
macroeconomic control-and there may be a conflict
between them-is indeed a serious one. A number of
peoplehaveargued that the more seriousyou think it is,
the more you might want to separate the role of lender
of last resort from the macroeconomicactivity of the
central bank. In the United States, this is done by sepa-
rating the FDIC (Federal Deposit Insurance Corpora-
tion), which provides stability against runs on banks,
from the Federal Reserve.Conceivablyyou might have
an independentinstitution to make lender-of-last-resort
loans under certain criteria; the central bank would off-
set the macroeconomiceffectof such loans through its
open market operation. It would haveone arm deciding
on a specificloan and another arm decidingto maintain
the monetary base and interest rates to affirm overall
price stability.In any case,this is a real problem that has
to be faced,although there are various waysof facingit.
Question:We have been working on some countries
wherelegislationrelatedto the centralbank's role is now
being introducedor considered.Braziland Argentinaare
two that cometo mind. Questionshavearisen as to how
they might restructure the relationship.One such ques-
tion, as an example,is what exactlyshouldbe done about
the extra-judicialliquidationfunction, which is usually
housed in the central bank. Should one keepthat func-
tion within the central bank or take it out of the central
bank and let the central bankworry about monetary pol-
icy, exchangerate policy,and so on. I am curious as to
what your thoughts are on this point.
Mr.Mathieson:I supposethe issue is what advantages
this separate institution would have over the central
bank. Whatwould be its source of funding?If you think
in terms of the ResolutionTrustCorporation,it is able to
get resources because its bonds carry the guarantee of
the U.S.governmentand can thereforeissue bonds onto
the world markets.
I don't see that any agencywill have that capacityin
developingcountries. In most developingcountries, go-
ing to the externalmarkets during a major crisis is typi-
callyjust out of the question. So the problem facedby
developingcountries is, "howdo you absorbthis type of
shock or meet this lender-of-last-resortfunction, when
you don't have the same tools of financeavailableto the
indtrial tres?" they ar aina the realmof
snd-st coies.
second-bestchoices.
In most developing countries, outcome
thebecause that the
noisother
central bank absorbs this function, in-
-stitutioni ablodos Fiscal plcyus not ibl
enough to generate the resources neededthrough, say,a
large hike in taxes or a sharp cut in expenditures. So
more or less by default, the shock is absorbed by the cen-
tral bank.
In the near term-which means about the next five
years-I don't see the situation changingfor most of the
developingcountries. Ideallythe capacityyou would like
these countries to have is what the UnitedStates can do,
which is to financeits shocksthrough time: spreadthem
out, take some costup front, take some over time. While
most of the industrial countries can do that, developing
countriesjust simplydo not havethat ability.
Mr. Caprio: But this liquidation function will be
housedwith the bank supervisoryfunction. Someargue
that, both becauseit is difficultto create independentin-
stitutions, and becausethere are some complementari-
ties between conducting monetary policy and bank
supervision,it is sensible to combinethe two powersin
one institution. But no hard and fast rules emerge from
cross-countryexperiences.

Question:I havea question on this point of spreading
costs over time. If a banking system or a bank failsin a
developingcountry, if the government then issues debt
instruments which that bank holdson the asset sideand
these instruments provide a spread to that bank over
their depositrates, does that help to addressthe issue?
Mr. Mathieson:When I say the central bank absorbs
the shock, what it reallydoes is to take over the bad as-
sets of the financialsystem. In reality,these assets often
do not earn anything, or they earn some small fraction
of their contractual return. That is the basic problem;
the central bank has issued liabilitiesof some facevalue,
but-if you want to think of it in the terms you used-it
has taken on assets that, although they have the same
facevalue, their real value is much less.
So the central bank has already taken a loss. If its as-
sets and liabilitieswere markedto market, it would end
up with a negativenet worth in its balance sheet, which
is goingto be workedout of the system in some sense. It
may be done through the inflation "tax," which erodes
may bealdonuetrou those "
outstandinfliation .s hi erods-
w will for a long time continue to use such tools does not
the real valueof those outstanding liablities. Theminis-
ters of finance and presidents of central banks in devel-
oping countries are quite good at thinking up different
waysto accomplishthis "workingout" process.Unfortu-
nately,most of it ends up affectingthe whole range of fi-
nancial instruments. But I really don't see them having
much flexibilityhere. For large shocks-to terms of
trade or somethinglike that-they just do not havethe
absorptive capacity except to take it onto the central
bank's books.
Mr.Honohan:When you say "takingit on the central
bank's books,"I take it that you do not necessarilymean
the actual accountingtreatment but more that the cen-
tral bank is goingto have to take care of the problem? It
can do so through any of a variety of measures. It can
print money;it can lower reserve requirements; it can
set interest rate ceilings. These are the only actions
availableto it. In other words, there must be some taxes
but, becausethe fiscalauthorities do not havethe means
to raise taxes, the residual course of action is to go
through the central bank.
Mr.Mathieson:Whatyou describeis not the only pos-
sible outcome. There could be a gradual transfer from
the fiscalauthorities over time. They could generate tax
revenuesthat couldbe used to buy up the bad debts from
the central bank. This would, in a sense, unwind the
shock through future taxes. So it does not necessarily
have to comethrough an inflation"tax."Althoughit of-
ten does, it need not.
Monetary Targets: European Experience 97
But my point is that this is a question of financing
technique. Right now,it is the only financingtechnique
for a large shock. If these countries really had accessto
international markets, then presumablythis would give
them an alternative way to finance a shock over time.
But for most of them, the accessis just not there, and I
doubt it will be there for most of the 1990s.
Mr.Caprio:Itis probablyworth addingthat, although
industrial countries havedeeper capitalmarkets that al-
low them to "stretch out" more easily various shocks,
they continued in the 1970sand 1980sto resort to direct
controls when they thought that indirect methods
would take too long to operate or prove insufficientto
the job. Thuswe saw"the corset"in the UnitedKingdom
in the 1970s,various credit control schemes in Italy in
the 1970s and 1980s,and a short-lived credit control
scheme in the UnitedStates in 1980.In either develop-
ing or industrialcountries, such programsmay be a mis-
take in that they inhibitthe efficientallocationof credit.
Thus Dr. Mathieson'spoint that developingcountries
ipyta nietmtossol o edvlpd
imply that indirectmethods should not be developed.
Rather, the inefficiencies of direct controls should
hemghtenonets resolveto assst in the movementtoward
market-basedmethods of monetary policy.
Mr.Queiroz: If you have real central bank indepen-
dence, absorbinga fiscalshock will remain the responsi-
bility of the fiscal authorities and the government,
because the central bank would not be financing the
government in any way.On the other hand, ifthe central
bank is not independent,it willtend to assume responsi-
bility for absorbing the shock. So the govemment in
general has a great interest in not having an indepen-
dent central bank.
Theextremecaseof central bank independencewould
be a bank that never financed any government paper,
only private paper. This creates an adversarialrelation,
with the central governmenton one sideand the central
bank on the other side. This is a war, a terrible war, the
worst war betweenthe two that could occur.In this case,
there wouldbe no inflationin any country in the world.
So my questionis, whichis the country,who are the pol-
iticians, the congress, the president, the first minister,
that would like to havean enemy of this kind?Nobody.
Mr. Goodhart: Well,the DeutscheBundesbankis in-
dependent.Although,we've seen in recent months the
limits of that independencewhen you havesomethingat
stake that is as important politicallyas German mone-
tary union. Where I do think you are right is that there
is a feelingamong commentatorsand economists-now
98 Charles Goodhart
quite fashionable-that central bank independencewill
help to delivergreater price stability,in part becausethe
Bundesbankhas been the most successfulcentral bank.
I think that your argument can also be reversed:one
tends to get central bank independencewhen the popu-
lation as a whole, and the politiciansas a whole, put the
highest priority on price stability. If they did not, they
would not make the central bank independent.So what
is reallycreating the price stabilityis the shift in priori-
ties of the public as a whole, not the constitutional
change to the central bank. Ultimatelythere will always
be some kind of politicaloverride of central bank inde-
pendence. There is a political override in the Reserve
Bank of NewZealandAct.And I have difficultybelieving
that most of the Europeancountries willaccepta "Euro-
Fed" (EuropeanFederalReserveBank)that is subject to
no politicalcontrol under any circumstances.But that
degree of independence is what the Delors committee
actuallyadvocated.
An Elaboration
Editors'note: The followingexchangetook place during
the questionand answersessionof BarryJohnston's pre-
sentation. It is reported here becauseit directlyrelatesto
the content of this session.
Question:Under the heading of helping the captain
steer the ship, Charles Goodhart suggested yesterday
that narrow monetary aggregates are not particularly
useful in guiding the ship. I would like to giveyou the
opportunityto rebut him, becauseI know that you have
done a lot of work on that.
Mr.Johnston: On the first point, I think Charlesand I
havehad some long-standingdisagreementson the mer-
its of differentmonetary aggregates.Charles has always
felt that broad moneywas the only reasonablysensible
aggregatein the economy.I believethe main reason for
this view is that broad money can be broken down into
credit counterparts; one can analyze broad money in
terms of the absorption of funds by the government, by
the banking system,and so on.
Myapproachhas alwaysbeen more empiricallybased.
It addressessuch questions as the underlyingstabilityof
the aggregatesthat are being examinedand the informa-
tion content, in any sense, of those aggregates. The
movement to MOin the United Kingdom, to which
Charles referred, was based on an extremelyextensive
study of all the monetaryaggregates.This studyalso in-
cluded measures of money not usuallystudied or pub-
lished,becausewe were trying to findwhat measurewas
the most stable.The results showedthat the demandfor
currency was consistentlythe most stable monetary ag-
gregate in the UnitedKingdom.
There werevery goodreasonsforthis result. Allof the
broader aggregates,includingMl, had been distorted by
significant financial innovation. In Ml, for example,
there had been the growth of interest-bearingsight de-
posits. The broader aggregateshad been affectedby the
developmentof the building societies. In the U.K.con-
text, the liberalizationof restrictionson creditexpansion
increasedthe demandfor holding broad money.
As I suggested in the presentation, the demand for
money,at least for broad money,dependsnot just on in-
come but also on broader portfolio restrictions. When
access to credit increases, the demand for money in-
creasesas wellas the demandfor other assets. This will
shift the demandfor money.This was indeedshownem-
piricallybythe U.K.experience,the detailsofwhichhave
been published.
Mr.Grenville:AsI understandwhat wassaidyesterday
by Mr.Goodhart,as an empirical,informationalvariable
to showwhat income is doing,a narrow monetary mea-
sure might be fine. But for a monetary aggregateto be
operationallyuseful, there needs to be a causalsequence
that beginswith M and finisheswith Y (in the quantity
identity), not the other way around. What the tests of
stability show is simply the informational content. I
know that the reason weget such good relationshipson
the narrow aggregatesin Australiais that we do not op-
erate policiesthrough them; when people demandcur-
rency,we simplygiveit to them. The critical thing is not
whether they are stable now but whether, if we try to
push on them, they would remain stable.
Mr.Johnston: I think the fundamentaldifferencelies
in the waythe monetarypolicyoperates.If the policyop-
erates through discriminatory restrictions, then your
aggregateswill not remain stable. The experiencein the
UnitedKingdom is that if policyoperates on an instru-
ment that is not restrictedor discriminatory,such as in-
terest rates, then the narrow aggregatesdo not become
structurallyunstable. Indeed,the experiencein the Unit-
ed Kingdomhas been that, although MOhas been target-
ed since 1984,the results still show that aggregatehas
maintainedits propertiesdespitehavingbeenthe princi-
pal monetary target.
I think the real problem with the broad aggregates
was that they used things like the corset, which really
made aggregatesrather meaninglessand led to a large
amount of disintermediation.While there has been a
trend or structural change in the currency aggregate
(MO),it has been takeninto account byincludingvarious
variables.

I should also point out that the choice of MOwas
basednot just on stabilityofdemandbut alsoon findings
of very extensive causalitytests. The work looked at a
very widerange of moneycomponents.I ran them in ev-
ery conceivablewaythat we could think of.We ran very
large statistical testing over subperiodsand for different
definitionsof the equations.Nevertheless,again the re-
sult was still that currency systematicallyturned out to
have the best information content of all the aggregates
we examined.
Question: One explanation for your results may be
that as interest rates go up, the economyslowsand the
narrow aggregates also slow. The narrow aggregates
slowbecausethere are financialflowsinto interest-bear-
ing securities. So one can find lots of reasons why the
Grangertest might be passed.The point that strikes me
is that there are such good substitutes for currency. If
you reallywant to press on that-and that is what mon-
Monetary Targets: European Experience 99
etary policy is all about, pressing.... some aggregate-
then peoplewill findwaysaround it.
Mr. Johnston: But how do you press on currency?
Certainlyyou do not stop people from coming to the
bank and withdrawingit. I wouldagree with your point
if pressing on the aggregatemeant that you said to the
populace,"Youcannot have any more currency today."
But raising interest ratesworks on the demandformon-
ey function, which is stable and does not deteriorate.
What you said about the Granger causality is precisely
why it is MOthat is targeted. The transmission mecha-
nism is the effectof interest rates on the economy.No-
body ever believedthat cutting back currency actually
affectedanything; it was designedsimplyas an interme-
diate target. This, I think, is the crux of my difference
with Charles.He feelsthat moneyshould do something,
not just be an indicator.
 PART III

BUDGETDEFICITSAND
MONETARY POLICY
 Introduction
Of the pressures which can cause central banks to lose
control over their balancesheet and thereforeover mon-
etary conditions, perhaps the most common is the need
of governments to finance large budget deficits. How
monetary policy has functionedwhen confronted with
large public sector deficits is the subject of this section.
Speakers recounted the experience of Italy, Chile, and
Brazil.
The late CesareCaranza,former ExecutiveDirector of
the World Bank and previouslywith the bank of Italy,
summarizedthe lessonsfrom Italy.Double-digitdeficits
(asa percentageofGDP)havebeen the rule in Italy since
the mid-1970s,and government debt-most of it held
internally-is equivalentto one year'sGNP.The author-
ities' abilityto market governmentpaper (helpedby high
household saving)has greatly helped to maintain a suc-
cessful anti-inflationary monetary policy. They have
been able to broaden the debt market by using instru-
ments of a variety of maturities and in small denomina-
tions. Indeed, this "structural" approach to monetary
policy,which is argued to have been essential to the
Bankof Italy'ssuccessin restraining inflation,shouldbe
of interest to those officialswhobelievethat onlycaptive
markets can be relied on to finance large deficits.
Togetherwith the shift from a pay-as-you-goto a fully
fundedsocial securitysystem,the indexationof all finan-
cial contracts accounted for much of the successin de-
velopingmedium-termand long-termfinancialmarkets
in Chile, according to Juan-AndresFontaine, who was
formerly Director of Researchat the Banco Central do
Chile. Indexationis controversial;some believeit weak-
103
ens the resolveof governmentsto control inflation.(Also
controversial have been debt-equity swaps, of which
Chileis the leadingexponent.Fontaine explainedwhy,as
operated by Chile, these swapsneed not have an infla-
tionary impact.)
Recentmonetarypolicyin Brazil,as describedby Car-
los Alberto Queiroz, Chief of the Open MarketTrading
Departmentat the Central Bank of Brazil, has had the
more limited objectivesof stabilizingbond market con-
ditionsand avoidingfinancialdistressof intermediaries.
The use of floatingrate instruments with dailyrevision
of yields has helped tide the Brazilianfinancial system
until budgetary control is restored enough to allow
monetarypolicyto resumethe role of stabilizingprices.
Theindependenceof centralbanks facedwith govern-
ment deficits has dependedon a widevariety of consid-
erations. Despite the 1981 "divorce,"which ended its
role as automatic purchaser of unsold treasury bills at
each auction,the Bancad'Italia is endowedwith relative-
ly limited de jure independence,inasmuch as monetary
policyis set by an ministerialcommittee chaired by the
TreasuryMinister.Yet its high professionalreputation
and its relativestabilityin the rough seas of Italian poli-
tics haveaccordedit considerablede factoindependence.
Dr. Fontaine underlined the role of greater central
bank independencein the Chileancontext, with the en-
actment in 1989 of a new law prohibiting the central
bank from lending to government;the lawcould proba-
bly be evadedbut only at a cost. The Braziliansituation
leavesthe central bank with less independence,at least
for the present.

Impact of Government Deficits on Monetary Policy:
The Caseof Italy
Cesare Caranza
Italy makes an interesting case for study becauseof all
the potential for conflict between fiscal and monetary
policy.With a public debt that roughlyequals GDPand a
fiscaldeficitof about 11percent of GDP,this conflictis a
fact of life.
I shallbeginwith somebasic factsand figuresthat are
essentialfor understanding the Italian experience.Then
I shall describethe policy responseof the monetary au-
thorities, the strategy they followedto finance these
growing fiscal deficits while maintaining a reasonably
sound monetarypolicy.Thirdly,I shallreviewthe results
of these policiesand the problemsstill unsolved.Finally,
I shall sketch the prospectsfor the future.
The Economic and Financial Picture
Webegin, then, with the factsand some figures. Dur-
ing the last two decades,Italy has experiencedhigh and
continuouspublic sector deficitsand a ballooningpublic
debt.At the same time, it has had a restrictivemonetary
policyaimedat maintainingthe exchangerate ofthe lira
within the EMSband. Althoughthe lira has been deval-
ued from time to time with respectto the German mark,
those changes did not compensatefor the inflation dif-
ferentialwith our major trading partners in Europe. In
effect,Italy maintaineda relativelystrong exchangerate
policyto reduce inflation to the level prevailing in our
major Europeantrading partners.
This objectiveof reducing inflation has been largely
achieved.Consumerinflation peakedin 1980at 21 per-
cent. From that extremelyhigh levelat the beginningof
the decade, it has declined to an average over the last
three years of between5 and 6 percent. The latest figure
is 5.8 percent, but some slowing is expectedover the
next fewmonths.
At the same time, real growth improved.In the first
half of the 1980s, GDP grew at an annual average of
slightlyless than 2 percent; in the second half of the de-
105
7
cade, it grew at slightlymore than 3 percent. Moreim-
portantly, the productive sector of the economy
recoveredduring the 1980s from the dark years of the
1970s.Private enterprises regainedproductiveefficien-
cy,competitiveness,and financialstrength.
Meanwhile,the status of public finances deteriorated
sharply.A few numbers will conveythe basic scenario.
The public sector borrowingrequirement (PSBR)aver-
aged 13 percent during the 1980s. From 14 percent in
1983,it decreasedto the present level of about 11 per-
cent. For the next fiscalyear,the governmentis aiming
to reduce the deficit to about 10 percent, but that goal
mayproveelusive.Asa consequenceof these continuing
high fiscaldeficits,the stock of public debt grewrapidly.
This year it will roughly equal GDP.Interest payments
on public debt grewfrom 5 percent of GDPin 1980to 9
percent this year.Therefore,of the 11 percent I quoted
earlier as the PSBR,9 percent representedinterest pay-
ments. Primary deficits (that is, net of interest pay-
ments) climbed to 6.5 percent of GDP in 1983, then
decreasedgradually to 1 percent this year.The primary
deficit,which is a keyelementin the story,has been cut.
Yetthis improvementhas beentoo slowand half-hearted
to break the viciouscircle of high deficits,higher debt,
higher interest rates, and still higher deficits.That, in
brief,is the scenarioof the 1980s:a dramaticallyimprov-
ing private sector; satisfactoryreal growth; a sharp re-
ductionof inflation;but deteriorationin public finances.
A New Strategy for Monetary Policy
Bywhat strategyhave the monetary authorities tried
to reconcilethe financingof this growingdeficitwith an
acceptablemonetary discipline?The monetary authori-
ties, and the central bank in particular,did not imple-
ment just a quantitative monetary policy, aimed at
control of the monetary base and control of aggregates.
They tried to implement a structural monetary policy,
106 Cesare Caranza
one that would developthe financial structure of the
country in order to copewith the huge demandfor funds
from the public sector, while also modernizing the in-
struments of monetary control. They sought, of course,
to keepthe creation of base moneyunder control.Atthe
same time, they sought to change the rules of the game
forthe financialsector.Theytried to deepenand broaden
the financialmarkets,to channel savings,and household
savingsin particular,into public securities.
A positive peculiarity of the Italian situation is the
high savingspropensityof its households.It is important
to assessthose high fiscaldeficits of 11 or 12 percent of
GDPagainst the high savingsin the other sectors, sav-
ings of between 20 and 21 percent of GDP.If a channel
can be activatedto transfer a major part of these savings
into funding of the public debt, money creation can be
reducedand kept under control.
In pursuit of more depth and breadth in the markets,
the authorities introduced an enormous amount of fi-
nancialinnovation.They floatednew kinds of securities:
financiallyindexedsecurities,short-term and medium-
term securities.Many new instruments appearedin the
financial markets, as well as new intermediaries.Some
years ago, Carlo Cottarelli and I wrote an article on fi-
nancialinnovationin Italy in whichwe calledit a lopsid-
ed process. Although there was all this financial
innovationduring the 1980s,it was lopsidedbecausethe
public sector was the major, and almost the only, inno-
vator. The reason was simply because its appetite for
funds was so huge that it had to find new sources of
funds.
So this broadening of the markets, to channel more
privatesavings into public debt,was the first major ele-
ment of the monetary strategy.The central bank in par-
ticular also tried to improve the instruments of
monetarycontrol,to regainsome maneuveringroom for
interest rate policy.One ofthe more significantdevelop-
ments in this direction was the "divorce"in 1981 be-
tween the Treasuryand the Bank of Italy,wherebythe
centralbank discontinuedthe commitment to act as the
residual buyer of treasury bills at auctions of public se-
curities.This divorce,whichstrengthenedthe role ofthe
central bank in setting short-term interest rates,
wrought a crucial change in the conduct of monetary
policyin Italy.
Results of the New Policy
These were the broad directions of policy; now we
cometo the results. The effortto channelprivate savings
into public debt wasa major success.The distributionof
public securities in the economychanged dramatically.
In the late 1970s,only about 20 percent of publicsecuri-
ties in circulationwas held in private sector portfolios;
by the late 1980s, this increased to more than 70 per-
cent.
Of course, concomitantwith this increase in the pri-
vate sector holdingswas an equallydramatic decreasein
the share of public securities held by commercialbanks
and the central bank. This redistributionmeant lesscre-
ation of base moneyand broad money.Andin fact,mon-
ey creation deceleratedthroughout the decade.In 1984
the Bankof Italybeganto announce moneytargets, such
as growthtargets for M2and for creditto the private sec-
tor. Although the credit target was usuallyovershot,the
M2 moneytarget was almost alwaysattained. The pre-
dominant tool for controlling moneygrowth was inter-
est rate policy, as credit ceilings on bank loans were
removedin 1983.
Real interest rates had, for the most part, been nega-
tive in the late 1970s. After the credit and monetary
crunch at the beginningof the 1980s,real interest rates
jumped to relativelyhigh positive levels. Over the re-
mainder of the decade,real rates remainedat a high lev-
el, although they showeda slight decliningtrend from
the earlyhighs. Even today real interest rates in Italy re-
main a bit higher than in other major industrial coun-
tries. This differential,by the way, explains the huge
influxof capital into Italy during the most recent years.
Theserelativelyhigh real interest rates notwithstanding,
Italy continued to enjoya very goodperformanceof pri-
vate investment.
I havealready mentioned that inflationwas reduced.
And last but not least among the major results of the
strategyhas beenthe gradual liberalizationof foreignex-
change controls. We took the final step just this week;
since May14, 1990,capitalcontrols havebeen complete-
ly removed,evenon short-term capital movements.The
EMShad agreed to remove these controls by the begin-
ning of July. France made the last step at the beginning
of this year.NowItaly has taken this last step. This liber-
alization is a crucial element of the story;from now on,
our financialmarketswillnot be protectedfromexternal
competitionby artificialbarriers.
This increasingfinancialopennessofthe Italianecon-
omy has thus far brought huge capital inflows.The rea-
son, simply enough, is the higher real interest rate,
relativeto other countries,combinedwith the prevailing
expectationof exchangerate stability.For foreign inves-
tors, Italian financial markets providegood buys. There
is a plentifulsupplyof bonds,treasury bills,and treasury
certificates,all with relativelyhigh yields.
 Impact of Government Deficits on Monetary Policy: The Case of Italy
The Remaining Problems
The results summarizedabove represent the positive
aspects of the Italian situation. But the coin has a dark
side as well.Of the two problems I want to discuss, one
is more technical, the other is more general.
The technical point concerns the implicationsof the
shortened average maturity of Italy's public debt. To
market that enormous amount of debt to the privatesec-
tor during a period of high and variable inflation rates,
the Treasuryhad to accept a major shortening of matu-
rities. This occurred not only as a direct shortening, by
which I mean an increasingamount of treasury bills,for
instance, but also as an indirect shortening. The Trea-
sury issued large quantities of so-calledtreasury certifi-
cates with maturities of between four and seven years.
However,their yieldis linked to short-term rates, to the
treasury bill rate, so the interest payments on them re-
flect the movement of short-term rates.
Todaythe average maturity on public debt is some-
thing less than three years.That is not drasticallyshort,
although it is much shorter than the average maturity
some years ago. Still, a shorter average maturity of the
publicdebt createsspecialpressureson monetarypolicy.
At the same levelof debt and of annual deficit,a shorter
averagematurity means that the debt stock must be re-
cycledmore frequently.Eachmonth, Italy'spublicissues
equal5.0 percent of GDP,which is ten times the average
stock of bank reserves.This high rate ofturnover consti-
tutes a major threat to monetary policy.If a treasury bill
auction is not completedsatisfactorily,a serious prob-
lem of monetary control ensues. If not enough bills or
bonds are sold in a month, bank reservesare out of con-
trol. So refinancing becomes a continual and delicate
game.
For the economyas a whole, this situationrepresents
a high degreeof financialfragility.The systemis acutely
exposed to external shocks coming from abroad and
from changingexpectationsin domesticmarkets. This is
one cost Italy has paid and will continue to pay for this
delicate equilibrium,which has so far been maintained
but remains a cost and a threat for the future.
The second and more general point is that Italy has
had a primarydeficitsince 1965.I remarked earlierthat
107
the annual deficits grew in the late 1970s and early
1980s;in fact, for the last twenty-fiveyears interest pay-
ments on the public debt havebeen coverednot bytaxes
but by the issueof new securities.In this situation, pub-
lic debt is no longer "irrelevant"in the Ricardo-Barro
sense. The private agents in the economyconsidertheir
holdingsofpublic debtas net wealthand its stream of in-
terest payments as part of their disposableincome with
expansionaryeffectson their spendingdecisions.
The Keynesian description of interest payments as
measuringthe conflict between producers and rentiers
does not hold in this situation. The absenceof a conflict
of interest between the taxpayersand the recipients of
interest paymentsweakensthe pressure of public opin-
ion on Governmentand Parliamentto cut the deficitand
act for the consolidationof public finance.
Potential Solutions
What solutionsto these problemsare possible?For a
number ofyears now,the central bank has been insisting
that the primary deficitmust be cut. Indeed,the deficit
should be reversed into a primary surplus, as other Eu-
ropean countries with high ratios of debt to GDP did
during the 1980s.Belgiummovedto a primarybudget
surplus in 1984; Ireland in 1987. If this first step is
achieved,one can expecta reinforcingcircle of improve-
ments. Lessdemand pressure on the financialmarkets,
becauseof the reduced deficit,would lessenpressure on
interest rates. Since interest payments at short-term
rates are a major componentof the deficits,lowerinter-
est rates would further lower the overall deficitand so
on. It is easy to say that the starting point for readjust-
ment should be cutting or even reversingthe primary
deficit,but it is difficultto do. In the end, the issues are
sociopoliticaland require a sociopoliticaldecision.If the
country reallywants to do this, it must choose to pay
more in taxes or to cut public expenditure.
Italyhas latelymadesomeprogress in the right direc-
tion. Thegovernmentplansto reducethe primarydeficit
to zero for the next fiscalyear.That forecastseems rea-
sonable.In any case,the only way out is to move ahead
in that directionand to move reasonablyfast.

Impact of GovernmentDeficitson MonetaryPolicy:
The Case of Chile
Juan Andres Fontaine
The Chilean experience is interesting because of the
manychanges that havebeen introducedover the last 15
years in financialmarketsand monetarypolicy.We have
movedfrom a repressedfinancialsector with substantial
govemmentintervention to something much more lib-
eral. Financial markets have grown and become much
deeper.Chilehas movedfrom high inflation-nearly 400
percent per year in the mid 1970s-to about 20 percent
a year,which is moderateby LatinAmericanstandards.
During this process, the country has undergone some
traumatic experiences.The fact that I am talking here
nowshowsthat a country can survivethose experiences.
One such traumatic experiencewas a severefinancial
crisis in the early part of the 1980s.It was solvedwith a
rescue plan in which the Central Bank took high losses.
These losses have created a quasi-fiscaldeficit in the
Central Bank, and I shall explain how this quasi-fiscal
deficitis being managed.
Asa starting point, I shall describea few institutional
featuresof the Chilean financialmarket that are closely
linked to this process of financial liberalization.These
features should be of interest, since many developing
countries have financialsystems similar to the Chilean
systembeforethe mid 1970s.
Factors in Chile'sFinancial Liberalization
The keyto successfullyliberalizinga financialmarket
is creating confidence.All financial markets depend on
confidence.The fundamental problem in the Mexican
situation, which was raised in the previousfloor discus-
sion, is exactlya lack of confidence.One way to create
confidence is to lift all exchange controls. Doing so
means granting free exit to those confident enough to
risk their capital on the domestic markets. Typically,
however,lifting exchangecontrols implies a substantial
devaluationof the domesticcurrency and a correspond-
ing reduction in real wages. For this reason, govern-
109
8
ments in countries with high inflation or with high
uncertainty are reluctant to followthis route.
In Chilein the mid 1970s,the alternativeof removing
all exchange controls was not even considered.Liberal-
izingthe financialmarket byopeningup the capitalmar-
ket to international financial flows was much more
controversialthen than now,after the experienceof Eu-
ropean countries such as Italy, Spain, and France. In-
stead,we took a rather long route that has createdsort
problemsbut, so far at least, is working.
That route had two basic elements. One consistedof
some legal changes in the status of the Central Bank.
The other was the creation of an efficient indexation
mechanism.
The legal changes in the Central Bank lawprohibited
it from lendingmoneyto the government,to the public
sector enterprises, or to any entity connected with the
government. That prohibition was enacted in the mid
1970sand has remainedin place.Althoughwaysofgoing
around this typeof legal restriction can alwaysbe found,
I think it playsan important role becausethose waysare
typicallycostly.Someone has to figure out how to do it
and take the risks, including the legal risks, of getting
around the prohibition. So I favorthis solution of simply
closing the legal window to Central Bank financing of
the government.Of course, once that windowis closed,
either the fiscaldeficitmust be reduced or it must be fi-
nanced by other means. This prohibitionwas taken to
the constitutional level in the 1980 constitution. Now
the Central Bankis constitutionallyforbiddento finance
the public sector.
Just last year,a new law governing the Central Bank
was enacted. It gavethe Central Bank independenceon
the model of the GermanBundesbank.This is a contin-
uation in the same directionas the earlier prohibitionon
public sector financing. Although the law is quite new
and the Central Bank is just starting to work indepen-
dently,things appearto be on the right path.
110 Juan Andres Fontaine
The secondchange to which I referredwas the index-
ation system. Whether this is a good solution or not is
still highly controversial. Chile had the problem of a
long history of very high inflation.Awayhad to be found
to create a financialmarket in that environment.So we
decidedto try to make financial markets function as if
inflation did not exist. The indexationmechanism that
was created was based on a unit of value, called the UF,
in whichalmost all financialassets and loansare denom-
inated. The UF changes its value daily,accordingto the
inflationof the previousmonth expressedas a dailyrate.
In essence,this nearly perfectmechanismcreated a new
currency,the UF,whose real value is constant. It is now
widelyused for most medium-termand long-term con-
tracts.
Creationof this mechanismmadepossiblethe devel-
opment of the medium-term and long-term financial
markets. Nowthe Treasuryand the Central Bank issue
paper,which is indexed,with a maturity of ten to twelve
years.Privatebanks issue mortgage bonds with maturi-
ties of twelveto twentyyears to finance housingloans of
equivalent maturities. Private firms are issuing long-
term bonds. So the market is functioningas if inflation
did not exist.
Of course, one must be aware of the risks of this sort
of institution. It may create some "inertial" inflation as
manycontracts start to be denominatedin this new cur-
rency. However,I do not agree with Mr. Caranza'sposi-
tion that this indexationnecessarilyreduces the political
pressures to fight inflation. The Chilean experience
shows just the oppositeas there are many debtors who
are affectedby the indexation.For example,when mort-
gage debtorshold housing loans denominatedin UF,the
monthly paymenton the loan is indexedto the consum-
er price index.In this situation, any increasein inflation
becomespoliticallya very difficultsituation. So in Chile
I think we havecreateda built-in mechanismto fight in-
flation, because so many individualsare personallyaf-
fected.
This mechanism of the UF has another effectthat is
more risky.It has made the demandfor money,for MO
and MI at least,very volatile.I shall addressthis in more
detail below.
A third institutional factor worth mentioning con-
cerns changes to the social securitysystem.In the early
1980s,the pensionsystem in Chilewas transformedsub-
stantially.It went from a standard pay-as-you-gosystem
to a fullyfunded private pension system.So the payroll
taxes that previouslyhad been collectedby the govern-
ment to finance pensions are now being channeled to
the capital market.
Becauseof this far-reachingreform,financialsavings
increased dramatically.Financial savings, which com-
prises not only M2 but also governmentbonds and sim-
ilar instruments, grew from 27 percent of GDP just
beforethe reform in 1980to 60 percent of GDPlast year.
This changewas inducedin large part bythe pensionre-
forms, although there were also other factors that con-
tributed. The stock of financialassets held by the newly
created private pension funds now amounts to 15 per-
cent of GDP.The funds havebecome long-term institu-
tional investors whose emergence has helped a great
deal in developingthe market.
This change would probably have been impossible
without the UFindexation,There had to be someway to
maintain a fullyfunded pension system in a high infla-
tion environment.
Managing Monetary Policy in Chile
Howis monetarypolicyconductedwithin this frame-
work?The Chileaneconomicpolicyof the last ten or fif-
teen years has gained a reputation for orthodoxy and
monetarism.In actual practice, monetarypolicyhas not
been monetarist at all, even though most of the econo-
mists in the Central Bank-myself included-have stud-
ied in Chicago.For the past fiveyears, monetary policy
has basicallybeen conducted through interest rate tar-
geting, which is somewhatheretical in the monetarist
world.
Our guiding principlehas been that the role of mac-
roeconomicpolicy,and specificallyof monetary policy,is
to provideinformationto the private sector.It is basical-
ly an informationalrole whoseaim is to guide the spend-
ing decisions of private economic agents so that
aggregate spendingwill be consistent with the produc-
tion capacity and the availabilityof foreign financing.
Our view has been that real interest rates conveymore
information than nominal monetary aggregates.This
seems especiallytrue when financial markets are rela-
tivelyyoung and it is difficultto determinethe right rate
of growth of monetary aggregates.
Everybusinessmanunderstandsvery wellwhat a real
interest rate means.The messageis clear.Wehavedone
this within the frameworkof a highly indexedfinancial
system, in which market-determinedreal interest rates
are evident.The criticismsof interest rate targeting tend
to be arguments against targeting of nominal interest
rates. Theseargumentsmaywellapplyin marketswhere
there is no indexation,but our UF-denominatedassets
allowagentsto see the real interest rates clearly.This en-
ables the CentralBank to target real interest rates. This
may be a capabilityunique to Chile.
There is a second reason whywe have focusedon in-
terest rates.AsI mentioned,the variationsofthe UFtend
to render the demand for MOand Ml very unstable.
 Impact of Government Deficits on Monetary Policy: The Case of Chile
Whenthe market incorporatesa nearly universal index-
ation system,nominal interest rates, whichare the argu-
ment in the demand for money functions, behave
according to a modifiedFisherian equation. Instead of
the nominal interest rate being equal to the real rate
plus expectedinflation,the nominal rate equals the real
interest rate plusthe changein the UF.Thechange in the
UFis determined by the previousmonth's inflation,and
month-to-month inflation rates fluctuate a lot. So the
nominal interest rate, which is alsothe cost of holding
demand deposits or currency, also tends to vary a lot.
Andthis renders the quantity of money demandedvola-
tile too. Under these circumstances,targeting a mone-
tary aggregatebecomesimpossible.
The way we do our interest rate targeting can be
viewedas a combinationof the Pakistani and Canadian
methods, as these were explainedby Paul Meek. The
Central Bank issues paper in a fixed-ratetender, so it is
prepared to sellwhatever amount of paper is demanded.
The reverseaction of adding liquidityto the market is
decidedcaseby case.Wedonot havean openwindow.In-
stead, we enter the markets and provide liquidity
through short-term and medium-term loans, a method
similar to the use of depositsin the Canadiansystem.
These aspects of monetary policy in Chile all have
some bearing on the quasi-fiscaldeficit I mentioned in
the introduction. As I said, Chile experienceda deep fi-
nancial crisis in the early 1980s.It was deep enough to
arouse fears that the whole financialliberalizationpro-
cess would be derailed. The Central Bank adopted and
careds woutd aestratlegyoferesug BanksandodebtosIn
carried out a strategy of rescuing banks and debtors. In
pursuit of this strategy,the Central Bank took on large
lossesfrom three sources:(1) liquidationof some banks;
(2) recapitalizationof the remainingbanks through sub-
ordinatedloans payinginterest at less than market rates;
and (3) acceptanceof the exchangerate risksassociated
with Chile'shigh foreigndebt, at a time of major deval-
uation of the peso.
In short, the strategy succeeded but was extremely
costly for the Central Bank. It created a Central Bank
deficitwhose preciseamount has been difficultto com-
pute becauseit dependson such things as the probability
of those subordinatedloans being paid.Anyway,this def-
icit has been estimatedat around 3 percent of GDP.
What is the implicationof this deficitfor the macro-
economicsituationin Chile?In the past, this issueof the
Central Bank's quasi-fiscaldeficitwas disregarded;now
it is receivingconsideration.Andyet, perhapsthe pendu-
lum has swung too far. For two reasons, the Central
Bank deficitcannot simplybe added onto the standard
fiscaldeficit.First, there is an accounting problemwith
whether costs are taken on an accrual basisor upon re-
alization. Second,the economicpoint is that much of
II1
the Central Bank quasi-fiscaldeficitis a consequenceof
decisionstaken in the past. Whatappears at the present
are just the financial implications of those decisions.
Thus, these amounts are not comparableto the current
subsidies,taxes, and other economicdecisionsthat are
typicallymanaged in the nonfinancialfiscal sector. In
other words, the standard fiscal deficit is directly rele-
vant for ascertaining fiscal influenceon aggregate de-
mand. In terms of effects on aggregate demand, the
quasi-fiscaldeficit, which comes from decisionstaken
perhaps ten years ago, cannot be simplyaddedonto the
standard fiscaldeficit.
This high quasi-fiscaldeficitin the Central Bankdoes
create some problemsfor monetary policythat are sim-
ilar to those describedfor Italy in the last presentation.
It generates a large stock of Central Bank debt in the
market, on the order of 19 percent of GDP.Most of it
used to have a maturity of 90 days,so amortization of
those debts posed a difficultproblem in terms of mone-
tary management.For example,there wasa time several
weeks before the 1988 plebiscite in Chile when banks
were reluctant to take a 90-dayposition. Becauseof the
political uncertainty,they wantedto remain liquid.As a
result, there was a huge expansionin the monetarybase,
since this mismatching automaticallyshows up in the
monetarybase.
That sort of liquidityproblem tended to recur until
the beginning of this year. We decided to address the
problemby placinglong-term bondsat a high premium.
Wehave been able to swapmost of the short-term paper
folngtrbnd.Tiaconsvethlquiy
for long-term bonds. This action solved the liquidity
problem, although at a significantprice in terms of a
higher rate.
Other considerations with this kind of quasi-fiscal
deficitare whether the increasein public debt impliedby
the deficit is reasonableand the extent to which it will
cause problemsin the future. In Chile'scase,a counter-
weight to the deficitwasthe change in the socialsecurity
system,which I describedearlier.Fortunately,the social
security reform created a high growth rate of financial
savings,on the order of 14 percent per yearin real terms.
Of course, this is not money dropping from heaven.
When a social security system changes from a pay-as-
you-gosystem to a fullyfunded pension system, the so-
cial security benefits of those who are already retired
must be paid. Those current benefitsare being paid by
the government.The net result of all these changes has
been a significantincrease in nonfinancialpublic sector
savings.These savingsin the nonfinancialpublic sector
offsetthe quasi-fiscaldeficit in the Central Bank. I be-
lieve this counterweightexplainswhy the Central Bank
deficithas not createdproblems.
112 Juan Andres Fontaine
Debt-Equity Swaps
I wouldnow like to turn for a fewminutes to the issue
of debt-equityswaps.The first thing to bear in mind is
that we in Chiledid not invent debt-equityswapsor oth-
er mechanisms of debt conversion.They did not result
from careful planning or discussions among govern-
ment officialstrying to devisean ingeniousplan to solve
the debt problem. They were basicallya market result,
which arose spontaneouslybecause there was a differ-
ence between the abilityor willingnessof debtors to pay
and the creditors' perceptionsof what the debtorswould
be paying.A common perception is that these are tools
of "debt management";actually,theyare just ways oflet-
ting the market work. Of course, had we thought sec-
ondary market prices for Chilean debt were "too high"
(in the sense of overestimatingour ability to pay) we
would not havegone ahead with debt conversions.
The questionwas whether to regulate or restrict this
market phenomenonor to allowit to run its owncourse.
Within the Central Bank,we did discussat length howto
design appropriate regulations to avoid some of the
problemsthese operationsmight cause.
In Chile,the processwas greatlyfacilitatedbythe fact
that most of the foreigndebt wasowedbythe privatesec-
tor, the private banks,and some large companies.There
was also some public sector debt, primarilyin the Cen-
tral Bank.Althoughwe did engage in some debt conver-
sion on that public sector debt, priority was given to
allowingthe private sector to carry out that kind of op-
eration.
Early in 1985,we analyzedthe costs and benefits of
these operations. On the cost side, wewere more or less
able to predict the kinds of problemsthat might arise if
the debt conversionswere allowed.On the benefit side,
however,we erred substantially in our estimate that
these would be small-scaleconversions.The figure we
hadin mind wasabout US$1billion.Sincethen, the total
value of conversionshas been US$9billion.
We put in place two basic mechanismsfor debt con-
version. One,which is referred to as Chapter 18, allows
Chileanresidents to buy debt paper abroad and convert
it into pesos through the debtor of that paper. The sec-
ond mechanism,knownas Chapter19,providesfor stan-
dard debt-equity swaps, in which foreign investors are
allowedto register as foreign investmentthe equivalent
in pesos of the equity they receivefrom convertingdebt
paper.Ofthe total ofUS$9billionof converteddebt,one-
third is the standard debt-equity swap; the remaining
two-thirds are operations carried out by Chilean resi-
dents, plus some relativelysmalloperationsby the Cen-
tral Bank itself in buying back its debt abroad,with the
agreement of the commercialbanks.
The benefits of these conversions are clear. First,
there is the capture of a portion of the discount at which
the paper is priced in the international markets. During
this period, the average discount on Chilean debt was
about 40 cents per dollar.The averagediscount actually
captured by Chile through the two conversion mecha-
nisms, Chapters 18 and 19,has been on the order of 30
cents. So we have been able to get most of the discount.
And 30 percent of US$9billion is US$2.7billion, which
representsa net gain to Chileof an amount slightlymore
than 10 percent of GDP.A program or policythat yields
10 percent of GDPis not easyto find, so I think the re-
sults havebeen impressive.
There is discussion now, in Chile and elsewherein
Latin America,of whether this debt conversionroute is
better than the alternative of working with the creditor
banks' steeringcommittee to agree on a debt conversion
scheme with a specificdiscount. Big plans of this type
havebeen negotiatedbyMexico,Venezuela,and the Phil-
ippines.Myimpressionis that the market mechanism is
much more efficient for the debtor country. When a
debtor country sits down with a creditor committee in
Washingtonor NewYorkto negotiate a discount, in ef-
fect a "bilateralmonopoly"is formed.Whendebt conver-
sions are worked through the market, the debtor
country has more options for regulating the amount of
conversionand for dealingwith small creditorswhomay
be willing to sell at differentprices.The debtor country
can better exploitthe market through mechanisms like
those used in Chile,as shownby the capture of the dis-
count in our operations.
As a secondbenefit,these debt conversionoperations
allowedus to reshape the country's"balancesheet." The
Chapter 18 operationsbrought flight capital back into
the country.In the Chapter19 conversions,a portion of
debt liabilitieswas converted into equity. Over all, we
havebeen able to reduce significantlythe stock of exter-
nal debt.At the start of the program,the total amount of
foreign debt was slightly larger than GDP.Now it has
been reducedto 60 percent of GDP.The ratio of foreign
debtto exportswas4 to 1; it is now 2 to 1. I think the dis-
advantage of a high levelof debt, comparedwith equity
liability,is that a high debt increasesthe country's risk
of interest rate or terms of trade shocks creating wide
fluctuationsin domesticoutput and consumption.
There is another benefit of these debt conversions
that is not readily quantified.Typically,when a country
has a high levelof foreigndebt,private economicactivity
is very low,due in part to expectations.Privateeconomic
agents feel there is no solution to the problem,and that
attitude tends to depressthe economy.Afew months af-
ter this conversionprogram started in Chile, the private
sector started to move again. Peoplecould see the light
 Impact of Government Deficits on Monetary Policy: The Case of Chile
at the end of the tunnel. They sawa way to reduce their
debts; they saw opportunities,which got the economy
moving again.
Of course, there are also costs to be considered,as
well as benefits. To capture those 30 cents on a dollar,
you must pay the other 70 cents. Waysmust be found to
finance that 70 cents.
Then there is the issue of the additionalityof the for-
eign exchangeused to paythat 70 cents. Both in Chile
and abroad,there have been long discussionson the ex-
tent to which the foreignexchangeinflowsattributed to
these debt conversionschemeswere additionalor would
have occurred anyway.In Chile's case, the balance of
paymentsimprovedsignificantly,so it seemsthe inflows
were for the most part additional.Not only did the bal-
ance of paymentsimprovegenerallybut cash investment
increasedsubstantially,despitethe successof the mech-
anism for investing through debt conversions.At the
least, there wasno costwith respectto foreignexchange.
We managed our conversion mechanisms to avoid
linkages that would affect our foreign exchange nega-
tively.With respect to Chapter 18-the program that
used residents' foreignexchange-a monthly quota was
established. We auctioned off portions of that quota.
With respect to Chapter 19,we screened each project to
determine whether it createdlinkages-for example,by
increasing imports-or wouldsubstitute for investment
alreadyin the country.
Macroeconomiceffects are also an issue when debt
conversionsare discussed.On the fiscalside, it has been
said that converting cheap foreign debt into expensive
domestic debt deteriorates the fiscal position of the
country. This is a validpoint, sothe discount on the debt
should be used as the basis for determiningwhether the
conversionmakes fiscalsense. For example,in Chileat
present interest rates have increased to the point that
there is little incentive to convert foreigndebt. The de-
cline in demandis putting pressure on the price of Chil-
ean debt in the international market. When the price
declines enough to make the discount attractive, the
market in debt conversionwillpick up again.
113
Inflation is another macroeconomic effect that is
sometimesmentioned as a potential cost of debtconver-
sion. This concern arisesfrom the assumptionthat debt
conversionmust operate through the central bank itself,
which buys and converts the debt by issuing domestic
currency.But the conversion mechanismsdo not work
that wayin Chile. Rather,there is usuallya privatenego-
tiation betweenthe debtorwho issued the paper and the
newcreditor whohas bought the paper on the secondary
market. No monetary expansion is associatedwith the
prepayment of that debt in local currency.There could
only be an effecton the moneysupplywhen the debt be-
ing converted is owedby the Central Bank.In that situ-
ation, however,rather than payingby issuing currency,
the Central Bank issues a long-term bond denominated
in local currency. The transaction is actually a swap of
foreigndebt for the long-term bond,which sterilizesthe
impacton money.
A third macroeconomiceffectto consider is pressure
on real interest rates. Certainly there is some effect,
since the substitution of domesticfor foreignfinancing
increases demandon domesticsavings.There may also
be an effecton real asset prices. When the debt conver-
sion operations began in Chile, real asset prices in-
creased significantly.Foreign investors and Chileans
with foreignexchangeabroadwere using the two mech-
anisms to buy real assets. So the initial effectwas more
on real asset pricethan on interest rates. In the past year
or so, we have seen interest rate increases rather than
real asset increases.
Weighingthese various costs against the benefits, I
believe the benefits outweigh the costs. To sum up, I
think our mechanismsfor debtconversionhavebeen ex-
tremely efficient in reducing problems that in 1985
seemed impossibleto solve.We have now reduced the
foreignbank debt to US$5billion,which is manageable.
Wewill probablycontinue to reduce it. Although it cre-
ates some problems for management of monetary and
fiscalpolicy,those problemsare manageable.

MonetaryManagementwith High Inflation:
The BrazilianExperience
CarlosAlberto Queiroz
I will begin by explaininghow the money market has
evolvedin Braziland howmonetarypolicyhas beencon-
ducted in that market. I will then look at Brazil'sexperi-
ence during the super-high inflation of 1989,in which
prices rose by as much as 84 percent in a single month.
In that kind of environment,open market operationsof
any kind becomedifficult.
Development of the Money Market and
Monetary Policy
Amarket for governmentsecuritiesbegan in Brazilin
the second half of the 1960sand grew during the 1970s
and 1980s. Beginning in the 1970s, this market grew
proportionatelymore than any other financial market,
including the stock market. The growthwasprimarilyin
the spot market; only in the 1980sdid the first, and not
very successful,attempts occur at a forwardmarket in
interest rate futures.
In a developingcountry like Brazil, there is a prefer-
ence for short-term investment of savingsbecauselong-
er-term expectationsface large uncertaintiesabout both
inflationand income.Themechanismthat madetrading
of long-term securities possible was the repurchase
agreement.' It enabled financialinstitutions to leverage
a long-run position on financial assets with funding
from the short-term market.
Financingthe publicdebt of Brazilduring the last de-
cade has come to depend on this mechanism of repur-
chase agreements and the associated system of
leveraging.Yetthis situation imposesenormous restric-
tions on the conduct of monetary policy.Because the
short-term market supports large and highly leveraged
positionsin governmentsecurities, anyunexpectedvari-
ation in short-term interest rates could engender a li-
quidity crisis. The possibilityof hedging against such
variations through futures on short-term interest rates
has been unavailable.
115
9
As a result, a change in the price of financial assets
could easily create a situation where private financial
markets would collapsewithout government interven-
tion. On severaloccasions,the Central Bankhad to enter
the market simply to provide a hedge for the market.
Overthe years that the Central Bank has livedwith this
situation, it has tried to avoid large-scalerescue opera-
tions of this kind. Attempts to improve the situation
were made in 1986,when Central Bank Bills,or LBCs,
were created,and later on when FinancialTreasuryBills,
or LFTs,were established.LFTsare variableinterest rate
securities; their daily interest yield equals the average
interest rate for that day on all repurchase agreements
forgovernment paper.In effectthe dailyyieldon an LFT
is equal to its own overnight repurchase rate. LFTsare
issued in public auctions every Tuesday,with transac-
tions cleared the followingday. The selling price, ex-
pressed as an annual discount rate, is determined by
competitivebidding.
Introduction of a security that yields its own over-
night rate has greatly diminishedthe capital value risks
from unexpectedfluctuationsin the interest rates. The
LFT mademonetary control much easier in a setting of
high inflationrates coupledwith no control of fiscaldef-
icits.
It is important to clarifythat the LFT does not solve
the problem of financingthe fiscaldeficit.Becauseof the
ever-growingneeds of the public sector for funding,an
increasingamount of governmentpaper must be sold at
auction.The market requires higher rates to clear the is-
sues, so the discount rate for the LFT becomes deeper
and deeper.All that the LFT yield mechanism has ac-
complishedis to insulate the effectofunexpectedlyhigh-
er inflation on the stock of LFT already held by the
private sector.
When open market operations started in the late
1960s,the government introduced a monetary budget
by consolidatingthe balancesheets of the Central Bank
116 Carlos Alberto Queiroz
and the Banco do Brasil, which is a commercial bank
owned by the government. Since then, this monetary
budget has been the basis for monetary policy.
Lendingbythe Bancodo Brasilto the nonbankpublic
has alwaysbeen a major source of monetarybase expan-
sion in Brazil. This credit was registered in the budget
and funded through subsequentsaleof governmentbills
and bonds.
In effect, the government was spending first, then
borrowingto fund the expenditure.Interest rates would
fall at first, from the credit expansion.Then the govern-
ment would sell securities to balancethe monetarybud-
get. Overall,interest rates did not rise, so the private
sector did not identifythe latter action as crowdingout
other borrowers.However,after severalmonths' lag, the
inflationrate wouldjump "unexpectedly."Onlythen did
the private sector realize the extent to which govern-
ment spendingwasexpandingthe moneybase.Andonly
when the monthly inflation rate, which is public infor-
mation, began to rise did the monetary authorities press
for interest rate changes.Theybegan to worryabout loss
of control over the monetaryaggregates.However,inter-
est rates could not be allowedto rise very far,since the
financialmarkets were so highly leveraged.Also,the de-
mand for investments might havebeen adverselyaffect-
ed.
This situation held for most of the time that the open
market was evolvingin Brazil.During the 1970s,the in-
flow of external capital made things easier. During the
1980s,particularlyafter the Mexicanbankruptcy,the in-
flowof external capitalhalted. From that time on, the is-
sue of how higher interest rateswould affectinvestment
becamean important constraint on monetary control.
Monetarypolicywas thereforebeing conductedindi-
rectly.Increasesin publicspendingwerejustifiedby reg-
istering future funding in the monetary budget.
Compensatingchanges in interest rates were too little
and too late. Built into this mode of CentralBank action
is a causal directionfrom variation in prices to variation
in the monetary aggregates.In other words, the causali-
ty was from P to M, not from Mto P.
The Year of Super-High Inflation
In many high inflations a point is reached (often
termed hyperinflationarycollapse)where the supply of
goods and services shrinks well below what had previ-
ously been available,resulting in a severeworseningof
economic conditions and a generalized shortage of
goodsand services.Typicallycurrency is no longer func-
tioning as a medium ofexchangeand the economyslides
into a costlysystem of barter. The reduction in national
output leads to exacerbationof poverty,and not merely
to a recessionfrom which an early recoverycould be ex-
pected.
Duringthe inflationdynamic,there can be a periodof
time when the real sector continues to produce goods
and services,despite high monthly or evenhigh weekly,
rates of inflation. Certainly,this high inflation rate re-
tains the negativeaspectsof an inflationaryprocess,with
all its allocativeand distributiveproblems.It is in fact an
intermediate stage that over time will lead to the hyper-
inflationarycollapseas describedabove.This intermedi-
ate stage is what I mean by a "super high rate of
inflation,"and this is what Brazilexperiencedin 1989.
Withthis backgroundI turn to the funding ofgovern-
ment debt and monetary policy in circumstanceswhen
inflation has reacheda monthlyrate of 70 percent.
By the end of the "Summer Plan" in May 1989,the
government admitted that nothing could be done to
control government spending. The goal was simply to
find a way to support the politicalprocessleading up to
the two rounds in the presidential election (the first
round in November,the second in December), while
avoidinghyperinflationarycollapse.
The monetary policy that was adopted in this situa-
tion was based not on the effect of prices on interest
rates, as in the past, but on control of interest rates to af-
fect prices. The nominal interest rate became the main
target for monetary policy.The real ex post interest rate,
i.e., as measuredafter the fact, had beenthe operational
target; it was now dropped from consideration.The ra-
tionale for this change was a Fisherian concept of the
nominal interest rate as having built into it the future
inflationrate. Thus the new policywouldtake the real ex
ante interest rate, i.e.,as estimatedbeforethe fact,as the
operationaltarget.
The operating procedurewas to set the monthly in-
terest rate so that the expectedinflation rate in the next
month would leavea real rate on governmentsecurities
that would be favorablerelativeto less liquid real assets
or speculativefinancialassets such as gold certificatesor
foreigncurrency.The hopewas to delaythe public'scon-
sumption decisionsas long as possibleby presentingan
opportunitycostsufficientto encouragedecisionsnot to
spend.
Auxiliaryindicators for the real interest rate target
were the price behaviorof the gold market, the parallel
market in foreigncurrencies,and the stock market. The
behavior of the monetary aggregates (base money and
MI) becameuseless.Therewas little sense in targetinga
monetary aggregatewhen the inflationrate was so high
and the government would not reduce spending. In
these circumstances,controlling moneysupply directly
could lead to interest rate variations-and resulting
fluctuations in prices of financial assets-so large that
 Monetary Management with High Inflation: The Brazilian Experience
the Central Bank would be obliged repeatedlyto enter
the market with rescue operations. Such rescue opera-
tions would only amplifythe uncertainties for the pri-
vate real sector, with further negative effects on the
production of goodsand services.
The real sector of an economywith inflationrates of
40, 60, or 70 percent monthly experiencesconsiderable
uncertainty.Asa result, there tend to be -brupt shifts in
savingspreferencesas between real stocks and financial
assets. Investmentdecisionsare delayedas the ability to
anticipate decisionsis eroded,since all the factors that
influence investment and production activitiesare sub-
ject to drastic revisionat any time.
In this situation, the stabilityprovidedby the Central
Bank became essentialto economiclife or death. With-
out some source of stability, the real economic sector
could haveexplodedat any moment.The trading desk of
the Central Bank providedthis stability in two key as-
pects: liquidityof the market and an adequatereal return
on financialassets.
Toachievethe desirednominal interest rate, the trad-
ing deskbought and sold repurchase agreementsand re-
verse repurchase agreements.It stood ready to provide
adequate liquidityat the target interest rates to a highly
leveraged financial market. In managing liquidity the
trading desk of the Central Bank madeoutright transac-
tions in governmentsecurities:buying those the market
did not want and selling those the market desired to
hold. At times the market demanded long-term paper;
on other occasions,long-termpaper wassold backto the
CentralBank. Similarly,demand for short-term notes
was met by the buyingand sellingoperationsofthe trad-
ing desk.
Thesetrading operationsin support of monetary pol-
icy made it possiblefor the private real sector to contin-
ue conveying its uncertainties to the financial sector.
Even so, the situation could not be sustained indefinite-
ly.Actionsthat are basedonlyon monetarypolicyto con-
trol interest rates, without adjustment of government
spending, eventually lose their efficacy.The stability
achievedby the policybecame merely a stabilizationof
the rate of growth of inflation.
Inflationaryprocessesarise from public deficits.With
inflationrates reaching70 percent per month, a govern-
ment can achievelong-term successonly by cutting its
owndeficit.In March1990,the governmentof Brazilin-
troduced a new program built on the three pillars of fis-
cal policy,monetarypolicy,and income policy.
117
The fiscalpolicygoal is to create a fiscalsurplus. It is
obviouslydifficultto transform a chronic public deficit
into a surplus. Unfortunately,it is not possibleto reduce
spending or generate more revenue overnight. In an
economyas complexas Brazil's,which is developedin
the south but very poor in the north, achievingthis fiscal
goal will require a long periodof hard work.
The new plan also acted to reduce demand in the
short run, through the compulsoryuse of private sav-
ings. The financialreturn on these savingsis basednot
on the old hyperinflationaryconditionsbut on the ex-
pectation of future price stability.The intent is to recov-
er a more stable framework in which traditional
monetary policywill again be effective.
After March 15, a new monetary base was created,
consisting at the outset only of cash in the hands of the
nonbank public. All other "new cruzado" (the old cur-
rency) assets became nonmonetary assets. Thus there
has beenan abrupt reduction of the stock of moneyand
free reservesto 20percent ofwhat it had been. The stock
of cruzadoswill graduallybe convertedto cruzeiros over
a period of eighteen months.
During this periodof a year and a half, the public def-
icit willbe managedbyreductionsin governmentspend-
ing and permanent increases in government revenues.
Public enterprises and government real estate will be
sold as well. Therefore,we hope to have recoveredthe
classicalinstruments of both fiscalpolicyand monetary
policyby the end of this transition period.
Note
1. The interbank market for reserves-one of two
main short-term markets in Brazil-is entirely basedon
repurchase agreements in government securities or on
outright trading in those securities. Therefore the key
interest rate for the national financialmarket is the over-
night repurchase rate for government securities. (The
other market, called the administrativecheck market,
trades funds availablefrom the floaton checksthat have
not yet beencleared,and is collateralizedwith private se-
curities. Becauseof the one-daydelayin clearing, inter-
est rates on the two markets are linked through
expectations:the rate on the administrativecheck mar-
ket is effectivelythe market's expectationof the next
day'srate on the market for reserves.
118 Carlos Alberto Queiroz
Discussion
Question:What Mr.Caranzahas describedreallydoes
seem to be an Italian miracle, at least compared with
Mexico,where the public debt is only 20 percent of GDP.
Mexicois running a budget surplus of around 7 percent
of GDP.Yetthe real interest rate on the public debt is
about 25 percent.This is creatinga tremendousthreat to
the stabilizationprogram in Mexico.We are lookingfor
financialinnovationsthat will extendthe maturity of the
debt, because the public debt is almost entirely short
term. The averagematurity is less than two months. In
effect,every two months or less the entire debt must be
renegotiated.
We think that this situation may be the result of ex-
pectationsof hyperinflationin the near future.Thoseex-
pectations may be grounded in the fear that at some
point that demandfor the public debt willbe too weak to
carry out the bimonthlyrefundingof that debt.
With respectto the financialinnovationsin Italydur-
ing the 1980s,did you investigatethe stabilityof demand
for long-term bonds issued by the public sector? Are
there other cost-saving schemes, such as contractual
savings,retirement funds,and pension plans,that could
create a stable demand for long-term debt?Or are there
other reasonswhy householdsin Italyare willingto hold
70 percent of the debt of the public sector?
Mr.Caranza:The link between your two questions is
inflation. Last summer, I spent several months in Sao
Paulo, Brazil,for I was fascinatedby certain similarities
betweenthe Italian and Braziliansituations. Brazil, like
Mexico,has extremelyhigh real interest rates and a very
short maturity on its public debt. The averagematurity
is practically24 hours, becausethey must go to the over-
night markets.
Why are real interest rates so high? The level of real
interest rates reflects the past inflation history of the
country.Mexicoand Brazilare payinga huge risk premi-
um to the savers.Afteryears of high inflation-not hy-
perinflation in the technical sense, but an ongoing
"mega-inflation"-how can the authorities convincepri-
vate savers to hold bonds or even bills? They must pay
tremendouslyhigh real interest rates, which have a bal-
looning effecton the debt, whosegrowth continues the
cycle.
In this situation, monetary policy alone is useless.
Monetarypolicy cannot be used without a sound fiscal
policy. If the monetary authorities try to push interest
rates higher,the deficitballoonsand the debtgrowseven
more. The result is a loss of confidencein the country.
Question:This is whyI say that Italymaybe a miracle.
If I were an Italianin 1965and saw that the government
was borrowingto pay interest and to cover the budget
deficit,the situation would look unstable. The Mexican
governmentis trying to addressthe problemof inflation-
ary expectationsby issuing indexedbonds, which cover
the inflation risk. Becausethis risk is removed,the rate
on indexedbonds is lower than on nonindexedbonds.
But there is still a substantial premiumin the real inter-
est rate because of the default risk. The problem seems
to be not the risk of inflationbut the lack of stable de-
mand for these indexedbonds.
Mr.Caranza:During the 1970sand 1980sin Italy,we
discussed all the possibilities, including the pros and
cons of issuing price-indexedbonds. We decidednot to
issueprice-indexedbondsbecauseindexationto prices is
a wayto livewith inflation;it is not the wayto fight in-
flation.Price-indexingthe debt anesthetizes the system.
It is like a drug; one liveswith the inflationbut losesthe
stamina to fight it.
You raised another point concerning the results we
had in Italy in creating a stable demandforlonger matu-
rities.In recent years,wehaveachievedthat result. Here
again, inflation calls the tune. Particularlyafter the fall
in oil prices in 1986,Italy had rapidly falling inflation;
then inflation stabilizedat a lower level.At that point,
the demand for longer-termbonds increased.Bylonger
term, I mean four to seven years, which is not all that
long, but there wasa large demandfor such bonds.
Question:Werethose bondstied to the short-term in-
terest rate?
Mr. Caranza:No, these new bonds are fixed-interest
bonds. In the past, from the late 1970suntil early in the
1980s,wedid use that kind of financialindexationin the
treasury certificates.We did not adopt price indexation,
but those financially-indexed certificateswere used as a
passageto more orderlyfinancialconditions.
Question:With the reduction in the primarydeficitin
Italy,what has been the impacton the private sector?
Mr. Caranza:That was one of the positiveaspects of
the story. Notwithstandingthe rise in real interest rates,
the private sector did not suffer much. In particular,real
investment did not suffer much. To be more precise,
construction activity slackened somewhat; there was
probablysome reallocationofsavingsawayfrom the tra-
 Monetary Management with High Inflation: The Brazilian Experience
ditionalinvestmentsin housingand dwellingsand to-
ward financial investments. Yet real, productive
investmentwasnot affectedmuch.Whenwelookat the
ratioof investmentto GDPgrowth,there is no signof
significantsuffering.
Weprobablywereableto achievethis resultbyhaving
interestrate and exchangerate policiesthat servedas
strongstimulito the enterprisesectorto cut costsand
improvecompetitiveness.
The most importantfactorwasthat the enterprise
sectorgeneratedsufficientinternalfundsto financenew
investment.Thatsectorunderwenta tremendousrecov-
eryin theearly1980s.Bythe middleofthe decade,their
financialsituationwassocomfortablethat a numberof
enterprisesbeganto investtheirextrafundsin thefinan-
cialmarkets.Theresultswerequitegoodbecauseofthe
high levelofinterestrates.
So this waspart of the bright side of the coin.We
soughtto managethe situationwithoutdampeningthe
realgrowthofthe economy.
Question:Didthe Italianrecoverybenefitfromcapi-
tal inflowswhenEuropeanbanksmovedawayfromLDC
lendingin thewakeofthe debtcrisis?
Mr.Caranza:Notverymuch,actually.Aninteresting
featureof the Italian experienceis that the treasury,
whichwasthe big borrower,didnot relyheavilyon ex-
ternalborrowing.Asa matterof policy,wepreferredto
sellthepublicdebtin the domesticmarkets.Onlyin the
morerecentyearshasthere beena significantinflowof
capitalfromabroad,but that hasbeencompletely auton-
omous.So I do not think Italybenefittedparticularly
fromthe debtcrisis.Therewasa majorinflowoffinan-
cial investmentfromabroadin the lastfewyears,espe-
ciallyfor purchasesof treasurysecurities.Thecrucial
point was the expectationof a reasonablystableex-
changerate,the EMSeffect.
Question:Youmentionedthat the governmentwas
borrowing11 or 12percentof GDPandthe savingsrate
was 20 to 21 percent of GDP.For investment,there
wouldalso havebeen foreignsavingsaddedfromthe
currentaccountsdeficit.Howmuchwasthe difference
betweenthe total savings,both foreignand domestic,
availableto the economyand the governmentdeficit?
Didthat differencechangeovertime?
Mr.Caranza:Duringthe secondhalfofthe 1980s,we
maintaineda smallcurrent accountdeficit.It wasless
that 1percentofGDP.Onbalance,then,the externalef-
fectwasmoreor lessneutral.Whenone looksat the fi-
nancial balances for Italy during this period, the
119
enterprisesectorand the foreignsectorweremore or
lessin equilibrium,in termsofthetrend,althoughthere
were significantfluctuationsat times. On the other
hand,the surplusfromthe householdsectorand the fi-
nancialdeficitin the budgetsectorformtwodiverging
trendsthatovertimerepresentan increasingpercentage
of GDP.Thesetwo trendscomplementeach other,but
usingthesurplusofoneto financethedeficitin theoth-
er requireda tremendousamountof financialinterme-
diation.Thatintermediationhad to be suppliedby the
bankingsystemand the financialmarkets.
Question:In termsof monetarypolicy,howdid you
handlethe inflowof resourceswithinthe contextof a
strengtheningofthe lira in the EMS?
Mr.Caranza:TheproblemItalyhas had in the past
twoyearswithcapitalinflowhas beena newexperience.
In the past, the treasurywastypicallycreatinga good
dealofmoney,whileoutflowsthroughthe externalsec-
tor resultedin net moneydestruction.Thelastfewyears
havereversedthe situation.Therehas beena lot ofin-
flowfrom abroad,and we havehad someproblemsab-
sorbingthis liquiditythroughopenmarketoperations.
Yetthis is just oneaspectofthe newrealitythat Italy
willbe facingevenmorein the future.I endedmypre-
sentationwith the presentsituation,but one can also
lookaheadto thenewframeworkofEuropeanmonetary
unificationwithinwhichItalywillbe operating.During
the lastdecade,thetypicalsituationwasthat ofonelead-
ing economy,Germany,whosemoneywasthe linchpin
ofthemonetarysystem.Theleaderwasthe countrywith
the sternestanti-inflationdiscipline.The other coun-
triestriedto followbylinkingtheirexchangeratesto the
moststablecurrency.Duringthe 1980s,the followers
had the samefinalobjectives as the leader:to reducein-
flation.There were costs, and much pain in certain
countries,Italyin particular.But theywere all deter-
minedto reducetheir domesticinflationratestowards
the Germanbaseline.
Also,a numberofthe largerfollowers-France,Italy,
and later on, Spain-were countrieswhoseeconomies
were relativelyclosedthroughapplicationof exchange
controls.Now,however, the economiesofFranceand It-
alyare entirelyopen,and Spainis graduallyliberalizing
its controls.
It is difficultto imaginethe newframeworkcontinu-
ing to workas it has duringthe past decade.Although
the largefollowers havenot achievedthelowlevelofin-
flationenjoyedbyGermany, their inflationratesarerea-
sonable.Andnowtheir economiesare totallyopen. I
doubt that Germanycan continueto playthe role of
leaderforthe followingreason.In a scenarioofcontin-
120 Carlos Alberto Queiroz
ued German leadership,the frameworkwould resemble
the old BrettonWoodssystem,but with a crucial differ-
ence. Germanylacksthe economicand financialweight
to havethe mark playthe role that the U.S.dollar played
under Bretton Woods.At that time, the dollar was the
currency of the economy of overwhelmingimportance
in the world.
The problem is not one of competingnationalisms;it
is a purely economicissue. The German economy can-
not defenditself from instabilitycoming from other ma-
jor Europeaneconomies.Evenwithin the last fewyears,
Germanyhas begun to experienceproblems related to
these pressures. When they have had major inflows of
capital from other European countries, they have had
trouble in controlling their moneysupply.
Furthermore, price stability in Germanyshould not
be taken for granted.What,for example,maybe the con-
sequencesfor price stabilityof German reunification?I
am not implying that there will be higher inflation in
Germany,just that we do not knowwhat may happen.
A preferable outcome to revisiting Bretton Woods
would, I argue, be a strong effort by keyparticipants to
coordinate their monetary policies. At least the major
European countries should coordinate monetary policy
ex ante and set mutually consistent monetary targets.
Then they must followdevelopmentsand make adjust-
ments to keep things on track.
This appears to be the direction in which the EMS is
heading. Beginningin the autumn of 1990,the four ma-
jor continental economies-Germany, France,Italy,and
Spain-and the two smaller economiesof the Nether-
lands and Denmark will announce coordinatedmone-
tary targets for the followingyear.They will attempt to
implement a system to monitor developmentsso that a
consistent monetary policy can be maintained within
the EMS.In the end, this is the only wayto achieve rea-
sonablystable exchange rates in Europe. Of course, the
whole exerciseassumes there is agreement on the final
objective of monetary policy.Clearly,that remains the
control of inflation in Europe.
Question:Towhat extentcan privatizationbe a poten-
tial solution of the debt problem in Italy?
Mr. Caranza:That is a very hot issue. Our treasury
minister is campaigningvigorouslyforthat kind of solu-
tion. Although it is not a completesolution for the debt
problem, it would be an important step forward.Major
portions of the public enterprise sector could be sold to
the private sector. But there is a lot of political resis-
tance, particularlywith respectto the banking industry,
much ofwhich is publiclyowned.
The situation is peculiar.Manyof these banks do be-
have likeprivate banks;they do not respondmuch to po-
litical pressure, at least so far. The problem lies in who
would be the new owners of the privatizedbanks. Italy
lacks a sufficientlydevelopedsector of institutional in-
vestors. Perhapsforeign interests could playa part, but
we do not want to sell our bankingsystem to foreignin-
terests entirely.The potential domesticbuyersare indus-
trial enterprises or privateinvestors.There are pros and
cons that can be argued,but the monetaryauthorities in
Italy oppose industry taking majority positionsin com-
mercial banks.
More generally,though, the issue of privatization is
important. While not a complete solution to the debt
problem, it could be an important step forward.
Question:I wasjust thinking back to when the Ger-
mans privatizedVolkswagen. They had a system through
which they distributed shares to a very wide public,
thereby avoidingconcentration.Presumablysomething
similar could be used in Italy, if the public authorities
were not hostile to it.
Mr.Caranza:Yes,that is another possibility.An addi-
tional problem in Italy is that we traditionallyhave had
an underdevelopedequity market. Of course, an injec-
tion of new securities, such as those resulting from a
privatization program, could help to revive the equity
market.
This weakness of the equity market is a problem in
other countries as well.I am thinking in particular ofre-
cent talks on privatization in the Eastern European
countries. They have problemswith how to market the
new shares to small investors. So there are technical
problems, but the privatization issue is fundamentally
linked to the politicaldecision whether to move in that
directionor not.
 PARTIV

THE INTERA CTION OF EXCHANGE

RATE AND MONETARY POLICY
 Introduction
The other main source of volatilityin the central bank's
balance sheet is foreign exchangeflows.Charles Freed-
man and Lin See Yan lookedat this problem from the
perspectiveof the central banks of a smallopen industri-
al economy(Canada)and a middle income country (Ma-
laysia),respectively.The speakersare deputy governors
of these respectivecentral banks.
Central banks can insulate their balance sheet from
international flowsonlyby adoptinga purelyfloatingex-
changerate regime, that is, byavoidinganypurchasesor
sales of foreignexchangefor its portfolio.'At the oppo-
site end of the spectrum, irrevocablyfixed rates entail
the sacrificeof all monetary policyautonomy;evenwith
somewhat less permanent arrangements, small coun-
tries are forcedto adopt the monetary policyof the larger
country to which their currency is pegged.In between
these two extremes, Dr. Freedmanargued that the in-
struments of monetary policy are more powerfulthan
exchangerate intervention in influencingthe path ofthe
exchangerate. Indeed,interventionis likelyto haveonly
a very short-run role.With less than perfect substitution
among assets, central banks can sterilize reserveflows,
that is, limit their impact on the domesticmoney stock
on interest rates.
Dr. Lin emphasizedthe extent to which a small open
economy can have its reserveswiped out if the central
bank does not act quickly.Occasionalspeculativeruns in
Malaysia'sopen capital market have beenstemmed only
by short periods of liquidity squeeze in which money
market rates have jumped as high as 100 percent (as
against single-digitinflation).In an exampleof the be-
haviordescribedby Dr.Mathiesonin Part II, the author-
ities have, on such occasions, acted to protect
nonspeculativeborrowerswith a closelycontrolledpro-
gram of subsidizedcredits.Someparticipantswereskep-
tical that such selectiverelief from the liquiditysqueeze
could be effectivelyimplementedwithout creating seri-
ous distortions.
123
It should be noted that Malaysianauthorities in the
earlyand mid-1980shad a distinct preferencefor defend-
ing the exchange rate, letting interest rates bear the
brunt of the adjustment to various shocks.Canadianau-
thorities, in contrast, chose to moderate interest rate
movementswith greater flexibilityin the exchangerate.
A variety of factors may govern this choice, including
countries' history, with Canadabeing an earlier experi-
menter with floating rates in the 1950s.In Freedman's
terms, these preferencesmay reflect relativelyless con-
cern on the part of Canadianauthorities with importing
the credibility of their main trading partner. These
choices might reflect as well the particular shocks to
whicheach economyhas been exposed,as wellas the de-
gree of priority givento combatinginflation.
Authorities of developingcountries should consider
their choices regarding (short-term) interest rate and
exchangerate volatility,given that in a liberalizedenvi-
exchang te vili gnt in a liberalized e
eclectic approach to monetary targeting suggested by
Lindseyand Goodhart is consistent with an eclectic ap-
proach toodharate a nclectin
proach to the exchange rate as well. Officialsmight in
the end focusmore on their ultimate objectivesfor infla-
tion and less on near-term targets for exchangerates if
an independentmonetary policyis being pursued. Alter-
nativelytheycan sacrificeall independenceand peg irre-
vocablyto a "stronger" currency. But either choice is
compatiblewith the move to market-based monetary
policy.Havingestablishedtheir credibilityregardingin-
flation, and being faced more with incipient capital in-
flows,Malaysianauthorities recentlyhavein fact moved
to a more eclectic approachto both interest rate and ex-
change rate policy. This lesson-that central banks
should pay most attention to restraining inflation, and
thereby reduce uncertainty facinginvestors-has much
to recommenditself to other developingcountries. Ma-
laysia'slong-termgrowthperformancesuggeststhat the
124 Introduction to Part IV

growth effectsof such a policyorientation may be posi- 2. Industrial country experience certainly suggests
tive and, indeed,could wellbe significant. that long-term interest rates cannot be controlled by
monetary authorities.
Notes
1. Steven Grenville (in Part I) discussesthe various
waysin which a central bank can lose control of its bal-
ance sheet.

ExchangeRate Policy
CharlesFreedman
This paper is in four sections. The first is definitional:
what is meant by exchange rate policy.As will be seen,
there are three types of policy that directly affect ex-
change rates: (i) the choiceof fixedversus flexiblerates
(withand without managedfloating);(ii) monetary pol-
icy;and (iii) intervention policy.In the second section I
discussthe choiceoffixedversus flexiblerates in a rather
broadbrush way.I then go on to discussin more detailed
fashionthe conduct of monetarypolicy and of interven-
tion policy.Finally,I discuss the issue of sterilization of
reservechanges in a system of fixedexchangerates or a
managedfloat.
Some Definitions
The term exchangerate policy is somewhat slippery
becauseit focuseson the variablebeing affectedby a va-
riety of policies,rather than on the policiesthemselves.
In this respect, it differsfrom terms like "monetarypol-
icy,""fiscalpolicy,"or "commercialpolicy,"which tend
to focuson the instrument ofpolicy(respectively,money
or interest rates, taxes and transfers or government ex-
penditures,tariffsor quotas). Hence,rather than talking
about exchangerate policy,it is more usefulto talk about
the elements of policythat havea direct influenceon ex-
change rates: the choices of exchange rate system,
monetary policy, and interventionpolicy. This is not to
argue that other policies,as wellas exogenousshocks,
do not affectexchangerates, e.g., fiscalpolicy,commer-
cial policy,or exogenouschanges in the relative price of
raw material on the world market. They do,but it is the
three policiesthat are under the control of the authori-
ties and that havea more direct effecton exchangerates
which are typicallyencompassedby the term "exchange
rate policy."Note alsothe hierarchyof these choices.The
choiceof the exchangerate system is fundamentaland
has a crucial implicationfor the scopeof monetary poli-
cy.Interventionpolicyis the least important of the three
125
10
policy elements and is much less powerfulthan mone-
tary policy.
FixedVersusFlexibleRates
There is a voluminous literature on this subject,
mostlyfor developedcountries, but some for developing
countries. There are a number of major strands in this
literature, some of which I would like to touch upon in
this paper.Amongthe important topics in the literature
are the relativepotencyof monetaryand fiscalpolicyun-
der fixed and flexibleexchangerates with various levels
of capitalmobilityand asset substitutability,' the mech-
anism via which policieshave their effectsin the two re-
gimes, the insulation properties of the different
exchange rate regimes in response to various kinds of
shocks,and the role of policycoordinationand concert-
ed interventionby the G-7countries.
Broadlyspeaking,a worldwidesystem of irrevocably
fixedexchangerates (i.e.,one currency)wouldhave sig-
nificant efficiencybenefits. Mostnotablythere wouldbe
no transactions costs and no need to expendresources
on coping with the risk of exchange rate changes. The
presumedadvantagesof flexibleexchangerates tend to
arise from macroeconomic and adjustment consider-
ations, in particular,the abilityofa country under a flex-
ible exchangerate regime to achievea better inflationary
performance than its potential partners and its en-
hanced abilityto respondto real shocksthat are specific
to it.
Let me begin a detaileddiscussionof some of the is-
sues surrounding the choice of fixedversus flexibleex-
change rates by a bald assertion.In a world with perfect
asset substitutability,no exchangecontrols,and fixedex-
change rates, there is virtually no autonomy in mone-
tary policy for a small country. There are two key
implications. First, a small country that fixes its ex-
change rate to the currency of a singlelarge country or
126 ChatrlesFreedman
to a basket of currencies ofa number of countries by and
large ties its inflation rate to that of its partner or to a
weighted average of its partners. Second, in the face of
real shocksto the terms of trade, adjustment of the real
exchangerate must take placethrough differentialprice
movementsrather than through nominal exchangerate
changes.
A country that fixesits exchange rate (permanently)
trades offits ability to influencedomesticnominal vari-
ables in return for the rate of inflationof its larger part-
ner. This decision is more sensible, the greater the
confidencea country has in the central bankof the coun-
try to which it is tying its currency and the greater the
similarityof the shocksfacedby the two countries.In the
caseof the EMS,for example,other countries havebeen
able to import the credibilityof the Bundesbankby tying
their currenciesto the German mark.And,indeed,there
has been a convergence of inflation rates over time
among those European countries that have associated
their currencieswith the mark. Thecase ofthose franco-
phone Africancountries that havemaintained their ties
to the French currency area servesas another example.
Note that to obtain the full credibilitybenefit of the
fixedrate in such circumstances,the country has to con-
vince the market that the fixed rate is close to irrevoca-
ble. Havinga fixedbut adjustable exchange rate, in my
view, does not yield the benefit of an irrevocablyfixed
rate. The country does not get the full advantageof the
credibility,and it may be facedperiodicallywith attacks
on the currency followedby large discrete adjustments
of the exchange rate peg, with concomitant effects on
prices of traded goods (or subsidies).
As mentioned earlier, the choice of an irrevocably
fixed exchange rate implies that, in the face of real
shocks to the terms of trade, adjustment of the real ex-
change rate must take place through differentialprice
movements rather than through nominal exchange
rates. This impliesthat, where possible,a country should
tie its currency to a large crediblecountry facingsimilar
shocks to those it faces. This will permit it to float
against other countries with differingexternal shocks.
Supposethere is no country that both faces similar
external shocks and has a credible anti-inflationary
stance.That is, supposea country faces sizable external
shockswhichare specificto it and which do not affectits
potential partners. A common shock of this type is the
shift in raw materials prices relative to manufactured
goodsprices. In such a case, the movement in the nom-
inal exchangerate can act to offsetin part the resulting
changes in aggregatedemand, to spread the costs and
benefitsof the change in product prices throughout the
economy,and to facilitatethe movement in the real ex-
changerate toward its equilibrium.Of course, evenwith
flexible exchange rates the adjustment is not all that
easy.There is alwaysthe risk that a currency deprecia-
tion in response to a negativeterms-of-tradeshock will
feedinto a wage-pricespiral.Andflexibleexchangerates
willsometimesmoveawayfrom equilibrium,not toward
it. Indeed,the misalignmentsin exchangeratesin recent
years have been the source of considerablediscussion
about currencyzones, concertedintervention,EMS,and
other ways of moving back towards the fixed exchange
regime. Nonetheless,in the caseof a country subject to
periodic sizeable external shocks that are specificto it
and do not affect its potential partners, it is difficultto
argue that fixedexchangerateswill dominateflexibleex-
change rates.
In this connection,it is worth noting the literature on
optimal currency areas, which focuseson such matters
as the mobility of labor, the size and openness of the
economy,the nature of shocks,and the flexibilityof real
wages. This literature reminds us that the decision re-
gardingfixedversus flexibleexchangeratesis a multifac-
eted one, particularlyfor small countries.
I arguedearlier that, if there is perfectasset substitut-
ability and no exchange controls, a country choosing
fixedexchangerates willhaveno monetarypolicyauton-
omy.I now turn to the situation in a fixedexchangerate
worldwith imperfectsubstitutability,exchangecontrols,
or both.
Imperfect substitutability does permit a country to
have some degree of autonomy in monetary policyeven
though its currencyis fixedto another country'scurren-
cy.The same holds true for exchangecontrols. However,
although the central bank in such circumstances can
followa somewhatdifferentpolicyin the short run from
its partner, it has relativelylittle longer-run autonomy
exceptif it is prepared to adjust its exchangerate from
time to time.
Consider,for example,a situation where small coun-
try Achoosesto followa somewhatlaxermonetarypolicy
than large country B to which it has tied its currency.
The lowerinterest rates in A can coexistwith higher in-
terest rates in B since the capital outflow is not over-
whelminglylarge (becauseof imperfectsubstitutability
or exchange controls). Over time, however,if A's infla-
tion rate is higher than B's, its currency becomes more
and more overvalued.This will leadto pressuresfor cap-
ital outflows from A as the public becomes more and
more convincedof the inevitabilityof a future devalua-
tion. Even if A'sgovernment is successfulin preventing
the outflowvia the use of controls2 the overvaluedcur-
rency can havedeleteriouseffectson the economy.It de-
presses returns in the tradeable goods sector of the
economy, artificiallyincreases the real wage rate (by
keepingimport priceslow),makesthe inevitableadjust-

ment harder to absorb, and raises the value of the ra-
tioned foreign exchange (increasingthe return to rent-
seeking behavior).
In sum, the pure logicof the arguments thus far leads
one to support either an irrevocablyfixedexchangerate
(or one close enough to irrevocableso as to enablethe
authorities to obtain all the credibilitybenefits)or a flex-
ible exchangerate (withor without management of the
float), rather than an adjustablepeg. The latter seemson
the surface to have a variety of disadvantageswithout
many compensating advantages.This particular judg-
ment is stronger,the greater is asset substitutabilityand
the less the relianceon controls. Havingsaidthis, I rec-
ognize that most small developingcountries have opted
for this intermediate outcome. There are a number of
reasons for such a result. Political considerations (in
particular, concern about the perceptions of indepen-
dence) maypreventa country from entering into a mon-
etary union, while economicconsiderations(its lack of
credibility)may lead it awayfrom exchangerate flexibil-
ity. Also, the experience under flexibleexchange rates
has made some countries wary of opting for this struc-
ture. The second-bestoutcome in such a case might be
to try to mimic monetaryunion by fixingthe currencyto
that of a crediblepartner and holding tightlyto that link.
Moreover,a country may be nervousabout tying irrevo-
cablyto another currency,becausethe country to which
it is tying may become less responsiblein the future.
Monetaryunion alsoprecludesthe adjustment of the ex-
change rate in case of a severe shock.Finally,the desire
to capture the seigniorage tax may prevent monetary
union.
Monetary Policy and Intervention Policy
This section considers more closelythe conduct of
monetary and interventionpolicy under fixedand flexi-
ble exchangerates. Considerfirstthe short-run situation
in a fixedexchangerate world.Ais the small country on
whichwe are focusingand B the larger country to which
it has tied its currency. Supposethat A decidesto relax
its monetarypolicyand ease interest rates. This willhave
both direct and indirect effects on its international re-
serves. The direct effectvia capital outflowsis generated
by the change in interest rate differentials (through
transactions of both residents and non-residents).The
indirect effectsoccur via the current account deteriora-
tion as output and prices rise in country A.It is obvious
that the higher the degree of asset substitutability,the
larger the changein internationalreserves(i.e.,the larg-
er the amount of interventionneeded)fora givenchange
in interest rates. In the limit, with perfect asset substi-
tutability,evena small changein interest rates results in
Exchange Rate Policy 127
an extremelylarge changein reserves,a reflectionof the
general point that there can be no autonomous mone-
tary policyin a fixedexchange rate system with perfect
substitutability.
Returning to the case of imperfectsubstitutability,I
would repeat and amplifyon the point made in the pre-
vious sectionthat although somewhatautonomouspol-
icy is possiblein the short run, it is not possiblein the
long run without a parity adjustment.The easier mone-
tary policyin Aresults in a capitaloutflowand a current
accountdeterioration.If Apersists in a more expansion-
ary monetarypolicy,there will probablybe some further
capitaloutflow(laggedadjustment to the earlier interest
rate change)and a continuallydeterioratingcurrent ac-
count, as domestic price inflation exceeds foreign
inflation,with a resulting declinein domesticcompeti-
tiveness.The continuing and growing current account
deficitwilleventuallyexhaustthe finiteamount of inter-
national reservesavailableto A.Moreover,wellbeforeA's
reserves are exhausted, the market will recognize the
nonviabilityof A'spoliciesand the inevitabilityof a fu-
ture devaluationif they are not changed.The resulting
capital outflow,driven by the expectationof a devalua-
tion of A'scurrency,can be considerablylarger than the
earlier interest-rate-drivencapital outflow because the
expectednet returns from getting out of investments in
a currency that one expectswill be devaluedcan be very
high (since the discrete changes tend to be very large).
In addition, if there is concern that the authorities will
impose capital or exchange controls in such circum-
stances, the outward movement of capital will be even
greater.3 Experiencesuggeststhat residentsplaya prom-
inent role in such capital outflows;it is not just foreign
investorsthat react.
In sum, in a fixed exchange rate structure a small
country'smonetary policy has to convergeto that of its
partner country in the long run, even if asset substitut-
abilityis low,and in the short run ifasset substitutability
is high.
I turn nowto the caseof a smallcountry with flexible
exchangerates. Monetarypolicycan be used in autono-
mous fashionby the authorities of the domesticcountry.
In the long run this autonomy enables the country to
choose its own inflation rate and facilitatesthe adjust-
ment to real (termsof trade) shocks.However,the ability
to choosean inflationrate can be used for goodor for ill.
Furthermore, at times the market has pushed exchange
rates too far in one direction or the other, resulting in
adjustment costs for the tradeable goods sector. These
erratic and bandwagoneffects providethe rationale for
intervention in a flexibleexchangerate system.
Howdoes monetary policy get transmitted in a flexi-
ble exchangerate world? In a small open economywith
 128 Charles Freedman
flexibleexchange rates, monetary policy is transmitted
via both interest rates and exchangerates. Andas econ-
omiesbecomemore open to foreignfinancialinfluences,
the greater will be the importance of the exchangerate
channel.In the typicalclosedeconomymodel,the tight-
ening of monetary policy operates to increase interest
rates; the higher interest rates in turn reduce interest-
sensitive expenditures.Typically,the focus is on invest-
ment expenditures,residential construction, and con-
sumer durables.In addition,spendingon other forms of
consumergoods is reducedvia the wealth effect,at least
in a world where long-term fixed-rateassets predomi-
nate. (Andin aworld with regulation-Qtypesof ceilings,
there wouldbe disintermediationand creditrationing by
financialinstitutions.) In the correspondingopen econ-
omy model with flexibleexchangerates, the tightening
of monetary policytends to increasethe value of the do-
mesticcurrency as well as to raise interest rates. The re-
sult is to shift expendituresby domesticresidents from
domesticallyproducedgoodsto imports. In addition,the
currency appreciation has a direct effecton prices, par-
ticularly in the case of the small open economywhere
the prices of both exportablesand importablesrespond
fairlydirectlyto exchangerate changes.
It is important to note that the central bank has very
little influenceon the split betweeninterest ratesand ex-
change rates of a given change in policy stance. Thus a
given tightening of policymay produce a sharp appreci-
ation and little interest rate change or little appreciation
and a significant interest rate increase.In large part the
split depends on expectationsin the foreign exchange
market, including most notably expectationsregarding
the length of time the tighter policyand higher interest
rates are expectedto last. To assist it in the conduct of
policyin these circumstances,the authorities can use a
"monetaryconditionsindex"that tries to weightboth in-
terest rate and exchange rate changes in terms of their
relativeeffecton aggregatedemand.
Beyondthe concern with its effecton aggregatede-
mand, the exchange rate is also important in the con-
duct of monetary policyfora coupleof other reasons.At
a time of inflationarypressure, one wouldwant to avoid
a sharp depreciationof the currency, because it would
feed into price changes fairlyrapidlyand have deleteri-
ous effects on expectationsof future rates of inflation.
Andthere is sometimesa concernthat marketswillover-
shoot and push exchangerates too far,especiallyif they
believethe authorities are taking a hands-offattitude to
the exchange rate ("benign neglect"). Both monetary
policyand intervention policycan be used to influence
the exchange rate in such circumstances, although
clearlythe former is by far the more powerfulinfluence.
More generally,in a world of flexibleexchangerates
with high asset substitutability,exchange market inter-
vention is not likelyto have long-run or even medium-
run influenceon the exchangerate, although it may be
useful in the short run. Thus,for example,the Canadian
authorities tend to think of intervention as, first and
foremost, a tool for promoting orderly markets and
moderating exchange rate movements in response to
shocks and temporary disruptions. The technique of
leaning against the wind is used to dampen short- run
volatilityand to offset random movements.Even in the
case of more persistent shocks and more fundamental
pressures,interventionis usedas a meansof buyingtime
in order to permit monetary policyto go to work. The
Jurgensen report, prepared some years ago for the G-7,
also reached the conclusion that intervention policy
could have some short-run effecton exchangerates but
could not be expectedto haveany lasting influence. 4 In-
tervention can be sporadicor more continuous, the lat-
ter being the case in Canada.
The use of exchange market intervention may also
have some significanceas a signaling deviceof forth-
comingmonetarypolicyaction, either automatic (in the
caseof unsterilizedintervention)or discretionary(in the
caseof sterilizedintervention).This role for intervention
can be of particular importancewhen a beliefhas devel-
oped in the market that speculativeforceshaveresulted
in a considerableovershootingof the exchangerate. If
market participants are alreadyconcerned about the vi-
ability of the currency at the prevailingexchange rate,
strong intervention, which signals that the authorities
hold the view that the currency is overpricedor under-
priced and that they are likelyto engage in more funda-
mental actionsto movethe rate, is likelyto havea direct
effecton market behavior.In the absence of clear evi-
dence of speculativeovershoot,the signalingaspect will
be less effective.
Sterilization of Changesin International
Reserves
This last section dealswith the technicalissue of ster-
ilization of reservechanges. The details of sterilization
will differfrom country to country dependingon the in-
stitutional structure (which,itself, is usuallythe result
of the historical developmentof markets and institu-
tions).The basic principleis, however,very simple.Ster-
ilization involvesaction (or nonaction in certain cases)
bythe authorities to preventchanges in internationalre-
servesfrom having secondaryeffectson domesticmon-
etary conditions through their influence on the cash
reservesof the domesticbanking system.

Considerthe followingsimpleexample.Supposethat
international reservesare held on the books of the cen-
tral bank (not a universal practice,as we shall see). The
other asset of the central bank is domesticbonds. The
central bank liabilitiesare currency, the depositsof the
banking system (reserves), and government deposits.
Supposethere is an increasein the desiredholdings by
foreignersof domesticcurrency assets, and that the au-
thorities interveneto preventan appreciationof the do-
mestic currency. The central bank issues a check upon
itselfin return forthe foreigncurrency (typicallya check
on a foreign bank). Whetherthe foreigner buys a bond
from a residentor holdsa depositin a domesticbank,the
domesticbankwill obtain the claim on the central bank
and its reservesat the central bank will rise. This will
lead to downwardpressure on short-term interest rates
as the banks with excess reserves act to expand their
portfolio of interest-bearing assets, and this downward
pressure maybe inappropriate.
What options are availableto the authorities to pre-
vent the secondary effects?They can do a number of
things: (1) The central bank can sell domesticassets on
the domesticbill or bond market. (2) It can sell the for-
eign exchangeto the domesticbanks on a swap agree-
ment. (3) It can shift government deposits from the
banks to the central bank. (4)The governmentcan issue
debt instruments and depositthe proceedsin the central
bank. (5)The reserverequirement ratio can be adjusted
upward. Note that the first and fourth options assume
the existenceof a reasonablywell-developedmarket for
domesticinstruments while options two, three, and five
do not.
It is sometimesargued that if the central bank or the
government sells domestic instruments to sterilize the
capital inflow,there willbe upwardpressure on domestic
interest rates, further capital inflows,further need for
open market sales for sterilizationpurposes,further up-
ward pressure on interest rates, etc. This argument in-
volvesa fallacybecauseit does not take into account the
destination of the original capitalinflow.One possibility
is that the foreigninvestorwants to hold domesticinter-
est-bearingassets, with the result that the central bank
indirectlyends up trading its own holdingsof, say,trea-
sury bills to the foreign investor in return for foreign
currency.Asecondpossibilityis that the foreigninvestor
wishesto hold a depositin the domesticbank.The cen-
tral bank or government can then supply an interest-
bearing asset for the bank to hold (treasury bill or
swappedforeign exchange)or it can reduce other (i.e.
government) deposits at the bank or it can force the
bank to hold more required reserves (either interest-
bearing or non-interest-bearing),all of which actions
prevent secondaryrepercussions.It is also possiblefor a
Exchange Rate Policy 129
country to impose a secondary reserve requirement,
which would force banks to hold certain assets such as
domestictreasury bills,in order to influencethe demand
by banks for the type of assets used in the sterilization
operation.
In some countries, such as Canada,international re-
servesare held on the booksof the governmentor a gov-
ernmental entity and not on the books of the central
bank. This doesnot changethe logicofthe abovediscus-
sion but rather makes sterilization automatic rather
than discretionary.In the case of Canada,for example,
international reservesare an asset in the ExchangeFund
Accountand are financedin the short run by a reduction
in governmentdepositsat the banks and in the long run
by an increase in treasury billsoutstanding. TheBank of
Canada acts as an agent. Note that if a government fi-
nanced its increased holdings of international reserves
by borrowingfrom the central bank, we would be right
back to the earlier example.
I would note in conclusion that sterilizationprevents
certain automatic responsesfrom taking place.Thus the
autonomous capital inflowjust discussedwould, if un-
sterilized,haveled to an expansionof the balancesheets
of the banks and put downwardpressure on interest
rates. Hence it would have induced a partly offsetting
capital outflow.Byengagingin sterilization,the author-
ities are trying to prevent the secondary repercussions
on the interest rate and hence insulate the domestic
economyfrom the capitalinflow.Similarly,ifthe capital
inflowwere the result ofa tightening of monetarypolicy,
the sterilizationtries to prevent downwardpressure on
interest rates, whichwouldoffsetthe originalaction. Re-
call, however,that in the limit in a fixed exchangerate
system with perfect substitutability,fully sterilized in-
tervention is not possiblebecauseof the infiniteelastic-
ity of capital flows, while nonsterilized intervention
simplyreversesthe original central bank tightening ac-
tion (leaving the central bank with more foreign ex-
change and less domesticassets on its balance sheet).
Notes
1. Capitalmobility is definedas the absenceof policy
restrictions on the movement of funds between coun-
tries. Asset substitutability is definedas the willingness
of investorsand borrowers to shift betweeninstruments
denominatedin differentcurrencies in response to very
small differencesin expectedreturns.
2. I wouldnote that no controlsare perfect.For exam-
ple, blackmarkets develop,and there are leads and lags
in current account transactions.
3. Foreign borrowingby the authorities can provide
more reserves,but market forceswill eventuallydomi-
130 Charles Freedman
nate. Interventionon the forwardforeignexchangemar-
ket has virtually identicaleffectsto intervention on the
spot foreignexchangemarket and appears on the surface
to be potentially unlimited in magnitude. However,ex-
tremely large intervention on the forwardmarket would
result in the authorities' taking on a very large risk and,
indeed,might cause counterpartiesto refuseto engagein
further transaction with the authorities because of the
perceivedrisk.
4. In a world with imperfectsubstitutability,interven-
tion can have last effects but these are quantitatively
small.

Interaction of ExchangeRate Policy and MonetaryPolicy:
The Caseof Malaysia
Lin See Yan,
After listening to Dr. Freedman,I think that what I am
going to say is very pedestrian.He has covereda lot of
ground, especiallyon the theoreticalside. I would like to
move awayfrom the theory and go into the nuts and
bolts of operations. One thing I havelearned over many
years is that, whilethe theory is nice to have, in develop-
ing countries where things are always imperfect,very
imperfect,even simple statements of theory cannot be
takenfor granted.Asan example,just nowDr.Freedman
referredso easilyto the option of using governmentde-
positsin commercialbanksto sterilizeforeignexchange
interventions.Andyet, it took me four years to convince
the governmentto allowthe centralbank to handle that.
First, then, I shall talk brieflyabout the Malaysianex-
perience,which has some unique aspects. Within that
context, I shall then describehow we operate monetary
policy,with particular regard to exchangerate policyand
the problemswe face.
M1alaysia'sEconomic Background
I am sure many ofyou haveheard of Malaysia,but you
mayknow little about it. A perspectiveis important, for
mayssonknowslittle
aboutit.Aprspmectiviswimhwportante,fo
it is difficultto appreciatewhat we havedone. The start-
ingpoint for this perspectiveon Malaysiais 1980.For the
twenty-fiveyearsprior to 1980,Malaysiahad been grow-
ing fast with practically no inflation. It is one of those
countries that some describeas allergicto inflation.We
regard inflation as being like toothpaste;once it's out, it
is difficultto push backin. At present, our inflation rate
is 2.2 percent,and alreadypeopleare talkingabout infla-
tionaryexpectations.So, yes,we are allergicto inflation.
I just came back from Geneva,where I was talking to
my Latin Americanfriends. It is very difficultto keep in
step; I am alwaysunsure whether they are speaking of
hourly,weekly,daily,or annual inflationrates. I say this
131
11
not to belittle but just to showthe differencein percep-
tions of the same thing.
AsI said, the starting point is 1980;until then wehad
been doing very well. Like all Western-trainedecono-
mists,I thought in 1980that the worldrecessioncoming
through then wouldbe short. Everybody-the big brains
all over the world including the World Bank and the
IMF-thought it would be a short recession.Likegood
Keynesians,wedecidedto spend our waythrough the re-
cession.That was a big mistake for us. By 1983,three
years later, we were running a government deficit of
close to 19 percent of GNPand a balance of payments
deficit of around 14 percent. Bythen we had increased
our externaldebt four times, and the public debtat that
time was close to 130 percent of GNP.
By the way,when the recession began in 1980,the
prices of everything that Malaysiaproducesfell at the
same time. I recall that the WorldBank had advisedus
that such a convergencewouldbe statisticallya near im-
possibility.In 1980,we had everythingfallingfor us at
the sametime,and thenit happenedagainin 1985.
Clearlythe situationwasnot sustainableforus, so we
decidedto pull back in 1983.This was done on our own
volition,without any help from the WorldBank or the
IMF.By 1984,dust beforethe onset of the secondreces-
percent of GNPto 8 percent. The balance of payments
deficitfell from 14percent almost to balance,with just a
small deficit.
Then the secondwhammycame through. We had to
manage it and so could not makefurther progress for a
while.But within another year and a half,we turned the
govemment deficit around to only 2-3 percent of GNP
and had a balance of payments surplus as large as the
deficitwe had in 1980.Allthis time, inflationwas from
3.5to 4.5 percent. During this period,we alsodecreased
the external public debt significantly.The ratio of exter-
nal debt serviceto exports had been 20 percent. Bythe
132 Lin See Yan
end of this year, we will have reduced debt service to 9
percent of exports.
ExchangeRate Policy and MonetaryPolicy
To put these changes in perspective,you must re-
member that Malaysiais a small country but very open
with respectto capitalflows.The only exchangecontrol,
ironically,is inward rather than outward.Malaysiahas a
floating exchangerate. It floats against a basket rather
than one particularcurrency.
Moreimportantly,we turned the economyaround by
completelyrestructuring it. Wemadethe private sector,
not the government, the engine of growth. In the pro-
cess, government expendituredecreasedfrom a high of
42 percent of GNPto less than 30 percent today,and it is
still falling. That means downsizinggovernment and
privatization.
The capital markets in Malaysia,the money market
and the foreignexchange market, are not highly devel-
oped.In addition,this open economyis situated only250
miles from Singapore,which welcomescapital inflows
with open arms. Hong Kong,which is just three hours
away,does the same. So money flows easily.Very few
monetary policy instruments are availablein such a sit-
uation. Mostof those whichwe learn from the textbooks
are not easy to implementin such an economy.
In Malaysiaas in most developingcountries,the me-
dium-term goal is clear: to have growth with stability.
Yet,in the processof goingfrom here to there, one must
remember that the means to that goalare never clear to
politicians, nor to many economists.In practice, there
are multiple objectivesand insufficientinstruments to
target them.
Certainlythough, the more fundamental objectives
include such things as low inflationwith high growth.
During the last four years, Malaysia'sgrowth rate has
been 8 to 8.5 percent in real terms. As I said, we have
handed the whole economy over to the private sector,
with exportsas the main engine of growth.
A matrix can be drawn that charts these objectives
against the availablemeans of achievingthem. On one
side are the objectivesof low inflation,high growth,and
dynamicexports.On the other sidewe list monetarypol-
icy,which for us works directly through liquidityman-
agement; credit policy,for which interest rates are the
main instrument; and exchangerate policy.Withobjec-
tives like those and limited instruments, and given the
sharp structural adjustments in our environmentduring
the past ten years, exceedinglydifficult dilemmas are
bound to occur again and again. The real problem of the
central bank is howto keepall these things in balance.
Withoutgoinginto too much detail,I want to provide
a taste of the dilemmaswe face. For example,as I just
mentioned we are now in our fifth consecutiveyear of
rapid growth. Wewill have growth of about 8.5 percent
in real terms. We hope inflation for the year as a whole
will not exceed4 percent. However,the moneysupply is
growing fast because of these four years of very rapid
growth.There was much pent-up demandthat had been
held back in the early 1980sby the strict structural ad-
justments. So money growth is running now at 18-20
percent, while a range of 14-15 percent is more to our
liking. In this environment,I think it is important to try
to pull back. Monetarypolicy has to become more re-
strictive. Creditpolicyhas to be somewhatrestrictive,so
interest rates must rise. Exchangerate policyshould be
at least neutral, even a bit below,so that the currency
will appreciate.
At the same time, we have a situation where the pri-
vate sector was brought in just a short time ago to take
over from government as the engine of growth. So we
must be verysensitivein managing monetarypolicy,es-
peciallyinterest rates, to ensure that a rise in rates does
not choke off these sharp increases in investment. On
the other hand, if we do not raise them enough, capital
outflowsmaybecomeso significantthat reservesare se-
riously drained.
The export orientation of the economycreates addi-
tional dilemmas.This year exportswill account for close
to 45 percent of total output. Onlyfiveor sixyears ago,
they were just 22 percent of total output, and the share
will rise to 65 percent in another three years. To ensure
that exportscontinue to grow,this export-drivenecono-
my requires policy to be relativelyeasy. Interest rates
cannot be too high, or they will choke offthe strong in-
flowof foreigninvestment.Wemust also ensurethat the
strong balance of payments position that has developed
over the last four to five years does not raise the ex-
change rate to the point that exports become noncom-
petitive.
Wherever we turn, these conflicts and dilemmas
arise. At present, we are trying to developa capital mar-
ket and a moneymarket that giveus sufficientflexibility
and adequate instruments to succeed in this balancing
game. Even so, it is practicallyimpossibleto maintain a
balance.
Overthe pastfewyears,wehavebeensterilizingthose
large inflowsof capital,both portfolioand direct invest-
ment capital, by prepayment of public debt. In conse-
quenceour debt has declinedenormously;as I said,total
external debt serviceby the end of this yearwill be only
9 percent of exports.But the capitalinflowsare still com-
ing.
 Interaction of Exchange Rate Policy and Monietaiy Policy: The Case of Malaysia
At the same time, Malaysiahas had huge outflowsof
capitalto take advantageof high interest rates abroad.To
giveyou an idea, Malaysiatoday probablyhas the lowest
interest rates in the world;our money market rates for
3-month or 6-month moneyare around 5.5or 6 percent.
We have a base lending rate of 6.75 percent, but busi-
nesses can finance themselvesthrough bankers' accep-
tances and export refinancingfacilitiesat 4.5 percent.
The current rate for 12-monthdepositsis 5 percent.
So here again we have a delicate situation. We have
managedso far becausethe inflowsof capital havebeen
so strong that the outflowfrom low interest rates was a
blessing.The outflowsworkedto sterilizea lot ofthe in-
flow.Still, our markets are not as developedas in the
West.There are not enough instruments to sterilize the
inflows,sowe needto createnew ones.Atfirstwhena lot
of capital came in, we just deflectedit through prepay-
ments of debt,but there is a limitto that. Not all debt can
be prepaid.Andyour creditors begin to complainwhen
you repay too much; they prefer to hold your paper.Nev-
ertheless, our government long-term inflowshave been
negativeforfour years.So we haveto create other means
of dealingwith the capital inflow.For example,we may
have to persuade our foreigndirect investorsto bring in
only M$20million,' rather than M$100million, and to
borrow monei locallyso that the banks can restructure
balance
oheir
sheets. Wehaveto think of things like that.
shereis.Wehavetoth of
Nevertheialae liquiithings
le that
Nevertheless,there IS still a lot of liquidityin the mar-
ket. So we do the same things that everyonedoes. We
raise required reserves or the liquidity ratios, and the
banks complainbecausethey haveto raise interest rates.
But do not forget that all the time the balance must be
Buintainedo
ntogthaltti thbacmse
maintained.
Sinceh egovenment wa insurs-th overn
ment had money flowingout of Its ears-on occasion
the treasury wantedto stop issuingbills.They askedwhy
wewere pressing them to issue more bills to mop up the
liquidity;that is the central bank's problem. So instead
of issuing bills,the next best thing we could think of was
to accept deposits.We began to mop up the liquidityby
acceptingdepositsfrom the banks. That is an easywayof
doing it, although it costs a lot. Our job is not to make
money.
Yeta simpleidea like this seemsterrible in the eyes of
a developingcountry. It is important that people have
confidencein the central bank. If they see your profits
fall, they do not understand that your motivationis not
to make a profit; they think something is going wrong
with the central bank. Outflowsof money will start be-
cause of that opinion. Therefore,such actions must be
carefullymanagedto ensure that your profitsdo not fall
too heavilywhenyou are moppingup reserves.
133
We think of many things to do; we do swaps.But do
not forgetthat swapsare not easyforour country.Malay-
sia is a smallcountry that has been in the shadowof Sin-
gaporefor manyyears.Wetry to be independentofthem
and create our own swapmarket.
These things we can manage;we can balance some
wayor other. But the most difficultpart occurswhen the
market gets itchy.The exchangemarket begins to feel
that it is time to speculateagainst the ringgit and does
so. Wehave not had this problem recently,but over the
last ten years it happenedmany times. Supposethe fun-
damentals and the exchangerate are not moving in the
same directions for obviousreasons:capital flowsand so
on. Thebitter lessonwehavelearned is that you must be
veryagile,whichthe textbooksdo not teach. Youhaveto
do it in such a way that, if necessary,you haveto accept
expensivetrade-offs.
Let me givean example.At a time when wewere go-
ing through that difficult structural adjustment, the
speculatorsthought it was time to take positionsagainst
the Malaysianringgit. Becausewe were in a recession,
interest rates were low, monetary policy was easy,and
ringgit were easily available.So the speculators start-
ed-and they alwaysstart in NewYork,while you are
sleeping.
To put the situation in perspective,we have reserves
today of roughly US$8 billion. That is equivalent to
about four and a half months of imports. In one day of
spcltoagitthcurnywcnlseablonf
speculationagamst the currencywe can lose a bllion of
For a country like Malaysia,with just US$8billionin re-
Forveco a si a, billion in re-
serves,losing a billion of it is no joke. The only way to
handlea speculativesituation like that is to hit the spec-
ulators hard, very hard. Whatwe do is makeit extremely
difficultfor them to get ringgit. We squeezethe money
market, though obviouslyat the expense of domestic
monetary policy,At that time our monetary policy was
easy,for we had to reviveconfidence.But in speculative
bouts like that-and over the last ten years we havehad
three or four such occasions-our experiencehas been
that unless you act very firmlyagainst them, the specu-
lators will never leaveyou alone. Andyou will never be
able to conduct monetary policy.
So even at the expenseof trade-offsagainst our mon-
etary policy,today we would not hesitate to squeezethe
money market to a 100 percent per day rate of interest.
Our past four experienceshavebeen that ifyou do that,
if you make life extremelypainful for the speculators,
then they will never touch you again. In fact the best
outcome would be to make them default and come to
you beggingon their kneesfor ringgit.Wehavenot gone
that far.The furthest we havegone is to raise the money
market rate to 100 percent and make presidentsof spec-
 134 Lin See Yan
ulative companiescometo me beggingfor ringgit. They
will never go after the ringgit again.
Theseare drasticmeasures,but you haveto improvise
like that. Being a small economy with not enough in-
struments, with openness in your economy,and dilem-
mas like that, you have to act very firmly.Andso, trade-
offs must be taken; you must be decisive.One result is
that we nowoperate24 hours a day.I run my foreignex-
change shop around the clock in three shifts, to make
sure nobodytakes advantageof us. In that way,we donot
lose hard-earned reservessimplybecausesomeonetries
to make a quick buck on us.
The point of this exampleis that in a small and open
economyit is very difficultto conduct monetary policy,
exchangerate policy,or interest rate policyby the meth-
ods the textbooksteach. Most of the time, the textbook
approachesdo not work. Youhave to think of whatever
way you can to work toward whatever objectives you
have. Youhaveto innovateall the time. We do that, and
I can givemore examplesthat show it can work.
Always,the key is to foster in the market, whether in
the local or foreign markets, a respect for your policies.
If somethingis working againstyou that you do not like,
if it is working against the effectiveattainment of your
policies,you must act firmly, even to the extent of ac-
ceptingseveretrade-offslike those I mentioned.
Wealsohavefound that whileyou mayhaveto accept
these trade-offs,there are ways to ameliorate their im-
pact. For example,when we had to squeeze the market
and raise interest rates very high to kill offthe specula-
tion, weset up new funds forwhat wecall priorityinvest-
ments. To bring down the interest rate for investments
that willincrease capacity,the central bank stands ready
Discussion
Question:Why doesn't Malaysialet the currency ap-
preciatemore to dealwith the inflationaryconsequences
of the favorablecurrent account and the large capital in-
flows?
Mr.Lin: Don't forget there are no inflationaryconse-
quences at this point in time. Weare sensitiveto the ex-
changerate becauseformany yearswehad an overvalued
exchangerate. Also,it is very difficultto explainto inves-
tors who havecome to you on the expectationthat your
exchangerate will be stablewhen suddenlythey are faced
with a 20 percent or 30 percent appreciationof your cur-
rency. I think that would be an act of bad faith. Foreign
investment,which was crucial in tuming our economy
around, accounted in 1987for about two-thirds of total
investment,so the exchangerate is very important.
to lend to them, either directly or through the redis-
count window,at rates that are more in line with normal
times. In our case, these priority investments included
manufacturing,plantations,and so on. In manyinstanc-
es, we haveset up funds like that. To ensurethat our ex-
treme tightening did not have an adverse impact on
investment,we thought of new ideas on how to help.
During the period when you have to handle an ex-
change rate problem,you haveto findwaysto makesure
life at home can go on as normally as possible.At the
right time, when the speculatorsare off for the day,you
just go backto normal.Fortunately,this approachis not
difficultto manage in a small economy.
Conclusion
Althoughwe have done many things in Malaysiathat
someof you haveprobablynever heard ofbeing done be-
fore, we do them becausethey work. With that general
point, ladiesand gentlemen, I shall stop. Mymessageto
you who operate in a small economyis that you do not
havethe luxuryto do many things my friend Mr.Freed-
man can do in Canada,things that any central bank in
the developedworld can do. To reach our objectives,we
must make do with what we have.
Notes
1. The author is DeputyGovernor,Bank NegaraMa-
laysia.
2. The currency ofMalaysiais the ringgit, whichis ab-
breviatedM$.
We don't think our current exchangerate is inappro-
priate. We moved back to this point when we went
through the structural adjustment. We went through a
periodof overvaluation,mainlybecauseof Singapore.We
were on a one-to-one peg to them for many years. Be-
cause of Singapore'sstrong economyas a financialcen-
ter, the political, and more importantly, the general
feelingwas alwaysto maintain this one-to-one relation-
ship. It took us fiveyears to convincepoliticiansand the
general public that it was all right to have a weaker ex-
change rate. So it is fundamentalfor private sector con-
fidence, in our context certainly, that we manage the
exchangerate in such a wayas to avoidsharp gyrations.
Dr.Freedmanearlierreferredto it as maintainingorderly
markets.
 Interaction of ExchiangeRate Policy and Monetary Policy: The Case of Malaysia
Question:In your exchangerate policy,am I correct
that you are assigning a real exchangerate on the basis
of a basket of currencies?And do you assign it weekly,
monthly or how often?
Mr.Lin: Yes[to the first question].Daily[to the sec-
ond question].Wehavea softwareprogram to handle it.
It is not difficultto developit. The problemis alwayshow
to get your economicspeople,who turn out all these bas-
kets and effectiveexchangerates, togetherwith your op-
erators. Although we are pegged to a basket, our
intervention currency is the U.S.dollar. If, for example,
the basket movesup too much [againstthe U.S.dollar],
moves up by a certain number of points; the operator
must know,by pressinga button, what he needsto do on
the U.S. dollar side to keep the rate stable. It is not too
difficultto developthat software.It allowsthe econom-
ics people,the research people,the monetarypolicypeo-
ple, to talk on the same wavelength as the operator.
That'svery important.
Question:Myquestion is on the instrument you use
to affectinterest rates. Is it through rediscounting?
Mr.Lin:Weaffectinterest rates through management
of liquidity,forwhichwe use the normal range of instru-
ments. For example,when we see the first signs of inter-
est rate changes, we will raise required reserves to
manage liquidity.However,with the capital inflowswe
have had, not just in direct investment but also in port-
folio investment, these traditional instruments are far
too inflexible.For example,to move required reserves,
we must wait for the middle of the month and dealwith
other complications.But in an open economy,actions
should be taken today or tomorrow. New instruments
that we havecreatedare more convenient.For example.
wejust go into the market and mop up [liquidity]bybor-
rowing.Nothingis faster than that, and it is very conve-
nient. Wejust movein and take out liquidity.
For many years, [our open market operations] have
faced difficultiesbecause the government is flush with
funds.Yetcontractionarystepstaken bythe central bank
tended to be offsetby Treasuryaction. It was very diffi-
cult to convincethe government that the central bank
should also manage the government'sso-calledexcess
reserves.But nowthe central bank has much better con-
trol of liquidity.To be able to marshal all the excessre-
serves of the public sector into the central bank is a
powerfulinstrument. It is an extremely useful instru-
ment to have.
Question: When speculators take positions against
your currency, you have increased the interest rate
135
sharply.On those occasions,you givesubsidizedcredits
to priority investments.How can you ensure that these
subsidizedcredits are not also used for speculation?
Mr. Lin: Oh, those funds are for actual projects. You
can only borrow from the special funds for actual
projects.
Question:But howcan you keepthem from joining in
the speculationagainst the currency?
Mr. Lin: I do not release my money until the investor
needs to paya bill. For example,if the investor is build-
ing a factory,I release the moneywhen the machinery
comes.
Question:Supposethey have their own resources to
apply to their activities, and they use the subsidized
moneyyou give them and speculateagainst you? What
willyou do?
Mr. Lin: I will come down very hard on them. But
these people knowthat.
Mr.Honohan: Haveyou ever consideredthis alterna-
tive. Instead of increasinginterest rates dramaticallyto
hit the speculator,have you ever allowedthe ringgit to
fall 3 or 4 percent, then suddenlyraise it 5 percent?That
approach would feed backless severelyon your priority
projects.
Mr.Lin: Yes,we have tried that. The problem is that
when you operate an exchange,you are never sure how
far you can go,becausethe market willmove you. If you
are not careful, the speculators will make money from
you.Andthat is the worst situation to everget into. Once
they havetasted blood,they never let go. So what you do
is to hit them reallyhard. Wehavetried other ways,but
as I said,the moment they make moneyout ofyou, they
know they can do it again.
Mr.Queiroz:Malaysiais small,so you can knowthese
things. In large countries it is not as easy.
Mr. Lin:Yes,I think our experiencewould not be ap-
plicableto everyone.Still,we don't makethese decisions
ourselves,we do it through the banks. For the banks, it
is part of standard procedure. In our banking system,
when the banks lend out for projects, we require
through our inspection system that the money goes
where it is supposedto go. Of course the system is not
perfect,but it is a systemimplementedwithinthe banks.
136 Lin See Yan
Mr.Queiroz:Howcan a centralbank at the same time
act as a developmentbank?Thosetwo functionsare well
separated.If you introduce a specialwindowfor project
finance,but fund speculationonly at a differentinterest
rate, overtime it all becomesone. So the strategyis very
risky.
We used to do this in Brazil. In December1979,the
government had a maxi-devaluation,followingwhich
they reduced interest rates. The objectivewas to boost
exports because the devaluation would make exports
cheaper.The problem was that peoplestarted to specu-
late, expectinganother devaluationbecausethe govern-
ment loweredinterest rates. That is just what you were
doing on those occasionsyou described.So I don't think
you havemuch chance of maintaining that strategyover
time. Interest ratescannot be high forsomebodyand low
for someoneelse. Either interest rates are low forevery-
bodyor high for everybody.
Mr. Lin: We maintain that distinction only for the
short term, the very short term [during which specula-
tion againstthe ringgit is active].Weare an unusual cen-
tral bank. For many years, we were a standard central
bank.Wedid everythingthat central bankers in the West
do.I remember,whenEugene Blackwaspresident of the
World Bank, my first governorwent to him for advice.
EugeneBlacksaid, "For heaven'ssake, you are a central
bank, neverget into developmentfinanceand planning."
But that is exactlywhat we did, and for severalreasons.
In a developingeconomy,where there is little or no
infrastructure, where everything must be developed
from scratch, the central bank cannot behave like a
Western central bank. That is reason number one. The
second reason may sound very hard, but I have been in
this game for twenty-eight years. I have learned that if
you don't take care ofyourself,nobodywill.
Let me pose the followingclassic example,which is
certainly controversial.The Group of Seven (G-7) can
make policy pronouncementsany way they want. They
get together oneweekendand decidethat the yen should
appreciate by 20 or 30 percent. People like us, sitting
milesaway,hear about it on the television;suddenlyour
debt has increased by 20 percent. And yet the central
banks say you must behavelike a normal central bank.
Don't rock the boat.
What doesone do?I learned from the game. If I know
that rates are going up, I take a position.That is not be-
havinglike a "normal"Western central bank, obviously.
But for heaven's sake,you tell me what I should do. Ev-
erybody knew that the yen was going to appreciate at
that time; what should the central bank do? Should we
sit backand, since a large part of our assets are denomi-
nated in U.S. dollars,see that portion of it lose 20 per-
cent of its value?
That is an interesting question.The "normal"central
banks ride it out over the long term. Things will be all
right. Butthe point is that things are not all right tomor-
row or the dayafter tomorrow.Youmust face the Minis-
ter of Financewho saysthat as a result of my decisionhis
debt goes up 30 percent; why didn't I do something
about it?
Ever since the world went off Bretton Woods,the
whole game of central banking has changed. It is well
and good for people,including the WorldBank and the
IMF,to giveadvice.Weappreciatethe advice,but much
of that advice comes without responsibility.I, on the
other hand, have to livewith the responsibilityif I take
your advice.If it goes wrong,I livewith it. AndI haveto
turn around a situation that has not workedor be trans-
ferred to another department. For us, this is the school
of hard knocks.
Centralbanks todayare not differentfrom other insti-
tutions. They have to be development-oriented.To the
extent that some action achievesthe national objectives,
we willdo it, just as we havedone it in the past. As I said,
we are very controversialtoday.The IMFdoes not like us
becausewe try to protect ourselvesfrom G-7 decisions.
Wehave explainedour positionto them and I think they
understand.We are sorry, but we believeit is important
fora central banker to be able to say,"Wewant to control
our destiny."
Mr. Freedman:Exceptfor the 1985case,where some
argue it was the market and not the G-5that drove the
dollar down,I don't think there is a singlecasewhere the
G-7decisionshad a major effecton the market.Andeven
in the 1985case the dollar was clearlyovervalued;every-
one knewit was.Martin Feldsteinargues that it did not
haveanything to dowith the G-7.
Usuallythe G-7pronouncementsare things like,"We
would like . . ." or "It isjust about right, but if it goestoo
far we are goingto move."It is alwaysmuch more vague
than [a decision on howmuch an exchangerate should
change.] In fact, it is the market that has been driving
the rate changes.
Mr.Lin:Whatyou say is quite true, but there is more
to it than that. Wedon't knowwhat goeson in G-7meet-
ings;we can onlyspeculate.Theymight not havedone it
to date, but they could do it tomorrow or the dayafter.
Mypoint is that within that group-and now the latest
writings talk ofa group of just three, the G-3now-they
can gang together and do whatever they want with re-
spect to interest rates and exchangerates. Whateverthey
dohas a directimpact on us. Wehaveto defendourselves
 Interaction of ExchangeRatePolicy anddMonetaryPolicy: Tle Case of Malaysia 137

in the market. I am sayingwe havebecomea streetwise
central bank, so to speak,over the years.
Mr.Freedman:Myonly point is that you are overstat-
ing the power of the central banks and their govern-
ments to affect these things. They are usually not
* preparedto take the fundamentalactions that it wouldin
fact take to have those effects.
Also,I supposethat if you are in that business,you are
part of the market. Youare going to win some and lose
some.Andyou are going to haveto take that responsibil-
ity.
Mr.Lin: Exactly;my point is that the traditional cen-
tral bank just sits it out.
The World Bank
Headquarters
1818H Street, N.W.
Washington, D.C. 20433,U.S.A.
Telephone: (202)477-J234
Facsimile:(202)477-6391
Telex: wui 64145WORLDBANK
RCA248423WORLDBK
Cable Address: INTBAFRAD
WASHINGTONDC
0 C)77-
EuropeanOffice Tokyo Office
66, avenue d'Iena Kokusai Building
75116Paris, France 1-1Marunouchi 3-chome
Chiyoda-ku,Tokyo 100,Japan
Telephone: (1)40.69.30.00
Facsimile:(1) 40.69.30.66 Telephone: (3) 3214-5001
Telex:640651 Facsimile:(3) 3214-3657
Telex:26838
Pol pietcroS,
j
ISBN 0-8213-1969-8