Official Foreign Exchange Intervention

O C C A S I O N A L PA P E R 249

Official Foreign Exchange Intervention

Shogo Ishii, Jorge Iván Canales-Kriljenko,Roberto Guimarães, and Cem Karacadag

INTERNATIONAL MONETARY FUND

Washington DC

2006

©International Monetary Fund. Not for Redistribution

© 2006 International Monetary Fund

Production: IMF Multimedia Services DivisionTypesetting and Figures: Choon Lee and Wendy Arnold

Cataloging-in-Publication Data

Official foreign exchange intervention / Shogo Ishii . . . [et al.] — Washington, D.C. : International Monetary Fund, 2006.

p. cm. — (Occasional paper ; 249)

Includes bibliographical references. ISBN 1-58906-421-6

1. Foreign exchange. 2. Foreign exchange — Developing countries.3. Foreign exchange — Mexico. 4. Foreign exchange — Turkey. I. Ishii, Shogo. II. Occasional paper (International Monetary Fund) ; no. 249HG3821.O34 2006

Price: US$25.00(US$22.00 to full-time faculty members and

students at universities and colleges)

Please send orders to:International Monetary Fund, Publication Services

700 19th Street, N.W., Washington, D.C. 20431, U.S.A.Tel.: (202) 623-7430 Telefax: (202) 623-7201

E-mail: [email protected]: http://www.imf.org

recycled paper


http://www.imf.org

Preface v

I Introduction 1Shogo Ishii

II Best Practices in Official Interventions in the Foreign 2Exchange MarketJorge Iván Canales-Kriljenko, Roberto Guimarães, and

Cem Karacadag

How Can Intervention Be Effective? 2Trends in Foreign Exchange Intervention 5Policy Issues 5Intervention Operations 12Conclusions 15

III Survey of Foreign Exchange Intervention in 18Developing CountriesJorge Iván Canales-Kriljenko

Prevalence of Foreign Exchange Intervention 18Sterilized or Not Sterilized? 18Relative Size of Foreign Exchange Intervention 20Information Advantage 25Conclusions 26

IV The Empirics of Foreign Exchange Intervention in Emerging 27Markets: Mexico and TurkeyRoberto Guimarães and Cem Karacadag

Empirical Analysis and Evidence on Intervention 28Policy Context of Intervention in Mexico and Turkey 30Effectiveness of Foreign Exchange Intervention 33Conclusions 40Appendix 41

References 43

Boxes2.1. Intervention Through Options 142.2. Choice of Counterparties and Transparency 16

Tables

3.1. Foreign Exchange Intervention by Exchange Rate Regime and Market Access, 2001 19

Contents

iii


fresher

Underline

CONTENTS

3.2. Characteristics of Foreign Exchange Intervention in Developing and Transition Economies, 2001 20

3.3. Developing and Transition Economies That Do Not Always Sterilize Foreign Exchange Intervention, by Exchange Rate Regime and Market Access, 2001 21

3.4. Magnitude of Foreign Exchange Intervention in Selected Developing and Transition Economies, 2000 22

3.5. Selected Regulations on Forward Foreign Exchange Transactions in Developing and Transition Economies, 2001 24

4.1. Turkey: Central Bank Foreign Exchange Intervention,March 2001–December 2003 32

4.2. Mexico: Descriptive Statistics 354.3. Turkey: Descriptive Statistics 364.4. Mexico and Turkey: Asymmetric Component GARCH Model

Estimates 384.5. Mexico and Turkey: Probit Model Estimates 39

A4.1. Analytical Methodologies of Empirical Studies on Intervention Effects on the Exchange Rate 41

Figures

2.1. Intervention and Exchange Rate Impact 84.1. Mexico: Exchange Rate, Interest Rate, and Inflation 304.2. Mexico: Exchange Rate and Foreign Exchange Intervention 314.3. Turkey: Central Bank Intervention and the Exchange Rate,

March 29, 2001–October 3, 2003 334.4. Turkey: Central Bank Intervention and Exchange Rate Returns,

March 29, 2001–October 3, 2003 344.5. Turkey: Exchange Rate Trend and Volatility 34

iv

The following symbols have been used throughout this paper:

. . . to indicate that data are not available;

— to indicate that the figure is zero or less than half the final digit shown, or that the itemdoes not exist;

– between years or months (e.g., 2004–05 or January–June) to indicate the years or monthscovered, including the beginning and ending years or months; and

/ between years (e.g., 2004/05) to indicate a fiscal (financial) year.

“Billion” means a thousand million.

Minor discrepancies between constituent figures and totals are due to rounding.

The term “country,” as used in this paper, does not in all cases refer to a territorial entity that is a state as understood by international law and practice; the term also covers someterritorial entities that are not states, but for which statistical data are maintained and providedinternationally on a separate and independent basis.


This Occasional Paper aims to develop a deeper understanding of foreign exchangeintervention in emerging markets. Central banks intervene in the foreign exchangemarket for several reasons, including to calm disorderly markets, correct misalign-ments, and accumulate reserves. The prevalence of central bank intervention in emerg-ing markets has led to renewed interest in how central banks should intervene tomaximize their efficacy. This paper sheds light on a number of operational aspects ofintervention. It also presents evidence on intervention practices and characteristicsbased on a survey on the organization of foreign exchange markets in developing coun-tries. The survey was carried out in 2001 by the International Monetary Fund. Finally,the paper presents empirical evidence on the effectiveness of intervention in Mexicoand Turkey, two countries where intervention data are publicly available. The authorsemphasize that intervention is not an independent policy tool and is most effectivewhen the exchange rate policy is consistent with other macroeconomic policies.

The authors are grateful to Stefan Ingves and Herve Ferhani for their support forthis project. We would like to thank Kai Barvell, Hali Edison, Shyamala Gopinath,Scott Roger, Barry Topf, and Mark Zelmer for insightful comments on earlier drafts.The authors are also indebted to Nirmaleen F. Jayawardane and Ranee Sirihorachaifor their excellent assistance in the preparation of the manuscript, and to Gail Berreand David Einhorn of the External Relations Department for editing and coordinatingproduction of the publication.

The views expressed in this paper are solely those of the authors and do not neces-sarily reflect the views or policies of the national authorities or the International Mon-etary Fund or its Executive Directors.

Preface

v


This page intentionally left blank


Central banks operating flexible exchange rateregimes intervene in the foreign exchange mar-

ket for a variety of reasons. In developing countries,prevalent among those reasons are to correct mis-alignment or stabilize the exchange rate, calm disor-derly markets, accumulate reserves, and supplyforeign exchange to the market when the public sec-tor is a prime foreign exchange earner and the centralbank is the public sector’s foreign exchange agent.1

Risks associated with central bank intervention arehigh, particularly in developing countries. Interven-tion puts the central bank’s credibility and scarce for-eign exchange reserves at risk. The depletion ofreserves by Mexico in 1994 and by Thailand in 1997while defending their currencies was an importantfactor in their respective financial crises.

Intervention is widespread in developing coun-tries, in contrast to its steady decline in advancedeconomies. This pattern may reflect the fact thatexchange rate stability commands a high premium ineconomies where liability dollarization and pass-through from exchange rate movements to inflationare higher. According to a 2001 survey on the orga-nization of foreign exchange markets carried out bythe International Monetary Fund, central banks inmany developing countries intervene in the foreignexchange market frequently and in amounts that arelarge relative to total market turnover. They also tendto back up their interventions with monetary policyto enhance their effectiveness in stabilizing theexchange rate. In other words, these central banksoften do not sterilize the impact of their foreignexchange purchases and sales on domestic moneysupply.

Despite a vast literature on the effectiveness ofintervention in advanced economies, empiricalresearch on its effectiveness in developing countriesis limited. Similarly, there are few sources of guid-ance on the operational issues and best practices in

this area for developing countries. This paperattempts to shed light on a number of questionsabout the mechanics of interventions by centralbanks, including the timing and amount, rules versusdiscretion, degree of transparency, and the choice ofmarkets and counterparties. It also presents empiri-cal evidence of the effectiveness of interventions inMexico and Turkey, where daily intervention dataare publicly available.

Throughout this paper it is emphasized that inter-vention is not an independent policy tool. Its successis conditional on the consistency of targetedexchange rates with macroeconomic policies. Thepaper also presents a number of reasons for centralbanks to be selective in their interventions and parsi-monious in their use of foreign exchange reserves:• Exchange rate misalignments and disorderly

markets—the most common justifications forintervention—are extremely difficult to detect.There is no consensus on a methodology to esti-mate the equilibrium exchange rate.

• Determining the timing and amount of interven-tion is a highly judgmental exercise, and the twodepend heavily on such factors as changing marketconditions, the nature of economic shocks, andavailable reserves.

• Empirical evidence on the effectiveness of inter-vention in influencing the exchange rate is mixed,and even where favorable evidence is found, theimpact of intervention on the exchange rate level isshort lived. Similarly, empirical studies find thatintervention tends to increase exchange ratevolatility under flexible exchange rate regimes.

It is important that intervention policies and objec-tives be transparent and clearly specified and thatdecisions to intervene be made after rigorous analy-sis of market conditions. Transparency in interven-tion objectives can enhance the credibility of thecentral bank by holding it accountable for its recordof policy implementation, even though the degree oftransparency in the tactical implementation of inter-vention policies may vary with the specific objec-tives involved.

I Introduction

Shogo Ishii

1

1In this paper, the term “developing countries” includes emerg-ing and transition economies.


This chapter provides an overview of the policy,technical, and administrative questions that

must be addressed to effectively intervene in theforeign exchange market, particularly in develop-ing economies with flexible exchange rate regimes.It includes a review of selected country experiencesand the academic literature on operational aspectsof official intervention. Key issues include thefollowing:• Amount and timing. When and in what amounts

should a central bank intervene in the foreignexchange market? Should official interventionsbe based on rules or discretionary? What marketfactors (liquidity, order flow, etc.) should be usedto help determine the timing and amount ofintervention?

• Degree of transparency (secret versus public).Should central bank interventions be announced orkept secret? What are the pros and cons of secrecyversus openness?

• Markets and counterparties. In which currencypair, instruments (spot or forward contracts), andtrading locations (onshore or offshore) shouldintervention take place? With whom should thecentral bank trade (any authorized dealer or pri-mary dealers), and how should it approach them(directly through agents, or through brokers) toachieve its intervention objectives?The focus in this chapter is primarily on interven-

tion under flexible exchange rate regimes. Undermore rigid exchange rate arrangements, includingvarious forms of pegs, central banks have little dis-cretion over intervention policies. Official foreigncurrency sales and purchases automatically bridgethe gap between supply and demand to ensure equi-librium at the predetermined exchange rate. The pol-icy trade-offs and operational issues discussed hereapply mainly to countries with independently float-ing or managed floating exchange rate regimes inwhich the monetary policy framework is notanchored by an exchange rate target.

Intervention can be defined as official purchasesand sales of foreign exchange to achieve one ormore of the following four objectives: (i) to moder-ate exchange rate fluctuations and correct misalign-

ment, (ii) to address disorderly market conditions,1

(iii) to accumulate foreign exchange reserves, and(iv) to supply foreign exchange to the market. Theaim is to capture what are known to be widelyadopted policy objectives of foreign exchange oper-ations in many developing countries to which thebest practices advocated here are primarily intendedto apply.

Following the convention in the literature, the def-inition of intervention in this paper is narrowed to“sterilized” intervention that does not affect domes-tic monetary conditions (base money or short-terminterest rates). To the extent that a foreign exchangeoperation is not, or is only partially, sterilized, thenthe component that is left “unsterilized” is equivalentto a monetary policy operation.2

How Can Intervention Be Effective?

Exchange rates are supposed to reflect basic sup-ply and demand conditions, which in turn should belinked to underlying macroeconomic fundamentals.The literature provides favorable evidence on therelationship between exchange rates and fundamen-tals in the long term in economies with full capitalmobility (Sarno and Taylor, 2002). The parity condi-tions also hold in developing economies with partialcapital mobility (Tanner, 1998).

Exchange rates deviate substantially from valuesimplied by fundamentals in the short term, even inwell-functioning foreign exchange markets (Sarnoand Taylor, 2002). Exchange rate movements violate

II Best Practices in Official Interventions in the Foreign Exchange Market

Jorge Iván Canales-Kriljenko, Roberto Guimarães, and Cem Karacadag

2

1Disorderly market conditions are characterized by illiquidity inthe foreign exchange market, wide bid-offer spreads relative totranquil periods, and sudden changes in foreign exchange marketturnover and order flow.

2In practice, intervention in the foreign exchange market doesnot affect money supply on the day of intervention but rather onthe day of settlement. Intervention is only one of many factorsaffecting base money—including movements in bank reserves andgovernment financial operations—and its monetary effect is takenas a given when managing domestic liquidity. Intervention andmonetary policy decisions and operations are thus independent ofeach other.


How Can Intervention Be Effective?

the uncovered interest rate and purchasing powerparity conditions and appear to be excessivelyvolatile compared with underlying macroeconomicfundamentals (Mark, 2001). Moreover, macroeco-nomic models of exchange rate determination gener-ally fail to outperform a naive random-walk model inout-of-sample forecasting at short time horizons(Rogoff, 1999).

The disconnect between short-term exchange ratelevels and macroeconomic fundamentals creates arole for sterilized intervention. In particular, inter-vention may be used, possibly in conjunction withmonetary policy, to stabilize market expectations,calm disorderly markets, and limit unwarrantedexchange rate movements resulting from temporaryshocks. Intervention may also be used in conjunctionwith policies to redress macroeconomic imbalances,and it can complement efforts to place macroeco-nomic policies on a sustainable path by resisting dis-ruptive changes in the exchange rate, but only ifthere is a credible commitment to, and tangibleprogress on, macroeconomic adjustment.

Intervention is not an independent policy tool. Itseffectiveness is based on the consistency of targetedexchange rates with macroeconomic policies. More-over, with a high level of capital mobility, exchangerate and monetary policies cannot be conductedindependently. Intervention is especially unlikely tobe effective when adverse exchange movementsreflect persistent macroeconomic imbalances. Pro-tracted, one-sided intervention on a large scale prob-ably indicates that the current policy mix isunsustainable and that changes in exchange rate pol-icy or other macroeconomic policies are necessary.3

Large capital inflows or fragility in the financial sec-tor, for example, may require adjustments on severalfronts, including the exchange rate, interest rates,and fiscal policies.

Channels of Influence

Intervention can affect the exchange rate throughvarious channels. Under the signaling channel, mar-ket participants may adjust their exchange rate expec-tations when they perceive intervention as signaling achange in future monetary policy. Under the portfoliobalance channel, the change in the currency composi-tion of asset portfolios associated with sterilizedintervention generates a change in the risk premium,which triggers an exchange rate adjustment as agentsrebalance their portfolios. Under the microstructure

approach, dealers are the price setters and base theirpricing decisions in part on the order flow theyobserve, which is private information. Intervention—central bank–generated order flow—may thus affectdealers’ expectations and the exchange rate if thosedealers view it as informative.

Signaling Channel

Under the signaling channel, intervention can beeffective if it is perceived as a signal of the futurestance of monetary policy. In models that support thischannel, the exchange rate is treated as an asset price,and is a function of the expected path of money sup-ply. To the extent that intervention, even when steril-ized, influences market expectations on future moneysupply, it can influence the exchange rate. For exam-ple, the sale of U.S. dollars by a developing countrycentral bank will lead to a local currency apprecia-tion, not because the intervention changes the funda-mental supply and demand conditions in the market,but because it signals a contractionary monetary pol-icy (i.e., higher interest rates) in the future if down-ward exchange rate pressures persist.

A central bank has an incentive to follow throughwith policy actions that justify intervention ex post tosafeguard its credibility and avoid financial losses.For instance, the central bank may tighten monetarypolicies if the domestic currency remains underdownward pressure. The central bank puts its reputa-tion and capital at stake either because it wants tosignal a policy change that would not be credibleotherwise or because it believes, based on its infor-mation advantage, that the level and direction of theexchange rate are unwarranted. Intervention, then,would aim to change expectations in line with thecentral bank’s assessment.

The signaling channel depends in part on the insti-tutional and policy credibility of the central bank.4

The effectiveness of intervention through signalingrelies on influencing market expectations by trans-mitting information on fundamentals or future policyactions. Interventions must be perceived as crediblesignals (or threats) of future monetary policies toinfluence expectations. The signaling channel is mosteffective when interventions are publicly announced,which enhances the visibility of intervention, thusstrengthening the central bank’s policy signal.

The signaling channel, however, may be lesseffective in developing countries. First, central banks

3

3In fact, the “Principles of Fund Surveillance over ExchangeRate Policies” mandated by the IMF’s Articles of Agreementinclude large-scale intervention as one of the developments thatcan trigger a discussion with a member country about its exchangerate policy (IMF, 2002b).

4A lack of credibility may also increase the likelihood of spec-ulative attacks against the central bank (Sarno and Taylor, 2001).For instance, when the central bank intervenes to defend an unsus-tainable exchange rate (which may be called “adverse signaling”),market participants may engage in speculative trading against thecentral bank.


II BEST PRACTICES IN OFFICIAL INTERVENTIONS IN THE FOREIGN EXCHANGE MARKET

in many developing economies are at a disadvantagewith respect to institutional and policy credibility.They tend to lack the record of prudent macro-economic management that underpins the strongcredibility of monetary authorities in advancedeconomies. As such, the relative size of their inter-ventions, all else being equal, may have to be greaterin order to “buy credibility” and signal their com-mitment to future monetary policies (Mussa, 1981).Second, ongoing structural shifts in many develop-ing economies—among them financial deepening,economic opening, private sector orientation, andshifts in the exchange rate regime—make it difficultto establish predictable and stable links between realand financial variables and, therefore, between inter-vention and future monetary policies.

Portfolio Balance Channel

Intervention can be effective by altering the cur-rency composition of agents’ portfolios. The keyassumptions are that domestic and foreign currency–denominated government securities are imperfectsubstitutes and that market participants are riskaverse. As a result, investors demand a risk premiumon the bonds denominated in the riskier currency(which constitutes a violation of the uncovered inter-est parity condition).5 Through the portfolio balancechannel, a sterilized intervention operation alters therelative supply of domestic versus foreign currencysecurities, leading agents to rebalance their portfo-lios to equalize risk-adjusted returns, which in turncauses a change in the exchange rate. The exchangerate serves as the adjustment mechanism for risk-adjusted returns when base money and interest ratesremain unchanged following sterilized intervention.6

Unlike the signaling channel, the portfolio balancechannel does not require credibility as a preconditionfor effectiveness. As such, it potentially can be morepotent in some developing economies, where policycredibility tends to be lower, domestic currency debtis an imperfect substitute for foreign currency debt,

and interventions are large relative to foreignexchange market turnover.7

Microstructure Channel

The microstructure channel provides a new win-dow into the functioning of foreign exchange mar-kets and the effectiveness of intervention (Lyons,2001).8 Microstructure finance analyzes the impactof order flow on exchange rates. Aggregate orderflow is the balance of buyer-initiated and seller-initiated orders; as such, it is a measure of net buy-ing pressure in the foreign exchange market (Evansand Lyons, 2002, 2005). In this framework, analysesof the effectiveness of interventions focus on theextent to which central bank trades affect aggregateorder flow.

According to the microstructure approach, centralbanks are uniquely positioned to affect the transmis-sion of fundamentals to the exchange rate throughorder flow. Central bank intervention can cause mar-ket participants to change their expectations on thefuture path of the exchange rate and lead them tomodify their net open foreign exchange positions,triggering a change in aggregate order flow well inexcess of the central bank’s contribution. The impactof official intervention on order flow and exchangerates can be greater in the presence of noise traders,which follow past trends, and often trade in a corre-lated fashion (Hung, 1997). Central bank interven-tion, even in small amounts, can trigger a tide of buyor sell orders by trend-chasing traders. Interventionsneed not be announced and should be timed to max-imize the exchange rate impact. Intervention in thiscontext may also lead to higher volatility, which canhelp promote a sense of two-way risk in the market.

The impact of official intervention on marketexpectations can be even greater if the central banktrades aggregate or disseminate information (Popperand Montgomery, 2001). When central banks areperceived to be more knowledgeable about futuremonetary and exchange rate policies or betterequipped to monitor and interpret fundamentals,such as balance of payments trends, market partici-pants may try to learn from central bank trades.9 In

4

5The literature on violations of uncovered interest parity givesindirect support to the portfolio balance channel (Obstfeld, 1990;and Dominguez and Frankel, 1993a).

6Consider the case in which the U.S. Federal Reserve (theFed) targets a stronger U.S. dollar against the Japanese yen. TheFed would sell the Japanese currency by liquidating its yen-denominated bonds and simultaneously buy domestic dollar-denominated bonds, leaving domestic base money and interestrates unaffected. These operations would increase the relative sup-ply of foreign (yen) debt relative to domestic (dollar) debt held bythe market, thus requiring an adjustment in the expected rate ofreturn of yen-denominated debt to induce investors to hold moreof it. Since monetary conditions have not been affected in eithermarket, the expected return on yen debt (and the risk premium)can change only through an exchange rate adjustment.

7The degree of substitutability between domestic and foreigncurrency debt is inversely related to the size of the risk premium.The lower the degree of substitutability, the more effective inter-vention is, all else being constant.

8The market microstructure literature emphasizes the effects oforder flow, market participants, information asymmetries, tradingmechanisms, liquidity, and price discovery in the foreign exchangemarket (Lyons, 2001).

9Even if the central bank does not have an information advan-tage, its interventions may still be effective as long as market par-ticipants believe that it has inside information and intervenes onthe basis of that information.


Policy Issues

this context, central bank intervention emits infor-mation to the market. Stated differently, order flowserves as the vehicle through which the marketaggregates information. To the extent that centralbank–initiated order flow transmits information, itcan potentially ignite an even greater flow of foreignexchange orders.

The microstructure channel also emphasizes thatthe size of intervention relative to market turnoverdetermines the effectiveness of that intervention. Inprinciple, the larger the intervention relative to mar-ket turnover, the higher its effect on the exchangerate. Thus, intervention has the potential to be moreeffective in developing countries, where foreignexchange markets are less liquid.

Trends in Foreign ExchangeIntervention

Central banks in most advanced economies andsome emerging market economies rarely intervene.The central banks of the countries issuing interna-tional reserve currencies—including the U.S. Fed-eral Reserve and the European Central Bank(ECB)—seldom intervene anymore. The trendamong other advanced economies is similar.Although the Bank of Canada actively intervened formany years, it has not done so since 1998. TheReserve Bank of New Zealand has not intervenedsince 1985.

Some emerging market economies have followedsuit. The Bank of Israel has not intervened since1997, despite its strong presence in the market in theearly 1990s. The South African Reserve Bank, whichpursued an active intervention policy in the 1990s,particularly in the forward markets, now purchasesproceeds mainly from government external borrow-ing and privatization and intervenes to strengthen itsnet reserve position. The above-mentioned centralbanks generally enjoy a high degree of credibilityand have monetary policy frameworks that do notuse the exchange rate as an intermediate operatingtarget.

In sharp contrast, many developing economies stillintervene actively in the spot foreign exchange mar-ket (see Chapter 3). Country experiences with cur-rency crises in the 1990s, however, illustrate thelimits of intervention as a policy instrument. Mex-ico’s year-long defense of its crawling peg in 1994ended suddenly when the market belatedly observedthe central bank’s depleted reserves position. Thai-land’s defense of the baht in the first half of 1997through interventions failed, virtually depleting thecentral bank’s net international reserves. Brazil’sdefense of the crawling band through spot and for-ward market intervention could no longer be sus-

tained against strong market pressures in late 1998and early 1999.

The effectiveness of intervention in the long run isthus primarily a function of the sustainability ofexchange rate and underlying macroeconomic poli-cies. However, despite their interdependence, it isimportant to distinguish between intervention andexchange rate policy. Currency crises shed light onthe sustainability of exchange rate policies but notnecessarily on the effectiveness of intervention perse. To the extent that intervention fails because ofunsustainable exchange rate policies, the source offailure lies not in intervention as an instrument itselfbut in the policy mix underpinning the targetedexchange rate.

The key lessons on intervention from the literatureand selected country experience are as follows:• Intervention is not an independent policy tool. It

cannot generate permanent changes in exchangerates when intervention objectives are inconsistentwith macroeconomic policies.

• Intervention may be used to address unwarrantedexchange rate movements stemming from tempo-rary shocks. Intervention can also complementefforts to place macroeconomic policies on a sus-tainable path by resisting disruptive changes in theexchange rate, but only if there is a credible com-mitment to, and tangible progress on, macroeco-nomic adjustment.

• Institutional and policy credibility is an importantdeterminant of the effectiveness of intervention.Credibility may enhance the effectiveness of inter-vention and even obviate the need for it.

• Central banks in most advanced economies andsome emerging market economies rarely inter-vene anymore, despite their strong institutionaland policy credibility. This reflects the limitedeffectiveness of intervention in deep and efficientforeign exchange markets, where market failuresare rare.In sum, intervention can play a role in stabilizing

exchange rates, provided that the rates are consistentwith underlying macroeconomic policies. Govern-ments should make efforts to build policy credibilityand develop liquid foreign exchange markets, whichcan help minimize instances of misalignment anddisorderly markets, and the need for intervention.

Policy Issues

The conduct of foreign exchange intervention, inpractice, requires the development of a comprehen-sive set of policies and guidelines on a wide range ofoperational issues at the policy, technical, andadministrative levels. At the policy level, decisionsmust be made on the objectives of intervention, the

5



criteria for determining the amount and timing ofintervention, and the degree of transparency.

Objectives

Central bank policies should define interventionobjectives in precise terms. Defining these objec-tives, in turn, involves two sequential steps: specify-ing the exchange rate measures to be used andselecting an exchange rate target, if any. Exchangerate measures may include the nominal bilateral,nominal effective, real bilateral, or real effectiveexchange rates. The exchange rate target may be apoint-in-time level or a range within which the cen-tral bank keeps the exchange rate. Even under inde-pendently floating exchange rate regimes, the centralbank may cap the rate of change it is prepared to per-mit over a certain period, such as one day. Ensuringprecision in intervention objectives is critical to suc-cessfully executing an intervention and to assessingits effectiveness ex post. In practice, intervention hasbeen pursued to (i) correct misalignment or stabilizethe exchange rate at predetermined levels or withintargeted rates of change, (ii) calm disorderly mar-kets, including exchange rate volatility and marketilliquidity, (iii) accumulate reserves, and (iv) supplyforeign exchange to the market.

Correcting exchange rate misalignments is a highpriority for central banks. Real exchange rate over-valuation can undermine export competitiveness andweaken a country’s external liquidity position, whilean undervalued exchange rate may create inflation-ary pressures.

Even without misalignment, sharp exchange ratemovements may be costly. They can raise the cost ofexternal trade and dampen trade flows, particularlyin economies where hedging opportunities are lim-ited.10 Moreover, where credibility is lacking and theexchange rate anchors inflationary expectations, theexchange rate is considered a symbolic and visiblemeasure of a government’s success in macroeco-nomic management; hence, its stability assumesparamount importance. The stronger pass-throughfrom exchange rate fluctuations to inflation in devel-oping economies makes them less tolerant ofexchange rate instability compared with advancedeconomies (Calvo and Reinhart, 2002). Intervention

thus can smooth volatility and avoid destabilizinginflationary expectations. Moreover, the balancesheet exposure of financial and nonfinancial enter-prises to exchange rate risk arising from foreign cur-rency debts adds to the costs of volatility.

Disorderly markets involve a collapse of liquidity,where market intermediaries face difficulties match-ing suppliers with end users of foreign exchange. Ifmarket illiquidity persists, it can have seriousadverse effects on the real economy and therefore isusually not neglected for extended periods. Tolerat-ing occasional episodes of illiquidity, however, isnecessary to allow markets to self-correct and toexpose market participants to the risks inherent intrading financial assets.

A number of indicators can detect market illiquid-ity, including an acceleration in the pace of exchangerate changes, unwarranted increases in exchange ratevolatility, a widening of bid-offer spreads, and sharpchanges in the level and composition of turnover.However, these indicators can result from changes ineconomic fundamentals or the arrival of new infor-mation, which may not warrant intervention. Possi-ble ways of interpreting trends in the foreignexchange market are discussed in Chapter 3.

Intervention aimed at addressing exchange ratelevels and disorderly markets should be constantlyassessed in light of the nature of the shocks to theeconomy, the macroeconomic policy mix, thedegrees of capital mobility and dollarization, andavailable international reserves. For example, resist-ing exchange rate depreciation out of fear of mis-alignment may not be feasible in the face of sharpcapital outflows when international reserves are low.The same applies when a bank run occurs in a coun-try where the central bank plays the role of lender oflast resort. In general, policymakers should con-stantly assess the feasibility of the objectives giventhe macroeconomic policy framework and the typeof shock that has led them to consider intervening.

Reserve accumulation can be a high priority, espe-cially in the aftermath of a currency crisis. It is par-ticularly prevalent in developing countries trying tobuild investor confidence and strengthen their debtrepayment capacity and external liquidity position.11

Most IMF-supported programs include targets on netinternational reserves to prevent them from beingdepleted in defense of an unsustainable exchangerate and to bolster the government’s creditworthinessthrough higher reserve coverage of imports andshort-term debt. Strengthening repayment capacityhelps countries reduce their dependence on bilateral

6

10However, these concerns may be unwarranted. Dominguezand Frankel (1993b) cite numerous studies that find the effect ofexchange rate volatility on trade to be small or nonexistent. Moregenerally, the research surveyed in Rogoff (1999) indicates thatthe exchange rate regime and exchange rate volatility do not havedetectable effects on output and trade. More recent studies on thesmall and open economies of Ireland and New Zealand also findthat exchange rate volatility has little or no impact on trade andinvestment (Bjorksten and Brook, 2002; and Bredin, Fountas, andMurphy, 2002).

11Strong external liquidity can help offset weak country funda-mentals and reduce vulnerability to external shocks. Moreover, acountry’s external liquidity position is a key determinant of itssovereign creditworthiness (Mulder and Perrelli, 2001).


Policy Issues

and multilateral funding by regaining access to inter-national capital markets.

Central banks in some developing countries inter-vene mainly to supply foreign exchange to the market.This is especially true when the central bank is theforeign exchange agent of the public sector, and state-owned enterprises (SOEs) are major exporters account-ing for a high share of the country’s foreign currencyearnings, as in oil-producing countries. In principle,SOEs are like any other corporate customer and candeal directly in the foreign exchange market throughcommercial banks. In practice, however, SOEs oftenlack the skilled personnel in treasury operations toeffectively manage their foreign currency assets andliabilities, including the timing of their foreignexchange purchases and sales. This task is usuallyperformed by the central bank, which has experiencein dealing with the foreign exchange market. The cen-tralization of the public sector’s foreign exchangeoperations in the central bank has the added benefit ofavoiding lumpy sales and purchases of foreignexchange by SOEs, which can disrupt the market.12

Amount and Timing

There is no simple rule for determining theamount and timing of intervention. It is a highly sub-jective exercise based on several factors, includingthe nature and duration of shocks, observable marketindicators, market intelligence, and availablereserves. However, some general principles can pro-vide guidance for decisions by the authorities.

Rules Versus Discretion

Central banks need to have some degree of discre-tion in determining when and in what amount tointervene. Discretion allows the central bank toaccommodate market conditions and gives it roomfor tactical maneuvering, although policy rules canserve as rules of thumb. Discretion is critical for sev-eral reasons:• A long-term commitment to an intervention policy

rule is not necessary when a commitment alreadyexists to another nominal anchor, such as in aninflation-targeting framework.

• Market participants may take advantage of thecentral bank when its operations are bound bystrict rules. Even if the authorities do not announcethe policy rule, market participants can often dis-cern it and speculate against the central bank, pos-sibly generating losses for it.

• The optimal intervention rules discussed in the lit-erature have limited application for practical pur-poses.13 Given an estimated model for exchangerate determination, an optimal policy rule thatyields the amount of intervention at a particularmoment in time can be estimated. However, thereis no consensus on the validity of these models andthe assumptions on which they are based.

• Qualitative information obtained from marketintelligence can help the authorities decide theamount and timing of intervention, but its interpre-tation requires considerable judgment.Rules-based intervention may be appropriate for a

short period under certain circumstances. Brazil’srules-based intervention policy, which ex ante lim-ited the central bank’s sales of foreign exchange to$50 million a day between July and December 2001,was effective in filling the estimated balance of pay-ments gap arising from a sharp reduction in capitalinflows, without giving the impression that the cen-tral bank was targeting the exchange rate. Similarly,in Turkey, foreign exchange sale and purchase auc-tions, whose timing and amount are determined andannounced ex ante, have been an effective and trans-parent mechanism for reinforcing the central bank’scommitment to a floating exchange rate regime.

Over time, however, many central banks that haveexperimented with rules-based policies have aban-doned or modified the rules to allow for some discretion. For example, Canada’s mechanical inter-vention policy of the 1990s was modified in 1995and abandoned in 1998 to provide the central bankwith greater discretion and to reduce the frequencyof intervention (Murray, Zelmer, and McManus,1996).14 Similarly, Brazil’s rules-based interventionpolicy, which was revived in mid-2002, was subse-quently relaxed to give the central bank more discre-tion over how, when, and by how much it couldintervene in the spot market as it responded to chang-ing market conditions.

Market Indicators and Intelligence

The authorities should monitor a combination ofmarket indicators and intelligence and use a varietyof analytical methods and economic models beforemaking intervention decisions. Criteria for interpret-ing observable market indicators and their implica-tions for the timing of intervention are discussedlater in this chapter.

7

12Centralizing the public sector’s foreign exchange operationsin the central bank has some drawbacks, however. It can give riseto conflicts of interest between the central bank’s policy goals andits fiduciary responsibilities to those entities it serves as an agent.

13See Roper and Turnovsky (1980) and Jones (1984).14A study by Beattie and Fillion (1999) on Canada also found

that rules-based intervention did not reduce exchange rate volatil-ity, though the existence of a nonintervention band provided asmall stabilizing influence.



Market intelligence—including information onthe underlying sources of foreign exchange demandand supply, large customer transactions, order flow,and the market’s view of the balance of paymentsoutlook—is a critical complement to observablemarket indicators. Market intelligence gathered frommarket participants enables monetary authorities toobtain information on critical but unobservable (on areal-time basis) determinants of exchange rates, suchas order flow. Traders are in constant contact withcustomers and, therefore, can offer key insights onlarge commercial transactions, the sources andunderlying reasons for customer orders, and marketreaction to a shock. Market intelligence can be gath-ered through daily telephone surveys, informal con-tact with traders, or a combination of the two.

Considerations on Amount

The amount of foreign exchange interventionshould not be determined from a policy rule. Centralbanks often determine effective amounts through trialand error. Intervention affects order flow, the risk pre-mium, and expectations, all of which interact withone another and can impact the exchange rate (Figure2.1). The intervention amount, in principle, should belarge enough to achieve the targeted exchange rateand usually must be a multiple of the typical marketorder. The intervention amount depends, in part, onits effect on exchange rate expectations. A change inexpectations can cause market participants to modifytheir net open foreign exchange positions and createorder flow in favor of the targeted exchange rate, low-ering the amount needed for intervention.

The cumulative amount of intervention in defenseof the exchange rate should be reviewed whenever it

approaches a predetermined benchmark over a givenperiod. Moreover, the central bank should avoid one-sided intervention on a continuing basis. It is vital toadjust policies to resolve the underlying causes ofimbalances in order flow. In this context, interventioncan provide an early warning indicator that the policymix is unsustainable. This is one of the rationales forsetting a floor on reserves under IMF-supportedprograms (Mussa and Savastano, 1999).

The stock of available reserves constrains thecumulative amount of intervention to defend thecurrency. Moreover, in some developing economies,the scarcity of reserves is a major constraint even inthe short run. When under currency depreciationpressures, the central bank faces a potential suddenloss of reserves, particularly when capital flows arelarge, which can undermine investor confidenceeven further.

Central banks generally should refrain from bor-rowing to finance intervention when net internationalreserves are low. Externally financed interventionunder such circumstances creates unacceptably highrisks for the central bank. For example, adjustmentefforts to reduce macroimbalances may be delayed iffinancing operations are opaque. Once access tofinancing dries up, both expectations and reserves arelikely to deteriorate sharply. This, in turn, can give riseto positive feedback trading and to an overshooting ofthe exchange rate. Moreover, large-scale externallyfinanced intervention exposes the central bank ornational authorities to exchange rate and rollover risks.

The scope for conducting large-scale official inter-vention is greater for foreign exchange purchasesthan for foreign exchange sales. Foreign exchangepurchases can always be financed by printing money,which may or may not be sterilized, depending on

8

Intervention Amount

ExchangeRate Impact

Expectations

Order Flow

Risk Premiums

Figure 2.1. Intervention and Exchange Rate Impact


Policy Issues

the monetary framework. However, printing moneycan either affect the inflation objectives of theauthorities (if unsterilized) or generate substantialinterest rate costs that would weaken the centralbank’s balance sheet (if sterilized).15

Considerations on Timing

Determining the timing of intervention is alsohighly judgmental and subjective. It involves ananalysis of observable market indicators and avail-able market intelligence, against the background ofthe central bank’s unique experiences and country-specific circumstances. Economic models and policyrules can be part of a decision-making process thatnevertheless requires considerable judgment and issubject to large margins of error. The timing of inter-vention ultimately depends on the central bank’sassessment of the presence of exchange rate mis-alignment and disorderly markets.

Exchange rate misalignment. Despite a volumi-nous amount of literature on exchange rate misalign-ment, there is no consensus on a methodology tocompute the equilibrium exchange rate. Readilyavailable indicators that can uncover signs ofexchange rate misalignment include the nominal andreal effective exchange rates, productivity and othercompetitiveness indicators, the terms of trade, thebalance of payments, and interest rate differentials.However, these indicators often do not allow policy-makers to identify the presence or magnitude of mis-alignment precisely enough to justify intervention.

Analyzing fundamental trends is vital to identify-ing the nature of shocks to the economy, which maybe important in determining the timing of interven-tion. Permanent shocks to domestic monetary condi-tions or the terms of trade, for example, would beexpected to generate a change in expectations and anadjustment in the exchange rate. A sharp change inthe exchange rate initiated by a permanent shockneed not be resisted by intervention, unless themovement threatens to trigger positive feedbacktrading. Then smoothing may be warranted. By con-trast, temporary shocks to the economy that do notsignificantly affect macroeconomic fundamentalsmay call for intervention if the shock causes unwar-ranted fluctuations in the exchange rate.

Disorderly market conditions. Detecting disor-derly markets is equally challenging. An accelerationin exchange rate changes is a prime symptom ofmarket illiquidity, where predominantly one-waycustomer orders become difficult to match, reducing

the number of transactions and ultimately leading toa complete market standstill. Rapid price movementscan also occur in a liquid market but still be a matterof concern because of their potential to create posi-tive feedback trading, which may cause the exchangerate to overshoot and set the stage for a sharp pricereversal. Under this scenario, it is the potentiallyself-fulfilling dynamics of price changes and theirpotential to create destabilizing shifts among multi-ple exchange rate equilibria that may necessitatecentral bank intervention.

Widening bid-offer spreads signal heighteneduncertainty in the level of the exchange rate, whichin turn may diminish market liquidity. Dealers typi-cally raise bid-offer spreads to protect againstexchange rate volatility and unexpected order flow.The high level of price uncertainty may keep marketparticipants from transacting until the direction ofthe exchange rate becomes clearer, possibly throughcentral bank intervention. Some central banks attachgreater importance to widening bid-offer spreadsthan to abrupt exchange rate changes, because theprice uncertainty implied by large spreads may be agreater threat to market liquidity. Larger bid-offerspreads may indicate that dealers are finding foreignexchange intermediation more risky or costly, a sit-uation that may lead them to withdraw from supply-ing liquidity to the market. Although wideningspreads are more likely to be associated with dimin-ishing market liquidity, their absence does not nec-essarily imply that markets are liquid. Whereone-sided trading produces sharp, unidirectional,and continuous changes in the exchange rate, themarket anomaly would be reflected in sharp, levelchanges while spreads remain low. Bid-offerspreads thus must always be analyzed in conjunc-tion with other indicators.

The composition and magnitude of turnover in theforeign exchange market also provide importantclues on liquidity and trading dynamics. In particu-lar, a rise in interbank trades relative to customer-bank turnover may indicate that dealers are facingdifficulties matching customer-initiated foreignexchange orders with final counterparties. This risemay also be symptomatic of “hot potato” trading,where interbank activity increases as dealers try topass on or hedge the exchange rate risk associatedwith a previous trade, with few or no participantswilling to remain exposed.16 Trading may once againcollapse if orders become entirely one-sided. In thiscase, central banks should intervene to prevent a col-lapse of liquidity in the foreign exchange market.

9

15High interest rates can be particularly damaging if the stock ofpublic sector debt is large and its service is affected significantlyby variations in nominal interest rates.

16Hot potato trading may be efficient when dealers share the riskborne out by trading with the end users of foreign exchange. Inthis case, excessive turnover need not be symptomatic of an inef-ficient market.



Exchange rate volatility is another vital indicatorof market conditions.17 However, it is not necessarilya source of concern unless other indicators such aswidening bid-offer spreads suggest disorderly mar-ket conditions. Volatility often reflects, among otherthings, uncertainty in economic policies and otherfundamental determinants of exchange rates, whichthe market may be struggling to price accurately. Tothe extent that price discovery and volatility occur inan orderly (liquid) market, central bank interventionwould be unwarranted. Moreover, tolerance for somedegree of volatility is essential. Insulating the econ-omy from volatility would deprive market partici-pants of learning to cope with volatility and manageexchange rate risks. The moral hazard problem asso-ciated with the perception of an implicit exchangerate guarantee, in turn, may lead to liability dollar-ization and balance sheet vulnerability. Therefore,central banks must strike a careful balance betweenexchange stability and volatility, confining potentialinterventions to only extreme degrees of price move-ments. By contrast, if volatility reflects wideningbid-offer spreads, heightens the risk of positive feed-back trading, or threatens price stability, then thecentral bank may have reason to act.

Distinguishing disorderly markets from normalmarket dynamics and setting predetermined triggerpoints for intervention is extremely difficult. The lackof consensus on determinants of exchange rates athigh frequencies heightens the challenge of ascertain-ing market conditions. Trends in volatility, spreads,and turnover can be interpreted only in the context ofevents and shocks that may be driving them. Forexample, volatility that amounts to a market distur-bance in one market can be typical behavior inanother market. As such, it is difficult to predeterminewhen volatility and spreads are excessive andturnover is inadequate with enough precision and cer-tainty to warrant intervention. The challenge is com-pounded by the constantly changing nature of shocksto the economy and of market dynamics, includingthe growing diversity of market participants.Nonetheless, the central bank should set benchmarksfor various market indicators to enhance its capacityto respond quickly to developments, when needed.18

To the extent that some flexibility exists within theday (or possibly week) in terms of timing, inter-vention should occur in a liquid market. When themarket is illiquid, intervention may have a large

price impact, but it can also disrupt trading condi-tions. Dominguez (2003) notes that interventionsduring heavy trading volume and closely timed toscheduled macroeconomic announcements are themost likely to have large and long-lasting effects onthe exchange rate.

Accumulating Reserves and Supplying Foreign Exchange

Unlike exchange rate–related objectives, accumu-lating reserves and supplying foreign exchange to themarket are much easier to define and measure. Thechallenge is to minimize the price impact of interven-tion, while accumulating reserves and supplyingforeign exchange. Periodic foreign exchange auctionsare one way the central bank can minimize interven-tion’s price impact, although such auctions, wherepossible, should not be automated according to a pre-determined schedule to avoid gaming by the market.

Market conditions should be taken into accountwhen intervening to accumulate reserves or supplyforeign exchange. For example, the supply of foreignexchange should be timed during downward pres-sure on the currency, while interventions to accumu-late reserves should coincide with an appreciatingcurrency, where possible.

Degree of Transparency

Transparency in intervention policies and objec-tives can enhance the credibility of the central bankby holding it accountable for its record of policyimplementation. When central bank credibility islacking, transparency can be instrumental in buildingreputation, provided there is a credible commitmentto prudent macroeconomic policies and central bankindependence.

The degree of transparency in intervention opera-tions may vary with the specific objectives of inter-vention. The optimal level of transparency alsodepends on whether the central bank undertakesintervention with the aim of affecting the exchangerate. For example, the central bank may prefer tointervene secretly while adjusting its own portfolioor replenishing reserves. Alternatively, the centralbank may not want to jeopardize its credibility overthe exchange rate that subsequently fails to hold(Enoch, 1998). By contrast, if the central bank hasstrong credibility and its intervention objective is toinfluence the exchange rate, its presence in the mar-ket should be revealed so that it can benefit from thesignaling effect.

Central banks usually have considerable discretionin choosing the degree of transparency of interven-tion. Nevertheless, regardless of whether the centralbank announces its presence in the market, it should

10

17There are several measures of exchange rate volatility. Widelyused measures include implied volatility from option prices andtime-series measures such as the generalized autoregressive condi-tional heteroscedasticity (GARCH) model (Jorion, 1996). Theanalysis below does not rely on any particular measure of volatility.

18Benchmarks on the level of exchange rate need to take intoaccount the degree of volatility in normal times.


Policy Issues

not reveal its trading tactics. Tactical ambiguity withrespect to the exact timing and amount of interven-tion will heighten prospects for achieving the inter-vention objective most efficiently, that is, with theminimum amount of intervention possible. Reveal-ing its trading tactics may create opportunities formarket participants to reap riskless profits (i.e., themarket could profit by anticipating and internalizingthe central bank’s actions in its own trading tactics).In some circumstances, the central bank can disclosecertain aspects of its interventions (e.g., whether itwill buy or sell), but it still must ensure that the infor-mation divulged does not enable market participantsto game it.

There are other reasons why the central bank maywish to intervene secretly. Central banks sometimesmay want to calm disorderly markets caused by herdbehavior by introducing a sense of two-way risk inthe market. In this instance, the central bank is notcommitted to a particular exchange rate level andmay not want to put its reputation at risk. It also maywant to retain an element of surprise, especially whenthe scale of intervention is modest relative to the dailyturnover in the market (Sarno and Taylor, 2001).

Similarly, central banks may aim to forestall a cur-rency from breaking a resistance level (i.e., the levelat which stop-loss trading models may trigger sell-ing) (Dominguez and Frankel, 1993a). In the samevein, where the exchange rate is managed within aband, the central bank may intervene intramarginallyto preempt speculative pressures from emergingonce the rate approaches the edge of the band(Enoch, 1998).

The extent to which tactical ambiguity is suc-cessfully maintained depends on the timing of disclo-sure, the choice of counterparties, and the structureand liquidity of the foreign exchange market. In termsof timing, official intervention may become public invarious ways, including by making policy announce-ments, revealing the central bank’s presence duringthe operations, selecting the most visible trading plat-forms and counterparties, confirming or denying expost press reports, and reporting statistical informa-tion on exchange rates, with a lag.

The central bank may conceal its identity beforestriking a deal to avoid discriminatory treatment andto secure the best possible rate for its trade. Thisanonymity can be achieved by trading in electronicbroking systems or through an agent, although thismethod may be difficult in other trading platforms.For example, as soon as the central bank asks for aquote, its identity is revealed, although routineinquiries can weaken the perceived link between itsinquiries and intervention. Intervening through voicebrokers raises other challenges. Concealing centralbank trades is much harder when the bulk of foreignexchange transactions is concentrated in a few mar-

ket participants. Even if successfully maintained,central bank anonymity can have adverse sideeffects. For example, market makers may widen thebid-offer spread when they attach some probabilitythat they may be trading with the central bank(which is an adverse selection effect of dealing withan informed agent).

The greater the size of intervention relative to mar-ket turnover, the more difficult it is to keep interven-tion secret. This is even more notable in manydeveloping countries, where foreign exchange inter-vention is large relative to turnover. In addition, inseveral developing countries, the banking system ishighly concentrated, allowing the main participants toinfer the central bank’s presence. Market participantsmay also get some indication of the central bank’sparticipation in the market through the evolution ofnet international reserves, particularly in countriesthat disclose their reserve data in a timely manner.

The transparency of intervention practices variesacross countries. The central banks in the UnitedStates, European Union, and Japan have enhancedthe transparency of their interventions. Since themid-1990s, interventions by the U.S. FederalReserve have been reported on a quarterly basis andhighly publicized in the financial press.19 The ECBannounces some of its interventions, although theinformation contained in the announcements is lim-ited and does not include amounts and timing. TheBank of Japan does not announce its interventionsbut reports the amounts and exchange rates ex post(Ramaswamy and Samiei, 2000; and Ito, 2002).

In emerging markets, central bank practices onsecrecy are mixed (Canales-Kriljenko, 2003). Abouthalf of the central banks in these economies thatresponded to a survey on intervention practicesannounce their presence in the market. About 25 per-cent of survey respondents indicated that they publishdata on their interventions, in some cases with a lag.

Market Microstructure

Central banks may organize the foreign exchangemarket—by defining core elements, including par-ticipants, information flows, and trading mecha-nisms (Canales-Kriljenko, 2003, 2004)—in a waythat enhances the effectiveness of intervention. Byrestricting participation to certain types of institu-tions, central bank regulations may temper specula-tive activities, albeit at the expense of narrowing therange of hedging options in the market.

11

19The U.S. Federal Reserve reports its intervention activity onits website quarterly and releases daily intervention figures with aone-year lag. Hung (1997) estimates that around 40 percent of theFederal Reserve’s foreign exchange interventions from 1985–89were not announced.



Central banks can also create an informationadvantage for themselves by, for example, selectingtrading mechanisms. Most central banks establishextensive reporting requirements on banks’ foreignexchange operations and net open positions andmanage the disclosure of information obtained fromthem. A few also allow only foreign exchange trad-ing mechanisms in which they have privilegedaccess to information from all transactions and canparticipate in the market like any other authorizeddealer. In a few countries where foreign exchangereceipts are surrendered to the central bank, central-ized auctions are conducted periodically for pricediscovery. In these cases, the central bank has insideinformation because it performs the role of auction-eer, but it can simultaneously participate in the auc-tion. Several central banks also have privilegedaccess to information from trading in selected elec-tronic broking and dealing systems.

Some countries use a variety of foreign exchange,monetary, and banking regulations that constrict thesize of the market, potentially increasing the relativesize of an intervention in it. For example, a few coun-tries explicitly ban interbank trading, while othersallow banks to trade with each other only on behalfof their customers. This way, countries are able tolimit competition in the process of price discoveryand reduce the scope for interbank trading. Othercountries achieve a similar outcome by performingthe role of market makers with narrow bid-offerspreads that undercut competition from authorizeddealers and transform the central bank into the prin-cipal foreign exchange intermediary (Canales-Kriljenko, 2004).

While some of these regulations can make inter-vention more effective, they can also cause the cen-tral bank to intervene more often than it otherwisewould. The central bank would have to shoulder theburden of smoothing discrepancies in foreignexchange orders, since other market participantswould not have an incentive to conduct stabilizingspeculation. This could increase exchange ratevolatility.

Intervention Operations

Intervention operations involve a number of tech-nical issues, including the choice of markets andcounterparties.

Choice of Markets

In choosing a market for intervention, the authori-ties must decide on the intervention instrument (ortype of foreign exchange contract), meaning eitherspot, forward, or other derivatives markets; the trad-

ing location (onshore or offshore); the currency inwhich to intervene; and whether to intervene in thewholesale transfer market or retail cash market.

Spot, Forward, or Other Derivatives Markets?

Intervention generally should take place in thespot market rather than in the forward market whenthe goal is to affect the spot exchange rate.20 The rea-sons are as follows:• Spot market intervention directly affects the spot

exchange rate. Forward market intervention relieson the transmission mechanism from forward tospot market rates, which is affected by moneymarket conditions as well as exchange and capitalcontrols.

• Spot market intervention is less susceptible toliquidity risk. Spot markets are usually more liquidand less constrained by counterparty limits thanare forward markets. Moreover, with increaseduncertainty, forward market trading may dry upfaster than spot trading owing to higher credit risksat longer maturities.In general, intervention financed through deriva-

tive instruments is not advisable because of the fol-lowing concerns:21

• Financing spot market intervention through swapsand other derivative contracts can lead to highlyleveraged net open foreign exchange positions,exposing the central bank to exchange rate androllover risks.

• Conducting intervention through derivatives,which enables the central bank to conceal inter-vention from the public at large, may delay neces-sary policy adjustments. Protracted intervention onone side of the market, in turn, can result in sub-stantial reserve losses without achieving interven-tion objectives.

• The use of derivative instruments obscures theinterpretation of the central bank’s balance sheeteven for the authorities. The complicated nature ofderivative contracts, especially option contracts,makes it difficult to exert internal control on for-

12

20The IMF’s 2001 survey on foreign exchange market organiza-tion found that 78 percent of the responding countries conductedintervention in the spot market, while only 4 percent reportedintervening in the forward market. Neely (2001) reports a similarfigure for spot market intervention but finds that about 50 percentof the respondents also intervened in forward markets. For a briefanalysis of Thailand’s experience with forward market interven-tion, see Moreno (1997) and IMF (1998). The South African expe-rience has also been documented in several South African ReserveBank Quarterly Bulletins (1997 and 1998) available via the Inter-net at http://www.resbank.co.za.

21For some of the risks associated with intervention in deriva-tives markets, see the Hannoun report (Bank for International Set-tlements, 1994).


http://www.resbank.co.za

Intervention Operations

eign exchange operations, which can elevate thecentral bank’s risk exposure, especially in coun-tries with weak governance.Nevertheless, under certain conditions, intervention

through derivatives markets may play a role.22 Forexample, derivatives markets may be an appropriatesource of financing intervention if the central bank:• understands the risks associated with derivative

positions and sets adequate systems for monitor-ing, controlling, and managing these risks;23

• avoids conducting protracted intervention on oneside of the market and building a persistent shortnet open foreign exchange position; and

• creates the right incentives for changing the policymix, including disclosing periodically its positionsin derivatives.At the operational level, it is preferable to sterilize

spot market intervention using swaps rather thanother derivative instruments such as forwards,futures, or options. The central bank can trade duringthe day in the spot market, tailoring its interventionaccording to market reactions, and then adjust itsreserve position with a swap at the end of the day.Futures contracts are subject to margin calls that candisrupt the authorities’ cash flows. Intervention inoption markets is an attractive alternative but needsto be approached with care (Box 2.1).

Trading Location: Onshore or Offshore Markets?

Intervention should normally take place inonshore markets where the bulk of foreign exchangetrading takes place. Concentrating interventions inthe domestic market helps maintain the primacy ofthe domestic market and may give the central bankgreater access to market information and intelli-gence.24 In addition, central banks can effectivelyaddress the order flow imbalance created offshore byintervening onshore.

In some cases, intervention in offshore currencymarkets may play a useful role. In particular, the cen-tral bank may intervene in offshore currency marketswhen the local currency trades offshore beyond thenormal working hours in the onshore market,exchange rate pressures emerge in offshore markets,and secret intervention is preferred and easier to con-duct offshore.

Several operational issues must be addressedwhen intervening in offshore markets. The centralbank may have to appoint an agent to act on itsbehalf, which can be a foreign central bank, the Bankfor International Settlements, or a foreign or domes-tic commercial bank. The central bank or its agentwould have to abide by the rules and regulations ofthe markets in which intervention takes place(including the trading protocols).

Intervention Currency

The principal intervention currency should be theinternational currency most widely traded against thedomestic currency to reduce costs and facilitate set-tlement in countries following flexible exchange rateregimes.25 Intervening in the most widely traded cur-rency pair makes it easier for the central bank to findcounterparties and reduces transaction costs as mea-sured by bid-offer spreads. In addition, settlementfacilities are usually more reliable.

For most developing economies, the interventioncurrency is the U.S. dollar, because foreign exchangetrading is concentrated in the dollar (Canales-Kriljenko, 2004).26 The U.S. dollar may be the mainintervention currency even in countries whose tradeis concentrated in the euro area and in countrieswhose currencies are pegged to a currency basket.Some major developing economies have publiclyannounced that they have included the euro as anintervention currency (see Reserve Bank of India,2002), but the U.S. dollar most likely remains theprincipal intervention currency.

The central bank should intervene in one foreigncurrency at a time to avoid the risk of cross-currencyfluctuations. Intervening in many currencies compli-cates foreign exchange operations and exposes thecentral bank to exchange rate risk. For example,rapid changes in cross-currency exchange rates canforce the central bank to realign its intervention

13

22Derivative instruments can be used to sterilize the monetaryeffects of intervention when monetary policy instruments arescarce (for example, when the money market is not well developedor the stock of government securities is small). The central bankmay also use swaps and other derivatives to manage domestic liq-uidity. Since foreign exchange swaps typically involve spot trans-actions and simultaneous reverse-forward transactions, the use offoreign exchange swaps for liquidity management can in principlegive rise to a large foreign exchange forward book, even if theswap operations do not have an exchange rate objective (Hooy-man, 1994; and Bartolini, 2002).

23The central bank needs to adopt tools to manage its exposureto market risks. For example, the authorities should, inter alia,open a new trading book, establish a method of measuring foreignexchange exposure, conduct regular stress tests, and frequentlyestimate value at risk.

24This distinction may be immaterial, since offshore currencytrading seldom takes place without the ultimate participation ofdomestic dealers and domestic banks.

25For countries with exchange rate regimes pegged to a singleforeign currency, however, the intervention currency should be thepeg currency. According to the International Monetary Fund’sAnnual Report on Exchange Arrangements and Exchange Restric-tions (IMF, 2002a), this rule is adopted by all countries withpegged exchange rate regimes that have disclosed their interven-tion currency.

26The main intervention currency for the ECB and the Bank ofJapan also is the U.S. dollar.



rates, and minor delays can provide arbitrage oppor-tunities to speculators at the expense of the centralbank. Intervening in only one currency at a time doesnot affect the counterparties that may need other for-eign currencies, because major international curren-cies can be converted at low cost in worldwidemarkets. Operationally, the central bank shouldannounce a limited number of currencies in which itwill conduct intervention, and that it will intervene inonly one currency at a time.27

Wholesale Transfers or Retail Cash Markets?

In the absence of exchange controls, central banksshould intervene only in the wholesale market fortransfers.28 Wholesale transfers reduce transactioncosts by economies of scale and by avoiding trans-portation and warehousing costs.29 In the presence ofexchange controls, however, central bank interven-

14

In theory, intervention through options may be effec-tive in reducing exchange rate volatility. Experiencewith this type of intervention, however, has been lim-ited and not well documented in the literature.

Option market intervention can have an immediateeffect on the exchange rate and reduce exchange ratevolatility, but it exposes the central bank to significantlosses (Breuer, 1999). For example, to hedge its posi-tion arising from a central bank sale of foreignexchange call options, a market maker can borrow for-eign exchange up to a fraction of the notional amount,sell it on the spot market, and invest the domestic cur-rency in the money market.1 Besides the immediateeffect on the spot market, the buyer of the call optionhedges its option position over time in a way thatreduces exchange rate volatility. In particular, the buyersells foreign exchange when the domestic currencydepreciates and buys it when the domestic currencyappreciates. In contrast, when the central bank sells for-eign exchange put options, the market maker buys for-eign exchange at the outset, but its hedging behaviorover time still reduces exchange rate volatility.2 As the

seller of the option, the central bank is exposed to sig-nificant exchange rate losses because it is contractuallyobliged to sell at below-market rates when the option is“in the money.”

Other Types of Option Contracts: Mexico and Colombia

Mexico auctioned U.S. dollar put options betweenAugust 1996 and June 2001, while Colombia set up asystem for the auction of U.S. dollar call and putoptions in 1999.3 The objective of the put option salesin Mexico and one type of option contract in Colombiawas to accumulate international reserves while mini-mizing the disruption in the spot market. Colombia alsoissued other types of options contracts to smoothexchange rate volatility. To stem downward pressureson the local currency, spot market intervention was con-ducted in Mexico during the period, while in Colombia,intervention was limited to the option market.

The Mexican and Colombian option contracts werenot standard. For example, the strike price of theoptions used to accumulate reserves was not fixed butcorresponded to the interbank rate of the day earlier andput (call) options could be exercised only if theexchange rate depreciated (appreciated) more than the20-day exchange rate moving average. Werner andMilo (1998) developed a model to price the Mexicanoption contract, and Mandeng (2003) suggestedimprovements to the Colombian ones. Werner (1997)found that the option intervention strategy in Mexico,as intended, did not have a significant effect onexchange and interest rates.

Box 2.1. Intervention Through Options

1A standard call option gives the buyer the right, but not theobligation, to buy foreign exchange. Similarly, a standard putoption gives the buyer the right, but not the obligation, to sellforeign exchange. American-style options can be exercised atany time before maturity, while European-style ones can beexercised only at maturity.

2Breuer (1999) argued that to reduce exchange rate volatil-ity, the central bank needs to sell (rather than buy) options. Hisargument assumed that market makers take a net long optionposition when they buy from the central bank. In hedging longoption positions over time, market makers reduce exchangerate volatility. In contrast, they increase volatility when theyhedge short option positions. Zapatero and Reverter (2003)developed and calibrated a theoretical model that comparedspot market with option market intervention and concludedthat the latter could result in lower exchange rate volatilities at

a lower reserve cost. This result, however, critically dependson the interaction between interest and exchange rates.

3The Colombian experience has been documented by Man-deng (2003). For Mexico, see Galán, González de Castillo,and Tames (1997); Werner (1997); Werner and Milo (1998);and Carstens and Werner (1999).

28Some electronic dealing systems distinguish among four prod-ucts (wholesale transfer, wholesale cash, retail transfer, and retailcash) that could have four different prices. However, transfers usu-ally trade in the wholesale market, and cash usually trades in theretail market.

29Differences in exchange rates between the wholesale transferand the retail cash market reflect mainly differences in conve-nience and cost.

27For example, the U.S. Federal Reserve is authorized to intervenein the nine currencies listed in the document “Authorization for For-eign Currency Operations,” which is reproduced in the annual reportof the Board of Governors of the Federal Reserve System.


Conclusions

tion in the retail cash market may rein in unwar-ranted depreciation expectations. The retail cashmarket, which is often linked to the parallel marketrate, can become the center for price discovery andthe formation of exchange rate expectations, even ifit is a small fraction of trading and highly volatile.Therefore, intervention in cash markets, includingparallel markets, can sometimes help prevent unwar-ranted shifts in exchange rate expectations.

Choice of Counterparties

The central bank should establish objective andtransparent criteria for choosing counterparties forintervention.30 For example, the central bank shouldtrade only with financial institutions that are sol-vent31 and that provide information on market devel-opments and conditions to the central bank and offerit competitive two-way exchange rates. More gener-ally, the central bank should trade mainly with mar-ket makers, but it may find it useful to extend therange of its counterparties when exchange controlsare present or there is little competition.

In competitive environments without foreignexchange controls, the central bank should trademainly with market makers.32 First, the central bankcan promote the development of a fledgling inter-bank foreign exchange market when it trades onlywith market makers that provide liquidity to the mar-ket by offering two-way (buying and selling)exchange rates on demand. Second, market makerscan efficiently distribute the foreign exchange pro-vided by the central bank by standing ready to tradewith other authorized dealers. Third, market makersare usually able to handle large trading volumes,avoiding the need for the central bank to conductmany foreign exchange operations.33 Fourth, thecentral bank minimizes the chances of dealing with

lesser-quality counterparties. Finally, trading withmarket makers that routinely interact with other mar-ket participants can provide greater control to thecentral bank on the degree of transparency of its for-eign exchange operations (Box 2.2).

When foreign exchange controls are present, thecentral bank may elect to trade directly with foreignexchange bureaus and in the parallel market.Although trading with bureaus may be a small frac-tion of total trading, a sharp depreciation in thebureau market can affect exchange rate expectationsand move exchange rates in the interbank market.34

However, the benefit of reducing spreads in the par-allel market through intervention should be balancedagainst the cost of dealing with riskier counterpar-ties. The central bank should make special settlementarrangements to protect itself against credit risk.

When there is little competition, the central bankmay consider directly trading with the public at largeto increase competition in the foreign exchange mar-ket. When the market is thin and a few authorizeddealers account for the bulk of trading and are notwilling to offer two-way quotes, the central bank’sdirect participation can intensify competition.

Central banks may want to deal directly with thepublic also when an unwarranted bid-offer spread isemerging in the retail market owing to insufficientcompetition. Some countries have experimentedwith authorized dealers acting on behalf of the cen-tral bank for conducting retail operations. In return,these banks receive a commission for transferringthe funds to end users.

Conclusions

This chapter has provided information to helpguide central banks in developing countries on howto design and carry out interventions in foreignexchange markets. Its primary goal has been toaddress the day-to-day operational issues faced bycentral banks in their interventions under flexibleexchange rate regimes.

The main themes of the chapter and the best prac-tices it has advocated here are as follows.• Intervention is not an independent policy tool. Its

success is based on the consistency of interventionobjectives with macroeconomic policies. It isunlikely to be effective, particularly over the long

15

30Some countries disclose the criteria. For details, see SverigesRiksbank (1999) and European Central Bank (2000).

31Trading with solvent institutions reduces settlement risks,while allowing the central bank to control the transparency ofintervention. Although the central bank can depart from standardmarket practice and insist on being paid first to avoid credit risk, itwould have to reveal its identity beforehand, which is not alwaysdesirable.

32In a few developing economies, the law establishes the per-missible central bank counterparties. Some central banks appointthe counterparties for their foreign exchange operations through aformal agreement (e.g., Sveriges Riksbank and the FederalReserve Bank of New York). Other central banks do not appointcounterparties but trade only with authorized dealers that belongto a market maker’s club. Under this arrangement, the private sec-tor indirectly chooses the central bank counterparties.

33In general, market makers are better positioned than the cen-tral bank to interact with the public at large, given their ongoingcustomer relations with providers and users of foreign exchange.They can perform retail foreign exchange operations at a lowercost than can the central bank.

34When foreign exchange controls restrict access to foreignexchange for some legally permitted international transactions, aspread of more than 2 percent in the parallel or bureau foreignexchange markets will give rise to a multiple-currency practice. Toavoid this possibility, some countries have directly forced, throughregulation, the bureau exchange rate to be set in relation to the offi-cial rate.



term, when adverse exchange movements reflectpersistent macroeconomic imbalances and deterio-rating investor confidence in the currency. In par-ticular, protracted, one-sided interventions shouldbe avoided.

• Intervention may be used, however, in conjunctionwith policies to redress macroeconomic imbal-ances. Intervention can complement efforts toplace the macroeconomic policies on a sustainablepath by resisting disruptive changes in theexchange rate, but only if there is a credible com-mitment to and tangible progress on macroeco-nomic adjustment.

• Central banks should be rigorous in their analysisof market conditions, judicious in their decisionsto use scarce foreign exchange reserves, andparsimonious in their interventions. Exchangerate misalignments and disorderly markets areextremely difficult to detect. Therefore, interven-tions to smooth volatility and correct for mis-alignment should not be used as cover fortargeting a particular exchange rate level under aflexible exchange rate regime. Infrequent recourseto intervention for exchange rate purposes, more-over, enhances its effectiveness by maximizingthe element of surprise.

• Determining the timing and amount of interven-tion is highly subjective. The two depend heavilyon the nature and permanence of economic shocksand exchange rate pressures, other observable mar-ket indicators, market intelligence, economic fore-casts, and available reserves. Policy rules andeconomic models can provide guidance and rulesof thumb for intervention, but they usually havelimited practical application and require a heavydose of judgment. Hence, some degree of discre-tion is often necessary. Discretion also allows thecentral bank to accommodate market conditionsand gives it room for tactical maneuvering. Rules-based intervention policies may be appropriateunder certain circumstances, including for accu-mulating reserves and supplying foreign exchangeto the market.

• Intervention policies and objectives should betransparent and clearly specified. Transparency inintervention objectives can enhance the credibilityof the central bank by holding it accountable for itsrecord of policy implementation, even though thedegree of transparency on the tactical implementa-tion of intervention policies may vary with the spe-cific objectives of intervention. Ensuring precisionin intervention objectives is critical to successfully

16

The degree of transparency in central bank interven-tions varies depending on whether the bank approachesthe final counterparties directly, through brokers, orthrough agents acting on its behalf. The central bankcan better control the transparency of its foreignexchange operations when it deals with selected marketmakers or when they act on its behalf.

Trading Directly with Market Makers

The central bank can impose a confidentiality require-ment on these institutions as a condition for dealing withthem. They will have little incentive to reveal theirclient’s presence for fear of driving the price againstthemselves before unloading the positions taken on fromthe central bank (Enoch, 1998). After having unloadedtheir position, they have the incentive not to disclose thepresence of the central bank if they want to keep dealingwith it on a regular basis. Moreover, the central bank canask its market-maker counterparties to act on its behalf toplace limit orders through brokers (in amounts that canbe considered normal for the turnover in the market) toincrease the secrecy of the transaction. The disclosure ofthe market maker’s identity once the deal is sealed willnot reveal the presence of the central bank. Nevertheless,if only one market maker takes a contrarian’s position ina one-sided market, market participants may presumethat the central bank is behind these operations.

Trading Indirectly Through Brokers or Agents

It may be more difficult to control the transparencyof the central bank’s presence in the market when itdeals with counterparties other than market makers.For example, the central bank will automatically revealits presence in the market when it trades with manydealers on a bilateral basis, since it will have to dis-close its identity to get or give a quote. The sameapplies when the central bank directly approachescounterparties through voice brokers. Brokers can ben-efit from disclosing the presence of the central bank inthe market by attracting more business from marketparticipants that value this type of confidential infor-mation. Confidentiality agreements with brokers aredifficult to enforce, because it is hard to prove whetherbrokers or the final counterparties disclosed the infor-mation. While brokers do not disclose the identity ofthe counterparties posting the best rates before the dealis struck, the identity must be disclosed for settlementto the final counterparties. Trades with state-ownedcommercial banks can be interpreted as interventionon behalf of the central bank. Finally, counterpartydiversification may obscure the degree of transparencyabout the size of intervention, but it is likely toincrease the market’s awareness of the presence of thecentral bank in the market because more individualsare involved.

Box 2.2. Choice of Counterparties and Transparency


Conclusions

executing an intervention and assessing its effec-tiveness ex post. Intervention objectives, more-over, should be reassessed constantly in light ofthe shocks faced by the economy, the macroeco-nomic policy mix, and available reserves.

• Central banks generally should refrain from fund-ing intervention by borrowing, when net inter-national reserves are low. Foreign currency debt–financed intervention—through either on-balance-sheet or off-balance-sheet operations—createsrisks that may not be justified by the uncertain andoften limited effectiveness of intervention, particu-larly when reserves and credibility are low andmarket access is limited.

• Interventions should normally be executed in themost liquid markets where price discovery occurs.

In particular, intervention for exchange rate pur-poses should normally occur in the spot market,where it is easier to find counterparties and whereintervention directly affects the exchange rate.Intervention should also normally be conductedonshore and in the currency most widely traded toreduce costs and facilitate settlement, but it may onoccasion be conducted offshore to prevent destabi-lizing trading activity from affecting domesticmarkets.In closing, it should be noted that the applicability

of the best practices put forth here ultimatelydepends on country-specific circumstances, includ-ing central bank credibility, the structure and depthof the foreign exchange market, and the macroeco-nomic and political environment.

17


Acomprehensive survey conducted by the Interna-tional Monetary Fund in 2001 systematically

documented the structure and main characteristics ofthe foreign exchange market in developing countries.The 90 countries classified as “developing” or “intransition” that responded to the survey accounted in2000 for 85 percent of exports, 91 percent of imports,and 85 percent of GDP of all the world’s developingeconomies.1 They also held about 90 percent of thedeveloping economies’ international reserves.

Based on the survey results, this chapter discussesthe prevalence of foreign exchange intervention indeveloping countries across different exchange rateregimes and degrees of market access; the degree ofsterilization practices; the size of foreign exchangeintervention relative to the foreign exchange market;and central banks’ information advantage.

Prevalence of Foreign ExchangeIntervention

The central banks in developed countries thatissue the major international currencies are notactive participants in the foreign exchange market.Their monetary policies target short-term interestrates, while foreign exchange intervention is limitedto calm disorderly market conditions. Foreignexchange intervention, in general, does not takeplace frequently. Although it can be large in absolutesize, it is estimated to be small relative to total for-eign exchange market turnover. In recent years, onlythe Bank of Japan was active in the foreign exchangemarket and intervened to any sizable degree. By con-trast, neither the U.S. Federal Reserve (the Fed) northe European Central Bank (ECB) conducted officialforeign exchange intervention on its own behalf.2

Most central banks in developing economies inter-vened in the foreign exchange market across allexchange rate regimes and degrees of market accessduring 2001 (Table 3.1). Most of the foreignexchange intervention took place in spot foreignexchange markets through foreign exchange transac-tions arranged by telephone conversations withbanks as main counterparties (Table 3.2).

Foreign exchange intervention is prevalent evenunder flexible exchange rate regimes.3 For example,in a managed floating exchange rate regime, themonetary authority influences the movements of theexchange rate through interest rate changes andactive foreign exchange intervention, without speci-fying (or precommitting to) a preannounced path forthe exchange rate. In an independently floatingexchange rate regime, foreign exchange interventionmay be conducted to moderate the rate of change andprevent undue fluctuations in the exchange rate,rather than establish a level for it.

Sterilized or Not Sterilized?

When the central banks that issue the major inter-national currencies intervene, they tend to sterilizeforeign exchange interventions to achieve theirshort-term operating targets of monetary policy,usually short-term interest rates. The Fed sterilizesits foreign exchange intervention to keep the amountof bank reserves at levels that are consistent withestablished monetary policy goals. In particular,liquidity is adjusted for consistency with the federalfunds target. The ECB has sterilized its foreign

III Survey of Foreign ExchangeIntervention in Developing Countries

Jorge Iván Canales-Kriljenko

18

1The survey was sent to all 160 IMF member countries classifiedas “developing” or “in transition.” For more detail on the surveyresults, see Canales-Kriljenko (2003). See also Cheung and Chinn(1999) and Neely (2001).

2The ECB drew from a foreign exchange swap arrangementsigned with the Federal Reserve Bank of New York in the after-math of the September 11, 2001, terrorist attacks, which alleviated

liquidity demands that could have otherwise generated large pres-sures on the exchange rate. However, it cannot be considered for-eign exchange intervention because the purpose of the operationwas not to affect exchange rates but to smooth potential disrup-tions in the payment systems. Moreover, although a foreignexchange swap theoretically involves simultaneous spot and for-ward foreign exchange operations, market makers hedge theirexposure by operations in the money market rather than in the for-eign exchange market.

3See Ishii and others (2003) for a discussion of exchange rateregime classification based on de facto policies.


Sterilized or Not Sterilized?

exchange intervention on the few occasions that ithas been in the market (Frenkel, Pierdzioch, andStadtmann, 2001).4 The Bank of Japan conducts for-eign exchange intervention as the agent of the min-istry of finance, with funds from a special accountof the Japanese government. Thus, foreign exchangeintervention does not affect the money base. Foreignexchange purchases are funded by issuing short-term, yen-denominated bills, and yen purchases arefunded by selling foreign exchange funds from thespecial account in the market (Bank of Japan, 2000;and Ito, 2002).

Unlike the Fed, the ECB, and the Bank of Japan,most central banks in developing economies do not

fully sterilize their foreign exchange interventions ona regular basis (Table 3.3). About 10 percent of thesurvey respondents reported that foreign exchangeintervention is never sterilized; about half indicatedthat it is sometimes sterilized; and about 20 percentsaid it is always sterilized. About 25 percent of sur-vey respondents did not answer the correspondingsection of the survey.

The frequency of sterilization varied slightly byexchange rate regime and market access. Countriesthat sometimes sterilize their foreign exchange inter-ventions can be found in almost all types ofexchange rate regimes. The countries that never ster-ilize are concentrated in the less flexible exchangerate regimes, as would be expected, but account foronly a small share of all countries following theseregimes. The countries that always sterilize their for-eign exchange interventions are more likely to befound in the more flexible exchange rate regimeswith market access, but they can also be found in softpeg exchange rate regimes.

19

Table 3.1. Foreign Exchange Intervention by Exchange Rate Regimeand Market Access, 2001(Percentage of countries answering the corresponding survey question in each category)1

Developing and Transition Economies________________________

Exchange Rate Regime2 Emerging markets3 Other Total

No country-specific currency — 100 100CAEMC4 — 100 100Other — — —

Country-specific currency 88 94 91Currency board 50 100 67Conventional fixed pegs against a single currency 60 88 77Conventional fixed pegs against a composite 100 100 100Pegs with horizontal bands within a cooperative arrangement — — —Pegs with horizontal bands within an IMF-supported program — 100 100Crawling pegs 100 100 100Exchange rates within crawling bands 80 100 83Managed floating, no preannounced path for exchange rate 100 91 96Independently floating 91 100 94

Total 88 94 91

Memorandum item:Number of countries answering question 41 35 76In percentage of survey respondents 93 76 84

Source: Survey on the organization of foreign exchange markets conducted in 2001 by the IMF.Note: — stands for not applicable, zero, or negligible amount.1Refers to the percentage of the 90 countries classified as “developing” and “in transition” that responded to

the IMF 2001 survey on the foreign exchange market.2Follows the IMF’s de facto exchange rate regime classification as published in the IMF’s International Financial

Statistics.3As so defined in the IMF’s quarterly publication Emerging Market Financing: A Quarterly Report on Developments

and Prospects.4The Central African Economic and Monetary Community (CAEMC) is a conventional fixed peg arrangement.

4The intervention operations conducted during 2001 by theECB, and not considered in Frenkel, Pierdzioch, and Stadtmann(2001), were also sterilized. In particular, the foreign exchangeintervention conducted on behalf of the Bank of Japan did notaffect the money base.


III SURVEY OF FOREIGN EXCHANGE INTERVENTION IN DEVELOPING COUNTRIES

Relative Size of ForeignExchange Intervention

Foreign exchange intervention by the centralbanks that issue the major international currenciesaccounts for a small fraction of annual foreignexchange market turnover. Even in the case of theBank of Japan, foreign exchange intervention againstU.S. dollars in 2000 accounted for less than 0.2 per-cent of estimated annual foreign exchange marketturnover. However, the size of foreign exchangeintervention on any given day may be substantial,reaching a peak of 16 percent of foreign exchangemarket turnover.5

Foreign exchange intervention by some centralbanks in developing economies accounts for a muchlarger portion of foreign exchange market turnover

than that conducted by the central banks that issuethe major international currencies, especially at thelevel of interbank trading (Table 3.4).6 In six of thecountries that responded to the survey, the size offoreign exchange intervention exceeds the volume ofinterbank foreign exchange market turnover (exclud-ing trades with the central bank). In contrast, the sizeof intervention is below 10 percent of the volume ofinterbank trading in four countries. The size of for-eign exchange intervention is usually significantlysmaller as a fraction of bank-customer trading,reflecting the fact that interbank trading in develop-ing economies usually accounts for only a smallamount of turnover in the bank-customer segment ofthe market.7

Several reasons may help explain the relativelylarge size of foreign exchange intervention by some

20

Table 3.2. Characteristics of Foreign Exchange Intervention in Developing and TransitionEconomies, 2001(Percentage of countries answering the corresponding survey question in each category)1

Market Access_____________________Exchange Rate Regime2

Emerging _____________________________________markets3 Other Pegged Intermediate Flexible Total

Foreign exchange intervention conducted

In the spot market 82 74 61 83 89 78In the forward market 2 7 6 0 4 4

Main counterpartsBanks 100 91 94 100 96 96Government 84 85 82 100 82 84Exporters and importers 5 4 3 0 7 4

Trading platformsTelephone orders 66 52 64 33 62 59Online trading systems

Reuters 2000-1 36 28 33 25 33 32Electronic broking system 18 9 12 8 16 13

Source: Survey on the organization of foreign exchange markets conducted in 2001 by the IMF.1Refers to the percentage of the 90 countries classified as “developing” and “in transition” that responded to the IMF’s 2001 survey on the foreign

exchange market.2Follows the IMF’s de facto exchange rate regime classification as published in the IMF’s International Financial Statistics.The pegged regimes are those

without a country-specific currency, currency boards, and conventional fixed peg arrangements. Intermediate regimes have pegged exchange rates withinhorizontal bands, crawling pegs, and exchange rates within crawling bands. Flexible regimes have managed and independently floating exchange rateregimes.

3As so defined in the IMF’s quarterly publication Emerging Market Financing: A Quarterly Report on Developments and Prospects.

5This calculation considers only total turnover in the spot mar-ket between yen and U.S. dollars, as documented by the TriennialCentral Bank Survey of the Foreign Exchange and DerivativesMarket Activity of the Bank for International Settlements (2001).About 95 percent of foreign exchange intervention by the Bank ofJapan is conducted against U.S. dollars.

6Only a few survey respondents provided information about theamounts of foreign exchange intervention and foreign exchangemarket turnover that allows for the comparison.

7In contrast, interbank trading accounts for most of the for-eign exchange market turnover among the major internationalcurrencies.


Relative Size of Foreign Exchange Intervention

central banks in developing economies. First, thesecentral banks are usually large customers in the for-eign exchange market. Second, they use a variety offoreign exchange, monetary, and banking regulationsto increase their relative size in the foreign exchangemarket, among other objectives. Third, central bankforeign exchange operating practices may preventthe development of an interbank foreign exchangemarket containing the growth of the turnover in theforeign exchange market.

Central Banks as Large Customers in the Foreign Exchange Market

Many central banks in developing economies areimportant players in the foreign exchange marketson their own behalf or on behalf of their govern-ments. These central banks may buy or sell foreignexchange as customers on their own behalf for sev-

eral reasons. Foreign exchange can be used to meettheir foreign expenditures, such as paying their ownexternal debt, or to sell the foreign exchangereceived from loans to support the balance of pay-ments, including those from multilateral lendinginstitutions. These central banks can also enter theforeign exchange market to adjust the actual level ofinternational reserves to the desired level, for exam-ple, to meet some reserve adequacy targets. In addi-tion, they often buy and sell foreign exchange todefend the level of the exchange rate or to reduceexchange rate volatility.

Many central banks in developing economies alsoconduct foreign exchange operations on behalf oftheir governments, state enterprises, and nonbud-getary government agencies. More than half of thesurvey respondents reported that the central bank isthe exclusive foreign exchange agent of the govern-ment, with the government trading foreign exchange

21

Table 3.3. Developing and Transition Economies That Do Not AlwaysSterilize Foreign Exchange Intervention, by Exchange Rate Regime andMarket Access, 2001(Percentage of countries answering the corresponding survey question in each category)1

Developing and Transition Economies________________________

Exchange Rate Regime2 Emerging markets3 Other Total

No country-specific currency — — —CAEMC4 — — —Other — — —

Country-specific currency 68 83 75Currency board — — —Conventional fixed pegs against a single currency 60 83 73Conventional fixed pegs against a composite — 67 83Pegs with horizontal bands within a cooperative arrangement — — —Pegs with horizontal bands within an IMF-supported program — — —Crawling pegs — — —Exchange rates within crawling bands 60 50 57Managed floating, no preannounced path for exchange rate 69 91 79Independently floating 60 83 69

Total 68 84 75

Memorandum item:Number of countries answering question 38 31 69In percent of survey respondents 86 67 77

Source: Survey on the organization of foreign exchange markets conducted in 2001 by the IMF.Note: — stands for not applicable, zero, or negligible amount.1Refers to the percentage of the 90 countries classified as “developing” and “in transition” that responded to

the IMF 2001 survey on the foreign exchange market.2Follows the IMF’s de facto exchange rate regime classification as published in the IMF’s International Financial

Statistics.3As so defined in the IMF’s quarterly publication Emerging Market Financing: A Quarterly Report on Developments

and Prospects.4The Central African Economic and Monetary Community (CAEMC) is a conventional fixed peg arrangement.



only with the central bank.8 State-owned enterprisesand nonbudgetary government agencies in manydeveloping and transition economies are alsorequired to trade foreign exchange exclusively withthe central bank. The survey found that about 8 per-cent of respondents had this exclusive arrangementwith state-owned enterprises and about 15 percenthad it with government agencies.

The governments and their agencies in developingeconomies are often an important source of foreignexchange, especially where the size of the govern-

ment in the economy is large. In particular, the gov-ernment, state enterprises, and nonbudgetary govern-ment agencies account for a large portion of foreignexchange traded in many countries. Concentrationarises where financial aid from foreign donors is themain source of foreign exchange. It also occurs incountries where state enterprises account for the bulkof the export receipts and in some open economieswhere foreign exchange traded domestically arisesmainly from taxes and royalties paid in foreignexchange. Finally, the government often becomes alarge supplier of foreign exchange in countrieswhere the fiscal deficit is financed abroad.

Moreover, many central banks in developingeconomies conduct foreign exchange operationswith government entities to achieve exchange ratepolicy objectives. On several occasions, govern-ments and state-owned companies have borrowed

22

Table 3.4. Magnitude of Foreign Exchange Intervention in SelectedDeveloping and Transition Economies, 2000(Percentage of foreign exchange market turnover at different levels of trading)

Foreign Exchange Intervention as a Percentage of:________________________________________________________________Foreign exchange

Foreign exchange market turnoverInterbank market turnover among banks and

foreign exchange between bank and between bank andCountries1 market turnover2 end customers3 end customers2

Country 1 5,153.4 0.1 0.1Country 2 1,351.7 239.0 203.1Country 3 161.8 — —Country 4 160.2 3.1 3.0Country 5 138.4 25.7 21.7Country 6 118.1 — —Country 7 90.2 9.6 8.7Country 8 59.3 — —Country 9 36.5 7.3 6.1Country 10 32.4 15.9 10.7Country 11 4.6 3.9 2.1Country 12 3.0 19.0 2.6Country 13 1.0 — —Country 14 0.0 0.0 0.0Country 15 — 7.4 —Country 16 — 68.1 —Country 17 — 5.5 —

Bank of Japan3 0.2 0.9 0.1

Sources: Survey on the organization of foreign exchange markets conducted in 2001 by the IMF; Bank for Inter-national Settlements (2001), Triennial Central Bank Survey of Foreign Exchange and Derivatives Market Activity; and Bankof Japan.

1The names of the countries are omitted for purposes of confidentiality.2The different levels of foreign exchange market turnover exclude transactions with the central bank.3Foreign exchange intervention conducted in U.S. dollars in 2000 versus spot market turnover of the yen against

the U.S. dollar on a yearly basis, as published in Table E-2 of the 2001 triennial survey statistical annex of the Bankfor International Settlements.To compute the figure on a yearly basis, 22 trading days are assumed each month.

8In some countries, the government is not allowed to hold foreigncurrency deposits and must surrender its foreign exchange to thecentral bank. In others, the government may keep foreign exchangedeposits in or out of the central bank, but when it decides toexchange them, it has to conduct the exchange through the centralbank.


Relative Size of Foreign Exchange Intervention

23

abroad with the main purpose of affecting the evolu-tion of the exchange rate rather than financing fiscalexpenditures or the companies’ operations. As docu-mented by Taylor (1982), this form of secret foreignexchange intervention was also practiced in somedeveloped countries in the late 1970s.

Regulations

Many central banks in developing economies useforeign exchange controls and monetary regulationsthat increase the size of intervention relative to theforeign exchange market. In contrast, none of themeasures described in this section are currently usedby the central banks that issue the major interna-tional currencies.

Foreign Exchange Controls

Foreign exchange controls increase the size of for-eign exchange intervention relative to the market byeither reducing the size of the foreign exchange mar-ket or concentrating the foreign exchange supply inthe hands of the central bank.

Capital controls can reduce cross-border move-ments of capital and the volume of foreign exchangemarket turnover, increasing the relative size of cen-tral bank foreign exchange intervention. Banningcross-border investments is a way of discouragingnonresidents from using the domestic currency andresidents from using foreign currencies, thus reduc-ing the potential volume of transactions in the for-eign exchange market. Capital controls can preventlarge capital movements and significant increases inforeign exchange market turnover that accompanydeviations from interest rate parity not explained bydifferences in risk premiums.

Surrender requirements to the central bank areobligations to sell foreign exchange proceeds (usu-ally from exports) within a specified time frame.When directed to the central bank, surrender require-ments concentrate the foreign exchange supply inthe hands of the central bank, turn the central bankinto the main foreign exchange intermediary, andincrease the relative size of foreign exchange inter-vention. In practice, surrender requirements exist inabout 40 percent of the survey respondents, but theyare seldom directed to the central bank.9

Prohibitions on interbank foreign exchange trad-ing limit the size of the foreign exchange market,increasing the relative size of foreign exchange inter-

vention. In a few developing economies, banks areallowed to conduct foreign exchange trading only onbehalf of their customers. Banks can still conductforeign exchange intermediation, buying fromsources of foreign exchange and selling to end usersof foreign exchange. But they cannot engage in mar-ket making activities.10

Many central banks in developing economies estab-lish a combination of measures to discourage financialinstitutions from taking net open foreign exchangepositions.11 The measures include limits on the leveland daily variations of the financial institutions’ netopen foreign exchange positions. The positions subjectto limits usually include open forward foreignexchange positions, since unhedged forward foreignexchange positions can trigger large pressures on thespot exchange rates when banks need to hedge theirexposure and cannot find an adequate counterparty totake the opposite forward foreign exchange position.

About half of the survey respondents have in placemeasures that restrict the operation of forward mar-kets, reducing the ability of nonfinancial institutionsto fund speculative positions and hedging exchangerate risk. In particular, about 15 percent of surveyrespondents explicitly prohibit banks from issuingforward contracts, and about 40 percent impose cer-tain requirements on banks that offer forward con-tracts, most notably the requirement that banks offerthese contracts only for hedging the exchange raterisk of legally permitted, underlying, internationaltransactions. Foreign exchange regulations in somecountries also control the maturity of the forwardcontracts offered to customers. About 45 percent ofsurvey respondents allow banks to issue forwardcontracts without any controls. However, the scopefor speculative net open position taking is limited bythe level of development, liquidity, and depth of themarket. In particular, only 9 percent of surveyrespondents consider their forward foreign exchangemarkets to be developed, liquid, and deep, while 30percent consider them to be undeveloped, illiquid,and shallow (Table 3.5).12

Monetary Regulations

Monetary regulations can increase the relative sizeof central bank foreign exchange intervention by

9Surrender requirements to the government may be motivatedby a desire to allocate foreign exchange to particular uses, makemore foreign exchange available to the central bank for foreignexchange intervention in periods of pressure on the exchange rate,or meet public foreign exchange expenditure commitments.

10The central bank thus reduces competition from the market insetting exchange rates, which increases the impact of foreignexchange intervention on the exchange rate.

11Net open foreign exchange positions also play an importantprudential role by limiting banks’ exposure to exchange rate risk.

12In contrast, about 40 percent of the survey respondentsperceived their spot foreign exchange markets to be developed,liquid, and deep, while only 6 percent perceived them to beundeveloped, illiquid, and shallow.



reducing the residents’ use of foreign currencies andnonresidents’ use of the domestic currency.

To reduce the scope for currency substitution,most countries that issue their own currencies havegranted a series of legal privileges to their domesticcurrency. Residents usually must use their domesticcurrency as means of payment. In particular, mone-tary regulations in many of the survey respondentsgive the domestic currency the exclusive role of

means of payment (forced tender) or, at least, theadvantage of legal tender so that it must be acceptedin payment for financial obligations. Moreover,about half of the survey respondents explicitly pro-hibit their residents from making payments to otherresidents in foreign currencies.

Most countries allow residents to use foreign cur-rencies as a store of value. Nearly all survey respon-dents allowed banks to accept foreign currency

24

Table 3.5. Selected Regulations on Forward Foreign Exchange Transactions in Developing andTransition Economies, 2001(Percentage of countries answering the corresponding survey question in each category)1

Market Access___________________Exchange Rate Regime2

Emerging __________________________________markets3 Other Pegged Intermediate Flexible Total

Forward markets allowed 89 63 70 58 84 76Forward markets not allowed 5 24 18 8 13 14Not able to determine 7 13 12 33 2 10

Types of derivative contracts allowedOutright forward contracts 89 63 70 58 84 76Nondeliverable forward contracts 59 28 33 42 51 43Futures 61 30 39 42 51 46Options 77 30 45 42 62 53

Requirements for offering forward contractsQuantitative limits 11 20 18 17 13 16Verification of existence of legally

permitted underlying current or capital transactions 27 33 39 17 27 30

Transaction made only on behalf of their customers 5 11 15 0 4 8

Freely 66 24 30 33 58 44Not able to determine 2 0 0 0 2 1

Subjective assessment of forward marketsDeveloped 34 7 21 8 22 20Undeveloped 48 52 42 67 51 50Other 11 2 6 0 9 7Not able to determine 7 39 30 25 18 23

Liquid 27 11 21 8 20 19Illiquid 43 35 39 42 38 39Other 18 0 6 8 11 9Not able to determine 11 54 33 42 31 33

Deep 18 7 9 8 16 12Shallow 55 37 45 50 44 46Other 14 2 9 0 9 8Not able to determine 14 54 36 42 31 34

Developed, liquid, and deep 14 4 6 0 13 9

Undeveloped, illiquid, and shallow 32 30 27 42 31 31

Source: Survey on the organization of foreign exchange markets conducted in 2001 by the IMF.1Refers to the percentage of the 90 countries classified as “developing” and “in transition” that responded to the IMF 2001 survey on the foreign

exchange market.2Follows the IMF’s de facto exchange rate regime classification as published in the IMF’s International Financial Statistics.3As so defined in the IMF’s quarterly publication Emerging Market Financing: A Quarterly Report on Developments and Prospects.


Information Advantage

deposits, especially from exporters.13 Some develop-ing countries explicitly prohibit other private sectorresidents from holding foreign currency deposits indomestic banks. Banks may also accept foreign cur-rency deposits from the public sector, especially fromstate enterprises. The number of countries allowingtheir financial systems to offer foreign currencydeposits to nonresidents is smaller. The degree of dol-larization of private sector deposits is above 10 per-cent in about half of the survey respondents, reachingthe 75 to 100 percent range in a few countries.14

In addition, about one-third of developing econo-mies impose controls on the use of their domesticcurrencies by nonresidents abroad. To avoid fuelingspeculation in foreign exchange markets, manycountries—about 30 percent of survey respondents—explicitly prohibit their banking systems fromextending short-term loans in domestic currency tononresidents.

Information Advantage

To infer exchange rate pressures embedded in for-eign exchange market activity, the literature on themicrostructure of such markets emphasizes theimportance of foreign exchange order flow. A posi-tive foreign exchange order flow reflects an excessdemand for foreign exchange that would tend todepreciate the domestic currency. Lyons (2001) sur-veys the literature that has empirically documentedthe positive relationship between order flow andexchange rate changes, and Vitale (2001) puts for-ward a theoretical argument in the context of foreignexchange intervention.

Some central banks in developing economiesmake use of their ability to issue regulations toobtain an information advantage over other marketparticipants, requesting, among other things,specifics about foreign exchange order flow. Theyrequire market participants to submit informationabout their foreign exchange activities, sometimes ingreat detail.15 The information advantage arisesbecause only a subset of the information collected ismade available to other foreign exchange market

participants. The data requested vary significantlyacross countries, ranging from all information oneach of the foreign exchange transactions made byeach authorized dealer to summary statistics, some-times weighted by the size of the transactions. Thecollected information available to central banksoften includes data for every licensed dealer onexchange rates (whose dispersion reflects the uncer-tainty in the foreign exchange market) and foreignexchange transaction volumes.

From the information on foreign exchange transac-tions, central banks can infer the size of foreignexchange order flow aggregated at some levels oftrading.16 For example, in transactions between banksand their customers, market turnover usually equalsaggregate foreign exchange order flow because cus-tomers are usually those initiating the foreignexchange transaction at the exchange rate quoted bydealer banks, especially in competitive foreignexchange markets where market makers operate. How-ever, in transactions among banks, foreign exchangemarket turnover usually differs from foreign exchangeorder flow. The lack of order flow data at the inter-bank level is less important in developing and transi-tion economies with shallow interbank markets sinceinterbank trading accounts for a smaller fraction of thetotal foreign exchange order flow in the market.

Information about the net open foreign exchangepositions of authorized dealers can be used to antic-ipate changes in order flow, as dealers with currencyexposure are likely to go to the market to cover theirpositions, affecting order flow, when changes in theexpected path of the exchange rate take place. Thisinformation also helps identify foreign exchangedealers that may be taking large net open foreignexchange positions and contributing to pressures onthe exchange rate. In most developing countries,banks must report to the central bank their net openforeign exchange positions more than once a month,but the information obtained is seldom published.17

Most reporting is done daily. Weekly reporting ismore common than monthly in all regions exceptEastern Europe. About 70 percent of the countrieswith net open foreign exchange position limits, how-ever, never publish this information.

In addition to obtaining information from reportingrequirements, some central banks in developing

25

13When the country imposes the requirement to surrenderexport receipts to the foreign exchange market, the exporter canoften keep foreign exchange earnings in a foreign currencyaccount for a period before it must sell the foreign exchange.

14The survey did not capture information on the number of coun-tries prohibiting financial contracts from being indexed to theexchange rate, which would preserve the store-of-value role of for-eign currencies and could give rise to exchange rate pressures.

15The provision of such information is often a condition for con-ducting foreign exchange intermediation, to the extent that some-times the obligation to provide information to the central bank isembedded in the foreign exchange license.

16Reporting requirements provide a good picture of foreignexchange market turnover in a country where institutions reportingto the central bank make up the bulk of foreign exchange marketturnover. This is more likely to be the case in about half of thecountries in the sample, which actually prohibit the offshoretrading of their currencies.

17The United States requests information on the net openforeign exchange positions of the internationally activeinternational banks every quarter. That information is available athttp://www.fms.treas.gov/bulletin/index.html.


http://www.fms.treas.gov/bulletin/index.html


economies obtain privileged information about for-eign exchange order flow in certain centralized trad-ing environments, for instance, when conductingforeign exchange auctions. Central banks conductmost foreign exchange auctions in 15 developing andtransition economies that responded to the survey.The central bank actively participated in the auctionsin three countries, but it indirectly participated inmany other auctions by deciding the amounts to beauctioned. A few central banks also have privilegedaccess to the information generated in electronicbroking systems.18 The central bank may be able tocompute foreign exchange order flow directly incountries where it observes foreign exchange transac-tions that take place among banks through an elec-tronic broking system. However, this would coveronly a fraction of the total foreign exchange orderflow, since banks can usually trade among themselvesoutside of those trading platforms.

Control of the payment and settlement systems inthe country can also give an information edge to thecentral bank, as many central banks in developingeconomies are directly involved in the settlement offoreign exchange transactions. In particular, controlof these systems allows the settlement of one or bothlegs of foreign exchange transactions at central bankaccounts. In many of the survey respondents wherefinancial institutions are often required to openaccounts at the central bank to meet reserve require-ments, the debiting and crediting take place at cen-tral bank accounts. The foreign currency legsettlement requires that foreign currency accountsbe opened at the central bank, a situation that oftenarises in dollarized economies in which the reserverequirements on foreign currency deposits aredenominated in foreign currency.19 However, the

information advantage obtained from the control ofthe payment and settlement systems may be difficultto obtain in practice unless special arrangements forthe settlement of foreign exchange transactions arein place to distinguish foreign exchange from othertransactions.

Conclusions

The evidence from the IMF’s 2001 survey on theorganization of foreign exchange markets indicatesthat foreign exchange intervention is a widely usedpolicy instrument in developing and transitioneconomies.

The survey results show that official foreignexchange intervention conducted by some centralbanks in developing economies differs in a numberof ways from that of the central banks of developedcountries that issue the major international curren-cies. First, unlike the U.S. Federal Reserve Board,the European Central Bank, and the Bank of Japan,not all central banks in developing economies fullysterilize their foreign exchange operations on aregular basis. Many developing-country centralbanks back up their foreign exchange interventionswith adjustments to their monetary policy stance.Second, unlike the central banks that issue the majorinternational currencies, some central banks indeveloping economies conduct foreign exchangeintervention in amounts that are significant relativeto the level of foreign exchange market turnover,the money base, and the stock of domestic bondsoutstanding. Third, some central banks in develop-ing economies have a greater information advantageover the central banks that issue the major interna-tional currencies because, among other reasons,they can infer the aggregate foreign exchange orderflow from reporting requirements. Finally, manycentral banks in developing economies also useforeign exchange and monetary regulations, as wellas their own foreign exchange operating practices,to increase their information advantage and the sizeof foreign exchange intervention relative to themarket.

26

18According to the survey, electronic broking systems are inplace in about 40 percent of emerging market economies.

19The fact that most countries also report that the foreigncurrency leg is settled at the accounts of correspondent banksabroad suggests that the central bank accounts serve as anintermediate netting scheme but that the final payments must bedone at the accounts of correspondent banks, transferring moneyto or from the central bank.


Central banks in emerging market countries inter-vene in the foreign exchange market frequently

and sometimes in very large amounts. Most inter-ventions are directed at the exchange rate—to cor-rect misalignments, smooth volatility, accumulatereserves, or supply foreign exchange to the market.Under flexible exchange rate regimes, the timing andamount of intervention—including whether to inter-vene at all—become critical policy decisions. Centralbanks have an overriding interest in the effectivenessof intervention, since intervention exposes them toreputational and financial risks. In many countries,intervention remains important even after moving tomanaged and independently floating exchange ratesfrom various forms of pegs (Bubula and Otker-Robe,2002; and Reinhart and Rogoff, 2003).

Even while adopting greater exchange rate flexi-bility, many countries are reluctant to allow theexchange rate to fluctuate. Exchange rate stabilitystill commands a high premium in emerging marketswhere policy credibility is lower and pass-throughfrom exchange rate movements to inflation is higher(Calvo and Reinhart, 2002). Liability dollarizationand an inability to borrow abroad in their own cur-rencies—which heighten domestic borrowers’ expo-sure to exchange rate risk—also lower countries’tolerance of exchange rate volatility (Hausmann,Panizza, and Stein, 2001). A small number of marketmakers, low turnover in the interbank foreignexchange market, and greater exposure to externalshocks are added sources of volatility in severalemerging market economies. Intervention thusremains widespread, as reported in the previouschapter.

Despite the prevalence of intervention in emerg-ing markets, empirical research on its effectivenessis limited.1 This reflects primarily the absence ofpublicly available data on intervention in emerging

markets. Moreover, it is difficult to model or controlfor changes in policy reaction functions and centralbank credibility in high-frequency time-seriesanalysis.

Intervention may be more effective in emergingmarkets than in advanced ones for several reasons.Many countries intervene in amounts that are largerelative to market turnover. They also use a variety offoreign exchange, monetary, and banking regulationsthat effectively constrict the size of the market,increasing the central bank’s size in it. The centralbank may also have an information advantage overthe market stemming from reporting requirements(Canales-Kriljenko, 2003).

Against this background, this chapter evaluates theeffectiveness of intervention in two emerging marketeconomies—Mexico and Turkey—that were chosenbecause of the availability of daily intervention dataand because they have flexible exchange rateregimes. Using standard methodologies in the litera-ture, effectiveness is measured in terms of the impactof intervention on the level and volatility of theexchange rate. In addition, this chapter differentiatesbetween the effects of intervention on short- versuslong-term exchange rate volatility.

This chapter finds mixed evidence on the effec-tiveness of intervention. In Mexico, foreignexchange sales (but not purchases) have a small butstatistically significant impact on the exchange ratelevel. Since the bulk of interventions by the Bank ofMexico during the period considered here consistedof foreign exchange purchases aimed at accumulat-ing reserves, this is broadly in line with the authori-ties’ objective of intervening without affecting theunderlying exchange rate trend. In Turkey, officialintervention does not appear to systematically affectthe exchange rate levels, a result consistent with theauthorities’ goal of maintaining a market-determinedexchange rate regime. The impact of intervention onexchange rate volatility also differs in both countries:it raises short-term exchange rate volatility in Mex-ico but reduces it in Turkey. These findings cannot begeneralized to other emerging markets, however, andcan be interpreted only in the context of specificcountry circumstances.

IV The Empirics of Foreign ExchangeIntervention in Emerging Markets:Mexico and Turkey

Roberto Guimarães and Cem Karacadag

27

1Domaç and Mendoza (2002) and Tapia and Tokman (2004) areexceptions. By contrast, there is a vast literature on advancedeconomies, which finds mixed evidence in favor of intervention.Where evidence of effectiveness is found, the impact is short lived,as discussed in the next section of this chapter.


IV THE EMPIRICS OF FOREIGN EXCHANGE INTERVENTION IN EMERGING MARKETS

Empirical Analysis and Evidence onIntervention

Analytical Techniques

An array of analytical techniques has been used toevaluate the effectiveness of intervention.2 Analyseshave examined the impact of intervention on theexchange rate level using ordinary least squares(OLS) regressions of the mean and risk premiumequations or through event studies of interventionepisodes.3 The impact of intervention on exchangerate volatility has been gauged through various formsof generalized autoregressive conditional hetero-scedasticity (GARCH) models.

Regression analyses all suffer from simultaneityproblems. In particular, the regression of exchangerate changes over intervention fails to disentanglethe degree to which intervention reacts to exchangerate movements rather than the other way around.As a result, the coefficient can assume the wrongsign or overstate the impact of intervention on theexchange rate.4 Another key shortcoming of existingtechniques, including event studies, is the short timehorizon—typically the same day—over which theeffectiveness of intervention is analyzed.

More recently, attempts have been made to over-come the simultaneity and time-horizon problemsthrough a joint analysis of monetary and exchangerate operations. Using an identified vector autore-gression (IVAR) framework, Kim (2003) andGuimarães (2004) empirically analyze the impact ofmonetary and intervention operations on theexchange rate and the extent to which interven-tion occurs in reaction to exchange rate movements.The IVAR framework can estimate the longer-termeffects of intervention through an analysis of cumu-lative impulse responses. Moreover, the use ofmonthly data eases the limitation imposed by theabsence of daily data (although intervention dataare still obtained through the aggregation of dailydata). However, there are some disadvantages tousing IVAR, including the limited degrees of free-dom (small sample and too many parameters), the

validity of the identifying restrictions, and theplausibility of the structural shocks (Guimarães,2004).

Another strand of the literature examines the linksbetween intervention and exchange rate volatility.The potential impact of intervention on volatility isworth studying, since many central banks interveneto smooth such volatility, even when they are not tar-geting a particular level of the exchange rate. In addi-tion, exchange rate volatility has often beenassociated with economic crisis and may signal alack of policy credibility, which gives rise to fear offloating (Calvo and Reinhart, 2002).5 Finally, volatil-ity may have harmful effects on trade and capitalflows, although evidence supporting this claim isweak (Rogoff, 1999).

The single most important impediment to empiri-cal work on the effectiveness of intervention is thelack of publicly available data on daily intervention.Attempts to use intervention proxies—for example,the change in the stock of central bank reserves—have not worked. Neely (2001) has shown that evenfor Group of Seven (G-7) countries, changes inreserves are a poor proxy for intervention: correlationcoefficients between foreign exchange interventionand reserve changes are usually less than 0.4. The useof such proxies in developing countries can be evenmore misleading, since reserve changes may reflect,all other things being equal, withdrawals of fundsfrom multilateral organizations, government debtrepayments, receipts from state-owned companies,and inflows of foreign aid.

Empirical Evidence

Empirical studies on the effectiveness of centralbank–sterilized intervention have focused almostexclusively on advanced countries. The research biastoward advanced countries reflects primarily theavailability of data and the depth and sophisticationof their foreign exchange markets assumed in manyintervention models.

Empirical tests have found mixed evidence infavor of the signaling and portfolio balance channels.For example, Dominguez and Frankel (1993b) esti-mate the effect of intervention on contemporane-ous exchange rate movements and on forecasts offuture exchange rates. Using survey data to measureexchange rate expectations, they find a significanteffect of intervention on market expectations, espe-

28

2See Edison (1993) and Sarno and Taylor (2001) for excellentsurveys of the methodologies used and of the empirical evidenceon the effectiveness of intervention. Empirical studies, however,often cannot disentangle the effects of the different channelsthrough which intervention affects the exchange rate; seeDominguez and Frankel (1993a, 1993b).

3See Appendix Table 4.1 for a selective survey of the variousmethodologies used to estimate the effects of foreign exchangeintervention on the exchange rate.

4Researchers have also analyzed central bank reaction functionsto gauge the extent to which intervention responds to the exchangerate. However, analyses of central bank reactions suffer from thesame weakness, but in reverse. Intervention is assumed to beineffective; otherwise, simultaneity arises once again.

5More generally, excessive volatility may be a symptom ofdisorderly markets, which involve a collapse of liquidity. How-ever, it is often difficult to empirically identify (ex ante) episodes ofdisorderly exchange rate movements (Canales-Kriljenko,Guimarães, and Karacadag, 2003).


Empirical Analysis and Evidence on Intervention

cially if interventions are announced and coordi-nated. They also show that secret interventions arelargely ineffective. Obstfeld (1990) finds that portfo-lio balance effects are statistically significant butsmall in size. The consensus in the literature untilrecently was that the portfolio effect gives a limitedrole for intervention to influence the exchange rate.One exception was a study that found a significantand potentially large portfolio effect between 1984and 1988, using survey data to measure exchangerate expectations and risk premiums (Dominguezand Frankel, 1993a).

Recent research using data on order flow, how-ever, identifies permanent price effects through theportfolio balance channel. Evans and Lyons (2002,2005) found that intervention has a significantprice impact in the most liquid currency pair mar-ket (before the introduction of the euro), the U.S.dollar–deutsche mark. The permanent effect of a$1 billion purchase was to appreciate the dollar byabout 0.35 percent.6 They also found that foreignexchange transactions have the largest impact on theexchange rate when the flow of macroeconomicannouncements is high.7

More generally, in a series of papers using anevent-study approach, Fatum (2000) and Fatum andHutchison (2003a, 2003b) find strong evidence infavor of intervention. In analyses of both the U.S.dollar–deutsche mark and U.S. dollar–Japanese yenbilateral exchange rates, they find that sterilizedintervention systematically affects the exchange ratelevel, regardless of whether it is secret or announced.The probability of success is much higher, however,when interventions are coordinated among centralbanks and when they are conducted on a large scale(i.e., greater than $1 billion). Also using an event-study approach, Edison, Cashin, and Liang (2003)find that the Reserve Bank of Australia’s interven-tions had some success—albeit modest—in moder-ating the depreciating tendency of the Australiandollar, but that the interventions also increasedexchange rate volatility.

Indeed, the impact of intervention on exchangerate volatility has been extensively researched.Intervention appears to be ineffective in reducingvolatility and, often, increases it.8 Both Dominguez

(1998) and Hung (1997) provide evidence that fol-lowing the Plaza Accord (September 1985) inter-vention tended to reduce exchange rate volatilityamong the Group of Three (G-3) currencies, butwhen the post-Louvre (1987–89) period is exam-ined, intervention increased volatility. Bonser-Nealand Tanner (1996) use implied volatilities from cur-rency option prices and find that intervention raisesexchange rate volatility. Beine, Bénassy-Quéré, andLecourt (2002) study a longer period of interven-tions spanning 1985 to 1995 and also find that inter-vention increases exchange rate volatility in theshort run. Cheung and Chinn (1999) conducted asurvey with foreign exchange traders, 60 percent ofwhom view intervention as increasing exchange ratevolatility.

Joint analyses of monetary and exchange rate pol-icy actions find that intervention is effective in thecase of the United States during the period 1973–96(Kim, 2003) and Japan (Guimarães, 2004). Theapproach is based on a VAR model similar to thoseused to study the monetary transmission mechanism.The identifying restrictions used in these modelsallow the exchange rate to have a contemporaneousimpact on intervention, which captures the “leaningagainst (or with) the wind” by the interveningauthorities. Moreover, the VAR model also permitsestimation of the impact of conventional monetarypolicy shocks (money or interest rate) on theexchange rate. The results also suggest that interven-tion in those two countries is sterilized and has animpact (small but significant) beyond the short termconsidered in most studies that use daily data.

In contrast with most findings for advancedeconomies, empirical evidence on the effects of inter-vention in emerging market economies has beenscant. In their empirical analysis of intervention inMexico and Turkey, Domaç and Mendoza (2002)conclude that central bank foreign exchange sales(but not purchases) were highly effective in influenc-ing the exchange rate and reducing volatility in bothcountries. In particular, they find that a net sale of$100 million appreciates the exchange rate by 0.08percent in Mexico and 0.2 percent in Turkey. A morerecent study on Chile found that intervention had asmall and generally insignificant effect on contempo-raneous exchange rate movements, but in contrast,public announcements on potential intervention had astatistically significant impact on the exchange rate(Tapia and Tokman, 2004).

29

6Their estimate for the immediate price impact of trades was0.44 percent per $1 billion (of which about 80 percent persistsindefinitely).

7The estimated impact of intervention is at best an indicator ofthe impact of intervention under normal market conditions. In aspeculative attack, for example, the credibility of the central bankis so low and liquidity so unpredictable that the estimates aboveshould not even be used as a first approximation.

8The measurement of exchange rate volatility is typically basedon two approaches. The first method is to use a statistical model,such as GARCH. This approach has the advantage of being simple

and is increasingly used in the market to estimate asset pricevolatility. Several market participants use GARCH-based modelsof volatility, such as Riskmetrics, to help monitor their positionsand calculate value at risk. Another approach is to use options-based measures of volatility. Options pricing models can beinverted to yield implied volatilities of the underlying asset.



Policy Context of Intervention inMexico and Turkey

Mexico

Despite the floating of the peso in 1994, the Mex-ican central bank has continued to intervene in theforeign exchange market to smooth exchange ratevolatility. The concern over exchange rate volatilityhas stemmed from the role of the exchange rate as akey monetary policy variable, even though it has lostits anchor role to inflation targets since 1999. As inother emerging markets, the exchange rate hasremained a determinant of inflationary expectations,even under the inflation-targeting framework(Carstens and Werner, 1999; Ho and McCauley,2003). Under inflation targeting, inflation and inter-est rates have come down substantially since themid-1990s (Figure 4.1).

The authorities intervened to accumulate reserves,given the low level of international reserves in theimmediate aftermath of the peso crisis. To this end,the central bank began auctioning put options withthe objective of gradually building up reserves inAugust 1996. The central bank sold put options onthe last business day of each month, allowing theholders of the option to sell U.S. dollars to the cen-tral bank anytime during the life of the option pro-vided that the exercise price, the exchange rate of theday before, was more appreciated than the 20-daymoving average of the interbank spot exchange rate.This condition limited the potential loss faced by theBank of Mexico, since the option could be exercisedonly if the peso was stronger than its 20-day movingaverage.9 The auction of put options continued untilJune 2001 and resulted in an accumulation ofreserves equivalent to 30 percent of reserves whenthe program ended (about $14 billion). Interventionamounts and the peso-dollar exchange rate areshown in Figure 4.2.

From 1996 to 2003, the central bank also inter-vened 14 times in a discretionary fashion, selling for-eign exchange to stabilize the exchange rate, but noparticular level was targeted. Sales of U.S. dollarswere large, totaling $2.9 billion, but still relativelylow compared with the dollar purchases madethrough auctions of put options.10

A significant accumulation of reserves promptedthe authorities to start selling foreign exchangedirectly to the market by May 2003. The amount tobe sold to the market in a given quarter is prean-nounced and equivalent to 50 percent of the reservesaccumulated in the preceding quarter. The dailyamount is based on the total amount for the quarterevenly distributed over the number of business daysof the period in question. The switch aimed toreduce the pace of reserve accumulation and the costof holding additional international reserves, whichhad reached $50 billion at end-2002, equivalent tomore than 120 percent of short-term debt (by resid-ual maturity) or about 50 percent of the monetarybase, up from less than $20 billion at end-1996.Since the switch, dollar sales have been com-

30

9The holder of the option could profit from exercising the optiononly if the exchange rate on the day of its exercise was strongerthan the day before, that is, the strike price.

10The average size of U.S. dollar sales approached $205 million,similar to a typical intervention by the U.S. authorities in the firsthalf of the 1990s. Importantly, official interventions (both options-based and discretionary) represented a sizable fraction of dailyturnover in the foreign exchange market.

0

10

20

30

40

50

0

2

4

6

8

10

12

0

10

20

30

40

Pesos per U.S. dollar(right scale)

Interest rates(3-month)

Inflation(annualized)

Interest ratedifferential

(3-month; left scale)

1996 98 2000 02

1996 98 2000 02

Perc

ent

Perc

ent

Peso

s pe

r U

.S. d

olla

r

Figure 4.1. Mexico: Exchange Rate, InterestRate, and Inflation

Sources: Bank of Mexico and Datastream.


Policy Context of Intervention in Mexico and Turkey

paratively small in magnitude (up to $32 milliondaily) but much more frequent relative to the earlierprogram. According to the authorities, a majorfeature of both the options mechanism and the pre-announced sales is that they minimize the impact on the market mechanism (e.g., pricing decisions)with negligible consequences for exchange ratevolatility. These claims can be tested empirically, atopic which is discussed in the next section of thischapter on the effectiveness of foreign exchangeintervention.

Turkey

Turkey offers important insights on the challengesand limitations of empirically analyzing the effec-tiveness of intervention. Among emerging marketeconomies, Turkey is one of the few countries with a(managed) floating exchange rate regime for whichdaily intervention data, although somewhat incom-plete, are available. During the period studied here(March 2001–October 2003), the country imple-mented substantial economic reforms, lived throughbouts of domestic political uncertainty, and was hitby financial market shocks.

Turkey’s exit from a crawling peg in February2001 shifted the burden of price discovery to the for-

eign exchange market at a time when it was stillundeveloped. During the crawling peg exchange rateregime, the foreign exchange market was heavilyinfluenced by the Central Bank of Turkey (CBT),with most banks trading bilaterally with the CBTrather than among themselves. At the time of the exitfrom the peg foreign exchange market, liquidity waslow, hedging instruments were virtually nonexistent,and financial institutions were caught with sizableshort foreign currency positions and with limitedcapacity to manage foreign exchange risk. As aresult, low turnover in the foreign exchange marketmay have been a dominant factor determining theexchange rate compared with the CBT’s interven-tions, at least during the early phases of the periodanalyzed here.

Since the flotation of the lira, the CBT’s interven-tions have passed through several phases (Table 4.1).The CBT initially sold foreign exchange throughauctions to sterilize the liquidity injections associ-ated with the Turkish Treasury’s use of externalfinancial resources. These were combined with dis-cretionary interventions to smooth exchange ratevolatility related to negative external developmentsand domestic political problems (Central Bank ofTurkey, 2001). The CBT began conducting prean-nounced (timing and amount) foreign exchange sale

31

0

2

4

6

8

10

12

03020120009998971996–600

–500

–400

–300

–200

–100

0

100

200

300

400

Mill

ions

of U

.S. d

olla

rs

Peso

s pe

r U

.S. d

olla

r

Exchange rate(right scale)

Intervention (left scale)

Figure 4.2. Mexico: Exchange Rate and Foreign Exchange Intervention

Source: Bank of Mexico.



auctions in March 2001. Preannounced auctionswere designed to enhance the transparency of officialintervention and minimize the price impact. Auc-tions remained the main form of interventionthroughout 2001, with brief interludes of discre-tionary intervention in lieu of or in parallel with pre-announced auctions. Throughout 2001, the CBT sold$6.5 billion in foreign exchange, enabling financialinstitutions to cover their short positions.

April 2002 marked the beginning of the second setof intervention phases, characterized by foreignexchange purchase operations. The move from for-eign exchange sales to purchases was driven inpart by reverse currency substitution engenderedby growing confidence in policy formulation andimplementation, and a pickup in capital inflows(Central Bank of Turkey, 2002). Foreign exchangepurchases—first through preannounced auctions,then on a discretionary basis—were suspended inJuly 2002 amid uncertainties before the November2002 general elections, but resumed in May 2003 asuncertainties faded and reverse currency substitutioncontinued.

Strong upward pressure on the lira caused theCBT to combine preannounced purchase auctionswith discretionary or optional foreign exchange pur-chases. As a result, the CBT accumulated close to $5 billion in reserves from May through October2003. It is also worth noting that there were lengthyperiods when the CBT did not intervene in the mar-ket at all (December 2001–March 2002 and July2002–April 2003).

Despite its frequent presence in the market over longperiods, the CBT has stated repeatedly through pressreleases and official policy statements that its inter-ventions do not target a specific exchange rate level(Central Bank of Turkey, 2003). These policy pro-nouncements, in principle, may be interpreted as hav-ing been designed to undercut the signaling channel bywhich intervention can influence the exchange rate.

Nevertheless, the CBT has been concerned aboutexchange rate misalignment and volatility (CentralBank of Turkey, 2001, 2002, 2003). The exchangerate has remained an important determinant ofinflationary outcomes and expectations even after thefloat, and the authorities have been vigilant againstvolatility and have resisted it, market conditions andinternational reserve levels permitting. Figure 4.3suggests that CBT interventions tended to “leanagainst the wind,” particularly during June–October2001 and May–October 2003, among the two periodsof heaviest intervention. Moreover, CBT interven-tions appear to have been more successful in temper-ing exchange rate movements when the lira wasunder upward rather than downward pressure.

Over the period of interventions analyzed here,market conditions and the CBT’s operating environ-ment became more favorable. First, the CBT gradu-ally regained credibility as it adhered to its monetaryprogram and exercised restraint and transparency inthe conduct of foreign exchange intervention. Sec-ond, the shift to the new nominal anchors, from basemoney to the inflation target, became increasinglyentrenched, reducing the pass-through from the

32

Table 4.1. Turkey: Central Bank Foreign Exchange Intervention, March 2001–December 2003

Amount Sold (+)Duration or Purchased (–)

Dates (Days) Type of Intervention Frequency (US$ millions)

2/23/01–3/28/01 . . . Discretionary auctions and sales Ad hoc . . .3/29/01–5/17/01 37 Preannounced sales Daily 2,0405/18/01–7/11/01 39 Discretionary sales1 Ad hoc 1,6787/12/01–8/31/01 38 Preannounced and discretionary sales Biweekly 1,5759/4/01–11/30/01 64 Preannounced sales Daily 1,18012/1/01–3/31/02 85 No intervention . . . . . .4/1/02–4/30/02 22 Preannounced purchases Daily –2805/1/02–6/28/02 43 Discretionary purchases Ad hoc –5157/1/02–5/5/03 212 No intervention2 . . . . . .5/6/03–8/31/03 83 Preannounced and discretionary purchases3 Daily –2,9609/1/03–10/1/03 29 Preannounced purchases with option for Daily –1,824

additional purchases

Sources: Central Bank of Turkey (CBT), annual reports, press releases, and website (www.tcmb.gov.tr).1Data on the CBT’s discretionary interventions are not available.2From July to December 2002, the CBT intervened three times (purchases) on a discretionary basis.3The amounts of discretionary purchases were not disclosed.


www.tcmb.gov.tr

Effectiveness of Foreign Exchange Intervention

exchange rate to inflation (Central Bank of Turkey,2003). Third, favorable external finances and reversecurrency substitution put upward pressure on thelira, aiding efforts to reduce inflation, lower interestrates, and bolster debt sustainability. As a result, thesupply of foreign exchange in the market consis-tently exceeded preannounced amounts set for thepurchase auctions. Exchange rate volatility alsodeclined, albeit gradually (Figures 4.4 and 4.5).Fourth, the emergence of a vibrant interbank foreignexchange market and the CBT’s less dominant rolein it created more room for market participants toprice foreign exchange.


Empirical work on foreign exchange interventionin emerging market countries has been limited,despite factors suggesting that intervention may bemore effective in these countries.11 The evidence sur-veyed by Canales-Kriljenko (2003) suggests that offi-cial intervention in developing countries may bemore effective because (i) its size is usually large rel-ative to the local market (order flow, bonds outstand-ing), (ii) domestic and foreign bonds are more likely

to be imperfect substitutes,12 and (iii) the central bankmay enjoy additional informational advantagesowing to the market size/infrastructure and reportingrequirements.

Data Description

Mexico

Daily data on foreign exchange interventionsare publicly available and cover the period August1996 through June 2003.13 Both option-basedand discretionary interventions are included. Thedata consist of 1,800 observations, including no-intervention days. The second part of the dataincludes the spot exchange rate (pesos per U.S.dollar), Mexican interest rate (Cetes 90-day),U.S. three-month treasury bill rate, yields on theBrady bond, and turnover and open interest on theMexican peso contract traded on the Chicago Mer-cantile Exchange’s International Money Market(IMM).14

Table 4.2 presents the descriptive statistics for theexchange rate, the (log) first-difference of the

33

12Cumby and Obstfeld (1983) present evidence supporting theimperfect substitutability between peso-denominated and foreigncurrency assets for Mexico using data from the 1970s.

13The data are from the Bank of Mexico’s website:www.banxico.org.mx.

14The second part of the data set is from Datastream.

0

200000

400000

600000

800000

1000000

1200000

1400000

1600000

1800000

2000000

–250

–200

–150

–100

–50

0

50

100

200320022001

Mill

ions

of U

.S. d

olla

rs

CBT intervention(left scale)

Turkish lira–U.S. dollar exchange rate(right scale)

Turk

ish

lira

per

U.S

. dol

lar

Figure 4.3. Turkey: Central Bank Intervention and the Exchange Rate,March 29, 2001–October 3, 2003

Sources: Central Bank of Turkey (CBT) and Datastream.

11Canales-Kriljenko (2003) and Canales-Kriljenko, Guimarães,and Karacadag (2003) discuss those arguments in more detail.


www.banxico.org.mx


exchange rate, interest rate differential, and spreadson Brady bond yields. The results for the AugmentedDickey-Fuller (ADF) unit root test for the variablesused in the regressions indicate that the series are

nonstationary (with the exception of the differencedseries), and some display other characteristics sharedby financial time series, including departures fromnormality (e.g., fat tails and skewness) and volatility

34

–300

–200

–100

0

100

200

300

–10

–8

–6

–4

–2

0

2

4

6

8

10

2001 2002 2003

Change inTurkish lira–U.S. dollar

exchange rate(right scale)

CBT intervention(left scale)

Mill

ions

of U

.S. d

olla

rs

Perc

ent

Figure 4.4. Turkey: Central Bank Intervention and Exchange RateReturns, March 29, 2001–October 3, 2003

Sources: Central Bank of Turkey (CBT) and Datastream.

0

250000

500000

750000

1000000

1250000

1500000

1750000

2000000

0

2

4

6

8

2001 2002 2003

Perc

ent

Turk

ish

lira

per

U.S

. dol

lar

Exchange rate(left scale)

Volatility(22-day moving standard deviation; right scale)

Figure 4.5. Turkey: Exchange Rate and Exchange Rate Volatility

Source: Datastream.



clustering, which are explored in the empirical mod-els used in this chapter.15

Turkey

The set of daily data on foreign exchange sale andpurchase auctions spans from March 29, 2001,through October 3, 2003, with more than 600 obser-vations.16 The set excludes discretionary foreignexchange sales and purchases through brokers andbanks, including the CBT’s large discretionarypurchases that exceeded those through auctionsbetween May and October 2003. However, theabsence of data on discretionary interventions,which we model through the use of a dummy vari-able, does not seem to be an important handicap inthe empirical analysis.

Descriptive statistics on variables used to analyzethe effectiveness of intervention are presented inTable 4.3. Distributions of all variables are asym-metric and nonnormal. ADF tests indicate that nearlyall variables are nonstationary; thus, we take first dif-ferences before including them in the regressions.

The data sample was divided into two subperiods,and separate GARCH regressions were run on thesubperiods as well as the entire sample. Dividing thesample facilitates the analysis of important differ-

ences in the type of interventions (sales versus pur-chases) and the environment in which they were con-ducted. The first subperiod, from March 2001 throughJune 2002, was dominated by foreign exchange salesand characterized by greater market and politicaluncertainty. The second subperiod, from July 2002through October 2003, represented a period of greaterconfidence in the policy environment, when virtuallyall of the CBT’s interventions were purchase opera-tions. Compared with the first period, interest ratesand spreads were substantially lower and, along withexchange rates, considerably less volatile.

Empirical Model

The effects of intervention on the level andvolatility of exchange rates are analyzed within theGARCH framework. The primary advantage ofGARCH models is that they provide a unifiedframework to gauge the impact of intervention onthe mean and conditional variance of exchange ratereturns simultaneously. The empirical model allowsthe estimated conditional volatility to enter themean equation (i.e., the “GARCH-in-mean” effect)and tests for asymmetric effects on volatility of“negative” shocks, defined as unexpected exchangerate depreciations. In addition to their computa-tional simplicity, GARCH models provide relativelygood forecasts of realized volatility and have proveduseful for modeling the volatility dynamics ofexchange rates and asset prices more generally(Andersen and Bollerslev, 1998).

The empirical analysis of the effectiveness ofintervention is based on the Asymmetric Compo-

35

Table 4.2. Mexico: Descriptive Statistics

StandardSample Period Mean Deviation Skewness Kurtosis Jacque-Bera Stationarity1

Total sample (August 1996–June 2003)Peso-dollar exchange rate 9.19 0.85 –0.32 2.42 56.2 NonstationaryFirst differential of log exchange rate 0.02 0.56 1.44 21.10 25,317 StationaryInterest differential (percent) 13.3 7.00 1.08 4.28 472 NonstationarySpreads (basis points) 11.1 1.54 1.06 4.15 436 Nonstationary

Subsample I (August 1996–June 2001)Peso-dollar exchange rate 9.0 0.81 –0.36 1.77 108 NonstationaryFirst differential of log exchange rate 0.01 0.60 1.60 22.5 20,795 StationaryInterest differential (percent) 16.3 6.25 1.29 4.92 553 NonstationarySpreads (basis points) 11.3 1.60 1.07 3.82 280 Nonstationary

Sources: Bank of Mexico and authors’ calculations.1According to the Augmented Dickey-Fuller test with a trend term and a maximum number of five lags selected according to the Bayesian Informa-

tion Criteria, at the 90, 95, 99 percent confidence intervals. Stationary (nonstationary) refers to stationarity (nonstationarity), that is, rejection (nonre-jection) of the unit root null at the 5 percent level.

15The ADF regression is run with up to five lags, and the num-ber of lags included is selected according to the Bayesian Infor-mation Criteria (BIC). The Dickey-Fuller Generalized LeastSquares tests yield similar results.

16The data are from the Central Bank of Turkey’s website:www.tcmb.gov.tr.


www.tcmb.gov.tr


nent Threshold GARCH (ACT-GARCH) specifica-tion, which jointly estimates the impacts of inter-vention on volatility at different time horizons. Thevolatility part of the model allows for asymmetricresponses of the conditional volatility to unexpectedexchange rate depreciations. Furthermore, themodel is consistent with stylized facts in asset pric-ing empirics, including persistence and volatilityclustering.

The baseline model is given by

Δst = β0 + β1It– + β2It

+ + β3Δdt + β4Δspt + εt (1)

ht = qt + α (ε2t–1 – qt–1)

+ τ(ε2t–1 – qt–1) zt–1 + δ(ht–1 – qt–1)

+ γ1It– + γ2It

+ + γ3Δdt + γ4Δspt (2)

qt = ω + ρ(qt–1 – ω) + φ(ε2t–1 – ht–1)

+ γ5It– + γ6It

+ + γ7Δdt + γ8Δspt, (3)

where Δst is the (log) first-difference of theexchange rate (expressed in terms of local currencyper U.S. dollar and in log form), It

– (It+) denotes

sales (purchases) of foreign currency in millions ofU.S. dollars by the central bank for intervention pur-poses, dt is the interest rate differential (domesticminus foreign), and spt is the yield spread on a sov-ereign foreign currency bond over a comparableU.S. treasury bond. The log exchange rate, interestdifferential, and yield spreads are first differenced to

obtain stationarity. The error term is the unexpectedreturn that is used to model the conditional volatil-ity of the exchange rate in the volatility equations(2) and (3).

Equation (1) of the empirical model (the “mean”equation) analyzes changes in the exchange ratereturn (depreciation or appreciation against the dol-lar) as a function of intervention, interest rate differ-entials, and yield spreads on a sovereign bond. Theinterest differential aims to capture the possibleimpact of monetary policy actions and local moneymarket conditions on the exchange rate. Yieldspreads on sovereign external debt over a compara-ble U.S. treasury bond are included as a measure ofcountry risk and foreign investor sentiment, whichare possibly key determinants of demand for localcurrency. It is hypothesized that a higher interest dif-ferential appreciates the domestic currency (Δst < 0),net purchases of foreign exchange depreciate thedomestic currency, and higher yield spreads are asso-ciated with depreciations of the domestic currency.17

36

Table 4.3.Turkey: Descriptive Statistics

StandardSample Period Mean Deviation Skewness Kurtosis Jacque-Bera Stationarity1

Total sample (3/29/01–10/1/03)Turkish lira-US$ exchange rate 1,468,505 153,143 –0.31 2.27 24.5 NonstationaryFirst differential of log exchange rate2 0.49 13.5 1.2 15.5 4,279 Stationary

Interest diff. (percent) 48.7 10.6 0.45 3.08 21.4 StationarySpreads (basis points) 790 153 0.26 2.02 32.4 Nonstationary

Subsample I (3/29/01–6/30/02)Turkish lira-US$ exchange rate 1,375,984 131,347 –0.10 2.54 3.3 NonstationaryFirst differential of log exchange rate2 1.37 17.4 0.94 11 873 Stationary

Interest differential (percent) 57.3 7.5 0.88 4.09 56.5 NonstationarySpreads (basis points) 809 161 0.24 1.87 19.8 Nonstationary

Subsample II (7/1/02–10/3/03)Turkish lira–US$ exchange rate 1,559,861 113,125 –0.53 1.66 38.8 NonstationaryFirst differential of log exchange rate2 –0.38 7.94 0.83 5.57 124 Stationary

Interest differential (percent) 40.4 4.98 –1.1 2.97 63.5 NonstationarySpreads (basis points) 772 144 0.19 2.02 114 Nonstationary

Sources: Central Bank of Turkey, Datastream, and authors’ calculations.1According to the Augmented Dickey-Fuller test with a trend term and a maximum number of five lags selected according to the Bayesian

Information Criteria, at the 90, 95, 99 percent confidence intervals. Stationary (nonstationary) refers to stationarity (nonstationarity), that is, rejection(nonrejection) of the unit root null at the 5 percent level.

2The mean and standard deviation are multiplied by 103 for presentational purposes.

17Generally, it is hypothesized that a higher interest differentialappreciates the domestic currency (Δst < 0) on impact. We do notdiscard, however, the possibility of reverse causality, since (i) thehypothesis holds only if the expected exchange rate is constant,and (ii) exchange rate depreciations may have a direct contempo-raneous impact on domestic yields if, for example, market condi-tions are tight.



37

In equation (2), ht is the conditional volatility ofthe exchange rate (log returns), and zt is a dummyvariable indicating unexpected exchange rate appre-ciations (i.e., if εt < 0, then zt > 0). The model allowsmean reversion of the short-term volatility, ht, to atime-varying longer-term volatility, given by qt, incontrast to the constant long-term volatility assumedin the standard GARCH model.18

Equation (2) models the short-term conditionalexchange rate volatility, ht, as a function of a time-varying long-term volatility, qt; lagged unexpectedshocks relative to lagged long-term volatility, givenby the term (ε2

t–1 – qt–1); lagged volatility relative to lagged long-term volatility, given by the term(ht–1 – qt–1); and the set of regressors of explanatoryvariables that were included in the mean equation(intervention, interest rate differentials, and sover-eign spreads). Lagged unexpected shocks andvolatility are included to capture volatility cluster-ing (as in standard GARCH-type models), sincehigh volatility periods tend to be clustered over time. The equation also includes a term [(ε2

t–1 – qt–1)z t–1] that allows for asymmetric impactsof past shocks (relative to long-term volatility) onshort-term volatility. If the estimated τ is less thanzero, then unexpected depreciations increase short-term volatility.

The model departs from the standard GARCH rep-resentation by assuming that the long-term volatilityis not constant. The long-term volatility equation isgiven by equation (3), and like its short-term coun-terpart, it depends on a set of explanatory variables(intervention, interest differentials, and sovereignspreads), its own lagged value (qt–1), and past shocks(ε2

t–1). Unlike the short-term volatility, qt converges toa constant ω.

Some general features of the model above arenoteworthy. First, it allows for asymmetric shocksin the conditional (short-term) variance equation. Inparticular, if τ < 0, then the impact of “negative”shocks (unexpected domestic currency depreciation,εt > 0) on short-term volatility is given by α, greaterthan the impact of “positive” shocks (unexpectedappreciation), which is given by (τ + α).19 Second,the short-term impact of foreign exchange inter-vention on exchange rate volatility may differ fromthe long-term impact. The empirical model may alsobe augmented with other exogenous variables, suchas sovereign spreads, market turnover, order flow,

and other relevant variables. For instance, in thecase of Mexico, futures market turnover (and openinterest) is included in the exchange rate volatilityequations.20

This chapter also analyzes the effects of volatilityon intervention, given the evidence in favor ofreverse causation between exchange rate returns andintervention (policy reaction function). FollowingDominguez (1998) and Baillie and Osterberg (1997),we apply the probit model to evaluate whether exces-sive volatility, defined as the deviation of estimatedvolatility from its recent trend, increases the proba-bility that the central bank will intervene in the for-eign exchange market.21 In addition to excessivevolatility, the estimated probit model includes thedeviation of the current exchange rate from its recentmoving average. Although other variables could beincluded in the model, the estimated model appearsto perform quite well and makes our analysis morecomparable to previous work. The estimated modelis given by

Pr{|It| ≠ 0} = Φ[α0 + α1(st–1 – Σj=1,kst–j /k)+ α2(h1/2

t–1 – Σj=1,kh1/2t–j /k)], (4)

where Pr denotes probability and Φ denotes the stan-dard normal transformation. If the estimated α1 (α 2)is statistically significant (different from zero), thendeviations from the k-day exchange rate trend(volatility) affect the probability of intervention.

Estimation Results

Mexico

The first set of regressions on Mexico (notreported here) highlight the importance of specifyingthe model and properly accounting for the simul-taneity problem. The regressions also show that theeffects of intervention on the exchange rate varyaccording to the sample period and whether inter-vention is lagged or contemporaneous.22 For exam-ple, the regression of exchange rate returns oncontemporaneous central bank purchases indicates

18Note that while h – q, the deviation of volatility from its long-term component, converges to 0 with powers of α + δ, the long-run component converges to ω with powers of ρ. This model canbe reparametrized as a nonlinear restricted GARCH(2,2) model.

The absolute value of the regressors is used in the varianceequations.

19Note that for τ < 0 and α > 0, α > τ + α.

20The number of outstanding contracts, measured in U.S. dollars.See also Jorion (1996) for the case of G-3 currency pairs.

21The estimated volatility is based on a simple GARCH(1,1)model, although the same qualitative results are obtained if theACT-GARCH model is used.

22The point estimates also differ significantly depending onwhether sales and purchases of foreign exchange are consideredseparately or in conjunction. The error term is assumed to have at-distribution to account for excess kurtosis in the data (fat tails).The estimates for the t-parameter (degrees of freedom) are alwayssignificant at the 5 percent level and indicate major departuresfrom normality.



38

that a $100 million purchase by the Bank of Mexicoappreciates the peso by 0.2 percent against the U.S.dollar, while an equivalent sale of U.S. dollars depre-ciates the peso by 1.4 percent (both statistically sig-nificant at the 1 percent level). This result could beerroneously interpreted as “leaning against thewind,” but, in fact, it is consistent with the rationalethat investors tend to exercise their put options whenthe domestic currency appreciates. This, in turn,would make the error term correlated with theexplanatory variable, which requires the applicationof instrumental variables/generalized method ofmoments or the use of lagged intervention to accountfor the correlation.23

To redress the simultaneity problem, the modelwas respecified by using two-day lagged interven-tion. The estimates from the second set of regres-sions using lagged intervention are presented in

Table 4.4. The upper half of the table shows the esti-mates for the mean equation.

The results indicate that the impacts of interven-tion on the level of the exchange rate are nontrivial.In particular, a two-day lagged $100 million saleappreciates the peso (against the U.S. dollar) by0.4 percent (statistically significant at the 5 percentlevel), but purchases of foreign exchange do nothave a statistically significant impact on the value ofthe peso.24 The results also underscore the impor-

Table 4.4. Mexico and Turkey: Asymmetric Component GARCH Model Estimates

Δst = β0 + β1It– + β2It+ + β3Δdt + β4Δspt + εt

ht – qt = α (ε2t–1 – qt–1) + τ(ε2

t–1 – qt–1) zt–1 + δ(ht–1 – qt–1) + γ1It– + γ2It++ γ3Δdt + γ4Δspt

qt = ω + ρ(qt–1 – ω) + φ(ε2t–1 – h t–1) + γ5It– + γ6It+ + γ7Δdt + γ8Δspt

Mexico Turkey1

Exchange rate level (mean) equationβ1 (Intervention—sale) –0.43** –0.34β2 (Intervention—purchase) 0.01 0.07β3 (Interest differential) 0.04** –2.03β4 (Spreads) 0.22** 1.72*

Short-term volatility equationτ (Negative shock) -0.24** 0.04γ1 (Intervention—sale) 0.06** –0.01***γ2 (Intervention—purchase) 0.00 –2.08E–07γ3 (Interest differential) 0.02 –1.25E–05γ4 (Spreads) 0.19** 3.66E–07

Long-term volatility equationγ5 (Intervention—sale) 0.01** 0.01**γ6 (Intervention—purchase) –6.73E–05 2.72E–08γ7 (Interest differential) –0.02** 3.17E–06γ8 (Spreads) 0.02 4.38E–07

Notes: * denotes significance at 10 percent, and ** denotes significance at 5 percent. GARCH = generalizedautoregressive conditional heteroskedasticity models.The results reported above are based on equations (1)–(3)after dropping the weekend dummy and the moving average parameter, which were insignificant in all specificationsestimated.The coefficient on intervention in the mean equation gives the impact for a $100 million purchase/saleof foreign exchange as a percentage.The sample sizes are, respectively, 1,800 and 1,278.

1Reported coefficients are multiplied by 105 for presentational purposes.

23Werner (1997) discusses this issue in detail.

24When the sample period studied by Domaç and Mendoza(2002)—August 1996 to June 2001—is used, the results are as fol-lows: in the case of contemporaneous intervention, both estimatesremain significant at the 5 percent level, with the same patterns ofsigns and magnitude as those reported above; in the case of laggedintervention, the point estimates are also about the same (qualita-tively and quantitatively), but only sales remain statistically sig-nificant at the 10 percent level. It is worth noting that the resultsmay not be comparable to those found by Domaç and Mendoza(2002), since the econometric specifications are different, as aresome of the variables used in the estimations.



tance of estimating the impact of sales and pur-chases of foreign exchange separately, particularlywhen there are systematic differences between pur-chases and sales.

The effects of intervention on volatility are alsosignificant and are shown in the lower two-thirds ofTable 4.4. Several factors account for the time-varying nature of exchange rate volatility, includingthe yield spread on Brady bonds, interest rate differ-ential, and intervention. The impact of interventionon exchange rate volatility is also estimated sepa-rately for sales and purchases.

The model indicates that foreign exchange salesand changes in the Brady bond yield spread increasethe short-term volatility of the exchange rate (ht) atthe 5 and 1 percent significance levels, respectively.The estimates indicate the existence of asymmetricshocks to the conditional variance (significant at the 5 percent level in both samples): unexpected domes-tic currency depreciations have a larger effect onvolatility than do unexpected appreciations. By con-trast, neither changes in the interest rate differentialnor foreign exchange purchases affect short-termexchange rate volatility. Foreign exchange sales alsoincrease the long-term component of exchange ratevolatility (significant at the 5 percent level). The esti-mated effect of purchases is negative but not statisti-cally significant.25 Changes in the Brady bond yieldspread do not have a significant impact on long-termvolatility, while changes in the interest rate differen-tial have a negative effect (significant at the 5 percentlevel).

The main empirical findings for Mexico may beinterpreted as follows.26 First, intervention seems tohave a nonnegligible effect on exchange ratechanges, with a $100 million sale of foreignexchange by the central bank estimated to appreciatethe peso (against the U.S. dollar) by 0.4 percent.However, foreign exchange purchases, which consti-tute the bulk of interventions in Mexico during theperiod covered here, do not appear to have had a sta-tistically significant impact on the value of the peso.This is consistent with the authorities’ objective ofaccumulating international reserves in a floatingexchange rate regime.

Nevertheless, intervention has a nontrivial impacton exchange rate volatility. The results indicate thatforeign exchange sales increase both the short- andlong-term volatility of the exchange rate, which in

part may be because sales are seen as less credible.27

Other factors (e.g., interest rate differentials) alsoappear to be important in explaining exchange ratevolatility at longer horizons. This finding has impor-tant implications for exchange rate policy, especiallyif (short-term) volatility-enhancing interventions caninfluence the expectations of market participants, asdescribed by Hung (1997).

The probit estimations for Mexico indicate that“excessive” exchange rate volatility decreases theprobability of intervention, in contrast with the find-ings of Dominguez (1998) and Baillie and Osterberg(1997) (Table 4.5).28 The estimations also reveal thatexchange rate depreciations increase the probabilityof intervention. The results, which are based on a21-day window (k = 21) and the full sample, aregenerally not robust to the choice of k and the sampleperiod. For instance, when the sample period1996–2001 is considered, increases in volatilityrelative to its recent trend raise the probability ofintervention. Nonetheless, the probit estimationresults underscore the importance of controlling forsimultaneity effects when estimating the impacts ofintervention on the level and volatility of theexchange rate.

39

Table 4.5. Mexico and Turkey:Probit Model Estimates

Pr{⎥ It⎥ ≠ 0} = Φ[α0 + α1(st–1 – Σj=1,k st–j /k)

+ α2(ht–11/2 – Σj=1,kht–j

1/2/k)]

Mexico TurkeyCoefficients 1996–2003 2001–03

α0 –1.36** –0.38**α1 8.58* 11.28α2 –0.03** –2.42

Notes: * and ** denote significance at 10 and 5 percent levels,respectively. Huber-White standard errors are used. The resultsare based on k = 21; st is the log of the nominal local currency–U.S. dollar rate; and ht denotes the estimated condi-tional volatility of the exchange rate return.

25In the case of sales, one could argue that attempts by themonetary authorities to smooth volatility with discretionary saleswere not fully credible, resulting in higher volatility.

26The model is also estimated under different assumptions aboutthe error term, including the t-distribution and the generalizedexponential distribution (GED) (Nelson, 1991).

27Generally, foreign exchange sales could be perceived as lesscredible because central banks often sell foreign exchange toprevent the domestic currency from depreciating, even if theultimate objective is to smooth volatility (since depreciations andheightened uncertainty are highly correlated).

28“Excessive” exchange rate volatility is defined as a deviation ofthe estimated volatility from its 21-day moving average. Theincrease in volatility, which precedes intervention (i.e., exercisingof put options) at time t1, could be the result of dynamic hedgingby market participants following the purchase of the options at t0.



Turkey

In Turkey’s case, regression estimates indicatethat official intervention does not influenceexchange rate levels (Table 4.4). Coefficients onthe sale and purchase of foreign exchange carrythe wrong sign (consistent with a “leaning againstthe wind” policy), but they are statistically insig-nificant. By contrast, sovereign spreads are highlysignificant. The regressions were also run on theentire sample as well as on the two subsamplesin an attempt to distinguish the more volatilemarket conditions of the earlier subperiod fromthe more favorable conditions in the later sub-period. A second set of regressions with dummiesfor the presence of the CBT in the market werealso run. A third set of regressions includedlagged intervention. Finally, a dummy variablewas used to account for the CBT’s discretionaryinterventions in unknown amounts. None of thevariants on the sample period, regressors, dummyvariables, or lags yielded results qualitatively dif-ferent from the ones reported here. Regressionswere also run under alternative GARCH spec-ifications (not shown here) to illustrate the sensitiv-ity of the results to model specification and thelimitations of empirically modeling the interven-tion’s effectiveness.29

The absence of evidence on the effectiveness ofintervention in influencing the level of the exchangerate may reflect the nature of Turkey’s interventionpolicies. Most official interventions were conductedin the context of preannounced foreign exchangesale (purchase) auctions, where the time andamounts were largely predetermined and known bymarket participants. Hence, the potential impact ofinterventions may have operated through the signal-ing channel well in advance of the actual interven-tions. Another interpretation of the results is that theauthorities succeeded in their stated goal of main-taining a market-determined exchange rate by imple-menting a transparent, rules-based interventionpolicy.

The regression results (lower two panels of Table4.4) suggest that intervention affects exchangerate volatility. Coefficients γ1 and γ5 on CBT foreignexchange sales in the short-term volatility equa-tion are statistically significant at the 5 percentlevel. In particular, CBT foreign exchange salesappear to reduce short-term volatility (which isconsistent with the findings of Domaç and Men-doza, 2002), but over the long term, sales increase

volatility. Unlike in Mexico, unexpected deprecia-tions do not appear to have asymmetric effects onvolatility.

The probit estimation for Turkey finds no evi-dence that the probability of intervention increasesin response to deviations of exchange rate vola-tility from its recent trend (Table 4.5). This result isconsistent with Turkey’s largely rules-based inter-vention policy during this period, during which timemost interventions were preannounced. The resultsalso show that exchange rate trends did not appearto have had an impact on the probability of inter-vention. Thus, the authorities rarely reacted tocontemporaneous market conditions, with the possi-ble exception of the few episodes of discretionaryinterventions.

Conclusions

This chapter has found mixed evidence on theeffectiveness of intervention in Mexico and Turkey.In Mexico, foreign exchange sales (but not pur-chases) have a statistically significant impact on theexchange rate level, while in Turkey, neither foreignexchange sales nor purchases are significant. In bothcases, these findings are broadly consistent with offi-cially stated policy objectives, which generally aimto minimize the effect of intervention on theexchange rate. The results are also broadly consistentwith the empirical analyses on advanced economies,where intervention is generally found to have little, ifany, effect on the exchange rate (Sarno and Taylor,2001).

The evidence also shows that intervention mayhave nontrivial effects on exchange rate volatility. InMexico, sales of foreign exchange are usually asso-ciated with increases in exchange rate volatility, incontrast to the often-stated objective of interventionto smooth volatility. The evidence is more mixed inTurkey, with foreign exchange sales (but not pur-chases) reducing volatility in the short term, butincreasing it in the long term. In both cases, theresults do not seem to substantiate claims that inter-vention smooths volatility.

The apparently limited effectiveness of interven-tion highlights the need for central banks to use theirscarce foreign reserves parsimoniously. The diffi-culty of identifying a strong link between interven-tion and exchange rate changes, however, may alsostem from model misspecification and a failure tocontrol for a variety of political and economic fac-tors. The hope is that this paper stimulates furtherresearch on the effectiveness of intervention—partic-ularly by the staff of central banks—to ensure thatinternational reserves are well spent when used forinterventions.

40

29The findings contrast with those of Domaç and Mendoza(2002) that foreign exchange sales (but not purchases) havesignificant effects on exchange rate returns.


Appendix

Appendix

41

Table A4.1. Analytical Methodologies of Empirical Studies on Intervention Effects on the Exchange Rate

Sources Economic Specification Data Requirements Advantages and Disadvantages

Dominguez and Frankel(1993a)

OLS Regression of Mean Equation

Δst = α + βIt + γ ′Xt + εt,

where Δs is the exchange rate change, I isintervention, and the X vector includes theinterest differential, the country risk premium,and possible dummies, including on seasonalityeffects (e.g., Monday effect), news items, andreports of central bank presence in the market.

• Intervention, exchange rate,interest rate, country riskpremium (e.g., spreads onsovereign bond) on a daily basis

• Compilation of news items,reports of central bankintervention, and other relevantinformation for dummy variables

• Simple technique

• Simultaneity between exchange ratechanges and intervention

• Analyzes only contemporaneouseffects with no insight on long-termeffects

• Provides no insight on the channelof transmission

Evans and Lyons(2002)

OLS Regression of Order Flow Equation

Δst = α + βΔrt + γxt + εt,

where Δrt is the change in the interestdifferential, and xt is interdealer order flow,which is defined as the net of buyer-initiatedand seller-initiated foreign exchange orders thatare consummated.

• Daily data on interest rates,exchange rates, and interdealerorder flow

• Order flow data are not alwaysavailable

• Effectiveness of order flow inaffecting exchange rates is anindirect measure of the effectivenessof central bank–initiated order flow(intervention)

• Simultaneity between exchange ratechanges and order flow

• Measures the impact of privatesector–generated order flow, whichis equivalent to a sterilized secretintervention

Fatum (2000)

Fatum and Hutchison(2003a, 2003b)

Event Studies

An event window of 2, 5, 10, and 15 days isdefined to include one or more interventionepisodes (interspersed with noninterventiondays), during which exchange rate changes areanalyzed compared with the pre-event window.

• Intervention and exchange ratedata on a daily basis

• Simple and analytically soundtechnique

• Provides no insight on the channelof transmission

• Analyzes only short-term effects(up to one month) with no insighton long-term effects

Edison (1993)

Dominguez and Frankel(1993a, 1993b)

Sarno and Taylor (2001)

OLS Regression of Risk Premium Equation

ρt = θ1Bt + θ2B*t,

where ρt = r – r* + E(st+1) – st,

where ρt is the risk premium, r is domesticinterest rates, r* is foreign interest rates, B islocal currency–denominated bonds, and B* isforeign currency–denominated bonds.

The risk premium, ρt, is measured by deviationsfrom uncovered interest parity, either assumingrational expectations or using survey data.

If ρt is nonzero (in violation of interest rateparity) and systemically responds to the relativesupplies of B and B* (i.e., imperfect substitutabil-ity), then the exchange rate necessarily changesin response to intervention and a change in therelative supplies of B and B*.

• Local and foreign currency bondsupplies, exchange rate, interestrate on a daily basis.

• Daily survey data on exchangerate expectations (if rationalexpectations are not assumed)

• In the absence of data on bondsupplies, intervention may beused as a proxy for the change inrelative bond supplies

• Assuming rational expectations orthe use of survey data has a numberof problems.

• Simultaneity between exchange ratechanges and relative bond supplies(or intervention).



42

Table A4.1 (concluded)

Sources Economic Specification Data Requirements Advantages and Disadvantages

Guimarães (2004)

Kim (2003)

Unified Approaches to Monetary Policy andIntervention (Structural Vector Autoregression)

β(L)γt = ut, where β(L) contains the structuralparameters, and the vector γ2 contains theexchange rate, intervention, and monetarypolicy variables.

The first two equations of the reduced formVAR can be expressed as

st = c1 + α1(L)st–1 + β1(L)It–1+ γ1(L)′Xt + ε1,t

It = c2 + α2(L)st–1 + β2(L)It–1+ γ2(L)′Xt + ε2,t

• Intervention, interest rate, trade-weighted exchange rate index (orbilateral exchange rate), moneysupply, inflation, industrialproduction, commodity prices ona monthly basis

• Accounts for endogeneity ofintervention, exchange rate, andinterest rate movements

• Analyzes short- and long-termimpacts of intervention

• Estimated impact of interventionmight not be robust to theidentification scheme used toidentify structural shocks

Dominguez (1998) GARCH Approaches to Measuring Volatility

Δst = α + βIt + γ1′Xt + εt

ht = β0 + β1 ε2t–1 + β2ht–1 + β3|It|

+ γ2′Xt

• In addition to the exchange rate,intervention, the variablescontained in the vector X, whichwould vary according to the casestudied

• Computationally simple.

• Ignores market expectationsembedded in option prices but notsubject to option pricing models,which have known biases/problems

Dominguez (1998);

Murray, Zelmer, andMcManus (1996)

Implied Volatilities from Option Prices

IVt = β0 + β1|It| + β2′Xt,

where IV is the measure of implied volatilitycalculated from option prices.

• In addition to the above, foreignexchange option prices

• Computationally more demandingand subject to data availability(especially in developing markets).

• Subject to the option pricing modelused.

• Subject to data problems related tooption market (illiquid contracts,etc.).


Andersen, Torben, and Tim Bollerslev, 1998, “Answeringthe Skeptics: Yes, Standard Volatility Models Do Pro-vide Accurate Forecasts,” International EconomicReview, Vol. 39, No. 4, pp. 885–905.

Bank for International Settlements, 1994, Macroeconomicand Monetary Policy Issues Raised by the Growth ofDerivatives Markets (Basel).

———, 2001, Triennial Central Bank Survey of ForeignExchange and Derivatives Market Activity (Basel:Bank for International Settlements).

Bank of Japan, 2000, “Outline of the Bank of Japan’sForeign Exchange Intervention Operations.” Avail-able via the Internet: www.boj.or.jp/en/about/basic/etc/faqkainy.htm.

Baillie, Richard T., and William Osterberg, 1997, “WhyDo Central Banks Intervene?” Journal of Inter-national Money and Finance, Vol. 16, No. 6,pp. 909–19.

Bartolini, Leonardo, 2002, “Foreign Exchange Swaps,” NewEngland Economic Review, Second Quarter, p. 11.

Beattie, Neil, and Jean-Francois Fillion, 1999, “An Intra-day Analysis of the Effectiveness of ForeignExchange Intervention,” Bank of Canada WorkingPaper 99–4 (Ottawa: Bank of Canada).

Beine, Michel, Agnès Bénassy-Quéré, and ChristelleLecourt, 2002, “Central Bank Intervention and For-eign Exchange Rates: New Evidence from FIGARCHEstimations,” Journal of International Money andFinance, Vol. 21, No. 1, pp. 115–144.

Bjorksten, Nils, and Anne-Marie Brook, 2002, “ExchangeRate Strategies for Small Open Developed EconomiesSuch as New Zealand,” Reserve Bank of New ZealandBulletin, Vol. 65 (March).

Bonser-Neal, Catherine, and Glen Tanner, 1996, “CentralBank Intervention and the Volatility of ForeignExchange Rates: Evidence from the Options Market,”Journal of International Money and Finance, Vol. 15,No. 6, pp. 853–78.

Bredin, Don, Stilianos Fountas, and Eithne Murphy, 2002,“An Empirical Analysis of Short-Run and Long-RunIrish Exports: Does Exchange Rate Volatility Mat-ter?” Central Bank of Ireland Research TechnicalPaper 01/RT/02 (Dublin: Central Bank and FinancialServices Authority of Ireland).

Breuer, Peter, 1999, “Central Bank Participation in Cur-rency Options Markets,” IMF Working Paper 99/140(Washington: International Monetary Fund).

Bubula, Andrea, and Inci Otker-Robe, 2002, “The Evolu-tion of Exchange Rate Regimes Since 1990: Evidence

from De Facto Policies,” IMF Working Paper 02/155(Washington: International Monetary Fund).

Calvo, Guillermo A., and Carmen Reinhart, 2002, “Fear ofFloating,” Quarterly Journal of Economics, Vol. 117(May), pp. 379–408.

Canales-Kriljenko, Jorge Iván, 2003, “Foreign ExchangeIntervention in Developing and Transition Econo-mies: Results of a Survey,” IMF Working Paper 03/95(Washington: International Monetary Fund).

———, 2004, “Foreign Exchange Market Organization inSelected Developing and Transition Economies: Evi-dence from a Survey,” IMF Working Paper 04/4(Washington: International Monetary Fund).

———, Roberto Guimarães, and Cem Karacadag, 2003,“Official Intervention in the Foreign ExchangeMarket: Elements of Best Practice,” IMF WorkingPaper 03/152 (Washington: International MonetaryFund).

Carstens, Agustín G., and Alejandro M. Werner, 1999,“Mexico’s Monetary Policy Framework Under a Floating Exchange,” Documento de InvestigaciónNo. 9905 (Mexico City: Bank of Mexico, May).

Central Bank of Turkey, 2001, Annual Report (Ankara).———, 2002, Annual Report (Ankara).———, 2003, “General Framework of the Monetary and

Exchange Rate Policy in 2003,” Press Release 2003–2(Ankara).

Cheung, Yin-Wong, and Menzie Chinn, 1999, “CurrencyTraders and Exchange Rate Dynamics: A Survey ofthe U.S. Market,” Journal of International Money andFinance, Vol. 20, No. 4, pp. 439–71.

Cumby, Robert, and Obstfeld, Maurice, 1983, “CapitalMobility and the Scope for Sterilization: Mexico inthe 1970s,” in Financial Policies and the World Capi-tal Market: The Problem of Latin American Coun-tries, ed. by Pedro Aspe, Rudiger Dornbusch, andMaurice Obstfeld (Chicago: Chicago UniversityPress), pp. 245–69.

Domaç, Ilker, and Alfonso Mendoza, 2002, “Is ThereRoom for Foreign Intervention Under Inflation Tar-geting Framework? Evidence from Mexico andTurkey,” CBT Discussion Paper (Ankara: CentralBank of Turkey, December).

Dominguez, Kathryn M., 1998, “Central Bank Interventionand Exchange Rate Volatility,” Journal of Interna-tional Money and Finance, Vol. 17, No. 1, pp. 161–90.

———, 2003, “The Market Microstructure of CentralBank Intervention,” Journal of International Econom-ics, Vol. 59, No. 1, pp. 25–45.

References

43


www.boj.or.jp/en/about/basic/etc/faqkainy.htm

www.boj.or.jp/en/about/basic/etc/faqkainy.htm

REFERENCES

———, and Jeffrey Frankel, 1993a, “Does ForeignExchange Intervention Matter? The Portfolio Effect”American Economic Review, Vol. 83, No. 5,pp. 1356–69.

———, 1993b, Does Foreign Exchange Intervention Work?(Washington: Institute for International Economics).

Edison, Hali J., 1993, “The Effectiveness of Central-BankIntervention: A Survey of the Literature After 1982,”Special Papers in International Economics, No. 18(Princeton, New Jersey: Princeton University).

———, Paul Cashin, and Hong Liang, 2003, “ForeignExchange Intervention and the Australian Dollar: HasIt Mattered?” IMF Working Paper 03/99 (Washing-ton: International Monetary Fund).

Enoch, Charles, 1998, “Transparency in Central BankOperations in the Foreign Exchange Market,” IMFPaper on Policy Analysis and Assessment 98/2(Washington: International Monetary Fund).

European Central Bank, 2000, “The Single Monetary Pol-icy in Stage Three: General Documentation onEurosystem Monetary Policy Instruments and Proce-dures” (Frankfurt).

Evans, Martin D., and Richard Lyons, 2002, “Order Flowand Exchange Rate Dynamics,” Journal of PoliticalEconomy, Vol. 110 (February), pp. 170–80.

––––––, 2005, “Are Different-Currency Assets ImperfectSubstitutes?” in Exchange Rate Economics: WhereDo We Stand? ed. by Paul De Grauwe (Cambridge,Massachusetts: MIT Press), pp. 1–38.

Fatum, Rasmus, 2000, “On the Effectiveness of SterilizedForeign Exchange Intervention,” ECB Working PaperNo. 10 (Frankfurt: European Central Bank).

———, and Michael M. Hutchison, 2003a, “Effectivenessof Official Daily Foreign Exchange Market Inter-vention Operations in Japan,” NBER Working Paper No. 9648 (Cambridge, Massachusetts: NationalBureau of Economic Research).

———, 2003b, “Is Sterilised Foreign Exchange Interven-tion Effective After All? An Event Study Approach,”Economic Journal, Vol. 113 (April), pp. 390–411.

Financial Markets Association, 2002, “The Model Code:The International Code of Conduct and Practice forthe Financial Markets” (October) (Paris). Availablevia the Internet: www.aciforex.com/.

Frenkel, Michael, Christian Pierdzioch, and Georg Stadt-mann, 2001, “The Foreign Exchange Interventions ofthe European Central Bank,” Banca Nazionale DelLavoro Quarterly Review, Vol. 54, No. 218, pp. 249–87.

Galán Medina, M., J. Duclaud González de Castillo, andA. Garcia Tames, 1997, “A Strategy for Accumu-lating Reserves Through Options to Sell Dollars,”Bank of Mexico, Mexico City. Available via theInternet: www.banxico.org.mx/siteBanxicoINGLES/bPoliticaMonetaria/FSpoliticaMonetaria.html.

Guimarães, Roberto, 2004, “Foreign Exchange Inter-vention and Monetary Policy in Japan: Evidence from Identified VARs,” Money, Macro, and FinanceResearch Conference, Cass Business School, London.Available via the Internet: www.cass.city.ac.uk/conferences/mmf2004/day1.html.

Hausmann, Ricardo, Ugo Panizza, and Ernesto Stein,2001, “Why Do Countries Float the Way They Float?”

Journal of Development Economics, Vol. 66, No. 2,pp. 387–414.

Ho, Corrinne, and Robert N. McCauley, 2003, “Livingwith Flexible Exchange Rates: Issues and RecentExperience in Inflation Targeting Emerging MarketEconomies,” BIS Working Paper No. 130 (Basel:Bank for International Settlements).

Hooyman, Catharina J., 1994, “The Use of ForeignExchange Swaps by Central Banks,” Staff Papers,International Monetary Fund, Vol. 41, No. 1,pp. 149–62.

Hung, Juann H., 1997, “Intervention Strategies andExchange Rate Volatility: A Noise Trading Perspec-tive,” Journal of International Money and Finance,Vol. 16, No. 5, pp. 779–93.

International Monetary Fund, 1998, “Reserve Impact ofForward Foreign Exchange Market Intervention,”International Capital Markets Report (Washington:International Monetary Fund).

––––––, 2002a, Annual Report on Exchange Arrangementand Exchange Restrictions (Washington: Interna-tional Monetary Fund).

––––––, 2002b, “Principles of Fund Surveillance Mem-bers’ Exchange Rate Policies,” in Selected Decisionsand Selected Documents of the International Mone-tary Fund, Twenty-Sixth Issue. Available via theInternet: www.imf.org/external/pubs/ft/sd/index.asp?decision=5392-(77/63).

Ishii, Shogo, Karl Habermeier, John Leimone, Inci Otker-Robe, and Jorge Iván Canales-Kriljenko, 2003,Exchange Arrangements and Foreign Exchange Mar-kets: Developments and Issues, IMF World Economicand Financial Survey Series (Washington: Interna-tional Monetary Fund).

Ito, Takatoshi, 2002, “Is Foreign Exchange InterventionEffective? The Japanese Experiences in the 1990s,”NBER Working Paper No. 8914 (Cambridge, Massa-chusetts: National Bureau of Economic Research).

Jones, Michael, 1984, “Optimal Foreign Exchange Inter-vention: Evidence from the Bretton Woods Era,”Review of Economics and Statistics, Vol. 66, No. 2,pp. 242–55.

Jorion, Phillipe, 1996, “Risk and Turnover in the ForeignExchange Market,” in The Microstructure of ForeignExchange Markets, ed. by Jeffrey Frankel, GiampaoloGalli, and Alberto Giovannini (Chicago: University ofChicago Press).

Killeen, William, Richard Lyons, Michael Moore, 2001,“Fixed Versus Flexible: Lessons from EMS OrderFlow,” NBER Working Paper No. 8491 (Cambridge,Massachusetts: National Bureau of EconomicResearch).

Kim, Soyoung, 2003, “Monetary Policy, ForeignExchange Intervention, and the Exchange Rate in aUnifying Framework,” Journal of International Eco-nomics, Vol. 60, No. 2, pp. 355–86.

Lyons, Richard, 2001, The Microstructure Approach toExchange Rates (Cambridge, Massachusetts: MITPress).

Mandeng, Ousmene-Jacques, 2003, “Central Bank For-eign Exchange Market Intervention and Option Con-tract Specification: The Case of Colombia,” IMF

44


www.imf.org/external/pubs/ft/sd/index.asp?decision=5392-(77/63)

www.imf.org/external/pubs/ft/sd/index.asp?decision=5392-(77/63)

www.aciforex.com/

www.banxico.org.mx/siteBanxicoINGLES/bPoliticaMonetaria/FSpoliticaMonetaria.html

www.banxico.org.mx/siteBanxicoINGLES/bPoliticaMonetaria/FSpoliticaMonetaria.html

www.cass.city.ac.uk/conferences/mmf2004/day1.html

www.cass.city.ac.uk/conferences/mmf2004/day1.html

References

Working Paper 03/135 (Washington: InternationalMonetary Fund).

Mark, Nelson, 2001, International Macroeconomics andFinance: Theory and Econometric Methods (NewYork: Blackwell Publishers).

Moreno, Ramon, 1997, “Lessons from Thailand,” FRBSFEconomic Letter, No. 97–33 (San Francisco: FederalReserve Bank of San Francisco).

Mulder, Christian, and Roberto Perrelli, 2001, “ForeignCurrency Credit Ratings for Emerging MarketEconomies,” IMF Working Paper 01/191 (Washing-ton: International Monetary Fund).

Murray, John, Mark Zelmer, and Des McManus, 1996,“The Effect of Intervention on Canadian DollarVolatility,” in Exchange Rates and Monetary Policy,proceedings of a conference held by the Bank ofCanada, October.

Mussa, Michael, 1981, “The Role of Official Interven-tion,” Group of Thirty Occasional Paper No. 6 (Wash-ington: Group of Thirty).

––––––, and Miguel A. Savastano, 1999, “The IMF Ap-proach to Economic Stabilization,” IMF WorkingPaper 99/104 (Washington: International MonetaryFund).

Neely, Christopher, 2001, “The Practice of Central BankIntervention: Looking Under the Hood,” Federal ReserveBank of St. Louis Review, Vol. 83, No. 3, pp. 1–10.

Nelson, Daniel, 1991, “Conditional Heteroskedasticity in Asset Returns: A New Approach,” Econometrica,Vol. 59, No. 2, pp. 347–70.

Obstfeld, Maurice, 1990, “The Effectiveness of Foreign-Exchange Intervention: Recent Experience,1985–1988,” in International Policy Coordinationand Exchange Rate Fluctuations, ed. by WilliamBranson, Jacob Frenkel, and Morris Goldstein(Chicago: University of Chicago Press),pp. 197–237.

Popper, Helen, and John Montgomery, 2001, “InformationSharing and Central Bank Intervention in the ForeignExchange Market,” Journal of International Econom-ics, Vol. 55, No. 2, pp. 295–316.

Ramaswamy, Ramana, and Hossein Samiei, 2000, “TheYen-Dollar Rate—Have Interventions Mattered?”IMF Working Paper 00/95 (Washington: InternationalMonetary Fund).

Reinhart, Carmen, and Kenneth Rogoff, 2003, “Evolutionand Performance of Exchange Rate Regimes” (unpub-lished; Washington: International Monetary Fund).

Reserve Bank of India, 2002, “Inclusion of Euro as an Additional Intervention Currency,” Press Release

No. 986/2001–2002 dated 04.03.2002, based onNotification S.O.235(E) dated February 27, 2002.Available via the Internet: www.gujaratchamber.org/notifications-rbi1.htm.

Rogoff, Kenneth, 1999, “Perspectives on Exchange RateVolatility,” in International Capital Flows, ed. byMartin Feldstein (Chicago: University of ChicagoPress), pp. 441–53.

Roper, Don E., and Stephen J. Turnovsky, 1980, “OptimalExchange Market Intervention in a Simple StochasticMacro Model,” Canadian Journal of Economics,Vol. 13, No. 2, pp. 296–309.

Sarno, Lucio, and Mark P. Taylor, 2001, “Official Interven-tion in the Foreign Exchange Market: Is It Effective,and If So, How Does It Work?” Journal of EconomicLiterature, Vol. 39 (September), pp. 839–68.

———, 2002, The Economics of Exchange Rates (Cam-bridge: Cambridge University Press).

Sveriges Riksbank, 1999, “Currency Market PrimaryDealer Agreement” (unpublished; Stockholm).

Tanner, Evan, 1998, “Deviation from Uncovered InterestRate Parity: A Global Guide to Where the Action Is,”IMF Working Paper 98/117 (Washington: Interna-tional Monetary Fund).

Tapia, Matías, and Andrea Tokman, 2004, “Effects ofForeign Exchange Intervention Under Public Infor-mation: The Chilean Case,” Central Bank of ChileWorking Papers No. 255 (Santiago: Central Bank ofChile).

Taylor, Dean, 1982, “Official Intervention in the ForeignExchange Market, or Bet Against the Central Bank,”Journal of Political Economy, Vol. 90 (April),pp. 256–68.

Vitale, Paolo, 2001, “Foreign Exchange Intervention, Pol-icy Objectives, and Macroeconomic Stability,” CEPRDiscussion Paper No. 2886 (London: Centre for Eco-nomic Policy Research).

Werner, Alejandro, 1997, “El Efecto Sobre el Tipo deCambio y las Tasas de Interes de las Intervenciones enel Mercado Cambiario y del Proceso de Esteril-ización,” Documento de Investigación No. 9706(Mexico City: Bank of Mexico).

––––––, and Alexis Milo, 1998, “Acumulación de Reser-vas Internacionales a Través de la Venta de Opciones:El Caso de México,” Documento de Investigación No. 9801 (Mexico City: Bank of Mexico).

Zapatero, Fernando, and Luis F. Reverter, 2003, “ForeignExchange Intervention with Options,” Journalof International Money and Finance, Vol. 22, No. 2,pp. 289–306.

45


www.gujaratchamber.org/notifications-rbi1.htm

www.gujaratchamber.org/notifications-rbi1.htm

OCCASIONAL PAPERS

Recent Occasional Papers of the International Monetary Fund

249. Official Foreign Exchange Intervention, by Shogo Ishi, Jorge Iván Canales-Kriljenko, RobertoGuimarães, and Cem Karacadag. 2006

248. Labor Market Performance in Transition: The Experience of Central and Eastern European Countries, byJerald Schiff, Philippe Egoumé-Bossogo, Miho Ihara, Tetsuya Konuki, and Kornélia Krajnyák. 2006.

247. Rebuilding Fiscal Institutions in Post-Conflict Countries, by Sanjeev Gupta, Shamsuddin Tareq, BenedictClements, Alex Segura-Ubiergo, Rina Bhattacharya, and Todd Mattina. 2005.

246. Experience with Large Fiscal Adjustments, by George C. Tsibouris, Mark A. Horton, Mark Flanagan, andWojciech S. Maliszewski. 2005.

245. Budget System Reform in Emerging Economies: The Challenges and the Reform Agenda, by JackDiamond. 2005.

244. Monetary Policy Implementation at Different Stages of Market Development, by a staff team led byBernard J. Laurens. 2005.

243. Central America: Global Integration and Regional Cooperation, edited by Markus Rodlauer and AlfredSchipke. 2005.

242. Turkey at the Crossroads: From Crisis Resolution to EU Accession, by a staff team led by RezaMoghadam. 2005.

241. The Design of IMF-Supported Programs, by Atish Ghosh, Charis Christofides, Jun Kim, Laura Papi, UmaRamakrishnan, Alun Thomas, and Juan Zalduendo. 2005.

240. Debt-Related Vulnerabilities and Financial Crises: An Application of the Balance Sheet Approach toEmerging Market Countries, by Christoph Rosenberg, Ioannis Halikias, Brett House, Christian Keller,Jens Nystedt, Alexander Pitt, and Brad Setser. 2005.

239. GEM: A New International Macroeconomic Model, by Tamim Bayoumi, with assistance from DouglasLaxton, Hamid Faruqee, Benjamin Hunt, Philippe Karam, Jaewoo Lee, Alessandro Rebucci, and IvanTchakarov. 2004.

238. Stabilization and Reforms in Latin America: A Macroeconomic Perspective on the Experience Since theEarly 1990s, by Anoop Singh, Agnès Belaisch, Charles Collyns, Paula De Masi, Reva Krieger, GuyMeredith, and Robert Rennhack. 2005.

237. Sovereign Debt Structure for Crisis Prevention, by Eduardo Borensztein, Marcos Chamon, Olivier Jeanne,Paolo Mauro, and Jeromin Zettelmeyer. 2004.

236. Lessons from the Crisis in Argentina, by Christina Daseking, Atish R. Ghosh, Alun Thomas, and Timo-thy Lane. 2004.

235. A New Look at Exchange Rate Volatility and Trade Flows, by Peter B. Clark, Natalia Tamirisa, andShang-Jin Wei, with Azim Sadikov and Li Zeng. 2004.

234. Adopting the Euro in Central Europe: Challenges of the Next Step in European Integration, by Susan M.Schadler, Paulo F. Drummond, Louis Kuijs, Zuzana Murgasova, and Rachel N. van Elkan. 2004.

233. Germany’s Three-Pillar Banking System: Cross-Country Perspectives in Europe, by Allan Brunner, JörgDecressin, Daniel Hardy, and Beata Kudela. 2004.

232. China’s Growth and Integration into the World Economy: Prospects and Challenges, edited by EswarPrasad. 2004.

231. Chile: Policies and Institutions Underpinning Stability and Growth, by Eliot Kalter, Steven Phillips,Marco A. Espinosa-Vega, Rodolfo Luzio, Mauricio Villafuerte, and Manmohan Singh. 2004.

230. Financial Stability in Dollarized Countries, by Anne-Marie Gulde, David Hoelscher, Alain Ize, DavidMarston, and Gianni De Nicoló. 2004.

229. Evolution and Performance of Exchange Rate Regimes, by Kenneth S. Rogoff, Aasim M. Husain, AshokaMody, Robin Brooks, and Nienke Oomes. 2004.

228. Capital Markets and Financial Intermediation in The Baltics, by Alfred Schipke, Christian Beddies, SusanM. George, and Niamh Sheridan. 2004.

46


Occasional Papers

227. U.S. Fiscal Policies and Priorities for Long-Run Sustainability, edited by Martin Mühleisen and Christo-pher Towe 2004.

226. Hong Kong SAR: Meeting the Challenges of Integration with the Mainland, edited by Eswar Prasad, withcontributions from Jorge Chan-Lau, Dora Iakova, William Lee, Hong Liang, Ida Liu, Papa N’Diaye,and Tao Wang. 2004.

225. Rules-Based Fiscal Policy in France, Germany, Italy, and Spain, by Teresa Dában, Enrica Detragiache,Gabriel di Bella, Gian Maria Milesi-Ferretti, and Steven Symansky. 2003.

224. Managing Systemic Banking Crises, by a staff team led by David S. Hoelscher and Marc Quintyn. 2003.

223. Monetary Union Among Member Countries of the Gulf Cooperation Council, by a staff team led by UgoFasano. 2003.

222. Informal Funds Transfer Systems: An Analysis of the Informal Hawala System, by Mohammed El Qorchi,Samuel Munzele Maimbo, and John F. Wilson. 2003.

221. Deflation: Determinants, Risks, and Policy Options, by Manmohan S. Kumar. 2003.

220. Effects of Financial Globalization on Developing Countries: Some Empirical Evidence, by Eswar S.Prasad, Kenneth Rogoff, Shang-Jin Wei, and Ayhan Kose. 2003.

219. Economic Policy in a Highly Dollarized Economy: The Case of Cambodia, by Mario de Zamaroczy andSopanha Sa. 2003.

218. Fiscal Vulnerability and Financial Crises in Emerging Market Economies, by Richard Hemming, MichaelKell, and Axel Schimmelpfennig. 2003.

217. Managing Financial Crises: Recent Experience and Lessons for Latin America, edited by Charles Collynsand G. Russell Kincaid. 2003.

216. Is the PRGF Living Up to Expectations?—An Assessment of Program Design, by Sanjeev Gupta, MarkPlant, Benedict Clements, Thomas Dorsey, Emanuele Baldacci, Gabriela Inchauste, Shamsuddin Tareq,and Nita Thacker. 2002.

215. Improving Large Taxpayers’ Compliance: A Review of Country Experience, by Katherine Baer. 2002.

214. Advanced Country Experiences with Capital Account Liberalization, by Age Bakker and Bryan Chapple.2002.

213. The Baltic Countries: Medium-Term Fiscal Issues Related to EU and NATO Accession, by JohannesMueller, Christian Beddies, Robert Burgess, Vitali Kramarenko, and Joannes Mongardini. 2002.

212. Financial Soundness Indicators: Analytical Aspects and Country Practices, by V. Sundararajan, CharlesEnoch, Armida San José, Paul Hilbers, Russell Krueger, Marina Moretti, and Graham Slack. 2002.

211. Capital Account Liberalization and Financial Sector Stability, by a staff team led by Shogo Ishii and KarlHabermeier. 2002.

210. IMF-Supported Programs in Capital Account Crises, by Atish Ghosh, Timothy Lane, Marianne Schulze-Ghattas, Ales Bulír , Javier Hamann, and Alex Mourmouras. 2002.

209. Methodology for Current Account and Exchange Rate Assessments, by Peter Isard, Hamid Faruqee,G. Russell Kincaid, and Martin Fetherston. 2001.

208. Yemen in the 1990s: From Unification to Economic Reform, by Klaus Enders, Sherwyn Williams, NadaChoueiri, Yuri Sobolev, and Jan Walliser. 2001.

207. Malaysia: From Crisis to Recovery, by Kanitta Meesook, Il Houng Lee, Olin Liu, Yougesh Khatri, NataliaTamirisa, Michael Moore, and Mark H. Krysl. 2001.

206. The Dominican Republic: Stabilization, Structural Reform, and Economic Growth, by a staff team led byPhilip Young comprising Alessandro Giustiniani, Werner C. Keller, and Randa E. Sab and others. 2001.

205. Stabilization and Savings Funds for Nonrenewable Resources, by Jeffrey Davis, Rolando Ossowski,James Daniel, and Steven Barnett. 2001.

Note: For information on the titles and availability of Occasional Papers not listed, please consult the IMF’s Publications Catalog or contact IMFPublication Services.

47


Official Foreign Exchange Intervention

Documents

Transcript of Official Foreign Exchange Intervention