UNCTAD Report on Exchange Rate Policy

8/3/2019 UNCTAD Report on Exchange Rate Policy

1/24

G-24 Discussion Paper Series

UNITED NATIONS CONFERENCE ON TRADE AND DEVELOPMENT

UNITED NATIONS

CENTER FOR

INTERNATIONAL

DEVELOPMENT

HARVARD UNIVERSITY

Exchange-rate Policies for Developing

Countries: What Have We Learned?

What Do We Still Not Know?

Andrs Velasco

No. 5, June 2000


2/24

G-24 Discussion Paper Series

Research papers for the Intergovernmental Group of Twenty-Four

on International Monetary Affairs

UNITED NATIONS

New York and Geneva, June 2000

CENTER FOR INTERNATIONAL DEVELOPMENT

HARVARD UNIVERSITY

UNITED NATIONS CONFERENCE ON

TRADE AND DEVELOPMENT


3/24

Note

Symbols of United Nations documents are composed of capitalletters combined with figures. Mention of such a symbol indicates areference to a United Nations document.

*

* *

The views expressed in this Series are those of the authors anddo not necessarily reflect the views of the UNCTAD secretariat. Thedesignations employed and the presentation of the material do notimply the expression of any opinion whatsoever on the part of theSecretariat of the United Nations concerning the legal status of anycountry, territory, city or area, or of its authorities, or concerning thedelimitation of its frontiers or boundaries.

*

* *

Material in this publication may be freely quoted; acknowl-edgement, however, is requested (including reference to the documentnumber). It would be appreciated if a copy of the publicationcontaining the quotation were sent to the Editorial Assistant,Macroeconomic and Development Policies Branch, Division onGlobalization and Development Strategies, UNCTAD, Palais desNations, CH-1211 Geneva 10.

UNCTAD/GDS/MDPB/G24/5

UNITED NATIONS PUBLICATION

Copyright United Nations, 2000All rights reserved


4/24

iiiExchange-rate Policies for Developing Countries: What Have We Learned? What Do We Still Not Know?

PREFACE

The G-24 Discussion Paper Series is a collection of research papers preparedunder the UNCTAD Project of Technical Support to the Intergovernmental Group ofTwenty-Four on International Monetary Affairs (G-24). The G-24 was established in1971 with a view to increasing the analytical capacity and the negotiating strength of thedeveloping countries in discussions and negotiations in the international financialinstitutions. The G-24 is the only formal developing-country grouping within the IMFand the World Bank. Its meetings are open to all developing countries.

The G-24 Project, which is administered by UNCTADs Macroeconomic andDevelopment Policies Branch, aims at enhancing the understanding of policy makers indeveloping countries of the complex issues in the international monetary and financialsystem, and at raising the awareness outside developing countries of the need to introducea development dimension into the discussion of international financial and institutionalreform.

The research carried out under the project is coordinated by Professor Dani Rodrik,John F. Kennedy School of Government, Harvard University. The research papers arediscussed among experts and policy makers at the meetings of the G-24 Technical Group,

and provide inputs to the meetings of the G-24 Ministers and Deputies in their preparationsfor negotiations and discussions in the framework of the IMFs International Monetaryand Financial Committee (formerly Interim Committee) and the Joint IMF/IBRDDevelopment Committee, as well as in other forums. Previously, the research papers forthe G-24 were published by UNCTAD in the collection International Monetary andFinancial Issues for the 1990s. Between 1992 and 1999 more than 80 papers werepublished in 11 volumes of this collection, covering a wide range of monetary and financialissues of major interest to developing countries. Since the beginning of 2000 the studiesare published jointly by UNCTAD and the Center for International Development atHarvard University in the G-24 Discussion Paper Series.

The Project of Technical Support to the G-24 receives generous financial supportfrom the International Development Research Centre of Canada and the Governments ofDenmark and the Netherlands, as well as contributions from the countries participatingin the meetings of the G-24.


5/24

EXCHANGE-RATE POLICIES FOR DEVELOPING

COUNTRIES: WHAT HAVE WE LEARNED?

WHAT DO WE STILL NOT KNOW?

Andrs Velasco

New York University,

University of Chile and

National Bureau of Economic Research

G-24 Discussion Paper No. 5

June 2000


6/24

viiExchange-rate Policies for Developing Countries: What Have We Learned? What Do We Still Not Know?

Abstract

The 19971998 Asian crisis, with its offshoots in Eastern Europe and Latin America, has

reignited the debate about appropriate exchange-rate policies for developing countries. One

widely shared conclusion from this episode is that adjustable or crawling pegs are extremely

fragile in a world of volatile capital movements. The pressure resulting from massive capital

flow reversals and weakened domestic financial systems was too strong even for countries that

followed sound macroeconomic policies and had large stocks of reserves. As a consequence, the

polar regimes of a hard pegs (such as a currency board), or a clean float, are enjoying new

popularity.

This paper argues that, while currency boards or even dollarization may be justified in some

extreme cases, they are not appropriate for all developing countries. The recommendations

formulated on the basis of the Mundell-McKinnon criteria for the optimum currency are

considered still sensible today. Currency boards face serious implementation problems. One is

the choice of the currency to peg to and at what rate; another is the need to ensure stability of

the domestic financial system in the absence of a domestic lender of last resort.

Floating appears to have wider applicability. As Friedman already argued in the early 1950s,

if prices move slowly, it is both faster and less costly to move the nominal exchange rate in

response to a shock that requires an adjustment in the real exchange rate. But for exchange-rate

flexibility to be stabilizing, it has to be implemented by independent central banks whose

commitment to low inflation is credible. Ongoing depreciations that follow from imprudent ofopportunistic monetary behaviour will surely come to be expected by agents, and hence will

have no real effect; occasional depreciations that respond exclusively to unforecastable shocks

will, almost by definition, have real effects. But floating also faces questions of implementation.

Given that no central bank completely abstains from intervention in currency markets, what

principles should govern such intervention? The paper elaborates on a number of points in this

regard on which recent experience is likely to be instructive, but on which more research is

needed.

Finally, any exchange-rate regime, and especially one of flexible rates, requires comple-

mentary policies to increase its chances of success. In this context, some have suggested the use

of capital controls; less controversial is the need for prudential regulation of the financial

system and for counter-cyclical fiscal policy.


7/24

ixExchange-rate Policies for Developing Countries: What Have We Learned? What Do We Still Not Know?

Table of contents

Page

Preface ............................................................................................................................................ iii

Abstract ........................................................................................................................................... vii

I. The new conventional wisdom ................................................................................................ 1II. Hard pegs: advantages, prerequisites and pitfalls ................................................................ 3

A. The credibility argument ...................................................................................................... 3B. The discipline argument....................................................................................................... 3C. Prerequisites for adoption .................................................................................................... 4D. Pegging to the right currency ............................................................................................... 5E. Combining exchange rate and financial stability................................................................. 5

III. Can exchange-rate flexibility help? ........................................................................................ 6

A. The basic case for flexibility ............................................................................................... 6B. Credibility versus flexibility ................................................................................................ 7

C. Politics and policy-making .................................................................................................. 8D. Exchange-rate flexibility and financial stability .................................................................. 9

IV. Making exchange-rate flexibility work in practice ............................................................. 10

A. Nominal anchors and inflation targets ............................................................................... 10B. Dealing with short-term exchange-rate fluctuations.......................................................... 11C. Dealing with long swings in the exchange rate ................................................................. 11D. Crafting monetary policy ................................................................................................... 12

V. Complementary policies: financial regulation, capital controls and fiscal institutions ........ 13

A. Financial liberalization and fragility .................................................................................. 13B. Capital inflows and short-term debt................................................................................... 14

C. Improving fiscal institutions .............................................................................................. 14Notes ........................................................................................................................................... 14

References ........................................................................................................................................... 15


8/24

1Exchange-rate Policies for Developing Countries: What Have We Learned? What Do We Still Not Know?

I. The new conventional wisdom

The 19971998 Asian crisis, with its offshootsin Eastern Europe and South America, revealed howlittle we still know about workable exchange-ratepolicies for developing countries. Arrangements thathad performed relatively well for years (think ofIndonesia and the Republic of Korea) came crashing

down with almost no advance notice; other arrange-ments that once seemed invulnerable (think of HongKongs currency board) almost tumbled down as well.Mid-course corrections and policy changes provedequally troublesome: in every country that abandoneda peg and floated (Brazil, Ecuador, Russian Federa-tion and Thailand, and again Indonesia and Republicof Korea) the exchange rate overshot massively, anda period of currency turmoil followed. And all of it,of course, with tremendous real costs: both the highinterest rates used to defend pegs and the massivedepreciations that followed abandonment playedhavoc with corporate balance sheet and wrecked largechunks of the domestic financial system.

But in spite of the confusion, pundits have notbeen shy about drawing conclusions. Past financialand currency crises bred new bits of conventionalwisdom, many of which were discarded when thenext crash hit; this latest meltdown is no exception.With analysts scrambled to extract a new set of policy

lessons, no tenet of conventional wisdom is morepervasive than the law of the excluded middle:there is apparently no intermediate exchange-rateregime suitable for developing countries. Currencyboards or free floating are, allegedly, the only options.

The reasoning behind this fashionable conclu-sion is simple. Adjustable or crawling pegs were inplace in almost every country that recently experi-

enced serious difficulties: Brazil, Ecuador, Indonesia,Republic of Korea, Russian Federation and Thailand.The pressure brought by massive capital flow revers-als and weakened domestic financial systems wastoo much to bear, even for countries that followedreasonably sound macro policies and had seeminglyplentiful reserves.

If lack of credibility and the resulting endemi-cally high interest rates was one of the factors thatbrought these pegs down, the logic goes, then theanswer is to ensure credibility at any expense: hardpegs such as a currency board or even full aban-donment of the domestic currency should helpconvince sceptics. After all, one cannot easily de-value a currency that does not exist, or one whoseexchange rate is set by law. Or if the conditions forsuch radical fixing are not present, one should go tothe other extreme and let the currency value fluctu-ate freely. The other way to ensure credibility is notto make any promises about the exchange rate at all.

EXCHANGE-RATE POLICIES FOR DEVELOPINGCOUNTRIES: WHAT HAVE WE LEARNED?

WHAT DO WE STILL NOT KNOW?*

* Parts of this document draw on Larrain and Velasco (1999).

Andrs Velasco


9/24

2 G-24 Discussion Paper Series, No. 5

As do most maxims of conventional wisdom,this one has a good deal of truth in it. Revocable pegs,whether of the crawling, adjusting or constant vari-ety, appear indefensible in a world of high and volatilecapital mobility. If this was true for rich countrieswith large reserves (Europe in 19911992), it is evenmore true for middle-income, reserve-constrained,developing countries.

But the new exchange rate orthodoxy alsoleaves a great deal to be desired. Its empirical foun-dations, for one, are weak. A good deal of the currententhusiasm for currency boards owes to the experi-ence of one country, Argentina, over a fairly briefperiod of time. All the other experiences, exceptfor Hong Kongs, have been too short-lived to beinformative.1 The endorsement of free floating simi-larly glosses over the fact that there are no central

banks in the world that completely abstain from in-tervention in the currency market. When assessingempirically the virtues of floating, therefore, one hasto look at mixed regimes that in many ways are nottoo different from systems of wide intervention bandsor periodically adjustable pegs. Chile, Colombia,Mexico and Peru are recent examples of this in LatinAmerica. And, overall, the jury is still out as to whichsystem hard pegs or floating performs better attimes of trouble. Early in the recent episode evidenceseemed to favour the Argentine/Hong Kong model:

a period of high interest rates seemed like a smallprice to pay to avoid the turmoil affecting countriesthat had let the exchange rate go. But both hard-pegcountries are today mired in major recessions, whilesome of the early devaluers (Mexico, Republic ofKorea and Thailand) seem to be back on the growthtrack. The enthusiasm for currency boards has di-minished accordingly.

The second shortcoming of the new orthodoxyis that it leaves it quite unclear which countries should

adopt which polar system. Once upon a time econo-mists familiar with the Mundell-McKinnon criteriafor optimum currency areas confidently recom-mended fixed exchange rates to small economieswide open to international trade (Mundell, 1961).Large economies, or small economies subjected toshocks uncorrelated to those buffeting the countryto whose currency they might have pegged, wereadvised to choose flexible rates. This prescription isnot antediluvian it was contained, for instance, ina special chapter on the subject in the 1997 IMFWorld Economic Outlook. But in the midst of theirrespective crises there was no shortage of punditsadvising Brazil and the Russian Federation (not ex-actly small countries) to adopt currency boards, as if

short-term credibility considerations should neces-sarily take precedence over all other considerations.That may well be so, but the abdication of monetaryindependence should be chosen after a careful ex-amination of pros and cons, not as a last-ditch effortto arrest economic collapse.

And there is, finally, the pesky problem of im-plementation. One question is not whether to floatfreely, but what kind of dirty float to have. Shouldthere be a monitoring band, as Williamson (1998)has suggested and some countries seem to employ inpractice? Should monetary policy react systemati-cally (either via aggregates or interest rates) tomovements in nominal or real exchange rates? Is aninflation target the best way to endow flexible sys-tems with a nominal anchor?

Currency boards also face serious implementa-tion problems of their own. Start with the choice ofwhat currency to peg to and at what rate. Pegging tothe wrong anchor in a world of great volatility in thecross-rates among the three major currencies can bedevastating, as the countries of South-East Asia re-cently discovered. And how to guarantee the stabilityof the domestic financial system in the absence of adomestic lender of last resort? A foreign alternativepresumably has to be found.

In this paper I review these and related issues.The goal is to highlight areas where more research isnecessary, both to clarify our understanding of com-plex problems and to help guide sound policy-makingin developing countries. Throughout, the emphasisis on designing exchange-rate systems for middle-income developing countries with reasonably modernfinancial systems and relatively high degrees of inte-gration into world capital markets that is, what istoday fashionably known as emerging markets. Low-income countries face quite a different set of issues.

Section II below focuses on the costs and ben-efits of currency boards, and tries to identify therelatively stringent conditions under which it is pru-dent to adopt such an arrangement. Section III triesto answer the question of whether regimes with sub-stantial exchange-rate flexibility can be effectivecounter-cyclical stabilizers in developing countries.Section IV studies how to make flexibility work inpractice, with special attention to inflation targetsand alternative monetary policy rules. Section V fo-cuses on useful complementary policies: prudentialbanking regulation, taxes on short-term capital move-ments, and reforms to the institutions that governfiscal policy.


10/24


II. Hard pegs: advantages,

prerequisites and pitfalls

There is no doubt that in the aftermath of theAsian crisis, hard pegs (especially currency boards)are becoming increasingly popular. After reviewing

some of the theoretical arguments behind this popu-larity, I ask two sets of questions. What kind ofcountry is best served by adopting a hard peg? Andwhat pitfalls should the adopting country strive toavoid?

A. The credibility argument

The main argument in favour of hard pegs restson the need to make monetary policy credible. If you

cannot build credibility for monetary policy at home,then you can presumably import it by fixing the valueof your currency to a hard-money country. This iswhat Club-Med countries attempted by pegging tothe deutsche mark, and what Argentina has tried withthe United States dollar. Many theoretical and prac-tical objections to the argument are well known.Where the political costs of abandoning a peg comefrom and whether they are large enough to preventunpleasant surprises is less than clear. Many an ir-reversible peg has come undone; the problems of

the European Monetary System in the early 1990sare but one example. Yet it also seems clear that ifthe political will is sufficient and if the institutionsdesigned to express that will are robust enough, in-terest rate spreads and other indicators of the publicsscepticism can come down sharply and stay there.Europe in the run-up to the Economic and MonetaryUnion is a good example.

The strength (and also the potential weakness)of hard pegs lies in the absence of escape clauses. A

fixed exchange rate may be thought of as an implicitcontract in which the Central Bank commits itself toretaining the peg unless one or more of several un-specified but painful factors kick in. If they do,devaluation need not be punished by a loss of cred-ibility, for in devaluing the authorities have adheredto the implicit contract. When the short-term pain ofdefending the peg is large enough to outweigh thelong-term benefits of retaining the fixed rates regime,the country could exercise an escape clause or en-gage in excusable devaluation.

Whether this is a plausible view of the worldhinges on difficult implementation problems. It is notclear whether there are excusable devaluations in

developing countries, just as there may not be or-derly devaluations either. This is probably becausethe exogenous shocks that could render them so arenot fully observable or perhaps not even fully ex-ogenous, in the sense that governments could try tomanipulate economic variables to justify an aban-donment of the peg. When in doubt, a weary publicmay justifiably choose to be sceptical.2

Obstfeld (1997) has raised an additional andcrucial argument against escape clauses in fixed ex-change rates: they can open the door to multipleequilibria. The government is allowed to devalue ifthe situation gets too nasty. But the expectation thatthe government might devalue could lead the privatesector to take actions (for example, by demandinglarge wage increases and high nominal interest rates)that could make the situation unpleasant to begin

with. If the government does not devalue, it has tolive with costly high real wages and real interest rates.But if it gives in, we have a self-fulfilling prophecysetting in: devaluation takes place exclusively be-cause agents expected it. This means that a govern-ment should think long and hard before hinting thatit views devaluation in some circumstances as ex-cusable. Equivalently, governments should adopthard pegs that make devaluation unthinkable.

B. The discipline argument

The other important reason that leads many toadvocate hard pegs is their alleged ability to inducediscipline whether fiscal or monetary. This argumentis a close cousin of the credibility story. Presumably,fixed rates induce more discipline because adoptinglax fiscal policies must eventually lead to an exhaus-tion of reserves and an end to the peg. Presumably,the eventual collapse of the fixed exchange rate

would imply a big political cost for the policy maker that is to say, bad behaviour today would lead to apunishment tomorrow. Fear of suffering this punish-ment leads the policy maker to be disciplined. If thedeterrent is strong enough, then unsustainable fiscalpolicies do not occur in equilibrium.

But, as Tornell and Velasco (1998, 2000) haveargued, the conventional wisdom fails to understandthat under flexible rates imprudent behaviour especially fiscal laxity has costs as well. The dif-ference with fixed rates is in the intertemporaldistribution of these costs. Under fixed rates unsoundpolicies manifest themselves in falling reserves orexploding debts. Only when the situation becomes


11/24


unsustainable do the costs begin to bite. Flexiblerates, by contrast, allow the effects of unsound fiscalpolicies to manifest themselves immediately throughmovements in the exchange rate and the price level.All of this means (as Tornell and Velasco 1998 and2000 show formally) that if inflation is costly for thefiscal authorities, and these discount the future heav-ily, then flexible rates, by forcing the costs ofmisbehaviour to be paid up-front, can provide morefiscal discipline.

Some empirical evidence supports this revision-ist view. Tornell and Velasco (1998) and Gavin andPerotti (1997) show that in Latin America fiscal poli-cies have been more prudent after controlling for ahost of factors under flexible than under fixed rates.Those were mostly soft pegs. Would hard pegsperform any differently? The evidence in this regard

is limited. Tornell and Velasco (2000) study the caseof sub-Saharan Africa, comparing the experience ofFrancophone countries that have pegged to the Frenchfranc versus the rest. Since pegs in the CFA zone arean artifact of colonial rule, they are supported by aFrench commitment to intervene and currency rateshave been changed only once since 1948; they couldconceivably be thought of as hard. The bad newsis that Francophone African countries operating un-der that regime seem, after controlling for a host offactors, to have exhibited less fiscal discipline de-

fined as average deficits than their Anglophonecounterparts.

The recent experience in Latin America is alsoambiguous. The fiscal performance of Argentina andPanama has not been outstanding, but in the case ofArgentina it represents a vast improvement fromthe hyper inflation-producing deficits of the 1980s.Would free-spending Brazilian congressmen havebehaved more prudently in 19971998 had theircountry been on a currency board? Some scepticism

is surely in order.

C. Prerequisites for adoption

Hard pegs therefore seem to have some impor-tant (though not unambiguous) advantages. But acurrency board or full dollarization are not for eve-ryone. A short list of conditions should to include:3

Optimal currency areas criteria must be satis-fied. This means, among other things, that largecountries are worse candidates than small coun-tries, and that pegging to a country subject to

very asymmetric real shocks is likely to proveproblematic.

Also along Mundell-McKinnon lines, the bulkof the adopting countrys trade takes place withthe country or countries to whose currencies itplans to peg. This means that, ceteris paribus,Mexico or Central America are much bettercandidates for dollarization than Argentina,Brazil or Chile. More on this below.

The adopting country must have preferencesabout inflation that are broadly similar to thoseof the country to which it plans to peg. Thismay be easily achieved in countries with a his-tory of high inflation, which now want pricestability at all costs (e.g. Argentina). It mayprove trickier in countries which have never

experienced a full-blown hyperinflation, andwhere the polis is less unanimous in its will-ingness to take pains to ensure stable prices(e.g. Brazil, Ecuador, Venezuela).

Flexible labour markets become essential: withthe exchange rate fixed, nominal wages andprices must adjust, however slowly, in responseto an adverse shock. Countries considering ahard peg are well advised to undertake labourreforms first. The argument is sometimes made

(especially in Europe) that the very presence ofa hard peg will create the political impetus forlabour market deregulation. That may well beso, but it seems like a very risky gamble to take,especially for countries with political systemsmore unwieldy than Europes.

Strong, well-capitalized and well-regulatedbanks are also essential, since a hard peg pre-vents the local central bank from serving as alender of last resort to domestic banks. More

on this below.

Hard pegs are most necessary for countries withweak central banks and chaotic fiscal institu-tions. But making hard pegs work requireshigh-quality institutions, and the rule of lawmatters in ways that are seldom discussed. Acurrency board for instance, is a commitmentto adhere to a set of very strict rules governingmonetary policy. It may also involve putting theexchange rate into the law, as Argentina hasdone. These arrangements only make sense incountries where governments adhere to theirown rules and where laws cannot be changedby fiat.


12/24


D. Pegging to the right currency

A key implementation problem is that, in aworld of floating rates, pegging to one currencymeans floating vis--vis most others. This is not aproblem for countries whose trade is geographically

very concentrated and which peg to the currency ofa large trading partner. But otherwise cross-rate fluc-tuations can do serious damage, as East Asianeconomies whose currencies were pegged to the dol-lar discovered in 1997. The sharp appreciation of thedollar vis--vis the yen caused substantial apprecia-tion in the real effective exchange rates of severalEast Asian countries, helping pave the way for thecrisis that followed (Corsetti et al., 1998). Of course,part of the problem followed from the fact that thesecountries pegged de facto or de jure to the dollar,

while their trade was quite diversified.

One way out is to peg not to a single currencybut to a basket. In principle, at least, this could helpinsulate countries from cross-rate instability. But theproblems of implementation are many and difficult.Under a currency board the weights used to calcu-late the basket would have to be public information;this is not the way in which banks have traditionallypreferred to manage such baskets. There is also theneed to change the weights in response to structuralchange. Who is to do that and according to what cri-teria? Discretional manipulation of weights can easilybecome arbitrary even when done by independentand respected central banks, as the recent experienceof Chile suggests.

Indeed, if simplicity, transparency and observ-ability are the main virtues of a currency board, 4

moving toward a complex and ever-changing basketsystem may undermine the very foundations of the

policy. And, of course, pegging to a basket meansthat pairwise exchange rates fluctuate as much asinternational cross rates do, and this adds risk to cer-tain kinds of transactions. Much of the appeal ofcurrent Argentine policy comes from the constantand one-for-one exchange rate, which all BuenosAires taxi drivers know and can brag about. A com-plex arrangement in which the price of the UnitedStates dollar fluctuated unpredictably every daymight not command the same kind of support andwould almost certainly not impose the same degree

of transparency upon monetary policy.

E. Combining exchange rate and

financial stability

The essence of a currency board is that it se-verely limits the ability of the authorities to extenddomestic credit. This may be good for preventing

inflation, but it can be bad for bank stability: under acurrency board or the gold standard, domestic bank-ing is left without a lender of last resort, and in aworld of fractional banking and imperfect depositinsurance this amounts to an invitation to self-ful-filling bank runs. A conclusion, couched in modernlanguage, that economists have known at least sinceBagehot: systems that tie the central banks handsand prevent it from printing money, also prevent itfrom coming to the rescue of banks at times of trou-ble. As Chang and Velasco (1998a) show formally

in a model of the Diamond and Dybvig (1983) type,a currency board makes balance-of-payments crisesless likely only at the price of making bank crisesmore likely. The price of low inflation may be en-demic financial instability.

An alternative is to use fiscal instead of mon-etary policy for helping troubled banks. But sincedeveloping countries are typically rationed at timesof crisis, it is not feasible for the government simplyto borrow against the present value of future tax re-

ceipts and then hand over the money to the bankers.Ready help at times of trouble requires that the fis-cal authority build, via sustained surpluses, a warchest to be kept in liquid form. For a country toself-insure its banking system in this way is, at leastin theory, perfectly possible but costly. Even if wegloss over the political difficulties, the financial costsare large. The following example is suggestive: im-agine a country with M2 equal to two thirds of GDP,which keeps half that amount in time deposits inZurich. Such deposits pay 50 basis points below

LIBOR, while domestic interest rates in the countryin question are 2.5 per cent above LIBOR. Hence,the lower bound for the net cost of holding the warchest is one per cent of GDP per annum.

Can the country do better by purchasing suchinsurance abroad? After all, if lenders can diversifyaway the risk of country-specific bank runs, suchinsurance need not be expensive. This is presumablythe logic of the Argentine policy of contracting a lineof credit (for which a premium is paid annually) tobe used in case of bank troubles. The idea is appeal-ing, but not without potential difficulties. First, ifthere is regional or global contagion, the risk of bankruns need not be easily diversifiable for lenders. Sec-


13/24


ond, the obvious potential for moral hazard makessuch contracts hard to write and enforce. Third is theissue of size: press accounts put the Argentine lineof credit at $6 billion, which is less than 10 per centof M2. Whether larger amounts may be provided bythe market at a reasonable premium is unclear.

Not everyone feels this is a problem. Dornbusch(1998) wrote: The counter argument that currencyboards or full dollarization sacrifice the lender of lastresort function are deeply misguided... Lender of lastresort can readily be rented, along with bank super-vision, by requiring financial institutions to carryoff-shore guarantees. But how exactly does one rentsuch a lender? We saw that contingent credit linesare not without problems. A currently fashionablealternative is to encourage foreign ownership of do-mestic banks, hoping that equity holders abroad will

serve as lenders of last resort. Again, this is prob-ably a good idea, but a completely untested one. WillCitibank U.S. ride to the rescue every time that Latinor Asian bank in which it has a 10 per cent equitystake gets into trouble? Perhaps. But hanging a wholefinancial systems health on that conjecture seemsrisky indeed.

III. Can exchange-rate flexibility help?

Currency boards and dollarization, then, are one but certainly not the only way forward. The alter-native is greater flexibility in exchange rates. That isindeed the direction in which many developing coun-tries, overwhelmed by the difficulties inherent in softpegs, have been moving. Is this a good idea?

A. The basic case for flexibility

The classical argument by Milton Friedman(1953) in favour of flexibility still holds much wa-ter: if prices move slowly, it is both faster and lesscostly to move the nominal exchange rate in responseto a shock that requires an adjustment in the real ex-change rate. The alternative is to wait until excessdemand in the goods and labour market pushesnominal goods prices down. One need not be anunreconstructed Keynesian to suspect that processis likely to be painful and protracted. The analogythat Milton Friedman used is revealing and accurate:every summer it is easier to move to daylight sav-ings time than to coordinate large numbers of peopleand move all activities by an hour.

The case for exchange-rate flexibility is espe-cially strong if the country in question is oftenbuffeted by large real shocks from abroad. The logichere is once again due to Mundell although in thiscase it is the somewhat later Mundell (1963) of themodel that linked his name to Flemings. If shocksto the goods markets are more prevalent than shocksto the money market, then a flexible exchange rate ispreferable to a fixed rate. And, of course, foreignreal variability is likely to be particularly large forexporters of primary products and/or countries highlyindebted abroad that is, a profile that fits manyemerging market countries. Indeed, the 1990s pro-duced large fluctuations in the terms of trade andinternational interest rates relevant for these coun-tries. Note also that the preference for flexibleexchange rates among countries with a heavy natu-ral resource base extends into the OECD: Australia,Canada, New Zealand and some of the Scandinaviancountries are good examples.

This old set of arguments in favour of exchange-rate flexibility for developing countries has recentlycome under attack from a number of fronts. One claimis that depreciations, like increases in the money sup-ply, only work if they surprise the public. And, ofcourse, no government can surprise all of the publicall of the time: repeated depreciations only causeinflation, without real effects. This claim is correct,but also perfectly irrelevant. The Friedman case forflexibility certainly does not advocate attempting touse the nominal exchange rate to keep real activityaway from its natural equilibrium level. On the con-trary, it advocates letting the nominal exchange ratemove to adjust relative prices to the new equilibriumlevel, after a shock has rendered the old constella-tion of relative prices obsolete.

A more relevant objection has been raised byHausmann et al. (1999). They argue that the classiccase may be right in theory, but wrong in practice fordeveloping countries. One problem, in their view,lies in the prevalence of wage indexation. And un-derstanding that nominal depreciation is unlikely tolead to real depreciation, central banks are reluctantto use it for counter-cyclical purposes. Another dif-ficulty lies with the classicpeso problem: in countrieswith a public rendered sceptical by decades ofcurrency debauchery, movements in the nominal ex-change rate tend to be anticipated by changes innominal interest rates, so that real rates do not fall(and may in fact rise) in response to adverse shocks.Hausmann et al. (1999) test these two claims withLatin American data, and find some qualified sup-


14/24


port. Their influential conclusion: exchange-rate flex-ibility does not deliver much insulation or monetarypolicy autonomy, while lacking the credibility valueof a hard peg. Currency boards are therefore a betteroption.

This revisionist view has a grain of truth, butdoes not generally invalidate the claim that exchange-rate flexibility, if properly managed, can be stabilizing.The key, as with fixed rates, lies in having credibility.Ongoing depreciations that follow from imprudentor opportunistic monetary behaviour will surely cometo be expected by agents, and hence will have noreal effect; occasional depreciations that respondexclusively to unforecastable shocks will, almost bydefinition, have real effects.5 The hard part is ensur-ing that the second case, and not the first, prevails.

Regimes with exchange-rate flexibility are rela-tively new to Latin America, and were almost alwaysadopted as the emergency response to an exchange-rate crisis (Mexico in 1994 and Brazil in 1999 aregood examples). Moreover, such regimes are run bycentral banks that have been legally independent foronly a few years. It therefore seems safe to conjec-ture that they lack credibility.6 If that is so, the policyconclusions extracted from the econometric exercisesin Hausmann et al. (1999) are vulnerable to the Lucascritique. What is being estimated are not structuralparameters linking the exchange rate with real inter-est rates and real exchange rates, but parameters thatwould change if the policy regime changed in thesense of becoming more credible over time. The de-gree of wage indexation, for instance, is almostcertainly a function of past inflation rates, and wouldprobably decline as inflation declines. The experi-ence of countries like Chile, where inflation has beenlow for over a decade, offers some support for thisconjecture.

Another way of approaching the same issue isto focus on the degree of pass-through from exchangerates to prices. If every movement in the nominalexchange rate is quickly reflected in an upward ad-

justment in domestic prices, then the insulation pro-vided by flexible exchange rates is nil, or close tonil. Both theory and evidence suggest that marketstructure and the degree of competition in goodsmarkets matter crucially for the degree of pass-through. But just as important is whether exchange-rate changes are perceived as permanent or transitoryand this, in turn, depends crucially on the averageperformance of inflation and monetary policy.Leiderman and Bufman (1996) investigate the issue

empirically for a number of countries (both devel-oped and developing), and conclude:

A different pattern arises in the Latin Ameri-can countries and Israel, where there is a muchweaker link between nominal and real ex-change rates, thus indicating a stronger pass-

through than in the foregoing countries. Thesefacts seem to be consistent with the notion that,other things being equal, the degree of pass-through is likely to be stronger in a high-infla-tion environment ...

B. Credibility versus flexibility

The standard theoretical debate on the virtuesof alternative exchange-rate regimes centres on thealleged tradeoff between credibility and flexibility.Start from the common assumption that full cred-ibility (technically, doing away from the time incon-sistency problem) can only be obtained through ahard fix. Combine that with a setting with pre-setwages or prices, so that unexpected movements inthe nominal exchange rate can have real effects. Then,as Rogoff (1985) convincingly showed, there is aclear tradeoff between the gains from low inflationand the those from counter-cyclical monetary policy(see also Velasco, 1996). An irrevocable fix robs acountry of one adjustment tool. If shocks buffetingan economy are sufficiently large (technically, if theirvariance exceeds some threshold), then fixing is notex ante welfare-improving. By contrast, if the infla-tion bias that occurs under discretional monetarypolicy is large enough, then flexing is not ex antewelfare-improving.

The earlier discussion suggests that while thistradeoff may well be relevant for developed econo-mies, it is not necessarily so for emerging marketeconomies. In this latter class of countries, credibil-ity appears to be a pre-requisite for flexibility to beuseful. In its absence, as Hausmann et al. (1999) use-fully stress, flexibility can be destabilizing.

The crucial policy question, then, is whether aregime of exchange-rate flexibility is compatible withsustained monetary credibility, or whether in coun-tries with a weak track record some kind of anexchange-rate anchor is needed. Conventional wis-dom has often chosen the latter option, emphasizingthe political and other costs of reneging on exchange-rate commitments. But, as we argued above, neithertheory or empirics are conclusive in this regard.7


15/24


Much hinges on the independence with which thecentral bank can carry out policy. And in turn thisdepends, to a large extent, on the degree of socialconsensus regarding the benefits of low inflation.

A number of small open economies have had

successful experiences with exchange-rate flexibil-ity, often coupled with inflation targeting. Australia,Chile, Colombia, Israel, New Zealand and Swedenare among them (Leiderman and Bufman, 1996). Inthese countries, moderate or low inflation has coex-isted with growing degrees of flexibility. In reviewingthe experience of these and other countries experi-menting with more flexible arrangements in the earlyand mid-1990s, Leiderman and Bufman (1996) con-clude: Despite fears that flexibility and enhancedmonetary policy autonomy would lead to uncon-trolled high inflation, there has been a substantialdecrease in the rate of inflation in most countries.

The more recent experience of Mexico andChile is also encouraging. In the years since the 1994crisis, Mexico has been running a money-basedpolicy with a de facto dirty float. The same is true ofChile, where an exchange-rate band has been wid-ened significantly. In both countries the central bankis legally independent. Several econometric studiesshow that in both Chile and Mexico policy has tight-ened systematically in response to expected inflation,and since the mid-1990s inflation has been trendingdownward.8 Their reaction to the Asian and then theRussian debcle is also encouraging. In the courseof 1998 both countries suffered large terms-of-tradeshocks, and their currencies came under pressure.Both countries allowed moderate depreciation (largerin Mexico than in Chile), which resulted in somereal depreciation as well. Inflation did not get out ofhand: it continued to fall in Chile, while it temporar-ily rose and then fell again in Mexico.9 The resulthas been a soft landing, with lower but still positivegrowth and reduced current-account deficits.10

But the evidence we have is limited. For one,there are still relatively few developing countries withdirty floats, and most of these have relatively shorttrack records. And in many of them, that record is stillcontaminated by the abrupt adoption of floating, of-ten in response to a crisis. But since, in response tothe most recent round of crashes, a number of so-called emerging markets (Brazil, the Republic ofKorea and the Russian Federation among them) havemoved to floating, much evidence will be producedin the near future. Researchers should start sharpen-ing their pencils and readying their computers.

C. Politics and policy-making

At one level, the current enthusiasm for hardpegs springs from a lack of enthusiasm for develop-ing countries ability to build institutions and togovern themselves soundly. Much of the currentconventional wisdom seems to say: just as war is tooimportant to be left to the generals, monetary policyis too important to be left to the central bankers especially if they hail from developing countries withweak political institutions. It is better to adopt a sys-tem that removes all discretion from domestic actorsand puts monetary policy on automatic pilot, withthe tough decisions transferred to the presumablysounder (or at least more politically insulated) bu-reaucrats in Washington or Frankfurt. In short, thecase for hard pegs rests ultimately on a political ar-

gument.

Whether that political argument is correct ornot is an empirical matter. Cynics can easily point todeveloping countries were an independent monetarypolicy is a chimera; optimists can readily point todeveloping countries where the track record suggestsotherwise. More problematic is the fact that, if theassessment of developing countries limited capac-ity for sound policy-making is empirically correct,that undermines the whole case for hard pegs. In-

deed, it renders it internally inconsistent.

The problem is this. Not even the most enthusi-astic advocate of currency boards would deny thatthey require soundly supervised banks and prudentfiscal policies. The literature is littered with calls forstrengthening bank supervision and eliminatingbudget deficits before adopting hard pegs. But bothof these prerequisites are technically taxing and po-litically troublesome. Why should a country withoutthe political wherewithal to set its interest rate pru-

dently be able to attain them?

Start with financial supervision. It was alwayshard, and globalization and innovation have made itmuch more so. In recent years, the United States,Japan and the Scandinavian countries have sufferedfinancial crashes that could be traced back, at leastpartially, to poor regulation. In the summer of 1998,the near collapse of Long-term Capital Managementrevealed a gaping hole in the regulatory arrangementscovering Wall Street. In short, there can be no doubtthat supervising banks is as technically demanding,if not more so, than implementing monetary policy.And the political constraints are just as large, as welearn from the uproar that invariably follows attempts


16/24


to close insolvent banks or to make good on earlierpromises of no government bailouts. This is exactlywhat the recent experience of Ecuador, Indonesia andJapan, to mention just a few examples, shows.

Much the same may be said about fiscal policy.Here the technical obstacles are fewer but the politi-cal pressures even greater. The record on unsustain-able budget deficits is just as checkered for bothdeveloped and developing nations. Theory, bothKeynesian and neoclassical, calls for a larger-than-average fiscal surplus at times of economic expan-sion and a smaller-than-average surplus at times ofcontraction. For many developing countries, though,these prescriptions are a far cry from reality. As aseries of papers produced by the Office of the ChiefEconomist of the Inter-American Development Bankshowed, fiscal policy in Latin America has been

clearly procyclical, in contrast both to theory and toobserved behaviour in the OECD (see Gavin et al.,1996, and references therein).

The problem, once again, has to do with cred-ibility. Policy makers would like to run a counter-cyclical fiscal policy, but they cannot. Doing so wouldinvolve borrowing large amounts at times of trou-ble. And, given political institutions, past record ofrepayment, the volatility of terms of trade, etc., lend-ers simply do not lend when the money is most

needed.11

Gavin et al. (1996) provide a useful illus-tration focusing on the post-Tequila effect experi-ence of Argentina and Mexico:

In 1995 both countries found themselves inthe midst of severe recessions. Despite this,both countries implemented strongly con-tractionary fiscal policies, almost certainlycontributing to the depth of the recession andpostponing recovery. This was not done be-cause officials in both countries would nothave liked to implement a more counter-cyclical policy. It was done because, in light

of investors loss of confidence in short-termprospects, financing of the deficits that wouldhave been implied by a counter-cyclical policywas simply not available.

The dilemma facing those who attempt to de-sign policy institutions for developing countries isstark. If politics prevents a country from managingits monetary policy soundly, then politics will belikely to prevent its banks and public finances frombeing properly managed as well. In that case, adopt-ing a hard peg solves part of the political problem,but leaves the country potentially exposed to finan-cial or fiscal crises. And these in time may also erodethe viability of the peg.

Alternatively, if a countrys politics and insti-tutions allow bank regulators some autonomy andlegislators some fiscal forethought, then that coun-try can also probably sustain an independent andcredible central bank. And if it can make itself cred-ible in the eyes of investors and markets, should notthat country also be entitled to enjoy the benefits ofexchange-rate flexibility?

D. Exchange-rate flexibility and

financial stability

A major lesson from recent crises in emergingmarkets is that financial factors are key in determin-ing an economys vulnerability to shocks. Anyadvocate of exchange-rate flexibility therefore has

to wrestle with the question of whether it is compat-ible with financial stability. After all, financialsystems do not respond well to sharp and unforecast-able changes in asset prices. Since the exchange rateis the price of that supremely important asset, do-mestic money, a regime of flexibility is nothing buta deliberate attempt to allow this asset price to fluc-tuate freely. Can this be an invitation to financialfragility? Yes and no.

The presence of dollar debt is often presented

as an argument against flexibility. Let us suppose thatdomestic firms have borrowed in dollars. Suppose,in addition, that at least some of them are in the non-traded goods sector and have earnings in localcurrency, and that the same is true of the government.Then a nominal devaluation, if successful in thechanging relative prices, drastically increases thecarrying costs of this debt, and can generate a wave ofcorporate bankruptcies along with a fiscal crisis. Thisdanger has been stressed in some interpretations of theAsian crisis particularly that of Corsetti et al. (1998).

Calvo (1999) also stresses that liability-dollarizedeconomies are highly vulnerable to devaluation.

But there are a number of important caveats tothis argument. Dollar debt can be hedged and, as dis-cussed in more detail below, a flexible exchange rategives borrowers an incentive to hedge that may beabsent under more rigid regimes. In addition, if aexternal shock calls for a real depreciation, this willhappen regardless of the exchange-rate system inplace. Policy will only determine the manner of ad-

justment. Under flexible rates the change in relativeprices occurs suddenly and sharply. Under fixed ratesor a currency board the real depreciation will takeplace slowly, as nominal prices fall. Throughout the


17/24


adjustment period the real depreciation will be an-ticipated by markets, and hence domestic real rateswill rise above world rates. And if there are doubtsabout the sustainability of the peg, interest rates willbe even higher. At the end of the day, the real valueof debt service will have risen relative to the price ofhaircuts. This process can wreck corporate and bankbalance sheets just as surely as a devaluation.

How steep the real devaluation/real interest ratetradeoff actually is we do not know, and this is cer-tainly a point that cries out for more empiricalresearch. What seems certain is that the answer willdepend heavily on specific country circumstances:strength of banks, currency denomination of assetsand liabilities, maturities, degree of hedging, etc.A real depreciation may be lethal in Indonesia andthe Republic of Korea, where unhedged short-term

foreign debt was the norm; the same is not true ofChile, for instance, where unhedged short-term for-eign debt is minimal.

A related and key point is that the circumstancesthat affect the slope of this tradeoff are not God-given,but often the result of deliberate policy design. Onecommon culprit is financial liberalization. Radeletand Sachs (1998) and Chang and Velasco (1998b)have argued, for instance, that changes in financialand tax policies in Thailand and elsewhere created

incentives for taking on dollar debt. Similarly, aninsistence on fixing, accompanied by frequent offi-cial assurances that exchange rates would never bedevalued, may have discouraged prudent hedging byprivate firms. Indeed, observers such as Radelet andSachs (1998) have claimed that the Asian pegs mayhave fostered a moral hazard problem among bor-rowers, who felt protected by the official guaranteeson the exchange rate.

Finally, flexible rates may also be helpful in

dealing with financial instability. Chang and Velasco(1998a) also show that a regime in which bank de-posits are denominated in domestic currency, thecentral bank stands ready to act as a lender of lastresort and exchange rates are flexible, may help fore-stall self-fulfilling bank runs. The intuition for thisis simple. An equilibrium bank run occurs if eachbank depositor expects others will run and exhaustthe available resources. Under a fixed rates regime,those who run to the bank withdraw domestic cur-rency, which in turn they use to buy hard currency atthe central bank. If a depositor expects this sequenceof actions to cause the central bank to run out of dol-lars or yen, then it is a best response for him/her torun as well, and pessimistic expectations become self-

fulfilling. On the other hand, under a flexible ratesregime plus a lender of last resort there is alwaysenough domestic currency at the commercial bankto satisfy those who run. But since the central bankis no longer compelled to sell all the available re-serves, those who run face a depreciation, while thosewho do not run know that there will still be dollarsavailable when they desire to withdraw them at alater date. Hence, running to the bank is no longerthe best response, pessimistic expectations are notself-fulfilling, and a depreciation need not happen inequilibrium.

In my view this represents a strong (thoughsurely not overwhelming) case in favour of flexibleexchange rates. But there are caveats. One is thatsuch a mechanism can protect banks against self-ful-filling pessimism on the part of domestic depositors

(whose claims are in local currency), but not againstpanic by external creditors who hold short-term IOUsdenominated in dollars. To the extent that this wasthe case in Asia, a flexible exchange-rate systemwould have provided only limited protection.12 Andproper implementation is subtle. If they are to be sta-bilizing, flexible rates must be part of a regime whoseoperation agents take into account when formingexpectations. Suddenly adopting a float because re-serves are dwindling as Mexico did in 1994 andseveral Asian countries have done more recently

may have the opposite effect by further frighteningconcerned investors.

IV. Making exchange-rate flexibility

work in practice

Giving up a peg, whether of the hard or softvariety, means that the economy gives up one nominalanchor. Finding and implementing an alternative anchor

is the first task of advocates of exchange-rate flexibil-ity. Other issues include the optimal degree of inter-vention in the foreign-exchange market (if any), andthe choice of instrument and rules for conductingmonetary policy. I shall discuss them in turn below.

A. Nominal anchors and inflation targets

The choices for nominal anchor under floatingboil down to two: monetary aggregates or inflationtargets. Among emerging market countries the latteris by far the most popular. To my knowledge, onlyMexico follows a policy of quantitative targets.


18/24


The popularity of inflation targets should notbe surprising. Given the instability of money demandin most economies, targeting aggregates is neithertheoretically optimal nor easy to do in practice. In-flation targets may also prevent the time inconsist-ency problem that leads to an inflation bias, whileavoiding the pitfalls of fixed exchange rates. Andinflation targets may also have some of the attributesof hard pegs, in particular transparency and observabil-ity. The inflation rate may be published with a lag, butit is just as accessible and comprehensible to the pro-verbial taxi driver as is the nominal exchange rate.

As mentioned above, a number of developedcountries, including Canada, Finland, New Zealand,Spain, Sweden and the United Kingdom, have ex-perimented with inflation target policies of slightlydifferent sorts. Performance has been reasonably

good, according to most published academic evalu-ations.13 Inflation targets are less common amongemerging market economies. According to Massonet al. (1997), Chile is the country that seems to comethe closest to conducting its monetary policy in amanner consistent with an inflation target. Colom-bia, Indonesia (before the crash), Mexico and thePhilippines have regimes that in some ways resem-ble an inflation target.14

What is the scope for a more widespread and

successful use of inflation targets among developingcountries? That is a difficult empirical question, onwhich much more research is needed. Masson et al.(1997) identify two requirements for successful in-flation targeting in such countries: freedom fromcommitment to another nominal anchor like the ex-change rate or wages, and the ability to carry out asubstantially independent monetary policy, especiallyone not constrained by fiscal considerations. Theformer is obviously less constraining to the extentthat many countries are moving towards exchange-

rate flexibility. There are also grounds to be optimisticon the second count: legally independent centralbanks are increasingly common, and the reliance onseigniorage to finance government spending has less-ened, even in traditionally inflationary regions likeLatin America.

B. Dealing with short-term exchange-rate

fluctuations

The conclusion that a clean float is the only al-ternative to a hard peg is largely academic. In thereal world clean floats do not exist. Major industri-

alized countries such as Canada and the United King-dom, smaller OECD countries such as Australia andNew Zealand, and middle income countries such asPeru and Mexico, all practice floating with varyingdegrees of dirt. Even the United States, usuallyregarded as the cleanest of the floaters, intervenesoccasionally in the foreign-exchange market.

The main reason for this is clear. Clean float-ing means high volatility of nominal exchange rates much higher than early advocates such as Fried-man (1953) and Johnson (1969) anticipated.15 And,as Mussa (1986) was the first to point out and manyhave documented since, that almost always meansgreater volatility of the real exchange rate, for pricesmove sluggishly. To the extent that this volatility inrelative prices is costly, either directly or because itcauses volatility in output or in the health of the fi-

nancial system, policy makers typically want tomitigate it.

Under inflation targeting there are additionalreasons for managing the exchange rate to some de-gree. The exchange rate affects inflation through twochannels, as Svensson (1998) has pointed out:

In an open economy, the real exchange rateaffects the relative price between domestic andforeign goods, which in turn affects both do-

mestic and foreign demand for domesticallyproduced goods, and hence affects aggregatedemand and inflation.

There is also a direct channel, in that the ex-change rate affects domestic currency pricesof imported foreign goods, which enter theconsumer price index.

Hence, any scheme to control the rate of inflation onthe short horizon must control, to some extent, thebehaviour of the nominal exchange rate. That helps

explain the prevalence of managed or dirty floats inthe real world.

C. Dealing with long swings in the

exchange rate

A harder question is whether authorities shouldattempt to mitigate not just short-term volatility butalso longer swings in the nominal and real exchangerate. The question has much practical and empirical

justification. Most observers agree that under float-ing the exchange rate can be subject to persistentmovements that are only weakly related to funda-


19/24


mentals. One often-mentioned example is the behav-iour of the dollar in the Reagan years. Obstfeld (1995)writes: Exhibit A in the case for irrational exchange-rate misalignment has long been the dollars massiveappreciation between 1980 and 1985, which amountedto somewhere between 40 and 60 per cent, depend-ing on the measure used.

Something similar could be said of the sharpreal appreciation suffered by most Latin Americancurrencies in the first half of the 1990s. Part of itcould be plausibly justified by the productivity gainsthat liberalizing reforms presumably brought; but agood part of it followed from very large capital in-flows, which kept coming because of the expectationthat currencies would appreciate even further. Whenexpectations reversed, so did the capital flows, andcurrencies crashed: Mexico in 1994 and Brazil in1999.

Such concerns have led to policies to limit ex-change-rate movements via flotation bands. And ifsuch bands crawl, so that their centre remains closeto an estimate of the equilibrium exchange rate,then medium-term misalignment can be avoided.Avoided, that is, to the extent that the edges of theband are defensible and, in the aftermath of theAsian, Brazilian, Mexican and Russian crises, theconsensus in the profession seems to be that theycannot be. Bands with hard edges eventually fallprey to the pressures of the market-place.

Williamson (1998) proposes monitoring bandsas a possible compromise solution. This is a bandthat attempts to target the real exchange rate, but witha twist. As he puts it:

The key difference between a crawling bandand a monitoring band is that the latter doesnot involve an obligation to defend the edge

of the band. The obligation is instead to avoidintervening within the band (except in a tacti-cal way, to prevent unwanted volatility). Thereis a presumption that the authorities will nor-mally intervene to discourage the rate fromstraying far from the band, but they have awhole extra degree of flexibility in decidingthe tactics they will employ to achieve this.

At one level, Williamsons proposal seems un-exceptionable. In practice, most central banks usebands of this sort in deciding their intervention policy,although the degree to which they do so explicitlyvaries widely. In any managed float, the authoritiesare likely to intervene if the exchange rate strays

too far from their perceived medium-term equilib-rium value.

However, two issues immediately arise. One ishow a central bank can avoid drawing a line in thesand, however fuzzy, if the exchange rate divergessystematically and in the same direction, from itsestimated equilibrium level. Consider again the caseof several Latin American currencies in the early partof the 1990s. The central banks of several countries including Brazil, Chile and Colombia were con-cerned about real appreciation. At the same time theyused fairly broad bands, and were not shy about wid-ening the bands from time to time when marketpressures demanded it. This avoided some of theproblems of hard-edged bands, but not all. On severaloccasions markets believed they identified thresh-olds for central bank intervention, and occasionally

mounted speculative attacks against these perceivedthresholds. When the monetary authorities retreated,as they often did, some credibility was lost.

The other key question, as Williamson himselfpoints out, is how much difference such a band wouldmake to the day-to-day movements in the exchangerate. The main result of literature on target zones pio-neered by Krugman (1991) was that the presence ofthe band may be stabilizing (in the sense of makingthe exchange rate less responsive to movements in

fundamentals) even when the currency price was wellwithin the edges of the band. But the less credible orthe less clearly defined the boundaries of the band,the weaker presumably is this stabilizing effect. Doesa band with very fuzzy edges approach, in the limit,the workings of a clearly floating exchange rate? Theanswer is probably yes, but the issue clearly meritsfurther research.

D. Crafting monetary policy

How should monetary policy be implementedand designed in this context? The Taylor rule oftenused by central banks provides a natural focus forthe discussion. In such a rule the nominal interestrate typically depends on the output gap and the de-viation of measured or expected inflation with respectto the target. In the open economy, several interest-ing issues arise in the design of this rule.

Mitigating short-term volatility in the exchangerate (and thereby in the rate of CPI inflation)requires that the nominal parity itself be in-cluded in the rule, either in rate of change form


20/24


or in deviations with respect to a target. Thelarger the coefficient on this argument, the moremanaged the exchange rate. As Svensson(1998) shows, putting the exchange rate in theTaylor rule is likely to be optimal for most speci-fications of social welfare function, and espe-cially when shocks are predominantly nominal.

Targeting quarterly or annual CPI inflation neednot be optimal. This is because in open econo-mies, as we saw above, the exchange rate has adirect impact on the CPI via import prices. Andto the extent that the nominal exchange rate fluc-tuates in response to shocks, stabilizing theshort-term CPI inflation could introduce exces-sive volatility in interest rates and output. Analternative is to target inflation in the non-tradeable sector, which is less influenced by ex-

change-rate movements; or, as Ball (1998)suggests, to target a modified inflation indexthat filters out the transitory effects of exchange-rate movements; or to use an average of CPIinflation over a longer period.16

Pure inflation targeting, in which only nominalvariables are included in the right-hand side ofthe Taylor rule, may well be inferior to a flex-ible targeting approach in which output or realexchange-rate deviations are also considered.

This is true in closed economies but even moreso in open economies again, because nominalexchange-rate volatility may cause excessivereal volatility. If pure inflation targeting is tobe pursued, it is better to target long-run oraverage inflation, as Ball (1998) shows.

These are preliminary results, using very gen-eral models. Conclusions are quite sensitive to modelspecification, the social utility function chosen, andthe relative variance of different shocks. Clearly,more research is warranted.

V. Complementary policies: financial

regulation, capital controls and

fiscal institutions

Any exchange regime, and especially a flexibleone, requires complementary policies to increase itschances of success. Some have suggested the use ofcontrols on capital flows. Less controversial is theneed for prudential regulation of the financial systemand for counter-cyclical fiscal policy. I review brieflyeach of these policies in this concluding section.

A. Financial liberalization and fragility

We saw above that weak banks can be a mainconstraint for monetary and exchange-rate policy.Only when banks are reasonably healthy can policybe used freely, without the fear that interest or ex-change-rate fluctuations will bring the bankingsystem tumbling down. Hence, identifying and tack-ling the sources of financial fragility is crucial formacro policy makers in developing countries.

In their 1996 paper on the twin crises,Kaminsky and Reinhart (1996) found that: (i) of the26 banking crises they studied, 18 were preceded byfinancial sector liberalization within a five-year in-terval; and (ii) financial liberalizations accuratelysignalled 71 per cent of all balance-of-payments cri-ses and 67 per cent of all banking crises. Theexperiences of Chile, Mexico, and now East Asia,strongly confirm this general tendency. Freeing in-terest rates, lowering reserve requirements, andenhancing competition in the banking sector aresound policies on many grounds and indeed, coun-tries in which they are applied often experience anexpansion in financial intermediation. But they canalso sharply reduce the liquidity of the financialsector, and hence set the stage for a potential crisis.This is the main finding ind Demirguc-Kent andDetragiache (1998).

Beyond the effects of liberalization on liquid-ity, a host of other potential ills have been mentionedin the literature. In particular, deregulation coupledwith explicit or implicit guarantees on banks and in-adequate oversight can generate a serious moralhazard problem. Overlending and excessive risk-tak-ing are likely results, as argued by Velasco (2000)for the case of Chile and by Krugman (1998) for therecent Asian episode. A lending boom and growingshare of risky or bad loans often result. As Gavinand Hausmann (1995) persuasively argue, the em-pirical link between lending booms and financialcrises is very strong. Rapid growth in the ratio ofbank credit to GDP preceded financial troubles not

just in Chile and Mexico, but also in Argentina(1981), Colombia (19821983), Uruguay (1982),Norway (1987), Finland (19911992), Japan (19921993) and Sweden (1991).17

The moral of the story is the same in both cases.Financial liberalization should be undertaken cau-tiously. Reserve requirements can be a useful tool instabilizing a banking system, as the experience of


21/24


Argentina in 1995 showed. Lowering them to zero,as Mexico did in the run up to the 1994 crash, smacksof imprudence.

B. Capital inflows and short-term debt

Short-term government debt proved to be dan-gerous in the case of Mexico; short-term externaldebt has proven to be risky in the case of Asia. Inboth cases, runs against this debt ultimately broughtthe exchange rate down. What can be done about it?

Restraining short-term borrowing involves nofree lunch, for both governments and banks haveperfectly sound reasons for wanting to make at leastsome of their liabilities short-term. At the same time,it is not clear whether decentralized decision-mak-ing delivers the optimal debt-maturity structure:governments may rely too much on short-term debtif they suffer from time inconsistency or high dis-counting; foreign creditors may only be willing tolend short because of imperfect information or moni-toring, or because of coordination failure with othercreditors (if each creditor expects the others will onlylend short, thus making a crisis possible, his best re-sponse is also to lend short in order to have a chanceto get out if the crisis comes). These suggest thatthere may be a case for a policy discouraging short-term debt.18

Exactly what policy is a tricky matter. High re-quired reserves on liquid bank liabilities (whether indomestic or foreign currency, and whether owed tolocals or foreigners) is an obvious choice. It may besound policy, even if it has some efficiency costs orif it causes some disintermediation. An obvious ca-veat is that if banks are constrained firms will dotheir own short-term borrowing, as happened mas-sively in Indonesia. Taxes on capital inflows wherethe tax rate in inverse proportion to the maturity ofthe inflow (and where long term flows such as FDIgo untaxed at the border) was used by Chile andColombia in the 1990s. They are often justified interms of findings such as those of Sachs et al. (1996b),who found that a shorter maturity of capital inflowswas a helpful predictor of vulnerability to the Te-quila effect in 1995, while the size of those inflowswas not. Valds-Prieto and Soto (1996), Larrain etal. (1997), and Montiel and Reinhart (1997), all findthat the restrictions have affected the maturity com-position of flows, though not their overall volume orthe course of the real exchange rate.

C. Improving fiscal institutions

We saw above that excessively procyclical fis-cal policies are the inevitable consequence of weakand deficit-prone fiscal institutions. Lenders do notlend when times are bad because they do not think

they will be repaid when times are good. This infor-mal view is corroborated by the formal evidence sug-gestive of fiscal borrowing constraints provided byGavin and Perotti (1997). And the weaker the coun-trys budgetary institutions the greater the problem,as shown by Gavin et al. (1996).19 The consequence:fiscal policy is not much use as a counter-cyclicaltool. If monetary policy is not available either per-haps because of an exchange-rate peg then coun-tries can be left bereft of a stabilization policy.

A simple and first step forward is to reduce thelevels of public indebtedness. With less initial debt,there is more room to expand in bad times withoutrunning into borrowing constraints. The ratio of pub-lic debt to GDP of East Asian and Latin Americancountries is low by OECD standards; but so prob-ably are their credit ceilings, for obvious politicaland institutional reasons.

A second step is to reform fiscal institutions tomake spending less cyclical and repayment more

likely. One possibility is the National Fiscal Councilproposed by Eichengreen et al. (1996), which wouldgive responsibility for the broad trends in fiscal policyto an autonomous body modelled after independentcentral banks. National congresses would still setspending levels and composition, but the size of thedeficit (or the allowable debt issuance) would be setby the autonomous Council. If this gave fiscal policygreater credibility, in the sense of ensuring that defi-cits today need not mean deficits tomorrow and intothe indefinite future, then fiscal policy would be moreuseful as a stabilization tool.

Notes

1 Other recent experiences with a currency board includeEstonia, Lithuania and Bulgaria.

2 One can think of exceptions. There may be shocks thatare so clearly observable and exogenous that they passthe test. For instance, Sachs et al. (1996a) argue that theassassination of presidential candidate Luis DonaldoColosio in Mexico in March 1995 could plausibly have

justified the abandonment of the exchange-rate band.3 Some coincide with the conditions put forth by Williamson

(1998).4 This case is made formally by Herrendorf (1997 and 1999).


22/24


5 For a formal model that yields this result, see Obstfeldand Rogoff (1995).

6 Some empirical evidence suggests exactly this. In the caseof Mexico, increases in the nominal exchange rate are fol-lowed by higher nominal interest rates, not lower as thestandard model would suggest. Inflationary expectationstend to rise as well. One explanation is that agents inferfrom temporary depreciations a permanent relaxation of

monetary policy.7 Buiter et al. (1998) concur: Is an exchange-rate commit-ment more easily established or more credible than a com-mitment to other nominal anchors? The short answer isthat we have no satisfactory theoretical arguments or em-pirical evidence to argue convincingly on either side ofthe issue.

8 On Chile, see Landerretche et al. (1998); on Mexico, seeEdwards and Savastano (1998).

9 If the real depreciation was not larger, it was not becauseof domestic inflation, but because of external deflation.

10 Mexico grew 7.0 per cent in 1997, 4.8 per cent in 1998,and is forecasted to grow around 2.8 per cent in 1999.Chile grew 7.1 per cent in 1997, 3.5 per cent in 1998, and

is likely to expand by 2.0 per cent this year. The currentaccount was 3.7 per cent of GDP in Mexico in 1998, andis forecasted at 2.8 per cent for 1999. The correspondingfigures for Chile are 6.2 and 3.5 per cent.

11 Formalizations of this story rely on the pitfalls of sover-eign borrowing, much studied in the 1980s. For a recentand interesting attempt, see Aizenmann et al. (1996).

12 Floating is not totally useless in this case, for panic byforeign creditors could perfectly well be triggered by arun by domestic depositors, with the outcome being self-fulfilling. For details on this line of argument, see Changand Velasco (1998a).

13 See Leiderman and Bufman (1996) and the references con-tained therein.

14 Mexico relies mostly on quantitative targets, but also an-nounces an inflation forecast that is meant as a loose guideto expectations. See Edwards and Savastano (1998).

15 See the insightful historical discussion in Obstfeld (1995).16 Of course, this issue only arises to the extent that the float

is reasonably clean. With active exchange-rate manage-ment, targeting CPI and non-tradeables inflation shouldhave practically identical effects.

17 In Mexico and Chile, as in the case of some Asian coun-tries more recently, the perception of government guaran-tees may have created a moral hazard problem and ledbanks to take on excessive risk. Velasco (1992) discussesevidence for this in the case of Chile. Krugman (1998)

stresses the role of moral hazard and over-investment inAsia.18 See Rodrik and Velasco (1999) for more details on the

argument.19 The quality of fiscal institutions is measured by the index

developed in Alesina et al. (1996).

References

AIZENMANN J, GAVIN M and HAUSMANN R (1996). Opti-mal tax policy with endogenous borrowing constraints(mimeo). Washington DC, Inter-American Development

Bank.BALL L (1998). Monetary policy rules for open economies.

NBER Working Paper, no. 6760. Cambridge MA: NationalBureau of Economic Research, October.

BUITER W, CORSETTI G and PESENTI P (1998). FinancialMarkets and European Monetary Cooperation. CambridgeMA, Cambridge University Press.

CALVO G (1999). Testimony on full dollarization. Paper pre-sented at a Joint Hearing of the Subcommittees on Eco-nomic Policy and International Trade and Finance,US Congress, Washington DC, April.

CHANG R and VELASCO A (1998a). Financial fragility and

the exchange rate regime. RR, no. 9805. New York, C.V.Starr Centre for Applied Economics, February. AlsoNBERWorking Paper, no. 6469. Cambridge MA, National Bu-reau of Economic Research.

CHANG R and VELASCO A (1998b). The Asian liquidity cri-sis. Working Paper. Atlanta, Federal Reserve Bank ofAtlanta, May.

CORSETTI G, PESENTI P and ROUBINI N (1998). Whatcaused the Asian currency and financial crises? Part I:Macroeconomic overview. NBER Working Paper,no. 6833. Cambridge MA, National Bureau of EconomicResearch, December.

DEMIRGUC-KENT A and DETRAGIACHE E (1998). Finan-cial liberalization and financial fragility. Working Paper.

Washington DC, International Monetary Fund andWorld Bank.

DIAMOND D and DYBVIG P (1983). Bank runs, deposit in-surance and liquidity.Journal of Political Economy, 91:401419.

DORNBUSCH R (1998). After Asia: New directions for theinternational financial system (mimeo). Cambridge MA,MIT, July.

EDWARDS S and SAVASTANO MA (1998). The morning af-ter: The Mexican peso in the aftermath of the 1994 cur-rency crisis.NBER Working Paper, no. 6516. CambridgeMA, National Bureau of Economic Research.

EICHENGREEN, B, HAUSMANN R and VON HAGEN J(1996). Reforming fiscal institutions in Latin America:The case for a national fiscal council (mimeo). Washing-ton DC, Inter-American Development Bank.

FRIEDMAN M (1953). The case for flexible exchange rates. Essays in Positive Economics. Chicago, University ofChicago Press.

GAVIN M and HAUSMANN R (1995). The roots of bankingcrises: The macroeconomic context (mimeo). Washing-ton DC, Inter-American Development Bank.

GAVIN M, HAUSMANN R, PEROTTI R and TALVI E (1996).Managing fiscal policy in Latin America and the Carib-bean (mimeo). Washington DC, Inter-American Devel-opment Bank, March.

GAVIN M and PEROTTI R (1997). Fiscal policy in Latin

America. NBER Macroeconomics Annual. CambridgeMA, National Bureau of Economic Research.HAUSMANN R, GAVIN M, PAGES-SERRA C and STEIN E

(1999). Financial turmoil and the choice of exchange rateregime (unpublished manuscript). Washington DC, Inter-American Development Bank, March.

HERRENDORF B (1997). Importing credibility through exchangerate pegging.Economic Journal, 107(442), May: 68794.

HERRENDORF B (1999). Transparency, reputation and cred-ibility under floating and pegged exchange rates.Journalof International Economics, 49(1), October: 3150.

JOHNSON H (1969). The case for flexible exchange rates, 1969.Federal Reserve Bank of St. Louis Review, 51, June.

KAMINSKY G and REINHART C (1996). The twin crises: The

causes of banking and balance of payments problems.In-ternational Finance Discussion Paper, no. 544. Wash-ington DC, Board of Governors of the Federal ReserveSystem, March.


23/24


KRUGMAN P (1991). Target zones and exchange rate dynamics.Quarterly Journal of Economics, 106(3), August: 669682.

KRUGMAN P (1998). What happened in Asia? (mimeo). Cam-bridge MA, MIT.

LANDERRETCHE O, MORAND F and SCHMIDT-HEBBEL K(1998). Inflation targets and stabilization in Chile (mimeo).Santiago, Central Bank of Chile, November.

LARRAIN F, LABAN R and CHUMACERO R (1997). What

determines capital inflows? An empirical analysis forChile.Development Discussion Paper, no. 590. CambridgeMA, Harvard Institute for International Development.

LARRAIN F and VELASCO A (1999). Exchange rate policyfor emerging markets: One size does not fit all. WorkingPaper. Cambridge MA, Harvard University, J.F. KennedySchool of Government.

LEIDERMAN L and BUFMAN G (1996). Searching for nomi-nal anchors in shock-prone economies in the 1990s: In-flation targets and exchange rate bands. Working Paper:1696. Tel Aviv, Foerder Institute for Economic Research,Tel Aviv University, June.

MASSON P, SAVASTANO M and SHARMA S (1997). Thescope for inflation targeting in developing countries.IMF

Working Paper, no. 97/130. Washington DC, InternationalMonetary Fund, Research Department, October.MONTIEL, P and REINHART C (1997). Do capital controls

influence the volume and composition of capital flows?Evidence from the 1990s. Paper prepared for UNU/WIDER project on Short-Term Capital Movements andBalance of Payments Crises, Sussex, 12 May.

MUNDELL R (1963). Capital mobility and stabilization policyunder fixed and flexible exchange rates. Canadian Jour-nal of Economics and Political Science, 29: 475485.

MUSSA M (1986). Nominal exchange rate regimes and thebehavior of real exchange rates: Evidence and implica-tions. Carnegie Rochester Conference Series on EconomicPolicy, 25.

OBSTFELD M (1995). International currency experience: Newlessons and lessons relearned.Brookings Papers on Eco-nomic Activity, no. 1.

OBSTFELD M (1997). Destabilizing effects of exchange rateescape clauses. Journal of International Economics,43(12), August: 6177.

OBSTFELD M and ROGOFF K (1995). Exchange rate dynam-ics redux.Journal of Political Economy, 103.

RADELET S and SACHS J (1998). The onset of the Asian fi-nancial crisis (mimeo). Cambridge MA, Harvard Institutefor International Development, March.

RODRIK D and VELASCO A (1999). Short-term capital flows.Annual World Bank Conference on Development Econom-

ics. Washington DC, World Bank.

ROGOFF K (1985). The optimal degree of commitment to anintermediate monetary target. Quarterly Journal of Eco-nomics, 100(4), November: 11691189.

SACHS J, TORNELL A and VELASCO A (1996a). The col-lapse of the Mexican peso: What have we learned?Eco-nomic Policy, 22: 1356.

SACHS J, TORNELL A and VELASCO A (1996b). Financialcrises in emerging markets: The lessons from 1995.Brookings Papers on Economic Activity, no. 1.

SVENSSON L (1998). Open economy inflation targeting (manu-script). Stockholm, University of Stockholm.

TORNELL A and VELASCO A (1998). Fiscal discipline andthe choice of a nominal anchor in stabilization. Journal ofInternational Economics, 46.

TORNELL A and VELASCO A (2000). Fixed versus flexibleexchange rates: Which provides more fiscal discipline?Journal of Monetary Economics, 45(2), April: 399436.

VALDS-PRIETO S and SOTO M (1996). The effectiveness o

UNCTAD Report on Exchange Rate Policy

Documents