Content uploaded by Roberto Chang
Author content
All content in this area was uploaded by Roberto Chang
Content may be subject to copyright.
- Bogotá - Colombia - Bogotá - Colombia - Bogotá - Colombia - Bogotá - Colombia - Bogotá - Colombia - Bogotá - Colombia - Bogotá - Colombia - Bogotá - Colombia - Bogotá -
Inflation Targeting, Reserves Accumulation, and
Exchange Rate Management in Latin America
Por: Roberto Chang
No. 487
2008
Inflation Targeting, Reserves Accumulation, and Exchange
Rate Management in Latin America
1
Roberto Chang
2
Rutgers University and National Bureau of Economic Research
October 2007
Abstract
This paper examines whether central banks in Latin America have implemented conventional
inflation targeting (IT) prescriptions, with a focus on foreign exchange intervention and official
reserves accumulation policies. To this end, the paper reviews the experiences of Brazil, Chile,
Colombia, and Peru, and finds significant departures from the dominant theory of IT. Foreign
exchange intervention has often been used to prevent excessive financial volatility, bubbles, and
panics. Ongoing patterns of reserves accumulation have been the outcome of an effort to build
“war chests” against speculative attacks and, more recently, of a fight against real exchange
appreciation. Possible justifications of the discrepancies between conventional IT theory and
practice are discussed and generally found unsatisfactory.
Keywords: Inflation Targeting, Exchange Rate Policy, International Reserves.
JEL Classification Codes: E5, F4
1
Financing for this paper was kindly provided by the Latin American Reserves Fund (FLAR) and Banco
de la Republica de Colombia. I thank Hernán Rincón, Lars Svensson, Saki Bigio, Ricardo Meirelles
Faria, Vitoria Saddi, seminar participants at Villanova, and participants at the FLAR Conference on
International Reserves in Middle- and Low-Income Countries for advice and suggestions. Adam Gulan
provided excellent research assistance. Of course, all errors and omissions are mine alone.
2
Department of Economics, Rutgers University, New Brunswick NJ 08901, USA. Email:
chang@econ.rutgers.edu
In flation Targeting, Reserv es A ccumulation, and
Exc hange Rate M anagem en t in Latin Am erica
∗
Roberto Chang
†
Rutgers Univ ersit y and NB ER
Revised Draft, 31 O ctober 2007
1. Introduction
An increasingly dominan t view on monetary policy suggests that central banks should adopt
inflation targeting (IT) regim es, as advocated b y Svensson (1999, 2000), Woodford (2003),
and others. Following this adv ice, sev eral Latin Am erica n central bank s now claim to hav e
established IT regimes. T his process has coincided with a period of low and d eclining inflation
in most countries, which has been cited as prima facie evidence of the superiority of IT o ver
other monetary rules.
How ever, actual IT regimes in Latin America appear to be rather differen t from their
theoretical coun terp arts, most mark edly on the exten t to which cen tra l banks have engaged
∗
Financing for this paper was kindly provided by the Latin American Reserves Fund (FLAR) and Banco
de la Republica de Colombia. I thank Hernán Rincón, Lars Svensson, Saki Bigio, Ricardo Meirelles Faria,
Vitoria Saddi, seminar participants at Villanova, and participants at the FLAR Conference on International
Reserves in Middle- and Low-Income Countries for advice and suggestions. Adam Gulan provided excellent
research assistance. Of course, all errors and omissions are mine alone.
†
Departmen t of Economics, Rutgers Universit y, New Brunswick NJ 08901, USA. Email:
chang@econ.rutgers.edu
on foreign exc ha nge interven tion and, m ore recently, accum u lated foreign exc ha nge reserv es.
Conv entional theory prescribes that an ideal IT regime should have an inflation target as its
primary objective
1
and, therefore, should not simultaneously pursue an exch ang e rate goal. It
follows that the exchange rate should be allowed to float. In contrast, Latin American central
banks have interv ened activ ely to stabilize foreign exchange rates. In addition, they have
recen tly increased their stoc ks of foreign exc ha nge rate reserves drastica lly. This represents
a challenge for dom in ant IT theo ries, whic h are t yp ically silent about the role of foreign
reserv es.
The objective of this paper is to documen t this state of affairs, iden tify possible interpre-
tations, and deriv e implications for IT and monetary m ana gem ent. To this end, the paper
includes a review of the inflation targe tin g experiences in Brazil, Colombia, Chile, and Peru,
with emphasis on the actual details of, and rationale for, reserv e accumulation and foreign
exc h an ge in terven tio n. Having presented the stylized facts, the paper th en discusses some
theoretical arguments that m ay explain the discrepancies between conventional IT regimes
and actual IT in Latin Amer ica.
To be sure, a con ven tio nal IT regime does not preclude exchange rates from ha vin g an
influence on monetary policy. As Svensson (1999) emphasizes, in an optimal IT regim e the
policy instrum ent should respond to any variable that marginally helps predictin g inflation.
Hence, m oneta ry policy ma y ha ve to respond to exchange rate m ovements if the latter are
correlatedwithfutureinflation, which is likely to be the case for most Latin Am e r ican
countries. This analysis is summ arized in Section 2 of this paper.
Section 3 reviews episodes in Bra zil, C olombia, Chile, and Peru and reveals, ho wever,
that actual policymaking departs from the preceding ideal view in substantial w ays. Latin
American cen tral bankers ha ve often reserved and exercised the right to adjust interest rates
1
This is not to say that the central bank cannot have objectiv es other than inflation. Ho wever, in ideal
IT regimes those other objectives are subordinate to inflation.
2
to influence exchange rates, not only when the latter hav e signalled impending inflationary
pressures, but also to curtail " ex cessive v olatility" in the foreign exc h an g e ma rket and
to "calm financial markets." In addition, Latin central banks have engaged in a pattern of
accumulation of international reserves with the stated objective of prev enting financial crises,
dealing with "sudden stops," and preven ting "excessiv e appreciations. "
According ly, Section 4 reviews some possible rationales for the actual practices of Latin
Am erican IT cen tr al banks, namely:
• The combination of "dollarized" debts, balance sheet effects, and financial imperfec-
tions ma y exacerbate and/or reverse the effects of exc hang e rate fluctuations on aggre-
gate demand, emplo y m ent, and income. If domestic firms ha ve net debts denominated
in foreign currency, a fall in the exc hange rate will significantly reduce corporate net
worth which, if financial frictions are presen t, may lead to a cut of foreign financing.
The final result is a halt to investment and a contraction of aggregate demand. This line
of argument strongly suggests that exc hange rate fluctuations may imply m uch larger
costs than usua lly recognized and, accordingly, that central bank er s should in tervene
often to limit those fluctuations (Calvo 2006).
• Recen t financial and exchange rate crises in Mexico 1994-95, Asia 1997-98, Argen tina
2001-2, and other emerging economies have been attributed, to a large exten t, to
a shortage of in ternational liquidity, that is, a large excess of potential short term
liabilities over assets in in ter nation al currency (Chang and Velasco 2000). Th e curren t
accumulation of reserves in Latin America and other emerging regions appears to be,
at least partly, an attempt to increase international liquidit y lev els to m inim ize the
lik elihood of crises.
• While IT theories are built upon rational behavio r, theoretical models have failed at
providing a convincin g account of exc h ange rates. He nce central bankers may justifiably
3
worry about the possibility of irrational exchange rate ma rket beha vior , bubbles, or
multiple equ ilib ria, and choose to limit exchange rate movements that may be out of
line with equilibrium fundamentals.
• Con ven tional IT argumen ts are built upon the implicit assumption that the cen tral
bank con trols only one instrument, usually a short term in terest rate or a monetary
aggregate. How ever, recent renew ed in terest on portfolio balance models adm its the
possibility that sterilized foreign exc hang e in te rvention may have real effects. If so,
exchange rate managemen t amoun ts to an independent instrument of policy, which
then could and should be taken advantage of as part of the o verall monetary strategy
(Bofinger and Wollmershauser 2003).
• Exchange rate flu ctua tion s m ay have strong distributional effects. Appreciations, in
particular, reduce the profitabilit y and competitiven ess of export and im port com pet-
ing industries, and ma y require costly adjustm ent in terms of reallocating labor and
other factors to alternative occupations. Not surprisingly, som e recen t experiences
of exc h ange rate appreciation hav e witn essed strong political pressures for the mon-
etary authorities to interven e to limit and reverse the appr eciation , resulting in the
accumulation of foreign reserves.
The discussion in Section 4 highlights that, at different times, uncon ventional exch an ge
rate and reserves management may ha ve responded to one or more of the concerns just
mentioned. But Section 4 also finds that, while eac h argument is intuitiv ely plausible, its
policy implicatio ns for the episodes reviewed in this paper appear somew h at limited and are
rarely straightforward.
4
2. Exc hange Rates in a Canonical Inflation Targeting Framew ork
To discuss ho w actual exchan ge rate m anag em ent by Latin American IT cen tral banks ma y
ha v e departed from con ventional IT, it will be useful to ha ve a specific view of the latter.
For our purposes, Svensson’s (2007) characterization of IT for the forthcomin g N ew Palgrav e
Dictionary of Econom ics is a suitable one:
Inflation targeting is a monetary policy strategy that w as intr oduced in New
Zealand in 1990...c haracterized by (a) an announced n um erical inflation target,
(b) an implementa tion of m on etar y policy that gives a major role to an inflation
forecast and has been called "inflation forecast targeting", (c) and a high degree
of transpa ren cy and accountability.
In practice, the num er ical inflation target is chosen in advance b y the go v ernmen t and/or
the central bank. Hence, in an IT regime, the central bank is not "goal independent" since its
ultimate objectiv e is fixed.
2
How ever, the cen tral bank is "instrument independen t," in the
sense that it is free to try to attain the inflation target b y adjusting an y policy instrument
it con trols. Often, the main operating instrument has been an overnight interbank interest
rate, although con tr ol of a m onetary aggregate has also been observ e d in IT experiences (e.g.
P eru bet ween 1994 and 2001).
In an inflation targeting regime, the inflation rate is (of course) the main target of policy.
In fact, in the Latin American episodes to be review ed below, the charters establishing IT
regimes clearly men tion that inflation should be the o verriding goal of m onetary policy, if
not the only one. Hence one could restrict discu ssion to the case in w hich IT is "strict,"
meaning that inflation is the only goal. Ho wever, the relevan t theory is easily extended to
thecaseof"flexible" IT regim e s, wh ich are concern ed not only with inflation but also with
2
The inflation target can, of course, be modified, but not in the short run. Peru’s central bank, for
instance, set an inflation target of 2.5 percent at the start of 2002, and only changed to 2 percent in 2007.
5
other indicators of aggregate real activity, such as the output gap or the unemplo yment rate.
This is of interest partly because, as Woodford (2003 ) has shown in a variety of contex ts,
the ma xim iza tion of the we lfar e of a t y p ical citizen is equivalen t, in some cases, to the
minimization of a loss function quadratic in inflation and the output gap.
In an y ev ent, in a con ven tiona l IT regime, the central bank is directed to minimize a
function of inflation and possibly the output gap or an alternativ e indicator of activit y. In
contrast, and significantly for our purposes, the ultimate goals of policy do not include other
variables, including real or nominal exc h ange rates, or in terna tional reserve lev els.
Does this view mean that neither exc han ge rates nor intern ationa l reserves play a ro le
in monetary policy m akin g? Not necessarily. But the view does mean that such a role, if
it exists, is limited to the w ay exchan ge rates or reserves either affect ho w monetary policy
translates into current and future inflation and output gaps, or provide informa tion about
futur e values of inflation and output gaps.
To understa nd this important result, it is con venient to refer to a styliz ed view of optimal
policy in an inflation targeting regime. A canonical formulation is developed in Sv ensson
(1999). In it, the central bank is man dated to minim ize a loss function of the form :
E
t
∞
X
s=0
δ
s
L
s
where E
t
(.) denotes expectation conditional on inform ation as of period t, δ is a discount
factor bet ween zero and one, and L
t
is a linear quadratic function of a vec tor of variables Y
t
from some specified targets. (A key example is
L
t
=(π
t
− π
∗
)
2
+ λy
2
t
6
where y
t
denotes the output gap, π
t
denotes inflation, and π
∗
the inflation target. The
parameter λ gives the relative importance of inflation v ersus output gap stabilization.)
In each period t, the cen tral bank chooses the curren t and future setting of a policy
instrument, i
t
, to minimize the loss function subject to the la w s of m otion of the economy,
given b y a linear m odel:
⎡
⎢
⎣
X
t+1
E
t
x
t+1
⎤
⎥
⎦
= A
⎡
⎢
⎣
X
t
x
t
⎤
⎥
⎦
+ Bi
t
+
⎡
⎢
⎣
υ
t+1
0
⎤
⎥
⎦
In the preceding equation, A and B are matrices of constants, whic h reflect the structure
of the econom y ; X
t
is a vector of predetermined state variables, suc h as capital; x
t
is a
v e ctor of jumping variables, such as exc ha nge rates; and {υ
t
} is a v ecto r of shoc ks, assumed
to have mean zero and to be independen tly and identically distributed. In this notation, the
ultimate goal or target variables are those whic h en ter the cen tral bank’s loss function (the
components of Y
t
), and are linear combinations of X
t
,x
t
, and i
t
.
3
In this context, Sv ensson (1999) sho ws that the optimal monetary policy is given b y a
system of equations of the form:
G
t
(Y
t
,E
t
Y
t+1
,E
t
Y
t+2
,...)=0
in volving the current value and future expected values of the tar get variables only.
4
At least
in pr inciple, this system can be solv ed to yield a path for Y
t
,E
t
Y
t+1
,E
t
Y
t+2
, and so on,
representing the optimal expected path of the target variables. So, in this analysis, the first
3
Note that this structure includes a wide class of dynamic stochastic equilibrium models. Hence, the
argumen ts to follow apply to that large class, and it will not be necessary to set up a more specificopen
economy model.
4
Note that the central bank is assumed to be able to commit to a time and state contingent policy at
t =0.In the absence of commitment, time inconsistency emerges, and it is not clear how to find a satisfactory
solution.
7
task in a monetary policy exercise is to iden tify the most desired values for the current value
and expected future values of target variables. The current value and future expected values
of the policy instrument (i
t
, E
t
i
t+1
,...) must then adjust so as to deliver the desired path for
thetargets.Thisiswhythesolutionissometimescalled "inflation forecast targeting".
These results imply that an y variable not included in the vector of targets Y
t
can affect the
current setting of the monetary instrument, i
t
, as long as that variable affects the forecasts
of the target variables. Moreov er, this is the only w ay in which nontarget variables should
affect the optimal instrum e nt path.
The case of most interest in our con text is when one of the components of the vector Y
t
is the inflation rate (that is, inflation is one of the ultimate goals of policy) but does not
include the exc h ange rate. For that case, t wo implications of the preceding analysis ha ve
special relevance for our subsequent discussion:
1. Even if the exchange rate is not an ultimate target of policy, setting the central bank
instrument optimally requires responding to information embodied in the exchange rate,
if that info rm a tion is marginally relevan t in foreca sting inflation. Hence, observing that a
cen tral bank, say, raises interest rates in response to an excha nge rate depr eciation does not
necessarily mean that the central bank is pursuing a target other than inflation. Suc h a
policywouldbeconsistentwithstrictinflation targeting if, for example, a depreciation is
expected to result in an acceleration of inflation above the target.
This observation, inciden tally, means that iden tifyin g wheth er m oneta ry policy is de-
parting from strict inflation targeting is more difficultthanhasbeenacknowledgedinthe
literature. Often , answ ers to this question has been based on the econometric estimation of
policy rules of the form
i
t
= α + βπ
t
+ γs
t
+ ...
8
say, where s
t
is the exc hange rate (nominal or real), and α, β, and γ are parameters to be
estimated.
5
It is not too uncom mon for γ to turn out to be statistically significant. In that
case, a pitfall in the in terp reta tion of the results is to conclude that the monetary authority
is targeting the exchang e rate. As we have seen, it is possible for in flation to be the only
target, yet for the optimal setting of the instrument i
t
to depend on the exc hange rate.
2. Iftheexchangerateisnotoneoftheultimatepolicytargets,however,optimal
policy should respond to exchange rates only insofar as the latter affect target forecasts.
In particular, under strict inflation targeting, allow ing for m one tary policy to depend on
exc han ge rate shocks can be justified only on the basis of such shocks having significant
implicatio ns for future expected inflation.
The latter implies that the central bank should also explain ho w exc hange rate shocks
are expected to affect inflation forecasts, particularly since "a high degree of transparency"
is a defining c haracteristic of inflation targeting. Typically, this m ust be done by disclosing
the econom ic model that the central bank uses in order to produce those forecasts.
Wh ile the central bank can in principle em b race any model for its purposes, in practice
inflation targeting regimes (and also the theoretical literature on IT) ha ve adopted versions
of what is kno w n as the "New Open Econom y Macroeconomics " paradigm. In that para-
digm,exchangeratenewsaffect forecasts of future inflation through a few specific channels,
including:
• Exchange rate shocks affect the prices of some imported goods that are included in the
consumer price index (this is know n as the direct channel);
• Exchange rates also affect the cost of imported in term ed iate inputs to dome stic pro-
duction, which in turn ma y affect aggregate supply relations between inflation and the
output gap;
5
See, for example, Mohanty and Klau (2004).
9
• Real exc h ang e rate movem ents, whic h may be due to nominal ones, can affect the
relative dema nd for do m estic ally produced goods vis a vis foreign goods, therefor e
affecting aggregate demand;
• Finally,exchangerateshocksmayaffect domestic interest rates through interest parit y
conditions, and hence in vestmen t demand.
Hence, one should expect IT cen tral bankers to justify an y policy reaction to exc ha nge rate
dev elop ments on the basis of one of the argume nts above. Notably, references to exc h ange
rate v o latility, market uncertainty, and the like , are absent in this fram ework. This is partly
because of the theory’s restriction to a linear quadra tic framework but, mor e importan tly,
to the fact that exc ha nge rates play a relatively restricted role in the New Open Economy
Macroeconom ics parad igm . Lik ew ise, there is no role for nor an y m ention of a need for
in tern a tional reserv es accumulation.
This analysis is the basis for the usual opposition, from IT proponents, to ha ve a monetary
policy response to exc h ange rate shocks over and abo ve the implications of those shoc ks on
future expected inflation . An d, clearly, stabilizin g exc ha ng e rates for its own sake is viewed
as being inconsistent with inflation targeting. M ishkin and Schm idt-Hebbel (2002), for
example, argue that
Thefactthatexchangeratefluctuations are a major concern in so man y
coun tries raises the danger that monetary policy ma y put too muc h focus on
limiting exc ha ng e ra te mo veme nts, even under an inflation targeting regime. The
first problem with a focus on limiting exc hange rate movements is that it can
transform the exchange rate into a nom inal anc hor that takes precedence over the
inflation target...T he second problem ...is that the impact of changes in exchange
rates on inflation and output can differ substantially depending on the nature
of the shock that causes the exc ha nge rate movem ent...Targeting the exc hange
10
rate is thus likely to w orsen the performance of m onetary policy. This does not
imply, ho wever, that cen tral banks should pay no attention to the exc han ge rate.
Theexchangerateservesasanimportanttransmissionmechanismformonetary
policy , and its level can have important effects on inflation and aggregate demand,
depending on the nature of the shock s. This is partic ular ly true in small, open
econom ies.
Likewise, in an independent review of the Central Bank of Norw ay (Norges Bank Watch
2002) one reads:
Inflation targeting in an open economy will include an element of implied
exchange rate stabiliz atio n...Importantly, this implie d exchange rate stabilization
is not for its own sake; it is der ived from the objectiv es of stabilizing inflation
and the output gap.
A separate issue is the desirability and feasibility of independent stabilization
of the exchange rate...Ex cep t in situations of fin an cial fragility with concern s
about the stabilit y of the paymen t and financial system, w e find it difficu lt to see
good reasons for such stabilizatio n at the cost of increased inflation and output
gap variabilit y.
The above considerations concern the optimal setting of the monetary policy instrument,
which is t ypically an interest rate or a monetary aggregate. In contr ast, sterilized foreign
exc h ange interven tion receives virtually no attention b y the conventional wisdom . This is
embodied on the fact that the models that dominate IT theories do not describe the asset
side of the central bank explicitly, reflecting the view that the division of the cen tr al bank’s
portfolio between foreign currency assets (reserves) and domestic currency assets is of no
consequence. Th is view can be giv en microfound ations, although it requires a set of implicit
11
or explicit assu mp tion s that ma y not be trivial.
6
Of cour se, the pra ctical justification for
ignoring the possibility and im plica tion s of sterilized foreign exc ha ng e interv ention is the
substantial body of empirical w ork that has failed to find systematic and du rable effects of
such inter ven tion in the data.
3. The L a t in Ame r ic an E xperien ce
To examine how Latin American inflation targeting regimes compare with the con v entional
IT framework of the previous section, w e now turn to a description of observ ed experiences
in Colombia, Chile, P eru , and Brazil. Our review is (necessarily) rather selective, and
empha sizes those episodes in which departures bet ween the con ven tion al IT w isdom and real
world policy appears to be greatest. We also try to identify what is the "official " position
of each IT regime with respect to foreign exchange in terven tion , exchange rate management,
and reserves accum ulation , so we can ascertain whether official deeds have follo wed words.
3.1. Colombia
Background. Colombia started imp lementing an inflation targeting regime after the 1991
Constitution and the 1992 Cen tral Bank La w (Ley 31) ga ve the central bank (Banco de la
Republica) the chief objective of "main taining the purchasing value of the peso," gran ting
it independence for this purpose. In addition, Ley 31 stated that the Banco de la Republica
must "ado pt specificinflation goals".
But arguably inflation targeting was fully em braced only in 1999 when, as a consequence
of the Russian crisis and the Brazilian devaluation, the Banco de la Republica announced
that it w ould allo w the exc hange rate to float. This reform replaced a system of preannounced
6
One such set is that the economy is small in world financial markets, that the central bank cannot issue
securities that are not available from the world capital market, and that domestic residents have unrestricted
access to the latter.
12
exchange rate bands that had been in place since 1994 and w as subject to speculative attac ks
during 1998-99. Vargas (2005) notes that, in the latter period, in ternation al reserves fell by
eighteen percen t, and the Banco de la Rep u blica had to sell US$ 2.6 billion to defend the
exc h ange rate bands. Yet, the pressu re on the peso did not subside, and the mid points of
the exchange rate bands had to be realigned twice, in Septem ber 1998 and June 1999, each
time by nine percent.
Finally, in September of 1999 the exc han ge rate bands were dropped and the Banco de
la Republica announced that, from then on, exc hange rates w ere to float. This commitment
was included a few weeks later (December 1999) in a Letter of Intent submitted to the IMF
in the context of a three ye ar Extended Fund Facility credit.
As stated in the Mem oran du m of Econ om ic P olicies attac hed to the Letter of Intent,
floa ting rates were to be implemented to help m oneta ry policy continu e the process of dis-
inflation:
Thecentralbankexpectsthatallowingthemarkettosettherateforthe
peso will enable the Colombian economy to better absorb external shocks and
facilitate the conduct of m on eta ry policy, as w ell as reduce speculation against
the curr en cy, while allowing the central bank to continu e its disinflatio n policy.
But, at the same time, the Banco de la Rep ublica explicitly "reserv ed the righ t to
interven e " to prevent unw a rranted v olatility in the foreign exch ang e market and to help
rebuilding the stock of international reserves:
Under the floating exchange rate regime, interven tion b y the Banco de la
Rep ub lica in the foreig n exc h a ng e market will be limited to ma intaining orderly
market conditions by smoothing short-term movem ents in the nominal exch ange
rate consistent with the program projections for the net in t ernatio na l reserves.
7
7
Colombia Letter of In t ent, December 3, 1999. The quotes here are from paragraph 26 of the Memorandum
of Economic Policies.
13
The claim for a righ t to interve ne has persisted since then. Indeed, the Banco de la
Repub lica w ebsite describes the objective of its exch ange rate policy in the following terms:
The strategy of mon etary policy has been implemented within a regim e of
flexib le exchan ge rates, subject to interven tion rules through wh ic h the follow ing
objectives have been sought:
* To main tain an adequate level of international reserves that reduce the
vulnerabilit y of the economy to foreign shocks, both on current account and on
capital account;
* To limit excessiv e vo latility of the exchange rate at short horizo ns, and
* To moderate excessiv e appreciations or depreciations that endanger at-
taining future inflation targets and the financial and external stability of the
economy.
8
The Inflation Targeting Regime. How, then, has inflation targeting in Colombia w o rked?
Chart 1 shows the ev olu tion of CP I inflation for the last twenty y ears.
9
As shown by the
c h ar t, the forma l adoptio n of inflation targeting in 1991-2 was preceded b y a period of
relatively high inflation, at rates around thirty percent per y ear. Since then, the inflation
rate has fallen gradually, to its curren t lev el of four and a half percent per y ear. Notably,
the fall in inflation has been more drastic since the start of floating exchange rates in 1999
(the subperiod sho wn as the shaded area in the chart). During that year, the inflation rate
fell to 9.2 percent, from 16.7 in 1998.
The reduction of inflation appears to be associated with the beha vior of the output gap
in the expected w ay. Betw een 1992 and 1995, GDP gro w th was better than fiv e percent per
year, and the inflation rate, while falling, was sluggish. A big fall in inflation from about
8
From the "Foreign Exchange Policy " entry of the Banco de la Republica web site, at
h ttp://www.banrep.gov.co/politica_cambiaria/index.html
9
Charts can be found at the end of the paper. Data for Charts 1-3 and 5-6 were taken from the Banco
de la Republica’s web site (www.banrep.gov.co). The source for Chart 4 is JP Morgan.
14
17percentin1998to7percentin2002wascoupledwithsubpargrowth. Thisisshownin
Chart 2, whic h decomposes 1996-2007 GD P quarterly gro w th between its H odrick-P resco tt
trend component and a cyclical component. Annual GDP growth was only one half of one
percent in 1998 and then fell dramatically, by 4.5 percent, in 1999. Recover y was very tepid
up to 2003: GDP gro w th was only 2.9 percent in 2000, and less than two percent in 2001
and 2002. Since 2003, ho wev er, gro wth has been rather robust, as evident from Chart 2.
Since 1999, the m ain instrument of monetary policy has been the overnight repo rate,
which has gen era lly followe d a down ward trend. This is sho w n in Chart 3. Relative to
inflation rates, interest rates w ere kept at a rather high lev el until July 2001, implying a
real interest rate of about four percent. In the following year the repo rate w as lowered b y
625 basis points, perhaps reigniting growth. Since then, the repo rate hovered bet ween 5.25
percent and 7.5 percent un til about a y ear ago, when a series of increases started aiming at
prev en ting overheating and bringing inflation down to target.
In ter est rate adjustm ents ha ve, therefore, been broadly consistent with a conven tion al IT
frame work. The Banco de la Repub lica, however, has exercised its option to in tervene in the
foreign exchange market in response to two episodes of significant foreign shock s.
200 2-3 : Political Uncertainty and Volatile Capita l Mar kets. The first episode w as asso-
ciated with the Brazilian presidential elections of 2002, which resulted in an increase in the
Braz ilian EMB I spread to almost 2500 basis poin ts, and political uncerta inty in Venezuela
following the popular revolt against Hugo C h av ez in A p ril 200 2. While m ost o ther Latin
Am erican economies w ere relativ ely unaffected b y the B razilian shock, C olombian spread s
increased b y almost 600 basis poin ts, probably reflecting the mark et expectations about the
impact of the Venezuelan turmoil on Colo mbia (Chart 4).
10
10
To be sure, events in Venezuela affected Colombia through several channels in this period. Weak oil
prices, for example, led to a fall in aggregate demand in Venezuela and to a reduction in Venezuela’s demand
for Colombian exports. This trend was exacerbated by the imposition of capital controls in Venezuela,
starting early 2003. Yet, as apparent from Charts 3 and 4, the sharpest movements in exchange rates and
EMBI spreads took place during the summer of 2002, thus following the political shocks emphasized in the
15
As shown in Chart 3, the Colombian peso depreciated sharply, losing thirt y percent of its
value between March of 2002 and March of 2003. One consequence w as that tradable goods
prices had to increase, pushing infla tion abo ve the target range of six percent for 2002 and
five and a half percen t for 2003. Indeed, ann ual CPI inflation w as to rise abo v e seven and a
half percen t b y mid-2003 .
The Banco de la Republica response had two dimensions. The first one was conventional:
repo rates, wh ic h had been follow ing a downward path since the beginning of 2001 and un til
June of 2002, we re increased b y 100 basis poin ts in Jan u ary 2003 and again in April of 2003.
The uncon v entional dimension was that the Banco de la Republica sold foreign exchange
call options between July 2002 and M a y 2003 for about US$ 745 million, through a pre-
announced auction mechanism that had been established after the peso w as floated. The
auctions that took place in 2002 were part of an automatic procedure that had been estab-
lished in 1999 to respond to exchan ge rate volatilit y. On the other hand, the 2003 auctions,
announced in February of that year, w ere discretiona ry mov es b y the Ba nco de la Republica,
and were explicitly intend ed as a response to expectations of higher inflation and deprecia-
tion.
11
It is noteworth y that the am ounts inv olved w ere v er y small r elative to the foreign reserves
of the Ban co de la Re pu blica (about US$ 11 billion at the time). Also, the private sector’s
appetite for the options was limited. As discussed b y Vargas (2005 ) and the Marc h 2003
Report from the B anco de la Republic a to the Congress, the auctions mec h anism stipulated
that there was a set and prea nnou n ced maximum amount of international reserves to be
sold. For the second half of 2002, the amount was $ 540 million, of wh ich US$ 414 w ere sold.
For the first half of 2003, the amount w as US$ 1 billion, of wh ich only US$ 345 million w e re
sold. (Cha rt 5)
text.
11
See, for example, section 2 of Banco de la Republica’s Informe sobre inflacion, Marzo de 2003.
16
By the end of 2003, the Banco de la Republica appeared to ha ve succeeded both in stop-
ping the deprecia tion of the peso and in reducing infla tion to target levels. The peso-dollar
exc h ange rate stabilized around February of 2003 and settled on a decidedly appreciating
trend from October 2003 on. Annual CPI inflation fell below 6.5 percent by October 2003,
and has continued to fall until very recently .
In discussing this episode, Vargas (2005) argues that
In terven tion of the central bank in the forex market wa s a very useful com-
plement to in terest rate policy. It wa s perceiv ed that the increases of the interest
rates alone were not sufficien t to correct the depreciation of the Col peso or to
curb inflation expectations. In other wo rds, the exclusive use of in terest rates
would ha ve required m uch larger movements than those observed, pr oba bly in-
troducing inefficien t output v olatility.
How ever, it is not clear exactly how forex in tervention m ay have been operated as a
comp leme nt to repo rate increases. By themselves, the increases were bound to strengthen
the peso, and at the same time put a brake on real activity and rein on inflation. So, one
could argue that the foreign exchange auctions w ere effe ctive simp ly because they signaled
future increases in repo rates. The sm all size of the auctions is consisten t with such an
in terp retatio n, as is the fact that the intervention of February 2003 w as discretionary and
follow ed b y a relativ ely large repo increase (100 basis poin ts) in April.
12
On the other hand,
the fact that the 2002 auctions were automatic rather than discretionary suggests that those
auctions could ha ve not been interpr eted as informative signals by the public.
2004 to Present: Persistent Real Appreciation. A second episode has been the strong
and contin ued appreciation of the Colombian peso since 2004. Several developments appear
12
Notice that this argument is consistent with the view that the repo rate was adjusted solely to attain the
inflation target, as seems to have been the case. But the intervention may have strenghtened the market’s
belief that more repo increases were coming.
17
to be behind pressures for the peso to appreciate. In Venezuela, Hu go Chav ez consolida ted
power b y defeating a recall referendum in August 2004. Colombia’s terms of trade k ept im-
proving. Domestic confidence improved, perhaps due to a com bination of impro v ed security
and macroeconomic stabilit y, which resulted in a vigo rous expansion of domestic demand,
led b y private in vestment (C hart 6).
13
Foreign direct investm ent has also increased as the
result of recent privatizatio n efforts. Transfers from abroad, especially work ers’ rem ittances,
increased significan tly during the period.
14
Finally, world in terest rates and risk premia
k e pt falling, reflecting the now famous "savin gs glut."
The exchange rate fell from 2790 pesos per US dollar to 2350 pesos per dollar between
the end of 2003 and the end of 2004, a 18.7 percen t appreciation . A policy response to the
appreciation w a s then deemed to be necessary for at least t wo reasons: the appreciation wa s
expected to reduce inflation significantly belo w the 2004 inflation target of 5.5 percent; and,
also, a real peso appreciation w as thought to be detrimen tal to the external competitiven ess
of Colombian exports.
The initial response of the Banco de la Republica was, again, two pronged. Auctions of
foreign exchange put option s were sold in January 2004 (for US$ 400 million) and A pril 200 4
(for US$ 20 0 million). This was followed by reductio ns in repo rates of 25 basis poin ts in
February and again in Marc h 2004.
This time, ho wev er, the policy response prov ed to be insufficient to stop the appreciation
of the peso. As a consequence, the Banco de la Republica announced its inten tion to purc hase
up to US$ 1 billio n d urin g the last quarter of 200 4. Significantly, this am ou nt w a s to be
bought under a new "discretionary" mechanism which freed the Banco de la Republica
from the constraints that had shaped foreign exchange auctions. A t the end of 2004, the
last constraint was dropped, as the Banco de la Republica announced that discretionary
13
As percentage of GDP, investment jumped from 15% in 2003 to 26.3% in 2006.
14
According to the Informe al Congreso de Julio 2007, workers’ remittances jumped from an annual
av erage of US$ 2.3 bn. in 2000-3 to US$ 3.7 bn. in 2004-6.
18
intervention would no longer be subject to limits in either transa ctions amounts or time
horizon.
Repo rates were reduced b y 25 basis points in Decem ber 2004, and by 50 more basis
points during 2005. These reductions were clearly insufficien t to curb the strength of the
peso, and in response the Banco de la Republica con tin ued purchasing dollars at health y
rates. Foreign exchange purc hases reached US$ 2.9 bn. in 2004 and US$ 4.7 bn. in 2005.
After purchasing US$ 1.2 bn. in the first quarter of 2006, the Banco de la Republica
stopped discretionary intervention in response to the rev ersa l of capital inflo ws in the second
quarter of 2006, during which the peso lost 15 percen t of its value. As is well known now,
foreign appetite for risky investments in emerg ing ma rkets dropped in that qua rter, due
to uncertain ty about the future course of the United Stated economy and Federal Reserv e
policy. This episode, howev er, proved to be short lived, and since the third quarter of 2006
the peso has settled on a path of frank appreciation. A t the end of June 2006, the exchange
rate was 2633 pesos per US dollar. By the end of Ma y 2007, it w as 1930 pesos per dollar
(an appreciation of 36 percent).
The recent peso appreciation has defied very large in terven tio n by the Ban co de la Repub -
lica (Charts 3 and 5). Between January and Ma y 2007, foreign exchange purc hases totaled
US$ 4.9 bn,
15
increasing the Banco’s stoc k of net foreign reserves by about one third.
In this case, and in contrast with previous experiences, the Banco’s dollar purchases ha ve
occurred against a backdrop of tigh tening monetary policy. The repo rate, whic h stood at
6 percent at the beginning of 2006, as been increased sev eral times since, and stood at 9
percen t as of July 2007. The rationale has been that Colom bia’s economy may be on the
verge of overheating, which would endanger the attainmen t of the inflation target for 2007.
And ind eed inflation has been accelerating recently, and ma rkets increasingly doubt that the
Banco de la Rep u blica will succeed in keeping 2007 inflatio n within its target range of 3.5-4 .5
15
Of which US$ 4.5 were discretionary purchases.
19
percent.
In fact, there seems to be a consensus now amo ng mark et analysts that the Banco de
la R ep ub lica is pursuin g not only an inflation objective, but an exc h ang e rate objectiv e
as well.
16
And, in a break with the past, these two objectiv es conflict. In this period,
the Ban co apparen tly used the in ter est rate as instrument to target inflation, and foreign
exc h ange in terven tion to target the excha nge rate. (Discretionary in tervention w as halted,
ho wev er, after May 2007.)
Current forecasts are that the Banco de la Republica will have to increase in ter est rates
even more in order to bring inflation under con trol. Giv en this, and not too surprisingly,
foreign exch an ge in terven tion has failed to prevent peso appreciation . In response, the B anco
de la Rep ublica has considered more radical measures, starting with the imposition of cap-
ital con trols in May.
17
Predictably, the mark et reaction to this mo ve has been mixed,
although there appears to be a widesp read belief that the existing capital controls will also
be insufficient to rev ert the appreciation of the peso.
A poten tially significant issue raised by the massiv e dollar purc ha ses b y the Ba nco de la
Republica is the impact on monetary and credit aggregates.
18
To a large extent, the Banco
de la R ep ub lica has mana ged to amelio rat e the expansiona ry implica tions of in tervention
b y selling foreign exc hang e for the Treasury to prepay external debt (US $ 3.5 bn in 2005
and US $ 1 bn in 2006). In addition, the Treasury’s deposits at the Banco de la Republica
increased significantly since 2003.
16
See the analysis by various investment banks at http://www.rgemonitor.com/367/americas
/andean_countries_
(colombia_ecuador_peru_and_bolivia)/?more=cluster&cluster_id=1929#11
17
Since early May, the Banco requires forty percent of all dollar borrowing from abroad to be held in non
in terest bearing accounts. This requirement was extended to include portfolio flows in late May.
18
Of course, dominant theories of inflation targeting emphasize that, under interest rate control, the
beha vior of monetary and credit aggregates has little bearing on the determination of inflation (Woodford
2003), and the Banco de la Republica’s management of repo rates seems consistent with the pursuit of its
inflation targets. However, it is fair to say that rapid monetary and credit growth remains a source of concern
in actual central banking practice.
20
Also to reduce the impact of credit growth, the Banco de la Republica decided in May
to increase reserv e requirements for ban k deposits (from 13% to 27% for checking accounts,
7% to 12.5% for savings accoun ts, and from 2.5% to 5% for time deposits).
It is too early to tell how this episode will end. But, as our discussion sho w s, recent
foreign exc hange intervention has worked at cross purposes with monetary m anagement,
raising serious questions about the abilit y of the Banco de la Republic s to attain its 200 7
inflation targets.
19
3.2. Peru
Background. Although the Peruvian Banco Cen tral de Reserva started announcing one y ear
inflation targets in 1994, it mo ved towards an explicit inflation targeting regime in January
2002. Then, the 2002 Monetary Program stated that the target for inflation would be 2.5
percen t, with a tolerance band of plus or minus one percentage point. That target w as to
be maintained until January 2007, when it w as lowered to 2 percent.
Under inflation targeting, the main monetary policy instrum ent has been the interest
rate for o vernigh t interbank loans. The 2002 Monetary Pr ogr am envisioned that interbank
rates would be adjusted solely to attain the inflation target, and hence the exc hange rate
had to float:
The management of the operational target is orien ted to wards attaining the
inflation target and not an y other nom inal variable suc h as a monetary aggregate
or the level of the exc ha nge rate. Therefo re, this regime [of inflation targeting] is
only compa tible with floating exchange rates.
In line with conve ntional views on inflation targeting regimes, the 2002 M onetary Program
did allow for monetary policy to respond to exc hange rates, but only as long as the latter
19
Again, see the market commentary available at the RGE web site, cited in footnote 16.
21
affected prospectiv e inflation:
The decisions about the operational target of moneta ry policy will be based on
an integral assessment of the determina nts of inflation, for w hich we will consider
surveys on inflatio n expectations, the ev o lution of economic activity, and the
exch ange rate, mainly.
However, this position w as quickly dr opped in fav or of intervening in the foreign exch ang e
market in case of need. The Septemb er 2002 Inflation Report stated this option explicitly:
In an exceptional w a y and ev en if there ma y not be inflationary pressures, the
Banc o Central can modify the availability of liqu idity for the banking system ,
the reference in terest rates, or intervene in the foreign exchang e market, in order
to ameliorate abrup t fluctuations in the exc hange rate or to preven t speculative
bubbles that may affect the normal performance of financial mark ets and the real
economy.
In fact, the Banco Central has been quite forthcoming in reporting its foreign exchange
intervention operations since 2002. At various times, it has justified interv en tion on at least
t wo related but conceptually separate ground s. First, as already mentioned, the Ban co
Central has often cited the need to curb excessively volatile exchan ge rates. Th is is not too
different from practice by other central banks. But Peru’s Banco Central has stressed that
the large exchange rate fluctuations, or speculative bubbles, may be more damaging for P eru
than for other coun tries because of the extent of financial dollarization and the associated
balance sheet effects:
Central Bank in terven tion in the foreign exc hange market seeks to minimize
the probability that an abrupt increase in the exc han ge rate, through the process
22
of dollarization of the economy, result in high costs in terms of the insolvency of
firm s and of a reduced stability in the fin ancia l system. This ty pe of in terventions
are, therefore, consisten t with a prud ent monetary policy.
20
Secondly, in justifying foreign exchan ge purchases, the Banco Central has often argued
that there are k ey benefits from holding a large stock of international reserves. These benefits
include the ability of the Banco Cen tral to serve as a lender of last resor t in dollars , and
hence to reduce the lik elihood of a bank run (Armas and Grippa, 2006).
The Peruvian macroeconomy under infla tio n ta rgeting. The period just before explicit
inflation targeting, in 2002, had been c haracterized b y mild deflation. Since then, the Banco
Central has been v ery successful in k e eping Peru’s inflationratewithinonepercentagepoint
of the inflation target. Violations of the target range hav e been small and short lived . (See
Chart 7)
21
At the same time, Ch art 8 shows that real activit y ha s been exceptionally fa vorable. The
growth rate of GD P, which had been close to zero between 1998 and 2001, jumped to 5.2% in
2002 and, after falling to 3.9% in 2003, accelerated in subsequent years, reac hing 7.9 percen t
in 2006. For 2007, the Banco Cen tral curren t projection is 6.8 percen t. (chart).
Sev eral factors appear to be driving the Peruvian performa nce. Clearly, low in t erest rates
in the wo rld economy have play ed a role. However , the behav ior of the current accoun t gives
little indication of excessiv e foreign borro w ing . As displa yed in Char t 9, the curren t account,
in fact, has c hanged from a deficit of 1.9 percen t of GD P in 2002 to a surplus of 2.3 of GDP
in 2006, improving every y ear in between.
Argua bly, better terms of trade, particu larly copper prices, are better explanations of the
Peruvian expansion. (Chart 10) Also, domestic confidence has improved strongly, perhaps
follow ing man y y ears now of monetary and fiscal stability. This has been expressed in health y
20
Jan uary 2006 Inflation Report.
21
Data for Charts 7-12 are from the Banco Central de Reserva’s web site (www.bcrp.gob.pe).
23
rates of private investm ent, whic h as a percen tage of G D P increased from 14.7 in 2007 to
16.8 in 2006, and is expected to reach 18.8 in 2007, as shown in Chart 11.
In this context, the Ba nco Central has had to change interest rates relatively infrequently.
But, as w e will discuss, there have been notable episodes of foreign exc hang e inter vention.
Repercussions fr o m the 2002 Brazilian election. A first notable episode was associated
with the 2002 Brazilian election. While the impact on capital flows to Peru wa s relatively
mild, the Peru vian sol lost six percen t against the US dollar bet ween May and Sep tember
of 2002. According to the September 2002 Inflatio n Report, anxiet y about future exchange
rates built up, putting pressure on the Banco Cen tral to increase interest rates:
On September 17 an unusual tension in the mark et w as faced, caused by ex-
pectations of sol depreciation, and the in terbank rate clim bed up to 16.1 percen t.
In the follo w ing da ys, after a stabilizing interven tio n b y the Ba nco Central, the
reduction of those expectations and the return of liquidity to the banks made the
in terbank rate to return to the reference interest rate corridor.
The "sta bilizin g interv e ntion" was a sale of US$ 127 millio n in the open market. In
addition, the Banco Central increased in terest rates by 100 basis points. After these policy
mov es, financial markets stabilized. By the end ofDecember,theexchangeratewasback
to the June 2002 lev el, where it w o uld remain for all of 2003. In Nov ember 2002, the Banco
Central bought bac k US$ 100 million. Th ese operations appear in Ch art 12.
Since the Banco Cen t ral’s net in tern atio nal reserve s w ere around US$ 9.8 billion at the
time, the size of the foreign exchan ge interven tion w as decidedly small. Wh at is noteworth y
is the Banco Central’s emphasis that interv ention w as not to be taken as a departu re from
floa ting rates:
The intent of policy was to smooth exc ha ng e rate vo latility, withou t com m it-
24
ment to any lev el [of the exch ange rate].
22
In addition, the record is clear in the objectiv e of the intervention w as to calm foreign
exc h ange markets rather than to figh t any inflatio na ry implica tions of the sol depreciation .
Indeed, a discussion of such implications is notably absent from the period’s Banco Central’s
documen t s and announcem ents.
2003 -2 006 : Accumu lation of Reserves, with Elections in the Middle. Since 2003, Peru ’s
external position has been rather strong and, as mentioned earlier and sho w n by Chart
9, driven by sev eral favorable domestic and foreign factors, including improving terms of
trade, decreasing cost of external credit, ma croeconomic stability, and im proving business
confidence. The impro vement has been most remarkable in the trade surplus, whic h increased
from 0.6 of GD P in 2002 to 9.6 of GDP in 2006. The current account balance, in turn,
improv e d from a deficit of 1.9 percen t of GD P in 20 02 to a surp lus of 2.8 percen t of GD P
in 2006. As a result of the implied capital inflows, the Peruvian sol has settled on an
appreciating path against the US dollar, except for a relatively brief period surrounding the
presidential elections of early 2006.
In this context, the Banco Cen tral has been very activ e in the foreign exchange rate
market. Between 2003 and the third quarter of 2005 , the Banco Central bought US $ 6.5
billion, increasing its stock of net foreign reserv e s from US $ 9.6 billion to US $ 13.6 billion.
The rationale for this strategy is w ell summarized by Grippa and Gondo (2006):
Accordin g to the inflation reports published by the BCR P [P eru’s Banco
Cen tral] between 2004 and 2005, the Banco an ticipated risks for the econom y of
possible increases in in te rnation a l in te rest rates, whic h would increase country
risk; electoral processes in Latin Am erica bet ween the end of 2005 and 2006,
which could also increase coun try risk according to their dev elo pment, especially
22
September 2002 Inflation Report.
25
the local electoral process in 2006; and the uncertain ty of how significant was the
transitory component of the increase in term s of trade (...) These risks, according
to the BCRP, justified the accum ulation of international reserves in a preven tive
ma n ne r...
Interestingly, Grippa and Gondo (2006) argue that, in fact, this episode shows that the
inflation objective had clear priority over foreign exchange interv ention:
Inflation expectations for 2005 showed an increase between July and Augu st
of 2004...In this situa tion, the Banco increased the reference in terest rate twice, in
Augu st and October, by 25 basis point each time. A smalle r monetary stim ulu s
was inconsistent with a putative wish to m aintain the exch an ge rate above some
determin ed lev el, given the tendency to wards a reduction then experienced. Also,
the Banco reduced its dollar purch ases in the market, and the exchange rate
experienced a greater fall [appreciation].
In this w ay, in a situa tion in which there w as foreig n exchange in tervention
in a buying position ...that w e r e in conflict with fulfillin g the inflation target, the
Banco Cen tral privileged the latter.
However, the available data presents a murkier pictu re. Althoug h the sol a pprec iation
became faster in the second half of 2004, the Banco Central’s dollar purc h as es, in fact,
accelerated, as shown in Chart 12. The purchases, which had added up to US $ 840 m illion
in the first six mo nths of 2004, increased to US $ 1.5 billion during the second half of that
year.
As anticip ated by the Ban co Central, the sol’s apprecia ting run was in terru pted b y the
run up to the presidential elections of April 2006. The ascen t of Ollanta Hum ala (a self
proclaimed ethno-nationalist candidate) in the polls wa s receiv ed with dismay b y the mark e ts
26
and reflectedinanincreaseinPeru’sriskpremiaandafive percent depreciation of the sol
in the second half of 2005.
In this juncture, the Banco Central raised interest rates by 150 basis points bet ween
Nov ember 2005 and M ay 2006. In addition, the Ban co reversed course and started selling
dollars, most inten sely during Decem ber 2005 and January 2006 when it sold US $ 700
million. (C hart 12)
The increases in interest rates may have been justified on the basis of strong real activity
and expectations of increasing inflation at the end of 2005. But the May 2006 Inflatio n
Report makes it clear that exchange rate volatility was also a consideration :
The increase in the reference in terest rate also h elped to prevent situations
of excessive v o latility in the foreign exc h an g e market whic h , given the economy’s
fin an cial dollar izatio n, could generate undesired effects on inflation and economic
activity.
With respect to the sales of foreign exchange, the 2006 Inflatio n Report states:
The foreign exc hange operations undertak en by BCRP ha ve sough t to prev ent
that an excessive volatility of the exc h ang e rate, in a context of high financial
dollarizatio n, nega tively affect the dyn amism of economic activity...[Interv e ntion]
also permits the foreign exc hange mark et to con tin ue operating under circum-
stances in which the vo lum e of supply or dem an d in said market con tracts sig-
nificantly, preventing the norm al formation of prices.
Market volatilit y fell after February 2006, reflecting Alan Garcia’s rise and ev entual
victory in the elections. Since then, pressures for sol apprecia tion ha ve in ten sified. In
response, the Ba nco Central has resum ed the accumulation of internationa l reserves, with
gusto. The Ba nc o purchased US $ 4.2 billion in the second half of 2006, and US $ 4.5 billion
27
bet ween January and May this yea r. T his policy resulted in an increase in net international
reserves from US $ 14.4 billion in June 2006 to US $ 21.2 billion in last May. The stated
rationale for the huge accumulation of reserv es, according to the latest (May 2007) Infla tion
Report, is still "to afford eventual negativ e external shocks in the future ".
3.3. Chile
Background. Chile was amon g the early adopters of New Zealan d-style inflation targets,
ha ving started to announce inflation goals since 1990. How ever, Chile maintained a system
of crawling exchange rate bands for mo st of the nineties, and exchange rate objectives played
an important role in the form u lation of monetary policy in sev eral occasions, suc h as the
period of the Russian crisis and the Brazilian 1999 dev aluation.
Theexchangeratebandswerefinally dropped in Septem ber 1999, and from then on the
Ban co Central de Chile has claimed to implem e nt a full inflation targeting regime along with
flexible exchan ge rates. Inflation was to be k ept between tw o and four percent, a range that
has been main tained to this da y. The main monetary policy instrumen t is the interest rate
for overnight in terba nk loans, and is supposed to be adjusted until the forecast for inflation
in the following t welve to twenty four m onths falls within the target range. (C hart 13)
23
While the Banco Central emphasizes that inflation targeting can only be com patible with
floa ting exch an ge rates, it also explicitly states that it reserves the right to intervene in the
foreignexchangemarketinsomecases. TheJanuary 2003 Inflation Report, for example,
sa ys:
The floating exc h an ge rate is a distinctive componen t of Chile’s macroeco-
nomic policy framework...In exceptiona l circum stances, nonetheless, the Central
Bank may ch oose to participate in the foreign exchan ge market.
23
Data for Charts 13-15 are from the Banco Central de Chile ’s web site (www.bcentral.cl).
28
"Exceptional circumstances" are to be understood as periods of heigh tened volatilit y of
the exchange rate not justified b y fundam entals:
The exchan ge rate over-reacts when, with no major chan ge in its main deter-
minants (e.g. terms of trade, conditions for accessing international financing) a
substantial depreciation is followed by a similar appreciation, or vice versa, in a
relatively brief period of time .
According to the Banco Central, suc h episodes are hard to deal with through norm al
mon etary adjustments (that is, c hang es in in terest rates), because of the collateral effects of
the latter on real activity:
For example, a sharp depreciation could generate inflation that would be
necessary to compensate with monetary policy tightenin g. But if the exchange
rate movem ent was an overreaction, this monetary tighten ing w o uld unnecessarily
deepen the cycle. (De Gregor io, Tokman, and Valdes 2005, p.12)
Of course, iden tify ing when excha n ge rate v olatility is not w arra nted by fundamentals is
rather tric ky. The Banco Central has ac knowledged this fact, and as a consequence it has
prom ised to limit exchange rate interven tio n in magnitud e and horizon , as well as m ak ing
in tervention operations as transparent as possible. It also em pha sizes that interven tio n is
not in tended to support specific levels of the exchange rate but, rather, to provide adequate
liquidity so that m arket forces can bring exchange rates to equilibrium lev els.
The Chilean Experience. As is well kno w n, Chile’s economic performance since 2000 has
been envia ble. An nual inflation has remain ed close to three percen t and, except for 2002,
GD P gro wth rates ha ve been 3.4 percen t or better (Charts 14 and 15).
In this period, the Banco Cen tral has abstained from intervening in the foreign exchange
market, except for two episodes in whic h it determ in ed that exchange rates w ere "o verre-
acting ". Th e first episode w as in 2001, during which the Chilean peso came under intense
29
pressure, losing about one fifth of its value between the end of December 2000 to early Au-
gust 2001. The fall in the value of the peso was associated with adv erse movements in the
price of copper as well as contagion from the deteriorating market outlook on Argen tina.
Then , as described by the September 2001 Monetary Policy Report,
The speed and quic k succession of movem ents that led to a decline in condi-
tions abroad exacerba ted peso volatility and depreciation , despite the strength
of Chile’s econom y and the coherence of its macroeconomic policies. This led the
Central Bank’s Boa rd to adopt m easures to improve the stability of the country’s
financial markets.
The measures, which were ado pted in August 2001, included the annou ncement that the
Banco Central wo uld sell up to US $ 2 billion dollars (out of U S $ 15 billion in inter na tional
reserv es) in the spot market in the next four months.
24
The hope appears to hav e been that the in terv en tion would be enough to prick a bubble
on the dollar, as expressed by De Greg orio and Tokman (2004):
The intervention appeared as the first line of defense against an inflation
caused by an excessiv e depreciation. The possibility of a bubble dominating the
market required actions to v erify if this was an overreaction. The success of the
sterilized intervention sho ws that the mark et reaction w as indeed exaggerated.
Otherw ise, the interv ention would hav e been ineffectiv e, and wou ld have required
a monetary contraction if inflation expectations had been very far from the target.
Happ ily, the B anco Central’s move w a s follo wed b y a good response b y the ma rkets, to
the point that th e fin al am o unt of dollars actually pur chased (US $ 803 million) was less
than half the maximum originally announced.
24
In addition, the Banco Central would sell up to US $ 2 bn in peso bonds indexed to the dollar.
30
The second episode was triggered by the Brazilian 2002 election s. B etw ee n August 2002
and October 2002, the Chilean peso embarked on a path of clear depreciation, losing almost
ten percent of its value. In response, the Banco Cen tral announ ced its willingness to intervene
to support the peso between October 200 2 and February 2003, as it had don e the year before.
This time, however, the pressures on the peso subsided quickly, and the Banco Cen tral ended
up not interv ening in the spot mark et.
25
In commenting on this episode, De Gregorio and Tokman (2004) stress the fact that the
operations w ere preannounced and transparent, and that they w ere designed to deal with
nonfundamen tal exchange rate volatility:
The specific form taken by the in terven tion package is of special in terest,
since it is not shar ed b y other countries. In C hile, when in tervention has been
announced, both start and end dates have been published. Also, the maxim um
amoun t of in terv ention in the spot market and the calendar of monthly issues of
Cen tral Bank Deposit Certificates have been disseminated...In fact, the in terven-
tion takes place to inform the market that the monetary authorit y considers that
theevolutionoftheexchangerateisnotjustified given its fundam entals...Thus,
the interv en tio n is a test to find a possible bubble [em p hasis added].
3.4. Brazil
Background. After the A sian and Russian crises of 1997-98, enormo us pressures on the
Braz ilian Rea l, which had been pegged to the dollar ever sin ce the 1994 Real Plan, led the
Banco Central do Brasil to adopt floating rates in Jan uary 1999. As the Banco was then free
to adopt an alternative nominal ancho r, a June 1999 presidential decree (3.088) directed the
Ban co to follow an inflation targeting regime.
25
It did sell, however, a small amount of Central Bank dollar denominated certificates.
31
Thedecreemandatedthatinflation targets, c hosen by a National Monetary Cou ncil,
would provide the "anchor for fixing the reg ime of monetary policy". The target horizon
was fixed at three y ear s, so inflation targets for 1999, 2000, and 2001 were announced (respec-
tiv e ly, at eight, six, an d four percent per y ea r). The target for 2002 were to be annou n ced
in June 2000, and so on. Tolerance bands of plus and m inus two percen tage points were
accepted.
From the start, the policy instrume nt was the in terest rate for o vernight in terb ank loans,
commonly known as the SELIC rate. T he setting for the SELIC rate is decided b y a M oneta ry
Policy Com m ittee (COP O M), whic h met once a mon th until 2006, and has m et eigh t times
ayearsince.
In contrast to the other cases discussed in this paper, the Banco C entral do B rasil has
refrained from an y explicit claims to an y "righ t to in terv ene" in the foreign exchange market.
Instead, official documents, speech es, and p resentations emp hasize that there is a floatin g
exc h ange rate regim e, mentioning only occasionallythatmonetarypolicymayhavetore-
spond to external developments insofar as they affected the inflation objective. For example,
a July 2, 1999 Letter of Intent to the IMF stated:
in its conduc t of interest rate policy, the BCB will respond to any significa nt
pressures on net international reserves or to excha nge r ate developments that
may threaten the achievemen t of the inflation target.
Similar statements, however, are hard to come by, and the record suggests that the Banco
do Brasil explicit em br acin g of floating exchange rates is part of its effort to build credibility
for the inflation targeting regime.
Overall Performanc e. AsgiveninChart16,
26
progress tow ards low and stable inflation
since the establishment of inflation targeting has been notable but unsteady. Annual inflation
26
Data for Charts 16-19 comes from the Banco Central do Brasil’s web site (www.bcb.gov.br). In addition,
the spreads data in Chart 19 was kindly provided by JP Morgan.
32
fell to alm ost nine percen t in 1999 (in spite of a h ug e Real depreciation) and six percen t
in 2000, and hence v ery close to the official targets. How e ver, inflation increased to 7.7
percent in 2001 and to 12.5 percen t in 2002; as we discuss next, this was mostly in response
to contagion from Argen tina, the imp act of the September 11, 2001 attac ks and the 2002
presidential elections. Since 2003, ho wever, the outlook has con tinuously improv ed, and
inflation is currently stable at around three percent per ye ar.
One w ell kno w n aspect of the Brazilian ex perience with inflation targeting is that interest
rates hav e been rather high (Chart 17). As discussed b y Fraga, Go ldfajn, and M inella (2003)
and others, the Banco do Brasil sees high interest rates as necessary giv en the lo w credibility
inherited from the past. In addition, very high interest rates were necessar y to figh t the
strong adverse shoc ks of 2001-2.
Perhaps as a consequence, the performance of the real econom y has been rather uneven
relative to other Latin Am erican coun tries. GDP gro wth was close to one percen t in 1999,
2001, and 2003. The other y ears in the period covered sa w reasonable but not spectacular
growth, with the exception of 2004 when growth w as 5.7 percent. (Chart 18)
Recently, ho wev er, Brazil has become the darling of the internation al in vestmen t comm u -
nit y. As the Lula administration has proven to be a w elcom e surprise in terms of macroeco-
nom ic management, fiscal and debt consolidation, and political stability, Brazil’s risk spread
has plummeted and intern atio nal capital inflows are flooding the financial system (Chart 19).
As discussed below , these flows have also presen ted the Ba nco do Brasil with a substantial
c hallenge in terms of exchange rate managemen t.
2001: Re al depreciation and a missed target
As already mentioned, inflation in 1999 and 2000 was in line with the targets set in 1999.
How ev er, inflation accelerated in the second half of 2001, to the point that by the end of
the year the inflation rate reached 7.7 percent, well outside the tolerance band (two to six
percen t).
33
The Banco do Brasil’s analysis of this event is best summariz ed in the 2002 Open Letter
to the Minister of Fina nc e which, as required by the inflation targeting rules, w as written to
explain the causes for missing the targ et as w ell as to describe policy m easures to address the
failure. As the Letter states, one proximate cause of inflation acceleration w as a com b ination
of external shoc ks:
On the external fron t, the deceleration of the world econom y, the negative fac-
tors stem m ing from the Argentine crisis, and the terrorist attac ks to the United
States generated strong pressure for the depreciation of the Real in 2001. In Oc-
tober, the Real peak ed at R$ 2.84 to the U S dollar, accumulating a deprecia tion
of 42.6 %...The depreciation of the Real mark e dly pressured domestic price s.
TheLetterwentontosaythatthedirecteffect of the Real depreciatio n on inflation was
2.9 percen tage poin ts, so that 2001 inflation w ould hav e been only 4.8% , hence within the
inflation tolerance bands, if the Real had not depreciated.
In this situation, the Banco’s main policy response was to adjust the SELIC rate, from
alowof15.25percentinJanuary2001to19percentattheendof2001. Inviewofthe
Banco ’s analysis, these in terest rate increases w ere consistent with the con ven tiona l wisdom
on inflation targeting , as they w ere a response to the inflationary effects of the external
shocks affecting the econom y.
Yet, the Ban co’s response also includ ed , less conv entionally, an element of intervention
in the foreign exc han ge m arket. The objectiv e of in t erven tion appears to ha ve been not
the preservation of the inflatio n goal, but stabilizing market volatility. This is cryptically
ac kn owledged in the Open Letter:
The exchange rate market sho wed instabilit y at som e mo m ents, especially
after the September attac ks. As a response, the Central Bank c h ose to com-
plement the in terest rate policy b y increasing the reserve requirement ratio on
34
time deposits and b y temporarily interven ing in the exchang e rate mark et, with
satisfactory results.
But foreign exchange in terven tion is not mentioned an yw h ere else in the Letter, which
closes by claiming that
The macroeconomic regime of inflation targets, floating exc hange rates and
fisca l responsibilit y pro ved to be effectiv e to deal with a perve rse sequen ce of
strong shoc ks...
Indeed, the Banco do Brasil rarely men tion ed the fact that there w as an significant
elementofinterventionintheforeignexchangemarketduring2001. Andwhenitdid, "in-
tervention" was meant to refer to the issuance of US dollar indexed debt in Real, instead of
direct sales of foreign exchan ge in th e market.
27
For example, the No vem ber 2001 Memo-
randum of Econom ic P olicies sent to the IMF states:
The BCB remains committed to the floating exchange rate regime, and will
con tinue to avoid targeting an y particular lev el of the exchange rate. Nev er-
theless, because of concern about the inflationary implications of exchange rate
deprec ia tion and v o latility, the BC B has sought to offset recen t pressures on the
exchange rate arising from the extraord inary circum stance s since early Septem -
ber by in tervening in the foreign exc ha nge mark et through the sale of foreign
exc hange-indexed debt.
In contrast, the March 2002 Mem or andu m reluctantly acknowledges that direct in ter ven-
tion had, in fact, tak en place:
27
See Bevilaqua and Azevedo (2005).
35
Theexperienceoflastyearconfirms the importance of the floating exc han ge
rate regime in helping the economy respond to shocks. Nev ertheless, the size
and scope of the shocks experienced last year and the high v ola tility and lo w
liquidity of the FX mark et required the Cen tral Bank to in tervene through the
issuance of foreign exc hang e-in dexed securities, as well as thro u gh pre-announ ced
spot market sales of limited quantities of foreign exch ang e. With excha nge rate
pressures largely ha vin g abated, the Centra l Bank has ceased direct and indirect
in terven tion in foreign exchange ma rkets,
Direct foreign exchange in te rven tion is not mentioned in the Banco de Brasil’s Inflation
Reports. But the Banco’s 2001 Ann ual Report includes (in page 142) what appears to be
the only explicit description of foreign exchange sales:
Banco Central interventions in the domestic exc hang e market resulted in net
sales of US$ 7.2 billion [in 2001]. Sales were sporadic in the first six months of
the year and totaled US$ 885 million. A s of the month of July, interventions
becam e m ore frequ ent, par ticularly in light of the excessive vo latility of exch an ge
rates in the period, as sales rose to a total of US$ 6.3 billion in the second ha lf
of the y ear.
This indicates that the amount of direct foreign exchan ge interv ention was substan tial:
since the Banco Central’s net in ternation al reserves at the end of 2000 were US $ 33 billion,
the Banco’s dollar sales in the second half of 2001 w ere bet ween on e fifthtoonesixthofits
reserv es. And, again, the reason stated was the volatility of the exchange rate, termed as
"excessiv e".
2002: Ele ctoral Uncertainty
The 2001 episode w as to be dw arfed b y the mark et v olatility surrounding the 2002 presi-
dential elections. M arket anxiety began to grow at the start of the summ er, fueled by the rise
36
of Lula and Ciro Gomes in the polls. Investors sta y ed away from Brazil, which was reflected
in a steep increase in coun try risk premia. Reflecting this situation, the Brazil EM B I Spread
jumped from 701 basis poin ts on March 20th to 2422 basis poin ts on July 30th. The impact
on exc hange rates w as correspondingly large. A t the end of Marc h, the Real w as trading at
aboutR$2.3perUSdollar;bymid-October,theexchangeratehaddepreciatedtoalmost
R$ 4 per US dollar. (See Chart 19)
The consequences for inflation and monetary policy were p red ictab le. Yearly inflation
in 2002 closed at 12.5 percent, wa y above the target of 3.5 percen t and outside the 1.5-
5.5 tolerance band. In addition, inflation increased every month after August, promptin g
concerns that the Banco Central had lost control of inflation.
Againasprescribedbytheinflation targeting rules, the Banco Central’s Go vernor had
to send an explanation letter to the Minister of Finance. The Open Letter, dated Jan ua ry
21, 2003, m en tions that there w as "a confidence crisis in the performance of the Brazilian
economy and an increase in risk aversion in in ter nation al mar kets. " The confidence crisis
was expressed in a drastic reduction in the deman d for public securities, as well as in a fall
of US$ 27.8 billion in foreign capital inflows. As a result,
The dom estic confidence crisis coupled with increasing global risk aversion had
significant impa cts on the economy...in vestment dropped to 18.8% of GDP in the
third quarter of 2002 (vs. 19.45% at the end of 2001)...consumption dropped
to 77.99% of GD P in the third quarter of 2002, from 79.82% of 2001 GD P...the
current account declined to US$ 7.8 billion in 2002 (1.7% of GDP) from US$ 23.2
billion in 2001 (4.6% of GD P)
As a result, the Letter argu es,
The difficulties faced by the coun try were reflectedmainlyintheexchangerate
37
and inflatio n expectations...It is estim ated that the exch an ge rate deprecia tion
con tributed 5.8 percen tage poin ts to inflation in 2002.
The policy response was based on a drastic adjustment of the SEL IC rate, as well as a
c hange in the inflatio n targets. The target for the SELIC rate, whic h had been lowered to
18 percent by July 2002, was increased in successive steps to close 2002 at 25 percen t. In
addition, Open Letter pro posed to raise the inflation targets for 2003 and 2004 to 8.5 percen t
and 5.5 percen t, from the original ones of 4.0 and 3.75 respectiv ely.
The decision to raise the inflation targets in mid course was, perhaps, the most notice-
able and debatable aspect of the Ban co Cen tral’s policy response (see, for instance, Fraga,
Gold fajn, and Minella 2003 and the discussion in the 2003 NBER Macroe c onomics Annual ).
For our purposes, how ever, the most relevan t feature of the response w as the interven tio n in
foreign exch an ge markets, starting in July.
Foreign exchange interven tion had three dimensio ns. First, the Ba nco Central agreed
to provide up to US $ 2 billion dollars in export cred it lines, of whic h US$ 1.4 billion we re
drawn in 2002. Second, the B anc o engaged in US$ 1.8 billion in foreign exch ang e repos.
Finally, there w ere spot sales of US $ 5.9 billion. A ll in all, the Banco Central sold US $ 9.1
billion in 2002, more than one quart er of its reserves at the end of 2001, as a resu lt of its
intervention policy.
28
Again , in spite of its substantial size, these foreign exc h ange interventions have received
little attention and are mentioned only sporadically by the Ban co Central. For example, the
August 2002 Technical Memorandum of Understanding to the IMF says
The experience of the current and past y ears confirms the importance of the
floa ting exc han ge rate regime in helping the economy respond to shoc k s. Never-
theless, the size and scope of the shoc ks experienced recen tly required the central
28
This information is taken from the Banco’s 2002 Annual Report.
38
banktoalleviatesomeofthepressureontheexchangeratebyinterveningin
the foreign exchan ge market. In the case of intervention reaching an accumu-
lated amount of U$3.0 billion, on a rolling 30-day basis, a genera l discussion of
mon eta ry and interv ention policies will be initiated with the staff.
Inflationary pressures began to subside only in the second half of 2003, after the Banco
Central had increased the SELIC rate even more, to 26.5 percent in February 2003. After
surpassing 16 percen t in June, year to y ea r inflation had dropped to 9.5 percent b y December,
and the exc hange rate appreciated to about R$ 3 per US dollar. By the beginning of 2004,
the Brazil EM B I spread had dropped to about 400 basis points. The return of confidence has
been widely attributed to political stability and disciplined monetary and fiscal man ag em ent.
2004 to the Present: Reserves Accumulat ion and Fighting App reciation
2004 turned to be a banner year for Brazil. GDP growth jumped to 5.7 percent, up from
only 0.5 percent in 2003. Annual inflation fell to an average of 6.6 percen t. Exports grew by
more than thirty percen t, from US$ 73.1 billion in 2003 to US$ 96.5 bn., helping the current
account surplus reach 1.9% of GDP.
Clearly, much of the credit goes to the success of the Lula administration is consolidat-
ing fiscal and monetary m anagement. On the fiscal side, credibilit y was enhanced by the
announcement of reforms of the social security and tax systems, and b y more stringent tar-
gets for primary surpluses. On the monetary side, the Lula administration kep t reaffirming
support for the independence of the Banco Central, in spite of widespread debate about the
desirability of the very high interest rates in place since 2003. In fact, the Ban co Central,
which had started gradually reducing the SELIC rate to a low of sixteen percent in April
2004, started a new round of increases in September 2004 in ord er to crush expectations
of reviving inflation. The SELIC rate would be increased several times up to 19.75 in Ma y
2005.
39
International capital started to flow bac k in to Brazil, and the Real kep t apprecia ting : it
would close 2004 at 2.7 per U S dollar, and would contin ue to appreciate until the current
value of about 1.95 per dollar.
It was in this context that the Banco Central a ch ange in its foreign interven tion policy,
which is summ a rized in the 2004 Annual Report as follo ws:
In 2004, the country maintained its strategy of reducing the share of the
internal public debt tied to the dollar and of purchasing exc hange in the mark et
for purposes of external debt paym en t and rebuilding of its in ternational reserv e
position. Here, the overall aim was to atten u ate balance of paymen ts and public
sector balance sheet vulnerability.
On 1.6.2004, a program of rebuildin g the country’s in ter nation al reserv e po-
sition w as announced. This program was designed to mak e it possible to take
advantage of liquidity conditions whenev er they turned fa vorable, in order to be
able to ensure a neutral imp act on exchange m arket volatility and floating ex-
c han ge rates. The BCB began purchasing exchange in the market in order to
gradually restructure its intern ationa l reserves and the National Treasury inter-
rupted its policy of acquiring exchang e in the m arket to be used for external debt
servicing in volving bonds and the P aris Club. Net BCB purchases in the market
totaled US$ 5.3 billion in 2004.
Since then, the Brazilian economy has continu ed to improv e . After falling to 2.9% in
2005, annu al GDP growth increased to 3.7% in 200 6 and is curren tly projected to be 4.7%
in 2007. Inflation has been falling, and is widely expected to be between 3.5 percent and
4 percent for 2007 and 2008. Export gro w th has continued to be robust, and the current
account surplus is expected to remain around one percent of GDP in 2007.
The Banco Central response to falling inflation rates has been to gradually reduce the
40
SELIC rate, an ongoing process that started in September 2005 and has brough t the SELIC
to 12 percent as of this writing. In spite of the interest rate reductions, however, the Real
appears stronger than ev er. T he Real appreciation is expected to influence monetary policy
in t wo w ays. First, appreciation is associated with deflationary pressures, allow in g for event
further SEL IC reductio ns. Second, the Banco Central stated policy of reserv es accumulation
has in tensified. The current level of reserv es is US $ 145 billion, more than double the US$
62.7 billion lev el of the end of June 2006.
The furious pace of reserves accumulation has prompted a lively debate on whether the
Banco Cen tral has abandoned its stated policy of floating exc hange rates, and wh y. Clearly,
one of the reasons for the curren t policy of accum ulating reserves is the belief that a large
"war ch est" can be useful as insuran ce against in ternation al shock s. T his w as acknowledged
recen tly b y the government:
Finance minister Guido Man tegna commemorated the fact [that reserves had
surpassed US $ 100 bn. in February 2007] stating that it "makes the coun try
more resistan t to the foreign turbu len ce tha t occasionally materia lizes " "(...)
"W ith the reserves, we will be vaccinated against inter na tional turbulence, " he
guaran teed.
29
On the other hand, the policy appears to respond also to concerns about competitiv en ess
and, hence, about the impact of the excha nge rate lev el on exporters. This is clearly expressed
b y President Lula in a recent inter view, sum m arized b y JP Morga n Chase:
Lula reaffirm ed that the floating fx regime will contin ue, along with a strong
pace of reserv e accum u lation. He mentioned that the government is concerned
about the effects of BRL appreciation on som e sectors, and that this is why the
29
Market analysts up their bets for Brazilian economy, by Stenio Ribeiro, Brazzil Magazine, Monday Ma y
28, 2007.
41
central bank will con tinue to purchase USD in the spot m arket (Lula me ntioned
US$ 200 billion of reserves as a target, vs the curren t level of US$ 145 billion).
The policy of giving sectoral fiscal incentiv es and tariff protection to sectors
most affected by the overvalued currency will con tinue. He denied any plan to
implem ent a more aggregate measure to con tain BRL appreciation, suc h as taxing
short-term capital inflows, wh ich the market has begun to talk about.
30
3.5. Summ ary and D iscussion
The four cases reviewed in this section share sev eral aspects, notable among them :
• In spite of havin g adopted inflation targeting, Latin American cen tral banks have
explicitly claimed a right to in tervene in the foreign exchange m arket under some
circum stan ces, most often related to instability in the foreign exchange market and the
fin an cial system. In additio n, Latin A m er ican central bank s consider it beneficial to
build a substantial stoc kpile of foreign exchange reserves.
• In the four cases review ed here, the most significa nt episodes invo lv in g capital outflows
and rapid depreciating trends w e re related to political shock s (the 20 01 crisis in Ar-
gentina, Lula’s 2002 election in Brazil, Humala ’s rise in P eru at the end of 2005). Th e
resulting exchange rate depreciation was a concern not only because of the impact on
prospectiv e inflation but also, and sometim es more significantly, because of the effect
on financial variables. T he policy response wa s t yp ic ally to figh t the depreciation via
increases in interest rates aided by sales of foreign exchange. The relative size of the
in terven tion operations differed from case to case.
• More recently, the countries under review ha ve witnessed strong currency appreciation
coupled with massive accumulation of inter na tion al reserves. There has been little
30
Global Data Watch, June 22 2007, p. 49.
42
attempt at consistency bet ween this accumulation policy and the o verall effort to man-
age interest rates to hit the inflation targets. References to overvaluation and loss of
competitiv eness have become more frequen t.
4. Rationalizing Actual Inflation Targeting Practices
Why have Latin American cen tral bank ers departed from the con ven tional inflation targeting
script in terms of exchange rate manag em ent and reserv es accumulation? The answer may
be that the applicabilit y of the conventional script to Latin American econom ies is believ ed
to be only partial. It turns out that good reasons for such a belief can be found in recent
academic work. In this section we review some of those reasons, and we also ask wheth er
they in fact w arra nt observ ed policies.
4.1. Balance Sheet Effects and D ollarization
The New Keynesian open econom y models underlying the conventional IT wisdom are built
upon the assumption that an exchange rate depreciation is expansionary. This is because,
in those models, the main c hannel through whic h a depreciation affects aggregate demand is
b y c hanging the relative price of domestic vis a vis foreign goods and b y reducing domestic
in terest rates.
The view that depreciations are expansionary, ho wev er, has been long at odds with
empirica l evidence, and has come under substan tia l challen ge follo win g the crises in Mexico
1995 and Asia 1997-9 8, the Russian 1998 de fault, and the Brazilian 1999 devaluation. Those
episodes w ere associated with widespread financial con tagion and exchange rate depreciation
in m any emer ging economies. And in several cases it was observed that a weak er currency
impaired the balance sheets of debtor firms, reducing the access of those firms to credit and,
as a consequen ce, restricting in vestment dem a nd. The balance sheet im pa ct of a depreciating
43
exc h ange rate w as greater for those agen ts that had taken debts in foreign currency (typically
the US dollar) wh en their rev enues were given in domestic currency.
These observations hav e motivated a substantial body of research focusing on the links
bet ween currency