Fraga InflationTargetingEmerging 2003

Inflation Targeting in Emerging Market Economies
Author(s): Arminio Fraga, Ilan Goldfajn and André Minella

Source: NBER Macroeconomics Annual , 2003, Vol. 18 (2003), pp. 365-400
Published by: The University of Chicago Press on behalf of the The National Bureau of
Economic Research
Stable URL: https://www.jstor.org/stable/3585264
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
The National Bureau of Economic Research and The University of Chicago Press are
collaborating with JSTOR to digitize, preserve and extend access to NBER Macroeconomics
Annual
This content downloaded from

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
All use subject to https://about.jstor.org/terms
Arminio Fraga, Ilan Goldfajn, and Andre Minella
PARTNER, GAVEA INVESTIMENTOS; PONTIFICAL
CATHOLIC UNIVERSITY OF RIO DE JANEIRO; AND CENTRAL
BANK OF BRAZIL
Inflation Targeting in Emerging

Market Economies
1. Introduction
The performance of inflation-targeting regimes around the world has

been positive. Average inflation in both emerging markets and devel-
oped economies is substantially lower after the adoption of the inflation-
targeting regime than immediately before its adoption (Figure 1).1
However, emerging market economies (EMEs) have had a relatively
worse performance. In these countries, deviations from both central tar-
gets and upper bounds are larger and more common.2 This outcome sug-
gests that either inflation targeters in EMEs are less committed to their
targets or inflation targeting in these countries is a more challenging task
than in developed ones. The latter explanation is related to the more
volatile macroeconomic environment and to weaker institutions and cred-
ibility in these countries, which in turn lead to more acute trade-offs th
the existing ones in developed economies.
We thank the participants in the NBER Eighteenth Annual Conference on Macroeconomi

especially Mark Gertler, Robert Hall, Frederic Mishkin, and Kenneth Rogoff, and semin
participants at Princeton University for their comments. We are also grateful to Fabia A.
Carvalho, Eduardo Loyo, and Marcio I. Nakane for their suggestions; to Thais P. Ferreira a
Myrian B. S. Petrassi for research assistance; and to Marcileide A. da Silva, Pedro H. S. d
Sousa, and Raquel K. de S. Tsukada for assistance with data. The views expressed are thos
of the authors and not necessarily those of the Banco Central do Brasil.
1. Bernanke, Laubach, Mishkin, and Posen (1999); Mishkin and Schmidt-Hebbel (2002); an
Corbo, Landerretche, and Schmidt-Hebbel (2002) have found evidence of additional gai
stemming from inflation targeting. Ball and Sheridan (2003) have found no evidence th
inflation-targeting countries have enjoyed better performance in the OECD.
2. The deviations of inflation from central targets and from the upper bounds of targe
were 81% and 167% higher than in developed economies, respectively (table available o
request to the authors).

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
366 - FRAGA, GOLDFAJN, & MINELLA
Under inflation targeting, EMEs have the challenge of breaking the

vicious circle between, on one side, low credibility and more fragile insti-
tutions and, on the other side, higher macroeconomic instability and vul-
nerability to external shocks. It is a long process that involves acquiring
credibility as a monetary policy institution committed to price stability in
the context of higher instability.
This paper assesses inflation targeting in EMEs compared to that in
developed economies and develops applied prescriptions for the conduct
of monetary policy and inflation-targeting design. We verify that EMEs
have faced more acute trade-offs than developed countries have: both
output and inflation are more volatile, and the inflation level is higher.
The explanation for the different performance of EMEs relies on the pres-
ence of more fragile institutions and imperfect credibility, and on the
nature and magnitude of the shocks that hit these economies.
There are several instances where these more acute trade-offs emerge.
Take the case of a sudden stop in the inflow of capital to an EME leading
to a substantial depreciation of the currency (e.g., in 2002, Brazil was faced
with a negative swing of US$30 billion--or 6% of gross domestic product
(GDP)-in capital flows relative to an already difficult 2001, which led to
a nominal depreciation of 50%). Even in a context of good initial condi-
tions-low pass-through and 12-month forward inflation expectations on
track-this event led to a breach of the inflation target and, given some
inertia, to a worsening of both inflation expectations and actual future
inflation (Brazil's target of 4% was breached with an inflation of 12.5%,
and 12-month forward inflation expectations were 11.0% at the end of
Figure 1 INFLATION BEFORE AND AFTER ADOPTION OF INFLATION

TARGETING (IT)
13.11 Emerging Market

Economies
14
12
10
, Developed Economies 5.95

. 4-2.50
Before IT After IT (U

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Inflation Targeting in Emerging Market Economies - 367
December 2002). In this context, what is the optimal response to these

types of shocks? Should one take 24 months or longer to converge to the
target, even when building credibility and reputation is still important?
In general, in the more volatile environment in EMEs, some applied
and theoretical issues deserve attention. (1) How can the country build
credibility when faced with larger shocks? And how can it balance
flexibility and credibility in such an instance? (2) How does an infla-
tion-targeting regime work in a disinflation process? And in a credibility-
building process? (3) How can a country deal with shocks that represent
important changes in relative prices? (4) Should bands be wider and the
central points of inflation targets be higher in EMEs? (5) How should
International Monetary Fund (IMF) conditionality be designed with an
inflation-targeting country?
This paper discusses these monetary policy issues. Its focus, therefore,
is more applied. For that purpose, our analysis is based on empirical find-
ings for EMEs and, in particular, on our own experience at the Banco
Central do Brasil (BCB), besides the use of simulations of a model to guide
our discussions. In a way, because the case of Brazil represents the first
stress test of an inflation-targeting regime, the lessons learned may some-
day be useful for other countries.
We stress the role of communication and transparency as crucial for the
process of building credibility. This paper lays out the main issues and, as a
by-product, also includes two applied proposals. The first is a transparent
procedure that a central bank under inflation targeting can apply and com-
municate when facing strong supply shocks. The second is a design of a mon-
itoring structure for an inflation-targeting regime under an IMF program.
The paper is organized as follows. Section 2 presents some stylized facts
about EMEs and developed countries. It contains a statistical comparison of
the conduct and results of monetary policy in EMEs and developed coun-
tries. Section 3 presents a theoretical model of a small open economy, which
is employed to simulate the effects of some shocks and changes in inflation
targets. Section 4 discusses some reasons why EMEs face higher volatility. It
analyzes the challenge of constructing credibility and reducing inflation lev-
els, analyzes the effect of large external shocks, and addresses issues on fis-
cal and financial dominance. Section 5 addresses how to deal with shocks.
2. Stylized Facts About Inflation Targeting in Emerging

Market Economies
In this section we present basic stylized facts comparing the volatilities of

inflation, output, exchange rates, and interest rates, and the average of
inflation and output growth in emerging markets and developed

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Table 1 INITIAL TARGETS AND INFLATION AROUND ADOPTION OF

INFLATION TARGETING (12-MONTH ACCUMULATED INFLATION)
Date of Inflation
adoption Inflation 12 months
inflation right before after
targeting First target IT adoption IT adoption
Developed economies
Australia Apr 1993 2% - 3% 1.22 1.74
Canada Feb 1991 3%- 5% 6.83 1.68
Iceland Mar 2001 2.5% (-1.5% + 3.5%) 4.05 8.72
New Zealand Mar 1990 3% - 5% 7.03 4.52
Norway Mar 2001 2.5 3.64 1.10
Sweden Jan 1993 2% (? 1%) 1.76 1.70
Switzerland Jan 2000 ? 2% 1.63 0.90
United Kingdom Oct 1992 1% - 4% 3.57 1.35
Average 2.8 3.72 2.71
Median 2.5 3.61 1.69
Emerging market e
Brazil1 Jun 1999 8% (? 2%) 3.15 6.51
Chile Jan 1991 15% - 20% 27.31 19.47
Colombia Sep 1999 15% 9.22 9.35
Czech Republic Jan 1998 5.5% - 6.5% 9.98 3.5
Hungary Jun 2001 7% (+1%) 10.78 4.87
Israel Jan 1992 14% - 15% 18.03 10.74
Mexico Jan 1991 ?13% 18.61 11.03
Peru Jan 1994 15% - 20% 39.49 13.71
Poland Oct 1998 ?9.5% 10.44 8.82
South Africa2 Feb 2000 3% - 6% 2.65 7.77
South Korea Jan 1998 9%(+1%) 6.57 1.46
Thailand Apr 2000 0% - 3.5% 1.04 2.47
Average 10.3 13.11 8.31
Median 9.3 10.21 8.30
1. In Brazil, the inflation of

of the overappreciation of t
2. First target established fo
economies.3 We hav
between the two c
EMEs; therefore, no
economies. Because
however, we call th
of inflation targetin
to apply econometr
3. The assessment of th
design of inflation targe
Truman (2003); Mishkin
Corbo and Schmidt-Heb

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Table 2 VOLATILITY AND AVERAGE OF SELECTED VARIABLES FOR

1997:1-2002:2
Volatility of basic variables Average

Exchange GDP Interest GDP
Countries Inflation rate* growtht rate growth Inflation
Developed economies
Australia 2.05 0.13 1.96 0.58 4.78 5.89
Canada 0.83 0.04 1.30 1.14 3.57 1.96
Iceland 2.45 0.15 3.13 3.02 4.17 4.05
New Zealand 1.21 0.16 3.61 1.47 3.09 1.65
Norway 0.77 0.10 2.25 1.46 2.66 2.44
Sweden 1.11 0.12 2.41 0.44 2.58 1.24
Switzerland 0.54 0.08 1.14 0.92 1.79 0.85
United Kingdom 0.92 0.06 0.79 1.13 2.61 2.46
Average 1.24 0.11 2.07 1.27 3.16 2.57
Median 1.02 0.11 2.11 1.13 2.88 2.20
Emerging market economies

Brazil 2.09 0.15t 2.06 7.06 1.81 5.89
Chile 1.30 0.17 3.25 - 3.11 3.88
Colombia 5.43 0.25 3.38 10.02 0.81 12.51
Czech Republic 3.46 0.09 2.73 5.81 1.81 5.31
Hungary 4.09 0.16 - 1.13 - 11.21
Israel 3.18 0.10 3.36 3.34 2.98 4.35
Mexico 5.98 0.07 3.17 7.26 4.05 11.72
Peru 3.04 0.11 3.45 5.50 2.11 3.89
Poland 4.13 0.11 2.40 4.14 3.85 8.40
South Africa 2.13 0.26 1.11 3.65 2.26 6.51
South Korea 2.36 0.14 6.38 5.52 4.31 3.73
Thailand 3.25 0.14 6.13 6.72 0.08 2.88
Average 3.37 0.15 3.40 5.47 2.41 6.69
Median 3.22 0.14 3.25 5.52 2.26 5.60
Data source: International Financial Statistics, IMF

* The coefficient of variation (standard deviation
Growth rate measured comparing the current quar
tThe period 1999:1-2002:2. For 1997:1-2002:2, the
emerging and developed economies

among developing and develope
respectively. Most of the developed
between 1990 and 1993, whereas th
adopted it from 1998 onward.
We consider two samples. The f
adoption of inflation targeting in
pare countries with the same regi
tries are different, however, the

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Figure 2 VOLATILITIES OF SELECTED VARIABLES OF THE PERIOD

1997-2002 (AVERAGE OF STANDARD DEVIATION)
Interest Rate
Inflation GDP Growth 5.47
" 4 3.37 3.40

CL Exchange Rate
31 2.07 (Standard DeviaionAverage)
2- 1.24 1.7
CL 0.11 0.15
0.
DE EME DE EME DE EME DE EME
DE = developed econominles.
EME = emerging market economies.
different as well. Then we consider a second

recent period, which includes Asian, Brazilian, a
to mid-2002, which we refer to as a fixed sam
countries are inflation targeters in the whole
are just in part of it. Table 2 and Figure 2 record
sample.4
In both samples, the data indicate that, in comparison to developed
economies, the volatilities of all variables (inflation, the exchange rate,5
output, and the interest rate) and the inflation level are higher in EMEs.
The more challenging trade-off faced by EMEs is illustrated in Figure 3,
which shows the combination of the variability of output growth and
inflation for each country for 1997:1 to 2002:2.
3. Model
We develop a small open economy model to illustrate the main points raised
in the paper. The objective is to simulate the effects of some shocks and
changes in inflation targets. The model combines features of Batini, Harrison,
and Millard's (2003) and McCallum and Nelson's (2000) formulations.
Imports enter as intermediate goods, in contrast to most of the
open economy literature, which typically uses a model with imports as
consumption goods. As stressed by McCallum and Nelson (2000), a
4. The data referring to the first sample is available on request to the authors.
5. For the inflation-targeting sample, the measured exchange-rate volatility is similar.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Figure 3 TRADE-OFF VOLATILITIES OF OUTPUT AND INFLATION (1997:1-2002:2)
7.0A 1 --.
SKorea Thailand
O 5.0 -/
0 4.5- /
o 4.0 Peru Colombia
0 3.5 Chile Icela e Average EME o
ICO 3.0 Israel Czech Republic
2.5 Norwa
a ,
eden Brazil aPoland
c 1.5
S2.0 AveragP
+CanadaA
1.0
U) Switzer ned Kth Africa
0.5 nited
0.0 rI
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5
Standard Deviation of Inflation
DE = developed economies.
EME = emerging market economies.
specification where imports are entered as intermediate goods captures

better the features of the data. In comparison to models where imports
enter as consumption goods, such as in Gall and Monacelli (2002),
McCallum and Nelson's (2000) model generates a lower and more
delayed correlation between exchange-rate changes and the inflation
rate, closer to that observed in the data. Furthermore, intermediate
goods are the major items in imports. Table 3 records the share in
imports of goods classified by use for five countries. On average, con-
sumption goods represent only 21.3% of the total imports, whereas cap-
ital and intermediate goods shares are 29.5% and 46.2%, respectively.
Because no imported consumption goods are included in the model,
there is no distinction between domestic and consumer price index
(CPI) inflation, which is different from Svensson (2000) and Gali and
Monacelli (2002).
The model is derived from the optimization of infinitely lived house-
holds and firms. We present directly the log approximation of the vari-
ables around the nonstochastic flexible-price steady state. Lowercase
variables represent log-deviations from their steady-state values.6 We
present here only the most important equations of the model.7 The economy is
6. x, = log(X,) - log(XSS), where Xss is the steady-state value for X,, and log is the natural log-
arithm. Because log(1 + r,) = r,, the lowercase variables represent percentage deviations
from the steady state.
7. The derivation of the model from the optimization of households and firms, and all the
equations resulting from the log-linearization are available on request to the authors.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
comprised of households, firms (owned by the households), and govern-

ment. Firms produce differentiated consumption goods using a Cobb-
Douglas production function:
yt = a, + can, + (1 - ca)m (1)
where yt is output, a, represents a s

(domestic) labor, and mt is imported
Production is either consumed by
(therefore, the economy exports con
mediate ones):
Yt = Sc Ct + Sx Xt (2)
where ct is domestic consumption, xt
Sx = Xss/yss.
The aggregate demand equations are:
Ct = Et ct + 1-- (it - Et +1) + +c,t (3)

Xt = nq, + yt (4)
where i, is the nominal inte
tion is nct = Pt - pt-,; Pt is th
ral elasticity of substitution
r1 is the elasticity of subst
is the real exchange rate, d
exchange rate, defined as th
eign currency; p* is the fore
Table 3 IMPORTS CLASSIFIED BY USE (2001)
Intermediate
Countries Consumption Capital (including fuel) Total
Australia 30.2 % 21.3 % 48.5 % 100.0 %
Brazil 12.8 % 26.6 % 60.6 % 100.0 %
Chile 19.8 % 21.0 % 59.3 % 100.0 %
Mexico 19.8 % 57.7 % 22.5 % 100.0 %
New Zealand* 24 % 21 % 40 % 85 %
Simple Average 21.3 % 29.5 % 46.2 %
*1999. Part of imports not classified by use.
Data source: Central banks and national institutes of statistics.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
of the world. The asterisk (*) indicates a variable of the rest of the world.
All variables of the rest of the world are treated as exogenous in the
model.
The model has domestic and foreign bonds, both private. Domestic
bonds are denominated in domestic currency and are held only by
domestic residents. Foreign bonds are denominated in foreign cur-
rency, and their prices include a stochastic country risk premium. The
derived uncovered interest rate parity condition (UIP) is presented
below:
it - it = E, st +, 1- s, + ?t (5)
where i,* is the interest rate in the rest of the wo
risk premium.
Firms maximize the difference between expected marginal revenue and
unit cost. There is price rigidity: only a fraction of the firms are allowed to
adjust prices each period. The choice of the optimum price for the firm
yields it, = E, ;t+l + v-v, where P is a discount factor, and vt is the real unit
cost given by:8
v, = Xa w, + (1 - a) pMt, - at - p, (6)
where wt is wage, and PMt is the price of imports in dom
defined as PMt = s, + p*,t (p*Mt is the price of imports in

Note that inflation is affected by the exchange rate via the
mediate goods.
Nevertheless, this formulation of the Phillips curve
counterfactual results, as stressed in Gall and Gertler (199
(1997). It implies that current changes in inflation are neg
to the lagged output gap; that is, a positive output gap wo
reduction in the inflation rate in the following period. In
empirical evidence is that a positive output gap is fol
increase in the inflation rate over the cycle. Moreover th
implies that, with perfectly credible announcements, a d
costless.
These empirical results have motivated some authors to work with a
hybrid Phillips curve: besides the expected inflation term, the equation
also contains a lagged term for inflation. In this paper, we do not
have a special concern about the specific derivation for the persistence in
8. X = (1- )( where 0 is the probability of the firm not adjusting its price in
period t.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
inflation.9 As in Gall and Gertler (1999), one possibility is to consider that

there is a fraction of backward-looking firms.10
In addition, we can also postulate a cost-push shock. This term could
reflect changes in the markup resulting from movements in the price elas-
ticity of demand over the cycle or in tax rates. It could also be used as a
proxy for the case of change in relative prices across sectors.11 The result-
ing aggregate supply equation is:
TC,= f Et1,nt+ I - yVtc " gb t (7)

where yb+ yf = 1.
We assume that the shocks follow stationary univariate autoregressive
processes of an order of 1.
The interest rate is the policy instrument. Monetary policy is given by a
Taylor-type simple rule or by optimal simple rules. In the latter, the opti-
mal coefficients are obtained by the minimization of the standard central
bank's intertemporal loss function that penalizes deviations of the output
gap (P,) and inflation from their targets:
i= 0
where wy is related negatively to the aversion to inflation var

is the inflation target, and P, = y, - Yt (Y, is the potential
as the output that would prevail in the case of full price f
9. Fuhrer and Moore (1995) have generated inflation persistence assumi

about relative wages over the life of the wage contract. Roberts (199
some empirical evidence that expectations are less than perfectly rati
the agents would have adaptive expectations or there would be a part
expectations (these would adjust only gradually to the fully ration
Gertler (1999) have found that the fraction of backward-looking firms i
nificant, although not quantitatively important.
10. For simplification, we are assuming that X is not affected by the pr
ward-looking term.
11. For example, in the case of Brazil, there has been an important chan
in the last years that is only partially related to the exchange rate and
international prices.
12. In the simulations, we have considered different values for w,. We
about the derivation of this parameter based on microfoundations. Th
tion can be derived from a household utility function in the presence
(Rotemberg, and Woodford, 1999; Woodford, 2003). Woodford (200
the objective function in the case where prices are indexed to a l
between the occasions on which they are reoptimized. The objective
quasi-differenced inflation rate term rather than the inflation rate itself
usual objective function based on the following grounds: (1) the ob
used only to get an idea of the optimal coefficients in the central bank's
(2) in practice, this objective function seems to be used more often by

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
interest rate is restricted to react to selected variables. We use the algo-

rithm in Dennis (2002) to estimate the optimal simple rule.13
4. Explaining the Higher Volatility

The conduct of monetary policy in EMEs faces at least three major chal-
lenges: (1) building credibility; (2) reducing the level of inflation; and (3)
dealing with fiscal, financial, and external dominance. The presence of
low credibility, inflation levels greater than the long-term goal, and large
shocks results in higher volatility of output, inflation, and the interest rate.
Fiscal and financial dominance issues also have implications for these
variables. In this section, we show how all these elements can help explain
the stylized facts presented previously. The explanation for the different
performance of EMEs relies on the presence of more fragile institutions
and imperfect credibility, on the necessity of reducing inflation levels, and
on the nature and magnitude of the shocks that hit these economies.
4.1 BUILDING CREDIBILITY AND REDUCING INFLATION RATE LEVELS
Institutions in emerging economies tend to be weaker than thos

oped economies. Central banks are no exception. In this context
tion of inflation targeting represents an effort to enhance the cred
the monetary authority as committed to price stability.
Nevertheless, building credibility takes time. During this tr
period, the central bank's actions not only have to be consisten
inflation-targeting framework, but they also have to take into acco
private agents do not fully trust that the central bank will act acco
Private agents have concerns about the commitment of the cent
the target itself and to its reaction to shocks. In the first case, giv
tory of low credibility, private agents assign some positive pr
that the central bank will renege on its commitment to the t
a result, the expected inflation and consequently the actua
tend to be higher than with a perfectly credible monetary au
Similarly, when the economy is hit by an inflationary sho
agents do not trust completely that the central bank will react stro
a consequence, the central bank incurs a cost of trust building
has to react to curb the inflationary pressures stemming from
bility and has to prove that it is committed to the new regim
some period, the volatility of the interest rate and output will
and, because the central bank also takes into account output c
inflation volatility also tends to be higher when compared to
of full credibility.
13. The explanation of the calibration of the model is available on request to th

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Figure 4 12-MONTH INFLATION RIGHT BEFORE IT ADOPTION
45Peru
40 Peru
35
a1 30 Chile
25
20 Israel Mexico
. 10
15 Poland Colombia Hungary
INew Zealand Canada Czech Republic o
S 5
Uited KingdoSouth
South Korea Brazil S
Korea Norway
0 Sweden* , Australia Switzerla
Dec. 1988 May 1990 Sep. 1991 Jan. 1993 Jun. 1994 Oct. 1995
Time
* Developed Economies a Emerging Market Economies
Imperfect credibility concerning the fulfillment of the targets becomes

more important when we consider that the role of inflation targeting in
emerging economies goes beyond assuring that inflation is kept at
approximately its long-term level. It must first assure that inflation con-
verges to low levels. In fact, emerging market countries have had to face
much higher initial inflation rates than have developed countries. Table
1 showed the inflation around the moment of adoption of inflation tar-
geting (right before and after), and the initial targets. When inflation tar-
geting was adopted, the average inflation in the developed countries was
3.7%, whereas in the EMEs, it was 13.1%.14 The values are also shown in
Figure 4. Half of the developing economies had a two-digit inflation rate
when implementing inflation targeting. In the case of Peru, Chile, Israel,
and Mexico, the inflation rate was 39.5%, 27.3%, 18.0%, and 18.6%,
respectively.
The differences are even clearer when we consider the first targets that
were established (Figure 5). In developed economies, the maximum
upper bound for the target was 6%, with an average of 2.8% for the cen-
tral target. For developing economies, however, the highest upper bound
reached 20%, and the average was 10.3%.
Because inflation was higher than the long-term goal, the targets were
decreasing in time. Figure 6 shows the evolution of the central target aver-
age for both country groups. They are relatively stable for developed
economies, and they are decreasing for EMEs.
If inflation targeting is adopted in an economy with an inflation rate
significantly higher than the long-term goal, the central bank has to con-
14. The low inflation rate in Brazil that prevailed prior to the adoption of inflation targeting
was partly a result of an overvalued exchange rate.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
duct an active policy to bring inflation down that leads to output costs.
The reduction in inflation faces two obstacles, which result in costly dis-
inflation and higher volatility of inflation and output: (1) the already men-
tioned imperfect credibility, and (2) the presence of some degree of
inflation persistence, resulting from some backward-looking behavior in
price setting. The presence of backward-looking behavior may be due to
factors such as indexed wage contracts, and adaptive expectations. In par-
ticular, in Brazil, the adjustment of regulated prices such as electricity and
telephone service follows a contractual rule that implies a high degree of
persistence.
Figure 5 FIRST CENTRAL TARGET ADOPTED

20
Chile
1 - Peru
14 - Mexico
CD Israel M c
) 12
S 12, Colom bia
10 South Korea Poland
Poland
() 8 - Brazil Hungary
- Czech Republic00
Canada CzechRepublic South Africa
4 -a Iceland
United Kingdom Australia Thailand +*
2 New Zealand * *
* Sweden Switzerland Norway
Dec. 1988 May 1990 Sep. 1991 Jan. 1993 Jun. 1994 Oct.
Time
. Developed Economies a Emerging M
Figure 6 INFLATION TARGET AVER

20
18
16
14
12"
W 10
w 8
6
2
2
0.
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year
- - Emerging Market Economies --Developed Economies

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Decreasing targets are a source of possible nonfulfillment of targets

because it is difficult to assess the current backward-looking behavior and
the speed of future convergence of inflation expectations. The higher
deviations from the targets in EMEs may then be related to the variabil-
ity of shocks as well as to the fact that inflation targeting was typically
adopted when inflation was significantly higher than the long-term
goal.
To have some indication of the speed of inflation convergence and of
the output costs involved, we have simulated the case of a reduction in
the inflation targets with imperfect credibility. To focus on this issue, we
have used the closed economy version of the model previously presented
(a = 1, s, = 1). The result is a standard model. The aggregate supply curve
without the backward-looking term can be written as:
;t = PEt nt +1+ I " Yt
where X* - k(yc + Yn). The central bank announces a reduction in the infla-
tion rate target from tTo to ItTn as of the current quarter, but private agents
do not fully believe that this change is permanent. They assign a proba-
bility bt that, at the following quarter, the central bank reneges on its
announcement and returns to nTo. Therefore, the expected inflation rate
target is given by E, t7z = bt To+ (1 - bt)7nTn. We have considered the case of
an optimal monetary policy under discretion.15 The central bank is
allowed to reoptimize every period. We have assumed that if the central
bank maintains the new target, the probability of reneging on its
announcement declines over time. We have used a simple law of motion:
bt+l = Pb b, where 0 < pb < 1.16
Figure 7 shows the impulse responses of the output gap and inflation to
a reduction of 3 percentage points in the inflation target for the cases of
imperfect and perfect credibility. If we compare the first announced tar-
gets to the inflation in the previous twelve months (Table 1), we verify
that many countries had initial targets more than 3 percentage points
lower than the previous inflation. We have assumed that the initial prob-
ability of reneging is 0.8, and Pb = 0.8. Because we are considering quarterly
data, the latter implies that, at the end of the year, b, = 0.41. The inflation
rate refers to the four-quarter accumulated inflation above the new target.
We assume that inflation was stable at the old target before the announce-
ment. At the end of the first year, in the case of imperfect credibility, the
15. We have not shown the results under commitment because it is less reasonable to assume
that private agents believe the central bank is committed and, at the same time, does not
keep its commitment.
16. The derivation is available on request to the authors.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Figure 7 IMPERFECT CREDIBILITY: IMPULSE RESPONSES TO A 3

PERCENTAGE POINT INFLATION TARGET REDUCTION
3.0 - Output Gap-imperfect Credibility

2.5-- --Four-Quarter Inflation--Imperfect Credibility
o - Output Gap--Perfect Credibility
2.0 ' -- Four-Quarter Inflation-Perfect Credibility
~c 1.5-
. 0.5-
0 .0 . . ......... .......... ----- ..... ....... .......... ....... .----...-- -
-0.5
-1.0
1 2 3 4 5 6 7 8 9 10 11 12
Time
inflation rate is still 1

output gap presented a
assuming a relatively ra
the inflation rate conv
perfect credibility, in
new target. As a resul
target, and this movem
Similar reasoning is a
shock. The possibility
time increases inflat
reduction. The result is
est rate.
As stressed by Svensson (2002), the economy incurs higher o

variability at the beginning of the regime to gain credibility, but i
efits later from an improved trade-off with lower output and inf
variability, and the central bank can then be a more flexible in
targeter. In the case of Brazil, the construction of credibility has
process that combines reactions to inflationary pressures and incr
transparency to the public. The Banco Central do Brasil has rea
inflation expectations in a way that is consistent with the inflation-targ
framework. Minella, Freitas, Goldfajn, and Muinhos (2003)18 have est
17. In this simulation, we are assuming that the relative output weight in the objective
tion is equal to 0.3. If we increase it to 1.0, the effect is significant on the output r
but low on the inflation path: the inflation rate is 1.45, and the output gap has a
tion of 0.15% on average.
18. That paper is an updated and shorter version of Minella, Freitas, Goldfajn, and M
(2002).

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
a reaction function for the BCB for the first three-and-a-half years of infla-
tion targeting. It relates the interest rate to deviations of the 12-month
ahead expected inflation from the target, allowing also for some interest-rate
smoothing and reaction to the output gap.19 Table 4 shows the estimations
using the inflation forecast of the BCB and of private agents.20 The point
estimates of the coefficient on inflation expectations are greater than 1 and
significantly different from 0 in all specifications. In most of the specifica-
tions, the coefficient is statistically greater than 1.21 Therefore, we can con-
clude that the BCB conducts monetary policy on a forward-looking basis
and responds to inflationary pressures.
We also simulate the case of reduction in the inflation targets in the
presence of a backward-looking component in the aggregate supply curve
equal to 0.4 (therefore, yf = 0.6). We use the open economy version of the
model. The central bank reacts according to a simple expectational rule:
i, = 1.5 Etnt+. The optimal coefficients found using two different weights
on the output gap in the objective function, wy = 0.3 and wy = 1.0, are 1.79
and 1.34, respectively.22 As in the case of imperfect credibility, inflation
decreases slowly to the new target, and the optimal output gap is nega-
tive. With a 3 percentage point reduction in the inflation target, inflation
is 0.78 percentage point above the target at the fourth quarter, and the out-
put gap reduction is 0.55% on average in the first year. In the case of a
purely forward-looking aggregate supply curve, inflation converges auto-
matically to the new target.
Because the inflation-targeting regime is supposed to affect inflation
expectations, we can consider the possibility that the backward-looking
component in the price adjustment becomes less important as credibility
increases. The share of backward-looking firms could become smaller
and/or firms could give less consideration to past inflation when adjust-
ing prices. Either situation would reduce the degree of persistence in
19. They have estimated the equation i, = ca it,1 + (1 - aC)(C + X2 (Et t+j - .rt+j) + 3 Yt-), where
i, is the Selic rate decided by the monetary policy committee (Copom), E, it+j equals
inflation expectations, and n~t+ is the inflation target, where j = 12, and yt is the output
gap.
20. Private agents' expectations are obtained from a survey that the Investors Relation
Group (Gerin) of the BCB conducts among banks and non-financial companies.
(Available at www.bcb.gov.br. Market Readout and Market Expectations Time Series of
Investor Relations Section).
21. The point estimates vary, however, across specifications. Using private agents' expecta-
tions (sample 2000:1-2002:12), the point estimates are around 2.1 to 2.3, whereas with
BCB expectations (sample 1999:6-2002:12), they are 3.5 and 5.7. The p-values for the test
that the coefficient is equal to 1 are 0.150, 0.101, 0.012, and 0.040 in specifications I, II, III,
and IV, respectively.
22. Cecchetti and Ehrmann (1999) have found a value between 0.32 and 0.41 for inflation-
targeting countries. Batini, Harrison, and Millard (2003) have used 1.0 (including also a
term for the interest rate in the objective function).

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Table 4 ESTIMATION OF REACTION FUNCTION OF THE BANCO

CENTRAL DO BRASIL: DEPENDENT VARIABLE: TARGET FOR THE
NOMINAL SELIC INTEREST RATE (MONTHLY DATA)'
Using central bank inflation Using market inflation

expectations expectations
(1999:7-2002:12) (2000:1-2002:12)
Regressors I II III IV
Constant 1.65 3.06* 4.58*** 5.38**
(1.08) (1.59) (1.52) (2.07)
Interest rate (t-1) 0.90*** 0.82*** 0.71*** 0.67***
(0.06) (0.09) (0.09) (0.12)
Inflation expectations 5.70* 3.54** 2.32*** 2.09***
(deviations from (3.20) (1.51) (0.53) (0.53)
the target)
Output gap(t-1) -0.36* -0.10
(0.21) (0.15)
R2 0.9129 0.9160 0.9205 0.9214
Adjusted R2 0.9084 0.9094 0.9157 0.9140
LM test for autocorrelation
of residuals (p-values)
1 lag 0.7853 0.7210 0.6586 0.6411
4 lags 0.6831 0.5298 0.5362 0.3991
1. Standard error in parentheses. *, ** and *** indicate the coe
levels, respectively.
Source: Minella, Freitas, Goldfajn, and Muinhos (2003).
inflation. Minella, Freitas, Goldfajn, and M

simple aggregate supply curve for the low
assess if the inflation-targeting regime w
tural change.23 They regress inflation rate
ment rate (lagged one period), and the exch
(lagged one period).
Table 5 records the results when includin
and when including two. The regressio
ables that multiply the constant and lagg
targeting period,24 and a dummy that assu
three months of 2002. Without adding th
both specifications present serial correlat
is a peculiar period, one that does not f
authors find that the backward-looking t
estimate of the autoregressive coefficient
23. The procedure is similar to the one in Kuttner an

24. Dummies for the inflation-targeting period that m
rate do not enter significantly; therefore, they were

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Table 5 ESTIMATION OF AGGREGATE SUPPLY CURVE: DEPENDENT

VARIABLE: MONTHLY INFLATION RATE (1995:08-2002:12)1
Regressors I II
Constant 0.65* 0.70*
(0.36) (0.36)
Dummy constantt 0.34*** 0.51***
(0.12) (0.14)
Inflation rate (t-1) 0.56*** 0.62***
(0.11) (0.15)
Inflation rate (t-2) -0.09
(0.14)
Dummy inflation rate (t-1)t -0.46*** -0.43*
(0.17) (0.19)
Dummy inflation rate (t-2)t -0.35*
(0.20)
Unemployment (t-1) -0.08 -0.09*
(0.05) (0.05)
Exchange rate change (t-1) (12-month average) 0.08* 0.09**
(0.04) (0.04)
Dummy 2002Q4t 1.42*** 1.47***
(0.26) (0.25)
R2 0.5593 0.6022
Adjusted R2 0.5271 0.5624
LM test for autocorrelation of resid
1 lag 0.6646 0.7022
4 lags 0.2218 0.3599
1. Standard error in parentheses. *,
level, respectively. Since exchang
1995:08 to avoid the inclusion of d
'Dummy has value 1 in the inflation
associated variable.
tDummy has value 1 in 2002:10-2002:12, and 0 otherwise.
Source: Minella, Freitas, Goldfajn, and Muinhos (2003).
the inflation-targeting period when compared to the previous period of

low inflation (specification I).25
4.2 DOMINANCE ISSUES: FISCAL, FINANCIAL, AND EXTERNAL
We deal here with three elements that seem to be potential features of EMEs:
weak fiscal regimes, the risks associated with poorly regulated financial sys-
tems, and large external shocks. Each of these problems can lead to a form
of dominance: fiscal, financial, or external. In the case of fiscal and financial
dominance, the problems that arise on the monetary policy front are quite
25. This result is in line with the findings in Kuttner and Posen (2001). Using a broad dataset
of 191 monetary frameworks from 41 countries, they found that inflation targeting
reduces inflation persistence.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
similar: the fear that one or both regimes will break down increases the
probability that the government will inflate in the future, and therefore
increases expected inflation. This in turn increases the challenge of estab-
lishing a solid monetary anchor. The external dominance refers to the vul-
nerability to external shocks, which results in higher macroeconomic
volatility.
4.2.1 Fiscal Dominance The success of inflation targeting or any other

monetary regime requires the absence of fiscal dominance. Therefore,
implementation of inflation targeting must be accompanied by a strong fis-
cal regime. But even with that, in the case of past weaknesses, it takes time
for government to gain the full confidence of private agents. This fear of
fiscal dominance affects inflation expectations, requiring a tighter mone-
tary policy, which in turn negatively affects the fiscal balance.
The challenge, therefore, is to build fiscal and monetary regimes that
reinforce one another. The evidence we have thus far on this issue is
promising, but it may be too early to celebrate. Schaechter, Stone
Zelmer (2000) show that the fiscal imbalance at the time inflation
ing was adopted was lower in developing countries.
4.2.2 Financial Dominance A problem for the conduct of monetary

can arise when there is fear that a tightening may lead to a financia
This may come as a consequence of a weak and/or overleveraged f
cial system and may bring about the expectation that monetary pol
not be conducted with the goal of defending the nominal anchor
economy. This problem can be characterized as a form of dom
which we can name financial dominance. For example, Goldfa
Gupta (2003) have found that, in the aftermath of currency crises, an e
omy that faces a banking crisis (1) has a lower probability of cho
tight policy, and (2) when tight monetary policy is adopted, the pr
ity of a successful recovery is lower.26
Banking sector weaknesses, and financial vulnerabilities in g
played a key role in the Asian crisis of 1997. This type of fragility may
be an issue when the financial system exhibits a significant prese
government-owned banks, either because these banks may thems
weak or because the government may use its banks unwisely. Jus
the case of fiscal dominance, it is necessary to work toward a stron
regime; in the case of financial dominance, care must be taken to
sure the regulation and supervision of the financial system is sou
permanent.
26. They have used a dataset of currency crises in 80 countries for the period 1980-1998.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Another issue is the predominance of short-term financial contracts and

lower financial depth in EMEs, which tend to weaken the monetary policy
transmission mechanisms.27 According to Schaechter, Stone, and Zelmer
(2000), the ratio of liquid liabilities to GDP was on average 51.0% in EMEs
and 72.0% in developed countries in 1997. The ratio of private credit to
GDP was 63.9% and 81.0%, respectively. As Taylor (2000) stresses, without
longer-term markets, short-term interest rates will have to move more
quickly. Therefore, we tend to observe higher interest-rate volatility.
4.2.3 External Dominance: Sudden Stops Another possible explanation for

the volatility shown in Section 2 is the existence of larger shocks in EMEs.
External shocks tend to play a more important role in EMEs than they do
in developed countries. EMEs are subject to sudden stops in capital
inflows. These shocks significantly affect the exchange rate, and conse-
quently the inflation rate, leading to higher interest rates to curb the infla-
tionary pressures. As a result, these economies tend to present a higher
volatility of interest rates and exchange rates.
Of course, sudden stops themselves may reflect weaker fundamentals,
which translate into lower credit ratings, among other problems. On the
other hand, the presence of large and frequent external shocks generates
greater instability in the economy and may jeopardize the fulfillment of
the targets, which in turn may negatively affect the credibility of the
regime. This may be seen as a form of external dominance. It must be
addressed through the strengthening of the fundamentals of the economy,
such as, in the case of inflation targeting, a fairly clean exchange-rate float
(by that, we mean the absence of an exchange-rate target) and in general
a sufficient degree of openness and flexibility.
The data presented in Table 2 confirmed the higher volatility of the
interest rate in EMEs. For the exchange rate, the data is less clear but
points to a higher volatility. The coefficient of variation (the ratio of the
standard deviation to the average) is 0.15 in EMEs and 0.11 in developed
economies. There are some differences, however, within the group of
developing economies. Brazil, Chile, Hungary, Peru, and South Africa
presented significantly greater exchange-rate volatility. The average of the
coefficient of variation of the five countries is 0.22.
To try to measure the importance of external shocks, we have run vec-
tor autoregression (VAR) estimations for selected countries. We have used
monthly and quarterly data of four variables: industrial production (or
GDP), the consumer price index, the interest rate, and the exchange rate.
27. Actually, there are two opposite effects in the case of short-term contracts: the wealth
effect of change in the interest rates is lower, but changes in the interest rates affect the
cost of outstanding debt more quickly.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
We have used a Cholesky decomposition with the order mentioned. We

have considered two specifications: (1) all variables in log level, and (2)
the price and exchange rate in first log-difference. We have estimated for
three periods: a large period (which varies across countries but in general
starts in the 1980s), the period before the adoption of inflation targeting,
and the inflation-targeting period. (All of them end in mid-2002.) Table 6
records the values for the variance error decomposition of the interest rate
and price level (or inflation rate), considering a 12-month or 4-quarter
horizon, for the inflation-targeting period. In particular, we show the per-
centage of the forecast error of the interest rate and prices (or inflation)
that is explained by shocks to the exchange rate. In Brazil and South
Africa, shocks to the exchange rate explain a significant part of the fore-
cast error of interest rate and prices. In Brazil, they explain 49% of the
interest rate forecast errors and 18% of the price forecast error (this figure
is not statistically significant though). South Korea has similar results
using monthly data, mainly for the interest rate. On the other hand, for
the developed economies and for Mexico, the estimations indicate that the
exchange rate does not play an important role.
Exchange-rate fluctuations do not always reflect the pressure of the
external shocks because of policy responses, which include interest-rate
changes and direct intervention in the exchange-rate market. Some
emerging markets opt not to allow the exchange rate to reflect the extent
of the external shocks. Some of the arguments are related to fear of float-
ing (Calvo and Reinhart, 2002).28 In addition to inflationary pressures,
significant exchange-rate fluctuations have other implications for the
economy, such as uncertainty concerning prices and the value of dollar-
denominated liabilities and assets. Huge depreciations of the domestic
currency may affect the financial solvency of firms and financial institu-
tions. In this case, the central bank may have additional goals in its
objective function. According to Amato and Gerlach (2002), several rea-
sons may make it appropriate for EMEs to give importance to the
exchange rate beyond that related to its inflationary effects: (1) with less
developed foreign exchange markets, large shocks or capital flows cause
significant volatility in the exchange rate if they are neglected by poli-
cies; (2) in economies with a poor history of monetary stability, the
exchange rate tends to be a focal point for inflationary expectations;
(3) exchange-rate fluctuations may have a large impact on the relative
profitability of firms across sectors; and (4) foreign currency borrowing
may be significant.
28. Reinhart and Rogoff (2002) have developed a system of reclassifying historical exchange-
rate regimes. They have found that regimes that were officially classified as floating in
reality use a form of de facto peg.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Therefore, it is worth examining the spread over treasuries of the for-

eign currency denominated debt of the country as another indicator of
external shocks. Table 7 shows the average and volatility of the spread of
the emerging market bond index plus (EMBI+) over the U.S. Treasury for
Brazil, Colombia, Mexico, Peru, Poland, and South Korea. Except for
Poland, the standard deviation is always higher than 118 basis points. In
the case of Brazil, it reaches 377 basis points in the inflation-targeting
period.
Table 6 VARIANCE ERROR DECOMPOSITION AFTER ADOPTION OF

INFLATION TARGETING (12-MONTH OR 4-QUARTER HORIZON)
(PERCENTAGE); VAR WITH FOUR VARIABLES: INDUSTRIAL PRODU
(GDP), CPI, INTEREST RATE, AND EXCHANGE RATE
Estimation in levels for price Estimation in first difference

and exchange rate for price and exchange rate
Shocks to the Shocks to Shocks to the Shocks to the
exchange rate the exchange exchange rate exchange
explaining the rate explaining explaining the explaining
forecast error the forecast forecast error forecast error
for the interest error for for the interest for inflation
rate inflation rate
Monthly Datat
Developed Economies
Canada 1 0 1 1
Sweden 2 8 4 6
United Kingdom 0 17 10 5
Emerging Market Economies
Brazil 49* 18 8 29*
Mexico 2 1 0 0
South Africa 36* 25* 36* 25
South Korea 26* 7 32 9
Quarterly Datat
Developed Economies
Australia 12 3 1 11
Canada 0 1 2 2
New Zealand 11 3 4 3
Sweden 4 1 4 10
United Kingdom 0 0 1 1
Emerging Market Economies
Brazil 57* 35 10 20
Mexico 1 0 1 12
South Korea 3 3 5 8
*Indicates that the value is sig

tNew Zealand and Australia d
tThe sample for South Africa

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
As a result of these external pressures, we tend to observe a higher

volatility of output, interest rates, and the exchange rate, and the possi-
bility of nonfulfillment of targets tends to be higher. In the case of Brazil,
the significant depreciation of the domestic currency was the main factor
behind the nonfulfillment of the inflation targets in 2001 and 2002. The
nominal exchange rate in Brazil accumulated an increase of 84.7% during
2001 and 2002 (a depreciation of the domestic currency of 45.9%), repre-
senting a real depreciation of 44.6%.
We use the model to simulate the effects of a shock to the real
exchange rate of 45%, but instead of using the UIP condition, we
autoregressive process for the real exchange rate (with an autoreg
coefficient equal to 0.5). As before, the central bank reacts accordi
simple expectational rule: i, = 1.5E, 7t,+, and the coefficient on la
inflation is 0.4.
Figure 8 shows the impulse responses. At the end of the fourth quarter,
the four-quarter accumulated inflation reaches 5.16%, and the output gap
reduction is on average 6.80% in the first year. Therefore, even with a sig-
nificant output gap reduction, any existing inflation target would be
breached. If we consider the inflation target for Brazil in 2002 of 4%, then
with this simulation, the inflation rate would reach 9.7%, distant even
from the upper bound of 6.5%.
The existence of this possibility of breaching the targets due to large
shocks leads us to analyze important issues in the design of inflation tar-
geting in emerging markets. The next section explores how to deal with
higher volatility in an inflation-targeting regime.
Table 7 EMBI+ (BASED ON MONTHLY AVERAGE)
1997:01-2002:12* After IT adoption until 2002:12t

Coefficient Coefficient
of variation of variation
Standard (s.d./ Standard (s.d./
Country Average deviation average) Average Deviation average)
Brazil 879.35 389.07 0.44 987.14 377.82 0.38
Colombia 651.18 118.48 0.18 648.04 123.83 0.19
Mexico 443.40 150.03 0.34 408.37 130.19 0.32
Peru 601.75 120.64 0.20 601.75 120.64 0.20
Poland 233.03 38.17 0.16 232.65 35.88 0.15
South Korea 236.37 145.39 0.62 236.37 145.39 0.62
*Colombia: 1999:05-2002:12. Peru: 1998:01-2002:12. Poland: 1998:01-2002:12. South Korea: 1998:

05-2002:12. For Brazil and Mexico, the data for 1997 refers to EMBI.
tSouth Korea: 1998:05-2002:12.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
5. How to Deal with Higher Volatility?

The central element of the inflation-targeting regime is the public
announcement of a numerical target for inflation. As Mishkin (2002)
stresses, however, inflation targeting comprises much more than that. It
also involves (1) an institutional commitment to price stability as the pri-
mary goal of monetary policy, to which other goals are subordinated; (2)
the use of many variables for the instrument policy decision; (3) increased
transparency of monetary policy strategy; and (4) an increased account-
ability of the central bank.
The issues presented in the previous section imply important chal-
lenges for monetary policy. The key aspect is how to build credibility in
the conduct of monetary policy and in the inflation-targeting regime itself,
and at the same time remain flexible enough to avoid unnecessary output
costs that could lead to a perception that the regime is too costly.
Communication and transparency become crucial.
One of the most appealing features of inflation targeting is the flexibil-
ity it allows monetary policy when the economy is confronted with
shocks. For instance, in dealing with a supply shock, the professional con-
sensus among academic economists and central bankers is that a central
bank should accommodate the direct price-level impact of the shock
while calibrating monetary policy to avoid further rounds of price
increases. In practice, however, this approach may come at a cost. The cen-
tral bank's commitment to low inflation may be questioned if the nature
of the shock and the appropriateness of the policy response are not clear
Figure 8 IMPULSE RESPONSES TO A 45 PERCENTAGE POINT REAL

EXCHANGE RATE SHOCK
50 - Output Gap
...40 .. ...... Four-Q
4-a-- Real Exc
30 ...... ... Interest Rate (Annualized)
2 20-
20 -
0-O
-10
-20 -
1 2 3 4 5 6 7 8 9 10 11 12
Time

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
to most observers. The solution to this conundrum is to provide enough

information to the public to clarify that the policy response is the right
answer to a well-understood problem.
In general, the optimal response depends on the nature of the shock, on
several economic parameters and elasticities, and on the preferences of
society relative to the inflation versus output gap volatility trade-off. At a
basic level, it is crucial therefore that the central bank make an effort to
identify the size and nature of the shocks, focusing in particular on
whether one is dealing with supply or demand shocks, temporary or per-
manent shocks, and the size and inflationary impact of the shocks. Once
the shocks are identified, the central bank can choose a monetary policy
response that will deliver the chosen feasible pair of inflation and output
gap paths.
Inflation-targeting central banks have developed several tools to deal
with these issues. They involve the inflation-targeting design, trans-
parency, adjusted targets, and a change in the IMF conditionality.
5.1 TARGET BANDS, HORIZONS, AND PERSISTENCE OF SHOCKS
The possibility that large shocks may cause target breaches leads to an
important feature in the design of the inflation-targeting regime: the size
of the band around the central point of the target. Tighter bands tend to
signal a preference for lower inflation volatility relative to lower output
volatility. The band is typically seen as a barrier not to be broken.
In a world of perfect information, however, where all shocks are pre-
cisely identified, there is no role for bands around the inflation target.
Deviations from the point target would occur as an optimal response to
shocks, given the parameters of the economy and the inflation aversion of
society. An optimal response to a very large shock may demand large
deviations from the central point of the target, sometimes beyond the
upper bound of the target. The same holds for the horizon over which
inflation is allowed to deviate from the target when the economy is hit by
shocks. This horizon should also be determined according to the type,
size, and persistence of the shock as well as the parameters mentioned
above.
So why do countries opt to include target bands? While some countries
may treat the band limit as a strict barrier not to be broken, in our view the
bands should be treated mainly as a communications device. The bands
should be considered mainly as checkpoints, with the central bank
explaining clearly the reasons for the nonfulfillment of the targets. This
discussion is not easily translated, however, into operational guidelines
that can be implemented by the central bank. As a result, it is necessary
that the assumptions underlying the decisionmaking process of the centra

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
bank be communicated clearly. This approach means being explicit about

a fairly precise path of inflation on the way back to the targeted level to
avoid losing the confidence of economic agents. Transparency therefore
plays the key role of imposing enough discipline on the central bank to
avoid the temptations depicted in the time-consistency literature.29
In practice, the size of the bands varies across countries: 1 percentage
point in Australia and Israel; 3 percentage points in South Africa; 3.5 in
Iceland; and 4 in Brazil for 2002, and 5 for 2003 and 2004. What should be
the size of the bands? One possibility would be to keep them large
enough to allow the inflation rate within the bands in most of the cir-
cumstances if monetary policy is conducted efficiently. Given some vari-
ability of shocks in the economy, inflation should be inside the bands in
most, say, 90%, of the cases. Using the model, and assuming some vari-
ance for the shocks, we obtain the standard deviation of inflation (given
some optimal rule). If we assume that the random shocks are normally
distributed, we would find a band size corresponding to 1.65 standard
deviations of inflation.
The recurrent presence of larger shocks may also recommend higher
central targets. As shown in Figure 6, the difference between the targets of
EMEs and developed countries has decreased, but it is still positive. In
2002, the central point target for EMEs was 3.7%, whereas for developed
countries it was 2.2%. The higher target reflects not only higher past infla-
tion but also the greater vulnerability to external shocks. One possible rea-
son is the asymmetric effects of supply shocks. Given greater downward
price rigidity, deflationary shocks tend to have a lower effect in inflation
than do inflationary shocks. As a consequence, with higher shocks, this
bias tends to be higher.
The issue of the magnitude of the response to shocks leads us also to the
discussion of the horizon to be used in the inflation-targeting framework.
As emphasized by Svensson (2002), in practice, flexible inflation targeting
(where some weight is given to output stabilization) means aiming to
achieve the inflation target at a somewhat longer horizon. Because EMEs
are more subject to larger shocks, their target horizon should naturally be
longer. A danger here, of course, is that if the central bank is still building
credibility, longer horizons could be interpreted as lenience, thus affecting
the central bank's reputation.
In practice, there is no magic number for the horizon that a central bank
should use to guide its reaction to supply shocks. It should be long
29. It seems these days that the important lesson of the time-consistency literature has reached
most central banks. In fact, one finds a substantial number of central bankers around the
world who seem to act like the inflation-averse central bankers in Rogoff (1985).

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
enough to allow the workings of the monetary policy transmission mech-

anisms and some degree of smoothing of the effects of the shock. On the
other hand, it should be short enough to neutralize part of its inflationary
effect and allow convergence of inflation expectations to the target. At any
rate, it is crucial that the central bank's response be clearly explained to
avoid reputational risk. Section 5.3 presents the procedure adopted by the
BCB in dealing with a series of large shocks during the years 2001 and
2002.
5.2 MONETARY POLICY COMMITTEES, MEETING MINUTES, AND AN

INFLATION REPORT
To stress the importance of the existence of a monetary policy c

(MPC) may sound unnecessary to most observers in more
economies, but it is not a trivial matter in the EMEs. Until one s
mittee was created in Brazil, for example, monetary policy decis
made on an ad-hoc basis, typically at the end of a board meetin
end of the day, when everyone was already quite exhausted, an
without the benefit of proper preparation and analysis. An
meets regularly, on a monthly basis, has created a proper environ
what, after all, is the key role of a central bank: to run monetar
The MPC meetings have become a ritual that provides those res
for setting policy with a well-informed decisionmaking enviro
moves board members away from their otherwise hectic da
schedules of meetings and phone calls and allows them to focus
task at hand.
A crucial aspect of inflation targeting is the ability to enhance the cred-
ibility of the policymaking process and, as a result, to achieve the desired
goals with minimum costs. The timely publication of the detailed minutes
of MPC meetings is a key ingredient for an effective communications
strategy. In emerging economies, where credibility is typically lower than
one would like, the benefits of publishing this information can be sub-
stantial.
In addition to monthly meeting minutes, most inflation-targeting cen-
tral banks also publish a quarterly inflation report where their views on
economic prospects and, in particular, on inflation trends are presented in
detail."0 Again, for EMEs, these reports play a key role, serving the pur-
pose of minimizing uncertainty about the central bank's analysis and
goals. Inflation reports are appropriate vehicles for the central bank to
present its views on complex issues such as the degree of exchange-rate
30. For an assessment of inflation reports by inflation-targeting central banks, see Fracasso,
Genberg, and Wyplosz (2003).

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
pass-through, the degree of inflationary persistence, the workings of the

transmission mechanism, and so on.
5.3 SHOCKS AND ADJUSTED TARGETS
The recent experience of Brazil with inflation targeting during turbulent

times serves to illustrate the practical application of the general guidelines
and principles discussed above. This section summarizes the methodol-
ogy currently used in Brazil. It calculates the inflationary impact of cur-
rent supply shocks as well as the secondary impact of past shocks (due to
inertia in the inflation process). The idea is simply to accommodate the
direct impact of current shocks and to choose a horizon to weed out the
secondary impact of past shocks.
When facing shocks, the BCB initially considers the nature and persist-
ence of the shock. Then it builds different inflation and output trajectories
associated with different interest-rate paths. Based on its aversion to infla-
tion variability, it chooses the optimal path for output and inflation. Banco
Central do Brasil (2003) has published this path and also the outcome of
different paths. This is in line with Svensson's (2002) recommendations.31
If shocks are large and/or persistent, however, their inflationary effects
may last one year or more. The optimal inflation path may imply a
12-month ahead inflation superior to the previous annual target.
Therefore, in this situation, because the BCB would not be targeting the
previous inflation target, it uses an adjusted target. More specifically, the
target is adjusted to take into account primary effects of change in relative
prices and of past inertia that will be accommodated. The new target is
publicly announced. Although there is a credibility loss stemming
from the target change itself, the gains in terms of transparency and com-
munication are more significant. Private agents know the target pursued
by the BCB. Keeping the old target would affect the credibility of the
BCB because it could be considered unattainable. In the concept of the
adjusted target, the primary effect of the shock to regulated-price inflation,
and the inflation inertia inherited from the previous year to be accommo-
dated in the current year, are added to the target previously set by the gov-
ernment. Facing cost shocks, such as the increase of regulated prices above
the inflation of the other prices of the economy, monetary policy should be
calibrated to accommodate the direct impact of shocks on the price level
but to fight their secondary effects. Furthermore, because the BCB
also takes into account output costs, the inertial impacts of the previous
year's inflation should not necessarily be fought completely.
31. Svensson's (2002) recommendations also involve publishing the corresponding instru-
ment-rate plan.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Indeed, significant shocks, such as increases in the price of regulated

utilities and the exchange rate, have been one of the main challenges faced
by the BCB. Since the implementation of the Real Plan in July 1994, regu-
lated-price inflation has been well above the market-price inflation for
various reasons. Since the start of the inflation-targeting period, the ratio
of regulated prices to market prices rose 31.4% (1999:7-2003:2). As long as
there is some downward rigidity in prices, changes in relative prices are
usually translated into higher inflation. If these increases are treated as a
supply shock, monetary policy should be oriented toward eliminating
only their secondary impact on inflation while preserving the initial
realignment of relative prices. Therefore, the efforts of the BCB to quantify
the first-order inflationary impact of the regulated-price inflation have
become particularly important because they help to implement monetary
policy in a flexible manner, without losing sight of the larger objective of
achieving the inflation targets.
The first-order inflationary impact of the shock to regulated items is
defined as the variation in regulated prices exceeding the target for the
inflation rate, weighted by the share of regulated prices in the IPCA (con-
sumer price index) and excluding the effects of the inflation inertia from
the previous year and of variations in the exchange rate:
ShA = (tadm -n*)* adm - (IA + CaA) (9)

where ShA = first-order inflationary impact of
?Iadm = inflation of regulated prices
n* = target for the inflation
Madm = weight of regulated prices in the IPCA
IA = effect of inertia in the previous year on
ulated prices
CaA = effect of the exchange-rate variation o
ulated prices
The effect of inflation inertia is excluded becau

mechanisms should be neutralized by the monet
deemed to be appropriate (again, based on inflat
parameters). The exchange-rate variation is exclu
by monetary policy and could reflect demand s
exchange-rate changes were automatically inc
would be validating any inflationary pressure co
rate (including demand pressures). Therefore, in
ulated prices, only the component of relative pri
mined or backward-looking, and therefore canno
policy in the short run, is preserved as a first-orde

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Table 8 ADJUSTED TARGETS FOR 2003 AND 2004
Itemization 2003 2004
(a) Target for inflation set by the government 4.0 3.75

(b) Regulated-price shocks' 1.7 1.1
(c) Inertia not to be fought in the current year2 2.8 0.6
Inherited inertia from the previous year (total) 4.2 1.0
of regulated prices 1.4 0.4
of market prices 2.8 0.6
(d) Adjusted targets = (a) + (b) + (c) 8.5 5.5
1. For the calculation of the shock, the effect of inertia and the exchange rate
is deducted.
2. The inertia not to be fought in the current year corresponds to % of the inertia inherited from the pr
vious year. Source: Banco Central do Brasil (2003).
Table 8 shows how this methodology was applied to Brazil for inflatio
in 2003 and 2004.32 The target established by the government for 2003
4%. However, the regulated-price shocks are estimated at 1.7%. This valu
represents the expected contribution for the overall inflation of a chan
in relative prices that is not related to the inflation inertia from the previ-
ous year and to the exchange-rate change. Since these first-round effec
should not be neutralized, they are added to the target of the BCB, lead
ing to an adjusted target of 5.7%.
Furthermore, the BCB also takes into account the nature and persist
ence of the shocks and the output costs involved in the disinflationary
process (the output weight in the objective function is greater than zer
In this case, the BCB decided to fight against one-third of the inertia inher-
ited from the previous year. This inertia is estimated at 4.2%. Therefor
we have to add 2.8% to the target, leading to an adjusted target of 8.5%
which was publicly announced.
The decision to pursue an inflation trajectory based on these adjusted
targets considers that monetary policy will be able to lead inflation to co
verge to the target tolerance interval in two years. We should stress, ho
ever, that two years is not a magic number. It depends on the size and ty
of the shock. Figure 9 draws the actual and expected path for inflatio
(Banco Central do Brasil, 2003). The trajectory is compatible with the (en
of-year) adjusted targets.
Other trajectories with steeper decreases of inflation imply an excessiv
loss of output. Simulations indicate that a trajectory of inflation that
reaches 6.5% in 2003, the ceiling of the target tolerance interval, woul
32. See Banco Central do Brasil (2003). For a more detailed explanation of the methodology
see Freitas, Minella, and Riella (2002).

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Inflation Targeting in Emerging Market Economies * 395
imply a 1.6% drop in GDP. A trajectory that reaches the center of the tar-
get, 4%, in 2003, would imply an even larger decline in GDP (-7.3%).
One could argue that the decision to neutralize the shock in a longer time
horizon, based on an evaluation of the size and persistence of the shock,
may lead to time-consistency issues: too much accommodation in the short
run could lead to a loss of credibility in the long run. In fact, it is essential
that the whole procedure be explained publicly in detail so that the agents
can judge effectively whether the size and persistence of the shock justify
the decision taken by the central bank. It is the transparency, therefore, that
imposes enough discipline to avoid time-consistency issues.
In the inflation-targeting design, a core inflation measure or the estab-
lishment of escape clauses has also been used or suggested as a way of deal-
ing with shocks and volatilities. The main argument contrary to the use of
core inflation is that it is less representative of the loss of the purchasing
power of money at a given point in time. Agents are concerned about the
whole basket of consumption. In the case of Brazil, exclusion of the regu-
lated price items would leave out more than 30% of the representative con-
sumption basket. In this sense, private agents may question a monetary
policy that is not concerned about the overall consumer price index.
In general, there are two advantages to the use of the adjusted target
procedure. First, the core inflation measure is not necessarily isolated
from the effect of shocks. For example, the large depreciation shock of the
Brazilian economy in 2002 led to a core inflation-calculated by the sym-
metric trimmed mean method-of 8.8%, way above the inflation target.
Second, the construction of the adjusted target is directly based on the
idea that monetary policy should neutralize second-order effects of supply
shocks and accommodate the first-round effects, and on the fact that some
Figure 9 TWELVE-MONTH INFLATION

16
14 Dec.:
12 - 8.5%Dec.:
I1 5.5%
2002 2003 2004

Year

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
weight to output volatility should be assigned in the objective function.

Therefore, the principles under which the monetary policy is conducted
become more transparent.
In the case of escape clauses, the circumstances under which the central
bank can justify the nonfulfillment of the targets are set in advance. It has
more similarities with the adjusted target procedure than with the use of
a core inflation measure because it does not exclude items from the infla-
tion target but defines circumstances in which the breach of targets can be
justified. The main advantages of the adjusted target procedure are the
following: (1) it is a forward-looking procedure, (2) it defines clearly
the new target to be pursued by the central bank, and (3) it explains how
the new target is measured.
5.4 IMF PROGRAMS AND CONDITIONALITY
This section focuses on the IMF conditionality in the case of a co

under an inflation-targeting regime.33 We stress two issues: (1) the
ished role of net domestic assets (NDA) conditionality, and (2)
insert inflation performance as a criterion for the assessment of
tary policy stance.
5.4.1 NDA Conditionality Versus Inflation Targeting Brazil was th

country under an inflation-targeting system to have an agreeme
the IMF. From a theoretical point of view, the NDA conditionality
is usually the one found in the agreements to evaluate the stance o
etary policy, is not adequate for an inflation-targeting regime bec
harms transparency and can force the central bank to take unne
monetary policy actions.
Money demand is unstable, and the monetary aggregates seem
poor predictors of inflation. Therefore, autonomous increases in the lev
money demand would require, in the case of NDA ceilings or mon
base targeting, an increase in the interest rate without any inflation-t
ing policy purpose. We would then observe higher volatility in the
rate than needed. Furthermore, the imposition of an NDA ceiling c
transparency in the sense that it would add to the inflation target
monetary policy goal. One of the main advantages of an inflation-t
regime is the definition of a clear target for monetary policy. The exis
of another target affects the credibility of the main goal of monetary
5.4.2 Inflation Performance As a Criterion for the Assessment of a Mon

Policy Stance In place of NDA targets, inflation performance emer
33. See also Blejer, Leone, Rabanal, and Schwartz (2002), and Bogdanski, Freitas, G
and Tombini (2001).

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
natural criterion for the assessment of a monetary policy stance. Of course,

because inflation targets focus on inflation forecasts, assessing inflation
outcomes has to account for the shocks that hit the economy. At least two
issues have to be addressed in the case of an IMF program: the frequency
of the assessments (reviews), and the criteria on which to base the targets.
IMF programs have quarterly reviews, whereas in an inflation-targeting
framework, inflation performance is assessed at longer horizons. The use of
annualized quarterly inflation figures as targets is not recommended because
they are more volatile and subject to the strong influence of temporary shocks.
Brazil has used a 12-month inflation as the monetary quarterly target in the
technical memorandum of understanding with the IMF. It also includes inner
and outer bands of 1.0 percentage point and 2.5 percentage points, respec-
tively, both above and below the central targets. Figure 10 shows the targets
agreed to by the IMF and the BCB in the second review of Brazil's performance
in March 2003. The target for the 12-month inflation of the quarter following
the agreement considers the inflation verified in the three previous quarters
plus an estimate for the next quarter inflation. If an important shock hits the
economy in the next quarter, however, inflation may breach the target.
The path of the targets with the IMF should be consistent with the annual
targets of the inflation-targeting regime. This approach is in line with the
forward-looking or pre-emptive nature of the inflation-targeting system. At
the same time, it eliminates the problem of the effect of quarterly figures
because the time horizon for the inflation target could be defined as four
quarters. Eventual (predicted) pickups of inflation during the year should
not alter the monetary policy committee decisions, provided inflation is
expected to converge to the target and provided shocks in one year keep
affecting inflation in the following year, but only through the direct chan-
nel, that is, via the autoregressive component of inflation.
Figure 10 INFLATION TARGETS (CENTRAL POINT AND UPPER AND

LOWER BOUNDS) AGREED WITH THE IMF
18
16
14
.14 .. , ... ................... ..
2 1
2002:3 2002:4 2003:1 2003:2 2003:3

Year: Quarter

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
The main difficulty arises with the necessity of agreement between the
IMF and the central bank concerning the model used and the associated
risks, and reaching this agreement may also be time consuming.
Therefore, it also involves the construction of credibility by the forecast-
making process of the central bank.
In the agreements between Brazil and the IMF, the inflation forecasts
made by the BCB have been used as an important criterion for the def-
inition of the targets. However, the revisions are still made on a quar-
terly basis based on the actual inflation. The reasons for the
nonfulfillment of targets have been explained thoroughly to the public
and to the IMF.
6. Conclusion
Inflation targeting in EMEs has been relatively successful but has

proven to be challenging. The volatility of output, inflation, the interest
rate, and the exchange rate has been higher than in developed coun-
tries. Several issues have led these economies to face this less favorable
trade-off. The process of building credibility, the necessity of reducin
inflation levels, the dominance issues, and the larger shocks have a
played an important role. To deal with this more volatile environment,
we recommend (1) high levels of communication and transparency, (2)
target bands treated mainly as communications devices, (3) a method
ology for calculating the convergence path following a shock (adjusted
targets), and (4) better IMF conditionality under inflation targeting.
REFERENCES
Amato, Jeffery D., and Stefan Gerlach. (2002). Inflation targeting in emerging
ket and transition economies: Lessons after a decade. European Economic R
46:781-790.
Ball, Laurence, and Niamh Sheridan. (2003). Does inflation targeting matt
Cambridge, MA: National Bureau of Economic Research. NBER Working Pap
No. 9577.
Banco Central do Brasil. (2003). Open-letter sent by Banco Central do Brasil's gov
ernor, Henrique de C. Meirelles, to the minister of Finance, Antonio Palocci
Filho. Available at www.bcb.gov.br (accessed March 6, 2003).
Batini, Nicoletta, Richard Harrison, and Stephen P. Millard. (2003). Monetary pol
icy rules for an open economy. Journal of Economic Dynamics and Control
27(11-12):2059-2094.
Bernanke, Ben S., Thomas Laubach, Frederic S. Mishkin, and Adam S. Posen
(1999). Inflation Targeting: Lessons from the International Experience. Princeton, NJ
Princeton University Press.
Blejer, Mario I., Alfredo M. Leone, Pau Rabanal, and Gerd Schwartz. (2002
Inflation targeting in the context of IMF-supported adjustment programs. IM
Staff Papers 49(3):313-338.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Bogdanski, Joel, Paulo S. de Freitas, Ilan Goldfajn, and Alexandre A. Tombini.

(2001). Inflation targeting in Brazil: Shocks, backward-looking prices, and IMF
conditionality. Central Bank of Brazil. Working Paper Series No. 24.
Calvo, Guillermo, and Carmen Reinhart. (2002). Fear of floating. Quarterly Journal
of Economics 117(2):379-408.
Cecchetti, Stephen G., and Michael Ehrmann. (1999). Does inflation targeting
increase output volatility? An international comparison of policymakers' pref-
erences and outcomes. Cambridge, MA: National Bureau of Economic
Research. NBER Working Paper No. 7426.
Corbo, Vittorio, Oscar Landerretche, and Klaus Schmidt-Hebbel. (2002). Does
inflation targeting make a difference? Inflation Targeting: Design, Performance,
Challenges, In Norman Loayza and Raimundo Soto (eds.). Santiago, Chile:
Central Bank of Chile, 221-269.
Corbo, Vittorio, and Klaus Schmidt-Hebbel. (2001). Inflation targeting in Latin
America. Central Bank of Chile. Working Paper No. 105.
Dennis, Richard. (2002). Solving for optimal simple rules in rational expectations
models. Federal Reserve Bank of San Francisco. Mimeo, Sept. Journal of
Economic Dynamics & Control (forthcoming).
Fracasso, Andrea, Hans Genberg, and Charles Wyplosz. (2003). How do central
banks write? An evaluation of inflation reports by inflation-targeting central
banks. Geneva Reports on the World Economy Special Report 2. Geneva,
Switzerland: International Center for Monetary and Banking Studies (ICMB);
London: Center for Economic Policy Research (CEPR).
Freitas, Paulo S. de, Andre Minella, and Gil Riella. (2002). Metodologia de Calculo
da Inercia Inflaciondria e dos Efeitos dos Choques dos Preqos Administrados.
Notas Thcnicas do Banco Central do Brasil No. 22. English translation in Banco
Central do Brasil. (2002). Inflation Report 4(2):123-128.
Fuhrer, Jeffrey. (1997). The (un)importance of forward-looking behavior in price
specifications. Journal of Money, Credit, and Banking 29(3):338-350.
Fuhrer, Jeffrey, and Georger Moore. (1995). Inflation persistence. Quarterly Journal
of Economics 110(1):127-159.
Gall, Jordi, and Mark Gertler. (1999). Inflation dynamics: A structural econometric
analysis. Journal of Monetary Economics 44:195-222.
Gall, Jordi, and Tommaso Monacelli. (2002). Monetary policy and exchange rate
volatility in a small open economy. Cambridge, MA: National Bureau of
Economic Research. NBER, Working Paper No. 8905.
Goldfajn, Ilan, and Poonam Gupta. (2003). Does monetary policy stabilize the
exchange rate following a currency crisis? IMF Staff Papers 50(1):90-114.
Kuttner, Kenneth N., and Adam S. Posen. (1999). Does talk matter after all?
Inflation targeting and central bank behavior. Federal Reserve Bank of New
York. Staff Report No. 88.
Kuttner, Kenneth N., and Adam S. Posen. (2001). Beyond bipolar: A three-dimensional
assessment of monetary frameworks. International Journal of Finance and
Economics 6:369-387.
McCallum, Bennett T., and Edward Nelson. (2000). Monetary policy for an ope
economy: An alternative framework with optimizing agents and sticky price
Oxford Review of Economic Policy 16(4):74-91.
Minella, Andre, Paulo S. de Freitas, Ilan Goldfajn, and Marcelo K. Muinho
(2002). Inflation targeting in Brazil: Lessons and challenges. Central Bank of
Brazil. Working Paper Series No. 53.

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00
Minella, Andre, Paulo S. de Freitas, Ilan Goldfajn, and Marcelo K. Muinhos.

(2003). Inflation targeting in Brazil: Constructing credibility under exchange
rate volatility. Journal of International Money and Finance 22(7):1015-1040.
Mishkin, Frederic. (2000). Inflation targeting in emerging market countries.
American Economic Review 90(2):105-109.
Mishkin, Frederic, and Klaus Schmidt-Hebbel. (2002). A decade of inflation tar-
geting in the world: What do we know and what do we need to know? In
Norman Loayza and Raimundo Soto (eds.). Inflation Targeting: Design,
Performance, Challenges. Santiago, Chile: Central Bank of Chile, 171-219.
Reinhart, Carmen M., and Kenneth S. Rogoff. (2002). The modern history of
exchange rate arrangements: A reinterpretation. Cambridge, MA: National
Bureau of Economic Research. NBER Working Paper No. 8963. Quarterly
Journal of Economics (forthcoming).
Roberts, John M. (1997). Is inflation sticky? Journal of Monetary Economics 39:173-196.
Roberts, John M. (1998). Inflation expectations and the transmission of monetary
policy. Board of Governors of the Federal Reserve System. Finance and
Economics Discussion Paper Series, no 98/43, October.
Rogoff, Kenneth. (1985). The optimal degree of commitment to an intermediate
monetary target. Quarterly Journal of Economics 100(4):1169-1189.
Rotemberg, Julio J., and Michael Woodford. (1999). Interest-rate rules in an esti-
mated sticky price model. In John B. Taylor (ed). Monetary Policy Rules. NBER
Business Cycles Series, Vol. 31. Chicago and London: University of Chicago
Press, pp. 57-119.
Schaechter, Andrea, Mark P. Stone, and Mark Zelmer. (2000). Adopting inflation
targeting: Practical issues for emerging market countries. International
Monetary Fund. Occasional Paper No. 202.
Schmidt-Hebbel, Klaus, and Alejandro Werner. (2002). Inflation targeting in Brazil,
Chile, and Mexico: Performance, credibility and the exchange rate. Central Bank
of Chile. Working Paper No. 171.
Svensson, Lars E. 0. (2000). Open-economy inflation targeting. Journal of
International Economics 50:155-183.
Svensson, Lars E. 0. (2002). Monetary policy and real stabilization. Princeton
University Mimeo. Paper presented at "Rethinking Stabilization Policy", a sym-
posium sponsored by the Federal Reserve Bank of Kansas City and held at
Jackson Hole, Wyoming, August 29-31, 2002.
Taylor, John B. (2000). Using monetary policy rules in emerging market
economies. Stanford University. Mimeo.
Truman, Edwin M. (2003). Inflation targeting in The World Economy. Washington,
D.C.: Institute for International Economics.
Woodford, Michael. (2003). Interest and Prices: Foundations of a Theory of Monetar
Policy. Princeton NJ: Princeton University Press.
Comment
ROBERT E. HALL
Stanford University and NBER
A nation dealing with its central bank is a leading example of pr

agent mechanism design in an imperfectly understood environm

50.231.1.2 on Mon, 14 Aug 2023 22:51:00 +00:00

Fraga InflationTargetingEmerging 2003

Uploaded by

Copyright:

Available Formats

Fraga InflationTargetingEmerging 2003

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fraga InflationTargetingEmerging 2003

Uploaded by

Copyright:

Available Formats

Inflation Targeting in Emerging Market Economies

Author(s): Arminio Fraga, Ilan Goldfajn and André Minella

Stable URL: https://www.jstor.org/stable/3585264

This content downloaded from

Inflation Targeting in Emerging

The performance of inflation-targeting regimes around the world has

We thank the participants in the NBER Eighteenth Annual Conference on Macroeconomi

This content downloaded from

Under inflation targeting, EMEs have the challenge of breaking the

Figure 1 INFLATION BEFORE AND AFTER ADOPTION OF INFLATION

13.11 Emerging Market

, Developed Economies 5.95

This content downloaded from

December 2002). In this context, what is the optimal response to these

2. Stylized Facts About Inflation Targeting in Emerging

In this section we present basic stylized facts comparing the volatilities of

This content downloaded from

Table 1 INITIAL TARGETS AND INFLATION AROUND ADOPTION OF

1. In Brazil, the inflation of

This content downloaded from

Table 2 VOLATILITY AND AVERAGE OF SELECTED VARIABLES FOR

Volatility of basic variables Average

Emerging market economies

Data source: International Financial Statistics, IMF

emerging and developed economies

This content downloaded from

Figure 2 VOLATILITIES OF SELECTED VARIABLES OF THE PERIOD

Inflation GDP Growth 5.47

" 4 3.37 3.40

different as well. Then we consider a second

This content downloaded from

Figure 3 TRADE-OFF VOLATILITIES OF OUTPUT AND INFLATION (1997:1-2002:2)

specification where imports are entered as intermediate goods captures

This content downloaded from

comprised of households, firms (owned by the households), and govern-

yt = a, + can, + (1 - ca)m (1)

where yt is output, a, represents a s

Ct = Et ct + 1-- (it - Et +1) + +c,t (3)

Table 3 IMPORTS CLASSIFIED BY USE (2001)

This content downloaded from

defined as PMt = s, + p*,t (p*Mt is the price of imports in

This content downloaded from

inflation.9 As in Gall and Gertler (1999), one possibility is to consider that

TC,= f Et1,nt+ I - yVtc " gb t (7)

where wy is related negatively to the aversion to inflation var

9. Fuhrer and Moore (1995) have generated inflation persistence assumi

This content downloaded from

interest rate is restricted to react to selected variables. We use the algo-

4. Explaining the Higher Volatility

4.1 BUILDING CREDIBILITY AND REDUCING INFLATION RATE LEVELS

Institutions in emerging economies tend to be weaker than thos

13. The explanation of the calibration of the model is available on request to th

This content downloaded from

Figure 4 12-MONTH INFLATION RIGHT BEFORE IT ADOPTION

* Developed Economies a Emerging Market Economies

Imperfect credibility concerning the fulfillment of the targets becomes

This content downloaded from

Figure 5 FIRST CENTRAL TARGET ADOPTED

. Developed Economies a Emerging M

defined as PMt = s, + p,t (pMt is the price of imports in

ShA = (tadm -n) adm - (IA + CaA) (9)