
 Start

4018
 Prefix

policy strategy for maintaining inflation at a relatively low and stable level
without sacrificing longterm growth.1Nonetheless, it is still an open question whether countries
that have adopted inflationtargeting regimes have lower inflation and better economic performance
than countries that follow other monetary frameworks, see for example the research on macroeconomic performance in
 Exact

Ball (2011), Ball and Sheridan (2005),
 Suffix

Gon ̧calves and Salles (2002), and Brito
and Bystedt (2010). Others have taken a different approach by looking for evidence on the extent
to which inflation expectations are well anchored using survey and financial market data.
 (check this in PDF content)

 Start

4091
 Prefix

stable level
without sacrificing longterm growth.1Nonetheless, it is still an open question whether countries
that have adopted inflationtargeting regimes have lower inflation and better economic performance
than countries that follow other monetary frameworks, see for example the research on macroeconomic performance in Ball (2011), Ball and Sheridan (2005),Gon ̧calves and Salles (2002), and
 Exact

Brito and Bystedt (2010).
 Suffix

Others have taken a different approach by looking for evidence on the extent
to which inflation expectations are well anchored using survey and financial market data. Because
of data limitations, however, most of the latter work has focused on the experience of industrialized
countries.
 (check this in PDF content)

 Start

6330
 Prefix

However, there are also differences among the three with respect to institutional settings and in
how their central banks explain to the public how they will strive to achieve the inflation goal.
Chile, for example, had already achieved considerable success in macroeconomic stabilization in
1According to
 Exact

Hammond (2012),
 Suffix

27 countries are considered tohave inflationtargeting frameworks: Armenia,
Australia, Brazil, Canada, Chile, Colombia, the Czech Republic, Ghana, Guatemala, Hungary, Iceland, Indonesia,
Israel, Korea, Mexico, New Zealand, Norway, Peru, the Philippines, Poland, Romania, Serbia, South Africa, Sweden,
Thailand, Turkey, and the United Kingdom.
 (check this in PDF content)

 Start

7186
 Prefix

Several years after the Bank of Mexico (BOM) adopted its inflationtargeting framework, it had
continued to formally target a money aggregate and, unlike most other inflationtargeting central
banks, did not publish its inflation forecasts, see
 Exact

Batini and Laxton (2006).
 Suffix

2
Our approach is a blend of a formal and informal analysis. In our formal analysis, we follow
the approach that was first used by G ̈urkaynak, Levin, Marder, and Swanson (2007a) by examining
evidence from financialmarketderived measures of longterm inflation expectations.
 (check this in PDF content)

 Start

15553
 Prefix

By
1998, the BOM’s monetary policy announcements could be seenas signaling the direction in which
the central bank wanted interest rates to move (RamosFrancia and TorresGarc ́ıa, 2005). In 1999,
BOM officials wrote that Mexico’s monetary policy framework was ”in a transition period towards a
clearcut inflation targeting scheme.”
 Exact

(Carstens and Werner, 1999;
 Suffix

cited in Mishkin and Savastano,
2001). The BOM formally adopted its inflationtargeting framework in 2001 and announced that
the inflation target would be 3 percent beginning in 2003.
Reflecting a growing consensus that central banks need to be free from political pressures
to pursue shortterm objectives, the central banks of Chileand Mexico had been granted legal
autonomy with price stability as
 (check this in PDF content)

 Start

15590
 Prefix

By
1998, the BOM’s monetary policy announcements could be seenas signaling the direction in which
the central bank wanted interest rates to move (RamosFrancia and TorresGarc ́ıa, 2005). In 1999,
BOM officials wrote that Mexico’s monetary policy framework was ”in a transition period towards a
clearcut inflation targeting scheme.” (Carstens and Werner, 1999; cited in
 Exact

Mishkin and Savastano, 2001).
 Suffix

The BOM formally adopted its inflationtargeting framework in 2001 and announced that
the inflation target would be 3 percent beginning in 2003.
Reflecting a growing consensus that central banks need to be free from political pressures
to pursue shortterm objectives, the central banks of Chileand Mexico had been granted legal
autonomy with price stability as their primary mandate, Chile in 1990 a
 (check this in PDF content)

 Start

18200
 Prefix

At that moment, weimagined that such level would be, in
a first step, something close to 3 to 4 percent (inspired by theChilean experience) and that, with
time, we would go to a rate close to the world average”
 Exact

(Fraga,2009,
 Suffix

the translation is ours).
After the Lula government took office, the inflation target wasset at 412percent in mid2003
and the target has remained at that level since then. However, in 2004, CBB President Meirelles
stated that he envisioned inflation falling to a longterm inflation target of 4 percent (Gomes,
2004).
 (check this in PDF content)

 Start

18514
 Prefix

After the Lula government took office, the inflation target wasset at 412percent in mid2003
and the target has remained at that level since then. However, in 2004, CBB President Meirelles
stated that he envisioned inflation falling to a longterm inflation target of 4 percent
 Exact

(Gomes, 2004).
 Suffix

In mid2007, in announcing the target for 2009, Finance Minister Guido Mantega stated
that ”the inflation targets for 2008 and 2009 should be seen asa transition in the direction of a
longterm inflation target that I judge appropriate to be in the neighborhood of 4 percent, given
the characteristics of the Brazilian economy” (Goldfajn, 2007, the translation is ours).
 (check this in PDF content)

 Start

18853
 Prefix

In mid2007, in announcing the target for 2009, Finance Minister Guido Mantega stated
that ”the inflation targets for 2008 and 2009 should be seen asa transition in the direction of a
longterm inflation target that I judge appropriate to be in the neighborhood of 4 percent, given
the characteristics of the Brazilian economy”
 Exact

(Goldfajn, 2007,
 Suffix

the translation is ours).
Dilma Rousseff, Lula’s protege and successor, took office in January 2011, and appointed Alexandre Tombini as the new central bank president. In October 2012, Tombini stated that ”[w]e have
to have the ambition of having inflation converge to [inflation] of our trading partners, as this, in
the medium and longterm, would make a difference.
 (check this in PDF content)

 Start

19349
 Prefix

In October 2012, Tombini stated that ”[w]e have
to have the ambition of having inflation converge to [inflation] of our trading partners, as this, in
the medium and longterm, would make a difference. Nonetheless, at the moment, we have to consolidate this level [referring to the 412percent inflation target].”
 Exact

(Grinbaum, 2012,
 Suffix

the translation is
ours.). As we detail below, there is some evidence that uncertainty about the longerterm inflation
goal has been feeding into survey and financial marketbasedreadings on the longerterm inflation
outlook for Brazil.
 (check this in PDF content)

 Start

28603
 Prefix

The nowoutstanding spectrum of both nominal and real sovereign bonds allows us to construct
nominal and real zerocoupon curves from these bonds, respectively. The zero curve estimation
method we apply is that of
 Exact

Nelson and Siegel (1987)
 Suffix

which has increasingly become the workhorse
method for estimating zero curves from bond prices.11
A zerocoupon yield curve consists of the collection of interest rates earned on noncouponpaying bonds with increasing maturities.
 (check this in PDF content)

 Start

29023
 Prefix

which has increasingly become the workhorse
method for estimating zero curves from bond prices.11
A zerocoupon yield curve consists of the collection of interest rates earned on noncouponpaying bonds with increasing maturities. Because zerocoupon yields are not directly observable but
are instead embedded in couponbearing bonds, we must resort to curve estimation techniques. Here
we use the
 Exact

Nelson and Siegel (1987)
 Suffix

model. This model postulates that the curve of continuouslycompounded zerocoupon yields at any given timetcan be well described by a smooth parametric
10In contrast, some developed economies, for example Germanyand Japan, while having extremely liquid nominal
bond markets, still have much less developed inflationlinked bond markets, with only a small number of bonds
outstanding at any given time
 (check this in PDF content)

 Start

29501
 Prefix

zerocoupon yields at any given timetcan be well described by a smooth parametric
10In contrast, some developed economies, for example Germanyand Japan, while having extremely liquid nominal
bond markets, still have much less developed inflationlinked bond markets, with only a small number of bonds
outstanding at any given time.
11For example, the Bank of International Settlements,
 Exact

(BIS, 2005),
 Suffix

reports that nine out of the thirteen (predominantly European) central banks that report their zerocoupon curve estimates to the BIS use either the Nelson and
Siegel (1987) model or an extension of it, the Svensson (1994) model, to construct zerocoupon yield curves.
8
function which is determined by just four parameters;
yt(τ) =β1,t+β2,t
1−exp
−τλt
<
+β3,t
1−exp
−τλt
<
<−exp
>
−
τ
λt
 (check this in PDF content)

 Start

29664
 Prefix

, while having extremely liquid nominal
bond markets, still have much less developed inflationlinked bond markets, with only a small number of bonds
outstanding at any given time.
11For example, the Bank of International Settlements, (BIS, 2005), reports that nine out of the thirteen (predominantly European) central banks that report their zerocoupon curve estimates to the BIS use either the
 Exact

Nelson and Siegel (1987)
 Suffix

model or an extension of it, the Svensson (1994) model, to construct zerocoupon yield curves.
8
function which is determined by just four parameters;
yt(τ) =β1,t+β2,t
1−exp
−τλt
<
+β3,t
1−exp
−τλt
<
<−exp
>
−
τ
λt
(1)
<
τ
λt
τ
λt
whereyt(τ) is the modelimpliedτperiod zerocoupon yield and{β1,t, β2,tβ3,t, λt}is the parameter
vector.
 (check this in PDF content)

 Start

29721
 Prefix

have much less developed inflationlinked bond markets, with only a small number of bonds
outstanding at any given time.
11For example, the Bank of International Settlements, (BIS, 2005), reports that nine out of the thirteen (predominantly European) central banks that report their zerocoupon curve estimates to the BIS use either the Nelson and Siegel (1987) model or an extension of it, the
 Exact

Svensson (1994)
 Suffix

model, to construct zerocoupon yield curves.
8
function which is determined by just four parameters;
yt(τ) =β1,t+β2,t
1−exp
−τλt
<
+β3,t
1−exp
−τλt
<
<−exp
>
−
τ
λt
(1)
<
τ
λt
τ
λt
whereyt(τ) is the modelimpliedτperiod zerocoupon yield and{β1,t, β2,tβ3,t, λt}is the parameter
vector.
 (check this in PDF content)

 Start

30264
 Prefix

These parameters can be interpreted as the level parameter,β1,t; the slope parameter,
β2,t; and the curvature parameter,β3,t, judging from the effect that a change in each of these
respective parameters has on the shape of the curve, see for example
 Exact

Diebold and Li (2006).
 Suffix

The
fourth parameter,λt, is a shape parameter that influences the factor loadings associated with the
slope and curvature parameters. We follow the approach of G ̈urkaynak, Sack, and Wright (2007b,
2010b) to estimate nominal and real zerocoupon curves fromobserved bond prices.
 (check this in PDF content)

 Start

35343
 Prefix

leave analyzing the effects of macroeconomic news surprises on the full term structure of forward inflation
compensation, such as is done in Beecheyet al.(2011), for future research.
14For Morgan Markets, seehttps://mm.jpmorgan.com/. For PiP, seehttps://www.precios.com.mx/.
15G ̈urkaynak, Sack, and Wright (2007b) show that for estimating zerocoupon curves from U.S. Treasury bonds,
one needs the
 Exact

Svensson (1994)
 Suffix

model to accurately fit bond prices in the very longest end of the curve. However, the
Svensson model requires estimating additional parameterscompared with the Nelson and Siegel model. Therefore,
due to the relatively small number of bond prices that we haveavailable for any given day in our sample, we only
consider maturities of up to fifteen years.
 (check this in PDF content)

 Start

52221
 Prefix

For Brazil we found outliers in the policy rate (the release of July 24, 2008), CPI (December 7, 2012),
IP (March 6, 2009), and GDP (March 10, 2009). For Chile these were in the policy rate (December
11, 2003), CPI (January 6, 2009), trade balance (September 7, 2007), and the unemployment
25See
 Exact

Cook and Weisberg (1982)
 Suffix

for details and a general discussion on outlier detection.
15
rate (November 27, 2003). Finally, for Mexico, these were inCPI (May 7, 2010), PMI (August
3, 2011), retail sales (May 26, 2003), and GDP (August 16, 2005).
 (check this in PDF content)