Post on 18-Jan-2019
Active or Passive Monetary Policy? Inflation Targeting in Brazil During the Financial Crisis of 2002
Patrice Robitaille
Patrice.robitaille@frb.gov
Board of Governors of the Federal Reserve System August 1, 2003
(Preliminary—Do not Quote) Abstract: In this paper, we explore two alternative interpretations of the Brazilian central bank’s behavior during the financial crisis of 2002. Under the first view, Brazil was in a money dominant regime, the central bank followed a Taylor Rule, and was surprised by the spurt of inflation that appeared in late 2002. Under the second view, the central bank operated under a fiscal-dominant regime, the central bank followed a passive Taylor Rule, and allowed inflation to rise. Our tentative conclusion is that the second view seems more consistent with the evidence from this crisis episode. Somewhat paradoxically, we postulate that, by following a passive Taylor Rule and even by avoiding default on the domestic debt, the central bank avoided an inflationary spiral.
Staff economist. I am grateful for the research assistance provided by Katherine Kelly, and benefited conversations with Jonathan Wright. All errors are my own. The views expressed in this paper reflect my views only and do not necessarily reflect views of the Board of Governors of the Federal Reserve System or its official staff.
2
I. Introduction The fiscal theory of the price level emphasizes that monetary authorities can lose
control of the price level under certain circumstances (Woodford 1996, Canzoneri and
Diba 1998, Canzoneri, Cumby, and Diba 2001). If the primary fiscal surplus (inclusive of
transfers from the central bank) adjusts to changes in the level of the debt to ensure fiscal
solvency, then monetary authorities retain control of the price level. A government
exhibiting this policy response is characterized as being in a money dominant (Ricardian)
regime. If the primary surplus does not or is not expected to adjust, however, then the
price level has to rise to make the present discounted value of future fiscal surplus equal
to the existing stock of public debt. Because a monetary authority looses control of the
price level under these circumstances, it is said to be operating in a fiscal dominant (non-
Ricardian) regime.
There seems to be no strong consensus on whether fiscal dominant or monetary
dominant regimes predominate. To some observers, it has seemed reasonable to
postulate that countries with a history of high inflation will have a more difficult time
controlling the price level. Sims (2003), who muses that perhaps too much emphasis has
been placed on the merits of inflation targeting in countries that have not addressed
institutional weaknesses, asks whether Brazil, which joined the inflation targeting camp
in 1999, will ultimately succeed in controlling the price level. This view is not shared by
Edwards (2002), who writes that ‘it is clear that in Brazil there is no “fiscal dominance.”’
The fact that Edwards expressed this view in mid-2002 is all the more remarkable given
that a financial crisis was in full-swing in Brazil at that time.1 The crisis, which
1 Edwards seems to base his view on the notion that most of the public debt was indexed. However, a far lower share of the debt was indexed than apparent at first glance, and the implications of this partial indexation are discussed below.
3
continued for several months into late 2002 and early 2003, was marked by serious debt
funding difficulties, sizable exchange rate depreciation, and a rise in inflation. In this
paper, we explore to what extent the fiscal theory of the price level can shed insight into
the Brazilian experience of 2002.
One interpretation of the events of 2002 follows from a reading the Brazilian
monetary policy committee meeting minutes and quarterly Inflation Report, vehicles used
by the Brazilian central bank to explain its actions. An admittedly simplistic
interpretation of the events of 2002, but which highlights the main points, would be the
following. Brazil was and continues to be in a money dominant regime. However, even
in a money dominant regime, a central bank cannot fully control the price level in the
face of adverse supply shocks. Most notably, the exchange rate depreciated substantially
between April and September, as seen by the plot of the real/dollar exchange rate in
Chart 1. The central bank, which follows an active monetary policy under a Taylor Rule,
was surprised by the rise in inflation in late 2002.
A second interpretation of developments last year, however, would be the
following: events associated with the electoral cycle raised prospects that future fiscal
policies would loosen—amounting to a fiscal policy shock. This situation interacted with
balance sheet risks and other institutional weaknesses, undermining the central bank’s
ability to control the price level. Because a substantial portion of the public debt was
linked to foreign currencies, the exchange rate depreciation led to a substantial rise in the
level of debt. With little possibility that the present discounted value of future fiscal
surpluses would rise sufficiently to equal the substantially greater level of debt, the price
level had to rise. Default risk also became an issue, evidenced by the rise in the Embi+
4
spread over U.S. Treasuries (Chart 2), but as a matter of social choice, authorities would
have ruled this option out. (This last argument is developed below.)
Now consider the reaction of the central bank. It may very well have
intentionally adopted a passive stance in the face of the inflation threat that was induced
by fiscal policy shocks. With a more passive stance, the price level would rise, but by
less than it would have had the central bank responded to the inflation threat
aggressively. This is basically an application of the “monetary paradox” model
developed by Loya (1999) to shed light into the Brazilian inflationary experience of the
1980s, but this time, going in reverse. That is, in 2002, the central bank could have
become more passive than otherwise the case. A more passive monetary policy stance
avoided the inflationary spiral seen in Brazil during the 1980s.
II. Overview of Brazil’s Inflation Targeting Experience
Brazil, with a history of high inflation through the mid-1990s, adopted inflation
targeting in mid-1999 in the wake of the collapse of its fixed exchange rate regime. It
was widely recognized that the inconsistency between fiscal policy and the fixed
exchange rate contributed to the collapse of the peg. In the wake of the real’s collapse,
however, the government of Henrique Cardoso made notable progress reforming
institutions and rules (i.e., the Fiscal Responsibility Law) that were aimed at promoting
fiscal discipline. In March 1999, an IMF program was signed, committing the
government to achieving sizeable primary fiscal surpluses. Brazil not only honored its
primary fiscal targets but also often surpassed them in subsequent years.
Under the leadership of Central Bank President Arimio Fraga, who took over in
5
early March 1999, an inflation targeting regime was launched the following June. The
top panel of Chart 3 plots the inflation targets and twelve-month inflation (the heavy line)
as measured by the headline CPI, called the IPCA. In June 1999, the midpoint targets
were announced for 1999, 2000, and 2001 and were set at 8, 6, and 4 percent. In mid-
2000 and mid-2001, inflation targets were set for 2002 and 2003 at 3.5 percent and 3.25
percent. Although there was an inflation target range of +/- 2 percentage points around
the midpoint, targeted inflation was on a downward trend as long as authorities did not
stray from the midpoint too far and too often. However, in late June 2002 and in mid-
January 2003, the upper bound of the inflation target range for 2003 was effectively
raised It is also clear from the graph that the inflation targets were met in 1999 and 2000,
but that twelve-month inflation has exceeded the targets since then.
III. A Counterfactural Experiment Using an Interest Rate Rule
Shown in the top panel of Chart 4 are the overnight interest rate that is targeted by
the central bank, the Selic, and twelve month inflation. The fact that that the Selic has
stayed well above twelve-month inflation does not mean that monetary policy has been
very tight all the time over the post-1999 period, as lagged twelve-month inflation is not a
good measure of inflation expectations in a country like Brazil, with a history of
persistently high inflation.
As a first pass at assessing the central bank’s behavior in 2002, we use the interest
rate rule estimated by Goldfajn et al (2002) to run a counterfactural experiment. Golfajn
et al use a foreign looking version of a Taylor Rule (Taylor 1993). It is a variant of the
forward-looking Taylor rule popularized by Clarida, Gertler, and Gali (1998 ) that relates
the desired interest rate to a model-based measure of expected inflation at the 12- month
6
horizon, and the output gap. Goldfajn et al (2002) instead use as an explanatory variable
the inflation forecasts of the central bank.2 It is this approach that we consider here as a
possible characterization of the behavior of the central bank.
Goldfajn et al estimated their rule over the 1999Q6 to 2002Q6 period, which is
one or two months more data in the sample than we would like for our counterfactual
experiment, but suitable enough to illustrate some issues at stake.
To deal with the fact that the inflation target is set on an annual basis, Goldajn et
al assume that policy makers’ twelve-month-ahead inflation target is a linear combination
of the current calendar year’s and the next calendar year’s inflation targets using moving
weights. The weight on the current year inflation target falls with the number of months
remaining in the current year and the weight on the next year inflation target rises with
the number of months in the next year that are included in the inflation target horizon.
The resulting estimation, which takes into account slow adjustment in the interest
rate, generates the following equation for the target interest rate for the period 1999Q6 to
2002Q4:
ygapi CBtt *47.0)(*84.157.17 *
12/* −−+= + ππ
(0.48) (1.19) (0.16)
Standard errors are in parentheses. Although the coefficient on the inflation term is
greater than one, one cannot reject the hypothesis that the coefficient zero, and the sign
on the output gap is the opposite of what is expected (a positive gap means output is
above capacity). We consider this equation as a starting point, but drop the output gap
2 Central bank forecasts are made public quarterly, so they interpolate the quarterly
7
term.
We use the estimated rule to fit an equation for the desired interest rate and
compare the fitted rate with actual interest rate over the 1999Q6 to 2002Q12 period. An
interest rate that is below the fitted value would suggest a less aggressive policy stance
over the period than was the case on average over the first three years of inflation
targeting regime. Chart 5 compares the actual interest rate with two fitted interest rates.
The higher path for the hypothetical interest rate assumes that the central bank targets the
midpoint of the inflation target range, while the lower hypothetical target path assumes
that the central bank targets the upper bound. According to then-Central Bank President
Arminio Fraga (Fraga 2000), the central bank focused on meeting the midpoint target as
of March 2000, whereas the limits of the inflation target range were there to guide
expectations in the face of supply shocks.
From mid-1999 to mid-2000, Selic rate was well above the hypothetical interest
rate. As this was the very early stage of the inflation target regime, this outcome might
have been expected for a central bank in the very early phases of building up its anti-
inflation credibility. Werlang (2002) identifies three “inflation scares” over the mid-1999
to May 2002 period, occurring in July 1999, October-November 1999, and March-August
2001 that were precipitated by supply shocks.3 The chart offers an explanation for why
the Selic was not raised further in 1999; the interest rate was already well above its
desired level, given where the central bank’s inflation forecast was relative to the
inflation target.
projections to arrive at monthly estimates. 3 The 1999 shocks were precipitated by jumps in the prices of regulated services (“administered prices”), food, and in a sales tax, while the main shock in 2001 was the rationing induced by domestic electrical energy shortage.
8
The Selic rate was well above its desired level from April 2001 to late 2002
assuming that the central bank had continued to target the midpoint. It is reasonable,
however, to postulate that around April 2001, it would have switched to the higher
inflation target. A switch at about April 2001 seems reasonable because April-May 2001
marked the early phases of a domestic electrical energy crisis (inducing rationing) that
lasted into early 2002.
The actual interest rate is above the fitted interest rate from about April 2001
through the third quarter of 2002. Over this period, the targeted interest rate lies below
the 17.57 percent that is the estimated long run target for the nominal interest rate. This
is consistent with the notion that the central bank believed that inflation was under
control; projected inflation lies below the inflation target. The Selic rate was in the 18-
18½ percent range from early 2002 until the evening of Oct 14, when the central bank
raised the Selic 300 basis points in an extraordinary meeting. In the fourth quarter, the
fittted target rate rises as much as the actual rate, the chart seems to well characterize the
behavior of the central bank over late 2002.
However, the Central Bank’s forecast is based on a consensus among the
members of the monetary policy committee, and therefore, judgment plays a role in
devising the forecast. This element of judgment is apparent in the June 2002 Inflation
Report, where the central bank takes a particular view about the assumed degree of
exchange rate pass-through and persistence of the exchange rate depreciation underlying
the inflation forecast:
an assumption was made that the economic agents do not use this actual exchange rate as a parameter for defining prices, but rather the rate that they think is most likely to be permanent. In the projections presented here, it was assumed that the exchange rate that is relevant for the formation of market prices is based on the average level observed
9
in the second quarter of 2002 [R$2.5/$], which gradually converges toward the actual exchange rate in the first quarter of 2003 [R$2.7/$]. (p. 115; italics added for emphasis).
These special assumptions seem to beg the question of what other possible scenarios the
central bank had devised, and in particular, how concerned they were that they would
lose control of the price level.
Chart 5 provides an informal assessment of the central bank’s forecasts by
displaying the 12-month inflation rate against the 12-month-ahead inflation projection of
the central bank, moved ahead 12 months. Therefore, by construction, the chart shows
the central bank’s inflation forecast error (realized inflation minus the projected rate of
inflation). For comparison, the average 12-month-ahead inflation projection from the
central bank’s survey of private sector forecasters is plotted the same way. (The private
sector’s inflation forecast begins January 2000 and thus there are no observations before
January 2001.) A notable feature of the graph is the positive forecast error that rises over
time in both the central bank’s and in the private sector’s inflation forecasts.
Systematically positive or negative inflation forecast errors are not what one would
expect if the inflation forecast were rationale ones, as the forecast error should be white
noise.
To investigate this question a little more formally, a simple tests of forecast bias
was conducted by regressing the forecast error against its mean and testing whether mean
was significantly different from zero. Given the small number of mostly overlapping
observations, one would be more likely to accept the null hypothesis of forecast
rationality. The hypothesis of forecast rationality is rejected at the 12-month horizon for
both the central bank and private sector foecasts (standard errors are autocorrelation-
10
robust). The mean forecast error for the central bank was about 3½ percent (for the
period 2000:2 to 2003:1) and 4 ¼ percent for the private sector (for the period January
2001 to April 2003). Although the number of observations is relatively small, the
forecast error is large enough to enable us to reject the hypothesis that the error has mean
zero.
IV. An Interpretation of the Crisis Based on the Fiscal Theory of the Price Level
Despite the inroads that Brazil’s government made into developing mechanisms
for committing current and future governments to following conservative fiscal policies,
most of these reforms (although not all) had been enacted under a single president who
had been in power since 1995. Over the course of 2002, the rise of a left wing candidate,
Lula, of the Workers’ Party (PT), during the presidential election campaign (elections
took place in October) raised concerns that future fiscal policies would turn expansionary
and worse, that the government might default on its debt. Public sector net debt relative
to GDP had grown markedly in recent years, from about 30 percent of GDP in 1995 to
over 50 percent of GDP at the end of 2001. That ratio continued to climb over much of
2002 as a result of the depreciation of the real and high real interests rates, reaching over
60 percent by October.
Foreign currency denominated debt and dollar-indexed domestic debt together
accounted for roughly 40 percent of the federal government’s debt as of May 2002, net of
the central bank’s international reserves. The exchange rate depreciation that took place
over the course of mid- to late- 2002 meant that an even higher stream of future fiscal
surpluses would be needed for the government’s intertemporal budget constraint to hold,
given the new (higher) level of debt and holding prices constant.
11
The fiscal responses to these developments proved insufficient to prevent the
situation from deteriorating further over most of 2002. In early 2002, the government’s
2003 target for the primary surplus was raised from 3¼ percent of GDP to 3½ percent of
GDP, and in June, the target was raised to 3¾ percent of GDP. An agreement on a new
IMF program was rumored to be imminent in late July, and was announced on August 7,
which contributed to a boost in asset prices, but that boost proved short-lived. Under an
IMF program, the government committed to maintain the primary surplus through 2003,
and the government pledged to raise the primary fiscal surplus further to 4 percent of
GDP through the end of 2002. There was always the possibility that the next government
would press to renegotiate the agreement to support a more expansionary fiscal policy.
The IMF program was designed to preclude this event by being heavily back loaded;
roughly $6 billion would be disbursed under the program over the rest of the year, and
another roughly $25 billion would be disbursed in 2003. But this was not enough to lead
to a marketed improvement in the financial conditions facing Brazil over the course of
late 2002.
Loyo (1999) uses a fiscal theory of the price level that highlight the dilemma
faced by monetary authorities when faced with a fiscal authority that moves
independently to set the primary fiscal surplus, so that dominance holds, in an
environment where the government must make sizeable interest payments on the
domestic nominal debt. A curious result is that a hard line monetary authority could
generate an inflationary spiral when raising the interest rate in response to inflationary
pressures sufficiently to raise the real interest rate. The problem is that a rise in real
interest rates, in a non-Ricardian (fiscally dominant) setting, generates a positive wealth
12
effect in the private sector; the public views the rise in the real interest rates as real
transfer from the government. But that positive wealth effect violates the equilibrium
condition that the private sector’s intertemporal budget constraint at the existing price
level be consistent with the level of available resources. Therefore, the price level must
rise. Loyo shows that the price level must not only rise but rise along an explosive path.
That model seems to fit well the Brazilian experience during the 1980s.
Brazil has not seen the inflation rates witnessed in the 1980s for several years.
Nevertheless, in 2001, although the primary surplus surpassed 3½ percent of GDP in
2001, the overall deficit was about 3½ percent of GDP. Thus, the interest bill was about
7 percent of GDP. The dollar-indexed and foreign currency-denominated debt could not
be inflated away (the dollar-indexed debt not easily), but a large portion of the domestic
debt in Brazil was not linked to the price level, leaving room for the central bank to
reduce the real value of this debt.
Aside from the dollar-linked debt, the largest form of debt was the floating rate
debt, the LFTs which were linked to the overnight policy instrument of the central bank.
LFTs accounted for about 40 percent of the government’s gross debt net of international
reserves as of May 2002, and had by far the longest maturities (often maturing in 2004-
2006). Interest accrues on the LFTs and is not paid out until they mature, making the
interest and principal both subject to default risk. Furthermore, LFTs are closer to
nominal bonds rather than indexed bonds. To say that interest rates rise on with expected
inflation assumes both a strong form of the Fischer hypothesis (that the interest rate on
the bonds rises with expected inflation) and that the policy rate, the Selic, is constrained
by market rates. Neither assumption necessarily holds.
13
Therefore, a desire to avoid a return to the inflationary spiral seen in the 1980s
may have been a concern for the central bank, leading it to raise interest rates less in 2002
than otherwise would have been the case. (A subsequent version of this paper will
develop this argument further.)
This interpretation of the events of 2002 seems consistent with the path that the
government’s net debt-to-GDP ratio, shown on a quarterly basis in Chart 3, took over the
course of the year. In the chart, the heavy line depicts the net debt-to-GDP ratio of the
general government (federal, state, and local governments) that is produced by the central
bank. This picture is likely to remain the same if we include only the federal
government. The striking feature of the chart is that after rising considerably over the
course of 2002, the debt-to-GDP ratio declines over the fourth quarter of 2002. Some of
this decline reflects a factor reducing the numerator: a partial recovery in the value of the
real over the fourth quarter of 2002 reduced the stock of debt valued in reais. Much of
the decline also came because of a rise in the denominator, that is, nominal GDP rose
considerably. It is the case that the estimate used by the central bank may have
overestimated the rise in the GDP deflator and thereby exaggerated the decline in the
debt-to-GDP ratio, a possibility we are currently investigating. The chart also shows two
alternative measures of debt-to-GDP, both of which show a decline.
V. The Role of Banks and Mutual Funds in Determining Policy Choices
This section briefly considers how the fact that banks and mutual fund investors
held roughly two-thirds of the domestic debt may have affected the government’s
inflation versus default incentives.
The mutual fund industry saw rapid growth over the 1990s that was driven not
14
only by regulatory changes that encouraged the development of the industry, but also by
tax policy. Notably, growth in the mutual fund industry over the late 1990s came at the
expense of a decline in the share of funds captured by bank savings accounts (cadernetas
de poupanca), which had traditionally been the investment vehicle of choice for small
savers.
Investors tended to favor mutual funds over bank deposits because mutual funds
investors were exempted from the financial transactions tax (the CPMF), which charged
as much as 0.38 percent to roll over a maturing deposit, which was typically of one
month’s maturity. Since all of the major banks were affiliated with mutual funds, which
earned sizeable fees, this market development met no opposition from the banking sector.
It is not clear to what extent investors gave up any transactions services by moving from
savings accounts to mutual funds, since, as stated above, mutual funds were affiliated
with banks, often closely so, with banks providing both mutual fund and banking services
under the same roof.
A notable feature of mutual funds is that roughly 80 percent of the investments for
the industry as a whole were in the form of government debt. Furthermore, most of the
securities held by mutual funds was in the form of the Selic-linked LFTs, which were
relatively long-term.
In well-functioning financial system, mutual funds amount to a desirable risk
sharing arrangement from a social standpoint. The nominal claim can theoretically go up
or down. If the nominal value of the claim falls in bad times for the issuer, well-
diversified investors absorb the losses. Losses suffered by mutual fund investors would
not generate a social demand for the government to reinstate the nominal claim that
15
would generate additional government liabilities that would have to be financed
somehow. There would also be no social demand to guarantee future nominal claims, as
is in the case in many instances with fixed claims issued by banks (e.g., government-
insured bank deposits).
In Brazil, however, the migration of so many small investors into mutual funds
created a different situation altogether. Many investors believed that their investments
were virtually risk free. In late May 2002, mutual fund portfolio managers engaged in
what seems to be modest markdowns in the value of mutual fund portfolios following
declines in the value of LFTs, but in some cases they were not so small. The main point
is that these losses they were large given investors’ expectations that their funds were
completely risk free. The resulting losses, furthermore, resurrected memories of the
government’s relatively recent record of visiting losses on holders of its debt via various
means (special taxes in the 1980s and an outright confiscation in 1990 for example).
Over the June-August period, about R$50 billion, net outflows from mutual funds
amounted to about 20 percent of the net asset value of the funds as of the end of May
(controlling for price effects). The monthly outflows (purged of price effects) on a
monthly basis are shown by the solid bars in Chart 9.
Over the course of late May through August, the central bank responded to these
developments by acquiring a substantial portion of the mutual funds’ relatively long-term
LFTs and exchanged them for relatively short-term LFTs. Some observers, moreover,
speculated that these operations created an implicit put option on future issues of
relatively long-term LFTs. It does not appear that the central bank acted as an agent of
the Treasury, but that rather that the central bank, acted in coordination with the Brazilian
16
Treasury and even led the effort. That is, there seems to have been no disagreement
between the central bank and the Treasury that it was a matter of public policy that small
savers, even mutual fund investors, should be protected from losses, especially if this
objective was consistent with the desire to support a demand for government debt.
The striped bars of the chart show the change in bank deposits (inclusive of
accrued interest). Therefore, about 60-70 percent of the outflows from the mutual fund
industry returned to the banking sector as deposits. In response, on August 14, the central
bank raised reserve requirements on bank demand, time, and savings deposits. Reserve
requirements were increased from 45 to 48 percent on demand deposits, from 20 to 25
percent for savings deposits, and from 15 to 18 percent for time deposits. Reserve
requirements were raised again on October 11.
To simplify things, let us assume for the moment that the government pays a
market interest rate on its reserves, and consider how who held the debt affected
incentives to default and inflate the real value of the debt. We have argued above that the
government was adverse to visiting small losses on mutual fund holders. Now consider
bank. Chart X displays two measures of banks’ exposure to the government. Exposure
to the government includes holdings of bonds and credit to the government, but does not
appear to include banks’ reserves at the central bank. Either way, banks’ exposure to the
government had risen substantially in recent years. By mid-2002, banks’ exposure to the
government was well above their equity capital and was high as a portion of their assets.
We assert here that since the mid-1990s, public policies in banking had essentially
protected creditors of relatively large banks against losses (this argument is being laid out
in other work in progress). A government default on its domestic debt would have
17
generated implicit government liabilities at a time when there was no credible means of
committing to running a path of substantial primary fiscal surpluses. The high
distributive and economic costs of default (e.g., disruptions in the payments mechanisms)
therefore seemed to have tilted incentives away from default.
VI. Conclusion (TBA)
18
References
Canzoneri, M., and Diba, B., 1998. Fiscal Constrains on Central Bank Independence and
Price Stability, in Monetary Policy and Inflation in Spain, Malo de Molina, Viñals, and Gutiérrez eds. New York, N.Y. : St. Martin's Press.
Canzoneri, M. B., Cumby, R. E., Diba, B.T., 2001. Is the Price Level Determined by the
Needs of Fiscal Solvency? American Economic Review, v. 91, iss. 5, pp. 1221-38. Clarida, R., J. Gali, and M. Gertler, 1998. Monetary Policy Rules in Practice: Some
International Evidence, European Economic Review, 42, pp. 1033-1067. Edwards, S., 2002. Commentary: Should the European Central Bank and the Federal
Reserve Be Concerned about Fiscal Policy? Fraga, A., 2000. Monetary Policy During the Transition to a Floating Exchange Rate:
Brazil’s Recent Experience, Finance and Development, Vol. 37, no. 1, March. Loyo, E. 1999. Tight Money Paradox on the Loose: A Fiscalist Hyperinflation, mimeo,
Kennedy School of Government, June. Goldfajn, I., Kfoury, M., Minella, A, Springer de Freitas, P., 2002. Inflation Targeting in
Brazil: Lessons and Challenges, Central Bank working paper no. 53. Sims, C. 2003. Limits to Inflation Targeting, mimeo, Princeton University. Taylor, J.B. 1993. Discretion versus Policy Rules in Practice, Carnegie-Rochester
Conference Series on Public Policy, 39, pp. 195-214. Werlang, S., 2002. Handout from “Three Years of Inflation Targets,” (in Portuguese)
from Central Bank Seminar on Three Years of Inflation Targets, May 17, 2002 (available on the central bank’s website (www.bacen.gov.br).
Woodford, M., 1996. Control of the Public Debt: A Requirement for Price Stability?
NBER Working Paper 5684.
19
Chart 1
Brazilian real -Dollar Exchange Rate (reais per dollar)
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
1-Ja
n-01
23-J
an-0
1
14-F
eb-0
1
8-M
ar-0
1
30-M
ar-0
1
23-A
pr-0
1
15-M
ay-0
1
6-Ju
n-01
28-J
un-0
1
20-J
ul-0
1
13-A
ug-0
1
4-S
ep-0
1
26-S
ep-0
1
18-O
ct-0
1
9-N
ov-0
1
3-D
ec-0
1
25-D
ec-0
1
16-J
an-0
2
7-Fe
b-02
1-M
ar-0
2
25-M
ar-0
2
16-A
pr-0
2
8-M
ay-0
2
31-M
ay-0
2
24-J
un-0
2
16-J
ul-0
2
7-A
ug-0
2
29-A
ug-0
2
20-S
ep-0
2
14-O
ct-0
2
5-N
ov-0
2
27-N
ov-0
2
19-D
ec-0
2
10-J
an-0
3
3-Fe
b-03
25-F
eb-0
3
19-M
ar-0
3
10-A
pr-0
3
20
Chart2
Brazil EMBI+ spread over US Treasuries
0
500
1000
1500
2000
2500
3000
Jan-99
Mar-99
May-99
Jul-99
Sep-99
Nov-99
Jan-00
Mar-00
May-00
Jul-00
Sep-00
Nov-00
Jan-01
Mar-01
May-01
Jul-01
Sep-01
Nov-01
Jan-02
Mar-02
May-02
Jul-02
Sep-02
Nov-02
Jan-03
Mar-03
basi
s po
ints
21
Chart 3
Brazilian 12-Month Inflation and Inflation Targets
0
2
4
6
8
10
12
14
16
18
Jan-98
Jul-98
Jan-99
Jul-99
Jan-00
Jul-00
Jan-01
Jul-01
Jan-02
Jul-02
Jan-03
Jul-03
Jan-04
Jul-04
Date
Perc
ent
12-month change in IPCA
Upper bound as of January 2003
Lower bound as of January 2003
Upper bound as of June 2002
22
Chart 4
Selic and Twelve-Month Inflation
0
5
10
15
20
25
30
35
40
45
50
Jan-97
Apr-97
Jul-97
Oct-97
Jan-98
Apr-98
Jul-98
Oct-98
Jan-99
Apr-99
Jul-99
Oct-99
Jan-00
Apr-00
Jul-00
Oct-00
Jan-01
Apr-01
Jul-01
Oct-01
Jan-02
Apr-02
Jul-02
Oct-02
Jan-03
Apr-03
Perc
ent
Selic12-month change in IPCA
Exchange rate peg abandonedInflation Targeting Introduced
Selic raised 300 bp 10/14/02; second round presidential election 10/27/02
New government takes office 1/03
23
Chart 5
Hypothetical Interest Rates vs Actual
10.00
12.00
14.00
16.00
18.00
20.00
22.00
24.00
26.00
28.00
Jun-99
Aug-99
Oct-99
Dec-99
Feb-00
Apr-00
Jun-00
Aug-00
Oct-00
Dec-00
Feb-01
Apr-01
Jun-01
Aug-01
Oct-01
Dec-01
Feb-02
Apr-02
Jun-02
Aug-02
Oct-02
Dec-02
Feb-03
10.00
15.00
20.00
25.00
30.00
35.00
i* if UB targetedselic"i* if mid-pt targeted"
24
Chart 6
Twelve-month-ahead inflation expectations (in month of realized 12-month inflation) vs realized 12-month inflation
0
2
4
6
8
10
12
14
16
18
20
Jun-00
Aug-00
Oct-00
Dec-00
Feb-01
Apr-01
Jun-01
Aug-01
Oct-01
Dec-01
Feb-02
Apr-02
Jun-02
Aug-02
Oct-02
Dec-02
Feb-03
Apr-03
Jun-03
Aug-03
Oct-03
Dec-03
Feb-04
Apr-04
Central Bank ForecastActualPrivate Sector Forecast
September 2002
25
Chart 7
Net Debt/GDP
30
40
50
60
70
1998:4 1999:4 2000:4 2001:4 2002:4
Perc
ent
Net Debt/GDP BACEN DefinitionAlternative Measure 1 of Net Debt/GDPAlternative Measure 2 of Net Debt/GDP
Notes: Debt is the central bank’s definition of net debt of the public sector. Net debt to GDP (solid line) was pulled from Haver database, and is the debt-to-GDP ratio that is published by the Brazilian central bank. Debt is deflated by the central bank’s nominal GDP series. Alternative measure 1 of debt-to-GDP = net debt in the last month of a given quarter, deflated by the nominal quarterly GDP series computed by the IBGE for the same quarter. Alternative measure 2 of debt-to-GDP = average net debt for a given quarter, deflated by the nominal quarterly GDP series computed by the IBGE for the same quarter. The IBGE is the data collection agency that compiles national income accounts. The decline in the GDP ratios would have been accentuated if instead the stock of debt outstanding at the end of the fourth quarter had been divided by the GDP deflator for the first quarter of 2003. That approach would more closely correspond to theory that the rise in the price level in period t reduces the outstanding stock of nominal debt.
26
Chart 8
Mutual Fund Flows and Change in Deposits
-30
-20
-10
0
10
20
30
Oct-01
Nov-01
Dec-01
Jan-02
Feb-02
Mar-02
Apr-02
May-02
Jun-02
Jul-02
Aug-02
Sep-02
Oct-02
Nov-02
Dec-02
Jan-03
Feb-03
Mar-03
Billions of reais
Mutual Fund Flows Change in stock of savings and time deposits
27
Net redemptions of Brazilian Public Debt
-10
0
10
20
30
40
50
60
Novem
ber 2001
Dec 2001
Jan 2002
Feb 2002
March 2002
April 2002
May 2002
June 2002
July 2002
August
Sept
Oct
Nov 2002
Dec 2002
jan 2003
Feb 2003
Mar-03
In b
illio
ns o
f rea
is
Net Redemptions*Purchases prior to maturity
Chart 9
Depository banks' exposure to the Brazilian Federal Government 1/
0
5
10
15
20
25
30
35
40
DEZ/1995
MAR
/1996JU
N/1996
SET/1996
DEZ/1996
MAR
/1997JU
N/1997
SET/1997
DEZ/1997
MAR
/1998JU
N/1998
SET/1998
DEZ/1998
MAR
/1999JU
N/1999
SET/1999
DEZ/1999
MAR
/2000JU
N/2000
SET/2000
DEZ/2000
MAR
/2001JU
N/2001
SET/2001
DEZ/2001
MAR
/2002JU
N/2002
SET/2002
DEZ/2002
1/ Credit to government = Series 1875 (titles of BC)+series 1877 (credit to federal govt). Capital = 1893 (conta de capital) . Assets = 1873 (activos totais).
0
20
40
60
80
100
120
140
160
180
exposure /assetsexposure/conta de capital
28