Monetary policy with a volatile exchange rate: the case of Brazil since 1999.
Barbosa-Filho, Nelson H.
INTRODUCTION
Brazil adopted inflation targeting and a floating exchange rate in
1999. This regime followed a brief period of fixed exchange rates, in
1994-98, when the Brazilian government used a currency peg to reduce and
stabilize inflation. The so-called 'Real Plan' was successful
in controlling inflation, but it also increased the financial fragility
of the economy through high domestic real interest rates, a rising ratio
of public debt to GDP and large current account deficits. When the
global situation changed, in the wake of the East Asian and Russian
crises of 1997-98, the Brazilian government abandoned the currency peg
and moved to another macroeconomic policy regime, which persists to
date.
The endurance of the current macroeconomic framework does not mean
that the Brazilian economy has experienced smooth sailing since 1999.
Quite to the contrary, a series of exogenous shocks have hit the economy
in the past 15 years, coming from the world economy and the climate,
which had a substantial impact on Brazil's exchange rate and power
prices, respectively. (1) Moreover, Brazil also experienced significant
domestic and government-induced structural changes, especially the
expansion of the government's social safety net, which resulted in
lower poverty, lower income inequality, credit expansion and financial
deepening.
Brazilian monetary policy managed to keep inflation under control
and maintain financial stability even in the face of the exogenous
shocks and structural changes mentioned above. More specifically, in 9
of the 15 years since the adoption of inflation targeting, the Brazilian
Central Bank (BCB) has managed to keep inflation within the bounds
defined by the government. Even though inflation stayed above the
government's central target during most of this period, it did not
get out of control either. The average annual rate of consumer inflation
was 7.1% in 1999-13, with a maximum of 12.5% in 2002 and a minimum of
3.1% in 2006. (2) The current central target is 4.5%, with an interval
of tolerance of plus or minus 2 percentage points, but the BCB has been
having difficulty meeting the target. (3)
The recent difficulty in controlling inflation comes mostly from
the depreciation of the Brazilian real (BRL). In fact, the Brazilian
average real exchange rate rose 20% in 2012-13, which in turn created a
temporary inflationary and recessive shock. (4) In fact, despite
Brazil's move to floating exchange rates in 1999, Brazilian prices
remain very sensitive to the exchange rate. Whenever there is a sharp
depreciation of the BRL, the BCB has difficulty meeting the inflation
target. Whenever the opposite happens, the target is met with relative
ease.
This paper analyzes Brazilian monetary policy since 1999 and its
relationship with inflation and economic growth. It will be argued that,
despite its limitations and problems, the current 'macroeconomic
tripod' composed of floating exchange rates, fiscal rules and
inflation targeting continues to be the best framework to manage the
Brazilian economy. Borrowing Churchill's famous line, the current
tripod of Brazilian macroeconomic policy is the worst form of
macroeconomic policy, except for all those other forms that have been
tried from time to time in Brazil. On a more technical level, it will be
argued that inflation targeting emerged as the best macroeconomic policy
in Brazil because it stabilizes both the real exchange rate and GDP
growth, although not necessarily at an optimal level.
The paper has six sections. The section 'The macroeconomic
"tripod" describes the institutional structure of Brazilian
macroeconomic policy. The section 'Macroeconomic policy since the
late 1990s' gives a brief macroeconomic history of Brazil since
1999. The section 'Monetary policy and exchange-rate policy'
presents the evolution of the main instruments of monetary and
exchange-rate policy since 1999. The section 'Exchange rate and
inflation in the short run' estimates the short-run impact of the
exchange rate on inflation and the section 'Exchange rate,
inflation and growth in the long run' does the same for the long
run, including GDP growth in the analysis. The section 'Monetary
policy and structural change' concludes with some comments on the
role of monetary demand management and structural reforms in
Brazil's economic policy.
THE MACROECONOMIC 'TRIPOD'
Four government ministries manage Brazil's macroeconomic
policy through three main committees: one for financial issues, one for
monetary issues and the third for fiscal issues. The four Ministries are
the Ministry of Finance (MoF), the already mentioned BCB, the Ministry
of Planning, Budget and Management (MPOG for short) and the Civil
Executive Office of the President (CCPR for short). (5) The three
committees and their respective members are as follows:
(i) The National Monetary Council (Conselho Monetario National or
CMN), formed by the MoF, the BCB and the MPOG.
(ii) The Monetary Policy Committee (Comite de Politico Monetaria or
Copom), formed by the President and the board of directors of the BCB,
which has eight members in total.
(iii) The Executive Budget Committee [Junta de Execuqao
Orqamentaria or JEO), formed by the MoF, the MPOG and the CCPR.
I will briefly describe the origins and functions of each
committee, starting with the CMN. This committee was created together
with the BCB, in the late 1960s. It went through many changes until it
reached its current three-member structure.
Its basic responsibilities are to set Brazilian financial
regulation, supervise the BCB and other government financial agencies,
and manage the volume, interest rates and financial guidelines for all
the government's credit and insurance programs, especially for
agriculture, housing, long-term finance and regional development.
More importantly for the macroeconomic focus of this paper, the CMN
determines the inflation target with an 18-month lead, that is, in June
of year 't' it sets the inflation target to be met in December
of year 't + 1'. The CMN can also revise the inflation target
at any time depending on the economic conditions, although revisions do
not occur frequently.
After the CMN sets the inflation target, it is up to the BCB to
meet it. The Presidential Decree that instituted inflation targeting in
Brazil does not make any special mention of the country's base
interest rate, the SELIC rate--it just states that 'the BCB should
execute all policies necessary to meet the targets set'. (6) Thus,
even though the SELIC rate is the BCB's main instrument of action,
Brazilian monetary policy also includes quantitative and prudential
controls of liquidity and credit to influence aggregate demand.
If the inflation target is not met, the President of the BCB has to
write an open letter to the Minister of Finance, explaining why this
happened and what actions will be taken to correct the problem. This
'penalty', together with the CMN's ability to revise the
inflation target, means that the Brazilian inflation-targeting regime is
a very flexible system to coordinate market expectations.
In complement to the CMN's regulatory role, the Brazilian
government can also change its tax on financial operations (IOF) through
a Presidential decree without Congress approval. (7) In practice, the
IOF functions as a form of credit and exchange-rate regulation, since
the government can alter the rate of return on many financial and
foreign exchange (FX) operations simply by altering the IOF rate.
Proposals to change the IOF rates fall under the jurisdiction of the
MoF, but they must be signed by the President.
Second, the Copom is a BCB committee created in 1996, as a late
part of the Real Plan. Even though Brazil still had a currency peg at
that time, the basic idea behind the Copom was to institute a formal
decision process for setting the base interest rate. The Copom gained
importance after the introduction of inflation targeting when the SELIC
rate became the main tool of monetary policy.
Currently, the Copom has eight meetings during a year. The minutes
of its decisions are published 1 week after each meeting, and each
member's decisions are disclosed. In addition, the BCB publishes a
quarterly inflation report, in which it analyzes the evolution of the
economy, makes forecasts for the near future and signals the guidelines
of monetary policy. The Brazilian President nominates the members of the
Copom, but they have to be confirmed by the Senate. Despite this, the
members of the Copom do not have a fixed mandate and can be fired
without any consultation with the Senate.
Finally, the JEO is an informal fiscal policy committee that
manages fiscal issues with more transparency and accountability than in
the past. (8) The JEO's basic task is to propose the
government's fiscal targets and budget to the President, who in
turn has to send these to Congress for approval. After Congress modifies
and approves the federal Budget, the JEO can manage its discretionary
part to achieve the fiscal target proposed for each year. In practice,
the JEO can cut, postpone or cancel discretionary expenditures approved
by Congress if it finds it necessary to meet the fiscal target. (9)
Since the end of the currency peg, Brazil's fiscal target has
consisted of a minimum value for the government's primary surplus,
in an attempt to control the country's net public debt as a
percentage of GDP. If the primary surplus falls below the
government's target, the JEO can slash spending and/or raise
revenues to correct it. However, since most public spending is
non-discretionary and tax laws cannot change much without Congress
approval, the JEO's powers are very limited. To deal with this
rigidity, the fiscal target also contains some escape clauses, as
special and optional 'deductions' due to investment in fixed
capital and tax cuts, which the JEO can use in times of high fiscal
volatility.
In practice, similar to what happens with monetary policy, the JEO
decisions are a formal but loose way to coordinate market expectations
about fiscal policy. The JEO meetings usually occur every other month,
starting in January, and they are followed by a joint fiscal report by
the MoF and MPOG, in which the two Ministries set the government fiscal
program for the remainder of the year. Since JEO decisions manage only a
small part of public spending, the effective guidelines for fiscal
policy are set on an annual basis, with a 1-year lead, by the President
and Congress.
Finally, on exchange-rate issues, there are no formal government
committees, targets or guidelines in Brazil because the country has had
a floating exchange rate since 1999. Despite this, the Brazilian
government does intervene in FX markets, but with no commitment to a
specific value for the exchange rate or a formal target for its
international reserves. This loose government approach to the topic
results in a dirty float of the BRL in which both the BCB and the MoF
try to reduce exchange-rate volatility through a combination of
taxation, regulation and operations in the spot and derivatives markets
for FX. The MoF does all of the taxation, the MF and the BCB share the
regulation, and the BCB does all of the government's FX
transactions in the spot and derivatives markets.
MACROECONOMIC POLICY SINCE THE LATE 1990s
Although monetary policy is usually the main instrument of demand
management in modern economies, it is just one part of macroeconomic
policy. Even where the central bank is independent, its actions
influence and are influenced by the other elements of economic policy.
It is therefore impossible to analyze monetary policy in isolation. In
the specific case of Brazil, the actions of the BCB have been strongly
influenced by exchange-rate issues, fiscal policy, and the government
regulation of credit and some key prices.
To facilitate our analysis, the recent Brazilian macroeconomic
policy can be divided into six 3-year periods. (10) Table 1 presents the
main macroeconomic indicators for these periods.
First, in 1997-99, the Brazilian authorities tried to defend their
exchange-rate peg with a very restrictive macroeconomic policy and
emergency loans from abroad, mainly from the US Treasury and the IMF. As
we already mentioned, this attempt was short-lived and the Brazilian
government decided to move to the current system of inflation targeting,
floating exchange rates and fiscal targets in 1999. The real exchange
rate moved up substantially at the end of this period, at which time GDP
growth slowed. Considering the 3 years as a whole, the average growth of
the economy was 1.2%. (11) Despite the sharp depreciation of the BRL in
1999, the average inflation rate was 5.2%.
Second, in 2000-02, the recovery of the Brazilian economy was
slowed down by three major adverse shocks: (i) the Argentine crisis of
2001, (ii) drought and power rationing (12) and (iii) a massive
speculative attack against the BRL during the Brazilian presidential
election campaign in 2002. All these factors resulted in another
substantial depreciation of the BRL, which in turn increased inflation
and did not allow a sustainable recovery of GDP growth. (13) Over this
period, the average GDP growth and inflation rates were 2.8% and 8.7%,
respectively. At the end of this period, the Brazilian government once
more had to resort to liquidity assistance from the IMF while the
economy adjusted to its new and high real exchange rate.
Third, in 2003-05, the Brazilian government adopted a restrictive
macroeconomic policy and benefited from the high real exchange rate
inherited from the previous period. The initially high real exchange
rate turned the economy's current account from deficit to surplus,
and helped the government to reduce the country's foreign financial
fragility. Then, as the real exchange returned to a lower level,
inflation also went down. In these 3 years, the average annual growth
rate of GDP was 3.3%, and annual inflation 7.5%. At the end of this
period, the Brazilian government paid all its debts to the IMF and
regained control over its macroeconomic policy.
Fourth, in 2006-08, the Brazilian economy benefited from the rapid
growth in the world demand for commodities pulled by China. The increase
in Brazil's terms of trade raised the economy's disposable
income and allowed growth acceleration with further disinflation and a
balanced expansion of the government revenues and expenditures. The
average GDP growth and inflation rates were 5.1% and 4.5%, respectively,
and during this exceptionally good period the Brazilian government
raised its income transfers to the poor, as well as the economy's
minimum wage, at a fast pace. Private consumption accelerated, but
investment accelerated even more in a virtuous cycle of wage expansion
and productivity growth. All these forces pushed the real exchange rate
down, and dragged the economy's current-account balance with it.
There were also massive capital inflows to Brazil during this period,
which allowed the BCB to accumulate a substantial amount of foreign
reserves.
Fifth, in 2009-11, the Brazilian economy was hit by the global
financial crisis. The initial impacts of the shock were a depreciation
of the BRL and a credit crunch, which led to a decline in GDP in 2009.
In the face of this, and given the lower financial fragility of the
economy, the Brazilian government adopted an expansionist policy to
bring the economy back to its pre-crisis level. The main initiatives
were fiscal and monetary expansion, mostly through the maintenance of
the government income transfers to the poor, increased public
investment, fiscal and financial incentives to private investment, and a
reduction in the SELIC rate. In parallel to all this, the
'quantitative easing' by advanced economies and the recovery
in commodity prices quickly resulted in another round of appreciation of
the BRL. Brazil's current account went further into deficit,
capital flows once more poured in, inflation accelerated and
macroeconomic policy turned very restrictive only at the end of this
period. Considering the 3 years together, the average GDP growth and
inflation rates were 3.3% and 5.6%, respectively.
Finally, in 2012-14, the effects of Brazil's restrictive
macroeconomic policy of 2011 were amplified by the slowdown of the world
economy--coming from the US, China and the Eurozone--and both forces
reduced Brazil's GDP growth substantially. (14) The slow growth was
also caused by domestic problems, since the economy faced another
drought and cost-push inflation--mainly in nonprocessed food items--as
well as a fall in private investment due to changes in government
regulation, and delays in government investments. Despite the
deceleration in GDP, inflation did not fall in 2012-14 because the new
international situation moved the Brazilian terms of trade down and its
real exchange rate up. In addition to this, demographic trends and the
government's social policies kept the rate of unemployment down
despite the reduction in economic activity, which in turn kept the
inflation in non-tradable goods and services at a high level. During
these 3 years, average GDP growth fell to 1.2%, and average inflation
rose to 6.0%. (15) In the face of slow growth and high inflation, the
government tried to manage the macroeconomy through price controls and
an expansionary fiscal, credit and monetary policy, which in turn
reduced its primary surplus and increased its debt-to-GDP ratio. (16)
[FIGURE 1 OMITTED]
MONETARY POLICY AND EXCHANGE-RATE POLICY
Based on the chronology in the previous section, we can now focus
on the main tools of monetary policy in Brazil: the SELIC rate and the
banks' compulsory deposits at the BCB. In order to have a complete
view of all BCB instruments, we will also include its operations in FX
markets in our monetary analysis, as well as the MoF's decisions on
FX taxation. Starting with the SELIC rate, Figure 1 shows the evolution
of its nominal and real values, with the latter calculated 'ex
post', that is, based on observed inflation. The data clearly shows
five monetary cycles of tightening and loosening monetary policy since
1999, measured by the troughs of the nominal SELIC rate, around a
downward trend. (17)
Focusing our attention on the real interest rate, we can identify
three main phases of Brazilian monetary policy in Figure 1. (18) First,
the end of the currency peg and the beginning of inflation targeting, in
1997-99, when the BCB had to raise the SELIC rate substantially to
control inflation and stabilize the country's real exchange rate
after the initial depreciation of the BRL. During the most critical
phase of the transition, at the end of 1998 and beginning of 1999, the
real SELIC rate reached 26%.
[FIGURE 2 OMITTED]
Second, the real SELIC rate fluctuated around 10% during the
consolidation of inflation targeting from mid-2000 to mid-2007. There
were two monetary cycles during this phase: one triggered by the
depreciation of the BRL, in 2002, and the other by the acceleration of
GDP, in 2004. The real SELIC rate reached a minimum of 4%, at the
beginning of 2003, and a maximum of 14%, at the beginning of 2004.
Third, the real SELIC rate fluctuated around 5% from mid-2007 to
mid-2012 when there was a combination of inflation targeting and low
financial fragility. The main difference between this phase and the
previous one was the massive accumulation of foreign reserves by the
BCB, which in turn reduced the economy's vulnerability to foreign
shocks. As shown in Figure 2, the BCB's foreign reserves jumped
from 5% to 16% of GDP between December 2005 and December 2008, and have
remained at a high level since then.
In addition to the accumulation of foreign reserves, the reduction
in the real SELIC rate in recent years happened together with a more
intensive use of quantitative tools of liquidity management by the BCB.
To illustrate this, Figure 3 presents the banks' compulsory
deposits at the BCB as a percentage of GDP. The data show that the BCB
reduced the banks' compulsory deposits to inject liquidity into the
economy immediately after the financial crisis of 2008. Then, as
monetary policy had to turn more restrictive to fight inflation, the BCB
raised the banks' compulsory deposits in 2010, and again in 2011.
[FIGURE 3 OMITTED]
The level of the banks' compulsory deposits at the BCB started
to fall only in 2012, but they are still much higher than their
pre-crisis level. Despite this 'quantitative easing', the real
SELIC rate fell to almost zero in mid-2013, which could indicate another
change in Brazilian monetary policy, but it is too early to know whether
this recent reduction in the real SELIC rate is a cyclical or a
structural change.
Finally, to complete our analysis of the BCB main macroeconomic
actions, Figure 4 presents its net 'long' position in FX
derivatives contracts. In this kind of contract the buyer of FX receives
the variation in the BRL/USD exchange rate, plus an interest rate in USD
(the 'cupom cambial' rate), in exchange for paying the SELIC
rate. The swap contract therefore mimics an arbitrage between the
domestic rates in BRL and USD, that is, the expected depreciation of the
BRL implicit in this kind of swap is the difference between the SELIC
rate and the domestic interest rate in USD.
Figure 4 shows that the BC has used FX swaps to influence the
exchange rate three times since the beginning of inflation targeting.
First, in mid-2008, the BCB bought USD through swaps to fight the
appreciation of BRL, but it quickly changed course after the financial
crash later that year. In fact, given the sudden liquidity constraint of
the Brazilian financial system in FX, the BCB changed its net long
position in USD from plus 1.4% of GDP to minus 0.9% of GDP between
September and December of that year. The total net change in the supply
of FX through derivatives contracts was therefore 2.3% of GDP in just 3
months, and this helped Brazilian financial firms to deal with the
depreciation of the BRL at the end of 2008.
[FIGURE 4 OMITTED]
Second, in 2011, the BCB once more used swaps to buy FX and fight
the appreciation of the BRL. The BCB's net long position in FX
swaps reached 0.5% of GDP in mid-2011, and this was complemented by
purchases in the spot market to increase the BCB's foreign
reserves. In fact, as shown in Figure 2, the BCB's foreign reserve
rose to a record-high level of 18% of GDP at the end of 2012. During
this period, the MoF also participated in the effort to fight the
excessive appreciation of the BRL by taxing any increased net long
position of private agents in BRL through derivatives contracts. The
taxation involved a new law to expand the IOF's reach to
derivatives contracts and caused great turmoil in the high-frequency
financial market. (19)
The third and final intervention of the BCB in the derivatives
market of FX started in mid-2013, when the perspective of monetary
tightening in the US pushed the currencies of many emerging markets
down, especially those of commodity-producer countries such as Brazil.
Similar to what it had already done after the 2008 crash, the BCB
increased the supply of FX by selling swaps in the derivatives market in
2013. However, unlike what it did in 2008, this time the BCB sold a very
high amount of FX through swaps. The BCB action was quickly followed by
the MoF, which cut all its previous taxation on capital inflows and
derivatives operations. (20)
In terms of GDP, the BCB's net long position in FX swaps
reached a record-low level of minus 4% of GDP in mid-2014, which
corresponds to approximately 22% of Brazil's foreign reserves at
that time. The magnitude of this intervention cannot be justified by
liquidity concerns alone. In addition to providing liquidity in FX, the
recent BCB swaps operations were also a clear attempt to avoid a sharp
depreciation of the BRL because of the impact of the exchange rate on
inflation.
EXCHANGE RATE AND INFLATION IN THE SHORT RUN
The BCB's most recent attempt to fight a sharp depreciation of
BRL comes from the impact of the exchange rate on consumer prices. I
estimate the impact with a vector auto-regressive (VAR) model of the
nominal exchange rate and consumer prices using monthly data since 1998.
(21) Results indicate that
(i) we cannot reject the null hypothesis that the nominal BRL/USD
'Granger' causes changes in the IPCA index at a 0.5% level of
statistical significance, but the same does not hold in the opposite
direction even at the 10% level of statistical significance; and
(ii) the long-run pass-through of the exchange rate to consumer
prices is 14% and most of the adjustment is completed in 24 months.
In economic terms, the statistical results mean that changes in the
nominal exchange rate precede and cause temporary changes in Brazilian
inflation, and that a permanent 10% depreciation of the BRL raises the
IPCA index by 1.4%. (22) Most of the price adjustment happens in the
first 24 months after the exchange-rate shock, that is, as soon as the
exchange rate stabilizes at a new level, the inflationary shock dies out
in 2 years.
The combination of a moderate exchange-rate pass-through to
consumer prices with a very volatile exchange rate means that the
variance of Brazilian inflation depends mostly on the variance of
Brazil's real exchange rate. In other words, when the real exchange
rate fluctuates 20% up and down, as has often been the case for Brazil
since 1999, the estimate of the long-run pass-through implies that real
exchange-rate volatility causes the consumer price index to move
approximately 3 percentage points up and down around its trend. (23)
Since the Brazilian inflation-targeting regime adopts an interval of
tolerance of 2 percentage points, the volatility of the real exchange
rate explains most of the volatility of Brazilian inflation since the
beginning of inflation targeting.
To illustrate the point above, Table 2 presents the evolution of
the Brazilian effective and target inflation rates since 1999, together
with the variation in the economy's average nominal and real
exchange rate. The data reveal that:
(i) since the beginning of inflation targeting, the BCB has met the
target in 11 out of 15 years (73% success rate);
(ii) if we consider the real exchange rate, the BRL appreciated in
8 of the 11 years in which the BCB met the inflation target (72% of the
success cases); and
(iii) if we consider the nominal exchange rate, the BRL appreciated
in 6 of the 11 years in which the BCB met the inflation target (55% of
the success cases).
The importance of the exchange rate for inflation targeting is even
greater than it appears from Table 2 if we consider that 1999, 2012 and
2013 were all very special cases. More specifically, because 1999 was
the first year of inflation targeting, the CMN set its target in June of
the same year, when it already knew 50% of the annual inflation. Then,
in 2012 and 2013, the BCB met the inflation target despite the
depreciation of the BRL with a great deal of help from the
government's control of some key regulated prices--especially fuel
and power. (24)
The main conclusion is that despite the end of the currency peg in
1999, inflation targeting still depends on exchange-rate targeting in
Brazil. The BCB does not have a formal commitment with a specific value
of the exchange rate, but it must nevertheless stabilize the
country's real exchange rate in order to keep inflation within the
interval of tolerance defined by the government. Given the estimate of
14% exchange-rate pass-through to consumer prices, the observed
exchange-rate volatility results in a very volatile inflation rate in
Brazil. That is, inflation volatility exceeds the interval of tolerance
around the government's central target for inflation.
EXCHANGE RATE, INFLATION AND GROWTH IN THE LONG RUN
In addition to the volatility of the real exchange rate, the level
at which the real exchange rate stabilizes is also important for
macroeconomic policy because the real exchange rate has a permanent
effect on both inflation and economic growth. More formally, our
estimates based on data from 1998 through 2014 imply that there is a
long-run relationship between the real exchange rate and inflation, on
one side, and between the real exchange rate and GDP growth, on the
other.
Consider first the inflation-real exchange rate relationship. In
addition to the 14% pass-through of the nominal exchange rate to
consumer prices mentioned earlier, there is also econometric evidence
that the level of the real exchange rate has a positive impact on the
long-run rate of inflation in Brazil. From a theoretical perspective,
the micro-foundations of this result come from a markup model of prices,
in which firms adjust their EBITDA to obtain a 'desired' or
'normal' rate of profit. (25) In this process, the level of
the real exchange rate influences the relative price of inputs and
capital, as well as the degree of indexation in the economy. After we
exclude demand shocks from the analysis, these hypotheses result in a
long-run equation in which inflation is equal to the markets'
expectation about the government's inflation target, plus a
function of the real exchange rate and other variables.
To represent this relationship, I use a univariate error correction
model of inflation as a function of the level of economic activity, the
change in the real exchange rate and the level of the real exchange
rate. (26) The appendix presents the estimated coefficients of this
model for 1998-2014. (27) For our economic discussion, the most
important result is that, after we exclude short-term shocks and demand
factors, the Brazilian inflation seems to be a third-degree polynomial
of the level of the real exchange in the long run. Figure 5 shows the
long-run 'inflation curve' that comes out of the econometric
model. The intuitive meaning of such a curve is that too much
depreciation (a high real exchange rate) pushes inflation up and too
much appreciation (a low real exchange rate) pushes it down in Brazil.
More importantly, for intermediate values of the real exchange rate,
inflation remains practically constant in Brazil.
[FIGURE 5 OMITTED]
Next, consider economic growth. The impact of the real exchange
rate on economic development is the topic of a long debate among
development and Latin American economists. (28) The literature is long
and the results are varied. To make a long history short, the main sides
of the debate can be summarized with two opposing hypotheses: (i) a high
real exchange rate is either good for growth, because it raises the
competitiveness of the tradable high-productivity sector of the economy,
or (ii) bad for growth, because it raises the relative price of capital.
If there is a nonlinear relationship between growth and the real
exchange rate, the validity of these two hypotheses depends on the
initial value of the exchange rate.
My estimates with the Brazilian data from 1998 through 2014
indicate that there is a long-run relationship between the level of the
real exchange rate and economic growth. (29) This relationship defines
economic growth as a second-degree polynomial--a
'concave-down' function--of the real exchange rate. In
economic terms this result means that both too much currency
appreciation and too much currency depreciation are bad for growth, as
shown in Figure 6.
[FIGURE 6 OMITTED]
To understand the theoretical meaning of the 'growth
curve' in Figure 6, assume that economic growth depends on the pace
of capital accumulation in the long run (an AK model), and that
investment depends on both the relative price of capital and the
expected rate of profit on it. (30) Since the real exchange rate has
opposite effects on the relative price of capital and the expected rate
of profit in an economy such as Brazil, economic growth becomes the
non-linear 'concave-down' function of the real exchange rate
shown in Figure 6. (31)
Now, if we combine the two nonlinear curves of Figures 5 and 6, the
result is a long-run structure in which inflation targeting stabilizes
both inflation and GDP growth. In other words, the floor and the ceiling
of the inflation target define a lower and an upper bound for the real
exchange rate in Figure 5. Then, in Figure 6, the upper and lower bounds
of the real exchange rate define the range of GDP growth rates
consistent with the inflation target. Assuming for the moment that both
the inflation and growth curves are fixed, the logical conclusion from
Figures 5 and 6 is that inflation targeting ends up stabilizing both the
real exchange rate and economic growth in Brazil.
The other natural conclusion from Figures 5 and 6 is that inflation
targeting may stabilize inflation and economic growth at a level below
the economy's growth potential. In fact, the growth rate consistent
with any given central inflation target is not necessarily the maximum
growth rate depicted in Figure 6. If the inflation target is either too
low or too high for the economy's structure, the GDP growth
consistent with it can actually be much lower than the economy's
maximum growth rate.
In theory, if the inflation and growth curves of the Brazilian
economy really do not move, the Brazilian government could pick an
inflation target that maximizes GDP growth. In practice, things do not
work out this way because a high inflation target increases the
indexation of the economy and pushes inflation up. In other words, the
inflation and growth curves move through time and one thing that can
move them is precisely a high inflation target. (32) Because of this
limitation, monetary policy cannot and should not be the instrument to
put the economy at its maximum growth potential. Demand management
should be complemented by microeconomic or institutional actions that
improve the economy's structure and raise its potential growth rate
for any inflation and real exchange rates.
MONETARY POLICY AND STRUCTURAL CHANGE
From the previous analysis, we can conclude that Brazilian monetary
policy managed to control inflation in the past 15 years, but not
necessarily in a way that is consistent with the maximum growth
potential of the economy. In theory, because of the nonlinear
relationships linking the real exchange rate, inflation and growth in
Brazil, the best possible action for the Brazilian government is to pick
an inflation targeting that is consistent with the maximum growth rate
of the economy. In practice, this theoretical solution may not be
feasible if the inflation target consistent with the maximum GDP growth
rate is too high.
A high rate of inflation tends to be unstable in Brazil because it
quickly increases the indexation of the economy--the inflationary
inertia measured by the autoregressive coefficient of inflation--which
in turn makes inflation rigid downwards. In fact, based on the
econometric model presented in the previous section, the real exchange
rate consistent with the maximum GDP growth rate currently results in an
annual inflation of 6%. Since the central target of inflation in Brazil
is 4.5%, the current structure of the Brazilian economy forces the BCB
to make the economy converge to a lower real exchange rate, which in
turn stabilizes GDP growth below its maximum value.
From the above we can conclude that the success of inflation
targeting in Brazil is mixed. On the one hand, the BCB has been able to
keep inflation under control. On the other, monetary policy has not been
able to do this and maximize the growth potential of the economy. The
solution for this dilemma is not to adjust the inflation target to the
economy's structure, but rather to adjust the economy's
structure to the adequate inflation target. High inflation is usually
unstable, and therefore the Brazilian economic policy should improve the
economy's structure in a way that makes it possible to achieve
maximum growth with a low and stable inflation. Monetary policy is one
part of economic policy. It certainly has structural and permanent
impacts on the economy, but it is also conditioned by the other
components of economic policy. (33)
More specifically, monetary policy operates to stabilize inflation
according to the inflation target defined by the government, based on
the growth and inflation curves of the economy. The positions of these
curves are not constant, since they depend on structural issues such as
tax policy, labor regulation and trade openness. By definition,
structural change is slow. Changes in the economy's coefficients of
production are slower than high-frequency adjustments in monetary policy
and exchange rates, but they nevertheless alter the trade-offs faced by
demand management. The best possible combination for Brazilian
macroeconomic policy is therefore to combine the short-run demand
management of inflation targeting and floating exchange rates with
long-run structural changes that move the trade-offs between real
exchange rates, inflation and growth in a favorable way.
In terms of the long-run curves presented in the previous section,
a positive structural change moves the inflation curve down, meaning
that it reduces the inflation rate consistent with any given value of
the real exchange rate. As for GDP, a positive structural change also
moves the growth curve to the left, meaning that it reduces the real
exchange rate consistent with the maximum growth rate of the economy.
These two changes are nothing more than the long-run effect of a
permanent increase in labor productivity, which can be achieved by many
policy instruments, not only by the real exchange rate. (34)
Putting all these pieces together, we can conclude that inflation
targeting has been the 'worst' form of macroeconomic policy
for Brazil, since it has not been able to maximize GDP growth, except
for all other forms of macroeconomic policy that have been tried before,
which failed at stabilizing inflation. Thus, instead of dropping
inflation targeting to maximize growth, the best alternative for Brazil
is to keep inflation targeting and try to maximize growth through other
instruments than short-run demand management. In such a 'division
of labor' of economic policy, the BCB should continue to pursue a
sensible inflation target, which depends on the given structure of the
economy, while the rest of economic policy adopts supply-side or
institutional initiatives that change the structure of the economy in a
way that promotes fast growth with stable and low inflation.
Acknowledgements
I would like to thank Roberto Frenkel, Martin Rapetti, Mario
Damill, Jose Antonio Ocampo and three anonymous referees for their
comments and suggestions on a previous version of this paper. The errors
and opinions remain mine only.
APPENDIX
Econometric appendix
The data series used in the models are:
* IPCA: consumer-price index, monthly frequency.
* GDPSA: real gross domestic product index, seasonally adjusted,
quarterly frequency.
* NER: nominal BRL/USD exchange rate, monthly frequency.
* RER: real exchange rate, as defined by the BCB, monthly
frequency.
* GAP: output gap, measured by the difference between GDP and its
Hodrick--Prescott trend, both series in natural logs, with a smoothing
parameter of 1600.
* DUMMY_POWER: dummy variable for the energy rationing of 2001,
equal to 1 in the third and fourth quarters of 2001.
* DUMMY_CRASH: dummy variable for the financial crash of 2008-09,
equal tol in the fourth quarter of 2008 and first quarter of 2009.
* DUMMY_WCUP: dummy variable for the World Cup of 2014, equal to 1
in the second quarter of 2014.
Table A1 presents the Granger causality test between IPCA and RER,
and Figure A1 presents the impulse-response function of the VAR model
for these two variables. The test and the model were estimated with 12
lags based on the data from January 1998 through September 2014.
Table A2 presents the statistics of a group of models, from general
to specific, of the second difference of IPCA (the change in inflation).
Based on the information criteria, the most general model seems to be
the best description of the error-correction process of inflation to its
long-run trend. Analogously, Table A3 presents the statistics of another
group of models, again from general to specific, of the second
difference of GDP (the acceleration of economic growth). Based on the
information criteria, the most general model seems to be the best
description of the data.
[FIGURE A1 OMITTED]
Table Al: Granger causality test between the first difference of
log(IPCA) and log(NER)
Null Hypothesis F-statistic Probability
DLOG(IPCA) does not Granger Cause 1.574 0.102
DLOG(NER)
DLOG(NER) does not Granger Cause 2.549 0.003
DLOG(IPCA)
Obs: DLOG(X) means the first difference of the natural log of X.
Source: Author's estimate
Table A2: Econometric models for the second difference of log(IPCA)
Model 1 Model 2 Model 3
Constant -24.198 0.582 -0.062
(0.0001) (0.302) (0.020)
D(DLOG(IPCA(-1))) 0.200 0.230 0.183
(0.056) (0.057) (0.085)
D(DLOG(IPCA(-2))) -0.119 -0.104 -0.141
(0.251) (0.340) (0.163)
DLOG(IPCA(-1)) -1.007 -0.841 -0.719
(0.000) (0.000) (0.000)
DLOG(REER(-1)) 0.025 0.026 0.029
(0.032) (0.082) (0.090)
LOG(REER(-1)) 15.903 -0.265 0.016
(0.0001) (0.284) (0.008)
LOG[(REER(-1)).sup.2] -3.480 0.030
(0.0001) (0.259)
LOG[(REER(-1)).sup.3] 0.253
(0.0001)
GAP(-l) 0.073 0.085 0.110
(0.094) (0.085) (0.051)
[R.sup.2] 0.624 0.544 0.523
Adjusted [R.sup.2] 0.572 0.490 0.475
Log-likelihood 251.380 244.894 243.360
Akaike info criterion -7.235 -7.071 -7.055
Schwarz criterion -6.939 -6.808 -6.825
Durbin-Watson stat 1.815 1.939 2.079
Model 4 Model 5
Constant 0.006 0.007
(0.007) (0.001)
D(DLOG(IPCA(-1))) 0.046 0.193
(0.676) (0.095)
D(DLOG(IPCA(-2))) -0.240 -0.247
(0.044) (0.056)
DLOG(IPCA(-1)) -0.416 -0.484
(0.014) (0.005)
DLOG(REER(-1)) 0.039
(0.064)
LOG(REER(-1))
LOG[(REER(-1)).sup.2]
LOG[(REER(-1)).sup.3]
GAP(-l) 0.042 0.038
(0.298) (0.444)
[R.sup.2] 0.454 0.359
Adjusted [R.sup.2] 0.409 0.318
Log-likelihood 238.854 233.474
Akaike info criterion -6.950 -6.820
Schwarz criterion -6.753 -6.655
Durbin-Watson stat 2.124 2.006
Note: DLOG(X) means the first difference of the natural log of X and
[X.sup.n] means the nth power of X. The probability for a one-sided
t-test is shown in parentheses below each coefficient estimate.
Source: Author's estimates
Table A3: Econometric models for the second difference of log(GDP)
Model 1 Model 2 Model 3
Constant -1.596 -1.048 -0.023
(0.001) (0.020) (0.545)
D(DLOG(GDPSA(-1))) -0.059 0.068 0.004
(0.692) (0.707) (0.978)
D(DLOG(GDPSA(-2))) 0.009 0.047 0.043
(0.950) (0.716) (0.778)
DLOG(GDPSA(-1)) -1.010 -0.924 -1.022
(0.000) (0.000) (0.000)
DLOG(REER(-1)) -0.019 -0.033 -0.023
(0.160) (0.070) (0.100)
LOG(REER(-1)) 0.689 0.450 0.007
(0.001) (0.022) (0.407)
LOG(REER(-1))2 -0.073 -0.048
(0.001) (0.024)
DAPAGAO -0.012 -0.011
(0.0002) (0.004)
DCRISE -0.037 -0.033
(0.000) (0.000)
DWCUP -0.018 -0.014
(0.000) (0.000)
[R.sup.2] 0.642 0.452 0.578
Adjusted [R.sup.2] 0.585 0.396 0.519
Log-likelihood 219.850 205.724 214.401
Akaike info criterion -6.359 -6.021 -6.224
Schwarz criterion -6.027 -5.789 -5.925
Durbin-Watson stat 1.760 1.995 1.700
Model 4 Model 5
Constant 0.008 -0.026
(0.0006) (0.383)
D(DLOG(GDPSA(-1))) 0.009 0.103
(0.954) (0.571)
D(DLOG(GDPSA(-2))) 0.046 0.069
(0.760) (0.595)
DLOG(GDPSA(-1)) -1.022 -0.941
(0.000) (0.000)
DLOG(REER(-1)) -0.021 -0.034
(0.118) (0.062)
LOG(REER(-1)) 0.007
(0.275)
LOG(REER(-1))2
DAPAGAO -0.009
(0.002)
DCRISE -0.034
(0.000)
DWCUP -0.014
(0.000)
[R.sup.2] 0.568 0.423
Adjusted [R.sup.2] 0.516 0.375
Log-likelihood 213.608 204.062
Akaike info criterion -6.230 -6.001
Schwarz criterion -5.965 -5.802
Durbin-Watson stat 1.676 2.000
Note: DLOG(AQ means the first difference of the natural log of X and
[X.sup.n] means the nth power of X. The probability for a one-sided
t-test is shown in parentheses below each coefficient estimate.
Source: Author's estimates
REFERENCES
Bacha, EL and Bonelli, R. 2005: Accounting for Brazil's growth
deceleration. Brazilian Journal of Political Economy 25(3): 163-189.
Barbosa-Filho, NH. 2010: Duas nao linearidades e uma assimetria.
Paper presented at the 7th Economic Forum of the Getulio Vargas
Foundation, Sao Paulo.
Barbosa-Filho, NH. 2014: A structuralist inflation curve.
Metroeconomica 65(2): 349-376.
Barbosa-Filho, NH, Silva, JA, Goto, F and Silva, B. 2011:
Crescimento economico, acumulacao de capital e taxa de cambio. In:
Holland, M and Nakano, Y. (ogs). Taxa de Cambio no Brasil: Estudos de
uma perspectiva do desenvolvimento economico. Elsevier: Sao Paulo.
Belaisch, A. 2003: Exchange-rate pass-through in Brazil. IMF
Working paper 03/141.
Bogdanski, J, Tombini, A and Werlang, S. 2001: Implementing
inflation targeting in Brazil. Banco Central do Brasil, Working paper 1.
Bresser-Pereira, LC. 2007: Macroeconomia da Estagnacao. Editora 34:
Sao Paulo.
Frenkel, R and Taylor, L. 2006: Real exchange rate, monetary policy
and employment. DESA Working paper no. 19.
Frenkel, R. 2008: The competitive real exchange-rate regime,
inflation and monetary policy. Cepal Review (96): 189-199.
Giambiagi, F, Vilela, A, Castro, L.B. de C. and Hermann, J.. 2011:
Economia Brasileira Contemporana: 1945-2010. Elsevier: Sao Paulo.
Rapetti, M, Skott, P and Razmi, A. 2012: The real exchange rate and
economic growth: Are developing countries different? International
Review of Applied Economics 26(6): 735-753.
Rego, JM. 1986: Inflacao inercial, teorias sobre inflacao e o Plano
cruzado. Paz e Terra: Sao Paulo.
Rodrik, D. 2008: The real exchange rate and economic growth.
Brookings Papers on Economic Activity 39(2): 365-412.
(1) In Brazil, power comes mostly from hydroelectric plants, which
makes the country's energy supply heavily dependent on the rainy
season.
(2) All inflation numbers mentioned in this paper refer to consumer
prices, measured by the IPCA index, on an annual basis.
(3) Inflation reached 6.4% in 2014.
(4) Throughout this paper, I define the nominal exchange rate as
the domestic price of foreign currency, that is, for example, the amount
of BRL per US dollar. The real exchange rate is this nominal rate
multiplied by the ratio of foreign to domestic prices, with the former
defined by the BCB based on the composition of Brazilian exports. As a
result, an increase in the exchange rate means a depreciation of the
domestic currency and vice versa.
(5) MPOG stands for Ministerio do Planejamento, Oricamento e Gestao
and CCPR stands for Casa Civil da Presidencia da Republica. Formally,
the BCB is still a government agency subordinated to the MoF, but its
President has a cabinet-level position. In practice, the BCB functions
as an independent agency. The CCPR is headed by the President's
Chief of Staff, who also has a cabinet-level position.
(6) Brazilian Presidential Decree 3.088, of 1999. For the beginning
of inflation targeting in Brazil, see Bogdanski et al. (2001).
(7) IOF stands for Imposto sobre Operates Financeiras. The IOF law
states what can be taxed and the maximum rates on each operation. Below
these ceilings, the President can change the IOF rates without Congress
approval.
(8) This JEO gained importance after Brazil's 'Fiscal
Responsibility Law' (Lei Complementar 101), adopted in 2000.
(9) This practice does not happen without political noise and, in
recent years, there have been many Congressional attempts to limit the
JEO's powers.
(10) For a more detailed analysis of the Brazilian economy during
this period, see Giambiagi et al. (2011).
(11) All growth rates are annual rates.
(12) Power prices rose 17.3% in 2001, while the IPCA index rose
7.7%.
(13) After the economy absorbed the maxi-depreciation of the BRL in
1999, GDP growth accelerated to 4.3 % in 2000, but the recovery was
short-lived.
(14) The slowdown in Argentina also pulled Brazil's GDP down
during this period.
(15) Assuming zero GDP growth in 2014.
(16) Credit expansion took the form of subsidized government loans
to its banks, especially to its development bank, which in turn lent the
funds at subsidized rates to the market.
(17) All interest rates were calculated as the cumulative rate over
the past 12 months.
(18) The BCB's data on the expected--'ex ante'--real
interest rate starts only in 2001, but it shows almost the same pattern,
with a one-year lead of the expected rate over the effective rate. To
simplify the exposition, I will focus only on the effective real
interest rate.
(19) This period marked the peak of the 'currency-war'
speech of Brazilian authorities, which not surprisingly coincided with
the peak in the Brazilian terms of trade, pulled by high commodity
prices. Until 2011, most FX actions by the MoF were focused on taxing
short-term capital inflows, with limited impact on the exchange rate.
Only when the MoF changed course and started to tax the net exposure in
FX derivatives markets did the BRL/USD move significantly.
(20) This marked the end of the 'currency war', at least
for Brazilian authorities.
(21) The VAR model with 12 lags is shown in the statistical
appendix. All series used in the econometric model are available from
the author upon request.
(22) For a previous estimate of the exchange-rate pass-through, see
Belaisch (2003).
(23) More formally, recall that, in log levels, the change in the
real exchange rate is the change in the nominal rate plus foreign
inflation minus domestic inflation. Therefore, if the variation in the
real exchange rate is 0.2 and the variation in domestic prices is 0.14
times the variation in the nominal exchange rate, the variation in
domestic prices is 0.2 minus foreign inflation times 0.14/0.86. For a
foreign inflation rate of 0.02, the result is 0.18 x 0.14/0.86 = 2.9%.
(24) As we commented in the previous section, the Brazilian
authorities relied on government subsidies to smooth the price
adjustment in some regulated markets in 2012-14, which in turn reduced
its primary balance.
(25) For the details of this model, see Barbosa-Filho (2014).
(26) Mathematically, since the data-generating process is the same,
the univariate model represents a linear transformation of a
vector-error correction model containing domestic price, the nominal
exchange rate and the foreign price, with the latter being exogenous.
Since the focus of this paper is economic policy rather than
econometrics, I restrict the statistical results to their univariate and
more intuitive representation.
(27) The econometric model is an updated version of the one
presented in Barbosa-Filho (2010).
(28) See, for instance, Frenkel and Taylor (2006), Bresser-Pereira
(2007), Frenkel (2008) and Rapettiefai. (2012).
(29) The statistical results are shown in the appendix. The model
is an updated version of the one presented in Barbosa-Filho et al.
(2011). By analogy with the inflation model, the univariate GDP model
can be interpreted as a linear transformation of a vector
error-correction model containing GDP, the domestic price, the nominal
exchange rate and the foreign price, with the latter being exogenous.
(30) For an application of the AK model to Brazil, see Bacha and
Bonelli (2005).
(31) More formally, define economic growth as g = su/p, where s is
the investment-GDP ratio at current prices, u the income-capital ratio
and p the relative price of investment in terms of the GDP deflator. For
a given capital productivity (u), the concave-down growth curve can be
justified from the fact that both s and p are positive functions of the
real exchange rate.
(32) The inflation target is one of the determinants of the
intercept coefficient of the inflation curve in Figure 5. For details on
this, see Barbosa-Filho (2014). For the importance of inflation inertia
in Brazil, see Rego (1986).
(33) For the substitution between the real exchange rate and
institutional or structural change in accelerating growth, see Rodrik
(2008).
(34) The usual candidates in the current Brazilian economic debate
are tax reform, social security reform, labor reform, higher investment
in education and infrastructure, financial deepening, and prudential
financial regulation--including the regulation of capital flows and
derivatives operations.
NELSON H. BARBOSA-FILHO
EESP, FGV, Rua Itapeva 474, Sao Paulo, SP 01332-000, Brazil.
E-mail: nelson.barbosa@fgv.br
Table 1: Summary of the Brazilian macroeconomic data in
1999-14
Variable 1997-99 2000-02 2003-05
Average GDP annual growth rate 1.2% 2.8% 3.3%
Average annual inflation rate (IPCA) 5.2% 8.7% 7.5%
Average RER (Jul/1994 = 100) 81.3 117.0 128.0
Change in the RER during the period 51.7% 26.0% -17.1%
Average SELIC nominal interest rate 26.4% 18.0% 19.5%
Average SELIC real interest rate 20.1% 8.5% 11.2%
(ex post)
Average current-account balance in % -3.9 -3.2 1.4
of GDP
Average primary surplus of the 0.8 1.9 2.5
federal government in % of GDP (b)
Variable 2006-08 2009-11 2012-14 (a)
Average GDP annual growth rate 5.1% 3.3% 1.2%
Average annual inflation rate (IPCA) 4.5% 5.6% 6.0%
Average RER (Jul/1994 = 100) 93.1 80.3 88.8
Change in the RER during the period -19.3% -15.6% 22.4%
Average SELIC nominal interest rate 13.1% 10.4% 9.0%
Average SELIC real interest rate 8.3% 4.6% 2.7%
(ex post)
Average current-account balance in % -0.1 -1.9 -3.3
of GDP
Average primary surplus of the 2.4 1.6 1.0
federal government in % of GDP (b)
(a) Based on the author's forecasts for 2014.
(b) Excludes balance-sheet operations with Petrobras and the
Brazilian Sovereign Wealth Fund.
Source: BCB, MF and IPEADATA; author's elaboration
Table 2: Effective inflation, target inflation, nominal exchange rate
and real exchange rate in Brazil: 1998-2013
Year Central Interval Effective Did the Nominal
inflation of inflation BCB exchange rate.
target tolerance meet the BRL/USD,
target? annual average
1998 NA NA 1.66 NA 1.16
1999 8.00 2.00 8.94 YES 1.81
2000 6.00 2.00 5.97 YES 1.83
2001 4.00 2.00 7.67 NO 2.35
2002 3.50 2.00 12.53 NO 2.92
2003 3.25 2.00 9.30 NO 3.08
2004 3.75 2.50 7.60 NO 2.93
2005 4.50 2.50 5.69 YES 2.44
2006 4.50 2.00 3.14 YES 2.18
2007 4.50 2.00 4.46 YES 1.95
2008 4.50 2.00 5.90 YES 1.83
2009 4.50 2.00 4.31 YES 2.00
2010 4.50 2.00 5.91 YES 1.76
2011 4.50 2.00 6.50 YES 1.67
2012 4.50 2.00 5.84 YES 1.95
2013 4.50 2.00 5.91 YES 2.16
Year % change in Real effective % change in
the nominal exchange rate, the nominal
exchange index, Jun/ exchange
rate 94 = 100 rate
1998 NA 70.55 NA
1999 56.37 105.72 49.86
2000 0.85 97.12 -8.13
2001 28.42 120.56 24.13
2002 24.28 133.19 10.47
2003 5.38 137.87 3.52
2004 -4.95 135.52 -1.71
2005 -16.77 110.47 -18.49
2006 -10.64 98.70 -10.65
2007 -10.49 91.60 -7.20
2008 -5.82 89.10 -2.73
2009 8.89 88.51 -0.67
2010 -11.89 77.17 -12.80
2011 -4.84 75.17 -2.60
2012 16.69 84.33 12.18
2013 10.39 90.12 6.87
Year Comments
1998 There was no inflation target
1999 Target defined in Jun/1999
2000 Target defined in Jun/1999
2001 Target defined in Jun/1999
2002 Target defined in Jun/2000
2003 Target defined in Jun/2001, later changed to 4%,
with an interval of tolerance of 2.5 pp, in Jun/
2002, and changed again to 8.5%, without
interval of tolerance, in Jan/03
2004 Target defined in Jun/02 and later changed to
5.5%, without interval of tolerance, in Jan/03
2005 Target defined in Jun/2003
2006 Target defined in Jun/2004
2007 Target defined in Jun/2005
2008 Target defined in Jun/2006
2009 Target defined in Jun/2007
2010 Target defined in Jun/2008
2011 Target defined in Jun/2009
2012 Target defined in Jun/2010
2013 Target defined in Jun/2011
Source: BCB and author's construction
